Hedonist philosophers believed that pleasure is the supreme goal of life and that behavior is determined by the pursuit of it (Aristotle, 2004). Hunger for tasty delicacies, for example, is a particularly potent motivator of behavior, as it can be promoted not only by a physiological need to consume nutrients, but also by the hedonic properties of food. Due to its dichotomous nature, defining behavioral and neural correlates of hunger is a challenge that has perplexed scientists for over a century. Early physiologists wondered how to explain hunger and where it might originate, and they eventually described hunger as a dull ache or gnawing sensation arising from the stomach or abdominal region (Cannon and Washburn, 1993). This historical description has aided the discovery of hormones like ghrelin, insulin, and others that communicate information from the periphery to the brain and have roles in controlling feeding in response to energetic or nutritional needs, a state we now refer to as homeostatic hunger. In the decades since these discoveries, it has become clear that peripheral signals and others are integrated in the central nervous system to promote changes in behavior and reward signaling, which we interpret as indicators of hunger, but the nature of the motivational brain systems that drive these outputs remains less straightforward to identify and quantify.

Homeostatic feeding, or ‘need-based’ intake, is defined as food intake that is necessary to ensure caloric and nutritional needs are met, and it has been quantified by measuring bulk food intake. Such measures may be of total caloric intake or consumption of specific nutrients, but it is unclear whether these measures are representative of hunger that is driven by energetic need, nutrient deficiency, or some other motivation, which has led to variation in the way measures of homeostatic hunger are reported and interpreted (for review of mammalian brain circuits that modulate feeding, see Alcantara et al., 2022; Andermann and Lowell, 2017; Saper et al., 2002). However, the observation that most animals, including humans, regulate food intake around a protein target and the identification of discrete neuronal populations that regulate protein-specific feeding in model systems suggest that levels of homeostatic hunger may be primarily determined by the amount of protein an animal requires and thus, in recent years, protein intake has become a common method for measuring need-based hunger (Gosby et al., 2011; Liu et al., 2017; Ro et al., 2016; Simpson et al., 2003; Weaver et al., 2023).

Our understanding of hunger has continued to evolve, due in part to the observation that the pleasurable effects of food also stimulate feeding independently of physiological need, a state referred to as hedonic hunger (Lowe and Butryn, 2007). Hedonic, or ‘need-free’, intake is thought to be evoked by the incentive salience or tastiness of food, and thus, the sweetness of the food environment is predicted to play a larger role (Lowe and Butryn, 2007). Indeed, hedonic hunger is likely the predominant driver of feeding in modern human environments where food is readily available and energy deprivation is seldom experienced. Hedonic responses to various food or environmental manipulations have been described as either ‘liking’ responses (relating to the pleasure of a rewarding experience) or ‘wanting’ responses (relating to the motivation to seek a reward), which arise from different neuronal substrates (reviewed in Berridge, 2009). Accordingly, ‘liking’ is measured in rats by tracking facial responses in response to sweet tastants, while ‘wanting’ is measured by tracking latency to search for and consume a tasty food treat or by using the novelty suppressed feeding assay that attempts to measure an animal’s ability to overcome an aversive environment to consume a food reward (Arcego et al., 2020; Peciña et al., 2006). While observations like these have greatly aided our ability to identify hedonic hunger states in mammals, much less is known about their neuronal and mechanistic origins relative to homeostatic hunger.

Although we now understand that hunger stratifies into two distinct states, one driven by need and the other by pleasure, we have a limited understanding of their behavioral and neuronal distinctions and connections. In mammals, brain circuits that modulate homeostatic or hedonic appetites have been identified. The well-described AGRP neurons in the arcuate nucleus of the hypothalamus, for example, are critical components of a circuit for homeostatic feeding (for review, see Sternson and Eiselt, 2017). They express receptors for peripherally released hormones that signal energetic state, are more active upon food deprivation, and transmit a negative valence signal that drives animals to find and consume food when the need arises (Betley et al., 2015; Denis et al., 2015; Willesen et al., 1999; Zigman et al., 2006). In addition to systems that seem to motivate homeostatic feeding, distinct circuits have also been associated with hedonic feeding including the monoaminergic populations that express serotonin and/or dopamine, and the mesolimbic dopamine pathway (Ahn et al., 2022; Altherr et al., 2021; Denis et al., 2015; Ghiglieri et al., 1997; Rossi and Stuber, 2018). While there is evidence to suggest that both branches of hunger are modulated by homeostatic state and hedonic feedback (e.g., Denis et al., 2015; Krashes et al., 2011; Zigman et al., 2006), the mechanistic origins that determine how various hunger drives work together and independently to generate feeding responses have yet to be fully described.

Simpler model systems, such as the fruit fly, Drosophila melanogaster, have proven valuable for insights into the neural circuitry that modulates feeding, but the nature of their hunger drives is unknown. Whether flies experience hedonic hunger, for example, is unclear, although it seems likely that they do; hungry flies learn to associate odors with a sucrose reward and when fed a high-sugar diet they overeat and become obese (Huetteroth et al., 2015; May et al., 2019). These behaviors are all thought to occur through mechanisms involving dopaminergic neurons in the fly brain which, when artificially activated, promote foraging and food intake (Tsao et al., 2018). Honeybees have also exhibited signatures of food wanting, a component of hedonic hunger, which manifests through activation of dopaminergic populations in the brain that are reflective of reward expectation (Huang et al., 2022).

To better understand the different motive forces that drive feeding, it seems clear that there is a need to move beyond measures of food intake or foraging behavior, which have been useful for identifying external and internal factors that change the way an animal eats, in order to dissect when and to what degree animals may be experiencing different types of hunger. Here, we describe our first efforts to address this need by generating clear behavioral definitions of hunger states in Drosophila with the hope of providing an attractive system for defining the molecular features of their independent roles and the relationships between them. We characterize a repertoire of feeding micro-behaviors in the fly and use them to argue that Drosophila experience both homeostatic and hedonic hunger. We identify neural correlates of hedonic hunger in specific lobes of the Drosophila mushroom body (MB) and show that they are required for hedonic feeding. Finally, we develop a framework to examine the relationships between hedonic and homeostatic hunger states and present evidence for the existence of a system that acts as a potentiator of the relationship between them. Translating findings about the nature of feeding drives and the relationships among them could have far-reaching implications for obesity and aging research, and other behaviors influenced by motivational drives, including drug addiction and eating disorders.

We set out to complement and enrich traditional measures of feeding by identifying new behavioral metrics that might be useful for classifying different hunger states. Fly feeding behavior is typically described as changes in volumetric food intake, which is difficult to accurately quantify at the individual level due to the small amount of food ingested (Deshpande et al., 2014; Shell et al., 2021; Shell et al., 2018), or by using automated systems that detect physical interactions with food sources (Itskov et al., 2014; Ro et al., 2014). These automated devices have revealed rhythmic patterns of feeding behavior in flies that mimic feeding strategies in other animals, suggesting that the behaviors are highly organized and centrally regulated. They are, however, unable to distinguish different stereotypical actions that a freely moving fly might exhibit when interacting with a food source. Detailed behavioral characterizations of this sort have proven to be useful read-outs of neural states in other contexts and have the potential to inform our understanding of the motivational states that drive feeding (Chen et al., 2002; Duistermars et al., 2018; Ning et al., 2022). Much like human infants, for example, facial expressions and tongue movements in rats in response to appetitive tastants have been used to identify hedonic reactions and the brain systems that mediate them (Berridge, 1996).

To identify nuances in individual fly feeding, we developed video systems for manual and high-throughput annotation of what flies do when they eat. In all cases, we recorded freely moving flies placed in feeding chambers of the Fly Liquid-Food Interaction Counter (FLIC) system, which is designed to continuously record individual flies’ interactions with food sources (Ro et al., 2014). We began by simply recording flies in the FLIC and manually curating each video in its entirety (Figure 1A). With the hope of observing the greatest variety of behaviors, we collected observations from flies ranging from 5 to 20 days old, from mated female and male sated and starved flies, from flies housed singly or in groups of up to three, and from flies interacting with 5% or 10% sucrose food. We observed 30 flies and nearly 300 ‘events’, which were demarcated by intervals greater than 1 s during which the fly was not in physical contact with the food. Events were easy to identify, and different behavioral characteristics associated with events were easy to distinguish from each other (Slooten, 1994). Rare events (observed in fewer than 5% of flies) were not considered further, and events during which sight of the fly was out of focus or obscured (typically 1–2 events per fly) were ignored. Based on the remaining events, we created a library of user-defined behaviors that captured common, distinct, and definable interactions of flies with their food (Table 1).

