A gut-derived hormone suppresses sugar appetite and regulates food choice in Drosophila

Animals must adapt their dietary choices to meet their nutritional needs. How these needs are detected and translated into nutrient-specific appetites that drive food-choice behaviours is poorly understood. Here we show that enteroendocrine cells of the adult female Drosophila midgut sense nutrients and in response release neuropeptide F (NPF), which is an ortholog of mammalian neuropeptide Y-family gut-brain hormones. Gut-derived NPF acts on glucagon-like adipokinetic hormone (AKH) signalling to induce sugar satiety and increase consumption of protein-rich food, and on adipose tissue to promote storage of ingested nutrients. Suppression of NPF-mediated gut signalling leads to overconsumption of dietary sugar while simultaneously decreasing intake of protein-rich yeast. Furthermore, gut-derived NPF has a female-specific function in promoting consumption of protein-containing food in mated females. Together, our findings suggest that gut NPF-to-AKH signalling modulates specific appetites and regulates food choice to ensure homeostatic consumption of nutrients, providing insight into the hormonal mechanisms that underlie nutrient-specific hungers.

Animals must be able to select the specific nutrients they need to consume.Food selection is governed by appetites for specific nutrients to ensure adequate ingestion of macronutrients needed to maintain nutritional homeostasis and optimal fitness 1,2 .This has given rise to the hypothesis that organisms can feel specific hungers or appetites for the type of nutrients they need 3,4 .Nutrient-specific appetite has been demonstrated in many organisms, including humans [3][4][5][6] .Such homeostatic nutrient consumption requires sensors that detect the internal nutritional state and mechanisms that translate this information into changes in feeding decisions.Food consumption is controlled by nutritional signals from the periphery, such as the adipokine leptin and a variety of gut hormones, that act together with circulating nutrients on the brain 7 .However, the hormones and mechanisms that govern nutrient-specific appetites that drive appropriate food choices to maintain or restore homeostasis are poorly defined.
The fruit fly Drosophila, like mammals, regulates feeding behaviours according to internal state 1,6,8,9 .The gut is one of the largest endocrine organs, releasing a number of different hormones from specialized enteroendocrine cells (EECs) in both flies and mammals 10,11 .Gut-to-brain signalling conveys important information about the nutritional nature of the intestinal contents, and enteric nutrient-sensing and signalling play key roles in regulating food intake 12,13 .For example, in the nutrient-deficient state, orexigenic or hunger signals from the mammalian gut such as the hormone ghrelin drive appetite to promote food consumption.Conversely, in response to food consumption, the mammalian gut releases glucagon-like peptide 1 (GLP-1), which acts as a satiety signal that reduces further food intake.Such satiety signals prevent excess nutrient intake, which can lead to the development of obesity and associated metabolic disorders, and GLP-1 therapy is effective in reducing body weight by lowering appetite 14 .The fly gut Article https://doi.org/10.1038/s42255-022-00672-zEEC knockdown of NPF also consumed significantly more sugar+yeast food than controls (Fig. 1c and Extended Data Fig. 1c).We also applied the capillary feeder (CAFÉ) assay 24 to quantify sugar intake over a longer period.Before this assay as well, we exposed animals to a 15-hour period of fasting to enhance consumption.Mated females with adult-restricted EEC knockdown of NPF consumed significantly more sugar over the 6-hour period than controls (Fig. 1d and Extended Data Fig. 1d).Since the sugar-feeding phenotype was observed in animals whether they were fully fed or preconditioned by 15-hour fasting (Fig. 1b,c,d), we chose to use animals that were 15-hour fasted for consistency in the following feeding assays, since this allows robust measurements in short-term food intake assays as well as longer-term feeding assays.Conversely, in males, loss of NPF in the EECs led to reduced food consumption over an even longer period of 24 hours (Extended Data Fig. 1e).Together, these results indicate that gut-derived NPF suppresses sugar intake in females.
To attribute these effects specifically to EEC-derived NPF, we measured the expression of NPF in dissected midguts and central nervous systems (CNS; brain and ventral nerve cord, VNC).NPF transcript levels were strongly reduced in the female midgut when EEC> was used to drive knockdown of NPF, whereas expression in the CNS was unaltered, which was confirmed by immunostainings (Fig. 1e,f,g and Extended Data Fig. 1f,g).To further support this, we used a second driver, NPF::2A::GAL4 (NPF>), a CRISPR-mediated insertion of T2A::GAL4 into the native NPF locus that drives GAL4 expression in only NPF-producing cells 25 .Since NPF is expressed in both neurons and EECs, we used pan-neuronal R57C10-GAL80, an optimized nSyb-GAL80 variant that suppresses neuronal GAL4 (ref. 18), to suppress the GAL4 activity of NPF> in the nervous system.We confirmed that this driver combination (R57C10-GAL80; NPF>), referred to hereafter as NPF gut >, efficiently knocks NPF down in the midgut without affecting CNS expression (Fig. 1h and Extended Data Fig. 1h,i).Knockdown of NPF using this gut NPF-specific driver caused a marked increase in intake of sugar+yeast food measured over 30 minutes and in consumption of sugar-only medium measured over 6 hours after 15-hour starvation (Fig. 1i and Extended Data Fig. 1j).Taken together, these data indicate that EEC-specific loss of NPF is responsible for the observed feeding phenotypes and indicate that gut NPF acts as a satiety signal that inhibits sugar consumption.
To examine the ability of NPF to promote satiety, we injected synthetic NPF peptide into circulation in mated females.EEC-specific knockdown of NPF induced hyperphagia, which was blocked by NPF injection (Fig. 1j and Extended Data Fig. 1k).NPF injection did not affect haemolymph sugar levels (Fig. 1k), indicating that the observed feeding effect was not a consequence of alterations in glycaemic levels.Next, we expressed the thermosensitive Transient receptor potential A1 (TrpA1) cation channel 26 in the NPF + EECs to enable induction of NPF release.Incubation at 29 °C, which induces TrpA1-mediated peptide release, inhibited food intake, an effect that was abolished by simultaneous NPF knockdown (Fig. 1l).These NPF-induced changes in food intake were likewise not associated with altered triacylglyceride (TAG) or is structurally similar to the mammalian gastrointestinal tract, and many gut-derived hormones are evolutionarily conserved 15,16 , making Drosophila an attractive model for unravelling the signals by which the gut controls feeding decisions and sex differences in feeding behaviour.Indeed, a great deal has been learned about gut-derived hormonal signalling in this system [17][18][19][20][21] .
Although substantial progress has been made in understanding the gut-hormonal signalling that controls metabolism 18,20,21 , much less is known about how the gut communicates the presence or absence of specific nutrients to adjust food choice, and gut-derived signals that regulate appetite towards specific nutrients have not been described.Here, we show that EECs in the adult female Drosophila gut sense sugar and in response release neuropeptide F (NPF), an ortholog of mammalian neuropeptide Y (NPY) hormones.NPF acts via several routes of tissue crosstalk to suppress sugar appetite and promote intake of protein-rich food in mated females, suggesting that NPF is important for regulation of food choices and prevention of excessive sugar consumption, which has been linked to obesity.

