Octopamine-mediated circuit mechanism underlying controlled appetite for palatable food in Drosophila

Significance

Proper control of appetite for palatable food is crucial for the prevention of overeating and development of obesity. A major factor that hinders the study of appetite control mechanisms is the complexity of the central nervous system of the traditional mammalian models. Genetically tractable Drosophila larvae have a relatively simple brain. Like humans, they display diverse adaptive feeding strategies in response to appetizing stimuli, changes of motivational state, and variations of food quality. In this work, we describe a unique neural circuit, defined by conserved norepinephrine-like octopamine and VEGF2-like receptor pathways, that exerts both positive and negative controls of larval appetite under fed and food-deprived conditions. Our findings provide fresh insight for the neurobiology and evolution of appetitive motivation.

Abstract

The easy accessibility of energy-rich palatable food makes it difficult to resist food temptation. Drosophila larvae are surrounded by sugar-rich food most of their lives, raising the question of how these animals modulate food-seeking behaviors in tune with physiological needs. Here we describe a circuit mechanism defined by neurons expressing tdc2-Gal4 (a tyrosine decarboxylase 2 promoter-directed driver) that selectively drives a distinct foraging strategy in food-deprived larvae. Stimulation of this otherwise functionally latent circuit in tdc2-Gal4 neurons was sufficient to induce exuberant feeding of liquid food in fed animals, whereas targeted lesions in a small subset of tdc2-Gal4 neurons in the subesophageal ganglion blocked hunger-driven increases in the feeding response. Furthermore, regulation of feeding rate enhancement by tdc2-Gal4 neurons requires a novel signaling mechanism involving the VEGF2-like receptor, octopamine, and its receptor. Our findings provide fresh insight for the neurobiology and evolution of appetitive motivation.

The adaptive control of foraging decisions is crucial to survival and reproduction and is mediated by complex brain mechanisms. For example, in hungry animals, feeding behaviors can be modulated by diverse neural systems including those responsible for receiving and processing sensory properties and assigning reward and motivational significance of food stimuli (1–3). At present, elucidation of molecular and circuit mechanisms underlying the adaptive control of feeding behavior remains highly challenging.

Our previous studies have shown that Drosophila larvae, like mammals, display diverse adaptive foraging strategies in response to appetizing odors or satiety state and food quality (4–6). For example, larvae fed for ad libitum intake tend to prefer soft, liquid sugar media that contain readily ingestible sugar solution but decline solid media in which sugar solution is embedded in gelled agar and is less accessible (5). However, as food deprivation is prolonged, larvae will become increasingly persistent in extracting the sugar solution from solid media (7). We have also shown that an evolutionarily conserved signaling cascade, involving neuropeptide F (NPF, the fly homolog of neuropeptide Y, or NPY) and insulin-like peptides (dILPs), selectively integrates motivational state (hunger) with persistence to pulverize solid food (5, 7).

The observation that the conserved NPY-like system selectively promotes food acquisition behaviors that require high energetic cost has led us to postulate that fly larvae may use other conserved neural mechanisms to regulate acquisition of readily accessible palatable food. In this work, we provide evidence which supports this hypothesis. We show that an octopamine (OA)/β-adrenergic-like receptor (Octß3R)-dependent circuit mechanism selectively regulates appetite for soft sugar media. This circuit mechanism seems to involve two subsets of tdc2-Gal4 neurons in the subesophogeal ganglia (SOG). One of them mediates the hunger-driven increase of feeding and is modulated by a novel activity of the VEGF2-like receptor pathway (8). The other is required for preventing excessive appetite in fed larvae. This and our previous findings provide fresh mechanistic insights into how brain mechanisms differentially organize appetitive motivations in responses to high- and low-quality food sources under different energy states.

Results

Hunger-Driven Appetite for Liquid Sugar Media in Fly Larvae.

Larvae fed for ad libitum consumption display a baseline level of feeding responses to readily accessible palatable food (e.g., 10% glucose agar paste), which can be quantified by counting the number of larval mouth hook contractions (MHCs) during a 30-s test period (Fig. 1). This baseline level of MHC rate increases in a dose-dependent manner when larvae are deprived of food (Fig. 1A) and is accompanied by an increase in ingestion rate (Fig. 1B). Hungry larvae display a slightly higher peak speed of MHC (11%) relative to fed larvae (maximum number of contractions per 3 s) (Fig. S1A), as well as a large increase in persistence of feeding, as evidenced by the shorter time intervals between bites (Fig. 1C and Fig. S1B). Together, these results indicate that fasted larvae, like hungry mammals, execute a behavioral program to effectively restore energy balance following food deprivation.

