Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Neuroscience. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 01 March 2024

Drosophila Reward Circuitsfree

Drosophila Reward Circuitsfree

  • John S. Hernandez, John S. HernandezBrown University
  • Tariq M. BrownTariq M. BrownBrown University
  •  and Karla R. KaunKarla R. KaunBrown University

Summary

The ability to sense and respond to a rewarding stimulus is a key advantage for animals in their natural environment. The circuits that mediate these responses are complex, and it has been difficult to identify the fundamental principles of reward structure and function. However, the well-characterized brain anatomy and sophisticated neurogenetic tools in Drosophila melanogaster make the fly an ideal model to understand the mechanisms through which reward is encoded. Drosophila find food, water, intoxicating substances, and social acts rewarding. Basic monoaminergic neurotransmitters, including dopamine (DA), serotonin (5-HT), and octopamine (OA), play a central role in encoding these rewards. DA is central to sensing, encoding, responding, and predicting reward, whereas 5-HT and OA carry information about the environment that influences DA circuit activity. In contrast, slower-acting neuromodulators such as hormones and neuropeptides play a key role in both encoding the pleasurable stimulus and modulating how the internal environment of the fly influences reward sensation and seeking. Recurring circuit motifs for reward signaling identified in Drosophila suggest that these key principles will help elucidate understanding of how reward circuits function in all animals.

Subjects

  • Molecular and Cellular Systems
  • Sensory Systems
  • Invertebrate Neuroscience

Introduction

Appetitive stimuli induce positive sensations that motivate animals to perform tasks, and these sensations are termed as “rewarding.” These experiences are fundamental to an animal’s ability to obtain and assess food and water, reproduce, and ensure fitness of its progeny. Perhaps unsurprisingly, reward circuits are strikingly similar in anatomy and function across species from flies to primates (Dvořáček & Kodrík, 2021; Landayan & Wolf, 2015; Scaplen & Kaun, 2016). It has traditionally been difficult to understand the structure and function of reward circuits at the level of individual cells. However, the relative simplicity, detailed anatomy, and genetic accessibility of the Drosophila melanogaster brain make the fly a powerful model to understand how reward circuits function (Honda, 2022; Hulse et al., 2021; Li et al., 2020; Scheffer et al., 2020; Takemura et al., 2017; Venken et al., 2011). The ability to investigate the structure and function of circuits in both larval and adult stages also provides a cross-developmental opportunity to extract general principles underlying the function of reward circuits.

In Drosophila, rewards include macronutrients like sugar, fat, protein, and water; social rewards like the act of copulation; and intoxicating substances such as alcohol (Dus et al., 2011; Fougeron et al., 2011; Kaun et al., 2011; Lin et al., 2014; Shohat-Ophir et al., 2012; Tempel et al., 1983; Toshima & Tanimura, 2012). These rewards and their cues are perceived through sensory structures in the Drosophila brain such as the optic lobe (vision), subesophageal zone (taste), antennal lobe (smell), and the antennal mechanosensory motor center (audition). Sensory information is then sent through divergent and convergent connections to central brain structures to dictate behavior. Artificial activation of reward-associated regions, including primary sensory brain regions and their direct outputs, can be inherently rewarding to Drosophila (Lyutova et al., 2019; Nuwal et al., 2012; Schneider et al., 2012; Seidenbecher et al., 2020; Thoener et al., 2022; Wen et al., 2005). For example, Drosophila will optogenetically self-stimulate sugar-sensing neurons (Deere & Devineni, 2022). Because reward behavior manifests in a variety of ways, a wide variety of reward contexts are discussed.

Internal state is intrinsically linked to reward response, as it changes the motivational response for reward. For example, thirsty flies find water rewarding, and hungry flies find food rewarding. This makes it difficult to parse out circuits that are responsible for inducing a particular internal state and encoding reward. Here, Drosophila circuits that directly encode reward behavior are reviewed within the appropriate context in which reward is perceived. Neural circuits and neurotransmitters underlying various rewards often overlap and share similar structural circuit motifs. The circuit function, however, may differ according to the internal state, reward stimulus, environmental context, and the appropriate output response. This review focuses on circuits that reside downstream of the primary sensory pathways and elicit a behavioral response in which a fly purposefully and/or repeatedly engages with a stimulus response.

Rewarding Stimuli

Most reward circuits in Drosophila have been characterized in the context of food reward. Food consumption behavior is broken down into three components: initiation, ingestion, and cessation. Food consumption initiation involves sensing food odors and nutrients within food (Aryal et al., 2022; Liman et al., 2014), whereas metabolic nutrient sensing occurs during the ingestion process (Yapici et al., 2016). Food consumption cessation occurs once an animal experiences satiety (Huetteroth et al., 2015). Experiencing both the taste and metabolic properties of food is perceived as rewarding to flies (Burke & Waddell, 2011; Huetteroth et al., 2015; May et al., 2020; Yamagata et al., 2015).

