A high-sugar diet, but not obesity, reduces female fertility in Drosophila melanogaster

The global obesity epidemic is a major public health concern (https://www.who.int/publications/i/item/9789240075634), and epidemiological and animal model studies show a link between obesity and many negative health outcomes, including loss of fertility (de Melo et al., 2021; Silvestris et al., 2018). The World Health Organization defines ‘overweight’ (body mass index over 25) and ‘obesity’ (body mass index over 30) as abnormal or excessive fat accumulation that presents a risk to health (www.who.int). In humans, the sharp increase in obesity incidence (largely owing to unhealthy diets and lack of physical activity) correlates with a corresponding decrease in fertility (Dağ and Dilbaz, 2015). In mammalian models showing that obesity lowers fertility through various proposed mechanisms (e.g. altered gonadotropin responses, reduced number of eggs and rates of fertilization), obese mice have been generated through obesogenic diets or genetic mutations that alter hormonal regulation and eating behavior (Haemmerle et al., 2006). In the Drosophila research field, obesity is more simply defined as increased accumulation of stored fat (Gáliková and Klepsatel, 2018; Musselman and Kühnlein, 2018). Drosophila females fed a diet with high sugar levels are obese and have reduced fertility (Brookheart et al., 2017; Morris et al., 2012). These studies in mammals and Drosophila have thus not distinguished whether effects on fertility are a consequence of the diet itself, the obesity (caused by the diet), or a combination thereof. This major shortcoming has limited our ability to investigate fully the root cause of the fertility decline often attributed to obesity.

The genetic model organism Drosophila melanogaster represents a powerful system for investigating fundamental aspects of metabolism and physiology (Chatterjee and Perrimon, 2021) implicated in obesity. Drosophila adipocytes reside alongside hepatocyte-like oenocytes in an organ called the fat body (Arrese and Soulages, 2010; Chatterjee and Perrimon, 2021), and lipid metabolism is controlled by evolutionarily conserved metabolic pathways (Chatterjee and Perrimon, 2021). For example, mutation of the Drosophila anti-obesogenic genes brummer (which encodes a triglyceride lipase) or adipose (which encodes a WD40/TPR-domain protein) leads to obesity, as does mutation of their mouse/human homologs Atgl/ATGL (Pnpla2/PNPLA2) and Wdtc1/WDTC1, respectively (Grönke et al., 2005; Häder et al., 2003; Lai et al., 2009; Schreiber et al., 2019; Suh et al., 2007). Similar to effects observed in mice and humans (Grönke et al., 2005; Suh et al., 2007), feeding Drosophila a diet high in sugar leads to obesity, insulin resistance, hyperglycemia, and cardiovascular abnormalities (Buescher et al., 2013; Guida et al., 2019; Morris et al., 2012), further underscoring the metabolic and physiological similarities between Drosophila and mammals.

The physiological regulation of Drosophila oogenesis has been extensively studied (Drummond-Barbosa, 2019). Drosophila females have a pair of ovaries, each subdivided into ∼15 ovarioles (Fig. 1A). Each ovariole has a germarium followed by progressively more developed egg chambers (also known as follicles) (Fig. 1B). Two to three germline stem cells reside within a specialized niche (composed primarily of cap cells) in the anterior portion of each germarium (Fig. 1C). Each germline stem cell division generates a germline stem cell and a posteriorly displaced cystoblast that undergoes four rounds of incomplete mitoses to generate a 16-cell cyst. Follicle cells surround the 16-cell germline cyst to form a new egg chamber that leaves the germarium. Egg chambers develop through 14 recognizable stages of oogenesis to form a mature egg, with vitellogenesis (i.e. yolk uptake) beginning at stage 8. Previous studies on the effects of dietary yeast and diet-dependent signaling pathways identified major points of physiological regulation of oogenesis (Drummond-Barbosa, 2019). For example, insulin signaling in the ovary and adipocytes is required for normal germline stem cell maintenance and proliferation, early germline cyst survival, egg chamber growth, and progression through vitellogenesis (Armstrong and Drummond-Barbosa, 2018; Hsu and Drummond-Barbosa, 2009; LaFever and Drummond-Barbosa, 2005). More recently, two groups reported that Drosophila females maintained on high-sucrose food for 7 days are obese, have insulin resistance and lay fewer eggs than those on low sucrose (Brookheart et al., 2017; Morris et al., 2012). One study reported that ovaries are smaller in females fed a high-sucrose diet (Brookheart et al., 2017). However, two important points remain unclear: which specific stages of oogenesis are affected in obese Drosophila females on a diet high in sucrose and how diet versus obesity contribute to their reduced fertility.

Here, we show that adult Drosophila females on a high-sugar diet become obese and produce fewer eggs as a result of increased death of early germline cysts and degeneration of vitellogenic egg chambers, and that laid eggs have lower hatching rates. By contrast, no negative effects on oogenesis result from obesity caused by adult adipocyte-specific knockdown of the conserved anti-obesogenic genes brummer or adipose, demonstrating that obesity is not sufficient to decrease female fertility. High-sugar-obese females, but not brummer-knockdown females, have increased levels of glycogen, trehalose and glucose, as well as insulin resistance in fat bodies (but not in ovaries). Intriguingly, when an extra source of dietary water is provided to females on a high-sugar diet, they remain obese and maintain high levels of trehalose and fat body insulin resistance, yet their egg production rates are largely restored in association with a dramatic reversal of elevated glucose levels. Altogether, our findings show that obesity, high glycogen, and fat body insulin resistance are not responsible for the reduced fertility of females on a high-sugar diet, indicating that insulin signaling remains above the threshold required for oogenesis. We further show that dietary water-dependent factor(s) (e.g. glucose and/or unknown molecules) are instead responsible for impairing female fertility on a high-sugar diet, providing a foundation for future research on the molecular mechanisms linking high levels of dietary sugars to disruption of specific processes during oogenesis.

