The Neuropeptide Allatostatin A Regulates Metabolism and Feeding Decisions in Drosophila

Coordinating metabolism and feeding is important to avoid obesity and metabolic diseases, yet the underlying mechanisms, balancing nutrient intake and metabolic expenditure, are poorly understood. Several mechanisms controlling these processes are conserved in Drosophila, where homeostasis and energy mobilization are regulated by the glucagon-related adipokinetic hormone (AKH) and the Drosophila insulin-like peptides (DILPs). Here, we provide evidence that the Drosophila neuropeptide Allatostatin A (AstA) regulates AKH and DILP signaling. The AstA receptor gene, Dar-2, is expressed in both the insulin and AKH producing cells. Silencing of Dar-2 in these cells results in changes in gene expression and physiology associated with reduced DILP and AKH signaling and animals lacking AstA accumulate high lipid levels. This suggests that AstA is regulating the balance between DILP and AKH, believed to be important for the maintenance of nutrient homeostasis in response to changing ratios of dietary sugar and protein. Furthermore, AstA and Dar-2 are regulated differentially by dietary carbohydrates and protein and AstA-neuronal activity modulates feeding choices between these types of nutrients. Our results suggest that AstA is involved in assigning value to these nutrients to coordinate metabolic and feeding decisions, responses that are important to balance food intake according to metabolic needs.

Imbalance between the amount and type of nutrients consumed and metabolized can cause obesity. It is therefore important to understand how animals maintain energy balancing, which is determined by mechanisms that guide feeding decisions according to the internal nutritional status. The fruit fly Drosophila melanogaster has become an important model for studies of feeding and metabolism, as the regulation of metabolic homeostasis is conserved from flies to mammals 1,2 . In Drosophila, hormones similar to insulin and glucagon regulate metabolic programs and nutrient homeostasis. Adipokinetic hormone (AKH) is an important metabolic hormone and considered functionally related to human glucagon and a key regulator of sugar homeostasis 3 . The release of AKH promotes mobilization of stored energy from the fat body, the equivalent of the mammalian liver and adipose tissues [3][4][5] . Neuroendocrine cells in the corpus cardiacum (CC) express and release AKH 3 that binds to the AKH receptor (AKHR), a G-protein coupled receptor (GPCR) expressed mainly in the fat body, and promotes mobilization of stored sugar and fat 6,7 . Insulin and glucagon have opposing effects important to maintain balanced blood glucose levels. The Drosophila genome contains 7 genes coding for insulin-like peptides (DILPs), called dilp1-7 8 , which are homologous to the mammalian insulin and insulin-like growth factors (IGFs). The seven DILPs are believed to act through one ortholog of the human insulin receptor that activates conserved intracellular signaling pathways [8][9][10] . The DILPs are important regulators of metabolism, sugar homeostasis 11,12 and cell growth 13,14 . DILP2, 3 and 5 are produced in 14 neurosecretory cells in the brain; the insulin producing cells (IPCs). Genetic ablation of the IPCs results in a diabetic phenotype, increased Scientific RepoRts | 5:11680 | DOi: 10.1038/srep11680 lifespan and reduced growth 12,13 . Because of the growth promoting effects, the activity of the DILPs is tightly linked to dietary amino acid concentrations.
Although metabolism has been extensively studied, the mechanisms that coordinate metabolism and feeding decisions to maintain energy balancing are poorly understood. Neuropeptides are major regulators of behavior and metabolism in mammals and insects 15,16 making them obvious candidates to coordinate these processes. Peptides with a FGL-amide carboxy terminus, called type A allatostatins, have previously been related to feeding and foraging behavior 17,18 . Four Drosophila Allatostatin A (AstA) peptides have been identified 19 that are ligands for two GPCRs, the Drosophila Allatostatin A receptors DAR-1 and DAR-2 20,21 . AstA peptides were originally identified as inhibitors of juvenile hormone (JH) synthesis from the corpora allata (CA) of the cockroach Diploptera punctata 22 . However, recently it was shown that AstA does not regulate JH in Drosophila 23 . Moreover, DAR-1 and DAR-2 are homologs of the mammalian galanin receptors 24,25 , known to be involved in both feeding behavior and metabolic regulation 26 .
We examined the function of AstA in Drosophila in an effort to determine whether it is involved in the neuroendocrine mechanisms coupling feeding behavior to metabolic pathways that manage energy supplies. Our data suggest that AstA is a modulator of AKH and DILP signaling. Dar-2 is expressed in both the IPCs and the AKH producing cells (APCs) of the CC. Silencing of Dar-2 in the APCs or IPCs resulted in changes in expression of genes associated with reduced AKH or DILP signaling, respectively. Moreover, loss of AstA is associated with increased fat body lipid levels, resembling the phenotype of mutants in the DILP and AKH pathways. We also investigated the connection between nutrients and AstA signaling, and found that AstA and Dar-2 are regulated differently in response to dietary carbohydrates and protein, and that activation of AstA-neurons increases the preference for a protein rich diet, while AstA loss enhances sugar consumption. Our results suggest that AstA is a key coordinator of metabolism and feeding behavior.