Drosophila food typically contains a mixture of macronutrients, including carbohydrates, protein, and fats, and micronutrients, including vitamins and minerals. Drosophila consume mixtures of carbohydrates like sucrose, fructose, and glucose in rotting fruits (Burke & Waddell, 2011; Lebreton et al., 2014; Lüersen et al., 2019), as well as proteins in the yeasts that grow on fermenting fruit (Becher et al., 2012). Nutritive sugars are sensed through two distinct mechanisms: (a) taste sensing via peripheral taste receptor cells that detect the palatability of sugar and (b) and taste-independent behavioral responses to the nutritional value of sugar (Clyne et al., 2000; Dus et al., 2011; Thorne et al., 2004; Wang et al., 2013). Sucrose is the most extensively studied reward in Drosophila, resulting in mechanistic knowledge of how circuits function during sugar reward encoding, how this alters sugar ingestion, and how olfactory cues associated with sugar help the fly navigate to a source of nutritious sugar. Less is understood about the rewarding properties of proteins, although a fly’s ability to sense and consume appropriate proteins directly contributes to their survival and fitness (Liman et al., 2014), and glucose consumption alters protein consumption (Lebreton et al., 2014). For this review, circuits for mixed media and individual macronutrients overlap in the context of food reward.

Water, essential for maintaining proper osmolarity of cells within living organisms, is rewarding to thirsty flies. In order to seek water, flies detect humidity gradients using humidity-sensing and dry-sensing receptors (Enjin et al., 2016; Ji & Zhu, 2015). Thirsty Drosophila are able to seek out sources of humidity through hygrosensation (Sayeed & Benzer, 1996), and upon encountering water, thirsty flies will extend their proboscis and ingest water rapidly using their pharynx (Enjin et al., 2016). Because thirsty flies find water rewarding, they actively seek out cues previously associated with water reward (Lin et al., 2014).

Intoxicating rewards such as alcohol induce an alternative internal state that is perceived as rewarding to Drosophila. Moderate doses of ethanol increase motor activity, whereas high concentrations of ethanol reduce postural control and induce sedation (Corl et al., 2005; Dzitoyeva et al., 2003; Moore et al., 1998; Rothenfluh et al., 2006; Wolf et al., 2002). Ethanol consumption evaluated by measuring proboscis extension (Scheiner et al., 2014), fluorometry (Peru Y Colón de Portugal et al., 2014), oligonucleotide tagging (A. Park et al., 2018), or capillary feeding demonstrates Drosophila preferentially consume alcohol (Adel & Griffith, 2021; Devineni & Heberlein, 2009; Ja et al., 2007). Flies remember pharmacologically stimulating doses of ethanol as rewarding and will walk over a 120-V electric shock to attain a cue previously associated with ethanol (Kaun et al., 2011).

Social behavior is also rewarding to flies in the context of courtship and copulation. In order for sexual engagement to occur in Drosophila, adult males court adult females in an intricate series of behaviors that involve auditory and visual sensory cues demonstrated to either a receptive or a dismissive female (Pavlou & Goodwin, 2013). If courtship is successful, the pair will copulate, which is the rewarding outcome of courtship behavior (Shohat-Ophir et al., 2012; Zer-Krispil et al., 2018). In contrast, repetitively suppressing copulation in male flies results in alternative reward seeking (Shohat-Ophir et al., 2012). These rewards are all encoded within layered neuropeptidergic and monoaminergic signaling within central brain structures, and these circuits function distinctly and cohesively to promote reward behavior, depending on context.

Figure 1. Reward circuits in adult Drosophila. Circuits identified in behavioral responses to a variety of rewards in Drosophila include (A) dopaminergic circuits; (B) dopamine mushroom body (MB) neurons implicated in reward synapse onto distinct MB compartments as follows: α‎1–α‎3, α‎′1–α‎′3, β‎1 and β‎2, β‎′1 and β‎′2, and γ‎1–γ‎5; (C) octopaminergic circuits, (D) serotonergic circuits, and (E) neuropeptidergic circuits.

Dopamine

Dopamine (DA) signaling is central to reward processing in Drosophila as DA neurons are activated by hunger, encode valence, and are required for cue-induced appetitive memories (Y.-C. Kim et al., 2007; W.-P. Lee et al., 2021; Pitman et al., 2011; Figure 1A). There are approximately 280 DA neurons in the Drosophila central brain (Budnik & White, 1987; Mao & Davis, 2009; Nässel & Elekes, 1992), half of which innervate the mushroom body (MB), a well-studied structure important for learning and memory (Mao & Davis, 2009; Nässel & Elekes, 1992). Two main types of DA neurons innervate the MB: 12 protocerebral posterior medial lateral (PPL1) neurons innervate the vertical (α‎, α‎′, γ‎1 and γ‎2) lobes, and ~100 protocerebral anterior medial (PAM) neurons innervate the horizontal (β‎, β‎′, γ‎ and α‎1) lobes. Together, these DA neurons innervate the parallel axons of intrinsic MB neurons, called Kenyon cells (KCs), and the dendrites of cholinergic, glutamatergic, or γ‎-aminobutyric acid (GABAergic) mushroom body output neurons (MBONs), which divide the MB axons into distinct compartments (Figure 1B; Aso, Hattori, et al., 2014; Stahl et al., 2022). DA neurons provide information about reward to both the KCs and MBONs to bias the fly toward engaging in behavior that leads to a rewarding outcome (Aso, Hattori, et al., 2014; Aso & Rubin, 2016; Aso, Sitaraman, et al., 2014; Huetteroth et al., 2015; Ichinose et al., 2021; Yamagata et al., 2015). The mechanisms by which DA neurons encode reward within the MB depends on the type and timing of reward, the internal state of the fly, and which behavior is being executed to attain the reward.

