Insulin signaling acts in adult adipocytes via GSK-3β and independently of FOXO to control Drosophila female germline stem cell numbers

Abstract

Tissue-specific stem cells are tied to the nutritional and physiological environment of adult organisms. Adipocytes have key endocrine and nutrient-sensing roles and have emerged as major players in relaying dietary information to regulate other organs. For example, previous studies in Drosophila melanogaster revealed that amino acid sensing as well as diet-dependent metabolic pathways function in adipocytes to influence the maintenance of female germline stem cells (GSCs). How nutrient-sensing pathways acting within adipocytes influence adult stem cell lineages, however, is just beginning to be elucidated. Here, we report that insulin/insulin-like growth factor signaling in adipocytes promotes GSC maintenance, early germline cyst survival, and vitellogenesis. Further, adipocytes use distinct mechanisms downstream of insulin receptor activation to control these aspects of oogenesis, all of which are independent of FOXO. We find that GSC maintenance is modulated by Akt1 through GSK-3β, early germline cyst survival is downstream of adipocyte Akt1 but independent of GSK-3β, and vitellogenesis is regulated through an Akt1-independent pathway in adipocytes. These results indicate that, in addition to employing different types of nutrient sensing, adipocytes can use distinct axes of a single nutrient-sensing pathway to regulate multiple stages of the GSC lineage in the ovary.

INTRODUCTION

Adult tissue homeostasis relies heavily on stem cells, especially in high turnover organs such as the intestine, skin, or blood (Barker, 2014; Goodell et al., 2015). Tissue stem cell lineages are physiologically regulated according to diet, stress, injury, and age, through mechanisms involving extensive inter-organ communication (Ables et al., 2012). Understanding the role of adipocytes in the regulation of stem cells is particularly relevant given their crucial roles in regulating whole-body physiology and the link between malfunctioning adipocytes and a number of pathologies, including wound healing defects and cancers (Fasshauer and Bluher, 2015; Rosen and Spiegelman, 2014).

Germline stem cells (GSCs) in the adult Drosophila ovary are a useful system for investigating how stem cell lineages are shaped by physiological inputs (Laws and Drummond-Barbosa, 2017). Each ovary contains 16–20 ovarioles, and the anterior region of each ovariole, the germarium, contains two to three GSCs in a somatic niche composed primarily of cap cells (Fig. 1A–C) (Laws and Drummond-Barbosa, 2017). Each GSC divides asymmetrically to self-renew and generate a cystoblast that forms two, four-, eight-, and 16-cell germline cysts through successive, synchronized mitotic divisions with incomplete cytokinesis. One of the cyst cells becomes the oocyte, which is supported by the remaining 15 nurse cells. Follicle cells surround each germline cyst to form an individual egg chamber, or follicle. The follicle develops through morphologically distinct stages, including vitellogenesis, to give rise to a stage 14 mature oocyte that is ovulated, fertilized and laid (Laws and Drummond-Barbosa, 2017). Oogenesis is markedly responsive to diet, resulting in up to 60-fold changes in egg laying rates on yeast-rich versus –free diets (Drummond-Barbosa and Spradling, 2001).

Fig. 1.

Insulin signaling in adult adipocytes cell autonomously controls cell size. (A) Diagram illustrating that adipocytes and oenocytes constitute the Drosophila fat body, which underlies the cuticle and surrounds the ovaries (composed of ovarioles). (B) Diagram of ovariole showing the anterior germarium (g) and developing egg chambers (or follicles), each composed of a germline cyst (one oocyte and 15 nurse cells) surrounded by follicle cells. “Pre-vitellogenic” and “vitellogenic” refer to stages prior to or after the onset of yolk uptake by the oocyte, respectively. (C) Diagram of germarium showing GSCs housed in direct contact with cap cells. GSCs generate cystoblasts, which in turn incompletely divide four times to form germline cysts containing 16 interconnected cells. The position and morphology of a special organelle, the fusome (orange), allows for the identification of GSCs, cystoblasts, two-, four-, eight-, and 16-cell germline cysts. (D–F) Adipocytes from females subjected to adipocyte-specific manipulation of insulin signaling. Gal80^ts; Lsp2-Gal4 drove expression of a UAS-nuclear GFP reporter in combination with other UAS transgenes (see Table S1), and females were raised at 18°C and switched to 29°C four days after eclosion for 10 days to induce transgene expression. Nuclear GFP intensity varies between adipocytes indicative of variable driver expression. Adipocytes with relatively robust nuclear GFP expression (arrowheads indicate some examples) were used for quantification shown in (G). GFP (green); E-cadherin (red), cell membranes. Scale bar, 25 μm. (G) Average cell area of adipocytes upon adult adipocyte-specific expression of various transgenes at 29°C for 10 days. GFP^dsRNA was used as RNAi control. Mean + s.e.m. Number of adipocytes quantified are shown inside bars. ****P < 0.0001; ordinary one-way ANOVA. See supplementary material Fig. S1 for additional transgenes tested.