Previous studies showed that Drosophila females on a high-sugar diet are obese and produce fewer eggs (Brookheart et al., 2017; van Dam et al., 2020); we therefore set out to pinpoint which specific stages of oogenesis are sensitive to high dietary sugar. We first confirmed that females fed a high-sugar diet become obese and lay fewer eggs under our experimental conditions (Fig. 1D-G). Based on Nile Red staining of lipid droplets in whole-mount adipocytes (Fig. 1D,E) and biochemical measurements of fat body triglycerides (Fig. 1F), we observed that females maintained on a high-sugar diet for 7 days had markedly larger lipid droplets and three-fold higher triglyceride levels compared with females on a control diet (Fig. 1D-F). To determine how quickly females develop obesity on a high-sugar diet, we measured their triglyceride content daily from 1-6 days of age on control versus high-sugar diets. Remarkably, whereas control females gradually lost triglycerides during their first 3 days of adulthood (presumably owing to the elimination of larval fat body cells present at eclosion; Aguila et al., 2007), females on a high-sugar diet displayed increased triglyceride content starting within 2 days of eclosion (Fig. 1G). We next measured the number of eggs produced by these high-sugar-obese females daily over a 15-day period. From the earliest time point until the end of the time course, high-sugar-obese females consistently laid fewer eggs compared with females on the control diet (Fig. 1H). Moreover, the eggs laid by high-sugar-obese females hatched at a ∼30% lower rate relative to control eggs (Fig. 1I), in apparent contrast to a previous study showing no effect of a high-sugar diet on hatching rates (Brookheart et al., 2017). [In addition, we note that our measurements of egg laying and hatching rates also fall within a higher range than in this previous study (Brookheart et al., 2017), likely owing to differences in methodology, such as density of flies and/or food composition.] Altogether, these experiments confirm that females on a high-sugar diet are obese and less fertile, and further show that obesity and loss of fertility develop concomitantly starting within a couple of days of being fed a high-sugar diet.

We then investigated whether and how a high-sugar diet affects specific stages and processes of oogenesis known to be regulated by organismal physiology (Drummond-Barbosa, 2019), starting with germline stem cells. Germline stem cell numbers in females on control and high-sugar diets remained similar at 0, 5, 10 and 15 days (Fig. 2A,B, Table S1), as did cap cell numbers (Fig. 2A,C, Table S2). There were no significant differences in the frequencies of germline stem cells labeled by 5-ethynyl-2′-deoxyuridine (EdU; an S phase marker) (Fig. 2D,E, Table S3) or phosphorylated histone H3 (an M phase marker) (Fig. 2F,G, Table S3). These results indicate that a high-sugar diet has no effect on either maintenance or proliferation of germline stem cells.

To determine whether a high-sugar diet affects the rate of egg chamber growth, we took advantage of a well-described lineage-tracing system (Drummond-Barbosa and Spradling, 2001; Gandara and Drummond-Barbosa, 2022; Harrison and Perrimon, 1993; Margolis and Spradling, 1995) (see Materials and Methods). We labeled dividing germ cells in the anterior region of germaria of females on control versus high-sugar diets and followed their progress through oogenesis after 1 and 3 days (Fig. 2H). The most developed labeled cysts reached similar stages in females on either diet, indicating similar growth and development rates (Fig. 2I). We also measured the proliferation rates of follicle cells surrounding these growing germline cysts based on the increase in the size of clones generated from single, labeled follicle cells and found no difference based on diet (Fig. 2J,J′,K). These results show that the growth of developing egg chambers is not affected by a high-sugar diet.

Having found that germline stem cell behavior and egg chamber growth are not altered on a high-sugar diet (Fig. 2), we asked whether high levels of dietary sugar might cause death of early germline cysts or of egg chambers entering vitellogenesis. We labeled dying early germline cysts using the ApopTag TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) assay (Fig. 3A) and observed a fourfold increase in the frequency of germaria with dying cysts in females maintained on high-sugar versus control diets (Fig. 3B, Table S4). We identified degenerating vitellogenic egg chambers based on the presence of pyknotic nuclei (Fig. 3C,D). Females on a high-sugar diet had twice as many ovarioles containing degenerating vitellogenic egg chambers as females on a control diet (Fig. 3E, Table S5). Therefore, we conclude that the reduced egg production of females maintained on a high-sugar diet is a direct consequence of the increased death of early germline cysts and vitellogenic egg chambers.

Many studies (including the present one) have shown that obesity induced by high dietary sugar leads to decreased fertility in Caenorhabditis elegans, Drosophila and mammals (Alcántar-Fernández et al., 2018; Brookheart et al., 2017; de Melo et al., 2021; van Dam et al., 2020) (Fig. 1D-I); however, it remains unclear whether these effects on fertility are a direct consequence of diet, obesity or both. To determine whether obesity alone is sufficient to impair fertility in Drosophila females, we generated obese females on a normal diet through genetic manipulation. We screened several genes involved in lipid metabolism using adult adipocyte-specific RNA interference (RNAi) knockdown (Armstrong et al., 2014) and fat body triglyceride measurements (Grönke et al., 2005) (Fig. 4A). Adult adipocyte-specific RNAi against brummer and adipose resulted in obese females (hereafter referred to as ‘genetically obese’ females) (Fig. 4A). Accordingly, the adipocytes of genetically obese females had much larger lipid droplets compared with control RNAi adipocytes (Fig. 4B-E). Notably, brummer and adipose affect fat storage through distinct pathways: brummer encodes the homolog of mammalian ATGL and is the main Drosophila lipase for mobilization of lipids from lipid droplets (Grönke et al., 2005), whereas adipose encodes the homolog of WDTC1, which binds to the DDB1-CUL4-ROC1 E3 ligase, targeting lipogenic enzymes for degradation (Groh et al., 2016). We used both types of genetically obese females in our experiments to ensure that our results would be interpretable based on their obesity (as opposed to specific genes or pathways that are disrupted).

We measured fertility and examined specific stages and processes of oogenesis in genetically obese females as described above. Genetically obese females did not display any reduction in egg production or hatching rates relative to control RNAi females (Fig. 4F,G). Accordingly, germline stem cell maintenance and proliferation (Fig. S1, Tables S6-S8), early germline cyst death (Fig. 4H, Table S9) and egg chamber degeneration (Fig. 4I, Table S10) were similar in control and genetically obese females. These results demonstrate that obesity alone does not negatively impact oogenesis.