Accumulating data suggest that the two main DA neuron types that innervate the MB provide information about opposing valences, where PAMs largely convey information about reward. PAM neurons that innervate the α‎1, γ‎4, γ‎5, and β‎′2 compartments of the MB are activated by water, sucrose, and alcohol (Cohn et al., 2015; Li et al., 2020; C. Liu et al., 2012; Otto et al., 2020; Scaplen et al., 2020; Yamagata et al., 2015; Figure 1B). This compartmentalization makes it possible for the flies to encode multiple forms of reward memories such as the short-term memory for sugar taste simultaneously with the long-term memory for sugar calories (Burke & Waddell, 2011; Chouhan & Sehgal, 2022; Huetteroth et al., 2015; Pardo-Garcia et al., 2023; Yamagata et al., 2015; Y. Zhang et al., 2015). The number of compartments involved appears to scale with the intensity of the reward. For example, alcohol indiscriminately activates all PAMs (Scaplen et al., 2020), whereas sucrose activates select compartments (Harris et al., 2015; C. Liu et al., 2012).

Some MB compartments encode multiple reward types, whereas others assess value or are specific to long-lasting cue-induced memories of reward (Huetteroth et al., 2015; Lin et al., 2014; Scaplen et al., 2020; Yamagata et al., 2015). An example of this is nicely highlighted with the β‎′2 compartment, which plays a key role in food reward but is also implicated in alcohol and water reward. β‎′2 DA neuron activity controls feeding rate for sucrose and subsequent satiation (May et al., 2020), whereas signaling from the neighboring γ‎5 PAM neurons is required for positive reinforcement of the nutritious quality of food/sucrose (Huetteroth et al., 2015; Yamagata et al., 2015). These neurons both innervate MBON-γ‎5β‎′2a, and optogenetic activation of this MBON is sufficient to induce sucrose memory formation (Otto et al., 2020). This process occurs through the interplay of neurons that encode both aversive and appetitive responses since the aversive GABAergic γ‎1pedc-MBON induces an appetitive response through feedforward inhibition to the γ‎5β‎′2a and β‎′2mp MBONs (Owald et al., 2015; Perisse et al., 2016).

Plasticity within this compartment is dependent on the previous food experience of the fly. Flies maintained on a high-sucrose diet show reduced activity of PAM-β‎′2 DA neurons in response to sugar (May et al., 2020). This increases activity of γ‎5β‎′2a and β‎′2mp MBONs in response to the odor paired with sucrose, causing an inability to form an associative sucrose memory (Pardo-Garcia et al., 2023). Subsequent activation of these PAM-β‎′2 DA neurons reduces overeating in flies on high-sucrose diets (May et al., 2020). This suggests that β‎′2 DA neurons encode the appetitive properties of sugar and receive information about satiety to control sugar seeking and feeding rate.

The β‎′2 MB compartment also plays a key role in alcohol response, which elicits a unique internal state. Activity in the β‎′2a PAM DA neurons and the corresponding compartmental MBON-β‎2β‎′2a neurons is required for cue-induced alcohol seeking (Scaplen et al., 2020). In contrast, inhibiting in the neighboring MBON-β‎′2mp overnight during memory consolidation enhances alcohol memory, a response dependent on the inhibitory dopamine 2-like receptor (Dop2R; Scaplen et al., 2020). β‎′2a PAM DA neurons likely inhibit MBON-β‎′2mp during sleep to promote memory consolidation, and that activity of the β‎′2a compartment during cue presentation is required to express reward memory. This suggests that alcohol acts on a natural reward memory circuit but alters dynamics within that circuit to induce long-lasting memories for alcohol reward.

Internal state significantly influences DA-dependent reward encoding in the MB. Hunger enhances activity of PAM DA neurons projecting to the α‎1, β‎1, β‎′2, and γ‎4 MB compartments in response to an odor (Siju et al., 2020) and recruits involvement of the PAM DA neurons in food occupancy (Landayan et al., 2018). Hunger also promotes yeast food seeking by recruiting the γ‎1pedc>αβ‎ MBON, which inhibits the β‎′2 MBON and α‎′1 MBON (Tsao et al., 2018), and the protocerebral posterior medial (PPM) DA neurons in protein seeking in adult females (Q. Liu et al., 2017). Likewise, DA input into the α‎1, β‎′1, β‎′2, and γ‎5 MB compartments inhibits the proboscis extension reflex to sucrose reward in hungry flies (Chia & Scott, 2020). Together, this suggests that hunger may act through DA MB circuits to both enhance reward perception and amplify behavioral responses to reward.