Many steps of oogenesis are highly regulated by diet to ensure that egg production is precisely coordinated with the physiology of the organism as a whole (Laws and Drummond-Barbosa, 2017). In well-fed females, robust GSC proliferation and development of their progeny provide a steady supply of new eggs. Upon removal of dietary yeast, GSCs and their progeny divide more slowly, early germline cysts die at higher rates, follicle growth is reduced, the vast majority of early vitellogenic follicles degenerate, and a partial block in ovulation causes prolonged retention of mature eggs in ovarioles (Drummond-Barbosa and Spradling, 2001). In addition, pre-vitellogenic follicles accumulate enlarged P bodies and undergo microtubule rearrangement (Shimada et al., 2011). Prolonged starvation also leads to increased rates of GSC loss (Hsu and Drummond-Barbosa, 2009). Many diet-dependent pathways, including insulin, Target of Rapamycin (TOR), AMP-dependent kinase, and nuclear hormone signaling function in the ovary to mediate these effects of diet (Laws and Drummond-Barbosa, 2017). Adipocytes, which in Drosophila co-exist with hepatocyte-like oenocytes in an organ called the fat body (Hoshizaki et al., 1995), also communicate their dietary status to the GSC lineage through as yet unknown systemic signals. Reduced amino acid transport into adipocytes causes increased GSC loss and a partial block in ovulation on a yeast-rich diet (Armstrong et al., 2014), whereas a number of diet-regulated metabolic pathways in adipocytes have specific effects on GSC number and early cyst survival (Matsuoka et al., 2017).

The highly conserved insulin/insulin-like growth factor pathway is a prominent and multifaceted actor in the regulation of oogenesis. In Drosophila, insulin-like peptides bind and activate a single receptor tyrosine kinase, the insulin receptor (InR) (Nassel et al., 2015). In response to InR activation, phosphoinositide-3-kinase (PI3K) phosphorylates phosphatidylinositol-4-5-diphosphate (PIP2) to produce phosphatidylinositol-triphosphate (PIP3), which recruits the serine/threonine kinase Akt1 to the membrane, allowing its phosphorylation and activation. Akt1 has many substrates that regulate a variety of cellular processes. For example, glycogen synthase kinase-3 β (GSK-3β) and the transcription factor Forkhead Box O (FOXO) are inhibited by Akt1, whereas phosphorylation of the tuberous sclerosis complex (TSC) by Akt1 leads to activation of the TOR-containing protein kinase complex mTORC1 (Manning and Toker, 2017; Nassel et al., 2015). Brain-derived insulin-like peptides act directly on the germline to control GSC proliferation, follicle growth and vitellogenesis (LaFever and Drummond-Barbosa, 2005). Distinct downstream effectors, however, are involved in the control of GSC division cycle (PI3K/FOXO) versus growth of their progeny at later stages (PI3K/TOR) (Hsu et al., 2008; LaFever et al., 2010). Insulin signaling acts through FOXO in somatic niche cells to promote GSC maintenance by regulating the number of cap cells and adhesion of GSCs to cap cells (Hsu and Drummond-Barbosa, 2009, 2011; Yang et al., 2013). It is also required in follicle cells for P body and microtubule responses in the underlying germline of previtellogenic follicles (Burn et al., 2015) and for the mitotic-to-endocycle switch (Jouandin et al., 2014). The reproductive function of the insulin pathway is conserved from C. elegans to mammals, where insulin signaling also plays important roles in oocyte growth, development and maturation (Das and Arur, 2017).

Insulin signaling in adult adipocytes has been shown to control fat storage and lifespan in Drosophila (DiAngelo and Birnbaum, 2009; Giannakou et al., 2004; Hwangbo et al., 2004); however, it has remained unknown how insulin signaling in adipocytes might modulate the adipocyte-ovary communication axis. In this study, we find that insulin signaling within adipocytes promotes GSC maintenance, survival of early germline cysts, and vitellogenesis. Akt1 and GSK-3β mediate the effect of adipocyte insulin signaling on GSCs independently of FOXO. By contrast, early germline cyst survival and vitellogenesis do not require either GSK-3β or FOXO in adipocytes. These results indicate that insulin signaling in adult adipocytes employs distinct downstream effectors to regulate different steps of GSC lineage development in the ovary.

MATERIALS AND METHODS

Drosophila strains and culture conditions

Fly stocks were maintained at 22–25°C on standard medium containing cornmeal, molasses, yeast, and agar. Standard medium supplemented with wet yeast paste was used for all experiments. The adipocyte-specific Gal4 line, 3.1Lsp2-Gal4 (Lazareva et al., 2007), and temperature-sensitive tub-Gal80^ts (McGuire et al., 2003) transgenes were used as previously described (Armstrong et al., 2014). UAS-RNA hairpin lines obtained from the Vienna Drosophila RNAi stock center (http://stockcenter.vdrc.at), and the Transgenic RNAi Project (http://www.flyrnai.org) collection at the Bloomington Drosophila Stock Center (http://flystocks.bio.indiana.edu) for manipulation of insulin signaling components are listed in Table S1. Other transgenic lines used were: UAS-InR (Brogiolo et al., 2001; Huang et al., 1999), UAS-Pten (II) (Potter et al., 2001), UAS-Pten (III) (Gao et al., 2000), UAS-Pten.ORF (FlyORF), UAS-foxo (Hwangbo et al., 2004), UAS-foxo (III) (Junger et al., 2003), UAS-sgg^S9A (Bourouis, 2002), ilp5-lacZ (Ikeya et al., 2002), and tGPH (Britton et al., 2002). Other genetic elements used are described in FlyBase (http://www.flybase.org).