Dar-2 is expressed in the APCs and IPCs.
To investigate the functional role of AstA, we examined the expression of AstA and its receptor Dar-2 in Drosophila. Immunostaining of 3 rd instar larvae, using a DAR-2 antibody, revealed that Dar-2 is expressed specifically in a small population of cells at the base of the ring gland, corresponding to the location of the CC, like previously reported 27 . To confirm that the expression was specific for the APCs, we used the APC-specific Akh-Gal4 (Akh> ) driver to express UAS-GFP (GFP) and show that anti-DAR-2 labeling co-localizes with the GFP expression in the APCs of the 3 rd instar larvae (Fig. 1A). To further support this, we expressed GFP using transgenic animals carrying Gal4 under control of a 4 kb promoter-element comprising the region upstream Dar-2 (Dar-2>) between the transcriptional start codon and the next gene CG10000. We observed overlap between GFP and anti-AKH antibody staining in Dar-2 > GFP animals (Fig. S1a). Moreover, we also found Dar-2 expression in the APCs of adults using both DAR-2 antibody staining and Dar-2 > GFP, even though GFP was only observed in a subset of AKH-positive cells. This could be due to the nature of the Dar-2 promoter construct or that the CC is a heterogeneous cell population and not all cells express the receptor at high levels ( Fig. S1b,c). To confirm expression of Dar-2 in the adult CC, we measured Dar-2 transcripts in the adult CC and found that the level of Dar-2 transcript was efficiently reduced using the CC-specific Akh> driver in combination with Dar-2-RNAi ( Fig S1d).
We also observed anti-DAR-2 staining in a population of neurons in the brain anatomically resembling the IPCs (Fig. S2a). To demonstrate that these DAR-2 positive neurons correspond to the IPCs, we used dilp2-Gal4 (dilp2>) to drive IPC-specific GFP expression and found co-localization with the anti-DAR-2 immunolabeling (Fig. 1B). We also confirmed that Dar-2> drives GFP expression in the IPCs using a DILP2 antibody that specifically labels the IPCs (Fig. S2b,c). To confirm expression of Dar-2 in the IPCs, we used the newly developed CRISPR/Cas9 technique to create a T2A-Gal4 reporter knock-in C-terminally in Dar-2 to tag the endogenous gene. This drives expression of Gal4 in the same pattern as the endogenous Dar-2 gene and by insertion of an intervening T2A sequence between Dar-2 and Gal4, the Gal4 can be translated independently of Dar-2, which allows Gal4 to enter the nucleus and activate transcription of UAS-GFP. This line with a reporter on the endogenous Dar-2 gene drives expression in DILP2 positive cells in the brain (Fig. S2e), demonstrating Dar-2 expression in the IPCs. Together these results show that Dar-2 is expressed in the IPCs of both the larvae and adults.
While the CC cells could be targeted by AstA released from endocrine cells of the gut, the IPCs are located in the brain suggesting that these neurons are regulated by a brain source of AstA. To examine this possibility, we used an AstA antibody to investigate if the neurites of the AstA producing neurons may contact the IPCs. In combination with the visualization of the IPCs (dilp2 > GFP), we found that varicose processes from the AstA neurons terminate in the vicinity of the presumed dendrites of the IPCs in the protocerebrum (Fig. 1C, S2d, S3a) and the IPC processes that branch in the tritocerebrum and subesophageal ganglion (SOG) (Fig. 1D, S3b) of both larvae and adults. Proximity between the AstA-immunoreactive varicosities and the IPCs, expressing Dar-2, was also seen with double labeling of anti-AstA and Dar-2 > GFP (Fig. S3c,d). The proximity of the AstA-positive varicosities and the IPCs together with the expression of Dar-2 in the IPCs and the APCs indicate that AstA may be involved in regulation of metabolism through modulation of DILP and AKH signaling. AstA regulates neuroendocrine signaling involved in energy metabolism. To investigate whether AstA regulates DILP and AKH signaling, expression of TeTxLC.tnt, the tetanus toxin light chain, which inhibits neurotransmitter exocytosis 28 , and NaChBac, a bacterial sodium channel that increases neuronal excitability 29 , was targeted to the AstA-producing neurons using AstA-Gal4 (AstA>). Previously, it has been shown that expression of NaChBac using AstA> , which expresses Gal4 specifically in AstA-positive neurons, is sufficient to change the activity of the AstA neurons 18 . To examine the effects of AstA neuronal activity on APC and IPC activities, we measured changes in expression of Akh, dilp2 and dilp3, and metabolic genes that are influenced by AKH and DILP signaling. Although the expression of Akh was not influenced in AstA > TeTxLC.tnt and AstA > NaChBac animals, the expression of the AKH receptor gene (akhr) was increased in AstA > TeTxLC.tnt animals where the activity of AstA-neurons is inhibited ( Fig. 2A). Increased akhr expression may indicate an upregulation of the receptor in response to reduced levels of AKH in circulation. Moreover, activation of the AstA-neurons (AstA > NaChBac) induced target of brain insulin (tobi), an α -glucosidase homolog which is stimulated by both AKH and the DILPs released from the IPCs 30 . Activation of the AstA neurons also promoted the expression of both dilp2 and dilp3 consistent with the increased expression of tobi (Fig. 2B). Although dilp2 was also upregulated by inhibition of the AstA neurons (AstA > TeTxLC.tnt), a dramatic drop in dilp3 expression was observed under these conditions. This suggests that AstA neuronal activity influences dilp expression in the IPCs, and especially dilp3 seems to be tightly associated with the activity the AstA producing neurons. In comparison, the moderate upregulation of dilp2 expression by both activation and inactivation of AstA neurons may indicate a more indirect and perhaps a compensatory regulation of dilp2 expression by AstA neuronal activity. The transcription of eIF4E-binding protein (4EBP), encoding an inhibitor of translation which is suppressed by DILP signaling 31,32 , was not significantly changed.