Figure 1 with 1 supplement see all

Download asset Open asset

Figure 1—source data 1

Excel spreadsheet containing source data used to generate Figures 1B-D.

https://cdn.elifesciences.org/articles/84537/elife-84537-fig1-data1-v1.xls

Download elife-84537-fig1-data1-v1.xls

We classified the behaviors we observed into five main categories: two non-feeding behaviors (‘other’ and ‘interaction’ events) and three feeding behaviors (‘fast’ feeding events, ‘long’ feeding events, and ‘long/quick’ feeding events) (Table 1). The first feeding event type comprised fast feeding events (termed ‘F’ or ‘FI’) during which we observed contact of the proboscis with the food for less than 4 s. These types of events were common (occurring in ~95% of flies, Figure 1B and C) and were sometimes simultaneous with the interaction of a fly’s legs with the food. Longer feeding events (termed ‘L’, ‘LI’, or ‘LQ’) were also identified, during which time we observed contact of the proboscis with the food for more than 4 s. Long events were less common than F-type events and could also include simultaneous interaction of legs with food. Long feeding events were either a continuous and sustained interaction of the proboscis with the food (‘L’ and ‘LI’, occurring in ~85% of flies, Figure 1B and C) or, less commonly, a continuous interaction during which the proboscis moved quickly in and out of the food (‘LQ’, occurring in ~60% of flies, Figure 1B and C). The first type of non-feeding behavior was termed an interaction event (‘IF’ or ‘IH’) during which the fly interacted with the food only using its legs; no proboscis contact was observed. These putative tasting interactions often occurred in a ‘pitter-patter’ motion, as if the fly were ‘playing’ with their food, and they occurred in greater than 90% of flies (Figure 1B and C). Lastly, we occasionally observed other non-feeding events (‘G’ or ‘UC’) that, nevertheless, often occurred just before or after feeding (occurring in ~60% of flies, Figure 1B and C). These tended to be various grooming behaviors, including using the forelegs to rub the head after interacting with food.

We next asked whether our behavioral classification could be used to distinguish hungry from sated flies. Recognizing a need to significantly increase the number of observed feeding events to provide sufficient statistical power to detect differences in behavioral tendencies, we designed a direct communication link between our custom DTrack video software and the FLIC system to trigger video capture only when there was physical contact between a fly and the food source, together with 5 s windows before and after the contact. This allowed us to create video compilations of feeding behaviors for fully fed or starved (24 hr) flies over several hours. Over 500 distinct events in these compilations were then manually scored using our classification of feeding behaviors, and comparisons were made between fully fed and starved flies for each behavioral category in Table 1. We observed that starved flies exhibited a greater frequency of feeding behaviors overall, as expected (sated event frequency = 0.606 events/min compared to starved event frequency = 0.788 events/min, p=0.061, data not shown), and that the relative frequency with which L-type feeding events occurred was increased (Figure 1D, quantified at right). The observation that starved flies exhibit more frequent long-duration events is consistent with recent reports that starved flies increase ‘sips per burst’ – a measure that is similar to our definition of L and LQ events (Itskov et al., 2014).

Qualitative aspects of the relationships between individual behaviors – that is, the architecture that results from sequences of behaviors – can reveal complexities that may also distinguish between different neural states (Casarrubea et al., 2019; Casarrubea et al., 2015; Flavell et al., 2022). Thus, we asked whether certain sequential patterns in fly feeding behavior appear repeatedly and whether they are affected by starvation. To do this, we used classic sequence analysis to map the probability that one event follows another in a sequence more often than what is expected to occur by chance alone – an approach that has been useful for identifying stereotyped patterns or structures of behaviors that tend to occur together (Ning et al., 2022; Slooten, 1994). We observed that hungry flies, whom we had starved for 24 hr, but not sated flies, followed F feeding events with additional F feeding events (transitional probability = 0.6 in starved flies with p-value <0.001, compared to a non-significant transitional probability = 0.55 in fully fed flies) and L-type feeding events with additional L-type feeding events (transitional probability = 0.439 in starved flies with p-value <0.001 compared to non-significant transitional probability = 0.1538 in fully fed flies) more often than would be expected by chance (Figure 1E). In contrast, fully fed flies were likely to engage in F-type events after grooming or interaction events and were more likely to groom following most other events, which may be reflective of a less goal-oriented behavioral state compared to that of starved flies (Figure 1E).

Having established that long-duration events occur more frequently and in a predictable sequence in starved flies, we sought to replicate this finding in flies during their normal circadian ‘meal-time’, when they are presumably physiologically hungry but not yet starved. When monitored over several days on the FLIC, flies exhibited circadian meal patterns that corresponded to the light cycle, with one peak in the morning when lights turn on and a second peak in the evening when lights turn off (e.g., Figure 1—figure supplement 1). We obtained 24 hr video compilations of nearly 700 feeding events in the FLIC and used the flies’ endogenous feeding rhythm to designate ‘meal’ and ‘non-meal’ times of day. We manually scored the videos during each time for feeding events (fast, long, and long/quick events). We found that during meal-times flies exhibited a significantly greater frequency of LQ events, while the frequencies of F and L events were statistically indistinguishable between times of day (Figure 1F). Thus, both starved flies and flies during their circadian meal-time exhibit an increase in the total number of feeding events together with an increase in the relative frequency of long feeding events.

The FLIC generates several quantitative signal measures when flies interact with food, and we asked whether any of these numerical metrics might substitute for direct visual observation and distinguish among fast feeding events, long feeding events, and interactions. To determine how signal measures change in accordance with a fly’s behavior, we scored 15 video compilations (resulting in 332 independent events) of flies interacting on the FLIC for F-type, L-type, or I-type events, and we matched each visually identified behavior to its corresponding FLIC signal data (e.g., Figure 2A–C). We evaluated differences in the characteristics of signal intensity for each event (i.e., its variance, minimum, maximum, average, and total) as well as the duration of the signal from each event. We found that only signal duration effectively distinguished long events from both fast events and interactions (Figure 2D–I). Thus, we established that longer signal durations from the FLIC are representative of L-type feeding events. We also noted that maximum signal intensity and the variance of the signal can distinguish interactions from both types of feeding events, which may be broadly useful to distinguish non-feeding interactions (i.e., tasting events) from feeding events (Ro et al., 2014).

Figure 2

Download asset Open asset

Figure 2—source data 1

Excel spreadsheet containing source data used to generate Figures 2D-I.

https://cdn.elifesciences.org/articles/84537/elife-84537-fig2-data1-v1.xls

Download elife-84537-fig2-data1-v1.xls

Development of an assay to identify homeostatic and hedonic feeding

Having established that hungry flies increase the duration of their feeding events as well as the total number of events (i.e., Figure 3A, top panel), we next asked if we might be able to identify distinct hunger drives in flies and, if so, whether our behavioral and quantitative metrics would distinguish between them. Recognizing that the original assays used for behavioral observation, in which flies were exposed to a single, homogenous food source, were likely insufficient to distinguish different types of feeding drives, we designed two types of food choice environments that we might expect would distinguish homeostatic from hedonic feeding. Homeostatic feeding drive in flies is thought to be measured by the extent to which flies choose to consume a yeast-containing food over one consisting of sucrose only, which we will refer to hereafter as a yeast preference (Liu et al., 2017; Ro et al., 2016). Therefore, for measures of homeostatic drive, we exposed flies to a choice of 2% sucrose vs. [2% yeast+1% sucrose], and we focused our analysis on the flies’ relative interactions with the yeast containing food. To measure hedonic drive, on the other hand, we incorporated theorized requirements to experimentally measure hedonic hunger in mammals: animals must be energy replete and they must be tested on highly palatable food (Lowe and Butryn, 2007). We therefore exposed flies to a ‘hedonic choice’ environment which mimicked the ‘homeostatic choice’ environment except that the 2% sucrose food was replaced with sweeter 20% sucrose, and we focused our analysis on comparing interactions with the sucrose food between the food environments (e.g., Figure 3A, bottom panel). This design ensured flies remained sated by access to yeast while also allowing feeding on the highly palatable 20% sucrose solution.

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Excel spreadsheet containing source data used to generate Figures 3B-M and Figure 3 - Figure Supplement 1A-C.

https://cdn.elifesciences.org/articles/84537/elife-84537-fig3-data1-v1.xls

Download elife-84537-fig3-data1-v1.xls

We observed that the 2% sucrose environment elicited an increased yeast preference in starved compared to fully fed female flies (Figure 3B), reinforcing the notion that the total number of events with yeast vs. sucrose can be used as a relative measure of homeostatic hunger. We also observed an increase in yeast preference in fully fed female compared to male flies (Figure 3C), which may be reflective of a known difference in protein requirements, and therefore homeostatic hunger levels, between male and female flies. We next evaluated female and male fully fed flies in the 20% sucrose food environment compared to the 2% sucrose environment, which elicited increased event durations with 20% sucrose compared to the 2% sucrose food (Figure 3D) and also elicited decreased event durations with yeast food (Figure 3E). The total number of events on sucrose or yeast wells was statistically indistinguishable between food environments (Figure 3F and G), suggesting that modulation of event duration, but not total number of events, may be a feature of hedonic food environments. We also measured sucrose intake in the control vs. hedonic environments using a method that determines how much volume a group of flies consume in 24 hr (termed ‘Con-Ex’ for consumption-excretion) (Shell et al., 2021; Shell et al., 2018), and found that flies consume twice as much sucrose volume in the hedonic environment (Figure 3—figure supplement 1A). We concluded that the measurable increase in sucrose event durations in the hedonic food environment may reflect increases in a hedonic hunger drive while total number of events on yeast vs. sucrose food may reflect homeostatic state.