To determine whether a high-sugar diet is sufficient to reduce fertility or if it requires co-occurring obesity, we would ideally maintain females on a high-sugar diet while preventing them from becoming obese. We attempted to genetically suppress obesity in females maintained on a high-sugar diet by using a well-characterized construct to overexpress the lipase-encoding gene brummer (Grönke et al., 2005; Zhang et al., 2020) while simultaneously inducing RNAi against Lsd-2 (see Fig. 4A) in adult adipocytes (‘high brummer low Lsd-2’). Lsd-2 encodes the homolog of mammalian perilipin, a protein that envelops lipid droplets and limits Brummer access (Grönke et al., 2003). Therefore, ‘high brummer low Lsd-2’ represents a powerful genetic manipulation for inducing leanness. Notably, the levels of adipocyte lipid storage in ‘high brummer low Lsd-2’ females were just as high as in control females on a high-sugar diet (Fig. S2); it remains unclear whether this might be due to differences in lipid droplet composition in high-sugar-obese females. Regardless, these results precluded our intended analysis of non-obese females on a high-sugar diet.

Females maintained on a high-sugar diet and adipocyte brummer knockdown females on a normal diet were similarly obese (Figs 1D-F and 4A-E), but only high sugar females had impaired oogenesis (Figs 1H,I and 4F,G). Previous studies showed that glycogen and trehalose levels are increased in high-sugar-obese females (Buescher et al., 2013; van Dam et al., 2020) and that the levels of glycogen do not change in brummer¹ global mutants (Grönke et al., 2005). Therefore, to investigate possible reasons for the difference in fertility between females on high sugar versus adipocyte brummer knockdown females, we compared their levels of glycogen, trehalose and glucose relative to their respective controls. In agreement with the literature, we observed that glycogen levels were doubled in high-sugar-obese females compared with their controls (Fig. 5A), but remained unchanged in brummer-knockdown-obese females compared with control RNAi females (Fig. 5B). In addition, we also found that trehalose levels in high-sugar-obese females were fourfold those of their controls (Fig. 5C), whereas trehalose was unaffected in genetically obese females relative to control RNAi females (Fig. 5D). Moreover, our results show a threefold increase in glucose levels in high-sugar-obese females (Fig. 5E), but no consistent change in brummer-knockdown-obese females (Fig. 5F). Thus, high-sugar-obese females, but not brummer-knockdown-obese females, had increased levels of glycogen, trehalose and glucose, showing a correlation between high carbohydrate levels in females and reduced fertility.

Sustained increases in circulating sugars are typically associated with insulin resistance (i.e. an impaired response to insulin) in adipocytes in Drosophila and mammals (Bayliak et al., 2019). We previously showed that insulin signaling in adipocytes and/or ovaries promotes the survival of early germline cysts and vitellogenic egg chambers, among its other roles (Armstrong and Drummond-Barbosa, 2018; Hsu et al., 2008; LaFever and Drummond-Barbosa, 2005). We therefore asked if the high trehalose and glucose levels we observed in high-sugar-obese females (Fig. 5C-F) might lead to increased expression of the Lipocalin-encoding gene Neural Lazarillo (NLaz), which is routinely used as a marker of insulin resistance in Drosophila larvae and adults (Hull-Thompson et al., 2009; Lourido et al., 2021; Pasco and Léopold, 2012; Ruiz et al., 2014). We found that NLaz mRNA levels were significantly higher in the fat body and ovary of high-sugar-obese females (Fig. 5G), but were unchanged in the fat body and reduced in the ovaries of brummer-knockdown-obese females (Fig. 5H) compared with their respective controls, initially suggesting that reduced insulin signaling might contribute to the increased death of early germline cysts and vitellogenic egg chambers in high-sugar-obese females. However, subsequent experiments (see below and Fig. S3) showed that NLaz is not a reliable marker of insulin resistance in the ovary, despite its reported requirement for high-sugar diet-induced insulin resistance in larvae (Pasco and Léopold, 2012).

We next addressed whether differences in food consumption might also contribute to the metabolic and ovarian phenotypes of high-sugar-obese females. Specifically, we compared the food intake of females on normal versus high-sugar diets using the consumption-excretion method, which measures the total amount of blue-dye-containing food ingested within a 24-h period based on the absorbance of the dye inside females plus the dye excreted by them (see Materials and Methods for details) (Gandara and Drummond-Barbosa, 2022; Shell et al., 2018). High-sugar-obese females ate one-quarter as much as their controls on a normal diet (Fig. 5I), in agreement with a previous study (van Dam et al., 2020). However, given that the concentration of sucrose in the high-sugar food is sixfold that of the control food, females on the high-sugar diet ingested higher levels of sugar compared with their controls. By contrast, genetically obese females showed similar food intake to control RNAi females (Fig. 5J). These results suggested that, in principle, increased sugar consumption, reduced intake of other nutrients, or both might potentially contribute to the reduced fertility of high-sugar-obese females.

A recent study showed that providing an additional source of dietary water reverses the negative effects of a high-sugar diet on the lifespan of Drosophila females, although these females remain obese with high levels of trehalose and insulin resistance (van Dam et al., 2020). We therefore asked if supplementation of a high-sugar diet with an ad libitum water source might also reverse the fertility defects of high-sugar-obese females. We first examined the effects of dietary water supplementation on the metabolic parameters of females on a high-sugar diet. As previously reported (van Dam et al., 2020), triglyceride levels were equally high in females on a high-sugar diet with or without water supplementation (Fig. 6A). Interestingly, glycogen levels also remained high in females on a high-sugar diet regardless of extra dietary water (Fig. 6B). By contrast, water supplementation reduced the levels of trehalose by 17% (Fig. 6C) and glucose by 58% (Fig. 6D) in females on a high-sugar diet. The levels of NLaz expression nearly doubled in the ovaries and were slightly elevated in the fat bodies of females on a high-sugar diet supplemented with water compared with those without water supplementation (Fig. 6E). However, it remained unclear whether NLaz expression levels accurately reported insulin resistance, especially in the ovaries, where, to our knowledge, no NLaz analysis has been published.