Midgut NPF suppresses sugar intake and energy breakdown
To identify gut-derived hormones and nutrient-sensing mechanisms that regulate feeding, we performed an in vivo RNA-interference screen of secreted factors and receptors in adult Drosophila.We focused on the EECs, which produce a variety of factors that play key roles in the coordination of food intake and metabolism 12,18,20,21 .We examined the effect of adult-restricted, EEC-specific gene knockdown on the sugar-water feeding behaviour of fed males and females (Fig. 1a) using the fly liquid-food interaction counter (FLIC) system, which allows automated monitoring of Drosophila feeding behaviours 22 .We used the driver voilà-GAL4 to target the RNAi effect to the EECs, in combination with ubiquitously expressed temperature-sensitive GAL80 (Tub-GAL80 TS ), together referred to as EEC> hereafter, which allowed us to induce gene silencing only in the adult stage 18,20 .Among our hits was the peptide NPF, knockdown of which increased the feeding time of mated females on sugar-only food while decreasing males' sugar-interaction time (Fig. 1b and Extended Data Fig. 1a).To rule out contributions of the UAS transgene itself to the phenotype, we crossed the UAS-NPF-RNAi KK (NPFi KK ) line to the control w 1118 background.This genotype showed results similar to those seen with the driver control (Extended Data Fig. 1b).This suggests that lack of NPF production in the EECs of mated females enhances their interest in or motivation to feed on sugar.
Knockdown of gut NPF throughout development has recently been associated with increased consumption of food containing both sugar and yeast in virgin female adults 21 .We therefore analyzed whether EEC-derived NPF also regulates intake of sugar+yeast food in mated females.We measured short-term (30-minute) food intake using a dye-consumption assay with standard adult fly food 23 containing both sugar (9%) and yeast (8%).To measure short-term intake, we preconditioned animals by fasting them for 15 hours to increase consumption.We confirmed that, like virgins, mated females with adult-restricted

Article
https://doi.org/10.1038/s42255-022-00672-zcirculating sugar levels (Extended Data Fig. 1l,m), supporting a direct role for gut-derived NPF in governing feeding behaviour, rather than effects of NPF on metabolism that then lead secondarily to altered behaviour.Together, these results indicate that NPF from these EECs is both necessary and sufficient to inhibit food intake and prevent food overconsumption.Feeding behaviours are tightly coordinated with physiology to maintain metabolic balance.Our findings indicate that NPF acts as a satiety signal, which suggests that it should act after a meal.In this scenario NPF would be expected to promote storage and inhibit mobilization of energy.As an indirect measure of energy storage and mobilization, we first assessed animals' starvation resistance.NPF knockdown in the EECs throughout development (voilà> without Tub-GAL80 TS (Fig. 1m)), as well as adult-restricted RNAi (Extended Data Fig. 2a,b,c), led to a decrease in starvation resistance in females but not males, in line with a recent study linking gut NPF to metabolic programs associated with energy storage 21 .Consistent with their shortened starvation survival, we found that females with constitutive (Extended Data Fig. 2d) or adult-restricted (Fig. 1n and Extended Data Fig. 2e) EEC knockdown of NPF showed a decrease in both TAG and glycogen levels, whereas TAG levels were not affected by EEC-specific NPF loss in males.These observations suggest that although NPF does affect metabolism in the adult stage, it also regulates early life history in ways that affect the adult.We found that haemolymph glucose levels increased more after re-feeding in animals with EEC suppression of NPF, consistent with their increased sugar consumption (Extended Data Fig. 2f), and showing that the mechanisms of sugar absorption and transport into circulation are functional.Together our findings indicate that, in addition to the metabolic findings described recently 21 , a main function of EEC-derived NPF in the adult stage is the regulation of feeding, particularly the inhibition of sugar intake in mated females.

