Food odors trigger an endocrine response that affects food ingestion and metabolism

Abstract

Food odors stimulate appetite and innate food-seeking behavior in hungry animals. The smell of food also induces salivation and release of gastric acid and insulin. Conversely, sustained odor exposure may induce satiation. We demonstrate novel effects of food odors on food ingestion, metabolism and endocrine signaling in Drosophila melanogaster. Acute exposure to attractive vinegar odor triggers a rapid and transient increase in circulating glucose, and a rapid upregulation of genes encoding the glucagon-like hormone adipokinetic hormone (AKH), four insulin-like peptides (DILPs) and some target genes in peripheral tissues. Sustained exposure to food odors, however, decreases food intake. Hunger-induced strengthening of synaptic signaling from olfactory sensory neurons (OSNs) to brain neurons increases food-seeking behavior, and conversely fed flies display reduced food odor sensitivity and feeding. We show that increasing the strength of OSN signaling chronically by genetic manipulation of local peptide neuromodulation reduces feeding, elevates carbohydrates and diminishes lipids. Furthermore, constitutively strengthened odor sensitivity altered gene transcripts for AKH, DILPs and some of their targets. Thus, we show that food odor can induce a transient anticipatory endocrine response, and that boosted sensitivity to this odor affects food intake, as well as metabolism and hormonal signaling.

Electronic supplementary material

The online version of this article (doi:10.1007/s00018-015-1884-4) contains supplementary material, which is available to authorized users.

Introduction

Accurate and sensitive detection of food-related odors is important in guiding animals to sources of energy and nutrition. Food odors also stimulate appetitive responses in hungry animals. In humans, the smell of food increases appetite and triggers salivation as well as release of gastric acid and insulin [1–4] emphasizing the link between the brain and gastrointestinal tract originally proposed by Pavlov [5, 6]. Conversely, it appears that sustained exposure to food-related odors can decrease appetite, and even induce satiety in human subjects [7–10]. This so-called sensory-specific satiety can occur while feeding or without food ingestion. Thus, food odors can mediate opposite effects on appetite and food intake depending on duration of exposure and other contextual factors. Here, we investigate the role of odor signaling in appetite, feeding and subsequent metabolic events, including insulin and glucagon-like signaling. In particular, we seek to determine mechanisms by which food smells trigger endocrine responses prior to intake of food and how sustained exposure to such odors affects food intake and metabolism.

Drosophila melanogaster with its genetic tractability is an excellent model animal for unraveling odor-associated behaviors and physiology [11, 12], as well as the endocrinology and signaling pathways that regulate feeding and metabolism [13–20]. Food odors are known to elicit strong innate attraction in D. melanogaster and mechanisms regulating food search behavior are being unraveled [21–27]. The attraction to food smells is known to be dependent on the nutritional state of the organism [17, 24, 28–32]. In D. melanogaster, the strength of synaptic signaling from specific olfactory sensory neurons (OSNs) in the antennae is increased by hunger and regulated by insulin-like peptides (DILPs) [24]. These authors showed that in hungry flies, where circulating DILP levels are low, subsets of OSNs relay increased signaling to the brain in response to food odor due to strengthened synaptic activity. This elevated activity is caused by the neuromodulator, short neuropeptide F (sNPF), and its receptor sNPFR1, both of which are expressed in subsets of OSNs [24, 33]. The level of sNPFR1 expression in OSNs is negatively regulated by the insulin receptor (dInR) and thus in hungry flies, where circulating levels of DILPs are low, the OSNs consequently express high levels of sNPFR1. Synaptic release of acetylcholine from the OSNs increases due to strengthened sNPF signaling which leads to increased odor inputs to higher order neurons and stronger attraction to food sources [24]. This peptidergic modulatory mechanism provides a genetic tool to manipulate odor sensitivity to mimic hunger or satiety in flies and, thus, providing means to investigate links between olfaction and metabolism and how odors affect feeding and endocrine responses.

In mammals, feeding is under complex control by neuropeptides and peptide hormones [34–36]. Also, in flies, a number of peptides are known to regulate feeding and metabolism [13, 14, 17, 26, 37–39] and the neuronal and neuroendocrine systems releasing these may be linked to the inputs from the olfactory system. Here, we focus on the key regulators of carbohydrate and lipid homeostasis and feeding: adipokinetic hormone (AKH), with a glucagon-like function, and the DILPs [39–43], as well as sNPF [44].

In the present study, we investigated food ingestion, endocrine responses and metabolism in flies after acute or sustained exposure to food-related odorants. Furthermore, we manipulated sNPF, sNPFR1 and dInR levels in OSNs to mimic nutritional state-dependent modulation of olfactory signaling. We first show that the strongly attractive food odor of apple cider vinegar [21, 22] acutely induces an endocrine response in hungry control flies (w¹¹¹⁸): a transient increase in circulating glucose and increased transcription of genes encoding DILPs, AKH and several genes in adipose tissue (fat body). Next, we demonstrate that sustained exposure to vinegar odor reduces food intake over several days and affects metabolism and transcription of the same set of genes as in the acute response. Our findings suggest that specific sets of odor channels, which utilize insulin-modulated sNPF signaling in a gain control mechanism, can alter food seeking and feeding in flies and thereby affect metabolism.