To further characterize the phenotype associated with AstA loss of function, we generated a mutation in the AstA locus using the CRISPR/Cas9 system 33 . We recovered a line carrying a mutation that deletes part of the AstA gene including the start codon, which we named AstA SK1 (Fig. S2f). Since ablation of either the APCs or the IPCs results in increased starvation resistance 3,6,12 , we examined the resistance of AstA SK1 mutant flies to starvation. Flies with AstA SK1 mutation survive significantly longer under starvation compared to the controls (Fig. 2C). We used y 2 cho 2 v 1 and y 1 w 1118 and a cross between these two lines as isogenic controls, since the mutation was generated in these genetic backgrounds. To determine whether AstA SK1 is a strong loss of function allele and ensure that the phenotype is caused by a specific mutation in the AstA locus, we tested the AstA SK1 allele over the Df(3R)BCS519 deficiency that covers the AstA locus. AstA SK1 /Df(3R)BCS519 flies are starvation resistant, like AstA SK1 homozygous flies indicating that AstA SK1 is a strong loss of function mutation. Since tobi is a target of both DILP and AKH signaling and that its expression was altered by AstA neuron activation ( Fig. 2A), we analyzed the expression of tobi in flies lacking AstA. We found that AstA SK1 homozygous flies and AstA SK1 /Df(3R)BCS519 flies exhibit a strong reduction in expression of tobi, similar to the decrease in tobi expression observed in Akh SK1 animals with a mutation in the Akh gene that disrupts the Akh reading frame (Fig. 2D, Fig. S2f). Like AstA SK1 mutants, the Akh SK1 mutant flies are also starvation resistant (data not shown) as previously reported for animals that lack Akh due to genetic ablation of the APCs 3 . Taken together this indicates a complex interplay between AstA, DILP and AKH signaling.
AstA is expressed both in the nervous system ( Fig. S2d) and in endocrine gut cells 18 and its release from these different subgroups of cells may be temporally distinct during natural feeding conditions. These different sources of AstA may affect DILP and AKH signaling independently. To determine whether the changes in expression of metabolic genes regulated by AKH and DILP signaling is caused by a direct effect of AstA on the IPCs, APCs or other metabolic changes that has an indirect effect, we analyzed the consequence of reducing Dar-2 expression in these cells on the survival under starvation. Using an RNAi line that efficiently reduces expression of Dar-2 ( Fig. S1d and 23 ), we silenced Dar-2 in either the IPCs (dilp2 > Dar-2-RNAi) or APCs (Akh > Dar-2-RNAi). Knock down of Dar-2 in the APCs increased resistance to starvation significantly compared to the control, indicating that these animals have reduced AKH signaling (Fig. 2E). On the other hand, overexpression of Akh in the APCs made animals significantly more starvation sensitive compared to the control suggesting that these animals release more AKH into circulation which promotes energy mobilization. The starvation resistance observed after knock down of Dar-2 in the APCs was not reversed by Akh overexpression indicating that the increased Akh transcription does not increase AKH release. This indicates that DAR-2 is important for the release of AKH consistent with the data showing that changing the activity of the AstA producing neurons does not influence akh expression. Reduction of Akh expression in the APCs made animals more resistant to starvation than when expression of Dar-2 was reduced in these cells. However, when Akh and Dar-2 expression was reduced simultaneously, the flies became even more starvation resistant than when these genes were knocked down individually. This suggests a genetic interaction between Dar-2 and Akh that supports DAR-2 modulation of AKH release.
Interestingly, reducing the expression of Dar-2 in the IPCs (dilp2 > Dar-2-RNAi) also increased starvation resistance compared to the control (Fig. 2F). This suggests that dilp2 > Dar-2-RNAi flies may have reduced insulin signaling, indicating that AstA is important for both AKH and DILP release.
animals survive longer under starvation compared to the control, they are less resistant to starvation compared to dilp2 > Dar-2-RNAi flies, which may be due to different strengths of the Dar-2> versus dilp2> drivers. Another possibility is that simultaneous silencing of Dar-2 in both the CC and IPCs using the Dar-2> may result in another AKH to DILP ratio than when silencing Dar-2 only in the IPCs using dilp2>, which may have a different effect on starvation survival. Together these observations suggest that AstA may be involved in energy mobilization through regulation of circulating AKH and DILP levels.