Gut NPF suppresses sugar intake and regulates food choice
Our findings suggest that NPF acts as a sugar-satiety signal.To test this hypothesis directly, we examined whether gut NPF affected animals' preference for dietary sugar when they were given the choice between two different sucrose concentrations (1 and 10%).NPF knockdown in the EECs increased feeding and preference for 10% sugar in mated females (Fig. 2a,b,c and Extended Data Fig. 3a,b).These results indicate that NPF is part of a postingestion sugar-sensing mechanism required to decrease sugar appetite.
Given that females increase their preference for protein-rich food after mating to meet the metabolic requirements of egg production 6 , we speculated that gut NPF might be important to reduce sugar appetite and increase intake of protein-rich food in mated females.Postmating sex peptide (SP) signalling within the female induces an increased preference for yeast 6 , and this peptide has also been shown to potentiate NPF release from the midgut 17 .We confirmed that mating induces NPF secretion from the midgut by measuring EEC NPF protein levels.After mating, NPF peptide levels were reduced in the midgut, consistent with increased release (Fig. 2d).Using a luciferase-based CaLexA reporter 27 , in which calcium induces the expression of luciferase, we found that NPF + EECs showed increased calcium-reporter activity after mating in an SP-dependent manner (Fig. 2e).We therefore proposed that gut-derived NPF might be involved in mediating the SP-induced increase in protein consumption in mated females.To test this possibility, we investigated whether NPF affects yeast preference by using a two-choice dye-based consumption assay to measure preference between sugar or protein-rich yeast food 6 .Animals were deprived of protein for 3 days before the experiment by keeping them on sucrose-only food to increase their preference for yeast food 6 , making any reduction in this preference easier to observe.We observed that mating increased control females' preference for yeast food after this treatment (Fig. 2f), consistent with previous findings 6 .However, animals with EEC-specific NPF loss displayed a reduced preference for yeast food that did not significantly increase after mating.This indicates that EEC-derived NPF is required to inhibit sugar intake in mated females, thereby promoting consumption of protein-rich food.Consistent with this, control animals increased their yeast consumption after mating, whereas mating did not significantly increase yeast consumption in females lacking gut NPF that consumed less yeast (Fig. 2g,h).To further test our conjecture that gut NPF is involved in mating-induced yeast consumption, we used a second automated behaviour-monitoring apparatus, the flyPAD 28 , to measure feeding preference in a two-choice assay.Behavioural results obtained with this assay indicate that gut NPF is important for the mating-induced increase in preference for yeast (Fig. 2i).Together, these data indicate that NPF from the midgut is involved in promoting yeast intake in mated females, an effect triggered by SP signalling 6 .Consistent with this notion, females with EEC-specific NPF knockdown mated to SP + males displayed a consumption pattern similar to that of control females mated to SP-mutant males: both consumed more sugar and less protein than control females mated to SP + males (Fig. 3a).Control females mated to SP-mutant males showed a yeast-preference phenotype that was intermediate between those of virgin females and females mated to SP + males, as previously reported 6 .Females with EEC-specific NPF knockdown mated to SP + males displayed lower yeast preference than control females mated to SP + males, and their yeast preference was not significantly different from that of virgin females (Fig. 3b).Thus, females upregulate their protein intake after mating in a partially SP-dependent manner, and our results indicate that NPF from the EECs is involved in mediating this SP-induced shift in food choice, independently of juvenile hormone (Extended Data Fig. 3c), which is known to affect gut remodelling after mating 29 .We therefore rationalized that injection of NPF into virgin females should induce an increase in their yeast preference.As expected, virgin females injected with NPF peptide exhibited an increased preference for dietary yeast (Fig. 3c).
The induction of yeast preference and the stimulation of gut NPF release after mating are both triggered by SP receptor (SPR) activity in reproductive-tract Ppk + neurons 6,17 .We found that mating did not significantly upregulate yeast preference in females with SPR knockdown in the Ppk + neurons (Fig. 3d), confirming that SP/SPR signalling is important for the preference change.However, injection of NPF was still able to increase the yeast preference of these mated females (Fig. 3d), suggesting that gut NPF acts downstream of SP-SPR signalling in mated females to regulate food choice.NPF gut >NPFi TRiP .d, Midgut NPF staining intensity in fed females on a per-cell basis and on a per-gut basis; n = 703 cells from seven guts for virgins, n = 883 cells from seven guts from mated females.e, NPF-cell activity measured in dissected midguts (two midguts per replicate) using a luciferase-based CaLexA calciumreporter system (NPF>LexA::NFAT::VP16; LexAop-luciferase); n = 4 from fed virgins, n = 4 from fed females mated to SP-deficient males (SP 0 /Df(3L)delta130), n = 6 from fed females mated to SP + males.f, Consumption preference for 10% sucrose versus 10% yeast after 3 days of yeast deprivation (3 days on sucrose-only medium) using two-choice dye assay; n = 25   Given these results indicating that NPF acts as a mediator of sugar satiety, we asked whether NPF-producing EECs might be activated by sucrose ingestion.In our initial analysis of a collection of RNAi lines (Fig. 1a), we found that knockdown of sugar transporter 2 (sut2), a member of the glucose-transporter class of solute carrier (SLC) proteins, in the EECs of fed mated females increased their sugar-feeding behaviour (Fig. 4a).Loss of sut2 in NPF + EECs increased animals' sugar-feeding  behaviour and sugar intake, similar to the effects observed in animals with knockdown of NPF itself (Fig. 4b,c and Extended Data Fig. 3d), suggesting that Sut2 might be required as part of a mechanism governing NPF production or release.
In Drosophila and mammals, the sugar-responsive transcription factor Mondo/ChREBP (carbohydrate-responsive-element-binding protein) contributes to many of the cellular responses to sugar 30 .To probe the molecular sugar-sensing mechanisms regulating NPF, we silenced Mondo specifically in NPF + EECs of mated females and found that this manipulation increased both sugar-feeding behaviour and sugar intake, although not as dramatically as the loss of NPF or sut2 (Fig. 4b,c).This suggests that although other mechanisms are probably involved, Mondo/ChREBP-mediated sugar sensing may contribute to NPF regulation in EECs.
Because NPF loss led to increased sugar intake and decreased protein feeding, we investigated whether Sut2 also affects sugar versus protein intake in mated females.We found that knockdown of sut2 in NPF + EECs led to a strongly increased intake of sucrose and a marked decrease in consumption of yeast when animals were presented with these foods separately (Fig. 4d).Similarly, when given a choice between these two foods, mated females with knockdown of sut2 in NPF + EECs displayed a reduced preference for dietary yeast (Fig. 4e).Although Sut1 has been linked to NPF secretion in virgins 21 , we did not observe significant changes in yeast preference in virgins or mated females with

Article
https://doi.org/10.1038/s42255-022-00672-zsut1 knockdown in NPF + EECs (Extended Data Fig. 3e).These results indicate that loss of sut2 in NPF + EECs shifts consumption towards sugar-rich food, similar to knockdown of NPF in these same gut cells, consistent with a role for Sut2 in regulating NPF production or release.Knockdown of sut2 in NPF + EECs led to a strong intracellular accumulation of NPF peptide, even though NPF transcript levels were reduced (Fig. 4f and Extended Data Fig. 3f,g), suggesting that sut2 loss in NPF + EECs leads to NPF retention and thus that Sut2 is required for normal NPF expression and release.We also found that sut2 transcript levels in the entire midgut were strongly reduced by knockdown targeted only at the NPF + gut endocrine cells (Extended Data Fig. 3g), demonstrating that sut2 is predominantly expressed in NPF + EECs.
To assess more directly whether sugar regulates NPF + EECs, we exposed mated females to different nutritional conditions and observed their calcium-signalling history, using the CaLexA reporter system, in which green fluorescent protein (GFP) expression reflects calcium signalling 31 .After 24 hours of starvation, we observed decreased calcium-induced GFP signal and increased NPF peptide staining in the NPF + EECs when measured on a per-cell basis, even though NPF transcript levels were reduced (Fig. 4g-i), indicating NPF retention.Although these measures were higher in starved animals on a per-cell basis (Fig. 4g,h), they were not significantly altered when analyzed on a one-mean-per-gut basis (Extended Data Fig. 3h).This might possibly reflect the inhibition of only a subpopulation of the NPF + EECs by starvation, which could be masked by averaging all the cells, or that starvation longer than 24 hours is required for strong inhibition, as also found by a recent report showing that NPF release is reduced after 48 hours' starvation 21 .Re-feeding with sucrose after starvation elicited a strong increase in calcium signalling within 2 hours in the NPF + EECs, associated with a decrease in NPF peptide staining within 6 hours of re-feeding (Fig. 4j,k and Extended Data Fig. 3i), as also reported by an independent study 21 .Since midgut NPF transcript levels were unaltered under these conditions (Fig. 4l), these results indicate that midgut NPF + cells are activated by dietary sugar, leading to their secretion of NPF peptide.
We then used genetic methods to mimic sugar sensing occurring in the EECs following a meal.We first induced RNAi-mediated silencing of Mondo/ChREBP by switching flies to 29 °C to inactivate GAL80 TS for 15 hours, to reduce sugar sensing in the EECs.We then reactivated Mondo/ChREBP signalling, to mimic sugar-sensing occurring after a meal, by switching animals from 29 back to 18 °C to renature GAL80 TS and thereby inactivate the RNAi effect.Reactivation of Mondo/ChREBP signalling in the EECs caused a decrease in NPF peptide levels in these cells without altering NPF expression (Fig. 4m  and Extended Data Fig. 3j,k), indicating increased NPF secretion, consistent with the notion that sugar sensing in the EECs is associated with NPF release.Taken together, our findings indicate that, in mated females, sugar intake leads to EEC NPF release through a process requiring glucose-transporter-family protein Sut2 and involving Mondo/ ChREBP-mediated sugar sensing.