Experimental procedures

Fly husbandry and transgenic flies

Experimental flies were kept at 25 °C and 12:12 L:D on food consisting of 5 % sucrose, 5 % yeast and 1.2 % agar. Nipagin (0.3 %) and propionic acid (0.3 %) were used as mold inhibitors. Mated 3–4-day-old female flies were used for all experiments, unless otherwise mentioned.

We used D. melanogaster w ¹¹¹⁸ from Bloomington Drosophila Stock Center (BDSC), Bloomington, IN, as control flies. The following Gal4 lines were used: Orco-Gal4, previously known as Or83b-Gal4 [45], was from BDSC), akh-Gal4 [46] from J. H. Park (Knoxville, TN, USA) and dilp2-Gal4 [41] from E. Rulifson (Stanford, CA, USA).

To manipulate gene expression and cell activity, we used the following lines: UAS-2xsNPF, UAS-sNPF-RNAi and UAS-sNPFR-RNAi [44, 47] from K. Yu (Daejeon, Korea), the UAS-dInR ^CA line [24, 48] from (Ping Shen, Athens, GA, USA), as well as UAS-NaChBac and UAS-Ork1 [49] from BDSC. All Gal4 and UAS lines were backcrossed to w ¹¹¹⁸ for at least four generations, except akh-GAL4 and UAS-InR^CA lines that were in yw background. UAS-mcd8-GFP flies were from BDSC. An Orco loss of function mutant, Orco ² (formerly known as Or83b ² [50]), was also obtained from BDSC.

Short-term fly treatment with food-related odors

For experiments, w ¹¹¹⁸ flies were starved for 18 h on 0.5 % agarose solution in groups of 15. Plastic tubes with either 0.5 ml of apple cider vinegar, yeast suspension (10 % dry heat-inactivated yeast in distilled water) or solvent (distilled water) applied on a 3 cm³ piece of Styrofoam were prepared 30 min before start of the experiment. Tubes with flies were attached with sticky tape to odor-containing ones. To prevent the flies from feeding, the two tubes were shielded by a mesh. Combined tubes were placed vertically in an incubator (25 °C) with the odor tube bottommost. After 15, 30, 60, 120 and 240 min, flies were collected either for immunostaining or frozen at −80 °C for further analysis.

Feeding

The amount of food eaten by single flies was measured in a capillary feeding assay (CAFE; [51]) modified according to Lushchak et al. [52]. To measure fly food ingestion in the presence of either vinegar or yeast, we used the following set-up: flies were kept in 1.5-ml centrifuge tubes with 20 holes made in the bottom with a thin needle. These tubes were inserted through tight-fitting holes into 5-ml vials with 0.5 ml of either odorant or solvent in the bottom. All tubes with flies were supplied with 5-μl capillary tubes with 5 % sucrose and 5 % yeast extract (heat-inactivated yeast). Flies were starved for 18 h before put in the assay tubes with access to food-filled capillaries. The amount of food consumed was measured every 24 h for 3–4 days (in some cases for 6 days) and mean daily consumption by the fly was given. Food capillaries were exchanged every 24 h.

Odor trap assay

The flies’ ability to find a food source was tested as described by Larsson et al. [50]. Briefly, flies were starved overnight on 0.5 % aqueous agarose solution. The following day the flies were placed into 5-L glass chambers supplemented with two trap vials. One trap vial contained 1 ml of apple cider vinegar and another one 1 ml of distilled water. After 24 h, the chambers were frozen and the number of flies in each trap and the glass chamber were counted. The percent of flies in each trap was calculated as amount of flies in the trap divided by total amount of flies in the experiment.

Hemolymph and stored carbohydrates and triacylglycerides

Female flies (3–4 days old) of different genotypes were used to measure concentrations of circulating glucose and trehalose together with whole body glycogen and trehalose according to [53]. Pre-weighed flies were decapitated and hemolymph was collected by centrifugation (3000 g at 4 °C, 6 min). Hemolymph was used to measure circulating glucose and trehalose, whereas whole bodies were used for determination of glycogen and stored trehalose. All parameters were measured with a glucose assay kit involving glucose oxidase and peroxidase (Liquick Cor-Glucose diagnostic kit, Cormay, Poland). Trehalose was converted to glucose by porcine kidney trehalase (Sigma T8778) and glycogen by amyloglucosidase from Aspergillus niger (Sigma 10115). Glucose and trehalose levels are expressed as concentration in hemolymph, whereas glycogen and body trehalose are given as amount per wet weight. The amount of triacylglycerides (TAG), and probably other glycerides (but for simplicity we refer to TAG in our results) in flies was determined with a Liquick Cor-TG diagnostic kit (Cormay, Poland) and calculated as amount per wet weight. All genotypes were tested in 4–6 independent replicates.