Reducing Dar-2 mRNA levels in the APCs and IPCs affects metabolic target genes. To further evaluate the effects of AstA on AKH and DILP signaling, we measured how cell specific Dar-2-RNAi affected transcript levels of key metabolic target genes influenced by AKH or DILP signaling. To investigate differences between the sexes, both males and females were analyzed. In male flies, reducing the expression of Dar-2 in the APCs did not affect expression of Akh consistent with the notion that AstA is involved primarily in the regulation release rather than expression of Akh from the CC (Fig. 3A). On the other end, expression of tobi has been shown to be dependent of AKH release from the CC 30 . Consistent with a reduction in AKH release in Akh > Dar-2-RNAi males, we observed a decrease in tobi mRNA combined with an increase of akhr mRNA. Furthermore, a strong increase in expression of phosphoenolpyruvate carboxykinase (PEPCK), coding for a key enzyme in the gluconeogenesis 34 , supports that glycogen breakdown, a process promoted by AKH, is impaired consistent with reduced AKH signaling and a compensatory increase in gluconeogenesis. Similar effects on the expression of PEPCK and tobi were observed in females, but Akh expression was also reduced (Fig. 3B).
Next, we analyzed expression of these AKH signaling target genes in flies where Akh was either overexpressed or depleted by genetic ablation (Akh > Grim) of the APCs 35 . Consistent with previous observations, we found a reduction in tobi expression in APC-ablated animals 30 , like observed in Akh SK1 mutants (Fig. 1D). Alterations in expression of the AKH-influenced metabolic genes observed in animals with reduced Dar-2 expression in the APCs mostly mimic those seen in APC-ablated animals and not flies overexpressing Akh (Fig. 3A-D). Furthermore, we found that transcription of the brummer (bmm) lipase is induced 6 hr after starvation in Akh > Dar-2-RNAi (Fig. S3e), like observed for animals lacking AKH or its receptor 7 . Together this indicates that lack of DAR2 in the APCs reduces AKH signaling, implying that AstA has a positive effect on AKH signaling.
Silencing of the expression of Dar-2 in the IPCs reduced dilp2 expression in females, but not in males (Fig. 3E,F). Although expression of dilp2 and dilp3 was not affected in dilp2 > Dar-2-RNAi males, increased the expression of 4EBP was observed in both sexes. Expression of 4EBP, a direct target of FOXO, is upregulated in response to repressed insulin signaling 31,32 and can be used as a proxy for global insulin signaling 36,37 . Expression of tobi is dependent not only on AKH, but also DILP signaling, which means that animals with reduced IPC release of DILPs have lower tobi transcript levels 30 . Consistent with the increase in 4EBP in dilp2 > Dar-2-RNAi animals, we observed that IPC-specific loss of Dar-2 was associated with a reduction in tobi expression. This indicates that silencing of Dar-2 in the IPCs causes a drop in systemic DILP signaling. Together these data suggest that AstA acts directly on the IPCs and APCs and has a positive influence on DILP and AKH signaling.

DAR-2 is involved in the coordination of AKH and DILP signaling. Since the balance between
AKH and the DILPs is important for nutrient homeostasis and interactions between the APCs and the IPCs have been suggested 30,38 , we examined whether AstA is involved in crosstalk between these two cell types. To investigate if DILP signaling is regulated by signals from the APCs, and examine if this regulation is influenced by DAR-2, we analyzed the expression of dilp2 and −3 and their target genes in Akh > Dar-2-RNAi animals (Fig. 4A,B). Expression of dilp2 was significantly increased in these animals. For dilp3 expression, we observed a reduction in females but not in males. While the reduced tobi expression in Akh > Dar-2-RNAi animals may be a result of low AKH signaling, the effects on 4EBP expression supports an effect on systemic DILP signaling, although it cannot be excluded that this is an indirect effect of changes in AKH signaling. Remarkably, we observed that 4EBP expression was reduced in Akh > Dar-2-RNAi females, while it was increased in males, suggesting a sexually dimorphic interaction between AKH and DILP signaling. Changes in expression of genes associated with DILP signaling in the APC-ablated flies was similar to those seen in Akh > Dar-2-RNAi animals, including the sex-specific differences in 4EBP expression and upregulation of dilp2. Together these results suggest that the APCs directly or indirectly play a role in regulation of systemic DILP signaling in a sex-specific manner (Fig. 4C,D). This also indicate that alterations in DILP signaling may indirectly contribute to the increased survival resistance associated with loss of Dar-2 in the APCs or with silencing of Akh (Fig. 2E). We observed no effect of lack of Akh on dilp3 mRNA levels. While our data are consistent with the findings of recent study 39 , an earlier study reported that loss of Akh induces dilp3 expression 30 . We also confirmed that dilp3 mRNA levels are unaffected in the Akh mutant (data not shown).