Fig. 1.

Quantification of hunger-driven feeding responses of Drosophila larvae. (A) The rate of larval MHC linearly increases as food deprivation prolongs up to 150 min; the correlation coefficient R = 0.99. (B) The amount of ingested food increased after food deprivation. N = 10 trials. *P < 0.05; **P < 0.01. (C) Each dot represents one bite. The typical feeding patterns sampled from eight fed or 150-min-deprived larvae are shown. (D) Inhibition of tdc2-Gal4, but not npf-Gal4, neuronal activity using UAS-Kir2.1 blocked hunger-induced feeding rate increases in liquid food (P < 0.01). Columns with identical letters indicate differences are statistically insignificant. (E) In contrast, inhibition of npf-Gal4 not tdc2-Gal4 neurons attenuated hunger-induced feeding rate increases in solid food; P < 0.01. (F) Transient inhibition of neurotransmission by tdc2-Gal4 neurons by expressing UAS-shi^ts1 at 30 °C abolished the hunger-elicited approaching response to liquid food; P < 0.01. (G) Stimulation of tdc2-Gal4 neurons by expressing UAS-NaChBac increased feeding response to liquid sugar but not sugar-free media; P < 0.01. All of the behavioral assays in this and following figures were analyzed under blind conditions. Unless stated otherwise, at least 12 larvae were tested for each group (n ≥ 12), and statistical analyses were performed using one-way ANOVA followed by Student–Newman–Keuls test in all figures. Error bars represent the SE (SEM) in this and all other figures.

Role of tdc2-Gal4 Neurons in Appetitive Motivation.

Our previous study of the NPF system suggests that feeding incentives of hungry larvae might be regulated by distinct neural mechanisms (4, 7). The insect OA system has been implicated in behaviors associated with seeking food rewards (9). The tdc2-Gal4 driver directs reporter expression in central neurons producing OA and/or tyramine (10). We found that blocking tdc2-Gal4 neuronal activity by expressing an inwardly rectifying potassium channel protein (UAS-Kir2.1) completely abolished hunger-induced MHC rate increases in liquid food (Fig. 1D) (11). Importantly, fasted tdc2-Gal4/UAS-Kir2.1 larvae showed normal hunger-driven feeding responses to less-accessible solid sugar food, opposite to the behavioral phenotypes of fasted larvae expressing UAS-Kir2.1 in NPF neurons (Fig. 1E and Fig. S2A). We also transiently inhibited tdc2-Gal4 neurons by expressing UAS-shi^ts1, a temperature-sensitive, semidominant-negative form of dynamin (12). At the restrictive temperature of 30 °C, tdc2-Gal4/UAS-shi^ts1 larvae failed to display hunger-driven feeding response to liquid food, but their feeding responses to solid food were normal (Fig. 1F and Fig. S2B). These findings suggest that tdc2-Gal4 neurons define an uncharacterized circuit mechanism that is acutely required for hunger-motivated feeding of readily accessible sugar media. In addition, we genetically activated tdc2-Gal4 neurons by expressing a UAS-NaChBac construct that encodes a voltage-gated bacterial sodium channel (13). Fed tdc2-Gal4/UAS-NaChBac larvae displayed enhanced MHC rates and decreased intervals between bites on liquid sugar food (Fig. 1G and Fig. S2C) but showed no detectable changes in feeding response to 10% glucose solid food or liquid media without sugar (Fig. S2D).

Role of OA in Appetitive Motivation.