NPF suppresses energy mobilization in adipose tissues
We then asked which target tissues might be involved in NPF-mediated appetite regulation by knocking down the NPF receptor (NPFR).
Whereas EEC-specific NPF knockdown induced overfeeding (Fig. 1d), adult females with neuronal knockdown of NPFR driven by elav-GAL4 (elav>) showed decreased food intake (Fig. 5a), consistent with previous reports that neuronal NPF/NPFR signalling promotes feeding 32 and indicating that other tissues mediate the downregulation of sugar intake induced by gut NPF.Global knockdown of NPFR driven by daughterless-GAL4 (da>) led to a feeding phenotype intermediate between those observed with gut or neuronal NPF signalling loss (Fig. 5a), probably reflecting opposing effects on feeding of NPF signalling within different organs.
To assess receptor expression in other target tissues, we used a CRISPR-mediated knock-in of T2A::GAL4 into the native NPFR locus (NPFR::T2A::GAL4, hereafter NPFR>) 33 to express UAS-mCD8::GFP.We observed reporter expression in the fat body (Fig. 5b), a tissue analogous to adipose tissue and liver in mammals.Although fat-body-specific NPFR knockdown driven by Cg-GAL4 (Cg>) in adult females did lead to increased short-term intake of food containing both sugar and yeast, it did not increase preference for sugar (Fig. 5c-e and Extended Data Fig. 4a).Indeed, suppression of NPFR in the fat body led to decreased sucrose intake and sugar-feeding behaviour (Fig. 5f,g).Thus, fat-body NPFR signalling does not appear to underlie the specific feeding phenotypes observed with gut NPF loss.We next asked whether fat-body NPFR mediates the effects of gut NPF on metabolism.Like animals with gut-specific NPF knockdown, fat-body NPFR knockdown animals were more sensitive to starvation and displayed metabolic phenotypes similar to those seen with loss of gut NPF (Fig. 5h and Extended Data Fig. 4b,c).These findings indicate that gut-derived NPF acts on NPFR in the fat body as part of a metabolic pathway that maintains energy homeostasis.

NPF regulates food choice through glucagon-like signalling
Our experiments indicate that gut NPF signalling regulates sugar appetite via tissues other than the CNS and fat body.In Drosophila, the brain cells that produce insulin express NPFR 21 and these cells also regulate aspects of feeding and satiety 34 .However, knockdown of NPFR in the insulin-producing cells (IPCs) did not change preference for yeast versus sugar in mated females (Extended Data Fig. 5a), suggesting that gut-derived NPF does not act through insulin to modulate preference for dietary sugar and protein.
To identify the tissue mediating this effect, we examined NPFR expression in other tissues, which revealed expression of the receptor in the cells producing the glucagon-like factor AKH (Fig. 6a).AKH is released from the AKH-producing cells (APCs) during starvation and acts through its receptor, AkhR, on the fat body to promote the mobilization of stored energy, and it is also thought to act as a hunger signal to drive feeding behaviours 18,31,[35][36][37] .However, whether AKH regulates sugar-or protein-specific feeding is unknown.We proposed that gut-derived NPF, released in response to sugar feeding, might suppress AKH release from the APCs in the fed state.Consistent with a recent report 21 , we found that knocking down NPFR in the APCs using AKH-GAL4 (AKH>) resulted in decreased AKH peptide levels within these cells in fed mated females (Fig. 6b), indicating that NPFR is required to suppress AKH release when the animal has ingested food.AKH promotes lipid and glycogen breakdown, and we therefore tested whether NPFR activity in the APCs regulates metabolism.As with knockdown of NPF in the midgut, adult females with knockdown of NPFR in the APCs showed reduced TAG and glycogen levels and increased susceptibility to starvation, as also reported recently 21 and consistent with an increase in AKH signalling (Fig. 6c,d).
We recently reported that the peptide hormone AstC released by midgut EECs promotes AKH release during starvation conditions 18 .We wondered whether NPF signalling might also inhibit AstC release from the midgut to suppress the activation of the AKH axis at multiple hierarchical levels.We found expression of NPFR reporter in AstC + EECs (Fig. 6e), consistent with single-cell RNA-sequencing data 38 .We silenced NPFR expression specifically in AstC + EECs using AstC-GAL4 (AstC>) with pan-neuronal GAL80 (R57C10-GAL80, AstC>-AstC gut > hereafter) to suppress nervous system GAL4 activity.This did not alter AstC expression, but it did lead to a reduction in the number of cells containing detectable AstC peptide (Fig. 6f,g), suggesting that EEC loss of NPFR cell-autonomously increases AstC release.Knockdown of NPFR with AstC gut > led to an increased number of active midgut AstC cells and to higher overall calcium-reporter activity (Fig. 6h-k), indicating that NPFR knockdown promotes AstC + EEC activation.This indicates that the decreased number of AstC immune-positive cells in adult females with NPFR knockdown in the AstC + EECs (Fig. 6f) is due to increased release of AstC peptide, which would be expected to promote AKH release, leading to more rapid depletion of energy stores and therefore reduced capacity to survive starvation 18 .Consistent with this, NPFR knockdown targeted to the AstC + EECs led to a clear reduction in the capacity of these animals to survive starvation (Fig. 6l).Taken together, our results indicate that in response to sugar intake, gut-derived NPF inhibits the AKH axis at three levels: blocking the release of adipokineticotropic AstC from midgut EECs, blocking the release of AKH itself from the APCs and counteracting AKH's effects on the fat body.
NPFR knockdown in the APCs was recently linked to increased consumption of food in virgin females 21 , an effect we confirmed in mated females (Fig. 7a).Next, we tested whether NPF regulates sugar-versus protein-specific feeding through NPFR in the APCs.Mated females with APC-specific NPFR knockdown exhibited elevated sugar-directed feeding behaviour, sugar consumption and preference for dietary sugar (Fig. 7b-e and Extended Data Fig. 5b-e), similar to animals with loss of NPF in the midgut.Next, we examined whether NPFR in the APCs might be involved in promoting yeast preference in mated females.Indeed, whereas APC knockdown of NPFR in virgin females did not detectably alter feeding preference, this manipulation had a strong effect in mated females (Fig. 7f and Extended Data Fig. 5f), similar to animals with knockdown of NPF in the gut, indicating that NPFR in the APCs is an important element for promoting consumption of protein-rich food in mated females.To determine whether AKH mediates the effects of NPFR loss on feeding, we examined the ability of AKH knockdown to rescue NPFR-RNAi's sugar-overeating phenotype.We found that knockdown of AKH completely abolished this effect (Fig. 7g), suggesting that AKH is the primary factor mediating the feeding effects of NPFR signalling.Consistent with NPFR's suppression of AKH release, AKH loss and NPFR knockdown induced opposite effects on sugar intake and yeast consumption (Fig. 7h,i).Together, our results indicate that, in mated females, gut-derived NPF acts on the APCs via NPFR to inhibit AKH release after a sugar-rich meal, to suppress further sugar feeding while promoting protein intake.