Quantitative real-time PCR (qPCR)

The amounts of mRNA from 3–4-day-old female flies were measured in heads (Dilps 2, 3 and 5) and whole bodies for all others by qPCR. RNA was isolated as described in [54], from four biological replicate samples of each genotype tested; 1 μg of total RNA was used for cDNA synthesis. cDNA was synthesized in triplicates, which were subsequently pooled and diluted for qPCR. Expression of genes of interest was measured relative to that of the reference gene rp49 using an ABI Prism 7000 instrument (Applied Biosystems) and a SensiFAST SYBR Hi-ROX Kit (Bioline) under conditions recommended by the manufacturer. Initial experiments were run also with another reference gene (actin 88F; Act88F), but since there was no difference between the data obtained with the two reference genes we ran the remaining replicates with rp49 only. Each analytical and standard reaction was performed in three technical replicates. The levels of transcripts were measured using primer pairs shown in Table S1. These primers were used extensively in a previous report [55].

Immunohistochemistry and quantification of immunofluorescence

Standard immunohistochemical protocols were used. In brief, dissected brains were fixed in ice-cold 4 % paraformaldehyde (in 0.1 M sodium phosphate buffer) for 4 h. After rinsing in 0.01 M phosphate-buffered saline with 0.25 % Triton-X, the brains were preincubated in 5 % normal goat serum overnight. Then, brains were incubated with the primary antibody for 72 h at 4 °C under gentle agitation. Secondary antibodies were used for 48 h before rinsing and final mounting in aqueous glycerol (80 %).

The following antibodies were used: rabbit anti-sNPF precursor (1:1000, kindly provided by J. A. Veenstra, Bordeaux, France [56]), rabbit anti-DILP2 (1:2000, kindly provided by Mark Brown, Athens, GA [57]) and mouse anti-GFP (1:1000, Invitrogen, Carlsbad, CA, USA). For detection of antisera, we used Alexa 488/546-tagged secondary antibodies (1:1000, Invitrogen, Carlsbad, CA, USA).

Specimens were imaged with a LSM 780 confocal microscope (Zeiss Jena, Germany). Confocal images were processed with the ZEN 2011 software (Zeiss). For quantitative measurements of anti-DILP2 staining, we used identical settings in all confocal stacks. Maximum projection images of stacked optical sections were used and the mean fluorescence intensity within outlines of anti-DILP2 stained cells was calculated for each animal.

Data analysis

Data were collected and analyzed in Microsoft Excel and statistical analysis was performed with Excel and Prism GraphPad 6. One-way ANOVA, followed by Dunnett’s multiple comparison test was used to compare differences between genotypes, in some experiments Student’s t test was used.

Results

Acute responses to food odors: anticipatory endocrine effects

Mated female flies (w ¹¹¹⁸ strain) that had been starved for 18 h were kept in tubes and exposed to odors of either apple cider vinegar, heat-inactivated yeast or solvent (water). Flies were collected every 15 min and hemolymph samples collected for analysis of glucose levels. In flies that were exposed to vinegar odor, there was a rapid and transient increase in hemolymph glucose (about 30 %) after 15 min (Fig. 1a). After the transient increase, the glucose levels remained relatively high over 240 min compared to flies exposed to water or yeast. In flies exposed to yeast, there was a small, but significant, increase in circulating glucose after 15 min that returned to levels close to control flies after 1 h (Fig. 1a). Thus, exposure to vinegar triggered an increase in circulating glucose that was sustained in spite of 4 h of continued starvation. We next tested glucose levels in hungry flies with impaired olfaction due to a mutation in the olfactory co-receptor Orco [50] in the same odor exposure experiment. These flies displayed low levels of glucose irrespectively of which odor they were exposed to and the levels remained relatively stable over 4 h (Fig. 1b). This indicates that exposure to vinegar odor triggers a transient and then sustained increase in circulating glucose in the flies.

Fig. 1.