To test whether OA is directly responsible for the observed phenotypes of tdc2-Gal4/UAS-Kir2.1 and tdc2-Gal4/UAS-NaChBac larvae, we first examined feeding responses of tßh^nM18 larvae carrying a null mutation in the tyramine β-hydroxylase (TβH) gene essential for OA synthesis (14). We found that blocking of OA synthesis completely abolished hunger-driven MHC rate increases in liquid food, phenocopying the tdc2-Gal4/UAS-Kir2.1 larvae (Fig. 2A). Moreover, prefeeding normal fed larvae with food containing OA also led to a detectable increase in the rate of MHC and ingestion of dyed liquid sugar media (Fig. 2 B and C). The same OA treatment of fasted tdc2-Gal4/UAS-Kir2.1 larvae largely restored hunger-driven feeding response to liquid food (Fig. 2D). These findings indicate that OA signaling underlies the activity of tdc2-Gal4 neurons in selective regulation of food acquisition. Given our previous findings that tdc2-Gal4 neuron activity stimulates a behavioral program distinct from that of npf-gal4 neurons, we investigated how simultaneous increases of OA and NPF signaling levels might affect food motivation in fed larvae. To this end, OA was introduced orally to fed elav-Gal4/UAS-npf^cDNA larvae that overexpress NPF. We found that fed elav-Gal4/UAS-npf^cDNA larvae treated with OA behaved similarly to OA-treated fed control larvae in liquid food (Fig. S3A). However, in solid food, OA-treated elav-Gal4/UAS-npf^cDNA larvae behaved similarly to untreated elav-Gal4/UAS-npf^cDNA larvae (Fig. S3B). These results support the notion that the OA-mediated circuit mechanism for feeding of liquid sugar media functions independently from the NPF circuit mechanism.

Fig. 2.

Control of appetite for liquid sugar food by the OA system. (A) The null mutant tßh^nM18 is deficient in hunger-driven feeding response; columns with identical letters indicate differences are statistically insignificant analyzed with one-way ANOVA followed by Dunn’s test. (B) Prefeeding of fed wild-type larvae with media containing10 mM OA for 30 min increased larval MHC rate (P < 0.01) and (C) ingestion rate on liquid glucose media (P < 0.05). (D) Oral OA treatment significantly restored the hunger-driven feeding response of fasted tdc2-Gal4/UAS-Kir2.1 larvae; P < 0.05. (E) Conditional knockdown of Octß3R, but not Oamb, in the nervous system attenuated hunger-driven approaching of liquid food. OA treatment failed to rescue the behavioral phenotype; columns with identical letters indicate differences are statistically insignificant (P > 0.05). (F) Octß3R^MB04794 larvae showed attenuated hunger-driven approaching of liquid food (P < 0.05).

OA Enhancement of Feeding Requires Octβ3R.

Four different OA receptors have been identified in Drosophila (15, 16). To determine the downstream effectors of the OA feeding pathway, we used a mifepristone (RU486)-inducible pan-neural GS-elav-Gal4 driver to perform dsRNA-mediated conditional knockdown of individual OA receptor activity (Table S1). We found that only disruption of the β-adrenergic–like Octβ3R receptor (16) blocked starvation-induced MHC rate increases. Unlike in tdc2Gal4/ UAS-Kir2.1 larvae, oral introduction of OA failed to rescue the defect of feeding response in fasted GS-elav-Gal4/UAS-Octβ3R^dsRNA larvae (Fig. 2E). Furthermore, Octβ3R^MB04794 larvae containing a transposable element that disrupts Octβ3R are also deficient in the hunger-driven feeding response (Fig. 2F). These results suggest that OA and Octβ3R define a circuit mechanism that enhances feeding of liquid sugar media in fasted larvae.

Subsets of tdc2-Gal4 Neurons Differentially Control Appetite.

