Excess dietary sugar impairs Drosophila adult stem cells via elevated reactive oxygen species-induced JNK signaling Free

Stem cells play a pivotal role in maintaining tissue homeostasis, replenishing cells lost as a result of aging, stress, or normal turnover. Their function is intricately linked to both intrinsic factors and external influences, such as nutrition and physiological status (Ables et al., 2012; Mihaylova et al., 2014). In humans, excess sugar consumption, often associated with a high-sugar diet (HSD), can lead to obesity and metabolic dysregulation, and can cause metabolic diseases, such as type 2 diabetes (Alexander Bentley et al., 2020; Chichger et al., 2016; Lean and Te Morenga, 2016). However, the impact of excess sugar and dysregulated metabolism on stem cell function remains unclear.

To address this question, we turned to Drosophila as an ideal model for several compelling reasons. First, established models of diabetes already exist in Drosophila, whereby 3 weeks of consumption of a 1 M sucrose-containing HSD induces metabolic syndrome, peripheral insulin resistance (a hallmark of type 2 diabetes), and organ dysfunction reminiscent of kidney and heart failure (Na et al., 2013; Rani et al., 2020). Second, the conservation of key signaling and metabolic pathways between flies and humans (Bharucha, 2009; Pandey and Nichols, 2011) enables us to draw meaningful parallels. Third, Drosophila offers the advantage of powerful genetic tools that facilitate in-depth studies of development and disease. Lastly, the Drosophila ovary and intestine harbor well-characterized ovarian germline stem cells (GSCs) and intestinal stem cells (ISCs) that respond rapidly to dietary changes (Drummond-Barbosa and Spradling, 2001; Lian et al., 2018; Micchelli and Perrimon, 2006).

In Drosophila, each female has two ovaries, each consisting of 16-20 ovarioles (Fig. 2A), the functional units for egg production (Spradling, 1993). Within the ovariole, the germarium, situated at the anterior end, houses two or three GSCs (Fig. 2A′). These GSCs are in direct contact with cap cells (the major component of the GSC niche) and contain a fusome located adjacent to the GSC–cap cell interface. The immediate progeny of GSCs, cystoblasts, undergo four incomplete divisions to form a 16-cell cyst, with germ cells interconnected by a branched fusome. These 16-cell cysts are subsequently encased by a single layer of follicle cells, ultimately developing into mature eggs.

The Drosophila adult midgut, a counterpart of the mammalian small intestine (Apidianakis and Rahme, 2011; Jiang and Edgar, 2011), consists of an epithelial monolayer and is divided into three main regions: the anterior midgut (AMG), the middle midgut (MMG) and the posterior midgut (PMG) (Fig. 3A). ISCs located near the basement membrane show a relative high proliferation rate in the PMG (Nászai et al., 2015) (Fig. 3A′). These ISCs give rise to both enteroblasts (EBs) and pre-enteroendocrine cells (preEEs). EBs subsequently differentiate into polyploid enterocytes (ECs), which comprise the majority of the midgut cells and are responsible for secreting digestive enzymes and absorbing nutrients. Meanwhile, preEEs differentiate into enteroendocrine cells (EEs) (Zeng and Hou, 2015), which release gut hormones that regulate gut motility and function in response to external stimuli and bacteria.

The insulin/insulin-like growth factor (IGF) signaling pathway, conserved in both flies and humans, regulates various biological processes (Barbieri et al., 2003; Brogiolo et al., 2001), including glucose uptake, metabolism, and cell division/growth. In Drosophila, insulin-like peptides (Dilps) activate the insulin receptor (encoded by InR), initiating downstream signaling through the recruitment and phosphorylation of Akt at the cell membrane. Drosophila has eight Dilps, but has only one gene for each downstream component of the insulin/IGF signaling pathway. Among the Dilps (called Dilp1-8; Ilp1-8), Dilp2, Dilp3 and Dilp5 are expressed in insulin-producing adult neuronal endocrine cells (Broughton et al., 2005), equivalent to vertebrate pancreatic β cells (Wang et al., 2007).

In this study, we show that excess dietary sugar impairs stem cell function, occurring before the onset of insulin resistance. Notably, tumorous GSCs or ISCs are more resilient to high sugar levels and do not respond to HSD in the same way as normal stem cells. This underscores the importance of maintaining a balanced, nutritious diet for metabolic health and tissue function, and highlights its impact on stem cells.

To distinguish the effects of obesity and type 2 diabetes induced by excessive dietary sugar on stem cells, we aimed to establish a Drosophila model of obesity without insulin resistance. We accomplished this by investigating the impact of sucrose diets at concentrations of 0.15 M (control diet, CD), 0.4 M, 0.7 M and 1 M on the lifespan of mated flies. Female flies fed diets containing 0.4 M sucrose exhibited similar survival rates to those on the CD, whereas those on 0.7 M or 1 M sucrose diets had reduced lifespans compared those on the CD and the 0.4 M sucrose diet (Fig. 1A), likely as a result of dehydration (van Dam et al., 2020). Notably, male flies experienced reduced lifespans only when fed 1 M sucrose diets (Fig. S1). As the survival rates of flies on the CD and the various sucrose diets were comparable during the first week (Fig. 1A′), we examined the physiological changes induced by excess dietary sugar in 1-week-old mated female flies, which were relatively healthy compared with older flies.

Given the association between increased sugar intake and obesity, we examined fat storage in adipocytes resided in the fat body using BODIPY (Fig. 1B,B′), a neutral lipid dye. Our results showed a dose-dependent increase in adipocyte size and lipid droplet accumulation in flies exposed to excess dietary sugar, indicative of obesity. Additionally, we examined the expressions of dilp2 (Ilp2), dilp3 (Ilp3) and dilp5 (Ilp5) in the brain (Fig. 1C). dilp3 and dilp5 transcripts were increased in flies fed diets containing 0.7 M and 1 M sucrose compared with controls, whereas dilp2, the most abundant Dilp gene in insulin-producing cells (Broughton et al., 2005), was slightly reduced. This result suggests that dilp3 and dilp5 expression may be sensitive to sugar levels. Furthermore, the expression of InR, which acts in a feedback loop when insulin signaling is low (Puig and Tjian, 2005), was unaffected by excess dietary sugar (Fig. 1C). These results suggest that insulin signaling in head tissues is not reduced following excess sugar consumption.