Food odors trigger rapid response in glucose levels and gene transcription in w ¹¹¹⁸ flies. a Exposure to apple cider vinegar results in a rapid increase in circulating glucose in w ¹¹¹⁸ flies. Yeast odor induces a transient increase at about 15 min and then falls close to levels in control flies exposed to solvent (water). b The same odors do not induce a glucose release in flies with impaired olfaction due to a loss of function mutation in the olfactory co-receptor Orco. c–f Vinegar odor exposure triggers a rapid increase in transcription of genes encoding adipokinetic hormone (akh, c), DILPs 2, 3 and 5 (dilp 2,3 and 5; d–f). Note that the Y-axes have different scales in these panels. g There is no effect on insulin receptor transcript, InR. h DILP2 immunofluorescence levels increase in the IPCs after 30 min and remain significantly higher for at least 1.5 h. In the above experiments, values of treated flies at given time points were compared to untreated flies and significance was analyzed by Student’s t test, * p < 0.05, ** p < 0.01, *** p < 0.001, for each treatment; n = 30–50 per time point (performed in 4–6 replicates)

A factor known to induce elevated glucose levels in the circulation of insects is adipokinetic hormone (AKH), the functional equivalent of glucagon [39, 40, 46, 58]. Therefore, we monitored the levels of akh transcript in flies exposed to vinegar or water over 4 h. We noted a rapid and transient increase in akh transcript after 15–30 min of odor exposure compared to control flies exposed to water (Fig. 1c). This increased transcription may reflect an increased AKH signaling caused by the food odor exposure, as supported by the elevated circulating glucose. Our data do, however, not provide direct evidence for AKH release.

We next monitored transcripts of several genes that may reflect changes in carbohydrate or lipid metabolism. Elevated circulating glucose levels are likely to induce release of insulin-like peptides (DILPs) in D. melanogaster [41, 59, 60]. Three of these peptides, DILP2, 3 and 5, are released from median neurosecretory cells of the brain, designated insulin-producing cells, IPCs [41, 60, 61]. Transcripts of all three brain-derived DILPs displayed a transient increase with dynamics similar to that of akh (Fig. 1d–f). Each transcript peaked after 30 min exposure to vinegar odor. dilp2 and 3 transcripts were about 4 times higher than in controls exposed to water (or at time 0) and dilp5 doubled. Over 4 h, the dilp2 and 3 transcripts decreased significantly below control levels. As expected, the expression of the insulin receptor (dInR) RNA remained stable over time both with vinegar and water exposure (Fig. 1g). For comparison with dilp2 RNA, we monitored DILP2 immunofluorescence in the cell bodies of IPCs. This immunofluorescence increased after 30 min and remained significantly higher than the control for 2 h (Fig. 1h). Note that the intensity of DILP2 immunolabeling in cell bodies is not likely to provide a measure of peptide release [62]; it was monitored here as another indication that odor stimulation affects the IPCs.

We also quantified transcripts of two factors mainly derived from the fat body, DILP6 and the leptin-like cytokine unpaired 2 (Upd2). Both of these were shown to be released from the fat body in response to nutritional signals in the circulation [63, 64]. The dilp6 transcript displayed a transient increase after 30 min of vinegar odor exposure (Fig. 2a), whereas the upd2 transcript transiently increases about five times after 30 min and thereafter decreases to a level significantly lower than controls after 4 h (Fig. 2b). Another gene known to be downstream of both AKH and DILPs is target of brain insulin (tobi), an α-glucosidase homolog [65]. The tobi transcript levels increased more slowly to a maximum at 60 min and remained high also after 4 h (Fig. 2c). Finally, the translational inhibitor 4E-BP (eIF4E-binding protein) RNA did not display significantly altered levels over 4 h of vinegar odor exposure (Fig. 2d).

Fig. 2.

Exposure to vinegar triggers rapid response in metabolic gene transcription in w ¹¹¹⁸ flies. a The fat body derived dilp6 transcript transiently increases in response to vinegar. b The RNA for the leptin-like cytokine upd2 displays a drastic and transient increase. After 240 min, the upd2 level decreases significantly below that in control flies exposed to water only. c Target of brain insulin (tobi) transcript increases to a steady high level over the 4 h tested. d The 4ebp RNA level does not change after vinegar exposure. In the above experiments, data were analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test, * p < 0.05, ** p < 0.01, *** p < 0.001, for each treatment n = 30–40 flies per time point (measured in 4 replicates)

These transient increases in transcript levels of akh, dilp2, 3, 5, 6 and upd2 after food odor exposure may reflect release of these molecules, although the possible link between transcription and release remains to be demonstrated. However, the transiently increased glucose levels in the circulation may be due to odor-induced systemic AKH release, known to also trigger appetite and energy-consuming food search [39, 46]. Likewise, the upregulated tobi transcript is probably a response to DILP and/or AKH release into the circulation [65].

Sustained exposure to food odors decreases food intake

Relatively brief exposures to food odors apparently have strong effects on food seeking in hungry animals and produce anticipatory endocrine responses. We next asked whether more sustained exposure to food odors affects food ingestion, as has been shown in humans where appetite is diminished [7, 8]. In other words, do flies feed normally when a strong food odor is superimposed on that of the food that is provided? We tested w¹¹¹⁸ flies in a capillary feeding (CAFE) assay over 3 days in tubes where they were exposed to odors of vinegar, yeast or solvent (water). Note that the CAFE assay probably monitors both vigor of food seeking and food consumption [51]. Flies exposed to yeast or vinegar odors fed significantly less than those exposed to water (Fig. 3a). Thus, ectopic food odors decrease feeding when applied over several days.