Adult adipocyte-specific genetic manipulations

For adult adipocyte-specific genetic manipulation, females of genotypes yw; tubP-Gal80^ts/+; 3.1Lsp2-Gal4/UAS-X or yw; tub-Gal80^ts/UAS-X; 3.1Lsp2-Gal4/+ were used. (UAS-X represents any of the UAS transgenes in this study.) Females were raised at 18°C, the permissive temperature for Gal80^ts (McGuire et al., 2003), to keep transgene expression off during development. Newly eclosed females were maintained at 18°C for three days and then switched to 29°C, the restrictive temperature for Gal80^ts (McGuire et al., 2003), for various lengths of time to induce transgene expression prior to dissection and analyses, as described (Armstrong et al., 2014). Females were housed with yw males and fed wet yeast daily.

Immunostaining and fluorescence microscopy of ovaries, fat bodies, and brains

All tissues were dissected in Grace’s medium (BioWhittaker) and fixed in 5.3% formaldehyde (Ted Pella) in Grace’s medium at room temperature for 13 minutes for ovaries, 20 minutes for abdominal carcasses (containing attached fat body), or 30 minutes for brains. Tissues were rinsed and washed three times in 0.1% Triton X-100 (Sigma) in phosphate-buffered saline (PBS), or PBT, and subsequently blocked in 5% bovine serum albumin (BSA; Sigma) and 5% normal goat serum (NGS; Jackson ImmunoResearch) in PBT, or blocking solution, for three hours at room temperature or overnight at 4°C. Tissues were incubated overnight at 4°C in the following primary antibodies diluted in blocking solution: rabbit anti-GFP (Torrey Pines, 1:2,500); mouse monoclonal anti-Hts (1B1) (DSHB; 1:10); mouse anti-α-spectrin (3A9) (DSHB; 1:50); mouse monoclonal anti-Lamin C (LC28.26) (DSHB; 1:100); rat monoclonal anti-E-cadherin (DCAD2) (DSHB, 1:10 for fat bodies and 1:3 for ovaries); rabbit anti-Dcp1 (#9578) (Cell Signaling, 1:200); rabbit anti-FOXO [(Slaidina et al., 2009), 1:500]; rat anti-ILP2 [(Geminard et al., 2009), 1:800]; chicken anti-β-galactosidase (β-gal) (Abcam ab9361, 1:2000); and rabbit anti-pMad (Smad 3, #1880) (Epitomics; 1:100). [This Smad 3 antibody is routinely used in Drosophila to detect pMad specifically (Ables and Drummond-Barbosa, 2010; Armstrong et al., 2014; Hayashi et al., 2009; Issigonis and Matunis, 2012; Laws et al., 2015; Ma et al., 2014; Matsuoka et al., 2017; Sulkowski et al., 2014)]. Tissues were washed in PBT and incubated for two hours in the dark at room temperature in 1:200 AlexaFluor 488- or 568-conjugated secondary antibodies (Molecular Probes). Samples were washed, and ovaries, brains, and fat bodies (scraped off from carcasses) were mounted in Vectashield containing 4′,6′-diamidino-2-phenylindole (DAPI) (Vector Labs). Data were collected with a Zeiss AxioImager-A2 fluorescence microscope or a Zeiss LSM700 confocal microscope.

Cap cells were identified based on their ovoid shape and Lamin C-positive staining, and GSCs were identified based on their juxtaposition to cap cells and fusome morphology and position, as described previously (de Cuevas and Spradling, 1998). For nuclear pMad quantification, the densitometric mean of individual GSC nuclei was measured from optical sections containing the largest nuclear diameter (visualized by DAPI) using AxioVision. Total E-cadherin levels at the cap cell-GSC junction were measured using morphometric analysis of maximum intensity projections spanning the width of contact in single z-planes. For consistency, ovary samples were dissected, fixed and stained in parallel under identical conditions. For each analysis, image acquisition settings were exactly the same for all images used for quantification.

Quantification of early germ cell death and degenerating vitellogenic follicles

For analysis of early germ cell death, the number of Dcp1-positive germ cells in the germarium was counted. Ovarioles containing vitellogenic follicles were easily distinguished from those with blocked vitellogenesis, which contained at least one dying vitellogenic follicle, recognizable by their position within the ovariole relative to neighboring follicles and by the presence of pyknotic nuclei visualized by DAPI.