We also measured the expression of Akh and akhr in animals with reduced Dar-2 expression in the IPCs, to examine if AstA regulation of the IPCs affects AKH signaling (Fig. 4E,F). Although a direct effect on Akh expression was not observed, dilp2 > Dar-2-RNAi animals exhibited a more than three-fold increase in akhr expression in females. Since a reduction in AKH signaling in APC-ablated animals was  associated with a compensatory increase in akhr expression (Fig. 3D), it is likely that AstA stimulated DILP signaling has a positive effect on AKH signaling. Another possibility is that DILP signaling regulates akhr transcription in the fat body, thereby affecting the response to AKH.

Loss of AstA results in accumulation of lipids in the fat body.
In Drosophila, fat is stored in lipid droplets in the fat body, which is dependent both on AKH and DILP signaling 7,40,41 . dilp and Akh pathway mutants have elevated lipid levels and loss of AKH signaling and insulin resistance have been associated with increased fat body lipid droplet storage. If AstA affects metabolism, as suggested by its influence on AKH and DILP signaling and key metabolic genes, then loss of AstA may lead to a change in the abundance of lipid droplets in the fat body. To test this possibility, we examined lipid droplets in the adult fat body using Coherent Anti-Stokes Raman Scattering (CARS) microscopy, a label-free technique that allows visualization of lipids by imaging 42 . Loss of AstA increased the number of fat body lipid droplet compared to the controls (Fig. 5), indicating that lack of AstA is associated with accumulation of high fat body levels of lipids. This finding is consistent with the notion that AstA is involved in the control AKH and DILP release, since reduced AKH and DILP signaling promotes lipid accumulation in the fat body. This provides evidence indicating that AstA mutants have metabolic defects and excess energy reserves which is consistent with their increased resistance to starvation. In contrast, changing the activity of the AstA neurons was recently reported not to affect starvation resistance and lipid levels in Drosophila 18 . A likely explanation for the observed differences is that changing AstA neuron activity may produce another phenotype compared to a mutation that affects AstA signaling in all tissues, including the endocrine gut cells.

AstA is affected differently by carbohydrate and protein rich diets and regulates preferences for these nutrients. Since the interplay between the DILPs and AKH has previously been suggested
to mediate metabolic programs related to carbohydrate versus protein rich diets 30 , we asked whether AstA signaling is regulated in response to feeding and the type of nutrients. We measured the expression of AstA and Dar-2 and key metabolic target genes in animals fed on either carbohydrate rich (cornmeal) or protein rich (yeast) food after 48 hours on a restricted low nutrient diet (1% sucrose) and compared it to control flies kept on cornmeal. After nutrient restriction, gene expression of AstA and Dar-2 was down regulated compared to the control and the expression of several metabolic genes was decreased in males indicating a general reduction of metabolism (Fig. 6A-D). When flies were fed on a high sugar diet after nutrient restriction, the expression of AstA and Dar-2 exhibited a strong response and increased to levels above that of control flies (Fig. 6A,B), while expression of the metabolic target genes only rebounded to levels similar to the control.
Feeding on a protein rich diet also increased AstA expression, but not as strongly as the high sugar diet, and Dar-2 expression did not increase (Fig. 6C,D). The type of diet, protein versus carbohydrate, also affected expression of the metabolic genes differentially. While the carbohydrate rich meal increased expression of 4EBP, the protein rich diet further reduced expression compared to the levels observed in nutrient restricted animals. The drop in 4EBP expression on the protein rich yeast diet indicates that dietary protein strongly increases systemic DILP signaling. Levels of tobi and PEPCK mRNA also increased more on the protein rich yeast compared to the carbohydrate rich cornmeal, suggesting an increase in glycogen breakdown and gluconeogenesis, processes expected to be important on a dietary source low on carbohydrates. The increase in DILP signaling as measured by changes in 4EBP and tobi expression on a protein rich diet is consistent with previous observations 30,43 . Taken together this shows that AstA and Dar-2 expression is regulated by nutrient intake and that the response depends on the carbohydrate and protein composition of the diet.
The differential response in AstA expression to carbohydrates and proteins prompted us to investigate whether AstA signaling is involved in feeding decisions related to intake of these different types of nutrients. To test if AstA affects the preference between sugar and proteins, we performed a two-choice feeding assay 44 , where flies were allowed to choose between a sucrose medium colored with a red dye and a yeast medium colored with a blue dye. We observed that ad-libitum-fed control flies kept on standard food conditions prefer sucrose (Fig. 6E) consistent with previous observations 44 . In contrast, ad libitum fed AstA > NaChBac flies with activation of the AstA neurons showed a stronger preference for the protein rich yeast diet compared to control flies. We also found a stronger preference for yeast in females compared to males. These data suggest that AstA is involved in assigning nutritional value to these dietary components and that increased AstA signaling may mimic sugar satiety or perhaps reflect a protein-hunger driven behavior. We therefore asked whether reducing AstA signaling promotes motivation to consume dietary sugar. To test this possibility, we measured sucrose feeding in the AstA SK1 mutants. We found that loss of AstA enhances sucrose intake compared to the control flies (Fig. 6F). These results also agree with a recent report showing that inhibition of the AstA neurons enhances the acceptance of sugar in an otherwise unpalatable food resource 18 . Altogether our data suggest that AstA respond to changes in nutrient intake and plays an important role in controlling metabolism and feeding (Fig. 6G). This suggests that AstA may be involved in a nutrient sensing mechanism that assign nutritional value to sugar and protein and guide feeding decisions to maintain the balance between energy requirements and food intake.