Fig. 3.

Feeding is reduced after food odor exposure. a Control flies (w ¹¹¹⁸) consume less food when exposed to vinegar or yeast odors compared to solvent alone (water) as measured in a CAFE assay over 3 days. In the above experiments, data were analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test, * p < 0.05, *** p < 0.001, for each treatment n = 60–80 flies per genotype (performed in 6–8 replicates). b, c Food consumption was tested in flies with depolarized or hyperpolarized DILP- and AKH-producing cells during exposure to food odors or solvent. Using a dilp2-Gal4 driver for IPCs and akh-Gal4 for AKH cells, we expressed an open inward rectifying K channel (Ork) for inactivation and a Na channel (NaChBac) for activation of these cells. Control flies were Gal4 lines crossed to w ¹¹¹⁸ (W). In b manipulations of IPCs combined with odor exposure are shown. In flies exposed to solvent only (white bars), hyperpolarization of IPCs diminished food intake (blue asterisks, *** p < 0.001) compared to controls (W). When IPCs are hyperpolarized by Ork expression, both vinegar and yeast odor significantly decreases feeding compared to water. In c manipulations of AKH cells are shown. In flies exposed to solvent only (white bars), depolarization of AKH cells increased food intake (blue asterisks, * p < 0.05) compared to controls (W). Both food odors significantly decrease feeding in control flies (W) and those with depolarized AKH cells (NaChBac). In the above experiments, data (in 3–4 replicates) were analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test, * p < 0.05, ** p < 0.01, *** p < 0.001, and Student’s t test for comparisons of solvent-exposed flies (blue asterisks)

This diminished food ingestion due to food odor exposure may be caused by sensory modulation of brain activity resulting in alterations in endocrine signaling. To explore this, we tested whether AKH or DILPs play a role in the decreased feeding. For this, we used the Gal4-UAS system [66] to manipulate activity in the AKH-producing cells and the IPCs. We expressed either a depolarizing bacterial sodium channel (UAS-NaChBac) or a hyperpolarizing open rectifying potassium channel (UAS-Ork) using akh-Gal4 or dilp2-Gal4 drivers. These flies were tested in the CAFE assay while exposed to vinegar, yeast or solvent (water) odors. When exposed to solvent, only flies with depolarized IPCs fed significantly less than controls (w ¹¹¹⁸), whereas hyperpolarization did not significantly change food intake (Fig. 3b). These findings are expected, since increased insulin signaling should induce satiety and a reduced food-seeking behavior (see [24]). With depolarized AKH-producing cells, food intake increased in water-exposed flies, whereas hyperpolarization diminished feeding (Fig. 3c). Our data are, thus, in line with previous findings that increased AKH release stimulates food search and feeding [39, 46].

Next, we monitored food ingestion in manipulated flies exposed to vinegar or yeast. The flies with hyperpolarized IPCs displayed significantly diminished feeding after vinegar or yeast odor exposures compared to water-exposed flies (Fig. 3b). With NaChBac-induced activation of IPCs, we only found a slight decrease in feeding after yeast, but not vinegar odor exposure compared to controls (Fig. 3b). In flies where we manipulated AKH cells, we found that vinegar and yeast odor decreased feeding significantly in flies with depolarized cells, whereas with hyperpolarized AKH cells only yeast odor induced a slightly reduced feeding compared to controls (Fig. 3c).

These findings suggest that the activity in both the IPCs and AKH cells is important for regulation of food intake under control conditions with no superimposed food odors. When vinegar and yeast odors are presented, the food intake decreases more drastically in flies with reduced activity in IPCs or with increased activity in AKH cells, i.e., with hormonal levels characteristic of hungry flies.

Effects of constitutively altered odor sensitivity on feeding, endocrine signaling and metabolism

If sustained exposure to food odors reduces feeding over a few days, then what is the effect of more long-term alterations in sensitivity to food odors? To test the effects of odor exposure on endocrine functions, we took advantage of recent findings showing that hunger-dependent modulation of synaptic activity in specific sets of OSNs by DILPs affects the fly’s sensitivity to food odors [24]. That study showed that overexpression of a constitutively active dInR (dInR^CA) in OSNs decreases the transcription of the sNPF receptor (sNPFR) in these cells. This leads to decreased signaling from the OSNs in response to food odors and thereby reduced food search. Thus, to change the sensitivity to food odors, we targeted sNPFR-RNAi or dInR^CA to OSNs. Furthermore, we altered the levels of sNPF in OSNs since this peptide ligand is also expressed in the OSNs [33, 67] (See Fig. S1A, B). Hence, we have three alternative means to change dynamics of sensitivity in OSNs and thereby mimic states of hunger and satiety in the olfactory system.