To assess insulin resistance independently, we measured the levels of phosphorylated AKT kinase (pAKT), the active form of a downstream component of the insulin pathway, which reflects insulin signaling levels (van Dam et al., 2020; Warr et al., 2018). As a control experiment, we first measured pAKT levels in fat bodies and ovaries (dissected and incubated ex vivo with or without insulin) from females on a normal diet with or without water supplementation (Fig. S3). In the fat bodies of these females, there was a 50% increase in pAKT levels in response to added insulin regardless of dietary water supplementation (Fig. S3A,A′). In the ovaries of females on a normal diet, the addition of exogenous insulin resulted in little to no increase in pAKT levels (Fig. S3B,B′). The apparent lack of response to insulin addition suggests that insulin signaling is already maximally activated in these dissected ‘normal diet’ ovaries. This interpretation (as opposed to the ovaries having insulin resistance) is based on evidence that insulin signaling is elevated in the ovaries of females on a normal diet (Hsu and Drummond-Barbosa, 2009) and that high levels of insulin signaling are required in the ovary for high levels of egg production by these females (Drummond-Barbosa and Spradling, 2001; Hsu and Drummond-Barbosa, 2009, 2011; Hsu et al., 2008; LaFever and Drummond-Barbosa, 2005; LaFever et al., 2010). We then analyzed ovaries and fat bodies from females on a high-sugar diet with or without water supplementation in a similar manner. In females on a high-sugar diet, the fat bodies did not show an increase in pAKT levels with added insulin (Fig. 6F,F′) (in contrast to ‘normal diet’ fat bodies; Fig. S3A,A′), and results were similar regardless of water supplementation (Fig. 6F′). These results are in agreement with previous reports that a high-sugar diet induces fat body insulin resistance (Pasco and Léopold, 2012; van Dam et al., 2020), and that this is not reversible by extra dietary water (van Dam et al., 2020). The overall levels of pAKT in the ovaries of females on a high-sugar diet appeared higher than those of females on a high-sugar diet supplemented with water (Fig. 6G,G′), possibly owing to higher expression of insulin pathway components, and the response of dissected ovaries to addition of insulin was statistically equivalent regardless of water supplementation (Fig. 6G,G′). These results argue against ovarian insulin resistance in females on a high-sugar diet as the cause for their reduced fertility. Moreover, the discrepancies between NLaz and pAKT results (Fig. 5G, Fig. S3B,B′, Fig. 6E,G,G′) indicate that NLaz is not a reliable indicator of insulin resistance in ovaries. Food intake on a high-sugar diet was not altered by water supplementation (Fig. 6H), as previously reported (van Dam et al., 2020). Altogether, our analysis indicates that dietary water supplementation drastically reverses the high glucose levels (with a smaller effect on trehalose) but not the fat body insulin resistance of females on a high-sugar diet, with no significant effects on their glycogen levels, food intake or obesity.

Next, we compared the fertility of high-sugar-obese females with or without dietary water supplementation. Interestingly, extra dietary water significantly increased egg production of females on a high-sugar diet (Fig. 6I), whereas hatching rates remained similar regardless of water supplementation (Fig. 6J). These results indicate that a high-sugar diet impairs egg production and hatching rates through different mechanisms. In accordance with the increase in egg production, water supplementation significantly decreased the frequency of dying early germline cysts and vitellogenic egg chambers in high-sugar-obese females (Fig. 6K,L, Tables S11, S12). Our finding that water supplementation reverses the negative effects of a high-sugar diet on egg production contrasts with the lack of rescue of fecundity by dietary water previously reported (van Dam et al., 2020). Multiple factors could potentially account for this discrepancy, including differences in mating status, genetic background and food composition, among other methodological differences. Notably, in the previous report (van Dam et al., 2020), females were kept under more crowded conditions and eggs were counted within vials at a single time point, which could also impact the experimental results.

Taken together, our findings lead to major conclusions regarding the potential mechanisms mediating the negative effects of a high-sugar diet on oogenesis. First, the specific correlation between high glucose levels and lower egg production in our study suggests the possibility of causality (for future investigation). Second, despite their reduced food consumption and elevated fat body insulin resistance, high-sugar-obese females have sufficient nutrient intake and insulin signaling to support oogenesis. Third, females on a high-sugar diet supplemented with water remain obese yet have largely restored egg production, strengthening our conclusion that obesity does not directly cause the oogenesis defects – namely, germline cyst death and egg chamber degeneration – observed in high-sugar-obese females.

From Drosophila to humans, many studies have established a strong association between obesity and reduced fertility (Alcántar-Fernández et al., 2018; Brookheart et al., 2017; de Melo et al., 2021; Espinós et al., 2020; Moholdt and Hawley, 2020; Morris et al., 2012). However, the methods for generating obese animals in these studies introduce major confounding factors, such as diets high in sugar and/or fat or hormonal changes leading to increased appetite. For example, a high-sugar diet causes obesity in Drosophila, and these obese females produce fewer eggs (Brookheart et al., 2017; Morris et al., 2012), making it unclear whether a high-sugar diet, obesity, or both reduce their fertility. In this study, we show that high-sugar-obese Drosophila females produce fewer eggs because of increased death of early germline cysts and degeneration of vitellogenic egg chambers (Fig. 7). Further, we demonstrate that this reduction in fertility is not caused by obesity, fat body insulin resistance, or high trehalose or glycogen levels (Fig. 7). Instead, we find a specific correlation between high glucose levels and the decreased egg production of females on a high-sugar diet, as supplementation with dietary water drastically improves both glucose levels and egg production (by decreasing death of early germline cysts and vitellogenic egg chambers), with much smaller or no effects on other metabolic parameters (Fig. 7). Our findings provide a foundation for future studies to investigate the causal relationship between high glucose levels (and/or other dietary water-dependent factors) and reduced Drosophila fertility. More broadly, they also highlight the importance of carefully controlling for confounding factors when investigating how obesity affects the risk of infertility and other obesity-associated disorders, including chronic inflammation, cardiovascular abnormalities, and cancers (Bayliak et al., 2019; Warr et al., 2018).