PAM DA neurons also play a key role in water reward response. The β‎′1, β‎′2, and γ‎4 DA PAMs are all active in response to water consumption in both water-sated and water-deprived files, with β‎′1 having the most robust response in thirsty flies and β‎′2 having a consistently high response regardless of state (Lin et al., 2014; Shyu et al., 2017). γ‎4 DA neurons are necessary for water memory following associations between an olfactory cue and a water reward (Lin et al., 2014; Shyu et al., 2017). Inhibition of signaling from the MBON-γ‎3β‎′1 GABAergic neuron does not directly affect water consumption but disrupts water long-term memory (W.-P. Lee et al., 2021). Flies lacking dopamine 1-like receptors (Dop1R1) also have decreased water consumption and decreased proboscis extension reflex to water reward (Lau et al., 2017). This demonstrates that water reward, like food or alcohol reward, is largely encoded through DA neurons that innervate several adjacent horizontal MB compartments that act in concert to elicit an appropriate behavioral response.

The vertical MB compartments also play a key role in memory for sugar, water, and alcohol reward. Interestingly, these compartments are more associated with 24-hour memory as opposed to immediate or 3-hour memory. The α‎1 PAM DA neuron is activated upon sucrose exposure and provides stimulatory input via the α‎1 KCs and the α‎1 MBON (Ichinose et al., 2015; Yamagata et al., 2015). During appetitive memory expression, the α‎1 MBON forms a feed-forward loop with α‎1 PAM DA neurons (Ichinose et al., 2015). Silencing the γ‎1pedc MBON and PPL1 γ‎1pedc DA neurons in hungry flies suppresses expression of appetitive memory formation. In contrast, satiated flies require activity of γ‎2a′1 MBON and the PPL1-γ‎2a′1 DA neurons for appetitive memory expression (Chouhan et al., 2021). Activity of the α‎2α‎′2 PPL1 DA neurons and α‎′2 MBONs is required for odor cue-induced alcohol seeking 24 hours after formation of alcohol–odor memories (Scaplen et al., 2020). Finally, the cholinergic α‎3 MBONs are required for odor-induced reward seeking 24 hours after an odor–sugar association (Ichinose et al., 2015; Plaçais & Preat, 2013). Similarly, inhibition of signaling from cholinergic MBON-α‎′2, or MBON-α‎′1, MBON-α‎′3m, and MBON-α‎3ap, does not alter water consumption directly but does disrupt preference for a water-paired odor during the first 4-hour period after training (W.-P. Lee et al., 2021). Although it is clear there are many connections between the horizontal and vertical MB compartments (Li et al., 2020; Scheffer et al., 2020), how these compartments interact in the context of reward is largely unknown.

Thus, it appears that a complex interplay of DA-MBON compartments along KC axons within the MB encodes information about reward, memory, and internal state and is required for appropriate behavioral responses. Although the mechanisms through which multiple rewards are encoded within PAM MB compartments seem complex, the relative simplicity of its anatomy allows for extraction of key principles of reward encoding. Study of the β‎′2 MB compartment in particular demonstrates that (a) multiple rewards activate this compartment through DA neurons, (b) odor cues encoded within KCs alter activity of the glutamatergic MBON as a consequence of concurrent DA activation, (c) activation of the glutamatergic MBON elicits learned reward seeking, (d) reward memory encoding within this compartment is dependent on activity of interconnected MB compartments, and (e) both internal state and previous experience alter dynamics within this compartment. Other compartments appear to be dependent on the type or timing of reward, or they are recruited as a response to a shift in internal state.

DA neurons outside the MB also contribute to reward response. Activity of DA neurons projecting to the fan-shaped body is required for ethanol consumption preference (Shamsideen et al., 2019). DA neurons that project to the ellipsoid body, another central complex structure, and their postsynaptic GABAergic R2/R4 ring ellipsoid body neurons contribute to both ethanol-induced hyperactivity and tolerance (Kong et al., 2010; Scaplen et al., 2019; Urizar et al., 2007), which are behavioral components of alcohol reward memory formation (Berger et al., 2008; Nunez et al., 2018). Interestingly, these circuits overlap with those required for the ethologically advantageous ethanol oviposition preference (Azanchi et al., 2013).

Mating is also rewarding to male flies (Shohat-Ophir et al., 2012; Zer-Krispil et al., 2018), and DA signaling is required for sex drive in both male and female Drosophila. Depleting DA via tyrosine hydroxylase inhibitor 3-iodo-tyrosine reduces the likelihood to copulate and increases the time to copulation in females (Neckameyer, 1998). Activation of protocerebral anterior lateral, protocerebral posterior lateral (PPL) 2ab and 2c, PPM 2, and 3 DA neurons drives courtship behavior and increases duration of copulation in male flies (Zhang et al., 2016). These DA neurons synapse on downstream circuitry acting through dopamine 1-like receptor 2 (Dop1R2) expressing P1 interneurons that directly contribute to male sex drive (Kimura et al., 2008; Zhang et al., 2016). Activated P1 neurons will generate courtship behavior in males and, upon successful courting, result in copulation (Kimura et al., 2008).