Analyses of systemic insulin signaling

Two approaches were used to measure systemic insulin signaling: 1) secretion of ILP2 from insulin producing cells (IPCs) in the brain, and 2) tGPH reporter activity in the ovary. Since IPCs secrete ILPs under nutrient-rich conditions (active insulin signaling) and retain ILPs under nutrient-poor conditions (reduced insulin signaling)(Geminard et al., 2009), ILP2 immunofluorescence intensity in IPCs was used as a measure of ILP retention versus secretion. Brains obtained from yw; tub-Gal80^ts, ilp5-lacZ/+; 3.1Lsp2-Gal4/UAS-X or yw; tub-Gal80^ts, ilp5-lacZ/UAS-X; 3.1Lsp2-Gal4/+ females were co-labeled with anti-β-gal and anti-ILP2 and images were acquired along the z-axis. Since IPCs express the ilp5-lacZ transcriptional reporter constitutively, β-gal was used as an independent IPC marker. ILP2 fluorescence intensity was measured by recording the total intensity of maximum intensity projections spanning the IPC cluster region. For control samples, fold change for individual brains was calculated by dividing the total IPC intensity of each individual brain by the average total IPC intensity of all control brains, resulting in the average fold change being equal to one. For UAS-X samples, fold change relative to control was obtained by dividing the total IPC intensity for each individual brain by the average total intensity of control brain IPCs. Insulin signaling activity in the ovary was measured using the tGPH reporter (Britton et al., 2002) by determining the membrane-to-cytoplasmic GFP ratio from nurse cells of stage 10 follicles. Using ImageJ, a line was drawn spanning the membrane between nurse cells and intensity values along the line were recorded and plotted (see Fig. S3). Thus, for each line a bell curve was generated with lower intensity values (cytoplasmic staining) flanking high intensity values (membrane staining)(see Fig. S3). Membrane GFP intensity was defined as the average of the following seven measurements on the curve: peak value, three values before and three values after the peak value. Cytoplasmic GFP intensity was defined as the average of the first three and last three values of the curve, allowing the GFP intensity for both nurse cell cytoplasmic regions to be measured. Membrane and cytoplasmic GFP intensities were taken for three nurse cell-nurse cell contacts per follicle. The membrane-to-cytoplasmic GFP ratio was calculated as the average membrane GFP intensity divided by the average cytoplasmic intensity. For consistency, ovary or brain tissues were dissected, fixed and stained in parallel under identical conditions. For each analysis, image acquisition settings were exactly the same for all images used for quantification.

Measurement of adipocyte cell size

ImageJ was used to measure the sizes of individual adipocytes in fat bodies. Within a single tissue, Gal4 expression varies as evidenced by variable GFP intensities; therefore, cell size measurements were taken of cells with qualitatively strong nuclear GFP expression. To measure adipocyte area, the largest cell diameter (visualized by E-cadherin) was selected for multiple cells (5–20, depending on how many had strong GFP expression) per image. For each genotype, three images were acquired resulting in measurement of 15–60 adipocytes per genotype.

RESULTS

Insulin signaling in adult adipocytes controls cell size

To investigate the role of adult adipocyte-specific insulin signaling in oogenesis, we used the adipocyte-specific Gal4 driver 3.1Lsp2-Gal4 in combination with a temperature-sensitive tub-Gal80^ts transgene (3.1Lsp2^ts) as previously described (Armstrong et al., 2014) to express transgenes targeting components of the insulin pathway (Fig. S1 and Table S1). InR/PI3K signaling promotes adipocyte growth during larval development (Britton et al., 2002); we therefore reasoned that insulin signaling might also regulate adipocyte size in adults, which could serve as a measure of transgene effectiveness. We combined the 3.1Lsp2^ts driver with a UAS-nuclear GFP reporter to focus our cell size measurements on adipocytes with the strongest levels of transgene expression as indicated by GFP intensity (Fig. 1D–G and Fig. S1). Adult adipocyte-specific RNAi knockdown of the insulin receptor (InR) or its effector kinase Akt1, or overexpression of a negative regulator of insulin signaling, Phosphatase and tensin homolog (Pten), resulted in notable reduction of average adipocyte cell size as compared to GFP RNAi controls or adipocytes overexpressing wild-type InR (Fig. 1D–G and Fig. S1). For most of the manipulations inhibiting insulin signaling, the average adipocyte size was decreased by at least 30% (Fig. S1). Based on these results, we performed the majority of our subsequent analyses using UAS-hairpin RNA lines against InR (InR^GD104 and InR^GL00139) and Akt1 (Akt1^GD1361, Akt1^KK100495, Akt1^HMS00007, and Akt1^HM04007), which have strong effects on adipocyte cell size.

Adipocyte insulin signaling controls GSC maintenance

To investigate the role of adipocyte insulin signaling in oogenesis, we first asked whether GSC maintenance is perturbed in response to knockdown of InR or Akt1 specifically in adult adipocytes. GSC numbers declined significantly faster over time in females with InR or Akt1 adipocyte knockdown relative to controls (Fig. 2A), with 25% and 50% of germaria, respectively, containing 0 or 1 GSCs compared to 12% in control females at 10 days of RNAi (Fig. 2B). Adipocyte-specific overexpression of Pten also led to a significant decline in GSC numbers over time relative to controls (Fig. 2A,B).

Fig. 2.

Inhibition of insulin signaling in adult adipocytes leads to increased rates of GSC loss. (A) Average number of GSCs per germarium at 0 or 10 days of adult adipocyte-specific RNAi against InR or Akt1 compared to GFP RNAi control or overexpression of Pten compared to Gal80^ts; Lsp2-Gal4 only control. Mean + s.e.m. *P < 0.05, **P < 0.01; two-way ANOVA with interaction. (B) Graph showing frequencies of germaria containing zero-or-one, two, or three-or-four GSCs representing the same data used to calculate average GSC number in (A). Number of germaria analyzed indicated inside bars. (C–F) Scatter plot of total E-cadherin intensity at the cap cell-GSC junction (C) with representative images for adult adipocyte-specific RNAi against GFP control (D), InR (E) or Akt1 (F). Mean + s.e.m. Scale bar, 2.5 μm. ****P < 0.0001, no statistically significant difference between Akt1 and control GFP knockdown; ordinary one-way ANOVA. (G–J) Mean nuclear phosphorylated Mad (pMad) intensity in GSCs (G) with representative images for adult adipocyte-specific RNAi against GFP control (H), InR (I) or Akt1 (J). Mean + s.e.m. No statistically significant differences; ordinary one-way ANOVA. Number of GSCs analyzed indicated below each data set (C and G). 1B1 (red), fusome; Lamin C (red), cap cell nuclear envelopes; E-cadherin (green, D–F); pMad (green, H–J). Scale bar, 5 μm. Asterisks indicate cap cells and visible portions of GSCs are outlined (D–F); GSC nuclei are outlined in (H–J).