Our previous published studies have shown that adipocytes are integral to the physiological regulation of Drosophila oogenesis (Drummond-Barbosa, 2019). For example, dietary yeast rapidly changes the expression of enzymes in multiple metabolic pathways in adipocytes, and adipocyte-specific knockdown of key components of these pathways leads to specific phenotypes, such as germline stem cell loss, early cyst death, or degeneration of vitellogenic stages (Matsuoka et al., 2017). Reduced amino acid sensing by adipocytes reduces germline stem cell numbers and inhibits ovulation (Armstrong et al., 2014), and blocking insulin signaling specifically in adipocytes increases loss of germline stem cells, death of early germline cysts and degeneration of vitellogenic egg chambers (Armstrong and Drummond-Barbosa, 2018). Our current findings indicate that none of these known roles of adipocytes is impaired, despite very high levels of fat accumulation in genetically obese females. We conclude that obesity in and of itself does not impair adipocyte function.

This work underscores the importance of designing studies that tease apart the contributions of obesity itself from obesity-independent effects of obesogenic diets. Between 1980 and 2018, the percentage of obese adults in the USA rose from 19% to 42% for women, and from 13% to 43% for men (Practice Committee of the American Society for Reproductive Medicine et al., 2021). This rapid rise in obesity was largely a consequence of unhealthy diets and/or lack of physical exercise (Bentley et al., 2020) (https://www.cdc.gov/obesity/index.html). Therefore, dietary and/or physical activity factors are likely contributing to the adverse health outcomes of a large fraction of adults that are currently obese. Meanwhile, ∼10-30% of obese individuals are clinically recognized as having metabolically healthy obesity (i.e. obesity with normal glucose and lipid metabolism parameters and absence of hypertension) for unclear reasons (Blüher, 2020). Yet, most studies in human populations or laboratory animal models examining the connection between obesity and infertility do not clearly distinguish adiposity and dietary contributions to their findings (Alcántar-Fernández et al., 2018; Bonfini et al., 2021; Brookheart et al., 2017; Buescher et al., 2013; de Melo et al., 2021; Diop et al., 2015; Lourido et al., 2021; Morris et al., 2012; Musselman et al., 2011; Pasco and Léopold, 2012; Santoro et al., 2022; Schoiswohl et al., 2015). For example, to our knowledge, the effect of obesity caused by mutation of Wdtc1 (Suh et al., 2007) or Atgl (Schoiswohl et al., 2015) on the fertility of mammalian females has not been studied. It would be interesting to compare findings from such types of studies to our findings of normal fertility in genetically obese females.

Insulin signaling is crucial for the regulation of Drosophila oogenesis. Insulin-like peptides act directly on the germline to control germline stem cell proliferation, egg chamber growth and vitellogenesis (Hsu and Drummond-Barbosa, 2011; LaFever and Drummond-Barbosa, 2005), on cap cells to control germline stem cell maintenance (Hsu and Drummond-Barbosa, 2009, 2011; Yang et al., 2013), and on adipocytes to control germline stem cell maintenance, early germline cyst survival and vitellogenesis (Armstrong and Drummond-Barbosa, 2018). We found no evidence of insulin resistance in ovaries of females on a high-sugar diet despite increased levels of NLaz, indicating that NLaz does not accurately report ovarian insulin resistance (and consistent with their normal rates of germline stem cell proliferation and egg chamber growth). By contrast, we showed that a high-sugar diet leads to fat body insulin resistance, as previously described (Pasco and Léopold, 2012; van Dam et al., 2020). However, germline stem cell maintenance is not affected in high-sugar-obese females, and germline cyst and vitellogenic egg chamber death are reversed despite persistent adipocyte insulin resistance in high-sugar-obese females supplied with extra dietary water. Altogether, we can definitively conclude that females on a high-sugar diet have sufficient levels of insulin signaling in their adipocytes to support robust rates of oogenesis. These findings raise the possibility that in other systems where insulin resistance is observed (Arruda et al., 2017; de Melo et al., 2021; Lourido et al., 2021; Morris et al., 2012; Pasco and Léopold, 2012; Schoiswohl et al., 2015; van Dam et al., 2020), there might be sufficient levels of insulin signaling for normal functioning of the tissues in question.

Water participates in various biological processes, including metabolism (Adhikari et al., 2020; Chaplin, 2006; da Silva and Soveral, 2023), and a correlation between obesity and dehydration has been reported in humans (Chang et al., 2016), although it remains unclear whether dehydration correlates with obesity itself or with obesogenic diets. Incidentally, associations between high salt intake (which causes dehydration) and increased risk of overweight or obesity in humans have been reported (Lee et al., 2023). In mice, a high-salt diet leads to obesity via a mechanism involving stimulation of endogenous fructose production and hyperphagia (Lanaspa et al., 2018). In Drosophila, dietary water supplementation reverses the shortened lifespan and dehydration (i.e. decreased hemolymph volume) caused by a high-sugar diet, even though flies remain obese and have high levels of trehalose and fat body insulin resistance (van Dam et al., 2020). The water-reversible reduced lifespan of flies on a high-sugar diet is the consequence of increased purine catabolism (which leads to urid acid accumulation and stone formation in Malpighian tubules by 28 days of age) (van Dam et al., 2020). Here, we further show that dietary water supplementation drastically reverses the high glucose levels and reduced fertility of Drosophila females on a high-sugar diet. Future studies should address whether high glucose levels have a causal relationship with reduced fertility and, if so, whether the ovaries sense glucose levels, decreased hemolymph volume (i.e. dehydration), or other signals. In addition, other potential mechanisms involving known candidates (e.g. dysregulated purine catabolism; van Dam et al., 2020) or as-yet-unknown factors (through unbiased approaches) should also be pursued.