AKH regulates appetites for sugar and protein-rich food
AKH is described as a generic hunger hormone released during nutritional deprivation 18,35,39 .However, our findings indicate that AKH regulates food choice.To confirm this effect, we examined the feeding behaviour of mated AKH mutant females and found that these animals exhibited significantly reduced sugar intake (Fig. 8a), suggesting that AKH promotes sugar preference.On this basis, this we expected an increase in yeast preference with loss of AKH, so we used assay conditions in which animals do not normally exhibit strong yeast preference:

Article
https://doi.org/10.1038/s42255-022-00672-z 15-hour starvation rather than 3-day yeast deprivation.Loss of AKH, including adult-restricted APC-specific knockdown (with GAL80 TS ; AKH>, together referred to as AKH ts >), led to a striking shift in preference towards yeast using the two-choice dye assay in mated females (Fig. 8b).Consistent with their increased intake of and preference for yeast food, mated AKH mutant females displayed a strong increase in the amount of time spent exploring patches of yeast food compared to sugar patches (Fig. 8c).This further supports a role for AKH in controlling feeding decisions, biasing behaviour towards sugar intake.Consistent with this, activation of the APCs to induce AKH release caused

Article
https://doi.org/10.1038/s42255-022-00672-zincreased sugar intake while decreasing yeast intake, effects that were blocked by simultaneous AKH knockdown (Fig. 8d,e), indicating that they were mediated by AKH.Circulating sugar levels and whole-body TAG levels were not altered (Extended Data Fig. 6a,b), suggesting that the observed AKH-induced feeding phenotypes are direct effects that precede detectable metabolic changes.Together, these findings indicate that AKH is a hormone that controls selective feeding decisions by increasing appetite for sugar and reducing intake of protein food.
AKH has recently emerged as a key factor in sex-specific metabolic regulation 40 .Unlike mated females, in which AKH loss enhanced yeast intake, males lacking AKH exhibited a decrease in yeast intake, indicating that AKH plays a sexually dimorphic role in feeding decisions (Fig. 8f and Extended Data Fig. 6c,d).Together these findings indicate that, in mated females, gut-derived NPF inhibits AKH secretion, and this inhibition suppresses sugar appetite and increases the consumption of protein-rich food.In this scenario, increased AKH signalling in virgins promotes preference for dietary sugar.We therefore examined whether mating reduces AKH release.Mating increased AKH peptide levels and reduced the calcium activity of the APCs, indicating repression of AKH release, and the effect on AKH was dependent on SP signalling (Fig. 8g).These findings suggest that SP signalling, through activation of gut NPF, contributes to the suppression of AKH release after mating.Injecting NPF peptide into virgin females increased AKH peptide levels within the APCs and reduced their calcium activity, indicating that NPF is sufficient to repress AKH secretion (Fig. 8g).Mating did not lead to AKH retention in females with knockdown of SPR in the Ppk + neurons (Extended Data Fig. 6e), further indicating that SP-SPR signalling is required to repress AKH signalling after mating.To test whether NPF functions downstream of SP-SPR signalling in the AKH-regulatory hierarchy, we injected NPF into mated females with reduced SP-SPR signalling (ppk>SPR-RNAi) and found that this led to increased AKH peptide levels in the APCs (Fig. 8h).This suggests that NPF is sufficient downstream of the SP pathway to repress AKH signalling after mating, which increases yeast intake.We therefore conjectured that loss of AKH would increase yeast intake in virgin females, whereas increasing AKH signalling in virgin females would have little effect on feeding behaviour.To assess the effect of loss of AKH, we examined behaviour under starved conditions, under which AKH signalling is normally high.As expected, we found that while control females increased their yeast consumption in response to mating, similarly conditioned AKH mutant virgins displayed a striking overconsumption of yeast food that was not significantly altered by mating (Fig. 8i).Activation of the APCs to induce AKH release in the fed state, in which AKH signalling is generally lower, did not alter yeast intake in fed virgin females (Extended Data Fig. 6f), presumably because of the already higher AKH signalling in the virgin state, whereas as mentioned above this treatment did reduce the yeast intake of fed mated females (Fig. 8e).Consistent with these findings, virgin AKH mutant females exhibited a strong preference for yeast food that did not increase significantly in response to mating as it did in control females (Fig. 8j,k and Extended Data Fig. 6g).Together, this suggests that, in mated females, NPF acts through NPFR in the APCs to repress AKH signalling, which increases yeast intake.Consistent with this, we found that inducing AKH release was sufficient to block the high yeast preference exhibited by mated females after 3 days of yeast deprivation (Fig. 8l).To demonstrate that NPF regulates food choice via AKH, we injected NPF into mated females with APC-specific NPFR knockdown.This manipulation did not increase yeast preference after 3 days' yeast deprivation (Extended Data Fig. 6h), suggesting that NPFR in the APCs is required for NPF to promote yeast preference.Furthermore, injection of NPF into AKH mutant virgin females also did not alter sugar versus yeast preference, indicating that AKH is required for mediating the effects of NPF-NPFR on food preference (Extended Data Fig. 6i).We propose that sugar-induced AKH-repressive NPF signalling from the gut constitutes a hormonal axis involved in suppressing sugar appetite and promoting intake of protein-rich food in mated females (Fig. 8m).