To challenge our measure of hedonic feeding, we reasoned that if increased event durations were indeed a result of hedonic hunger, then sweetness alone should be sufficient to increase event durations without modulating measures of homeostatic feeding. We tested two additional food choice environments where we altered either the sweetness or caloric content of the sugar choice. Sweetness was manipulated by using the non-metabolizable sweetener arabinose ([2% sucrose+18% arabinose]), which mimics the caloric and nutritive content of the 2% sucrose but contains the sweetness of the 20% sucrose choice. Caloric content was manipulated using the nutritive, but not sweet, sugar alcohol, sorbitol ([2% sucrose+18% sorbitol]), which mimics the caloric content of the 20% sucrose but contains the sweetness of the 2% sucrose (Burke and Waddell, 2011; Hassett, 1948). We observed that the sweet arabinose food environment elicited increased event durations that were comparable to flies in the 20% sucrose environment, with no statistical difference in total event number (Figure 3H and I, Figure 3—figure supplement 1B) and that the caloric but not sweet sorbitol food environment was not sufficient to increase event durations or total event number (Figure 3J and K). Finally, we tested several additional food choice environments containing a range of sucrose concentrations and observed that event durations were shortest when the sucrose choice contained only water or 1% sucrose and increased linearly as the sucrose concentration increased, while event durations with yeast food and total number of sucrose events were not statistically different (Figure 3L and M, Figure 3—figure supplement 1C). Thus, sweet taste was sufficient to evoke longer feeding event durations, while calories alone were not, indicating that this metric obtained from our hedonic food choice paradigm may reliably quantify hedonic feeding.

Whether and how homeostatic and hedonic feeding mechanisms interact or overlap is controversial, and thus, we asked whether our metrics of homeostatic and hedonic hunger are correlated among individual flies. To obtain flies with a range of homeostatic needs, we used fully fed and starved flies to obtain feeding measures in our control or hedonic food choice environments. Starvation elicited an increase in yeast interactions compared to 2% sucrose (Figure 4A) that was more pronounced in flies starved for 48 vs 24 hr (Figure 4B), which is consistent with our expectation for increased homeostatic drive. We found that starvation seemed to marginally increase differences in sucrose event durations in hedonic vs. control environments, indicating that homeostatic hunger may enhance hedonic feeding (Figure 4C). For each fly, we next calculated metrics for homeostatic drive (# interactions with yeast food) as well as hedonic drive (sucrose event durations) and plotted their relationship along the x- and y-axes, respectively. We reasoned that if the circuits or programs that drive hedonic and homeostatic feeding were completely independent, then we should observe a slope close to zero. A significantly positive or negative slope, however, would suggest that increasing physiological need might also modulate hedonic hunger, or vice versa. We observed a significantly positive correlation between our measures of homeostatic and hedonic feeding (Figure 4D), suggesting that increasing physiological need might also potentiate hedonic hunger, or vice versa, which is consistent with the observation in mammals that hunger increases the reward value of food (Fulton, 2010). Interestingly, we also noted that separating regression lines from flies in the control vs. hedonic food environments revealed parallel lines whose slopes were not significantly different but were separated by a shift along the y-axis, indicating again that a sweeter food environment promotes longer event durations but may not modulate the relationship between hunger drives. These findings indicate that homeostatic state is a potentiator of hedonic feeding and establish that inferences of hunger states and measures of feeding should include considerations of each component independently as well as their interaction.

Figure 4

Download asset Open asset

Figure 4—source data 1

Excel spreadsheet containing source data used to generate Figures 4A-D.

https://cdn.elifesciences.org/articles/84537/elife-84537-fig4-data1-v1.xls

Download elife-84537-fig4-data1-v1.xls

Identification of neural correlates of hedonic hunger

A good deal is known about neural circuits associated with homeostatic hunger in Drosophila, and we wondered whether our behavioral and quantitative classification metrics could aid in identification of representative neural correlates of a hedonic hunger state. We began by conducting an unbiased screen for neurons that respond to hedonic food environments using the genetically encoded calcium indicator CaMPARI driven by a pan-neuronal GAL4 driver (Nsyb-GAL4). CaMPARI is a green fluorescent protein that irreversibly photoconverts to red when flies are simultaneously exposed to increases in intracellular calcium and UV light provided by the experimenter (Edwards et al., 2020; Fosque et al., 2015). We placed groups of three naïve, fully fed flies in the hedonic food choice environment or the control food environment and allowed them to habituate for 1 hr. Flies in each chamber were then exposed to cycles of UV light (10 s on/10 s off) for a total of 1 hr to label neuronal populations that were differentially active when flies are in the hedonic vs. the control food environment. Brains were dissected and imaged immediately, and the ratio of red:green fluorescent protein was analyzed, with a higher ratio indicating increased neuronal activity. We observed that the MB had a higher ratio of red:green fluorescence when flies were in the hedonic food environment, indicating that this population was more active (Figure 5A).

Figure 5 with 13 supplements see all

Download asset Open asset

Figure 5—source data 1

Excel spreadsheet containing source data used to generate Figures 5A-G and Figure 5 - Figure Supplements 1A-C.

https://cdn.elifesciences.org/articles/84537/elife-84537-fig5-data1-v1.xls

Download elife-84537-fig5-data1-v1.xls

To validate the findings of our unbiased screen and as a direct test of MB activity in hedonic food environments, we restricted the expression of CaMPARI to the MB using the 238Y enhancer-trap line (Yang et al., 1995). We observed that the hedonic food environment increased red:green fluorescence in the horizontal MB lobes compared to the control food environment (Figure 5B). The Drosophila MB is a well-described structure known to be involved in reinforcement and motivational control and is innervated by dopaminergic populations that are thought to encode reinforcing properties of food, including relaying information about sweet taste (Aso et al., 2009; Aso et al., 2014; Das et al., 2016). Thus, our observation that MB neurons are activated in hedonic food environments is consistent with prior knowledge of their roles in motivation and reward (Landayan et al., 2018).

To determine whether MB neurons are involved in hedonic hunger beyond correlative increases in activity, we tested for their requirement in hedonic and homeostatic feeding. We used the genetically encoded optogenetic tool GtACR to selectively inhibit MB neurons in our food choice assay. GtACR is a channelrhodopsin that is activated upon green light exposure and increases chloride conductance, resulting in hyperpolarization and silencing of neuronal populations (Mohammad et al., 2017). We expressed GtACR in either the γ, α/β, or α’/β’ lobes of the MB (Aso et al., 2014) and found that communication from the γ and α/β lobes is dispensable for both hedonic and homeostatic feeding (Figure 5C and D, Figure 5—figure supplement 1A and B). In contrast, signaling specifically from the α’/β’ lobes was required for hedonic, but not homeostatic, feeding (Figure 5E, Figure 5—figure supplement 1C). The MB horizontal lobes are innervated by dopaminergic neurons from the protocerebral anterior medial (PAM) cluster, which were also required for hedonic feeding (Figure 5F; Mao and Davis, 2009). These results indicate that a distinct PAM to α’/β’ MB circuit regulates hedonic feeding, which is consistent with known descriptions of anatomical and functional segregation of MB lobes (Aso et al., 2014; Landayan et al., 2018).

We next sought to determine whether communication from the α’/β’ lobe was required for the potentiating effects of homeostatic hunger on hedonic feeding. We reasoned that if the α’/β’ lobe also functions as a potentiator between the two hunger states, then inhibition of these neurons would be expected to prevent increases in hedonic feeding when flies are starved. Instead, we observed that starvation restored hedonic feeding in flies whose α’/β’ lobes were inhibited (Figure 5G), suggesting that potentiation of hedonic feeding by homeostatic state involves the recruitment of additional and independent circuits for hedonic feeding.

Animals exquisitely regulate feeding according to internal needs, but they are also subject to their desires. Thus, food intake only sometimes occurs in response to homeostatic hunger and can also occur in response to emotional stress, social environment, illness, and pursuit of reward (Siemian et al., 2021; Xu et al., 2019). Because animals eat for many reasons other than hunger, food intake alone may be a poor indicator of hunger states and may obscure investigations into the motivational forces that drive behavior. Here, we observed the feeding micro-behaviors of flies and developed a behavioral classification system to assist in the identification of feeding drives. We established that flies exhibit behavioral signatures of two hunger states, homeostatic and hedonic, and we used them to identify neurocircuits that are required for hedonic, but not homeostatic, feeding. We also showed that homeostatic and hedonic feeding are both independent and overlapping depending on environmental context, and we developed a simple interaction framework that researchers can use to investigate these relationships in detail. The development of this paradigm presents entry points into the discovery of new homeostatic and hedonic hunger circuits in flies, better classification of feeding circuits which have already been described, and dissection of mechanistic relationships among them.