Drosophila stocks were maintained at 22°C on standard medium composed of 4.64% w/v yellow cornmeal (Quaker), 5.8% v/v unsulfured cane molasses (Sweet Harvest Feeds), 1.74% w/v active dry yeast (Red Star), 0.93% w/v agar (Apex BioResearch Products), 1.05% Tegosept (Apex Chemicals) and 0.36% propionic acid (Apex Chemicals). (Note: The concentration of 5.8% molasses in our standard medium corresponds to ∼5% sugar, mostly sucrose.) The y w ‘wild-type’ strain was used for all experiments, except where indicated. The adipocyte-specific 3.1Lsp2-Gal4 (Lazareva et al., 2007) and tub-Gal80^ts (McGuire et al., 2003) were used in combination to induce adult-specific RNAi as described (Armstrong et al., 2014) using the UAS-RNA hairpin lines listed in Table S13 (Dietzl et al., 2007; Diop et al., 2015; Dong et al., 2021; Garrido et al., 2015; Matsuoka et al., 2017; Pandey et al., 2021; Perkins et al., 2015; Wat et al., 2020). The UAS-brummer construct has been described (Grönke et al., 2005). The X-15-29, X-15-33 and MKRS, hs-FLP lines used for lineage-tracing experiments have been described (Harrison and Perrimon, 1993). For most experiments, flies were raised at 22°C on standard medium and 0- to 1-day-old females (with males) were incubated at 25°C, ≥70% humidity, on standard medium (∼5% sugar content; control diet) or standard medium with added sucrose (∼32% sugar content; high-sugar diet). For adult adipocyte-specific experiments, flies were raised at 18°C on standard medium and 0- to 1-day-old adult females (with males) were incubated at 29°C, ≥70% humidity, on standard medium. For dietary water supplementation, a barrier tip (20 µl; Genesse Scientific) filled with 1% agar was inserted in the food and taped to the wall of the vials containing high-sugar medium. For all experiments, fly medium was supplemented with dry yeast and changed daily (along with the 1% agar-filled tip, if applicable).

To measure food intake, we used the consumption-excretion dye-based method (Shell et al., 2018) with minor modifications, as described previously (Gandara and Drummond-Barbosa, 2022). Briefly, ten 0- to 1-day-old couples were maintained on control or high-sugar medium for 6 days at 25°C (or 29°C, for RNAi experiments). Females were then transferred to empty vials closed with plugs containing (on their internal surface) ∼0.5 ml control or high-sugar medium with 0.25% FD&C Blue No. 1 (Spectrum Chemicals), respectively, for 24 h. (A similar procedure was followed to control for background absorbance, except that media on plugs did not have the blue dye.) To recover the internal dye, females were homogenized in 1.5 ml water, homogenates were centrifuged at 10,000 g for 1 min, and supernatants were collected. To collect the dye excreted by females, 3 ml water were used to rinse the vials and water extracts were vortexed for 10 s. Absorbance at 630 nm was measured using a Synergy H1 spectrophotometer (Agilent BioTek) and converted to amount of medium consumed (µg/fly) based on a standard curve of pure FD&C Blue No. 1 in water. Experiments were carried out in triplicate and unpaired, two-tailed Student's t-test (GraphPad Prism) was used to determine statistical significance.

We counted the number of eggs laid per female over time using our well-established methodology with minor modifications (Armstrong et al., 2014; Drummond-Barbosa and Spradling, 2001, 2004; Gandara and Drummond-Barbosa, 2022; Ma et al., 2020; Weaver and Barbosa, 2021; Weaver and Drummond-Barbosa, 2019). Briefly, five couples were maintained in inverted perforated plastic bottles closed with control or high-sugar medium plates supplemented with dry yeast, in six replicates, at 25°C (or 29°C, for RNAi experiments). Plates were changed twice a day, and eggs laid within a 24-h period were counted daily for 15 days to calculate the average number of eggs produced per female per day. Experiments were performed in triplicate and statistical analysis was carried out using F-test of third order polynomial (GraphPad Prism) fitted curves.

To quantify hatching rates, eggs laid overnight from the 7- to 8-day time points of the egg count experiments above were collected as previously described (Gandara and Drummond-Barbosa, 2022). For each experiment, ten groups of ten eggs per condition were placed on a molasses plate around a dab of yeast paste and incubated at 25°C for 24 h in a humid chamber in triplicate. The unhatched eggs were counted and subtracted from the total to calculate the number of hatched eggs. Experiments were carried out in triplicate and unpaired, two-tailed Student's t-test (GraphPad Prism) was used to determine statistical significance.

Ovaries were dissected and ovarioles were teased apart in Grace's Insect Medium (BioWhittaker). Samples were fixed for 15 min at room temperature in fixing solution [5.3% paraformaldehyde (Ted Pella) in Grace's medium], and then rinsed and washed three times for 15 min each in PBST [0.1% Triton X-100 in PBS (10 mM NaH₂PO₄/NaHPO₄, 175 mM NaCl, pH 7.4)]. Samples were blocked for 3 h at room temperature in blocking solution [5% normal goat serum (MP Biochemicals) plus 5% bovine serum albumin (Sigma-Aldrich) in PBST] and then incubated overnight at 4°C in primary antibodies diluted in blocking solution: 1:20 mouse monoclonal anti-α-Spectrin [3A9, Developmental Studies Hybridoma Bank (DSHB)]; 1:20 mouse monoclonal anti-Lamin C (LC28.26, DSHB); 1:200 chicken anti-β-gal antibody (ab9361, Abcam); or 1:200 rabbit anti-phospho-histone H3 (Ser10) (06-570, Millipore). Ovaries were washed as described and incubated with secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 568 (A11034 or A11004, respectively, Molecular Probes/Invitrogen; 1:400) in blocking solution for 2 h (protected from light) at room temperature. Samples were washed (protected from light) and mounted in Vectashield containing DAPI (4′,6-diamidino-2-phenylindole, a fluorescent stain specific for DNA) (Vector Laboratories). For lipid droplet visualization, abdominal carcasses (without ovaries or guts) dissected in Grace's medium were fixed for 20 min at room temperature in fixing solution, and then rinsed and washed three times for 15 min each in PBST. Carcasses were then incubated with 1:200 Alexa Fluor 488-conjugated phalloidin (a bicyclic peptide that binds F-actin) (A12379, Invitrogen) for 40 min at 4°C and washed as described above. Samples were then stained with 25 ng/ml Nile Red (a lipophilic fluorescent dye that stains lipid droplets; 19123, Sigma-Aldrich) in Vectashield containing DAPI for at least 48 h at 4°C. All experiments were performed in triplicate, and data were collected using a Zeiss AxioImager-A2 fluorescence microscope or Zeiss LSM700 or LSM900 confocal microscopes.

Cap cells were identified based on their ovoid nuclei and Lamin C-positive staining, whereas germline stem cells were identified based on their juxtaposition to cap cells and typical fusome morphology and position (de Cuevas and Spradling, 1998). Experiments were carried out in triplicate and statistical analysis was performed using two-way ANOVA with interaction (GraphPad Prism).