As the 1 M sucrose diet had the strongest effects on lifespan and lipid accumulation, we decided to use this as the HSD to feed flies for 1 week for subsequent experiments. Consistently, HSD-fed flies exhibited increased levels of TAGs (triglycerides, the main form of stored fat) in the fat body and whole-body extracts compared with CD-fed flies (Fig. 1D). However, glucose and trehalose levels (α-glucose disaccharide, the major form of sugar in circulation; Matsushita and Nishimura, 2020) in the hemolymph of HSD-fed flies did not increase (Fig. 1E).

To assess insulin resistance in these HSD-fed flies, we examined expression of Akt phosphorylation at Ser505 (equivalent to Ser473 in mammals) in response to exogenous insulin (1 µM, 15 min exposure) (Musselman et al., 2011; Sarbassov et al., 2005). Surprisingly, the fat bodies of flies fed the HSD for 1 week showed a greater response to exogenous insulin compared with that of CD-fed flies (Fig. 1F), indicating an increase of insulin sensitivity. We confirmed this result with immunostaining (Fig. 1G). In CD-fed flies, endogenous expression of phospho-Akt (pAkt) in adipocytes was minimal, but significantly increased upon insulin treatment. In contrast, pAkt signals were already detectable in the adipocytes of HSD-fed flies before insulin treatment and was significantly enriched at the adipocyte membrane after insulin treatment, indicating the high responsiveness of HSD-fed adipocytes to insulin. The increase in insulin sensitivity was still present in flies fed the HSD for 2 weeks, but no increase in insulin sensitivity was observed in flies fed the HSD for 3 weeks (Fig. S2), suggesting that insulin resistance may be imminent. Taken together, our results thus far indicate that mated young female flies fed an HSD for 1 week develop obesity without exhibiting insulin resistance.

The Drosophila ovary responds rapidly to diet and contains well-characterized GSCs (Drummond-Barbosa and Spradling, 2001) (Fig. 2A,A′). We first examined the impact of HSD on egg production. Flies laid an average of 37-51 eggs daily on the CD but only 8-11 eggs daily on the HSD (Fig. 2B). Ovaries from 1-week HSD-fed flies were smaller and contained fewer late-stage egg chambers, including stage 10 (Fig. 2B′), compared with CD-fed counterparts. Interestingly, when flies were switched to the CD after 1 week on HSD, egg production rates and ovarian size returned to control levels (Fig. 2B,B″), indicating that the HSD disrupts oogenesis reversibly. The HSD also increased lipid droplets in germ cells, including GSCs (dashed outlines) (Fig. 2C,C′). However, total TAG levels in ovaries of HSD-fed flies were lower than those of controls (Fig. 2D), likely because of the reduced number of late-stage, TAG-enriched egg chambers (Sieber and Spradling, 2015). This suggests an imbalance in lipid metabolism caused by the HSD. Notably, like the fat body, the ovaries of HSD-fed flies remained responsive to exogenous insulin (Fig. 2E,F). These results indicate that 1 week of excess dietary sugar impairs lipid metabolism and ovarian function, but the tissue retains its responsiveness to insulin.

The gut is expected to be the primary target of HSD feeding because sucrose, composed of one glucose and one fructose molecule, is hydrolyzed and absorbed by the intestinal epithelium before entering different cells through the bloodstream (Goodman, 2010). The Drosophila midgut (counterpart of the mammalian small intestine) contains the well-characterized ISC lineage (Fig. 3A,A′). One week of HSD feeding did not affect gut length (Fig. 3B,B′), but it reduced the thickness of the PMG (Fig. 3B″). In a previous report (Pereira et al., 2018), ∼7% of HSD-fed flies exhibited a leaky gut phenotype. However, a smurf assay revealed that the non-absorbable blue dye was restricted to the guts of flies fed with either CD or HSD for at least 2 weeks (Fig. 3C), indicating the maintained integrity of the gut epithelium. Like the fat body and ovaries, gut cells accumulated lipid droplets when flies were fed with an HSD (Fig. 3D), but these cells retained the ability to respond to exogenous insulin (Fig. 3E). These results suggest that after 1 week of HSD exposure the intestine shows imbalanced lipid metabolism but preserved insulin response.

Next, we examined the impact of HSD feeding on stem cell function. We assessed GSC proliferative capacity by labeling proliferating GSCs with the thymidine analog 5-ethynyl-2′-deoxyuridine (EdU). In the HSD group, we observed a 2.2-fold reduction in the frequency of EdU-positive GSCs [2.7±1.4 (s.d.) (n=962 GSCs, from three individual experiments)] compared with CD-fed flies [5.9±1.2 EdU-positive GSCs (n=950 GSCs from three independent experiments); P<0.05] (Fig. 4A,A′), with no evidence of GSC loss in HSD-fed flies (Fig. 4A″). Similar phenomena were observed in another laboratory wild-type strain (w¹¹¹⁸), albeit with a slight reduction in GSCs, and in male yw (Fig. S3). To test whether the reduction in GSC proliferation could be attributed to excess sugar or excess calories in the HSD, we provided flies with isocaloric diets in which the additional calories were supplied by fat (high-fat diet, HFD) or protein (high-protein diet, HPD) (Table S1). Although food intake was reduced in the flies fed with HSD, HFD or HPD (Fig. S4A), these high calorie diets all induced varying degrees of obesity (Fig. 4B,B′) with reduced fertility (Fig. S4B). However, GSC proliferation (Fig. 4C) and number (Fig. S4C) remained unaffected by HPD or HFD. These results suggest that reduced GSC proliferation is closely associated with the HSD, either by its own effect or through its induced obesity.