Increased GSC loss could result from reduced numbers of cap cells, changes in bone morphogenetic protein (BMP) signaling or E-cadherin levels [both of which are required for GSC maintenance (Song et al., 2002; Xie and Spradling, 1998)], or GSC death. Cap cell numbers remained comparable to those of controls in females with adult adipocyte-specific knockdown of InR or Akt1 (Fig. S2). Quantification of E-cadherin levels at the cap cell-GSC junction, and of the BMP signaling reporter phosphorylated Mad (pMad) (Kai and Spradling, 2003) in GSCs revealed no reduction in adhesion or niche signaling (Fig. 2C–J). Finally, we did not detect apoptotic GSCs in any of the genotypes based on the cleaved Dcp-1 marker (McCall and Steller, 1998) (338, 300, and 136 germaria analyzed for adipocyte RNAi against GFP control, InR and Akt1, respectively). These data suggest that adipocyte insulin signaling does not influence GSC niche size or function, and that insulin signaling in adipocytes likely controls other factors that act in parallel to BMP signaling and E-cadherin to promote GSC maintenance. We cannot rule out, however, that dying GSCs might be rapidly cleared and/or use a Dcp-1-independent mechanism of cell death in the absence of adipocyte insulin signaling.

Insulin signaling in adipocytes promotes the survival of early germline cysts and vitellogenic follicles

We next examined whether early germline cysts are influenced by insulin signaling levels in adipocytes. In contrast to our findings for GSCs, we observed increased death of early germline cysts in the germarium, particularly in region 1, which contains GSCs and their mitotically dividing progeny (Fig. 3A–D). Upon InR or Akt1 knockdown in adult adipocytes, a significantly higher percentage of germaria showed cleaved Dcp-1 immunoreactivity in the germline relative to GFP RNAi controls (Fig. 3A–D; 39% for GFP RNAi control compared to 67% for InR and 49% for Akt1 RNAi). In addition, females with InR or Akt1 knockdown in adult adipocytes contained two to three times as many germaria with two or more Dcp1-positive early germline cysts as control knockdown females (Fig. 3D, hatched bars; 11% for GFP RNAi control, compared to 35% for InR RNAi and 24% for Akt1 RNAi).

Fig. 3.

Reduced insulin signaling in adult adipocytes causes increased death of early germline cysts and degeneration of vitellogenic follicles. (A–C) Germaria from females at 10 days of adult adipocyte-specific knockdown of GFP control (A), InR (B) or Akt1 (C). 1B1 (red), fusome; Lamin C (red), cap cell nuclear envelopes; Cleaved Dcp-1 (green), dying germline cysts. Scale bar, 10 μm. (D) Percentage of germaria with Dcp-1-positive germline. Hatched and solid regions indicate the percentage of germaria with two or more Dcp-1-positive germline cysts, or just one dying cyst, respectively. Data combine three replicate experiments each for control GFP and InR knockdown and two replicate experiments for Akt1 knockdown. Mean + s.e.m. Total number of germaria analyzed are shown inside bars. **P < 0.01, ***P < 0.001; ordinary one-way ANOVA. (E–G) DAPI-stained ovarioles from GFP control (E), InR (F), and Akt1 (G) knockdown as described in (A–C). Arrow indicates pyknotic nuclei in degenerating follicle. Scale bar, 100 μm. (H) Percentage of ovarioles with degenerating vitellogenic follicles. Mean + s.e.m. Data combine nine replicate experiments each for control GFP and InR knockdown and four replicate experiments for Akt1 knockdown. Total number of ovarioles analyzed are shown inside bars. ****P < 0.0001; ordinary one-way ANOVA.

We found that adipocyte insulin signaling also regulates the survival of follicles undergoing the energy and resource intensive process of vitellogenesis. Upon InR knockdown in adult adipocytes, a higher percentage of ovarioles contain dying vitellogenic follicles compared to RNAi control, based on pyknotic morphology of nuclei recognized by DAPI staining (Fig. 3E–H). Surprisingly, Akt1 knockdown in adult adipocytes did not lead to increased numbers of dying vitellogenic follicles (Fig. 3G and H). These data indicate that insulin signaling in adipocytes controls survival of the germline at two major dietary checkpoints (Drummond-Barbosa and Spradling, 2001), although the survival of vitellogenic follicles appears to be independent of the downstream effector Akt1.