Discussion
To maintain nutritional homeostasis, animals need to match their ingestion of specific nutrients to their needs.This is achieved by modulating appetite towards the specific nutrients needed.A number of factors, including gut hormones, that regulate food consumption have been identified in both flies and mammals, and reports have also described central brain mechanisms that induce ingestion of protein food in response to amino-acid deprivation, that sense amino acids and promote food consumption and that reject food lacking essential amino acids 1,6,12,41,42 .However, little is known about the hormonal mechanisms that regulate nutrient-specific appetite, and gut hormones that regulate selective food intake are completely unknown.Our findings indicate that, in mated female Drosophila, gut-derived NPF is a selective driver of sugar satiety and protein consumption, providing a basis for understanding these mechanisms.Hormone-based therapies that inhibit appetite offer promising new directions for weight-loss treatment 14 .For example, Fibroblast growth factor 21 (FGF21) is a liver-derived hormone that promotes protein consumption, and it is emerging as a promising target for metabolic disorders 43 .Uncovering appetite-regulatory hormones such as gut-derived NPF that specifically inhibit sugar consumption while promoting the intake of protein-rich foods could provide effective new weight-management strategies by promoting healthier food choices.
The SLC2-family sugar transporter Sut2 is the closest Drosophila homologue of human SLC2A7 (GLUT7), a transporter expressed mainly in the intestine whose function is poorly defined 44 .In flies, GLUT1 is important for Bursicon secretion from the EECs, and Sut1, another SLC2-family sugar transporter protein, was recently shown to be involved in midgut NPF release in virgin females 20,21 .Our results implicate Sut2 in the release of NPF from EECs in mated females and thus link it to the mechanism by which NPF-mediated gut signalling  Article https://doi.org/10.1038/s42255-022-00672-zintestinal glucose-sensing mechanisms therefore has become a focus of weight-management therapies because of its potential in regulating appetite and incretin effects 46 .Future studies should investigate whether GLUT7, like its Drosophila homologue Sut2, affects appetite-regulatory mechanisms in the mammalian gut.
NPF is orthologous with the mammalian NPY family of gut-brain peptides, including peptide YY (PYY), pancreatic polypeptide and NPY itself, that regulate food-seeking behaviours and metabolism 47,48 .Like mammalian NPY-family hormones, Drosophila NPF is expressed in both the nervous system and the gut.While NPY is abundant in the nervous system and, like brain NPF, promotes food intake, PYY is mainly produced by endocrine cells of the gut as a satiety factor.Gut-expressed PYY is homologous to NPY, and both act through specific G-protein coupled receptors, called NPY receptors (NPYRs), that are orthologous with Drosophila NPFR 49 .Thus, in mammals, multiple NPY-family peptides from different tissues sources exert their functions on target organs through several related NPYRs, while in Drosophila, these functions may be regulated through the single peptide-receptor pair of NPF and NPFR 47 .
Our results indicate that gut-derived Drosophila NPF fulfils the function of mammalian PYY.PYY is produced by the endocrine l-cells of the gut, which, like the EECs of Drosophila, produce a context-dependent combination of multiple hormones 49 .The physiological role of PYY in feeding regulation has been difficult to clarify, but it is believed to act through different NPYRs on tissues including the hypothalamus and the pancreatic islets to suppress appetite.Our findings show that, in flies, NPF injection strongly reduces the intake of sugar-containing food and promotes the ingestion of protein-rich food.In humans, PYY infusion also been shown to strongly reduce food intake.Although the satiety function of human PYY has made it a prime therapeutic target for potential weight management, it is not clear whether PYY regulates nutrient-specific appetite, which would be important from a therapeutic perspective.Our results indicate that Drosophila gut NPF, perhaps filling the role of mammalian gut PYY, acts to mediate sugar-specific satiety, illustrating a key hormonal mechanism that underlies selective hunger by which animals adjust their intake of specific nutrients.
Feeding decisions are based on internal state and exhibit sexual dimorphism.In Drosophila, males and females differ in their preference for and intake of dietary sugar and protein 6 .Our findings define a complex interorgan communication system through which mating influences food choices in females.We have found that midgut NPF is involved in mediating SP-induced postmating responses in females, inhibiting sugar appetite and promoting the ingestion of protein-rich yeast food, and we have further shown that AKH is required for mediating the effects of NPF.When mated females consume dietary carbohydrates, NPF is released from the EECs and inhibits the AKH axis by directly suppressing AKH release from the APCs as well as by inhibiting the release of midgut AstC, a factor that stimulates AKH secretion 18 .Furthermore, NPF acts directly on the fat body through NPFR to inhibit energy mobilization, thereby antagonizing AKH-mediated signalling in the adipose tissue.Likewise, mammalian NPY-family peptides also regulate metabolism by direct actions on adipose tissue via NPYR 50 .Although a number of studies have demonstrated that AKH is a regulator of metabolism (reviewed in ref. 2 ), our findings uncover a key role of AKH in governing nutrient-specific feeding decisions.It is becoming clear that the APCs integrate many signals that affect AKH rele ase 18,20,21,36,51,52 , and these signals may therefore also affect food choice.The APCs therefore seem to function as a signal-integration hub, similar to the IPCs, which receive many different inputs to control insulin production and release.AstC, Bursicon and NPF from the gut control AKH expression and secretion, indicating that multiple signals, even from the same organ, converge on the APCs.These signals presumably convey different aspects of nutritional status and may act with different dynamics to regulate AKH production and/or release, or even in a redundant manner to regulate AKH signalling.Likewise, many signals released from the fat body convey similar and seemingly redundant nutritional information to the IPCs 2,53,54 .
Recent work has also revealed a sex-specific role of AKH, with lower activity in females underlying differences in male and female metabolism 40 .Consistent with this notion, our results indicate that in mated females the midgut NPF system inhibits AKH signalling, suppressing intake of sugar-rich food.Furthermore, we recently showed that in mated females, midgut-derived AstC acts in a sex-specific manner through AKH to coordinate metabolism and food intake under nutritional stress 18 .Our work here shows that NPF also works sex-specifically to sustain physiological requirements in mated females by signalling from the gut to control AKH, suggesting that the gut-AKH axis occupies a central link in the hormonal relays underlying sex-specific regulation of physiology.A recent report showed that female germline cells modulate sugar appetite, but this effect is not induced by mating and does not affect yeast feeding 55 as we have found here for gut NPF and AKH, suggesting that it is an independent mechanism.
How nutrient signals from the gut modulate feeding is key to understanding how nutritional needs are translated into specific feeding actions to maintain balance.We have identified a homeostatic circuit triggered by gut-derived NPF that limits sugar consumption.Similar mechanisms for sugar-induced satiety that promote protein consumption may also enable mammals to balance their intake of different nutrients with their metabolic needs.Explaining how nutrient-responsive gut hormones such as NPF affect dietary choice is important to better understand hunger and cravings for specific nutrients that may ultimately lead to obesity.

Article
https://doi.org/10.1038/s42255-022-00672-zfrom A. von Philipsborn (Aarhus University).UAS-LexA::VP16::NFAT; LexAop-luciferase was a kind gift from M. Rosbash (Brandeis University).The fly lines used are listed in Supplementary Table 1.No ethical approval is needed for the use of the fruit fly Drosophila.For standardizing the genetic background and generating controls with proper genetic background, all GAL4 lines and GAL80 lines used this study were backcrossed for several generations to the same w 1118 genetic background population before they were used in a final outcross with the genetic background of the RNAi lines and used as controls 18 .

Starvation-survival assay
Flies were transferred without anaesthesia to vials containing starvation medium (1% agar in water) and kept either at 29 or 25 °C, depending on whether they carried GAL80 TS .Forty to 150 animals, at 10-15 flies per vial, were assayed for each genotype/sex.Dead animals were counted every 4-8 h.The statistical significance of survival differences was determined by using the Kaplan-Meier log-rank survival test or Gehan-Breslow-Wilcoxon survival test in the Prism software package (GraphPad v.9).