To change the expression of sNPF, sNPFR and dInR in OSNs, we used an Orco-Gal4 driver, known to be expressed in a wide array of OSNs, including food odor-sensitive ones [24, 45]. We tested 3–4-day-old female flies (unless otherwise stated) of different genotypes in various assays. The efficiency of these targeted manipulations on odor sensitivity was confirmed in an odor trap assay [50]. Diminished signaling in odor channels after knockdown of sNPF or sNPFR or overexpression of the dInR^CA in OSNs rendered flies less attracted to vinegar than controls, whereas overexpression of sNPF had no significant effect (Fig. 4a).

Fig. 4.

Manipulations of sNPF signaling in olfactory sensory neurons (OSNs) affect feeding and metabolism. a An odor preference trap assay was used to test flies with manipulations of sNPF signaling in OSNs: 3–4-day-old female flies were given a choice between apple cider vinegar and solvent (water). Orco-Gal4 flies were crossed to UAS-sNPF, UAS-sNPF-RNAi (Ri), sNPFR-RNAi and UAS-InR^CA (a constitutively active dInR) and control flies (WT; w¹¹¹⁸). The preference for vinegar odor was significantly reduced compared to controls (WT) when sNPF signaling was downregulated, but no effect was seen after sNPF overexpression. b Food consumption measured in CAFE assay over 6 days. c–f Effects of sNPF manipulations in OSNs on levels of glucose, trehalose glycogen and triacylglycerides (TAG) in 3–4-day-old female flies. g sNPFR-RNAi and InR^CA trigger increased TAG levels. Further effects of receptor manipulations on metabolism are shown in Fig. S3. h Levels of DILP2 immunofluorescence in IPCs of flies with altered sNPF signaling in OSNs. i Effects on feeding of decreased sNPF signaling in OSNs by receptor manipulations: both sNPFR-RNAi and InRCA leads to increased food consumption. In the above experiments, data were analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test, * p < 0.05, ** p < 0.01, *** p < 0.001, for each treatment n = 60–80 flies per genotype (performed in 6–8 replicates)

Next, we tested effects of sNPF signaling manipulations in OSN on food intake in flies exposed only to the intrinsic odors of the food provided. Overexpression of sNPF in OSNs leads to decreased food consumption in the CAFE assay (Fig. 4b). Conversely sNPF-RNAi targeted to OSNs induced increased food intake (Fig. 4b). Also, manipulation of signaling in OSNs by targeted sNPFR-RNAi and overexpression of the active insulin receptor, dInR^CA, induced increased feeding (see Fig. 4i). Thus, diminished signaling from the OSNs by diminished sNPF activity and thus reduced odor sensitivity leads to increased food intake and vice versa.

Using transgene flies with the same type of manipulations of sNPF signaling in OSNs, and ad lib access to food, we next monitored effects on carbohydrate and lipid metabolism. First, we analyzed the outcome of changing sNPF levels in OSNs of 4-day-old flies. We noted a steady-state increase in circulating levels of glucose and trehalose after overexpression of sNPF in OSNs and decreased levels after sNPF knockdown (Fig. 4c, d). The whole body glycogen was increased in flies with sNPF overexpression in OSNs, but not affected by sNPF knockdown (Fig. 4e). Finally, whole body triacylglyceride (TAG) levels were diminished after sNPF overexpression, but not significantly changed after sNPF-RNAi (Fig. 4f). Both sNPFR-RNAi and dInR^CA expression, however, increased TAG levels (Fig. 4g). Thus, carbohydrate and TAG levels are clearly affected by the strength of sNPF signaling in food odor channels. Since DILP signaling is known to affect carbohydrate and lipid homeostasis, we monitored DILP2 immunofluorescence in IPCs after interference with sNPF signaling in OSNs. DILP2 levels decreased significantly after sNPF overexpression and increased after sNPF-RNAi (Fig. 4h).

A set of experiments was conducted on 25-day-old flies to determine long-term effects of the sNPF manipulations in OSNs. These flies had ad lib access to food and were not exposed to odors other than those in the food. As seen in Fig. S2A–D, the steady-state levels of carbohydrates and TAG in normally fed older flies changed in a fashion very similar to flies that were 3–4 days old. The one difference was that glycogen levels were significantly decreased after sNPF-RNAi in 25-day-old flies (Fig. S2C), whereas they were not altered in the younger flies (Fig. 4e).

In the odor trap assay, the manipulations of sNPFR and dInR in OSNs did not yield as prominent phenotypes as the sNPF-RNAi (Fig. 4a), and similarly we noted that sNPFR-RNAi and expression of the dInR^CA in OSNs did not induce significant changes in all the other assays. As mentioned, food consumption and TAG levels increased after sNPFR-RNAi and dInR^CA expression (Fig. 4g, i). However, glucose, trehalose, and glycogen levels did not change significantly (Fig. S3A–C). On the other hand, both receptor manipulations lead to increased DILP2 immunofluorescence in cell bodies of brain IPCs (Fig. 4h).