The role of DA receptors in appetitive response is less well understood, in part because cell type–specific manipulation of these receptors in the context of reward behavior is a fairly recent approach. Ablation of Dop1R1 leads to deficits in odor–sugar memory in adults but does not affect sucrose consumption (Y.-C. Kim et al., 2007), suggesting these behaviors are mediated by different DA receptors. More recent evidence suggests that Dop1R1 can act in parallel to Dop1R2 in the formation and updating of memories, as well as shaping behavioral responses (Cohn et al., 2015; Handler et al., 2019). Dop2R plays a key role in increased sucrose sensitivity and consumption in Drosophila raised on diets lacking sugar (Ganguly et al., 2021).

As a general rule, DA signaling plays a role in arousal, learning, and memory, as well as valence encoding, for a wide-variety of reward modalities. Some DA neurons encode multiple types of reward, whereas others are reward specific, and the amount of DA neuron involvement scales with reward intensity. Modularity of DA circuits within the MB plays a pivotal role in encoding different types of reward and eliciting appropriate behavioral responses. Finally, behavioral responses are shaped by the excitatory and inhibitory properties of DA and how this is altered by experience, internal state, and environmental context.

Octopamine

In parallel with dopamine (DA) signaling, acquisition of the olfactory cues associated with sucrose consumption also requires octopamine (OA) signaling (Figure 1C). Mutations in tyramine-β-hydroxylase (Tβh), the enzyme required for OA synthesis, or the OA receptor in mushroom bodies (OAMBs) reduces the ability of flies to form a positive odor–sucrose association (Y.-C. Kim et al., 2013; Schwaerzel et al., 2003) and decreases ethanol preference (Scheiner et al., 2014). Moreover, Drosophila will optogenetically self-stimulate OA neurons (Schneider et al., 2012). These OA neurons simultaneously act on Kenyon cells (KCs) and protocerebral anterior medial DA neurons through OAMBs in KCs and OA receptor Octopamine Beta-2 Receptor (OctB2R) in γ‎1pedc mushroom body (MB) DA neurons to signal positive reinforcement for sucrose consumption (Burke et al., 2012).

OA also plays a key role in how the internal state influences reward response. For example, hunger significantly suppresses the activity of OA neurons (LeDue et al., 2016). In addition, food- and sex-deprived flies will choose to participate in feeding behavior over courtship behavior, and this behavior is caused by a decrease in excitatory tyramine, a metabolite of OA, acting on P1 neurons to activate courtship and decreased inhibitory tyramine acting on Posterior Lateral Protocerebrum (PLP) neurons that activate feeding behavior (Cheriyamkunnel et al., 2021). Positive reinforcement from OA-releasing neurons is necessary and sufficient for ethanol attraction, with a very specific subset of OA neurons being required for manifestation of ethanol preference (i.e., G3a/AL2 and VMI–VMIII; Schneider et al., 2012). When Drosophila lack the rate-limiting enzyme for producing OA, they are unable to form ethanol preference, but this lack of ethanol preference is reversed if flies are prefed ethanol food at least 3.5 hours prior to testing (Claßen & Scholz, 2018). Together, these data suggest OA is required and necessary for innate ethanol attraction, but ethanol exposure itself could alter non-OA circuits to induce ethanol preference.

As a general rule, OA seems to play a role in general state setting that contributes to the direction in which other neurotransmitter systems influence reward behavior.

Serotonin

Serotonin (5-HT) also appears to play a key role in reward in the Drosophila brain (Figure 1D). Although evidence suggesting 5-HT directly encodes a reward response is lacking, current evidence points to its role in mediating how the internal state influences reward. Activation of a subset of 5-HT neurons induces starvation-like feeding in sated flies, suggesting that activation of these neurons elicits a hunger response, thereby increasing motivational response for sucrose (Albin et al., 2015). Similarly, serotonergic ellipsoid body R4 neurons increase food consumption when artificially activated (J.-Y. Park et al., 2016). In the subesophageal zone (SEZ), 5-HT neurons that encode stimuli with opposing valences dictate balanced sucrose consumption through feedback signals from enteric neurons (Yao & Scott, 2022).

Two different populations of 5-HT neurons are required for sucrose memory: neurons that project outside of the mushroom body (Albin et al., 2015) and the Dorsal Paired Medial (DPM) neurons that innervate Kenyon cell axons (Krashes & Waddell, 2008; Sitaraman et al., 2012). The DPM neurons also promote sleep (Haynes et al., 2015), which is required to learn an appetitively reinforced operant task in Drosophila (Wiggin et al., 2021). Finally, pharmacologically increasing 5-HT signaling decreases ethanol odor attraction, whereas prolonged signaling from the 5-HT neurons provides greater attraction to higher concentrations of ethanol (Xu et al., 2016). This may be because restriction of the 5-HT transporters to different neuronal compartments differentially alters ethanol attraction (Kasture et al., 2019).