Insulin signaling within adipocytes does not influence systemic insulin signaling

GSC numbers and survival of vitellogenic follicles are controlled by brain-derived insulin-like peptides acting on the GSC niche and germline, respectively (LaFever and Drummond-Barbosa, 2005). In addition, recent studies have shown that brain insulin-like peptide secretion is modulated by fat body-derived factors (Nassel and Vanden Broeck, 2016). We therefore considered that the effect of adipocyte insulin signaling on GSC maintenance and vitellogenesis might involve changes in systemic insulin signaling. To directly address this possibility, we tested if adipocyte insulin signaling alters insulin-like peptide 2 (ILP2) secretion. ILP2 levels in brain insulin-producing cells reflect insulin-like peptide secretion because reduced or increased ILP2 secretion leads to accumulation or depletion of ILP2 in those cells, respectively (Geminard et al., 2009; Rajan and Perrimon, 2012). We therefore measured anti-ILP2 immunofluorescence intensity in brain insulin-producing cells. Females in which control GFP, InR, or Akt1 was knocked down in adult adipocytes had comparable levels of ILP2 in brain insulin-producing cells (Fig. 4A–D). As a complementary approach to assess changes in circulating insulin-like peptides, we measured insulin signaling in ovarian nurse cells using the tGPH reporter, which is recruited to the plasma membrane in response to high PI3K activity (Britton et al., 2002). In accordance with the ILP2 immunofluorescence data, GFP control, InR, and Akt1 knockdown females showed no differences in the membrane-to-cytoplasmic ratio of tGPH in nurse cells (Fig. 4E–H). These results suggest that insulin signaling in adipocytes is not part of a positive feedback loop regulating brain insulin-like peptide secretion. Further, they demonstrate that the increase in GSC loss and degeneration of vitellogenic follicles does not result from changes in circulating insulin-like peptides downstream of adipocyte insulin signaling. This conclusion is further supported by our findings that other aspects of oogenesis known to be controlled by systemic insulin signaling, including E-cadherin levels at the GSC-niche interface and cap cell numbers, remain unaffected upon inhibition of insulin signaling in adult adipocytes (see Fig. 2C–F, and Fig. S2).

Fig. 4.

Insulin signaling in adult adipocytes does not regulate insulin-like peptide secretion or systemic insulin signaling. (A–C) Clusters of median neurosecretory cells from females carrying ilp5-lacZ (a marker for insulin-producing cells) at 10 days of adult adipocyte-specific knockdown of GFP control (A), InR (B) or Akt1 (C). ILP2 (green); βgal (red), encoded by the ilp5-lacZ transgene. ILP2 shown in grayscale in (A′-C′). Scale bar, 20 μm. (D) Quantification of total ILP2 fluorescence intensity in insulin-producing cells. Mean + s.e.m. Number of adult brains quantified are shown inside bars. No statistically significant differences; ordinary one-way ANOVA. (E–G) Nurse cells of stage 10 follicles from females as described in (A–C) but carrying the tGPH reporter instead of ilp5-lacZ. In the tGPH reporter, GFP is fused to a pleckstrin homology domain such that under conditions of high insulin signaling, PI3K activity recruits GFP to the plasma membrane (Britton et al., 2002). GFP shown in grayscale. Scale bar, 50 μm. (H) Quantification of average membrane-to-cytoplasmic GFP ratio. Mean + s.e.m. Number of follicles quantified are shown inside bars. No statistically significant differences; ordinary one-way ANOVA. See supplementary material Fig. S3 for an explanation of the quantification method.

GSK-3β, but not FOXO, is required in adult adipocytes for proper GSC maintenance

The transcription factor FOXO is a major downstream effector of insulin signaling (Webb and Brunet, 2014). Under active insulin signaling, Akt1 phosphorylates FOXO, leading to its cytoplasmic retention, while low insulin signaling and thus reduced Akt1 activity allow FOXO nuclear translocation and control of target genes (Webb and Brunet, 2014). Accordingly, InR or Akt1 knockdown in adult adipocytes results in robust nuclear translocation of FOXO (Fig. S1D). To test if the effects of reduced adipocyte insulin signaling on GSC numbers might be mediated by FOXO, we overexpressed foxo in adipocytes. Not surprisingly, foxo overexpression led to severely reduced adipocyte size (Fig. 5A and B, and Fig. S1). Unlike InR or Akt1 knockdown, however, foxo overexpression in adult adipocytes had no effect on GSC loss over time as compared to controls (Fig. 5C and Fig. S4A), indicating that the control of GSC number is independent of FOXO activity in adipocytes. Further, these results demonstrate that decreased GSC number upon reduction of adipocyte insulin signaling is not a secondary consequence of reduced adipocyte size.

Fig. 5.

GSK-3β, but not FOXO, function is required in adipocytes for proper maintenance of GSCs, whereas early cyst survival and vitellogenesis do not require either insulin signaling effector. (A and B) Adipocytes from females at 10 days of adult adipocyte-specific control (A; lacking UAS transgene) or foxo overexpression (B). FOXO immunoreactivity shows that adipocytes with high levels of FOXO expression have markedly reduced cell sizes (arrows). FOXO (green); E-cadherin (red), cell membranes; DAPI (blue), nuclei. Scale bar, 25 μm. (C and D) Average number of GSCs per germarium at 0 or 10 days of adult adipocyte-specific overexpression of foxo (C) or constitutively active sgg (sgg^S9A) (D) compared to Gal80^ts; Lsp2-Gal4-only control. Mean + s.e.m. *P < 0.05; two-way ANOVA with interaction. See supplementary material Fig. S4 for sample sizes and distribution. (E and F) Percentage of germaria with cleaved Dcp1-positive germline cysts (E) or ovarioles with degenerating vitellogenic follicles (F) in females overexpressing foxo or sgg^S9A in adult adipocytes as described in (A and B). Mean + s.e.m. Total number of germaria (E) or ovarioles (F) quantified are shown inside bars. *P < 0.05; ordinary one-way ANOVA.