The Drosophila MB supports a wide variety of behaviors and functions, in part due to its complexity and heterogeneity in anatomical structure (Aso et al., 2009). The lobes of the MB are formed by the interconnected axons of ~2000 Kenyon cells and 34 MB output neurons, and they receive modulatory input from ~20 dopaminergic cell types (Aso et al., 2014). The MB lobes have been anatomically and functionally divided into three classes: γ, α/β, and α’/β’, which have been further divided into functionally distinct compartments (Aso et al., 2014; Crittenden et al., 1998; Tanaka et al., 2008). Our data suggest that compartments of the α’/β’ MB lobes, but not the γ or α/β lobes, are required for hedonic feeding. Although our experiments isolated a discrete location within the functionally heterogenous MB, the α’/β’ lobe still contains more than 350 Kenyon cells that are divided into middle (m) and anterior-posterior (ap) compartments (Aso et al., 2014). Future investigations that utilize existing and emerging intersectional technologies will be useful for discerning the specific cells within the α’/β’ MB lobes that promote hedonic feeding, as well as their modulatory inputs and outputs, and for evaluation of the cell-intrinsic mechanisms that determine how hedonic hunger is encoded.

Connections from dopaminergic populations in the PAM cluster with MB-intrinsic Kenyon cells have been implicated in appetitive learning, signaling reward in response to sugar, motivation to overcome environmental stress, and promoting wakefulness (Burke et al., 2012; Haynes et al., 2015; Hermanns et al., 2022; Liu et al., 2012). PAM projections to Kenyon cell axons in the β’ compartment of the MB respond to sugar by increasing calcium transients, which has led to the idea that these synapses transmit information about sweet taste (Liu et al., 2012; May et al., 2020). Interestingly, these responses are increased twofold by starvation, suggesting that they scale in response to environmental context or degree of need (Liu et al., 2012), and artificial activation of PAM neurons also increases foraging behavior (Tsao et al., 2018). Our finding that the PAM>α’/β’ circuit is required for promoting hedonic hunger builds upon a rich body of literature, and together these observations offer a model that may unite the many described functions of this node. One possibility is that this circuit acts as a permissive or motive force that gates or fuels downstream mechanisms involved in behavioral responses to environmental stimuli. Indeed, learning, reward, and wakefulness are all processes that might benefit from or require a force that sustains them. This view offers an attractive entry into understanding the seat of motivation in the fly brain.

Decades of research on the neural control of feeding have resulted in the identification of mammalian neurocircuits involved in the control of feeding, and technological advances have allowed researchers to manipulate them with an increasing degree of specificity. We are, however, approaching the limits of what we can learn from manipulations and technologies of these sorts and thus, complimenting them with discoveries from model systems in which more specific manipulations can be tested may accelerate our ability to understand how the brain generates motivated states. The hedonic circuits we have identified here in flies, for example, can be immediately investigated at multiple levels to assess the neuronal and molecular mechanisms that determine their role in feeding, including spatiotemporal manipulation of cell-signaling pathways, electrophysiological recordings in live animals in response to various food environments or energetic deficits, and optogenetic manipulation paired with close behavioral monitoring. These types of experiments would fulfill several main goals in the study of the neural control of appetite regulation (Ahn et al., 2022) and, despite differences in the precise identity of feeding neurocircuits between flies and mammals, discoveries about foundational principles and conserved signaling pathways may inform and direct mammalian studies.

All data generated or analyses during this study are included in the manuscript and supporting files. FLIC data were analyzed using custom R code, which is available at https://github.com/PletcherLab/FLIC_R_Code (copy archived at swh:1:rev:9713cfeb88c26e7aa4c0d87b26fcf3361500c670).

1. 1. Ahn BH
   2. Kim M
   3. Kim SY

   (2022) Brain circuits for promoting Homeostatic and non-Homeostatic appetites

   Experimental & Molecular Medicine 54:349–357.

   https://doi.org/10.1038/s12276-022-00758-4

   • PubMed
   • Google Scholar
2. 1. Alcantara IC
   2. Tapia APM
   3. Aponte Y
   4. Krashes MJ

   (2022) Acts of appetite: neural circuits governing the Appetitive, Consummatory, and terminating phases of feeding

   Nature Metabolism 4:836–847.

   https://doi.org/10.1038/s42255-022-00611-y

   • PubMed
   • Google Scholar
3. 1. Altherr E
   2. Rainwater A
   3. Kaviani D
   4. Tang Q
   5. Güler AD

   (2021) Long-term high fat diet consumption Reversibly alters feeding behavior via a dopamine-associated mechanism in mice

   Behavioural Brain Research 414:113470.

   https://doi.org/10.1016/j.bbr.2021.113470

   • PubMed
   • Google Scholar
4. 1. Andermann ML
   2. Lowell BB

   (2017) Toward a wiring diagram understanding of appetite control

   Neuron 95:757–778.

   https://doi.org/10.1016/j.neuron.2017.06.014

   • PubMed
   • Google Scholar
5. 1. Arcego DM
   2. Krolow R
   3. Lampert C
   4. Toniazzo AP
   5. Garcia EDS
   6. Lazzaretti C
   7. Costa G
   8. Scorza C
   9. Dalmaz C

   (2020) Chronic high-fat diet affects food-motivated behavior and Hedonic systems in the nucleus accumbens of male rats

   Appetite 153:104739.

   https://doi.org/10.1016/j.appet.2020.104739

   • PubMed
   • Google Scholar
6. Book

   1. Aristotle T

   (2004)

   The Nicomachean Ethics

   Penguin Books.

   • Google Scholar
7. 1. Aso Y
   2. Grübel K
   3. Busch S
   4. Friedrich AB
   5. Siwanowicz I
   6. Tanimoto H

   (2009) The mushroom body of adult Drosophila characterized by Gal4 drivers

   Journal of Neurogenetics 23:156–172.

   https://doi.org/10.1080/01677060802471718

   • PubMed
   • Google Scholar
8. 1. Aso Y
   2. Hattori D
   3. Yu Y
   4. Johnston RM
   5. Iyer NA
   6. Ngo T-TB
   7. Dionne H
   8. Abbott LF
   9. Axel R
   10. Tanimoto H
   11. Rubin GM

   (2014) The neuronal architecture of the mushroom body provides a logic for associative learning

   eLife 3:e04577.

   https://doi.org/10.7554/eLife.04577

   • PubMed
   • Google Scholar
9. 1. Berridge KC

   (1996) Food reward: brain substrates of wanting and liking

   Neuroscience and Biobehavioral Reviews 20:1–25.

   https://doi.org/10.1016/0149-7634(95)00033-b

   • PubMed
   • Google Scholar
10. 1. Berridge KC

    (2009) Liking' and 'wanting' food rewards: brain substrates and roles in eating disorders

    Physiology & Behavior 97:537–550.

    https://doi.org/10.1016/j.physbeh.2009.02.044

    • PubMed
    • Google Scholar
11. 1. Betley JN
    2. Xu S
    3. Cao ZFH
    4. Gong R
    5. Magnus CJ
    6. Yu Y
    7. Sternson SM

    (2015) Neurons for hunger and thirst transmit a negative-Valence teaching signal

    Nature 521:180–185.

    https://doi.org/10.1038/nature14416

    • PubMed
    • Google Scholar
12. Book

    1. Blumstein DT
    2. Daniel JC

    (2007)

    Quantifying Behavior the JWatcher Way

    Sinauer Associates.

    • Google Scholar
13. 1. Burke CJ
    2. Waddell S

    (2011) Remembering nutrient quality of sugar in Drosophila

    Current Biology 21:746–750.

    https://doi.org/10.1016/j.cub.2011.03.032

    • PubMed
    • Google Scholar
14. 1. Burke CJ
    2. Huetteroth W
    3. Owald D
    4. Perisse E
    5. Krashes MJ
    6. Das G
    7. Gohl D
    8. Silies M
    9. Certel S
    10. Waddell S

    (2012) Layered reward signalling through Octopamine and dopamine in Drosophila

    Nature 492:433–437.

    https://doi.org/10.1038/nature11614

    • PubMed
    • Google Scholar
15. 1. Cannon WB
    2. Washburn AL

    (1993) An explanation of hunger. 1911

    Obesity Research 1:494–500.

    https://doi.org/10.1002/j.1550-8528.1993.tb00033.x

    • PubMed
    • Google Scholar
16. 1. Casarrubea M
    2. Jonsson GK
    3. Faulisi F
    4. Sorbera F
    5. Di Giovanni G
    6. Benigno A
    7. Crescimanno G
    8. Magnusson MS

    (2015) T-pattern analysis for the study of temporal structure of animal and human behavior: a comprehensive review

    Journal of Neuroscience Methods 239:34–46.

    https://doi.org/10.1016/j.jneumeth.2014.09.024

    • PubMed
    • Google Scholar
17. 1. Casarrubea M
    2. Aiello S
    3. Di Giovanni G
    4. Santangelo A
    5. Palacino M
    6. Crescimanno G

    (2019) Combining quantitative and qualitative data in the study of feeding behavior in male Wistar rats

    Frontiers in Psychology 10:881.

    https://doi.org/10.3389/fpsyg.2019.00881

    • PubMed
    • Google Scholar
18. 1. Chen S
    2. Lee AY
    3. Bowens NM
    4. Huber R
    5. Kravitz EA

    (2002) Fighting fruit flies: a model system for the study of aggression

    PNAS 99:5664–5668.

    https://doi.org/10.1073/pnas.082102599

    • PubMed
    • Google Scholar
19. 1. Crittenden JR
    2. Skoulakis EM
    3. Han KA
    4. Kalderon D
    5. Davis RL

    (1998)

    Tripartite mushroom body architecture revealed by Antigenic markers

    Learning & Memory 5:38–51.