We further examined the effects of HSD on the ISC lineage by analyzing the number of ISCs and their progeny according to specific cell markers with nuclear sizes (Fig. 4D,D′). esg-GAL4 driving UAS-mCD8gfp (esg> mCD8gfp) labeled ISCs and progenitors (EBs and preEEs), whereas Pros labeled preEEs and EEs. Thus, ISCs/EBs were GFP positive (+) but Pros-negative (−); PreEEs were GFP⁺ and Pros⁺; EEs were GFP⁻ and Pros⁺; and polyploid ECs were GFP⁻ and Pros⁻ with relatively large nuclei. The ovaries of esg>mCD8gfp and wild-type yw flies responded similarly to HSD (Fig. S5), indicating that the intestines of esg>mCD8gfp flies can represent the intestinal response to HSD in wild-type flies. We did not find any difference in cell densities in the PMG between HSD- and CD-fed flies (Fig. 4E). Compared with CD-fed flies, the percentages of preEEs, EEs and ECs were not changed (Fig. 4F,G), whereas ISCs/EBs were reduced by about 30% after HSD feeding (CD, 20.4±2%, n=10 guts; HSD, 13.5±5%, n=10 guts; P<0.01) (Fig. 4G). Interestingly, a population of cells with the nuclear size of ECs but expressing esg>mCD8gfp was increased by 2.3-fold in HSD-fed flies; these cells were termed ‘mis-differentiated ECs’ (CD, 6.7±3%, n=10 guts; HSD, 15.1±4%, n=10 guts; P<0.001) (Fig. 4D″,H), as they appeared to indicate a defect in ISC-to-EC differentiation. Similar phenomena were observed in the AMG and MMG (Fig. S6). We next examined ISC numbers using Delta (Dl)-GAL4 to drive UAS-mCD8gfp and quantified the number of ISCs per unit area (Fig. 4I,I′). HSD-fed flies had a 40% reduction in ISC number compared with CD-fed flies (CD, 13±2.9, n=10 guts; HSD, 7.9±1.8, n=11 guts; P<0.001), indicating that HSD disrupts ISC maintenance. To determine whether HSD also reduces ISC division (as it does in GSCs), we labeled replicating cells with EdU, and counted EdU⁺ cells in the PMG (Fig. 4J,J′). Both the numbers of proliferating ISCs (with small nuclei) and of endoreplicating ECs (with large nuclei) were unaffected by HSD feeding, consistent with a previous report (Zhang et al., 2017). Notably, an HFD of similar caloric value to the HSD increased ISC proliferation but did not affect ISC number or ISC-to-EC differentiation (von Frieling et al., 2020). Our findings collectively indicate that HSD has a specific effect on stem cells.

Previous studies have shown that culturing cells in high-glucose medium increases the levels of reactive oxygen species (ROS) (Cheng et al., 2016; Ha and Lee, 2000). These increased ROS levels activate stress-responsive c-Jun N-terminal kinase (JNK) signaling (Shen and Liu, 2006), which controls various cellular processes (Pinal et al., 2019; Semba et al., 2020). We therefore detected ROS levels in GSCs from HSD-fed flies using dihydroethidium (DHE), a superoxide fluorescent probe (Robinson et al., 2006). Germaria (including GSCs) in CD-fed flies had low DHE signals, whereas strong DHE signals were observed in the germaria of HSD-fed flies, or of CD-fed flies treated with paraquat (PQ), a superoxide generator (Castello et al., 2007). Additionally, germaria (including GSCs) in HSD-fed flies exhibited stronger JNK signaling compared with those on the CD, as indicated by the expression of pJNK (Fig. 5B,B′) and puc-lacZ, a JNK signaling reporter (Wang et al., 2019), (Fig. S7). To determine whether germ cells from HSD-fed flies had an increased tendency to take up sugar, we performed a glucose uptake assay using a non-metabolized fluorescent glucose analog, 2-NBDG (Yoshioka et al., 1996). Germaria from HSD-fed flies showed higher 2-NBDG signals than those from CD-fed flies (Fig. 5C), including GSCs (Fig. 5C′). These findings were consistent with biochemical assays confirming higher endogenous glucose levels in the ovaries of HSD-fed flies compared with CD-fed flies (Fig. 5D). Furthermore, elevated ROS levels, JNK signaling, and glucose uptake were all observed in intestinal cells under the HSD (Fig. 5E-G).

In order to examine whether excess dietary sugar was responsible for the increased ROS levels and decreased GSC proliferation, we decreased dietary sugar by switching flies from HSD to CD. When flies were fed with the HSD for 1 week followed by CD for another week, the HSD-induced obesity and increased ROS levels in GSCs reverted to the levels seen in CD-fed flies (Fig. S8A,B). In addition, GSC proliferation, which was decreased by HSD, also tended to increase towards levels seen in GSCs under CD (Fig. S8C). These results collectively indicate that stem cells exhibit increased sugar uptake when flies are fed an HSD, and the excess dietary sugar correlates with activation of ROS-induced JNK signaling (hereafter ROS/JNK signaling).

We postulated that activation of ROS/JNK signaling might underlie the disruption of GSC and ISC function caused by HSD. To test this, we explored whether removing ROS or inhibiting JNK signaling could prevent the HSD-induced effects on GSC proliferation and ISC maintenance. We overexpressed Superoxide dismutase 1 (SOD1), an antioxidant enzyme (Ighodaro and Akinloye, 2018), in the germline. In a control experiment, PQ treatment increased ROS in the germanium of CD-fed flies, and overexpression of SOD1 decreased this PQ-induced ROS level (Fig. 6A), demonstrating the antioxidant capacity of SOD1. Compared with control GSCs under CD, SOD1-expressing GSCs in CD-fed files showed a lower proliferation rate, although this was not a significant difference (Fig. 6B). However, in contrast to control GSCs, HSD exposure did not decrease the proliferation of SOD1-expresssing GSCs (Fig. 6B), suggesting that increased ROS levels contribute to decreased GSC proliferation. Next, we disrupted JNK signaling in the germline by decreasing the expression of JNK (encoded by basket, bsk) (Herrera and Bach, 2021), or the expression of JNK kinase (encoded by hemipteruous, hep) (Herrera and Bach, 2021), and examined GSC proliferation under HSD conditions. We found that the HSD-induced decrease in GSC proliferation did not occur in flies with bsk or hep knockdown GSCs (Fig. 6C). These results indicate that the HSD reduces GSC proliferation through ROS/JNK signaling.