Tdc2-Gal4 is expressed in multiple clusters of OA neurons in the larval brain lobes and the ventral ganglia (Fig. 3A). Immunostaining of the larval central nervous system with anti-TβH antibodies further suggests that the somata of all central OA neurons are located to the ventral ganglia (Fig. 3B) (17). OA neurons in the larval SOG respond to gustatory inputs from gustatory receptor neurons, and unlike the OA neurons in the abdominal ganglia, they are not motor neurons (17, 18). Anatomical mapping of tdc2-Gal4 neurons at the single-cell resolution in adult flies revealed that the ventral unpaired median neurons (VUMs) in the anterior compartment (VUM1) and middle compartment (VUM2) of SOG seem to project to several common areas of the protocerebrum, whereas those in the posterior compartment of SOG (VUM3) project to the ventral ganglia (19, 20). The tsh-Gal80 construct blocks GAL4 activities in the thoracic and abdominal ganglia (21) (Fig. 3C). Fasted tsh-Gal80/tdc2-Gal4/UAS-shi^ts larvae showed reduced MHC rates in response to liquid glucose media (Fig. 3D), suggesting that tdc2-Gal4 neurons in the SOG are required for hunger-driven response to liquid food. The SOG is proposed to act as a feeding control center in the central nervous system of insects (22, 23). To test which tdc2-Gal4 neurons in the SOG are important for OA-dependent feeding activity, we generated targeted lesions in the subsets of tdc2-Gal4 neurons using focused laser beams (24). Targeted lesions in five VUM1 with or without four ventral paired median (VPM)1 neurons in SOG1 caused a significant increase in the feeding activity of fed larvae (Fig. 3E and Fig. S4A). However, this increased feeding activity was abolished when additional neurons from SOG2 (five VUM1 plus six VUM2) were lesioned. In fasted larvae, targeted lesions in six VUM2 neurons alone attenuated hunger-elicited increases of feeding response (Fig. 3F), suggesting that proper control of appetitive motivation under fed and fasted conditions may require the negative and positive regulatory activities from VUM1 and VUM2 neurons, respectively. Lesions in all OA neurons (five VUM3 plus two VPM3) in SOG3 had no effects on larval feeding response (Fig. S4 A and B). In addition, conditioned excitation of tdc2-Gal4 neurons in fed larvae by expressing UAS-dTrpA1 triggered increased feeding response to liquid sugar media. However, this dTrpA1-stimulated effect was completely abolished by targeted lesions in VUM2 and VPM2 neurons (Fig. 3G). Together, our findings suggest that VUM1 neurons may restrict appetite for liquid sugar media in fed larvae by suppressing feeding enhancement by VUM2 neurons.

Fig. 3.

Functional mapping of tdc2-Gal4 neurons by targeted laser lesion. (A) Expression of nuclear LacZ in the tdc2-Gal4 neurons in the central nervous system of tdc2-Gal4/UAS-nlsLacZ larvae. The tissue is counterstained with phalloidin. (Inset)The distribution of tdc2-Gal4 neurons in three SOG compartments (SOG1–3). Five VUM1 and four VPM1 neurons (arrows) in the SOG1; six VUM2 and two VPM2 neurons (arrowheads) in the SOG2; five VUM3 and two VPM3 neurons (asterisks) in the SOG3. (Scale bar, 50 μm.) (B) Expression of TβH in the larval brain lobes and ventral nerve cord. No TβH-positive somata were detected in the brain lobes. (Scale bar, 50 μm.) (C) Expression of a nuclear GFP in the central nervous system of tdc2-Gal4/tsh-Gal80/UAS-nlsGFP larvae (lateral view). The GFP expression in the OA neurons from the thoracic and ventral ganglia is blocked by tsh-Gal80. The numbers denote SOG1 to 3, respectively. (Scale bar, 50 μm.) (D) At 30 °C, fed larvae expressing tdc2-Gal4/UAS- shi^ts1/tsh-Gal80 remained deficient in hunger-driven feeding response; **P < 0.01. (E and F) Analysis of feeding-related activities of OA neurons in the SOG in fed and fasted larvae by generating targeted lesions using focused laser beams (P < 0.01). (G) tdc2-Gal4/UAS-dTrpA1 larvae showed abnormally high feeding response, which was blocked by targeted lesions in VUM2 and VPM2 neurons; P < 0.01. At least 12 larvae were tested for each group (n ≥ 12) in three separate trials.

Drk Mediates Hunger-Induced Appetitive Motivation.

From a previous genetic screen, we isolated a candidate gene downstream receptor kinase (drk), the fly homolog of human growth factor receptor-bound protein 2 [Grb2 (25)], whose mutations affect larval feeding response to liquid, but not solid, food (Fig. 4). Under fed conditions, larvae transheterozygous for three independent loss-of-function drk alleles, drk^ΔP24/drk^R1, drk^ΔP24/drk¹⁰⁶²⁶ (26, 27), showed basal levels of feeding activity similar to wild-type larvae. However, after 150-min deprivation, the mutant larvae exhibited significantly attenuated feeding responses to liquid food (Fig. 4A). In addition, expression of UAS-drk^dsRNA driven by elav-Gal4 also led to significantly reduced hunger-driven responses to liquid food. Conversely, overexpression of drk cDNA in fed larvae (elav-Gal4/UAS-drk^cDNA) caused excessive feeding response to liquid food (Fig. 4B). Importantly, both fasted elav-Gal4/UAS-drk^dsRNA and fed elav-Gal4/UAS-drk^cDNA showed normal responses to solid food (Fig. S5A). These findings suggest that the neural activity of drk is a positive regulator of hunger-driven feeding response to liquid sugar food.