    • PubMed
    • Google Scholar
20. 1. Das G
    2. Lin S
    3. Waddell S

    (2016) Remembering components of food in Drosophila

    Frontiers in Integrative Neuroscience 10:4.

    https://doi.org/10.3389/fnint.2016.00004

    • PubMed
    • Google Scholar
21. 1. Denis RGP
    2. Joly-Amado A
    3. Webber E
    4. Langlet F
    5. Schaeffer M
    6. Padilla SL
    7. Cansell C
    8. Dehouck B
    9. Castel J
    10. Delbès A-S
    11. Martinez S
    12. Lacombe A
    13. Rouch C
    14. Kassis N
    15. Fehrentz J-A
    16. Martinez J
    17. Verdié P
    18. Hnasko TS
    19. Palmiter RD
    20. Krashes MJ
    21. Güler AD
    22. Magnan C
    23. Luquet S

    (2015) Palatability can drive feeding independent of Agrp neurons

    Cell Metabolism 22:646–657.

    https://doi.org/10.1016/j.cmet.2015.07.011

    • PubMed
    • Google Scholar
22. 1. Deshpande SA
    2. Carvalho GB
    3. Amador A
    4. Phillips AM
    5. Hoxha S
    6. Lizotte KJ
    7. Ja WW

    (2014) Quantifying Drosophila food intake: comparative analysis of current methodology

    Nature Methods 11:535–540.

    https://doi.org/10.1038/nmeth.2899

    • PubMed
    • Google Scholar
23. 1. Duistermars BJ
    2. Pfeiffer BD
    3. Hoopfer ED
    4. Anderson DJ

    (2018) A brain Module for Scalable control of complex, multi-motor threat displays

    Neuron 100:1474–1490.

    https://doi.org/10.1016/j.neuron.2018.10.027

    • PubMed
    • Google Scholar
24. 1. Edwards KA
    2. Hoppa MB
    3. Bosco G

    (2020) The Photoconvertible fluorescent probe, Campari, labels active neurons in freely-moving intact adult fruit flies

    Frontiers in Neural Circuits 14:22.

    https://doi.org/10.3389/fncir.2020.00022

    • PubMed
    • Google Scholar
25. 1. Flavell SW
    2. Gogolla N
    3. Lovett-Barron M
    4. Zelikowsky M

    (2022) The emergence and influence of internal States

    Neuron 110:2545–2570.

    https://doi.org/10.1016/j.neuron.2022.04.030

    • PubMed
    • Google Scholar
26. 1. Fosque BF
    2. Sun Y
    3. Dana H
    4. Yang CT
    5. Ohyama T
    6. Tadross MR
    7. Patel R
    8. Zlatic M
    9. Kim DS
    10. Ahrens MB
    11. Jayaraman V
    12. Looger LL
    13. Schreiter ER

    (2015) Neural circuits. labeling of active neural circuits in vivo with designed calcium Integrators

    Science 347:755–760.

    https://doi.org/10.1126/science.1260922

    • PubMed
    • Google Scholar
27. 1. Fulton S

    (2010) Appetite and reward

    Frontiers in Neuroendocrinology 31:85–103.

    https://doi.org/10.1016/j.yfrne.2009.10.003

    • PubMed
    • Google Scholar
28. 1. Ghiglieri O
    2. Gambarana C
    3. Scheggi S
    4. Tagliamonte A
    5. Willner P
    6. De Montis MG

    (1997) Palatable food induces an Appetitive behaviour in Satiated rats which can be inhibited by chronic stress

    Behavioural Pharmacology 8:619–628.

    https://doi.org/10.1097/00008877-199711000-00018

    • PubMed
    • Google Scholar
29. 1. Gosby AK
    2. Conigrave AD
    3. Lau NS
    4. Iglesias MA
    5. Hall RM
    6. Jebb SA
    7. Brand-Miller J
    8. Caterson ID
    9. Raubenheimer D
    10. Simpson SJ

    (2011) Testing protein leverage in lean humans: a randomised controlled experimental study

    PLOS ONE 6:e25929.

    https://doi.org/10.1371/journal.pone.0025929

    • PubMed
    • Google Scholar
30. 1. Hassett CC

    (1948)

    The utilization of sugars and other substances by Drosophila

    The Biological Bulletin 95:114–123.

    • PubMed
    • Google Scholar
31. 1. Haynes PR
    2. Christmann BL
    3. Griffith LC

    (2015) A single pair of neurons links sleep to memory consolidation in Drosophila melanogaster

    eLife 4:e03868.

    https://doi.org/10.7554/eLife.03868

    • PubMed
    • Google Scholar
32. 1. Hermanns T
    2. Graf-Boxhorn S
    3. Poeck B
    4. Strauss R

    (2022) Octopamine mediates sugar relief from a chronic-stress-induced depression-like state in Drosophila

    Current Biology 32:4048–4056.

    https://doi.org/10.1016/j.cub.2022.07.016

    • PubMed
    • Google Scholar
33. 1. Huang J
    2. Zhang Z
    3. Feng W
    4. Zhao Y
    5. Aldanondo A
    6. de Brito Sanchez MG
    7. Paoli M
    8. Rolland A
    9. Li Z
    10. Nie H
    11. Lin Y
    12. Zhang S
    13. Giurfa M
    14. Su S

    (2022) Food wanting is mediated by transient activation of dopaminergic signaling in the honey bee brain

    Science 376:508–512.

    https://doi.org/10.1126/science.abn9920

    • PubMed
    • Google Scholar
34. 1. Huetteroth W
    2. Perisse E
    3. Lin S
    4. Klappenbach M
    5. Burke C
    6. Waddell S

    (2015) Sweet taste and nutrient value Subdivide rewarding dopaminergic neurons in Drosophila

    Current Biology 25:751–758.

    https://doi.org/10.1016/j.cub.2015.01.036

    • PubMed
    • Google Scholar
35. 1. Itskov PM
    2. Moreira JM
    3. Vinnik E
    4. Lopes G
    5. Safarik S
    6. Dickinson MH
    7. Ribeiro C

    (2014) Automated monitoring and quantitative analysis of feeding behaviour in Drosophila

    Nature Communications 5:4560.

    https://doi.org/10.1038/ncomms5560

    • PubMed
    • Google Scholar
36. 1. Krashes MJ
    2. Koda S
    3. Ye C
    4. Rogan SC
    5. Adams AC
    6. Cusher DS
    7. Maratos-Flier E
    8. Roth BL
    9. Lowell BB

    (2011) Rapid, reversible activation of Agrp neurons drives feeding behavior in mice

    The Journal of Clinical Investigation 121:1424–1428.

    https://doi.org/10.1172/JCI46229

    • PubMed
    • Google Scholar
37. 1. Landayan D
    2. Feldman DS
    3. Wolf FW

    (2018) Satiation state-dependent dopaminergic control of foraging in Drosophila

    Scientific Reports 8:5777.

    https://doi.org/10.1038/s41598-018-24217-1

    • PubMed
    • Google Scholar
38. 1. Liu C
    2. Plaçais PY
    3. Yamagata N
    4. Pfeiffer BD
    5. Aso Y
    6. Friedrich AB
    7. Siwanowicz I
    8. Rubin GM
    9. Preat T
    10. Tanimoto H

    (2012) A subset of dopamine neurons signals reward for odour memory in Drosophila

    Nature 488:512–516.

    https://doi.org/10.1038/nature11304

    • PubMed
    • Google Scholar
39. 1. Liu Q
    2. Tabuchi M
    3. Liu S
    4. Kodama L
    5. Horiuchi W
    6. Daniels J
    7. Chiu L
    8. Baldoni D
    9. Wu MN

    (2017) Branch-specific plasticity of a Bifunctional dopamine circuit Encodes protein hunger

    Science 356:534–539.

    https://doi.org/10.1126/science.aal3245

    • PubMed
    • Google Scholar
40. 1. Lowe MR
    2. Butryn ML

    (2007) Hedonic hunger: a new dimension of appetite?