Thus, 5-HT generally functions to modulate sucrose behavior and ethanol response, which could reflect its role in both natural and drug reward encoding. However, most evidence suggests that like OA, 5-HT modulates state setting, which contributes to how other neurotransmitter systems influence reward behavior.

Neuropeptides and Hormones

Similar to mammals, stress hormones and peptides modulate reward behavior in Drosophila (Figure 1E). Drosophila will optogenetically self-activate neurons expressing neuropeptide F (NPF), the Drosophila ortholog to neuropeptide Y (Shao et al., 2017; Wen et al., 2005). Activation of NPF neurons enhances preference for an odor associated with sucrose (Krashes et al., 2009), whereas knockdown of NPF receptors in shock-encoding dopamine (DA) neurons reduces preference for an odor associated with sucrose (Aso, Hattori, et al., 2014; Aso et al., 2012; Krashes et al., 2009).

NPF signaling is also critical for the increase in ethanol intake that occurs in males as a consequence of sexual rejection, and this occurs in concert with corazonin (Crz) signaling. During and after copulation, Crz neurons are activated, ultimately resulting in ejaculation and cessation of sexual behavior (Tayler et al., 2012). Flies will optogenetically self-activate Crz neurons and approach odors associated with optogenetic activation of these neurons (Zer-Krispil et al., 2018). Successful mating or activation of Crz neurons ultimately generates an increase in NPF levels in the brains of males (Zer-Krispil et al., 2018), which decreases the likelihood of engaging in reward-seeking behavior (Shohat-Ophir et al., 2012). Thus, optogenetic activation of Crz neurons in males is rewarding and gates release of NPF, which reduces ethanol self-administration in males by inducing a hedonic-like state (Shohat-Ophir et al., 2012). Together, these data suggest that monoaminergic signaling influences alcohol reward and preference, and peptidergic signaling appears to modulate alcohol attraction due to complex social interactions.

Crz-releasing neurons also contribute to food reward. RNA interference–mediated Crz receptor knockdown in adipocytes and salivary glands, as well as Crz production knockdown, significantly reduces food intake in flies (Kubrak et al., 2016). Similarly, gain of function of short NPF (sNPF) increases food consumption, whereas the loss of function of sNPF suppresses food consumption (K. S. Lee et al., 2004). Knockdown of sNPF in the mushroom body and sNPF receptors on protocerebral anterior medial (PAM) neurons reduces the efficacy of odor–sugar memory encoding (Knapek et al., 2013; Lyutova et al., 2019).

Hormones in Drosophila also play a key role in sugar reward. Neurons releasing diuretic hormone 44 (Dh44), orthologous to the mammalian stress hormone corticotropin releasing factor (Cannell et al., 2016; Lovejoy & Jahan, 2006), are key sensors for promoting food consumption (Yang et al., 2018). Dh44, released from neurons in the central brain, is necessary and sufficient for the detection and consumption of nutritive sugar (Dus et al., 2015). Dh44 binds to the Dh44 receptor in R1 neurons and R2 cells in the gut, which are required for behavioral responses to the nutritional value of sugar (Dus et al., 2015).

Allatostatin A (AstA) neurons function in response to hunger and thirst. Activating AstA neurons in food-deprived animals suppresses feeding, and this effect is reversed by activating NPF neurons (Hergarden et al., 2012). Additionally, AstA neurons act on the allatostatin A receptor 1 (DAR1) in PAM-γ‎3 DA neurons to suppress their activity after sucrose consumption. γ‎3 PAM inactivation mediates reward during sucrose consumption and associative learning (Yamagata et al., 2016). AstA also enhances water seeking while simultaneously suppressing feeding via AstA receptor 2 in NPF neurons (Landayan et al., 2021).

The neuropeptides drosulfakinin (Dsk) and tachykinin contribute to reward indirectly. Activating or silencing Dsk neurons (Agrawal et al., 2020; Wu et al., 2020) elicits social isolation-induced fighting behavior phenotypes in Drosophila (Y.-K. Kim et al., 2018). Activation of tachykinin-releasing neurons is sufficient to induce same-sex aggression in males (Asahina et al., 2014). Since winning or losing a fight influences subsequent risk taking, social behavior, and reward behavior (Y.-K. Kim et al., 2018; Trannoy et al., 2016), activity from both Dsk and tachykinin neurons could disrupt reward perception.

As a rule, hormone, peptidergic, and neurotransmitter circuits function together to directly modulate reward behavior, although the exact mechanisms through which this occurs remains to be understood.

Larval Reward Circuits

One of the advantages of studying Drosophila is the unique ability to compare circuit frameworks within the same animal across different life stages. Despite having 10-fold fewer neurons than adults, larvae appear to have remarkable similarities in reward circuit form and function. Emerging research suggests a rich network of feeding circuits in larvae, making larvae an ideal model to study the mechanisms of food reward. Much of the larval feeding circuitry has been mapped with high resolution and is organized into a series of parallel pathways that connect sensory to motor neurons with interneuron signaling to the mushroom body (MB; Miroschnikow et al., 2018, 2020).