Feeding assays
Short-term food consumption was measured by using a spectrophotometric dye-feeding assay 65,66 , and all food intake experiments were performed during the time when animals have their morning meal (1 h after lights on; 12/12 h light/dark cycle).During the morning meal (after lights on), flies were transferred without anaesthesia to adult-optimized food containing 0.5% erioglaucine dye (brilliant blue R, FD&C Blue No. 1, Sigma-Aldrich, no.861146) and allowed to feed for 30 min, if the flies had been 15 h starved to stimulate food intake or for 2-3 h if not.Another set of flies was fed with undyed food to measure the baseline absorbance of fly lysates.For two-choice assays, the protocol of Ribeiro and Dickson 6 was used with some modifications.Briefly, 25 flies were lightly anaesthetized with CO 2 before being placed into a 60-mm Petri dish with a chequerboard array of 20-μl patches of alternative diets containing either 100 g l −1 of sucrose and dyed red with 0.5% amaranth (Sigma no.A1016), or 100 g l −1 yeast (dyed with 0.5% erioglaucine) and allowed to eat for 2 h in the dark.For each genotype, 10-25 samples of 1-2 flies each were homogenized in 100 μl of phosphate buffer, pH 7.5, using a TissueLyser LT (Qiagen) bead mill with 5-mm stainless-steel beads (Qiagen, no.69989).Homogenates were centrifuged at 16,000g for 5 min and 50 μl of each cleared supernatant was loaded into a 384-well plate.Sample absorbance was measured at 520 nm (amaranth) and at 629 nm (erioglaucine) on an Ensight multimode plate reader (PerkinElmer).Standard curves for erioglaucine and amaranth were used to convert absorbance values to food consumption amounts.
Long-term food intake was monitored using the CAFÉ capillary-feeding assay 24 .For one-choice consumption assays, assay tubes were constructed by inserting a 5-μl microcapillary (Hirschmann) through a hole in the lid of a 2-ml Eppendorf tube.The capillary was filled with a liquid sugar or yeast-extract medium 24 containing 100 g l −1 sucrose or 100 g l −1 yeast extract, with 1 ml l −1 propionic acid and 1 g l −1 methyl-4-hydroxybenzoate preservatives, before the start of the experiment.For sugar-preference assays, two capillaries were inserted into each tube, one filled with 10 g l −1 sucrose solution and the other filled with 100 g l −1 sucrose solution.Individual flies were briefly anaesthetized on ice and placed into assay tubes, and the tubes were placed inside a moist chamber within a standard fly incubator.The level of the meniscus in each tube was measured at intervals.Tubes containing no flies were used as controls for evaporation; the amount of meniscus movement in these tubes was subtracted from the other measurements.
To monitor feeding behaviour, interactions with food were measured over a 20-24 h period using the FLIC assay 22 .Drosophila feeding monitors (DFMs) (Sable Systems) were installed in an incubator (25 or 29 °C if GAL80 TS was present; 70% humidity, 12/12-h light/dark cycle).Feeding wells were filled with a 10% sucrose solution, and individual flies were placed in each of the 12 chambers of the DFMs in the afternoon (after the morning meal) and left to acclimate for several hours, after which evening feeding data were recorded.The next morning, at lights on, fresh sugar solution was added to the DFMs and morning meal data were recorded.In two-choice sugar-preference experiments, half of the DFM wells were filled with 1% sugar solution and the other half with 10% sugar solution.The data were recorded using the manufacturer's software and analyzed in R Studio using the published package, available at https://github.com/PletcherLab/FLIC_R_Code.
For flyPAD 28 two-choice behavioural experiments, 4-7-day-old female flies, mated or virgin, were either starved for 15 h on 1% agar (to establish a low yeast-preference baseline for experiments in which a manipulation was anticipated to increase this preference) or yeast-deprived for 3 days by keeping them on medium containing 10% sucrose and 1% agar (to establish a higher baseline yeast preference, for experiments in which yeast preference was expected to decrease).Flies were briefly immobilized on ice and transferred with a brush into flyPAD behavioural arenas.They were left to acclimate for several minutes before data acquisition was started.The flies were allowed to choose between food droplets containing 1% agarose and either 10% sucrose or 10% yeast.The food was aliquoted into 1.5-ml tubes and kept frozen at −20 °C.Before each experiment, the tubes were placed into a heat block for melting at 90 °C and 3 μl of food was loaded into each well.A package created by the developer for the Bonsai data-stream processing program was used to acquire the data.Data processing was done by using the developer's software, which is available at http:// www.flypad.pt.

Video recording of feeding behaviour
Behaviour chambers (40 mm in diameter) were coated with fluon on the top and sides to prevent flies' walking on these surfaces.Fifteen-microlitre patches of either 10% sucrose or 10% yeast (with no dyes) were placed in a circular pattern within the arena.Twelve 15-hour-starved animals per genotype were introduced into the chamber and were allowed to acclimate in darkness for a few minutes.The behaviour chambers were placed on an infrared-light transilluminator viewed by a Basler camera, and half-hour videos were recorded at 15 Hz using an imaging setup described elsewhere 67 .Flies were tracked using the Ctrax software 68 , and locomotion data were analyzed using custom MATLAB code (Code availability statement).

Immunohistochemistry and confocal imaging
Adult midguts, CNSs and fat bodies were dissected in cold PBS and fixed for 1 h in 4% paraformaldehyde at room temperature with agitation.Anterior parts of guts, containing APCs (CCs) were first fixed for 30 min, finely dissected and fixed for a further 30 min.Fixed tissues were quickly rinsed once with PBST (PBS with 0.1% Triton X-100, Merck no.12298) and washed in PBST three times for 15 min each.Washed tissues were incubated in blocking solution (PBST containing 5% normal goat serum (Sigma)) for 30 minutes at room temperature and incubated with primary antibodies diluted in blocking solution overnight (or 2 d for CNS samples) at 4 °C with gentle agitation.Primary-antibody solution was removed, and the tissues were rinsed once and washed three times, 20 minutes each, with PBST.Tissues were incubated with secondary antibodies diluted in PBST overnight at 4 °C, washed three times with PBST and mounted in Vectashield mounting medium containing 4,6-diamidino-2-phenylindole (Vector Laboratories, no. H-1200) on slides treated with poly-l-lysine (Sigma, no.P8920).Tissues were scanned on a Zeiss LSM-900 confocal microscope using a 20× air objective using the Zen software package.Image analysis was carried out using the open-source program ImageJ 69 .For quantification of NPF, AstC, AKH and GFP (CaLexA reporter) staining intensity, samples to be compared were stained simultaneously using the same reagent preparations and imaged with the same settings.Relevant Article https://doi.org/10.1038/s42255-022-00672-zmidgut regions (the entire NPF-expressing region or the AstC-positive region) were tiled at 20× with 10-20 Z-stacks of at least 100 planes separated by 1 μm.Tiled stacks were stitched into a single large stack for each gut using the Stitching function of Zeiss Zen Blue v.3.1.For quantification of NPF and CaLexA staining, a binary mask containing identified cells was created using local thresholding in ImageJ.This mask was manually curated in ImageJ by comparison with the raw image data, and incorrectly joined cells were manually resegmented.In a custom MATLAB script (Code availability statement), this mask was applied to the image data to segment out each cell.Staining intensity within each cell was summed, and the local background of each cell was removed by measuring the signal around the circumference of each cell.For AstC quantifications, stacks were Z-projected using the sum method.Cells were manually segmented, and their intensity was measured using ImageJ with local background subtraction.We integrated a UAS-tdTomato transgene into the CaLexA system to normalize calcium-dependent GFP fluorescence.Antibodies used included a rabbit antibody against the processed AKH peptide 37 , a kind gift of J. Park, University of Tennessee, 1:500; rabbit anti-NPF (Ray BioTech, no.RB-19-0001-20), 1:500; rabbit anti-AstC 70 , kindly given by J. Veenstra, University of Bordeaux and M. Zandawala, Brown University, 1:500; mouse anti-GFP (ThermoFisher, no.A11120), 1:500; rat anti-mCherry (used against tdTomato; ThermoFisher, no.M11217), 1:2,000; mouse anti-Prospero (University of Iowa Developmental Studies Hybridoma Bank, no.MR1A), 1:20; Alexa Fluor 488-conjugated goat anti-mouse (ThermoFisher, no.A32723), 1:500; Alexa Fluor 555-conjugated goat anti-rabbit (ThermoFisher, no.A32732), 1:500; Alexa Fluor 555-conjugated goat anti-rat (ThermoFisher, no.A21434), 1:500 and Alexa Fluor 405-conjugated goat anti-rabbit (Ther-moFisher, no.31556), 1:500.