    Physiology & Behavior 91:432–439.

    https://doi.org/10.1016/j.physbeh.2007.04.006

    • PubMed
    • Google Scholar
41. 1. Mao Z
    2. Davis RL

    (2009) Eight different types of dopaminergic neurons Innervate the Drosophila mushroom body Neuropil: anatomical and physiological heterogeneity

    Frontiers in Neural Circuits 3:5.

    https://doi.org/10.3389/neuro.04.005.2009

    • PubMed
    • Google Scholar
42. 1. May CE
    2. Vaziri A
    3. Lin YQ
    4. Grushko O
    5. Khabiri M
    6. Wang QP
    7. Holme KJ
    8. Pletcher SD
    9. Freddolino PL
    10. Neely GG
    11. Dus M

    (2019) High dietary sugar reshapes sweet taste to promote feeding behavior in Drosophila melanogaster

    Cell Reports 27:1675–1685.

    https://doi.org/10.1016/j.celrep.2019.04.027

    • PubMed
    • Google Scholar
43. 1. May CE
    2. Rosander J
    3. Gottfried J
    4. Dennis E
    5. Dus M

    (2020) Dietary sugar inhibits Satiation by decreasing the central processing of sweet taste

    eLife 9:e54530.

    https://doi.org/10.7554/eLife.54530

    • PubMed
    • Google Scholar
44. 1. Mohammad F
    2. Stewart JC
    3. Ott S
    4. Chlebikova K
    5. Chua JY
    6. Koh TW
    7. Ho J
    8. Claridge-Chang A

    (2017) Optogenetic inhibition of behavior with anion Channelrhodopsins

    Nature Methods 14:271–274.

    https://doi.org/10.1038/nmeth.4148

    • PubMed
    • Google Scholar
45. 1. Ning J
    2. Li Z
    3. Zhang X
    4. Wang J
    5. Chen D
    6. Liu Q
    7. Sun Y

    (2022) Behavioral signatures of structured feature detection during courtship in Drosophila

    Current Biology 32:1211–1231.

    https://doi.org/10.1016/j.cub.2022.01.024

    • PubMed
    • Google Scholar
46. 1. Peciña S
    2. Smith KS
    3. Berridge KC

    (2006) Hedonic hot spots in the brain

    The Neuroscientist 12:500–511.

    https://doi.org/10.1177/1073858406293154

    • PubMed
    • Google Scholar
47. 1. Ro J
    2. Harvanek ZM
    3. Pletcher SD
    4. Roman G

    (2014) FLIC: high-throughput, continuous analysis of feeding behaviors in Drosophila

    PLOS ONE 9:e101107.

    https://doi.org/10.1371/journal.pone.0101107

    • PubMed
    • Google Scholar
48. 1. Ro J
    2. Pak G
    3. Malec PA
    4. Lyu Y
    5. Allison DB
    6. Kennedy RT
    7. Pletcher SD

    (2016) Serotonin signaling mediates protein valuation and aging

    eLife 5:e16843.

    https://doi.org/10.7554/eLife.16843

    • PubMed
    • Google Scholar
49. 1. Rossi MA
    2. Stuber GD

    (2018) Overlapping brain circuits for Homeostatic and Hedonic feeding

    Cell Metabolism 27:42–56.

    https://doi.org/10.1016/j.cmet.2017.09.021

    • PubMed
    • Google Scholar
50. 1. Saper CB
    2. Chou TC
    3. Elmquist JK

    (2002) The need to feed: Homeostatic and Hedonic control of eating

    Neuron 36:199–211.

    https://doi.org/10.1016/s0896-6273(02)00969-8

    • PubMed
    • Google Scholar
51. 1. Shell BC
    2. Schmitt RE
    3. Lee KM
    4. Johnson JC
    5. Chung BY
    6. Pletcher SD
    7. Grotewiel M

    (2018) Measurement of solid food intake in Drosophila via consumption-excretion of a Dye Tracer

    Scientific Reports 8:11536.

    https://doi.org/10.1038/s41598-018-29813-9

    • PubMed
    • Google Scholar
52. 1. Shell BC
    2. Luo Y
    3. Pletcher S
    4. Grotewiel M

    (2021) Expansion and application of Dye tracers for measuring solid food intake and food preference in Drosophila

    Scientific Reports 11:20044.

    https://doi.org/10.1038/s41598-021-99483-7

    • PubMed
    • Google Scholar
53. 1. Siemian JN
    2. Arenivar MA
    3. Sarsfield S
    4. Aponte Y

    (2021) Hypothalamic control of Interoceptive hunger

    Current Biology 31:3797–3809.

    https://doi.org/10.1016/j.cub.2021.06.048

    • PubMed
    • Google Scholar
54. 1. Simpson SJ
    2. Batley R
    3. Raubenheimer D

    (2003) Geometric analysis of Macronutrient intake in humans: the power of protein?

    Appetite 41:123–140.

    https://doi.org/10.1016/s0195-6663(03)00049-7

    • PubMed
    • Google Scholar
55. 1. Slooten E

    (1994) Behavior of Hector’s dolphin: classifying behavior by sequence analysis

    Journal of Mammalogy 75:956–964.

    https://doi.org/10.2307/1382477

    • Google Scholar
56. 1. Sternson SM
    2. Eiselt AK

    (2017) Three pillars for the neural control of appetite

    Annual Review of Physiology 79:401–423.

    https://doi.org/10.1146/annurev-physiol-021115-104948

    • PubMed
    • Google Scholar
57. 1. Tanaka NK
    2. Tanimoto H
    3. Ito K

    (2008) Neuronal Assemblies of the Drosophila mushroom body

    The Journal of Comparative Neurology 508:711–755.

    https://doi.org/10.1002/cne.21692

    • PubMed
    • Google Scholar
58. 1. Tsao CH
    2. Chen CC
    3. Lin CH
    4. Yang HY
    5. Lin S

    (2018) Drosophila mushroom bodies integrate hunger and satiety signals to control innate food-seeking behavior

    eLife 7:e35264.

    https://doi.org/10.7554/eLife.35264

    • PubMed
    • Google Scholar
59. Software

    1. Weaver KJ

    (2023) R code for simple FLIC analyses, version swh:1:rev:9713cfeb88c26e7aa4c0d87b26fcf3361500c670

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:8049eac1502e50f7d0728d68c69a9a3bc56e6802;origin=https://github.com/PletcherLab/FLIC_R_Code;visit=swh:1:snp:14c5ef7c09555c3df00b6c8b38cda53a00ae3ce3;anchor=swh:1:rev:9713cfeb88c26e7aa4c0d87b26fcf3361500c670
60. 1. Weaver KJ
    2. Holt RA
    3. Henry E
    4. Lyu Y
    5. Pletcher SD

    (2023) Effects of hunger on neuronal histone modifications slow aging in Drosophila

    Science 380:625–632.

    https://doi.org/10.1126/science.ade1662

    • PubMed
    • Google Scholar
61. 1. Willesen MG
    2. Kristensen P
    3. Rømer J

    (1999) Co-localization of growth hormone Secretagogue receptor and NPY mRNA in the Arcuate nucleus of the rat

    Neuroendocrinology 70:306–316.

    https://doi.org/10.1159/000054491

    • PubMed
    • Google Scholar
62. 1. Wu JS
    2. Luo L

    (2006) A protocol for Dissecting Drosophila melanogaster brains for live imaging or immunostaining

    Nature Protocols 1:2110–2115.

    https://doi.org/10.1038/nprot.2006.336

    • PubMed
    • Google Scholar
63. 1. Xu Y
    2. Lu Y
    3. Cassidy RM
    4. Mangieri LR
    5. Zhu C
    6. Huang X
    7. Jiang Z
    8. Justice NJ
    9. Xu Y
    10. Arenkiel BR
    11. Tong Q

    (2019) Identification of a Neurocircuit underlying regulation of feeding by stress-related emotional responses

    Nature Communications 10:3446.

    https://doi.org/10.1038/s41467-019-11399-z

    • PubMed
    • Google Scholar
64. 1. Yang MY
    2. Armstrong JD
    3. Vilinsky I
    4. Strausfeld NJ
    5. Kaiser K

    (1995) Subdivision of the Drosophila mushroom bodies by enhancer-trap expression patterns

    Neuron 15:45–54.

    https://doi.org/10.1016/0896-6273(95)90063-2

    • PubMed
    • Google Scholar
65. 1. Zigman JM
    2. Jones JE
    3. Lee CE
    4. Saper CB
    5. Elmquist JK

    (2006) Expression of Ghrelin receptor mRNA in the rat and the mouse brain

    The Journal of Comparative Neurology 494:528–548.

    https://doi.org/10.1002/cne.20823

    • PubMed
    • Google Scholar

The following is the authors' response to the original reviews.

In brief, we incorporated all wording and clarity suggestions into the manuscript. We also updated figure legends to include additional details, including replicate numbers. New data have been added in response to requests from the reviewers. Volumetric intake data are included as a supplemental figure (Figure 1–Figure Supplement 1A) and we will include movies of the confocal stacks from our CaMPARI imaging. We worked hard to address all the reviewers’ concerns and provide a detailed response below to the reviewers’ public comments as well as their author-specific comments.