Discussion
In order to adjust energy homeostasis to different environmental conditions, feeding-related behavior needs to be coordinated with nutrient sensing and metabolism. Our data suggest that AstA is a modulator of AKH and DILP signaling that control metabolism and nutrient storage, but also affects feeding decisions. The positive effect of AstA on AKH signaling indicated by our observations is supported by the recent finding that expression of a presumably constitutive active mu opioid receptor, a mammalian GPCR which is also closely related to DAR-2, stimulates AKH release from the APCs in Drosophila 45 . Moreover, AstA-type peptides have also been shown to stimulate AKH release in Locusta migratoria 46 . AKH is primarily regulated at the level of secretion to allow a rapid response to metabolic needs 47 . Considering that we only observed a minor effect of Dar-2 silencing in the APCs on Akh transcription, it is likely that AstA primarily acts at the level of AKH release in Drosophila.
Our data suggest regulation of both the DILPs and AKH by AstA indicating a close coupling between the activity of these two hormones. Consistent with this notion, our results also indicate a feedback relationship between the IPCs and APCs. The IPCs have processes that contact the CC cells 13,14 and it is possible that DILPs released from these affect AKH release. Our findings are supported by a previous study that identifies a tight association between DILPs and AKH secretion in Drosophila 30 . Furthermore, it was recently found that AKH regulates DILP3 release from the IPCs, and that sugar promotes DILP3 AstA neuron activation were tested in a two choice feeding assay for their preference for sucrose versus yeast (protein). Negative values indicate a preference for sucrose and positive values indicate a preference for yeast. Ad-libitum-fed AstA > NaChBac flies had a stronger preference for protein rich yeast compared to the control. Ctrl: AstA-Gal4/+. Error bars indicate standard errors (n = 3). (F) Sucrose consumption of adult males lacking AstA (AstA SK1 and AstA SK1 /Df(3R)BSC519) compared to control animals (y 2 cho 2 v 1 , y 1 w 1118 and y 2 cho 2 v 1 x y 1 w 1118 : offspring from crossing these lines). Sucrose intake was measured using the Capillary Feeding (CAFE) assay. Error bars indicate standard errors (n = 10). *: P < 0.05; **: P < 0.01; ***: P < 0.001 (Student's t-test). (G) Model: The proposed role of AstA in coordinating metabolism and feeding decision to balance nutrient intake according the internal nutrient state. AstA is affected differentially by dietary sugar and protein and guide metabolic programs, through the balance between AKH and DILPs, important to maintain homeostasis on food with different sugar:protein ratios. Activation of AstA neurons change the preference for protein rich food, showing that AstA plays a key role in the decision-making processes that assign value to dietary sugar and protein and guide food choice.
Scientific RepoRts | 5:11680 | DOi: 10.1038/srep11680 release, while DILP2 release is amino acid dependent 39 . Interestingly, our data, which suggest that AstA is involved in the cross-talk between DILPs and AKH related specifically to sugar and protein, also indicate that AstA has a strong influence on dilp3 expression. Why is the relationship between the DILPs and AKH so tight? Even though insulin-like peptides reduce hemolymph sugar, they also reduce the content of stored glycogen and lipids 11,12 , like AKH 3,6 . Consistent with this, both AKH and the DILPs stimulate expression of tobi, which encodes a glycosidase believed to be involved in glycogen breakdown 30 . However, since AKH and the DILPs have opposing effect on hemolymph sugar levels, a balance between these hormones is presumably required to maintain homeostasis. It is likely that different sources of AstA affect these two hormones, since the IPCs are located in the brain in proximity of AstA-positive neurites, while AstA-positive processes do not innervate the CC 48 . Thus, it is likely that neuronal-derived AstA affects DILP secretion from the IPCs, while circulating AstA, which may be released from the endocrine cells of the gut, may be the source of AstA that acts on the APCs to regulate AKH. AstA regulation of DILP and AKH release may therefore not occur simultaneously and could also depend on the type of nutrient ingested, or be sequential. Since our data suggest feedback regulation between AKH and DILP, the overall outcome of simultaneous AstA induced activation of both cell types will not necessarily be a strong and equal increase in both hormones in the hemolymph. It is possible that AstA is involved in metabolic balancing, adjusting the ratio between AKH and DILPs in response to different dietary conditions. In mammals, glucagon and insulin are secreted simultaneously when the animal feeds on a protein-rich diet, to prevent hypoglycemia and promote cellular protein synthesis, since insulin is strongly induced after ingestion of amino acids [49][50][51] . A similar mechanism has been proposed to explain the relationship between DILPs and AKH in Drosophila 30 . The balance between DILP and AKH therefore may be important for resource allocation into growth and reproduction.