To test steady-state effects of altered sNPF signaling in OSNs on AKH and DILP signaling, we monitored gene transcripts in 4-day-old flies. The akh transcript increased about four times after sNPF overexpression in OSNs, but was not affected by peptide knockdown (Fig. 5a). The dilp2, 3 and 5 transcripts were differentially affected by sNPF and sNPF-RNAi expression in OSNs (Fig. 5b–d). dilp2 and 3 were downregulated by sNPF overexpression and upregulated by RNAi (Fig. 5b, c), whereas dilp5 was regulated in the opposite direction (Fig. 5d).

Fig. 5.

Manipulations of sNPF signaling in olfactory sensory neurons (OSNs) affect gene expression in brain and other tissues. a The akh transcript increases fourfold after sNPF overexpression. b–d dilp2, 3 and 5 transcripts are differentially affected by sNPF overexpression and RNAi. e The dilp6 transcript decreases after sNPF overexpression and increases after sNPF-RNAi. f–i Transcripts of target of brain insulin (tobi), the leptin-like upd2, 4e-bp and pepck are also up- or downregulated by sNPF manipulations in a manner suggesting that sNPF overexpression decreases DILP2 and 3 signaling and sNPF-RNAi increases it. Effects of receptor manipulations in OSNs on transcripts are shown in Fig. S3. In the above experiments, data were analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test, * p < 0.05, ** p < 0.01, *** p < 0.001, for each treatment n = 35–40 flies per genotype (tested in 4 replicates)

Next, we monitored transcripts of dilp6, tobi, upd2, 4e-bp and pepck (PEPCK is a phosphoenolpyruvate carboxykinase, a key enzyme in the gluconeogenesis) to determine effects downstream of AKH and DILP signaling. Here, we found that dilp6 and tobi are downregulated with increased sNPF expression in OSNs, whereas dilp6, tobi and upd2 levels are upregulated with diminished sNPF signaling (Fig. 5e–g). Both 4e-bp and pepck are upregulated after increased sNPF signaling and downregulated by sNPF-RNAi (Fig. 5h, i). These data support the idea that constitutively increased signaling from OSNs (that diminishes feeding) increases AKH and DILP5 signaling, but diminishes DILP2 and 3 signaling with ensuing effects on target genes. With diminished sNPF signaling in OSNs, and thus less sensitive food odor channels, we see the opposite effects on transcript levels, suggesting increased DILP2 and 3 signaling and reduced AKH signaling. We also made an analysis of the same transcripts after expression of sNPFR-RNAi and dInR^CA in OSNs (to decrease odor sensitivity). dilp6 was not affected (Fig S3D), whereas tobi and upd2 were upregulated by sNPFR knockdown and dInR^CA expression (Fig. S3E,F), 4e-bp was unaltered (Fig. S3G) and pepck was downregulated by both manipulations (Fig. S3H). As expected, these phenotypes were the same as after sNPF-RNAi and the opposite of those obtained by overexpression of sNPF.

Discussion

Our study presents novel findings revealing that olfactory signaling plays different important roles in food attraction and feeding dependent on state of satiety. In hungry flies, food odors induce an acute anticipatory response that may prepare the organism for food search and ingestion. Furthermore, flies with access to food respond to sustained exposure to food odors with reduced food intake, and with genetically increased odor sensitivity this induces steady-state changes in metabolism and gene expression suggestive of reduced insulin signaling.

In hungry flies, the short-term effect of food odor exposure is a rapid and partly transient increase in expression of several genes involved in regulation of carbohydrate and lipid homeostasis and feedback from the fat body and intestine. This is accompanied by an early transient increase in circulating glucose, perhaps to provide fuel to support increased food search behavior in the fly (see [17, 46, 68]). The transient glucose surge may be due to the early activation of the glucagon-like peptide AKH [39, 40, 68] whose transcript increases significantly within 15 min and peaks at 30 min after odor exposure. Next, and possibly as a consequence of the increased blood glucose, we see increased transcription of dilp2, 3 and 5, and a few target genes of AKH and DILPs as well as targets of nutrient sensing in the fat body. Thus, starved flies display a rapid change in metabolic signaling when exposed to the attractive food odor vinegar. It is known that anticipatory cues, such as odor or sight of food, can trigger behavioral and physiological responses in humans such as inducing release of gastric acid and hormones [1, 3, 4, 28, 31, 69, 70]. Apparently, the hungry fly displays a conditioned appetitive response to food-related odors [34, 71], and we show that this response involves activation of the endocrine system. Such an anticipatory mechanism may speed up the subsequent digestion and nutrient uptake. This is beneficial while the organism’s vulnerability increases during the process of digestion as a consequence of reduced ability to escape predation [72].