We next considered the potential roles of other downstream effectors of insulin signaling in adipocytes in the control of GSC maintenance. Although the TOR kinase is a well-known downstream effector of InR/Akt1 signaling in many contexts (Manning and Toker, 2017), we previously showed that reducing TOR signaling specifically in adult adipocytes has no effect on GSC numbers (Armstrong et al., 2014). Similar to FOXO, GSK-3β (encoded by shaggy, sgg, in Drosophila) is negatively regulated by Akt1 and regulates cell survival, growth, proliferation and differentiation (Manning and Toker, 2017). In the Drosophila ovary, sgg controls the number and differentiation of follicle cells (Song and Xie, 2003) and mediates the shift in mitochondrial metabolism that is required to establish and maintain oocyte quiescence prior to fertilization (Sieber et al., 2016). As was the case for InR or Akt1 knockdown (see Fig. 2A), overexpression of constitutively active sgg^S9A in adult adipocytes led to significantly faster GSC loss compared to the 3.1Lsp2^ts driver alone control (Fig. 5D and Fig. S4B). Taken together, these data suggest that the serine/threonine kinase GSK-3β, not the transcriptional factor FOXO, is the downstream effector of insulin signaling in adipocytes that controls GSC maintenance, although its relevant phosphorylation targets remain to be identified.

FOXO and GSK-3β do not act in adipocytes to control early germ cell survival or vitellogenesis

We next examined if early germ cell survival, vitellogenesis, and GSC maintenance involve common or distinct downstream branches of insulin signaling in adipocytes. We found that females overexpressing foxo or sgg^S9A in adult adipocytes showed no increase in germline cyst death or dying vitellogenic follicles compared to 3.1Lsp2^ts driver alone control females (Fig. 5E and F; compare to Fig. 3D and H). Thus, the control of early and late germline survival by adipocyte insulin signaling appears to work independently of FOXO or GSK-3β. These results are consistent with differences in Akt1 requirement in adipocytes for GSC maintenance and vitellogenesis (compare Figs. 2A and 3H) and indicate that insulin signaling in adipocytes acts through distinct effectors, and presumably separate downstream secreted molecules, to control different aspects of oogenesis.

DISCUSSION

Our previous studies revealed the importance of amino acid sensing and diet-regulated metabolic pathways operating in adipocytes in the control of the Drosophila female GSC lineage (Armstrong et al., 2014; Matsuoka et al., 2017). Yet, we are just beginning to unravel the complexities of how multiple diet-dependent pathways act within adipocytes to affect oogenesis. Our data in this study demonstrate that, independent of its cell autonomous role in regulating cell size through FOXO, insulin signaling within adult adipocytes controls GSC number, survival of early GSC progeny, and vitellogenic follicles through distinct downstream mechanisms. GSC maintenance is modulated by the serine/threonine kinase GSK-3β, whereas germline survival during early and late oogenesis involves as-yet-unidentified downstream effectors (Fig. 6). In addition to advancing our understanding of the adipocyte-ovary connection, this study underscores how the apparently seamless role of the insulin pathway in controlling diet-sensitive steps of oogenesis entails a range of cell autonomous and non-autonomous mechanisms involving multiple organs. This work illustrates the challenges of investigating the physiological control of stem cell lineages: the crosstalk between multiple organs; the involvement of many systemic signals and their respective downstream pathways; the potential for each organ to respond directly or indirectly to each of the systemic signals; and the variety of mechanisms employed by each signal to regulate any given stem cell lineage.

Fig. 6.

Model for how insulin signaling within adipocytes regulates the GSC lineage. Our previous published work showed that insulin signaling within the ovary (grey arrows) controls GSC maintenance (through the niche) and proliferation, follicle growth, and vitellogenesis (reviewed in Laws and Drummond-Barbosa, 2017). In this study, we show that insulin signaling within adult adipocytes controls multiple steps of oogenesis through distinct effectors: 1) GSC maintenance is modulated by the InR/Akt1/GSK-3β axis (red); 2) early germline cyst survival downstream of Akt1 (blue) and Akt1-independent progression through vitellogenesis (purple) are controlled by as-yet-unidentified downstream effectors of the insulin signaling pathway. As reported previously (DiAngelo and Birnbaum, 2009), the InR/Akt1/FOXO axis controls adipocyte growth (orange).