Several differences in the expression of genes involved in energy mobilization were observed between males and females, which possibly reflects sex-specific strategies for energy mobilization and allocation of resources towards reproduction. Interestingly, 4EBP expression was significantly decreased in females with reduced AKH signaling, but upregulated in males. This suggests that in females AKH has a strong negative influence on DILP signaling that is not present in males. Why does the interaction between AKH and DILPs differ between sexes? An interesting possibility is that this sexually dimorphic interaction is related to the different preferences and requirements for sugar and protein in males and females. Males generally have a higher preference for sugar compared to females that prefer more dietary protein and show strong correlation between amino acid uptake, insulin and reproduction 44,52,53 . In both mammals and Drosophila the balance between insulin and glucagon/AKH is important for nutrient homeostasis in response to high-protein versus high-sugar diets 30,50,51 . This balance ensures that insulin promotes protein synthesis in response to dietary amino acids, while maintaining sugar levels stable, a function possibly important in females to allocate the high consumption of amino acids into reproduction. Thus, the sex-specific interplay between DILPs and AKH likely reflects difference in the metabolic wiring of males and females that underlie the sexually dimorphic reproductive requirements for dietary sugar and protein.
Interestingly, AstA expression showed a general increase after feeding with a stronger transcriptional response of both AstA and Dar-2 to the carbohydrate rich diet compared to the protein rich diet. AstA may therefore be important for coordinating carbohydrate and protein dependent metabolic programs. The strong response to carbohydrates indicates that AstA may be involved in signaling related to carbohydrate feeding, although increased transcription may not necessarily result in elevated release of the mature AstA peptide. Nonetheless, the data indicate that feeding regulates AstA-signaling and that the response is influenced by the food composition. Consistent with the notion that AstA is involved in different responses to dietary carbohydrate and protein, we found that flies with increased AstA neuronal activity increase their protein preference on the expense of their natural preference for sucrose. The AstA regulated circuitry may therefore be important for guiding the decision to feed on protein or sugar, a decision influenced by metabolic needs 44 . The AstA neurons have projections that may contact the Gr5a sugar sensing neurons and AstA > NaChBac flies with increased activity of the AstA neurons display reduced feeding and responsiveness to sucrose under starvation 18 . Thus, the increased preference for dietary protein in AstA > NaChBac flies observed here may be caused by reduced sucrose responsiveness. If AstA signaling is high after feeding on carbohydrates as indicated by our data showing increased expression of AstA and Dar-2, then an increase in AstA signaling might mimic carbohydrate satiety. In line with this view, our data show that animals lacking AstA enhance their intake of dietary sugar. AstA signaling may therefore increase the animals preference for essential amino acids, as suggested by a recent study indicating that amino acid depleted flies increased their taste sensitivity for amino acids, even when they were replete with glucose 54 . Based on our data, we therefore propose that AstA plays a central role in a circuitry important for encoding nutritional value related to these distinct nutrients and the regulation of feeding decisions and metabolic programs. Excess dietary sugar is associated with obesity 37,41 and we find that flies lacking AstA enhance intake of sugar and have increased lipid storage droplets in their fat bodies, like animals lacking AKH or its receptor 7 . Thus, our data implicate AstA in regulation of appetite and food intake related to sugar, which is relevant for understanding obesity.
Our study suggests that AstA affects metabolism through its action on two key players, the DILPs and AKH. AstA expression is induced by feeding, but exhibits a differential nutritional response to dietary sugar and protein and influence metabolic programs and feeding choices associated with the intake of these nutrients. Interestingly, the homolog of AstA, galanin, regulates both feeding and metabolism in mammals and in Caenorhabditis elegans loss of the Allatostatin/galanin-like receptor npr9 affects foraging behavior and nutrient storage 55 . Altogether our data suggest that AstA is part of a conserved mechanism involved in coordinating nutrient sensing, feeding decisions and metabolism to ensure adequate intake of amino acids and sugar to maintain nutrient homeostasis under different feeding conditions.