We also overexpressed SOD1 or Catalase (Cat), another antioxidant enzyme (Ighodaro and Akinloye, 2018), in ISCs/progenitors (EBs/preEEs) using esg-GAL4 (Fig. 6D). Overexpression of SOD1 or Cat reduced ISCs/EBs in CD-fed flies (Fig. 6D′), consistent with a previous report showing that ROS is essential for ISC self-renewal (Morris and Jasper, 2021). However, in contrast to control GSCs, HSD feeding did not cause a reduction in ISCs/EBs when SOD or Cat was overexpressed (Fig. 6D′), nor was an increase in mis-differentiated ECs observed (Fig. S9A). Blocking JNK signaling in ISCs/EBs by knocking down bsk or hep also suppressed the HSD-induced reduction in ISC/EB numbers (Fig. 6E,E′). Although decreasing hep expression did not increase mis-differentiated ECs under HSD, a slight increase of mis-differentiated ECs was observed in esg>bsk^RNAi PMG of HSD-fed flies (Fig. S9B). This result is likely due to the fact that very few mis-differentiated ECs are present in CD-fed flies on this genetic background. Nevertheless, our results suggest that an HSD induces ROS/JNK activation to perturb GSC and ISC function.

Cancer cells are known to favor sugar utilization for their rapid proliferation, relying primarily on glycolysis to produce lactate for energy (Liberti and Locasale, 2016). To test the effects of HSD on cancer stem cell proliferation, we induced tumorous GSCs by mutating bag of marbles (bam), which promotes GSC differentiation (Fig. 7A) (McKearin and Spradling, 1990). Interestingly, HSD did not affect the division of tumorous GSCs (Fig. 7A′) or the germarial size (reflecting the number of germ cells carried) (Fig. 7A″). Germline-specific knockdown of bam also led to no decrease in GSC proliferation when exposed to HSD conditions (Fig. 7B). Although HSD feeding caused obesity in bam mutant flies (Fig. S10A-A″), the tumorous GSCs did not show increased lipid droplets (Fig. 7C,C′) or higher ROS levels (Fig. 7D-E). [Note that bam mutation does not affect food intake throughout the whole adult life (Wang et al., 2022)]. These results suggest that bam mutant tumorous GSCs are resistant to the effects of HSD.

Autophagy plays a crucial role in promoting stemness in some cancer stem cells (White, 2015), and bam mutant GSCs exhibit enhanced autophagy that promotes niche occupancy (Zhao et al., 2018). This raises the possibility that autophagosomes may remove mitochondria damaged by HSD consumption, thereby reducing cellular ROS levels. To test this, we examined the mitochondrial membrane potential in isolated bam mutant GSCs using TMRE, a fluorescent probe (Perry et al., 2011). GSCs from CD-fed bam mutant flies treated with FCCP, a potent mitochondrial oxidative phosphorylation uncoupler (To et al., 2010), showed very low TMRE signals compared with those without FCCP treatment (Fig. 7F). In contrast, under HSD conditions, bam mutant GSCs showed only a slight decrease in TMRE signal compared with GSCs from CD-fed bam mutant flies (Fig. 7F), indicating that the mitochondria of bam mutant GSCs remain active when exposed to HSD. We also treated bam mutant flies with chloroquine (CQ), an autophagy inhibitor (Manic et al., 2014), and examined GSC proliferation in both CD- and HSD-fed flies. GSC proliferation was decreased in CD-fed flies after CQ treatment compared with CD-fed flies without CQ treatment (Fig. 7G,G′; see also Fig. 7A′), consistent with a previous study showing that autophagy promotes GSC division (Zhao et al., 2018). However, CQ treatment did not decrease GSC proliferation in HSD-fed flies compared with CD-fed flies (Fig. 7G′), indicating that autophagy is not involved in the GSC response to HSD.

Low ROS levels observed in tumorous GSCs might result from reduced oxidative phosphorylation, which generates ROS as a by-product (Nolfi-Donegan et al., 2020). To investigate this, we measured lactate levels in bam tumorous GSCs using a genetically encoded fluorescence resonance energy transfer (FRET) lactate biosensor (San Martín et al., 2013). The sensor consists of a bacterial lactate-binding protein, LldR, sandwiched between two fluorescent proteins, mTFP and Venus (Fig. 8A). In the presence of lactate, FRET from donor (mTFP) to acceptor (Venus) is low, resulting in higher mTFP signals. Tumorous GSCs in CD-fed flies showed elevated lactate levels compared with normal GSCs (Fig. 8A,A′), suggesting a preference for glycolysis. However, there was no further increase in lactate levels in tumorous GSCs under HSD, in comparison with the higher lactate levels shown by normal GSCs under HSD (Fig. 8A,A′), suggesting that they consume excess sugar by producing lactate. Consistent with this, under CD, bam mutant tumorous GSCs had higher glucose uptake compared with normal GSCs from both CD- and HSD-fed flies (Fig. 8B,B′). Notably, tumorous GSCs did not exhibit increased glucose uptake under HSD (Fig. 8B′), probably because their glucose uptake was already saturated under the CD.

Interestingly, tumorous ISCs induced by overexpression of a constitutively active form of Egfr (Egfr^A88T) (Proske et al., 2021), also exhibited increased glucose uptake under CD compared with normal ISCs from CD flies (Fig. S11A) with no further increase in glucose uptake under HSD. In addition, the size of ISC tumor (the occupancy of tumorous ISCs per gut area) was also not affected by HSD exposure (Fig. S11B). In contrast to tumorous GSCs, ISC tumors under CD showed higher pJNK expression than normal ISCs under HSD, whereas this high pJNK expression was not further increased, but decreased, under HSD (Fig. S11C). These results suggest that oxidative phosphorylation is promoted in tumorous ISCs under CD, thereby generating ROS to activate JNK signaling, whereas oxidative phosphorylation may be attenuated under HSD in the tumor context. Taken together, our results suggest that tumorous stem cells may require more energy or employ distinct metabolic strategies to withstand the deleterious effects of excess dietary sugar.