Our findings highlight the variety of mechanisms through which a given endocrine signaling pathway can regulate Drosophila oogenesis. Insulin signaling in the germline controls GSC proliferation through PI3K/FOXO and cyst growth through (PI3K/TOR) (Hsu et al., 2008; LaFever et al., 2010), whereas insulin signaling in cap cells regulates GSC maintenance through FOXO regulation of Notch signaling and E-cadherin levels (Hsu and Drummond-Barbosa, 2009, 2011; Yang et al., 2013). This study adds to the complexity of insulin signaling roles by showing that in adipocytes an Akt1/GSK-3β-dependent mechanism controls GSC numbers, whereas Akt1-dependent or –independent regulation of germline cyst survival and vitellogenesis, respectively, are both GSK-3β-independent. Moreover, these oogenesis processes are independent of FOXO in adipocytes (this study), which instead controls adipocyte size/fat storage and lifespan (DiAngelo and Birnbaum, 2009; Giannakou et al., 2004; Hwangbo et al., 2004). Other systemic signals likely have equally complex mechanisms for regulating oogenesis. For example, ecdysone signaling cell autonomously controls GSC maintenance and proliferation through the downstream transcriptional factor E74 (Ables and Drummond-Barbosa, 2010), whereas germline cyst differentiation requires the action of ecdysone in neighboring somatic escort cells through the co-activator Taiman and E75 instead (Konig et al., 2011). We speculate that the complex web of nutrient-sensing physiological mechanisms regulating oogenesis facilitates extremely fine-tuned cellular responses to changes in nutritional input.

Control of GSCs requires the Akt1/GSK-3β branch of the insulin pathway in adipocytes, although downstream mediators remain to be identified. GSK-3β, a serine kinase, is a shared component of many signaling pathways, including the canonical Wnt, TOR, Notch, Hedgehog, TGFβ, and PI3K/Akt1 pathways (Patel and Woodgett, 2017). Consequently, the kinase activity of GSK-3β can be regulated by insulin signals, growth factors, amino acids, and other inputs (Sutherland, 2011). Not surprisingly, GSK-3β is involved in a wide range of biological processes including apoptosis, cell survival and growth, cell cycle progression, and stress responses (Maurer et al., 2014). Many predicted substrates of GSK-3β have been identified, and their phosphorylation can either activate or repress protein function (Linding et al., 2007). For example, GSK-3β inhibits eIF2B and β-catenin, which are positive regulators of translation and transcription, respectively (Maurer et al., 2014). Depending on the substrate, GSK-3β can be pro-apoptotic, acting on targets such as Tip60 and pro-apoptotic proteins, or can be pro-survival, acting on targets such as NF-κB and p100 (Maurer et al., 2014). GSK-3β has also been implicated in negatively regulating mouse hematopoietic stem cell (Huang et al., 2009) and embryonic stem cell self-renewal (Bone et al., 2009; Sato et al., 2004; Ying et al., 2008). More recent data suggest that fine-tuning GSK-3β inhibition regulates the balance between self-renewal and differentiation of rat embryonic stem cells by modulating the amount of stable β-catenin (Chen et al., 2013), although these studies reveal cell autonomous roles of GSK-3β in stem cell regulation. Future studies should identify downstream substrates of GSK-3β that non-cell autonomously mediate the effects of adipocyte insulin signaling on GSC number, and also address whether or not other systemic inputs cross-talk with the insulin pathway through GSK-3β.

Interestingly, the control of germline cyst survival and vitellogenesis by insulin signaling in adipocytes is independent of both FOXO and GSK-3β. InR/Akt1 signaling in adipocytes controls germline cyst survival, indicating that other Akt1 targets (Manning and Cantley, 2007) are required instead. By contrast, the survival of vitellogenic follicles does not require Akt1 function in adipocytes, implying yet distinct insulin signaling effectors. In addition to the PI3K/Akt1 branch, insulin receptor activation signals through the Ras/MAPK downstream pathway (Boucher et al., 2014), and several studies provide evidence that Akt1-independent insulin signaling controls endothelial cell proliferation in mice (Qiao et al., 2004), GLUT4 vesicle fusion at the plasma membrane (Gonzalez and McGraw, 2006), mammalian adipocyte lipolysis (Choi et al., 2010), and proliferation and survival of mouse-derived colon cancer cells (Teng et al., 2016). Recently, it was shown that Ras/MAPK signaling in the Drosophila fat body contributes to the extended lifespan observed in insulin pathway mutants (Slack et al., 2015). It will be important to investigate if Ras/MAPK might be required downstream of adipocyte insulin signaling to control vitellogenesis.

Identifying the signaling pathways and specific effectors downstream of insulin receptor activation in adipocytes that correspond to distinct stages of oogenesis is a key initial step towards a better understanding of the molecular mechanisms involved in adipocyte-to-ovary crosstalk. Also critical will be to identify the actual systemic signals produced by adipocytes downstream of the respective insulin pathway branches (e.g. through changes in their expression, secretion, or transport), and to determine their mechanisms of action. Such adipocyte-derived circulating factors might signal directly to specific ovarian cell types or have more indirect mechanisms through intermediate organs/tissues/cell types. Together with our previous reports showing that adipocyte amino acid sensing and metabolic pathways influence the GSC lineage (Armstrong et al., 2014; Matsuoka et al., 2017), and studies of adipose tissue inter-organ communication (Droujinine and Perrimon, 2016), our data add to the growing body of evidence that adipocytes serve as a major integrator of whole-body nutritional status.

HIGHLIGHTS.

• Insulin signaling in adipocytes remotely controls the GSC lineage in Drosophila

• Adipocyte insulin signaling effects on the GSC lineage are independent of FOXO

• GSC maintenance is modulated by the insulin receptor/Akt1/GSK-3β axis in adipocytes

• Early germline cyst survival is controlled by adipocyte Akt1 but not GSK-3β

• An Akt1-independent branch of adipocyte insulin signaling controls vitellogenesis

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at …