Fig. 4.

The neural activity of drk in appetitive motivation. (A) The drk transheterozygous mutant larvae are deficient in hunger-driven food response; *P < 0.05 and **P < 0.01, compared with ^+/+. (B) Knockdown of drk in the nervous system reduced hunger-driven food response (P < 0.05), whereas its overexpression increased feeding response in fed larvae (one-way ANOVA followed by Dunn’s test; P < 0.05). (C) Knockdown of drk activity in tdc2-Gal4 neurons, but not npf-Gal4 neurons, attenuated hunger-driven food response. One-way ANOVA followed by Dunn’s test; P < 0.05. (D) OA treatment of tdc2-Gal4/UAS-drk^dsRNA larvae rescued the deficiency of hunger-driven response to liquid food. Columns with identical letters indicate differences are statistically insignificant.

Because loss of the neural activity of drk or OA signaling led to similar feeding behavioral defects, we investigated whether drk regulates OA neuronal signaling. Indeed, larvae expressing UAS-drk^dsRNA in tdc2-Gal4 but not npf-Gal4 neurons displayed attenuated hunger-driven approaching response to liquid food, but showed normal food response under fed conditions (Fig. 4C). Moreover, oral administration of OA to tdc2-Gal4/UAS-drk^dsRNA larvae largely restored their deficiency in food motivation (Fig. 4D). In addition, both fasted tdc2-Gal4/UAS-drk^dsRNA and fed tdc2-Gal4/UAS-drk^cDNA showed normal responses to solid food (Fig. S5B). These findings suggest that drk regulates feeding of liquid food through its modulation of OA neuronal signaling.

Tdc2-Gal4 Neuronal Signaling Requires Pvr.

The fact that Drk is a SH2/SH3 adaptor protein that directly binds to activated receptor tyrosine kinase (RTK) strongly implicates the involvement of a yet-unknown RTK in modulation of OA neuronal activity. At least 14 RTK genes have been identified in the Drosophila genome (28–34). To identify which RTK(s) are involved, we performed dsRNA-mediated knockdown of the 14 known RTK genes in tdc2-Gal4 neurons (Table S2). This initial screening has led to the identification of three candidates, Eph receptor tyrosine kinase (Eph), heartless (htl), and PDGF/VEGF-receptor-related (Pvr), which is a VEGF2-like receptor (8). To assess whether the effects of these three RTKs on tdc2-Gal4 neurons are physiological, we conditionally knocked down the individual activity of Pvr, Eph, or htl using the GS-elav-Gal4 driver. We found that only GS-elav-Gal4/UAS-Pvr^dsRNA larvae showed significantly attenuated hunger-driven response to liquid food (Fig. 5A). Similarly, GS- elav-Gal4/UAS-drk^dsRNA larvae also showed significantly attenuated hunger-driven response to liquid food. Moreover, the phenotype of tdc2-Gal4/ UAS-Pvr^DN larvae expressing a dominant negative form of Pvr provides further verification of the essential role of Pvr in appetitive motivation (29) (Fig. 5B and Fig. S6 A and D). Furthermore, expression of a dominant-active Pvr (Pvr^ACT) in tdc2-Gal4/UAS-Pvr^ACT fed larvae caused excessive feeding of liquid sugar media, as evidenced by increased rate of MHC and ingestion of dyed food (Fig. 5C and Fig. S6C). These findings indicate that the Pvr pathway in tdc2-Gal4 neurons has a previously uncharacterized role in the physiological regulation of hunger-driven food motivation. The Drosophila genome encodes three PDGF/VEGF homologs (Pvf1–3) that function as the ligands of Pvr (29). We tested the three different mutant larvae (Pvf1^MB01242, Pvf2^d02444, and Pvf3^EY09531), each carrying a transposon that disrupts pvf1, pvf2, and pvf3, respectively (Table S3). We have found that pvf2^d02444 larvae showed attenuated hunger-driven feeding response (Fig. 5D).