A balanced diet is crucial for proper stem cell function and tissue integrity. Excess dietary sugar has been linked to obesity and diabetes, which are associated with stem cell dysfunction function (Oestreich et al., 2020; Xu and Zuo, 2021), although the mechanisms are complex and unclear. In this study, we induced obesity in flies through exposure to an HSD without inducing insulin resistance, revealing high ISC loss and a low GSC division rate. This is specific to the HSD, and cannot be recapitulated by HFD or HPD, both of which contain similar caloric contents to the HSD and cause obesity in flies. A switch from HSD to a regular diet reduces stem cell glucose uptake and reverses most of the HSD-induced phenomena, suggesting a direct effect of HSD on stem cells (Fig. 8C). Under the HSD, normal adult stem cells exhibit increased glucose uptake, which may cause mitochondrial overload, leading to overproduction of acetyl-CoA and ROS via the TCA cycle and oxidative phosphorylation, respectively. The increased acetyl-coA can be used to synthesize fatty acids and thus be stored as lipids in the cytoplasm; ectopic lipid storage in non-adipose tissues can lead to cellular lipotoxicity (Ahmed et al., 2021). The increased ROS also activates JNK signaling, which negatively affects stem cell function. We do not know how the HSD activates insulin signaling in the GSCs, and which glucose transporter is used for glucose uptake in GSCs. Interestingly, germline depletion of InR rescued HSD-induced phenomena in the germline, including the increased glucose uptake and ROS levels and decreased GSC proliferation rate (Fig. S12), suggesting that insulin signaling may control glucose uptake in GSCs, and that excess dietary sugar may directly affect GSCs. Our study implies that excess dietary sugar can be directly harmful to stem cells, and that the stem cell dysfunction observed in HSD-induced obese and diabetic patients is likely due to prolonged exposure to high sugar levels.

In contrast, tumorous stem cells differ in their response to HSD, depending on their energy needs and metabolic preference. They exhibit increased glucose uptake even under a CD. Tumorous GSCs may have adopted glycolytic-lactate production as their main energy source, so there is no increased ROS, lipid accumulation, or decreased GSC proliferation. By contrast, tumorous ISCs may favor oxidative phosphorylation rather than glycolysis for energy production, thereby generating more ROS to enhance JNK signaling under the CD. However, this enhanced JNK signaling decreases under HSD, likely owing to some unknown feedback or systemic regulation. Nevertheless, our study potentially delineates the detrimental effects of excess sugar on stem cells and shows that tumor stem cells have a high tolerance to sugar, shedding light on the link between obesity, diabetes, and tumor growth.

In multiple studies, an HSD with 1 M sucrose induced insulin resistance in both Canton-S and w¹¹¹⁸ larvae and adult flies (Morris et al., 2012; Musselman et al., 2011). Additionally, adult w¹¹¹⁸ and Oregon R flies develop insulin resistance and display failures of cardiomyocytes and retinal tube after 3 weeks of an HSD (Na et al., 2013; Rani et al., 2020). In one study, w¹¹¹⁸ female virgins maintained on an HSD for 1 week displayed insulin resistance (Brookheart et al., 2017). However, in our study, HSD feeding of newly eclosed yw female flies co-cultured with males for 1 week did not develop insulin resistance. The lack of resistance to exogenous insulin was demonstrated by western blot and/or immunostaining of fat body, gut and ovarian tissues. We therefore speculate that prolonged HSD exposure may facilitate the development of insulin resistance. Furthermore, female virgin flies have relatively slow oogenesis with fewer mature egg chambers where TAGs are stored; it is not clear whether the development of insulin resistance in these female virgin flies under HSD is due to the lack of TAG-stored tissues to buffer circulating sugars. However, our HSD led to obesity in yw flies without signs of insulin resistance before week 3, suggesting that HSD-induced obesity does not always model diabetes comprehensively.

yw and w¹¹¹⁸ strains were used as the wild-type controls. The following fly strains were used in experiments: bam^Δ86 (McKearin and Ohlstein, 1995), bam¹ (McKearin and Spradling, 1990), UAS-bsk^RNAi [Bloomington Drosophila Stock Center (BDSC), 32977] (Wang et al., 2019), UAS-hep^RNAi (BDSC, 35210) (Houtz et al., 2017), UAS-bam^RNAi (BDSC, 33631) (Carbonell et al., 2017), UASp-dInr (BDSC, 35251) (Pletcher et al., 2019), UASp-Cat (Pan et al., 2007), UASp-SOD1 (Pan et al., 2007), UAS-egfr^A88T (BDSC, 9534; a gift from Dr Chen-Yuan Tseng, Institute of Molecular Biology, National Chung Hsing University, Taiwan), puc-lacZ^[A251] (BDSC, 11173) (Wang et al., 2019), nos-GAL4 (Rørth, 1998), Dl-GAL4 (Zeng et al., 2010) and esg-GAL4 (Loza-Coll et al., 2014). Other genetic elements are described in FlyBase (http://flybase.org).

The lactate sensor-containing plasmid was purchased from Addgene (#44238). The fragment containing the lactonic sensor was amplified using a pair of primers with XbaI and XhoI restriction enzyme sites (Table S2), and subcloned into the UASz vector (a gift from Dr Tzu-Yang Lin, Institute of Cellular and Organismic Biology, Academia Sinica, Taiwan). Transgenic lines were then generated according to standard procedures (Spradling and Rubin, 1982; Rubin and Spradling, 1982).

Flies were maintained at 25°C with standard medium, unless otherwise indicated. Control, HSD, HPD and HFD preparation was performed according to a previous study (Table S1) (Musselman et al., 2011) in which diets were modified from standard Bloomington Semi-Defined Food. For chloroquine treatment, 100 µl of ddH₂O (control) and 50 mM chloroquine (Sigma-Aldrich) was added onto the surface of the food and dried in a chemical hood before use. Newly eclosed flies were cultured on different diets as indicated; food was changed daily until dissection. For the lifespan assay, ∼30 pairs of yw flies were cultured in the bottle in triplicate at 25°C; flies were removed after 2 days. The bottles were kept at 25°C until eclosion. Newly eclosed flies were collected from the bottles for 2 days (0-2 days old); ten pairs of the flies were transferred in a plastic vial containing the indicated diet in ten replicates. Food was changed and dead flies were counted and removed every day; two independent experiments were performed. For the egg laying assay, newly eclosed flies (Day 0) were cultured in plastic bottles (for Fig. 3A) or vials (for Figs S4B and S5B). For the plastic bottles, five pairs of newly eclosed flies were cultured in plastic bottles in triplicate. The bottles were capped with a Petri dish filled with the indicated diet and placed upside down. The Petri dish with the indicated diet was changed daily, and the number of eggs on the dish was counted every 24 h. For the vials, ten females were cultured with five males; flies were transferred daily to the new vials containing the indicated diet, and a photograph was taken from the top of each old vials after the flies were transferred to the new vial. Eggs were counted using Adobe Photoshop software. Flies expressing RNAi, egfr^A88T or lactonic sensor were cultured at 18°C to suppress GAL4 activity during development; flies were switched to 29°C after eclosion to activate GAL4 activity. Flies expressing Cat or SOD1 in the gut were cultured at 25°C; flies expressing SOD1 in the germline were cultured at 18°C during development and switched to 25°C after eclosion. One-week-old female flies fed the indicated diets for 1 week were used for analysis unless otherwise indicated. The genotype of the flies used in each figure are listed in Table S3-1.