The transcriptional changes of akh and dilps probably reflect production associated with release of the peptides into the fly circulation [40, 60]. That peptide release occurred is suggestive since the gene tobi, encoding an α-glucosidase, is upregulated after odor exposure. tobi is targeted by both AKH and DILPs and responds in opposite ways to dietary protein and carbohydrate [65]. DILP6 and Upd2 are released from the fat body upon feeding [63, 64] and we show that their transcripts display transient upregulation after vinegar exposure, or by other unknown signals. These transcriptional responses may be triggered by the increase in circulating glucose that follows vinegar odor exposure. In summary, so far, we propose that exposure to vinegar odor sets in motion an anticipatory response in the hungry fly that prepares it for food search and subsequent food ingestion and metabolic events.

Appetite suppression by food odors is a known phenomenon in humans [1, 7, 9], but has not been investigated experimentally in model animals. Thus, our findings on the suppression of food intake by food odor exposure in D. melanogaster are novel. The effect of sustained exposure to food odors in D. melanogaster can be observed as reduced food intake and a number of alterations in endocrine responses, metabolism and gene transcription. We also observed that when vinegar or yeast odors are presented the food intake decreases more drastically in flies with reduced activity in IPCs or with increased activity in AKH cells, suggesting that endocrine status influences the response to olfactory input.

We altered the signaling strength in food odor-sensitive OSNs by manipulations of a peptidergic (sNPF) co-transmitter system and utilized this as an efficient means to affect food seeking, feeding and metabolism in flies. Targeted changes of sNPF signaling in OSNs produced strong changes in odor attraction, food intake, metabolism, and expression of several genes. Thus, diminishing sNPF signaling by knocking down sNPF, sNPFR or overexpressing an active dInR in OSNs resulted in decreased attraction to vinegar, increased food intake and lowered levels of circulating and stored carbohydrates as well as increased TAG. Conversely, increased local sNPF signaling in OSNs decreased food intake and had the opposite effects on carbohydrate and lipid levels. The increased sNPF signaling in OSNs and ensuing reduction in food intake increases circulating and stored carbohydrates and decreases TAG levels, suggesting decreased insulin signaling. This is in concordance with gene transcript data where akh is upregulated, whereas dilp2 is downregulated together with dilp6 and tobi. Furthermore, the mRNA for the translational regulator 4E-BP and the gluconeogenesis enzyme PEPCK are upregulated as expected with decreased circulating insulin. Thus, it is sufficient to increase sNPF signaling locally in OSNs to induce a steady-state downregulation of systemic insulin signaling in the fly. Hence, food odor sensitivity affects the metabolic state of the fly due to decreased food seeking and intake that in turn affects AKH and DILP signaling.

How does the olfactory information induce changes in the endocrine system? Odor inputs are relayed from the antennal lobe via projection neurons (PNs) that have axon terminations in the calyx of the mushroom bodies and a protocerebral region designated the lateral horn [11, 71, 73]. There are no available data suggesting direct connections between antennal lobe PNs and neurosecretory cells of the brain that produce DILPs or cells producing AKH in the corpora cardiaca. However, a recent paper on larval D. melanogaster suggested that olfactory inputs, specified by the odorant receptor Or42a, via PNs control hormonal release of GABA which regulates maintenance of blood cell progenitors [74]. This study identified subesophageal neurosecretory cells as the source of GABA release, but did not report neuronal pathways connecting the olfactory PNs to these GABA-producing cells. It therefore remains to identify the connections between PNs and neuroendocrine cells, which probably arise from the lateral horn of the brain where PNs terminate [11, 71]. Candidate systems are available: several neurotransmitters and neuropeptides are known to regulate activity in IPCs, but the neuronal pathways releasing these have not been identified anatomically in any detail except for a set of brain neurons, DLPs, producing the peptides corazonin and sNPF [53, 75]. A subset of these DLPs expresses a gustatory receptor, Gr43a, known to be a fructose sensor [76, 77]. Since fructose levels in the circulation increase only after a carbohydrate meal [76], it is not likely that this neuronal pathway accounts for odor-induced responses prior to feeding. In mammals, odors that predict ingestion and flavor of food are processed in the amygdala and mediodorsal thalamus [31], and neuronal and hormonal pathways regulating appetite, feeding and metabolism have been charted, and circuitry in the hypothalamus is the main control center of the brain [32, 35, 36]. In flies, the brain region considered to correspond functionally to the hypothalamus is the pars intercerebralis where the IPCs are located (see [78, 79]).

In conclusion, we demonstrated that food-related odors alter appetitive behavior, feeding and metabolism in opposite directions depending on state of satiety and that odor inputs can activate the endocrine system. Interestingly, hungry flies exposed to food odors display an anticipatory endocrine response that presumably prepares them for food ingestion.

Electronic supplementary material

Below is the link to the electronic supplementary material.