Injection and methoprene-treatment experiments
Synthetic amidated NPF peptide (SNSRPPRKNDVNTMA-DAYKFLQDLDTYYGDRARVRFamide) was a kind gift from F. Hauser (University of Copenhagen).Peptide was dissolved at 25 μM in a synthetic haemolymph-like buffer 71 containing 5 mM glucose, 5 mM trehalose and 110 mM sucrose (inert osmolyte).Flies were reared at 18 °C as described above.After 4 days at 29 °C to permit GAL4 activity, flies were starved on 1% agar for 15 h or protein-deprived on sugar-agar for 3 days.Flies were immobilized on ice and 50 nl of haemolymph-like solution with or without NPF was injected into each animal at the lateral mid-thorax ventral to the wings using a Nanoject II injector (Drummond Scientific).Assuming each injected animal contained 1 μl of haemolymph, the final NPF concentration in the injected animals was increased by 1.25 μM, a level that should strongly activate NPFR (IC 50 roughly 60 nM, ref. 72 ).Animals were allowed to recover for 30 min at 29 °C before use in dye-feeding assays as described above.
Because the juvenile hormone analogue methoprene (Sigma-Aldrich no.33375) is not thermally stable, the working solution (0.01 μg μl −1 in acetone) or vehicle (pure acetone) was applied to the surface of cooled, solidified fly medium instead of being mixed into melted medium before solidification.We applied 32 μl of hormone or vehicle solution to the surface of 2 ml of fly medium in 25-mm diameter plastic vials (in total, 0.32 μg per vial), an amount that should be effective while also being well tolerated by the animals 73 .Treated media were kept at room temperature for roughly 12 h to allow acetone evaporation before flies were added.

Metabolite measurements
Triglyceride and glycogen levels were measured using established protocols 23,74 .For each genotype, ten batches of three flies each were homogenized in PBS containing 0.05% Tween-20 (Sigma no.1379) in a TissueLyser LT (Qiagen) bead mill with 5-mm stainless-steel beads.Glycogen was measured by hydrolysing glycogen into glucose by using amyloglucosidase (Sigma, no.A7420) followed by colorimetric glucose measurement (Sigma, no.GAGO20).TAG levels were assayed by cleaving their ester bonds using Triglyceride Reagent (Sigma, no.T2449) to obtain free glycerol, the level of which was then colorimetrically measured using the Free Glycerol Reagent (Sigma, no.F6428).For determination of circulating glucose concentration, haemolymph was extracted as described previously 23 and glucose was measured using the colorimetric assay (Sigma, no.GAGO20).Each sample's absorbance at 540 nm was measured in a 384-well plate using an Ensight multimode plate reader (PerkinElmer) and converted to metabolite concentrations using glycerol and glucose standard curves.Measurements are reported on a per-fly basis.

Luciferase assay
Female guts were dissected into lysis buffer (Promega, no.E2920).For each condition, 4-7 replicates with two guts in each were homogenized in 50 μl of lysis buffer in 2-ml round-bottomed Eppendorf tubes using a TissueLyser LT (Qiagen) bead mill with 5-mm stainless-steel beads (Qiagen, no.69989).Homogenates were centrifuged at 21,000g for 5 min, and the supernatant was transferred into new tubes and centrifuged a second time.Ten microlitres of the cleared supernatant were loaded into a 384-well plate, and 10 μl of Dual Glo Stop & Glo Reagent (Promega) was added.The plate was left to incubate for 15 min at room temperature to allow for the reaction to pass from the burst phase into the glow phase, after which luciferase activity was measured using the luminescence mode of an Ensight multimode plate reader (PerkinElmer).

Transcript measurement using quantitative PCR
Six tissue replicates (each containing five CNSs, five midguts or five CC-containing anterior parts of guts) per condition or genotype were homogenized in 2-ml Eppendorf tubes containing lysis buffer with 1% beta-mercaptoethanol using a TissueLyser LT bead mill (Qiagen) and 5-mm stainless-steel beads (Qiagen no.69989).RNA purification was performed using the NucleoSpin RNA kit (Macherey-Nagel, no.740955) according to the manufacturer's instructions.Complementary DNA was synthesized using the High-Capacity cDNA Synthesis kit (Applied Biosystems, no.4368814).Quantitative PCR was done using RealQ Plus 2× Master Mix Green (Ampliqon, no.A324402) on a QuantStudio 5 (Applied Biosystems) machine.Results were normalized against the housekeeping gene Rp49 using the delta-delta-Ct method.The oligos used are listed in Supplementary Table 2.

Statistics
All statistics were computed using the Prism analysis package (Graph-Pad v.9).Starvation-survival curves were analyzed using Kaplan-Meier log-rank tests or Gehan-Breslow-Wilcoxon test.Other data were assessed for normality before comparisons were performed.For normally distributed data, pairwise comparisons were made using two-tailed unpaired Student's t-tests and multiple samples were compared using one-way analysis of variance (ANOVA) with post hoc multiple-comparisons tests.Other data were compared using two-tailed unpaired Mann-Whitney U-tests or one-way Kruskal-Wallis ANOVA followed by multiple-comparisons tests.Bar plots show the mean plus or minus the standard error of the mean (s.e.m.).Box plots that show the median and the first and third quartile, with whiskers indicating the full range of values.No data were excluded.Sample size was chosen on the basis of similar previously published studies of Drosophila behaviour and metabolism 17,18,20,55 .No sample-size calculations were performed.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material.If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.