To label germline stem cells in S phase, we used an EdU (a nucleoside analog of thymidine) incorporation assay. Intact dissected ovaries were incubated for 1 h at room temperature in 100 μM EdU (Molecular Probes) in Grace's insect medium, then washed and fixed as described above. Following incubation with primary antibodies (anti-phospho-histone H3, anti-Lamin C and anti-α-Spectrin), samples were subjected to the Click-iT reaction according to the manufacturer's instructions (Life Technologies) for 30 minutes at room temperature. Ovaries were washed, incubated with secondary antibodies, and washed again prior to mounting and microscopy, as described above. We calculated the fraction of EdU-positive or phospho-histone H3-positive germline stem cells as a percentage of the total number of germline stem cells analyzed per condition. Experiments were carried out in triplicate and statistical analysis was performed using Chi-square analysis.

Egg chamber development through oogenesis involves the growth of the germline cysts (originally produced by four incomplete mitotic divisions of cystoblasts in the anterior portion of the germarium) in coordination with mitotic divisions of surrounding follicle cells (until stage 7 of oogenesis, when follicle cells transition to endoreplication) (Spradling, 1993). β-Galactosidase (β-gal)-positive clones from single mitotically dividing cells were produced as previously described (Drummond-Barbosa and Spradling, 2001; Gandara and Drummond-Barbosa, 2022). Newly eclosed y w; X-15-29/X-15-33; MKRS, hs-FLP/+ females (with 0- to 1-day-old y w males) were maintained on control or high-sugar diets for 2 days at 25°C. Flies were then heat-shocked at 37°C for 1 h to induce flippase expression (and the generation of single β-gal-positive cells at day zero) and subsequently transferred to their respective media (changed daily) for 1 or 3 days prior to ovary dissections. To assess egg chamber growth, we analyzed β-gal-positive germline and follicle cell clones at both time points. For germline clones (i.e. partially or fully labeled 16-cell cysts generated from mitotically dividing germline stem cells, cystoblasts, or two-, four- or eight-cell cysts present at day 0), we identified the most developed egg chamber stage containing a partially/fully β-gal-labeled cyst in each ovariole analyzed. Statistical analysis was performed using Chi-square analysis. For follicle cells (i.e. β-gal-positive clones generated from single follicle cells labeled at day zero), the number of labeled cells per clone was counted in egg chambers at stages 4-6. Doubling times were calculated using regression line equations (GraphPad Prism). Experiments were carried out in triplicate and statistical analysis was performed using two-way ANOVA with interaction (GraphPad Prism).

To quantify germaria containing dying germline cysts, we used the ApopTag Fluorescein Direct In Situ Apoptosis Detection Kit (S7160, Millipore Sigma) as previously described (Drummond-Barbosa and Spradling, 2001). Degenerating vitellogenic egg chambers were identified in DAPI-stained ovarioles based on the presence of pyknotic nuclei, which are not present in healthy vitellogenic egg chambers. Experiments were carried out in triplicate and statistical analysis was performed using Chi-square analysis.

For triacylglycerol (TAG) quantification, ten whole females or ten pairs of ovaries (dissected in cold PBS) were homogenized in 500 μl cold PBST at 4.0 m/s for 20 s in a FastPrep-24 Classic homogenizer (MP Biomedicals) using 2 ml Lysing Matrix A tubes (MP Biomedicals). Samples were vortexed, transferred to 1.5-ml tubes and centrifuged at 16,000 g for 5 min at 4°C. One-fifth of the supernatant was collected without disturbing the lipid layer to measure protein content. The remaining supernatant was used to collect the lipid layer and heated for 10 min at 70°C in a dry bath. For all experiments (except in Fig. 1G), TAG levels were then measured using the Serum Triglyceride Determination Kit (TR0100-1kt, Millipore) according to the manufacturer's protocol, adapted for 96-well plates. Briefly, 25 μl supernatant from each sample (or 25 μl standards) were transferred in four replicates to 96-well plates, 200 μl TAG reagent was added to half of the wells and Free Glycerol reagent was added to the other half of the wells, and the plate was incubated at 37°C for 30 min. Absorbance at 492 nm was measured using a plate reader (Synergy H1, Agilent BioTek). The TAG content was determined by subtracting free glycerol from TAG, and then the fat body TAG content was determined by subtracting ovary-pair TAG amounts from whole-body TAG amounts. (Note: We did not measure TAG in dissected carcasses directly because fat body cells are often lost during dissection, which introduces technical variability and can lead to inaccurate ‘per fly’ measurements.) For the experiment shown in Fig. 1G, a more sensitive Triglyceride Quantification Assay kit (AB65336-1001, Abcam) was used following the manufacturer's protocol. Briefly, 6 μl samples were used per well in four replicates; 2 μl lipase reagent were added to two replicates and 2 μl lipase buffer were added to the other two replicates. Samples were incubated at room temperature for 20 min, after which 50 μl of triglyceride reaction mix were added to all wells and incubated at room temperature for 60 min. The absorbance of the plate was read at 570 nm using a plate reader (Synergy H1, Agilent BioTek). Fat-body TAG content was determined as described above. Experiments were carried out in triplicate and unpaired, two-tailed Student's t-test (GraphPad Prism) was used to determine statistical significance.