Total RNA from 35 heads of 7-day-old females was extracted using TRIzol (NovelGene). cDNA synthesis was performed using a Transcriptor First Strand cDNA Synthesis Kit (Roche), following the manufacturer's instructions. Quantitative real-time PCR was performed in both technical and biological triplicates using Lightcycle 480 probes master mix (Roche) on a Roche Light 480 machine. Primers and probes were acquired from Roche Universal ProbeLibrary Assay Design Center and are listed in Table S2.

Food intake assays were performed as previously described (Moraru et al., 2017), with modification. Ten pairs of newly eclosed flies were kept on CD, HSD, HFD or HPD in the plastic in triplicate vials until 1 week of age; food was changed daily. The same female flies were then transferred to an empty plastic vial with a cap from a 15 ml falcon tube containing the corresponding diet with 3.75% (w/v) Acid Blue (erioglaucine disodium salt; Sigma-Aldrich) for 24 h. After feeding, females were homogenized in 400 µl PBS following incubation on ice for 10 min. After homogenization, the supernatant was harvested by centrifugation (8000 g) for 20 min. After removing the cap containing the food, the dye excreted by the flies on the walls of the vials was collected by adding 3 ml of dH₂O followed by vortexing. The absorbance of the dye in the fly lysate and in the excretion was determined at 630 nm using a spectrophotometer (Spectra Max M5, Molecular Devices). Volumes of medium consumed were calculated from absorbance values by interpolation from standard curves of pure dyes. Extracts from flies fed medium without dye controlled for background absorbance.

Smurf assays were performed as previously described (Rera et al., 2012), with modification. Ten pairs of newly eclosed flies (day 0) were fed with CD or HSD mixed with 2.5% (w/v) blue dye (Sigma, MKBS2204V) for 1 week and 2 weeks; food was changed daily. After 1 or 2 weeks of feeding, female flies were anesthetized with CO₂, and images were captured with a ZEISS Axiocam ERc 5 s camera.

Total TAG assay was performed as previously described (Tennessen et al., 2014), with minor modifications. Adult female flies were rinsed with cold PBS in a 9-well glass plate to remove food attached outside of the animal bodies. The entire bodies, fat bodies or ovaries of groups of five females in triplicate were quickly homogenized in 100 µl 0.1% PBST on ice and heated at 70°C for 10 min to inactivate endogenous lipase and then centrifuged (21,130 g for 3 min). Ten microliters of the supernatant were used to measure protein content using Bio-Rad protein assay reagent (Richmond). The TAG in the supernatants was measured by absorption at 540 nm by SpectraMax M5 (Molecular Devices) after applying the Serum Triglyceride Determination Kit (Sigma-Aldrich, TR0100). Serially diluted samples of glycerol standard (Sigma-Aldrich, G7793) were treated simultaneously and used to quantify TAG levels in each sample.

The glucose and trehalose assay was performed as previously described (Tennessen et al., 2014), with minor modifications. For hemolymph collection, triplicate samples of 40 female flies in a group were pierced through the thorax using insect pins, and then placed in 0.5-ml microcentrifuge tubes with small holes in the bottom. The smaller tubes were placed in a 1.5-ml microcentrifuge tubes and centrifuged at 2300 g for 5 min to collect about 1 µl hemolymph. For ovarian lysate preparation, five pairs of ovaries were dissected and lysed with cold 100 µl trehalase buffer (TB; 5 mM Tris pH 6.6, 137 mM NaCl, 2.7 mM KCl); 10 µl of lysate was used for measurement of protein concentration and the remaining lysate was stored at −80°C until analysis. Heated (70°C) hemolymph and the desired amounts of ovary lysate were mixed with reagents of the Glucose (HK) Assay kit (Sigma-Aldrich, GAHK20), and absorbance was measured at 340 nm using a SpectraMax M5 (Molecular Devices). Trehalose was measured using the same reagent after digestion with trehalose (Sigma-Aldrich). The samples were diluted tenfold for the trehalose measurement because trehalose levels are higher than glucose levels. Glucose and trehalose standards were used to quantify the sugar levels in hemolymph and ovary lysates.

Immunostaining was performed as previously described (Chen et al., 2019; Micchelli, 2014; Tseng et al., 2014). In brief, ovaries and fat bodies were dissected in Grace's insect medium (GIM; Gibco) and fixed with 5.3% formaldehyde (FA) in GIM for 13 min at room temperature (RT). Then, the tissues were washed with 0.1% PBST (1× PBS with 0.1% Triton X-100). Testes were dissected in PBS on ice and fixed with 4% paraformaldehyde (PFA) in PBS for 20 min at RT, followed by three washes with 0.3% PBST. Testes were then treated with 0.3% sodium deoxycholate in 0.3% PBST for 30 min and washed three times with 0.3% PBST. Guts were dissected in PBS and fixed in a cold 4% PFA overnight (O/N) at 4°C without agitation. The next day, samples were washed with cold PBS for 8 h at 4°C. After washing, samples were incubated with secondary antibodies for 3 h at RT or O/N at 4°C. The following primary antibodies were used: mouse anti-Hts [Developmental Studies Hybridoma Bank (DSHB), 1B1, 1:50], mouse anti-Lamin (Lam) C (LC28.26) (DSHB, 1:25), mouse anti-FasIII (DSHB, 7G10, 1:200), rat anti-Vasa (DSHB, 1:20), mouse anti-Prospero (DHSB, MR1A, 1:100), chicken anti-β-gal (abcam, ab9361, 1:1000), chicken anti-GFP (abcam, ab13970, 1:500), rabbit anti-Phospho-Histone H3 (Ser10) (Merck Millipore, 06-570, 1:250), rabbit anti-Drosophila pAkt Ser505 (Cell Signaling Technology, 4054, 1:1000), rabbit anti-pJNK (Thr183/Tyr185) (Cell Signaling Technology, 9251, 1:200) and rabbit anti-Akt antibody (Cell Signaling Technology, 4016, 1:1000). After washing, samples were incubated with secondary antibodies for 3 h at RT or O/N at 4°C. Alexa Fluor 488-, 568- or 633-conjugated goat species-specific secondary antibodies (Table S4, 1:500-1000) were used. After washing, samples were stained with 0.5 µg/ml DAPI (Sigma-Aldrich) and mounted in 80% glycerol containing 20 µg/ml N-propyl gallate (Sigma-Aldrich). Images were taken using a Zeiss LSM 700 or LSM 900 confocal microscope.