Fig. 5.

A unique neural activity of the Pvr pathway in appetitive motivation. (A) Conditional pan-neuronal expression of drk^dsRNA and Pvr^dsRNA but not Eph^dsRNA or htl^dsRNA reduced hunger-driven food response. One-way ANOVA followed by Dunn’s test; P < 0.05. (B) Expression of UAS-Pvr^dsRNA and UAS-Pvr^DN in tdc2-Gal4 neurons attenuated hunger-driven feeding response to liquid media. Columns with identical letters indicate differences are statistically insignificant. (C) Expression of UAS-Pvr^ACT-caused increased feeding in fed larvae; P < 0.01. (D) Fasted Pvf2^d02444 larvae showed attenuated feeding response; P < 0.01. (E) Fasted larvae expressing both drk^cDNA and Pvr^dsRNA in tdc2-Gal4 neurons displayed normal liquid food response; P < 0.01. (F) Fed larvae expressing both drk^cDNA and Ras^DN in tdc2-Gal4 neurons also displayed normal liquid feeding response; P < 0.01.

To provide evidence for the functional interaction between drk and Pvr in tdc2-Gal4 neurons, we coexpressed UAS-Pvr^dsRNA and UAS-drk^cDNA under the direction of tdc2-Gal4. We found that expression of UAS-drk^cDNA in fasted tdc2-Gal4/UAS-Pvr^dsRNA larvae completely restored the deficiency in approaching response to liquid food (Fig. 5E). Because drk signaling is mediated by Ras85D GTPase (35), we also coexpressed a dominant-negative form of mammalian Ras protein (UAS-Ras^DN (36);) with UAS-drk^cDNA in tdc2-Gal4 neurons. As expected, expression of UAS-Ras^DN blocked the excessive food response in fed tdc2-Gal4/UAS-drk^cDNA larvae (Fig. 5F). Together, these findings raise the possibility that Drk and Ras may function in the Pvr pathway to regulate OA neuronal signaling.

Discussion

A Neural Network That Differentially Regulates Appetitive Motivations.

Modulation of feeding responses to food sources is heavily influenced by nutritional quality, taste, and the energy costs of foraging. Our findings suggest that Drosophila larvae have evolved a complex neural network to regulate appetitive motivations (Fig. 6). In hungry fly larvae, OA neurons seem to mediate a specialized circuit that selectively promotes persistent feeding of readily ingestible sugar food. This OA circuit functions in parallel to the previously characterized mechanism coregulated by the fly insulin and NPY-like systems that drives feeding response to nonpreferred solid food (5, 7). Because food deprivation triggers simultaneous activation of both circuits, hungry larvae become capable of adaptively responding to diverse energy sources of high or low quality. It remains to be determined how OA signaling promotes persistent feeding response to liquid sugar food in hungry larvae. One possible scenario is that OA neurons in the SOG may be conditionally activated by gustatory cues associated with rich palatable food to promote appetitive motivation.

Fig. 6.

A schematic model for differential regulation of two hunger-driven appetitive motivations. The OA-mediated circuit, regulated by the Pvf2/Pvr signaling pathway, selectively controls appetite for readily accessible liquid sugar media. This circuit functions independently of the previously identified the dILPs/insulin receptor-regulated NPF circuit that selectively promotes appetite for solid sugar media.

Two Opposing OA Activities in Regulation of Appetite Control.

We have provided evidence, at both molecular and neuronal levels, that the OA-mediated feeding circuit has two opposing effects on food motivation. When surrounded by liquid sugar media, the OA circuit is essential to prevent fed animals from excessive feeding. Because targeted lesions in VUM1 neurons caused excessive feeding response, these neurons may define an inhibitory subprogram within the OA feeding circuit (Fig. 3E). However, targeted lesions in VUM2 neurons attenuated hunger-induced increases of feeding response, suggesting that VUM2 neurons, along with the OA receptor Octβ3R, may define a subprogram that enhances feeding in fasted larvae (Fig. 3F). Several lines of evidence suggest that the VUM2 neuron-mediated subprogram may be suppressed by the VUM1 neuron-mediated subprogram. First, fed larvae with double lesions in both VUM1 and VUM2 neurons failed to display excessive feeding, suggesting that increased feeding response of fed larvae deficient for VUM1 neuronal signaling requires VUM2 neurons. Second, targeted lesions in VUM2 neurons of fed tdc2-Gal4/UAS- dTrpA1 larvae completely blocked the increased feeding response induced by genetic activation of tdc2-Gal4 neurons. Finally, the anatomical data also show that VUM1 and VUM2 neurons project to many common regions of the larval brain implicated in the control of feeding (Fig. 3). Future work will be needed to determine whether VUM1 neurons inhibit directly or indirectly the activity of VUM2 neurons.