In larvae, as in adults, a complex interplay of neuropeptidergic and dopaminergic circuits mediates the behavioral response to food (Figure 2). Feeding initiation is regulated by insulin signaling in the MB (Zhao & Campos, 2012), whereas subsets of neurons expressing cyclic guanosine monophosphate (cGMP)-dependent protein kinase regulate the amount of food consumed and triglyceride levels in larvae, suggesting they play a role in feeding cessation (Allen et al., 2018). Adipokinetic hormone (AKH)–releasing cells release AKH and receive modulatory input from midgut neuropeptide F (NPF) and Allatostatin-C (AstC) neurons to alter hunger and satiation (Nelson et al., 2021). Similarly, cholinergic neurons that also release the neuropeptide hugin suppress feeding behavior by increasing searching behavior (Melcher & Pankratz, 2005; Schlegel et al., 2016). Interestingly, these neurons project to distinct parts of the larval brain (Figure 2), although it is unclear how these circuits function together to guide food consumption in larvae.

As with adults, the dopamine (DA)–NPF circuit also plays a role in reward response in larvae. Neurons expressing NPF contribute to gustatory stimulation by sugar (Shen & Cai, 2001), contribute to odor-induced appetitive arousal (Pu et al., 2018), and interact with DA neurons to gate appetitive response (Wang et al., 2013). The Dopamine Neuron (DAN) k1 DA neurons (DAN-k1) directly contribute to fructose seeking and consumption (Saumweber et al., 2018). Similarly, ablation of dopamine 1 receptor 1 (Dop1R1) reduces odor–sugar memory of larvae but does not affect sucrose consumption (Honjo & Furukubo-Tokunaga, 2009; Selcho et al., 2009). Larval DA neuron activation is sufficient to induce reward memory (Thoener et al., 2022), suggesting this induces a pleasurable interoceptive state. DA neurons may be mediated by octopamine (OA) neurons (Eichler et al., 2017), as OA is required to rescue reduced preference for consuming sucrose in larval Drosophila exposed to artificial repeated failure in reward pursuit (Fei et al., 2018). Larval Drosophila will optogenetically self-stimulate Kenyon cells (KCs; Lyutova et al., 2019) and sugar-sensing gustatory neurons (Nuwal et al., 2012; Seidenbecher et al., 2020). The γ‎-aminobutyric acid (GABAergic) anterior paired lateral neuron, which broadly innervates KC axons and contributes to odor discrimination (Amin et al., 2020; Prisco et al., 2021), is also critical for appetitive memory in larvae (Amin et al., 2020; Sabandal et al., 2020; Saumweber et al., 2018).

Drosophila larvae display diversity in sensing and subsequent preference for amino acids (Kudow et al., 2017), which requires glutamatergic input to larval taste neurons (Croset et al., 2016). DA neurons contribute to Drosophila larvae by rejecting food that lacks essential amino acids (Bjordal et al., 2014). DL1 DA neurons in larvae are necessary and sufficient for the expression of preference for amino acids (Bjordal et al., 2014). When Drosophila larvae are reared in a protein-deficient diet, they reduce food seeking and consumption, which is dependent on OA neurons in the subesophageal zone (Bjordal et al., 2014; Fei et al., 2018). Together, this suggests that glutamate, OA, and DA function together to shape amino acid, and thus protein, preference. 5-HT also conveys information on the nutritional value of protein. Of the five known 5-HT receptors, serotonin 2A (5-HT2A) receptors are the main mechanism by which Drosophila establish the value of dietary protein (Ro et al., 2016), and 5-HT2A receptors are necessary for normal larval feeding (Gasque et al., 2013). Together, the above circuits function in larvae to sense and consume amino acids and proteins required for proper development.

Despite larvae being a different developmental stage, with large differences in morphology and behavior, their reward circuits look remarkably similar to those in adults. Peptidergic, hormone, OA, and 5-HT neurons signal information about arousal and internal state to DA neurons, which modulate valence and drive responses in cue-encoding neurons and premotor neurons required for eliciting reward behaviors.

Figure 2. Reward circuits in larval Drosophila. Acetylcholine-releasing neurons as well as GABA and DAergic neurons contribute to the expression of reward behavior. Acetylcholine-expressing Hugin neurons differ in their projection to protocerebrum (PC), pharynx (PH), and two populations that project to the ring gland (RG).

Discussion

As understanding of the structure and function of Drosophila reward circuits develops, it is increasingly clear why Drosophila is an effective model for comprehending key basic principles underlying reward circuit function. In both larvae and adult Drosophila, slow, long-acting neuropeptidergic neurons connect to neurons that release faster-acting neurotransmitters that influence premotor and motor neurons to control directionality (goal orientation). Reward circuits thus act in concert to integrate internal state and reward sensation to produce reward behavior, much like a five-piece band works together to produce music for an audience (Figure 3).