For carbohydrate measurements, five 7-day-old females (per sample) were washed in PBS and transferred to a 1.5-ml tube. All liquid was removed, and females were homogenized in 100 µl cold PBS (on ice) using a motorized pestle (749521-1500, Kontes). Samples were heated for 10 min at 70°C on a heating block and centrifuged at 16,000 g for 5 min at 4°C, and supernatants were collected and stored at −80°C. The Glucose (GO) Assay Kit (GAGO20-1KT, Sigma-Aldrich) was used to measure the amount of free glucose, trehalose (after enzymatic breakdown into two glucose molecules) or glycogen (after enzymatic breakdown into many glucose molecules). For glycogen measurements, 30 µl 1:10 diluted supernatants were transferred to a 96-well plate in six replicates; 100 µl GO reagent containing 2.3 units/ml amyloglucosidase were added to three of the replicates, and 100 µl GO reagent alone were added to the remaining replicates. The plate was incubated at 37°C for 60 min, after which reactions were stopped by addition of 100 µl 6 N sulfuric acid. The 540 nm absorbance was measured using a Synergy H1 plate reader (Agilent BioTek), and glycogen levels were calculated based on a glucose standard curve after subtracting the absorbance measured for free glucose in the untreated samples from the absorbance of the samples digested with amyloglucosidase. For trehalose and glucose measurements, 30 µl 1:8 diluted supernatants were transferred to 1.5-ml tubes containing 30 µl Trehalase Buffer (5 mM Tris pH 6.6, 137 mM NaCl, 2.7 mM KCl) with 2.7 units/µl porcine trehalase or 30 µl of Trehalase Buffer alone. Tubes were incubated at 37°C for 24 h, after which samples were transferred to 96-well plates and processed as described above for glucose measurement using the GO Assay Kit. Free glucose levels (from samples not treated with trehalose) were calculated based on a glucose standard curve. Trehalose levels were calculated based on a glucose standard curve after subtracting the absorbance measured for free glucose in the untreated samples from the absorbance of the samples digested with trehalase. All experiments were carried out in triplicate and unpaired, two-tailed Student's t-test (GraphPad Prism) was used to determine statistical significance.

Ten adult female fat bodies (scraped off carcasses) or pairs of ovaries were dissected and incubated in RNAlater Stabilization Solution (Thermo Fisher Scientific) for 10 min. After RNAlater removal, 250 μl lysis buffer from the RNAqueous-4PCR Total RNA Isolation Kit (Thermo Fisher Scientific) were added and samples were homogenized using a motorized pestle and RNA extraction proceeded according to the manufacturer's instructions. Complementary DNA was synthesized from 1 mg total RNA using oligo (dT) primers and SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific) according to the manufacturer's instructions. Complementary DNA was amplified through a 40-cycle reaction (95°C for 3 s, 55°C for 3 s and 72°C for 20 s for Nlaz; 95°C for 3 s, 55°C for 3 s and 72°C for 20 s for RpL32) using previously described primers for NLaz (5′-GGACAACCCTCGAATGTAACT-3′ and 5′-GACGGCGTATGACTCGTAATC-3′; Lourido et al., 2021) and for RpL32 (also known as RP-49; used as a normalization control; 5′-CAGTCGGATCGATATGCTAAGC-3′ and 5′-AATCTCCTTGCGCTTCTTGG-3′; Weaver and Drummond-Barbosa, 2018). Mock reactions using water without complementary DNA served as negative controls. Experiments were carried out in triplicate and statistical analysis was performed using unpaired, two-tailed Student's t-test on the relative ΔΔCT quantification method (GraphPad Prism).

Six fat bodies or pairs of ovaries (per sample) from 7-day-old females were dissected in Grace's medium and transferred to another well containing Grace's medium with or without 0.5 μM insulin and incubated for 20 min at room temperature. Samples were then transferred to 1.5-ml tubes and 60 μl Protein Lysis Buffer [50 mM Tris-HCl, 0.1% Triton X-100, 200 mM NaCl, 1 mM EDTA, 0.2 mg/ml sodium azide, 1:100 Protease Inhibitor Cocktail (P8340, Sigma-Aldrich)] were added. Samples were homogenized with a motorized pestle on ice and centrifuged at 16,000 g for 5 min at 4°C, and supernatants were collected. Protein quantification was performed using a BCA Protein Assay (23225, Pierce) according to the manufacturer's instructions. From each sample, 10 µg protein in Fluorescent Compatible Sample Buffer (Invitrogen) was heated at 95°C in a heating block for 10 min prior to loading. Samples were electrophoresed on NuPAGE 4 to 12%, Bis-Tris, 1.5 mm, Mini Protein Gels (Invitrogen WG1401BOX) in NuPAGE MOPS SDS running buffer (Invitrogen) at 200 V for 42 min. Samples were then transferred to Low Fluorescence PVDF (Azure) membranes in NuPAGE Transfer Buffer (Life Technologies) at 20 mA for 60 min using a Semidry Blotter (Thermo Fisher Scientific). Membranes were stained for total protein using Red Protein Stain (Azure) and washed three times in methanol and twice in PBS. Membranes were blocked with AzureSpectra Protein Free Blocking Buffer (Azure) for 60 min and incubated with primary antibodies rabbit anti-pAKT [Phospho-Akt (Ser473), 9271, Cell Signaling Technology; 1:1000] and mouse anti-α-Tubulin (12G10, DHSB; 1:500) at 4°C overnight. After rinsing twice and washing three times with PBST, the membranes were incubated with 1:10,000 secondary antibodies Azure 700 mouse and 800 rabbit for 2 h, rinsed twice and washed three times with PBST then washed twice with PBS, according to the manufacturer's protocol. Membranes were imaged using an Azure 600 western blot imaging system. Quantification and normalization of the bands (to generate the raw data) were made using the BandPeakQuantification macro made by Kenji Ohgane and Hiromasa Yoshioka on Fiji (https://www.protocols.io/view/quantification-of-gel-bands-by-an-image-j-macro-ba-bp2l6n4bkgqe/v1) with the following settings: background width (pixels) 3; estimate background from top/bottom; background estimation by mean. Experiments were carried out in triplicate and statistical analysis of the raw data was performed using two-way ANOVA with interaction (GraphPad Prism), which, simply stated, calculates the significance of any differences in how distinct samples respond to insulin (i.e. differences in pAKT level changes without versus with insulin addition among samples).

We thank the Developmental Studies Hybridoma Bank for antibodies, and the Bloomington Stock Center (National Institutes of Health P400D018537), Vienna Drosophila Resource Center, and Ronald Kühnlein for Drosophila stocks. We are grateful to Phil Newmark and his lab members for generously sharing their Azure 600 western blot imaging system and Synergy H1 plate reader. We also thank members of the Newmark and D.D.-B. labs for helpful discussions during lab meetings. We thank Emily Wessel, Sabi Nagarajan, Ana Caroline Gandara, and Alicia Williams for careful reading of the manuscript and helpful editing suggestions.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201769.reviewer-comments.pdf