EdU incorporation was performed with the Click-iT Edu imaging kit (Invitrogen). In brief, ovaries and guts were dissected in pre-warmed GIM and incubated with 10mM EdU in GIM for 30 min, 1 h or 2 h at RT prior to fixation and immunostaining as described above. Before staining with DAPI, the EdU detection reaction was performed according to the manufacturer's instructions.

Ovaries bearing germ cells expressing the lactate sensor were fixed and labeled with 1B1 and LamC antibodies, as described above. mTFP was excited at 438/24, and emission was divided by a splitter with bandpass filters at 480/20 (mTFP) and 535/15 nm (Venus).

Ovaries or guts were dissected in 1× PBS, teased apart, and incubated with 1 mM 2-NBDG (a fluorescent glucose analog; Selleck, S8914) for 45 min at 25°C. Ovaries were washed in cold 1× PBS twice (each for 5 min) and fixed with 4% formaldehyde in 1× PBS for 15 min at RT. After washing, ovaries were incubated with DAPI and mounted as described above. For glucose uptake analysis with tumorous GSCs or ISCs, 2-NBDG was substituted , which shows higher fluorescence signals than 2-NBDG.

The western blotting assay for insulin response was performed as previously described (Musselman et al., 2011). Seven-day-old CD- or HSD-fed flies were switched to plastic vials containing Kemwipes soaked with ddH₂0 for 6 h starvation (from 11:00 to 17:00), before dissection. Guts, ovaries and fat bodies were separately dissected from five female flies in pre-warmed GIM and then incubated with or without 1 μM human recombinant insulin peptide (Sigma-Aldrich) with vigorous shaking at 25°C for 15 min. Samples were rinsed with GIM, frozen in liquid nitrogen adding 50 μl of 2-mercaptoethanol (Amresco, 60-24-2) into 950 μl of 2× Laemmli Sample Buffer (Bio-Rad,161-0737). After centrifugation, the supernatants were harvested for western blotting. The same set of samples was loaded in duplicate on two gels, transferred to PVDF membranes; one membrane was probed with anti-pan Akt antibody and an internal control antibody, and one membrane was probed with anti-pAkt antibody and an internal control antibody.

Antibodies against pAkt included rabbit anti-Drosophila pAkt Ser 505 (Cell Signaling Technology, 4054, 1:1000) or rabbit anti-human pAkt Ser 473 (Cell Signaling Technology, 4060, 1:1000). The antibody for pan Akt (Cell Signaling Technology, 4691, 1:1000) was a rabbit anti-mouse Akt antibody. Rabbit anti-actin (abcam, ab8227, 1:2500), rabbit anti-human Histone H3 (abcam, ab1791, 1:5000), or mouse anti-human α-tubulin antibody (Merk Millipore, DM1A, 1:5000) was used as a loading control. Secondary antibodies included horseradish peroxidase-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch, 111-035-003, 1:5000) and horseradish peroxidase-conjugated goat anti-mouse IgG (Millipore, AP124P, 1:5000). The ratio of pAkt/internal control expression was normalized to the ratio of pan-Akt/internal control to obtain a pAkt/pan-Akt ratio for each sample. Insulin response was obtained by calculating the fold change of the pAkt/pan-Akt ratio for insulin treatment compared to without insulin treatment.

For the immunofluorescence insulin response assay, samples were taken from flies after insulin treatment. The samples were fixed and immunostaining was performed as described above.

DHE staining was performed as previously described (Rera et al., 2011; Vaccaro et al., 2020). Ovaries and guts were dissected in GIM and incubated with 30 μM DHE (Invitrogen) in GIM (ovaries, 20 min; guts, 7 min) in the dark at RT. Ovaries and guts were incubated with 50 μM PQ (Sigma-Aldrich) for 20 min to induce cellular ROS as a positive control. For ovaries, immunostaining was performed after washing and fixation as described above. For guts, staining with Hoechst 20098863 (Invitrogen, 1:1000) for 2 min at RT was performed, followed by fixation for 5 min with 4% PFA; images were then captured immediately.

To validate the ability of SOD1 to functionally remove ROS, 3-day-old nos>SOD1 flies cultured on a CD were starved for 6 h, then fed with 5% sucrose with or without 10 mM PQ (soaked on Kimwipes) for 48 h until dissection (food was changed daily).

Ovaries, fat bodies and guts were dissected and fixed as described above. Fat bodies were incubated in Rhodamine Phalloidin (Invitrogen, R415, 1:80) for 20 min at RT and washed with PBST for 20 min. Then, samples were incubated with BODIPY (Molecular Probes, D-3922, 1:500) for 20 min, washed with PBST, stained with antibodies and DAPI, and then stored in mounting solution at −20°C.