Functional Parallels Between OA and Norepinephrine Systems.

We have obtained genetic and pharmacological evidence for the critical role of OA in the regulation of acquiring readily accessible sugar media. OA has been reported to mediate diverse neurobiological functions including appetitive memory formation and modulation of the dance of honey bee foragers to communicate floral or sucrose rewards (9, 37, 38). We postulate that the different OA receptors may mediate diverse OA-dependent behavioral responses to high-quality foods.

Norepinephrine (NE), the vertebrate counterpart of OA, has been shown to promote ingestion of carbohydrate-rich food at the beginning of a natural feeding cycle (39, 40). This feeding activity of NE resides in the paraventricular nucleus (PVN) of the feeding control center. In the PVN, α1 and α2 adrenergic receptors are organized in an antagonistic pattern (41). Activation of α1 receptor inhibits food intake (42), whereas activation of the α2 receptor stimulates food intake (43). Our results suggest that the insect OA system, like the NE system in mammals, exerts both positive and negative effects on the intake of preferred food. The activity of NE in PVN has been shown to antagonize that of 5-HT, which suppresses intake of carbohydrate-rich food (44). In Drosophila, 5-HT is also known to suppress feeding response (45). These findings suggest that the homeostatic control of intake of preferred food is likely mediated by a conserved neural network in flies and mammals.

The Role of Pvr in OA Neurons.

We have identified a unique role of Pvr in physiological regulation of hunger-motivated feeding of preferred food (liquid sugar media). The feeding-related activity of the Pvr pathway involves two regulatory proteins, Drk and Ras, and oral introduction of OA restores the hunger-driven feeding response in tdc2-Gal4/ UAS-drk^dsRNA larvae. Together, these results suggest that the Pvr pathway positively regulates OA release by tdc2-Gal4 neurons. Among the three identified ligands of Pvr (8, 29), Pvf2 is enriched in the larval CNS. Our finding suggests that Pvf2 regulates the feeding-related activity of the Pvr pathway. It is possible that Pvf2 may transduce a metabolic stimulus to Pvr/tdc2-Gal4 neurons that signals the energy state of larvae. In the honey bee brain, OA neurons from the SOG have been reported to respond to sugar stimulation (46, 47). Therefore, it would be interesting to test whether the Pvf2/Pvr pathway is responsive to sugar stimuli.

Our previous studies showed that the fly insulin and NPY-like systems coregulate hunger-elicited motivation to acquire solid sugar media (5, 7). We have now provided evidence that the fly VEGFR2- and NE-like systems control larval motivation to acquire liquid sugar media. These findings strongly suggest that the neural activities of different RTK systems play critical roles in different aspects of adaptive feeding decisions under various food and metabolic conditions. Therefore, further investigation of the mechanistic details of the food-related functions of RTK systems in the Drosophila model may provide novel insights into the neurobiology and evolution of appetitive control as well as pathophysiology of eating-related disorders.

Materials and Methods

Fly Strains, Media, and Larval Growth.

The fly rearing and the egg collections were performed as previously described (48). After a 2.5-h synchronized egg collection, eggs were kept in a 12-h light/dark cycle in an incubator at 25 °C. Larvae were transferred to a fresh apple juice plate with yeast paste at the age of 48–52 h (<80 larvae per plate). The fly lines used include tdc2-Gal4 (10), npf-Gal4, UAS-npf ^cDNA (4), UAS-Kir2.1 (11), UAS-shi^ts1 (12), tßh^nM18 (14), drk^R1, drk^∆P24 (26), tsh-Gal80 (21), UAS-drk^cDNA (discussed below), UAS-drk^dsRNA (49), GS-elav-Gal4 (50), UAS-Pvr^DN, UAS- Pvr^ACT (29), UAS-dTrpA1 (51), UAS-Octß3R^dsRNA, drk¹⁰⁶²⁶, actin-Gal4, UAS-GFP.nls, UAS-NaChBac, elav-Gal4, UAS-Ras^DN, UAS-dTrpA1, Pvf1^MB01242, Pvf2^d02444, Pvf3^EY09531, and Octß3R^MB04794 (Bloomington Drosophila Stock Center at Indiana University). The following lines were obtained from the Vienna Drosophila RNAi Center: UAS-Pvr^dsRNA (105353), UAS-Eph^dsRNA (4771), UAS-htl^dsRNA (27180), and UAS-Oamb^dsRNA (106511).