Figure 3. General model for reward signaling using a metaphor of a five-member band. Band components such as percussion (A1), bass (A2), guitar (A3), synthesizers (A4), and vocalists (A5) function similarly to state-setting (B1), arousal-eliciting (B2), valence-encoding (B3), cue-encoding (B4), and premotor/motor (B5) neurons to produce music or reward-related behaviors, respectively. Our inset circuit motif demonstrates the synapse between modulatory, sensory, and output neurons that contribute to reward behavior expression.

Octopaminergic or serotonergic state-setting neurons (Figure 3B1) act as percussive instruments (Figure 3A1), setting the tempo for the rest of the players. Slow-acting peptidergic neurons (Figure 3B2) work together with these state-setting neurons to layer rhythm, similar to the slow tones of a bass guitar (Figure 3A2). Bass guitars can alter the perception of music by providing notes that act as a bridge between treble (guitar) and percussion, similar to how neuropeptide F (NPF) can induce wakefulness that alters reward perception. Dopaminergic neurons (Figure 3B3) provide valence in the same way the lead guitar (Figure 3A3) drives the direction of the music. Cue-encoding Kenyon cells (KCs; Figure 3B4) sparsely activate in response to sensory information in the same way that a synthesizer (Figure 3A4) sporadically punctuates the music. Sequential activation of KC compartments helps encode the memory the way keys on a piano/synthesizer are activated to produce a melody (Owald & Waddell, 2015). Finally, the mushroom body output neuron (MBONs; Figure 3B5) are the star of the show as they guide the behavior in the same way the vocalist of the band communicates to the audience an overall message of the music through lyrics (Figure 3A5). Together, the percussion, bass, guitar, synthesizer, and vocalist act in concert to produce a song in a similar way that octopamine (OA), NPF, dopamine (DA), acetylcholine, and fast-acting neurotransmitter circuits directly produce reward behavior. Although this metaphor is based on mushroom body (MB) circuits, other circuits in the Drosophila brain encompass the described overlapping motifs for responses to sugar, water, and alcohol reward, suggesting it is a generalized model (Ichinose et al., 2015; Lin et al., 2014; Scaplen et al., 2020; Shyu et al., 2017; N. Yamagata et al., 2015).

This pattern of connectivity is repeated across different types of reward in Drosophila (Figure 4). Remarkably, this also appears to be functionally similar across development, across different structures within the same brain, and across species. This suggests that key principles extracted from how reward circuits are structured and function in flies can enhance knowledge of the function of reward circuits and brain systems pivotal for reward behavior in mammals (Borst & Helmstaedter, 2015; Hunter et al., 2021; Kaiser, 2015; Scaplen & Kaun, 2016).

Figure 4. Circuits for different rewards in Drosophila. Illustrations of neurons that are directly involved in (A) food, (B) water, (C) sugar, (D) alcohol, and (E) social reward in the adult Drosophila central brain (gray). Neuron types are identified by color (see key), and coverage is based on peer-reviewed publications referenced in the article or from the Janelia FlyLight Split-Gal4 Driver Collection.

Some of the key principles that can already be extracted include the following:

Rewards are encoded through the actions of monoaminergic, peptidergic, and hormone signaling within the brain.

Reward response is dependent on the internal state of the animal.

Reward encoding is dependent on timing and intensity of reward.

A compartmentalized code within reward structures defines reward response.

There are common circuits for all forms of reward.

Unique circuits for different rewards are dependent on the context in which the reward was received.

A noticeable gap in understanding includes identifying which neurons directly contribute to appetitive internal states and which elicit the appropriate appetitive response. This is further complicated by how prior experiences can influence internal state and reward experience. Sleep is an excellent example of this. Hungry flies sleep less than satiated flies, and hunger-induced sleep deprivation during memory consolidation reduces sucrose-associated odor cue seeking (Chouhan et al., 2021). This effect is dependent on the activity of the γ‎2α‎′1 DA neuron–MBON circuit (Chouhan et al., 2021).

DA neurons also likely contribute reward prediction error (RPE). Although direct data demonstrating this RPE are still lacking, the anatomy and function of the DA–MB circuit suggest it would be an ideal model to identify the detailed circuit mechanisms underlying this phenomenon (i.e., Adel & Griffith, 2021; Bennett et al., 2021). Future research using synthetic manipulation of neurons will help elucidate this and further interrogate how reward-encoding and reward-expressing circuits function distinctly and collaboratively to modulate appetitive behavior. Ultimately, understanding how experience and internal state alters neuromodulation and neural dynamics is key to understanding the mechanisms underlying motivational behavior for reward seeking and consumption.

Conflict of Interest

The authors report no conflict of interest in the development of this review article. The authors alone are responsible for the content and writing of this article.

Acknowledgments

The authors thank Dr. Kristin Scaplen (Bryant University) and the Kaun lab for their helpful comments on earlier versions of this review article, as well as the frontline health care workers and essential workers who, throughout the pandemic, made it possible to conduct research and writing. This work was written within the ancestral homelands of the Narragansett Indian Tribe and funded by the National Institute for Alcoholism and Alcohol Abuse (R01AA024434 to K.R.K., F32AA-29595-01 to J.S.H., and F31AA030219-01 to T.M.B).

References