The bam mutant GSCs were isolated and subjected to flow cytometry analysis as previously described (Kao et al., 2015). In brief, vasa-gfpbam^Δ86/vasa-gfpbam¹ (vasa-gfp was used to detect germ cells) were grown at 25°C in standard media until eclosure. One- or two-day-old flies were collected and subsequently cultured in 0.1 M and 1.0 M sucrose diets for 7 days. Seven to ten pairs of ovaries were dissected from flies on each diet in pre-warmed GIM with 10% fetal bovine serum (FBS) (GIM–FBS). Ovaries were then dissociated in freshly prepared dissociation solution containing 0.45% Trypsin (Solution 10X, 9002077, Sigma-Aldrich) and 2.5 mg/ml collagenase (17018–029, Gibco) in GIM-FBS. Samples were dissociated on a rotator at 25°C for 25 min with vigorous shaking in a dark chamber and were gently vortexed every 5 min. Digested ovaries were filtered through a 40-µm nylon mesh and then centrifuged at 1000 g for 7 min to harvest the cell pellet. The pellets were re-suspended in 500 µl GIM–FBS containing 10 nM mitochondrial membrane potential probe TMRE (Thermo Fisher Scientific, T669) or 30 µM cytoplasmic superoxide probe DHE (Invitrogen, D11347) and 0.5 µg/ml of DAPI with vigorous shaking at RT for 10 min in a dark chamber. For a positive control in the membrane potential experiment, dissociated cells were co-treated with 10 nM TMRE and 10 µM FCCP (to depolarize the mitochondrial membrane, Sigma–Aldrich C2920) for 10 min under the same conditions as described above. For a positive control in the ROS experiment, dissociated cells were treated with 30 µM DHE and 100 µM PQ (to induce cellular ROS; Sigma–Aldrich, #3752782) for 10 min under the same conditions as described above. After incubation, cells were kept on ice and subjected to flow cytometry within 30 min.

The stained cells were detected using an Attune NxT acoustic focusing cytometer (Thermo Fisher Scientific) to measure GSCs carrying vasa-GFP (488/530 ex/em), DAPI-labeled dead cells (405/440 ex/em), DHE (488/590 ex/em) and TMRE (561/585 ex/em). TMRE and DHE intensities were measured from GFP-positive and DAPI-negative GSCs. Signals from at least 10,000 GSCs were measured and averaged for one replicate; three replicates were performed for each measurement.

Female GSCs were defined by the position of the fusome (labeled by 1B1), which is adjacent to cap cells (labeled by LamC). For female GSC proliferation and maintenance assays, at least three independent experiments were performed, and at least 100 GSCs were counted in each experiment. Male GSCs were recognized by their direct contact with hub cells (labeled with FasIII). For male GSC proliferation and maintenance assays, ten testes from each diet were dissected and analyzed; two independent experiments were performed. Stage 10 egg chambers were defined as the oocyte occupying 50% of the egg chamber volume.

To quantify the average intensity of DHE, pJNK, 2-NBDG and Glucose Uptake Probe-Green and the number of LDs, the z-stack section with the largest GSC area was chosen for quantification.

For FRET-based measurement of lactate levels, the FRET ratio of mTFP/Venus were obtained from integration of the ratio signal over the entire germ cells using ImageJ software (version 1.8, National Institutes of Health).

To image the specimens for these analyses, 10-15 germaria were randomly selected from a slide on which five ovaries (approximately 80-100 germaria) randomly selected from ten pairs of dissected ovaries were mounted; at least two independent experiments were performed.

The PMG was located between the hindgut and MMG (identified by the presence of Malpighian tubules to the first narrow region encountered). For gut length and thickness, images of the gut were taken after fixation with a Zeiss SV11 Apo stereoimager and analyzed using ImageJ. A freehand line was drawn through the entire gut to measure gut length, and PMG thickness was measured by drawing three straight lines vertically through the anterior, middle and posterior regions of the PMG.

Four or five images were taken of the PMG, each consisting of 16-20 z-stack sections covering half of the thickness of the gut. For ISC lineage analysis, numbers of each cell type were counted according to markers from two randomly selected areas (100 µm×100 µm) in each gut image, which was merged from 16-20 z-stack sections covering half of the gut thickness. For mitotic cell analysis, PHH3- or EdU-positive cells were counted in the entire z-stack.

To quantify DHE, pJNK, 2-NBDG and glucose uptake probe signals in the PMG, all z-stack sections of a gut image were merged to generate a 2D PMG image, and five square areas (of 1000 μm²) were randomly selected for average intensity measurement using ZEN software (ZEISS).

To measure the occupancy of tumorous ISCs in the gut, all the z-sections of each gut image were merged to create a 2D image. The gut area in each gut image was measured. The threshold of each 2D image was set to define tumorous ISCs (esg>mCD8gfp positive), using ImageJ (version 1.8, National Institutes of Health). The total area of tumorous ISCs was measured and divided by the gut area to obtain the occupancy of tumorous ISCs per gut image. For gut image analysis, seven to ten guts were analyzed for each genotype with the indicated diet or treatment; two to three independent experiments for each analysis were performed.

To measure the size of lipid droplets (LDs), number of LDs, LD content and adipocyte size, the biggest area of the adipocyte was used for measuring or counting using ZEN software. Several adipocytes were measured in five fat bodies, which were randomly chosen from ten dissected fat bodies. Two or three independent experiments for each measurement were performed.

Images were processed using ZEN, Photoshop and ImageJ. For FRET signal images of germ cells in the germarium, the background (the area without FRET signals, e.g. nucleus) were removed using remove.bg (https://www.remove.bg/zh), and then pasted onto the same germaria with 1B1 and LamC labels using Photoshop software.

Graphs and statistical analyses were performed using either Microsoft Excel or GraphPad Prism 10. Sample and replicative numbers and statistical analysis performed are summarized in Table S3-2.

We thank C. Y. Tseng, J. C. Hsu, S. E. Bickel, G. C. Chen, T. Xie, Bloomington stock center, VDRC Stock Center and DSHB for Drosophila stocks and antibodies. We thank the Taiwan fly core for help with the purchase of fly lines and antibodies from the above mentioned stock centers and DSHB. We thank T. J. Baranski (Division of Endocrinology, Metabolism and Lipid research, Washington University School of Medicine, MO, USA) and L. P. Musselman (Binghamton University state University of New York, NY, USA) for helpful discussion regarding insulin response assay. We thank W. C. Chu from the Image Core at the Institute of Cell and Organismic Biology for assisting with image analysis, T. Kaur for help with data analysis, Q. Y. Tang for assistance with microinjection, and M. Calkins for English editing.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201772.reviewer-comments.pdf.