Methods
Drosophila stocks and mutants. The (BDSC, 25023). A mutation was generated in the AstA and Akh loci using the CRISPR/Cas9 system as described 33 . The target guide RNA (gRNA) sequences GCAGTAGGAGGTGGGCGTGAAGG for AstA and GGCCAGCAGCATGAAGAGCACGG for Akh, including the underlined 3 bp Protospacer Adjacent Motif (PAM) sequence, were used to generate U6-gRNA vectors. Transgenic U6-gRNA flies were established and indel mutations in the AstA and Akh loci were recovered in offspring from nos-Cas9 and U6-gRNA flies. For each gene, six to eight independent F1 males were mated to balancer females. After three days, the male was removed from the culture vial and its genomic DNA was extracted and used as a template for PCR amplification of the target region. The PCR product was directly sequenced to identify a potential heterozygous indel mutation. For AstA, four out of six F1 males examined carried a small indel mutation. For Akh, a small indel mutation was present in all of eight F1 males examined. The AstA SK1 mutation removes 19 nt of the AstA locus including the ATG start codon, while the Akh SK1 mutation introduces a frameshift downstream of the ATG start codon (Fig. S2f). As controls for the AstA SK1 and Akh SK1 mutants, crosses between the y 2 cho 2 v 1 and y 1 w 1118 lines, as well as the individual lines, were used since mutations were introduced into these genetic backgrounds. The y 2 cho 2 v 1 and y 1 w 1118 lines had been generated by crossing the y 1 v 1 , y 2 cho 2 , y 1 and w 1118 (BDSC). To generate a knock-in GAL4 reporter line for Dar-2, a T2A-GAL4 56 transgene was integrated into the Dar-2 locus by CRISPR-mediated homologous recombination. T2A-GAL4 was placed immediately downstream of the final coding codon of Dar-2, such that the engineered locus encodes a bicistronic transcript that produces DAR-2 and GAL4 proteins, thereby allowing GAL4 to be expressed exactly in the same spatiotemporal pattern as endogenous DAR-2. Details of the transgene cassette used in the targeting vector will be described elsewhere. All flies were reared under 25 °C and 60% humidity and with 12:12 h light-dark cycle on standard cornmeal unless otherwise stated. To ensure similar nutritional conditions, an equal number of adult flies were kept in each vial for all experiments.
Quantitative PCR. Total RNA was isolated using RNeasy Mini kit (Qiagen) and treated with RNase-Free DNase to avoid genomic DNA contamination. RNA was extracted from four pooled females or five pooled males. For isolation of total RNA from adult CC, RNA was extracted from 10 adult CC for each sample. For measuring dietary effects, male and female w 1118 were separated after eclosion and kept on normal cornmeal for 3 to 6 days before they were transferred to a restricted diet of agar containing 1% sucrose 43,57 for two days. Hereafter they were fed either pure yeast or normal cornmeal for 24 hours. Control flies were kept at standard cornmeal and collected at the same age. For all other experiments RNA was extracted from animals 6 days after eclosion. cDNA was synthesized from oligo(dT) primers using the Superscript III reverse transcriptase (Invitrogen). Levels of transcripts were quantified using quantitative real time PCR (qPCR) with SYBR green as the reporter. Primers are given in the Table S1 and qPCR conditions were: 95 °C 10 min followed by 45 cycles of 95 °C for 15 sec, 60 °C for 15 sec and 72 °C for 15 sec. Melting curve analysis was carried out for all PCR reactions to verify homogeneity of the PCR products. As an internal standard, expression of the ribosomal protein rpL23 was used for normalization as described 58 . Relative quantification of transcript levels was determined using the comparative Ct-methods.
Starvation assay. Flies were kept on standard cornmeal on a density of 20 animals per vial for 4-7 days after eclosion before they were starved by transferring to vials containing agar (1%). Prism software (GraphPad Prism software, version 4.02, San Diego California USA) was used to examine the significance of the starvation survival using the Kaplan-Meier method and a Mantel-Haenszel logrank test. Feeding assays. Two-choice feeding preference assay was performed as described 44 . In brief, flies were kept on standard medium for 3-5 days after eclosion before they were offered the choice between sucrose medium (100 mM) containing a red dye and a yeast medium (5%, rich in protein) with a blue dye for 3 hours. Flies were scored according to the color of their abdomen. The "yeast preference index" was calculated and each experiment was repeated 3 times and each included 13-23 flies.
Quantification of sucrose intake was determined using the Capillary Feeding (CAFE) assay as described 59 . Males were kept on standard medium for 3-5 days after eclosion and then transferred to vials with 1% agar (to allow access to water) and a glass capillary containing 5% sucrose and 0.6 g/l Allura Red AC dye (Sigma-Aldrich) for 2 hours at 25 °C and 60% humidity. Blank controls without flies were included in each experiment to determine evaporation.
Determination of lipid storage droplets using Coherent Anti-Stokes Raman Scattering (CARS) microscopy. Lipid droplets were determined in fat body from adult flies 4-5 days after eclosion using CARS microscopy. Fat bodies were dissected in PBS and immediately imaged using a Leica TCS SP8 microscope with a CARS laser picoEmerald (OPO, > 600 mW at 780 nm to 940 nm, pulse width 5 to 6 ps, 80 MHz;Pump, > 750 mW at 1064 nm, pulse length 7 ps, 80 MHz) and the LAS AF/X software. Lasers were adapted to the CH stretching vibrational range by tuning the pump beam to 816.4 nm while keeping the Stokes beam constant at 1,064.6 nm. The output of the lasers was set to 1.3 W and the scan speed was 400 Hz. Signals were collected from the epi-CARS (E-CARS) and epi-SHG (E-SHG) detectors. Quantification of lipid droplet number and size was done using ImageJ (NIH).