Reviewer #1 (Public Review):

1. All feeding data presented in the manuscript are from the interactions of individual flies with a source of liquid food, where interaction is defined as 'physical contact of a specific duration.' It would be helpful to approach the measurement of feeding from multiple angles to form the notion of hedonic feeding since the debate around hedonic feeding in Drosophila has been ongoing for some time and remains controversial. One possibility would be to measure food intake volumetrically in addition to food interaction patterns and durations (e.g. via the modified CAFE assay used by Ja).

We acknowledge that our FLIC assays address only one dimension of feeding behavior, physical interaction with liquid food. However, there is clear evidence that interactions are strongly predictive of consumption, and it would be technically difficult to measure feeding durations at the resolution of milliseconds using a Café assay. Nevertheless, we appreciate the spirit of this comment and agree that expanding our inference to other measures of feeding, as well as feeding environments, is an important next step. To this end, we now include measures of feeding on more traditional solid food, using the ConEx assay, and find that flies in the hedonic environment consume twice as much sucrose volume compared to flies in the control environment. These have been added as supplemental data (Figure 1 – Figure Supplement 1A), and the text has been updated to reflect our findings.

2. Some of the statistical analyses were presented in a way that may make understanding the data unnecessarily difficult for readers. Examples include:

a) In Table I the authors present food interaction classifications based on direct observation. These are helpful. However, the classification system is updated or incompletely used as the manuscript progresses, most importantly changing from four categories with seven total subcategories to three categories and no subcategories. In subsequent data analyses, only one or two of these categories are assessed. It would be helpful, especially when moving from direct observation to automated categorization, to quantify the exact correspondences between all of the prior and new classifications, as well as elaborate on the types of data that are being excluded.

We appreciate the feedback on our usage of the behavioral classification system and have made several adjustments to improve it. We renamed some of the behaviors to make them more intuitive (see Reviewer #2, comment #1), and updated the main text and Table 1 to reflect these changes. We updated the text and figures to be more transparent about when we group subcategories into main categories for quantification and when we quantify all subcategories separately. Because these videos required manual scoring by an experimenter, after our initial characterizations we opted to score only main categories (which contain subcategories). We agree that it would be useful to quantify correspondence between subcategories and the automated FLIC signal. However, we believe this task is better suited for more advanced and automated video tracking software, and, incidentally, more sophisticated analysis of FLIC data, which has a very high-dimensional character that has yet to be properly exploited. At the moment, therefore, we are not confident in the ability to understand the data at the desired resolution.

b) The authors switch between a variety of biological and physiological conditions with varying assays, which makes following the train of reasoning nearly impossible to follow. For example, the authors introduce us to circadian aspects of feeding behavior to introduce the concept of 'meal' and 'non-meal' periods of the day. It is then not clear in which of the subsequent experiments this paradigm is used to measure food interactions. Is it the majority of the subsequent figure panels? However, the authors also use starved flies for some assays, which would be incompatible with circadian-locked meals. The somewhat random and incompletely reported use of males and females, which the authors show behave differently, also makes the results more difficult to parse. Finally, the authors are comparing within-fly for the 'control environment' and between flies for their 'hedonic environment' (Figure 3A and subsequent panels), which I believe is not a good thing to do.

We apologize for our difficulties conveying our inference, which was also noted by Reviewer #2. We have worked hard to improve this component in the revision. With respect to the confusion about circadian feeding, we introduced circadian meal-times to complement starvation as a second (perhaps more natural) way to measure behaviors associated with hunger. Importantly, we do not use circadian meal-times beyond Figure 1; all subsequent FLIC experiments were conducted during non-meal times of day for 6 hours, which avoids confounding our data with circadian-locked meals even when we use starved flies. We have clarified this point in the revision.

The reviewer also points out that we make both within-fly and between-fly comparisons, which is a point that we note. Perhaps some concern arises, again, from the challenges that we faced in properly delineating our inferences about different types of feeding measures (and motivations). Inference about homeostatic feeding was made using within-fly measures, comparing events on sucrose vs. those on yeast. Inference about hedonic feeding was made using between fly measures (average durations of different flies on 2% vs. 20% sucrose). Treatment comparisons to control always used measures of the same type, such that inference was not made using between-fly measures for treatment and within-fly for control (i.e., all of our figure panels were either within-fly or between fly). We have worked to clarify this in the revision.

Importantly, our approach to all experiments avoided confounding by used randomized design at multiple levels (e.g., randomizing control and hedonic environments to FLIC DFMs, alternating food choice sidedness in the DFMs), by ensuring that flies in both environments are sibling flies that came from the same vial environment before being tested, and by performing each experiment multiple times.

c) Statistical analyses are not always used consistently. For example, in Figures 3B and C, post hoc test results are shown for sucrose vs. yeast interactions, but no such statistics are given for 3E and 3F, preventing readers from assessing if the assay design is measuring what the authors tell us it is measuring.

We report p-values for two-way ANOVA interaction terms for all appropriate experiments. If (and only if) the interaction term is significant, we conduct post-hoc tests for more detailed statistical analysis and report the p-values. The reviewer points out that we do not perform post-hoc tests in figures 3E and 3F. These figures had a non-significant interaction term, and thus, we did not feel a post-hoc test was warranted.

Reviewer #2 (Public Review):

1. The dissection of feeding into distinct behavioral elements and its correlation with electrical FLIC signals that allow interpreting feeding types is a fundamental new method to dissect feeding in flies. However, the categories of micro-behaviors in Table 1 are not intuitive.

We agree and have updated the Table, figures, and main text. Please see also our response to Reviewer #1, comment #1.

2. The details for the behavioral data analysis are not clear and should be made more obvious. For example, how many males and females were used in each experiment? Were any of the females mated or were they all virgins? If all virgins, why not use mated females? Mating status may have an effect on the feeding drive. If mated and virgin females were used, are there any differences between them? Similarly, for diurnal feeding experiments, it is not immediately clear from the graphs how many animals were used and how the frequencies were obtained (Fig. 1F, presumably averages for each category per fly but that is inconsistent with the legend in the supplement for this figure). Why does the transition heat map not include all micro-behaviors (Fig. 1E, no LQ data which are significant in diurnal feeding)?

We have clarified the number of flies and events for each behavioral experiment in Figure 1, and we updated the figure legend appropriately. We note that these behavioral datasets are non-overlapping, and each time we mention the number of events scored in the text, that number includes only “new” videos. Female and male flies for all experiments were mated, and we have clarified this in the main text and methods.

For the diurnal experiment in Figure 1F, we scored over 700 events from new (non-overlapping) video compilations and updated the number of flies and event number in the figure legend. The diurnal data we present in the supplement for this figure is a separate experiment conducted on 38 flies, intended only to demonstrate the circadian nature of fly feeding.

For the transition heat map, analysis of this sort seems to require a large amount of data to have sufficient power to return a transition matrix. LQ events are relatively low in frequency, so we opted to combine them with L events for this analysis. We have updated the figure and figure legend to reflect this.

3. The CaMPARI images do not look great, particularly in the pan-neuronal condition (Fig. 5A). It would be useful to include the movie of the stack. Did any other brain regions show activity differences, such as SEZ or PI? These regions are known to be involved in feeding so it seems surprising they show no effect.

We find that CaMPARI imaging is subject to high levels of noise and background, especially when using a broad driver as the reviewer has pointed out. This is why we opted to follow-up our pan-neuronal CaMPARI experiment using a more specific mushroom body driver and to test our correlational findings of increased MB activity in hedonic environments with genetic approaches in the remainder of Figure 5. We have included movies of the confocal stacks for both CaMPARI experiments, as requested.

Reviewer #1 (Recommendations For The Authors):

Main concern:

No measurements of intake, either in volume or in caloric value. Hence, 'hedonic' feeding is only indirectly supported.

I would like to suggest to the authors that they measure intake volumetrically in addition to food interaction patterns and durations. For example, William Ja developed a modified CAFE assay that measures consumption volume in real-time in freely behaving flies (http://dx.doi.org/10.1038/nprot.2017.096). Liming Wang has another capable assay. Additional values of expanding measurement methods for feeding are that it helps tie the research more directly to that of others, and it helps remove the concern that any one assay may introduce unknown biases.

For the CaMPARI, it would be helpful to provide a demonstration of its effectiveness by recapitulating a deep brain neural pathway known to be engaged by a stimulus by GCaMP or electrophysiology.

Additional concern:

The authors assume satiety states during different circadian periods (line 253, for example). It seems critical to directly measure the satiety state.

Technical concerns:

Figure 5 A, B: there is reported near zero UV transmission through the head: https://doi.org/10.1364%2FBOE.6.000514, hence the CaMPARI measurements are suspect. It appears that there may be an effect in the optic lobes that may receive greater UV illumination by being more peripheral. A positive control to demonstrate deep brain access by UV is needed.

Y-axes vary for the same measurement types within figures, for example, Figure 5 C-G. Also Figures 3F, G, I, K, M and Figures 3D, E, H, J, L. This hinders direct comparisons.