Transgenic Constructs.

The UAS-drk^cDNA was made by ligating drk^cDNA construct into the pUAST vector. The drk^cDNA clone was acquired from Berkeley Drosophila Genome Project. This vector was digested with EcoRI and XhoI (New England Biolabs), and the resulting 1,558-bp sequence covered the whole drk coding sequence. This sequence was subcloned into the EcoRI and Xhol sites of pUAST vector, which is at the downstream of the UAS promoters. Purified drkcDNA/pUAST was injected to w¹¹¹⁸ (BestGene Inc.).

Behavioral Assays.

The rate of larval food intake was quantified by following a previously published protocol with slight modification (7, 52). SI Materials and Methods details assay plate preparation. Early third-instar larvae (10–20 ) were transferred to the center of the assay plate, and then each plate was videotaped for 2 min. The number of MHCs per 30 s was scored and analyzed. The dynamic patterns of larval MHC were generated with a computer program using the MatLab software (MatLab Inc.). All assays were analyzed under blind conditions. At least three separate trials were conducted for each test. Statistical analyses were performed using one-way ANOVA followed by the Student–Newman–Keuls or Dunn test. The food ingestion assay was carried out by feeding a group of 30 larvae 10% (wt/vol) glucose liquid media containing 1% food dye FD&C No. 1 (Sigma-Aldrich) for 3 min. After rinsing with a copious amount of water, larvae were quickly frozen in liquid nitrogen and homogenized in 100 µL 0.1 M phosphate buffer (pH 7.2). The homogenates were centrifuged at 30,000 × g for 10 min, and the supernatants were analyzed spectrophotometrically for absorbance at 625 nm.

When the fly line UAS-shi^ts1 was used in the liquid food assay, the permissive temperature was room temperature, 23 °C, and the restrictive temperature was 30 °C. Twenty minutes before video recording, the experimental larva groups were prewarmed on a 30 °C heat plate. The liquid food plate was prewarmed on the heat plate for 3 min before use, and the videos were recorded on the heat plates. The control group assays were performed at 23 °C.

To activate GS-elav-Gal4, larvae were fed with deactivated yeast paste containing 1 mM RU486 (Cayman Chemical) and 4% (vol/vol) ethanol as solvent, and the mock group used deactivated yeast paste containing only 4% ethanol. Ten to 20 second-instar larvae were transferred to RU486 or ethanol yeast plates 20 h before assay.

Pharmacological Treatment.

OA stock solution was 500 mM OA in double-distilled H₂O. When used, the solution was diluted to desired concentration with deactivated yeast or with double-distilled H₂O, dependent on the larval food deprivation conditions. Larvae treated with OA were fed 5 mM or 10 mM OA, in yeast or water, 30 min before video recording.

Immunohistochemistry.

Brains from larvae 76 h after egg lay were dissected out and the immunostaining were performed as previously described (24) by using Chicken anti-GFP (1:1,000; Invitrogen), rabbit anti-ßgal (1:1,000; Promega), rabbit anti-tyromine-β-hydroxylase (1:500; gift from Vivian Budnik, University of Massachusetts Medical School, Worcester, MA), Alexa 488-goat anti-chicken (1:2,000; Invitrogen), Alexa Fluor-568 goat anti-rabbit (1:2,000; Invitrogen), and rhodamine conjugated phalloidin (1:1,000; Cytoskeleton). Images were collected using a Zeiss LSM510 META confocal microscope.

Targeted Laser Lesion.

The 337-nm nitrogen laser unit (337-USAS; Micro Point) was calibrated and adjusted as previously described (24). The laser lesion was performed using a previously published protocol with slight modification (6, 53). SI Materials and Methods gives more details.