Figure 2: why are there no statistics to distinguish interaction (I) events from F and L? Why are the example graphs presented using different scale x-axes? For A-C, why no averaged response graphs for the classifications? Were there other events that did not fit these classifications?

In lines 224-226, the claim of statistical significance at p=0.061 makes the reader suspicious of the statistical interpretations throughout the manuscript.

Figure 3B starved looks the same as Figure 3C sated for females, using the same assay and conditions. This implies a huge amount of variance in behavior between experiments.

We appreciate the recommendations from Reviewer #1 and have done our best to address many of their concerns. Regarding their main concerns, we have added volumetric feeding data to the manuscript, included movies of the confocal stacks for the CaMPARI experiments, and clarified the circadian timing of our behavioral experiments. These details are outlined in our public response to both reviewers. The reviewer also expressed a few technical concerns, mostly regarding statistical analyses. We agree that there seems to be a large amount of biological variability between experiments, which we do indeed find to be the case with behavioral experiments of this sort. For this reason, we avoid making direct comparisons on absolute values between experiments, as the reviewer suggests, and thus allow our Y-axes to vary for each figure to better facilitate within-experiment comparisons. The reviewer also points out that, in one instance, we refer to a p-value of 0.061 as statistically significant in the text. While we have changed our language to reflect the perceived convention, we note that there is little inferential difference between these values, and we report exact p-values to allow the reader to make an informed decision.

Reviewer #2 (Recommendations For The Authors): The writing and data presentation in this paper is somewhat dense and confusing at times. Comments and questions below are intended to help improve data presentation and resolve questions that will help the reader navigate and understand the data to better appreciate the significance of the findings.

Comments and questions:

Line 160 cites Chen et al, 2002 as an example of behavioral characterization that is useful for read-outs of neural states, but no neural states were defined in that work. A better example where a circuit was linked to a specific behavioral category is PMID30415997 (Duistermars et al., 2018).

Line 171: were the females mated or virgin or was it variable?

The classification system in Table 1 is a bit confusing. For example, the distinction is made between Fast and Long feeding events as well as interactions with food and other events. FH meet the requirements of F and H, presumably meaning that flies are fast feeding and touching the food with their front legs. Why are front legs and hind legs touching food abbreviated H and FF respectively instead of something more obvious like IF and IH (referring to Interaction with Front legs or Interaction with Hind legs)?

Also was there never any tasting with the middle legs? In Fig1B, all the I events are grouped. Are most of these H or FF events? The frequency in Fig. 1B is shown as normalized as a frequency of all events. The statistical analyses are all parametric. Are these data normally distributed?

Lines 224-229: the relative frequency of L-type feeding is increased in starved flies and the relative frequency of F feeding is decreased. Is the relative L- or F-type feeding frequency considered on total behavior or just the sum of long and fast feeding or the sum of all types of feeding?

The events that are analyzed vary throughout the paper. Line 173 mentions 300 events, line 222, 500 events, and line 257, 700 feeding events. Are these all independent experiments, or are these overlapping data sets analyzed for different parameters?

For diurnal feeding behavior, the authors analyzed 700 events and found significantly more LQ events during meal time (i.e. at the beginning and end of the day). Based on the figure legend in the supplement to Figure 1, it appears that these data were collected on 38 female flies. But in Fig 1F, there are ~8 points per feeding type (F, L, and LQ) during meal and non-meal conditions. Shouldn't all 38 flies have an average frequency for each type of feeding during meal and non-meal times? Were these females mated or not? Is this effect also true for males? To help the reader understand the data better, it would be helpful to note the number of flies used in each experiment or in each analysis in the different figures and wherever the data are mentioned in the manuscript. It also seems likely that the mating state may have an effect on feeding so knowing the result in mated versus unmated would be a useful analysis.

It is interesting that there is a difference in feeding in starved flies versus diurnal feeding in the presumably hungry versus sated phase (meal versus non-meal phases). As mentioned by the authors earlier in the manuscript, starved flies have a relative increase in L-type feeding. However, they perform less LQ feeding than sated flies, and yet LQ feeding is the only significantly different type of feeding in the hungry state of diurnal feeding. In the morning, the transition to feeding is very abrupt compared to the gradual increase in the evening. Is there any difference between the type of feeding or the transition matrix in the evening versus morning meal times? Also, why is LQ feeding not included as a category in the transition matrix in Fig 1E?

In Fig 2, the authors examine FLIC signals with video data to identify feeding types from FLIC signals. Why are there signal durations for F-type feeding that are longer than 3 seconds when it is defined as 1-3 sec of the proboscis contact with food and conversely signals of L-type feeding shorter than 4 seconds when it is defined as >4 seconds of continuous proboscis contact? Does this mean that signal can be longer or shorter than the actual time the proboscis is in the food?

With these parameters, the authors develop an assay to identify homeostatic and hedonic feeding by applying the signal analysis to food choices representing homeostatic (2% sucrose versus yeast) and hedonic (2% sucrose versus 20%) conditions. In Fig 3C, they show that fully-fed females show a stronger preference for yeast food than sugar food compared to males (line 335). Is this in fully fed animals? The yeast preference in females looks almost the same as in the starved females in Fig 3B.

The CaMPARI images shown in Fig 5A (and to a lesser extent Fig 5B) are not particularly convincing although the quantification looks clear. Providing the movies of the stacks may help the reader better appreciate the difference in MB red signal in the hedonic state. It would also help to show the number of flies that were tested in these experiments as well as the sex and mating status. Provide the n in the figure legend and in the relevant sections in the text.

Were the mushroom bodies the only brain region with significant, measurable activity changes? One might expect changes in other feeding areas, such as the subesophageal zone (SEZ) and the peptidergic regions of the brain (PI), which are both known to affect feeding in flies. This may also be a useful method to examine differences in mated versus unmated flies.

In Fig 5C the caption reads MB lambda lobe inhibition. Shouldn't this be gamma lobe inhibition as suggested in the figure legend?

The paper largely distinguishes homeostatic from hedonic feeding only. It may be useful to discuss other non-homeostatic mechanisms as well or at least make the distinction in the introduction and or discussion.

We thank reviewer #2 for their thoughtful suggestions to improve the clarity of the manuscript. They suggest several improvements, which we implemented, including that we improve the classification system in Table 1 to make it more intuitive, state how we normalized observed behavioral frequencies, clarify that the number of events we cite for each experiment are non-overlapping, and explain the use of circadian meal vs. non-meal times. We also noticed, as did this reviewer, that the usage of L vs. LQ events differs between starved flies and flies observed during meal-time. We agree that it may be interesting to sort out the nuances of why and how these differences occur, as it suggests that starvation may in some ways be different from physiological hunger. However, our method of manually observing flies would make this difficult at present. We hope to utilize more advanced video tracking software in the future to investigate this question. The reviewer also posed several questions about the hunger/satiety state of flies that we used for each experiment, which we clarified throughout the main text, figure legends, and methods.

This reviewer points out two technical concerns, which we have addressed. The concerns about our CaMPARI imaging are noted, and we have discussed them in response to reviewer #1 and in our public response. We now include movies of the confocal stacks, as requested. There was also a question about FLIC durations of F and L events in Figure 2, with some visually identified F events producing FLIC signals longer than 4 seconds and some L events producing FLIC signals shorter than 4 seconds. Although we show that population averages from the FLIC can reliably recapitulate our visual metrics, there is occasional noise at the individual level. For example, although a fly may have contact of its proboscis with the food for less than 4 seconds, the FLIC signal may persist slightly beyond that interaction due to sustained contact with a non-proboscis body part or due to liquid food contacting the signal pad. We also occasionally observed L events that we visually identified to last longer than 4 seconds, but nevertheless did not produce a FLIC signal of equal length. This can occur when a fly feeds on the liquid food but transiently loses contact with the signal pad. Although there is some noted technical noise, we show that population-level data is sufficient to reflect our visual observations.

Author details

1. Kristina J Weaver

   Department of Molecular and Integrative Physiology and Geriatrics Center, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   kjweaver@umich.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2496-3568

2. Sonakshi Raju

   College of Literature, Science, and the Arts, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States

   Contribution

   Formal analysis, Investigation, Writing – original draft

   Competing interests

   No competing interests declared

3. Rachel A Rucker

   Neuroscience Graduate Program, University of Michigan, University of Michigan, Ann Arbor, United States

   Contribution

   Formal analysis, Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1434-9401

4. Tuhin Chakraborty

   Department of Molecular and Integrative Physiology and Geriatrics Center, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

5. Robert A Holt

   College of Literature, Science, and the Arts, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

6. Scott D Pletcher

   Department of Molecular and Integrative Physiology and Geriatrics Center, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States

   Contribution

   Resources, Software, Supervision, Funding acquisition, Methodology, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   spletch@umich.edu

   Competing interests

   is a share holder in the company, Flidea, which has developed technology related to the FLIC feeding system

   "This ORCID iD identifies the author of this article:" 0000-0002-4812-3785

• 1,044

  Page views

• 66

  Downloads

• 0

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.