Come Fly with Me: an overview of dopamine receptors in Drosophila melanogaster

Abstract

Dopamine (DA) receptors play critical roles in a wide range of behaviours, including sensory processing, motor function, reward and arousal. As such, aberrant DA signalling is associated with numerous neurological and psychiatric disorders. Therefore, understanding the mechanisms by which DA neurotransmission drives intracellular signalling pathways that modulate behaviour can provide critical insights to guide the development of targeted therapeutics. Drosophila melanogaster has emerged as a powerful model with unique advantages to study the mechanisms underlying DA neurotransmission and associated behaviours in a controlled and systematic manner. Many regions in the fly brain innervated by dopaminergic neurons have been mapped and linked to specific behaviours, including associative learning and arousal. Here, we provide an overview of the homology between human and Drosophila dopaminergic systems and review the current literature on the pharmacology, molecular signalling mechanisms, and behavioural outcomes of DA receptor activation in the Drosophila brain.

Introduction

G protein-coupled receptors (GPCRs) form a diverse superfamily of seven-transmembrane (TM) proteins that relay information from external signals by transducing the binding of a variety of ligands to the initiation of intracellular signalling cascades. The human genome codes for over 800 GPCRs, many of which are conserved among multiple species. They represent the most abundant therapeutic drug targets, with 34% of all drugs approved by the U.S. Food and Drug Administration acting at 108 unique GPCRs [1]. Of these, the dopamine D2 receptors act as targets for essentially all current clinical antipsychotic drugs [2]. Furthermore, aberrant DA signalling is associated with numerous neurological and psychiatric disorders, including Parkinson’s disease, restless leg syndrome, attention deficit hyperactivity disorder, depression, addiction, bipolar disorder and schizophrenia [3].

Upon binding of their ligand, DA receptors transduce information to the inside of the cell through conformational changes, which activate heterotrimeric G proteins and initiate downstream signalling pathways through the recruitment and activation of effector proteins [4]. Different DA agonists have been shown to engage distinct downstream effectors, as described by the phenomenon of biased agonism [5, 6]. Therefore, understanding the mechanisms underlying receptor signalling in vivo can provide critical insights to guide the development of pharmacological treatments of neuropsychiatric disorders. Over the past two decades, Drosophila has emerged as a powerful model to study DA signalling in vivo [7, 8]. In this MiniReview, we describe the homology between human and Drosophila DA systems and summarize the current literature on the pharmacology, signalling mechanisms and behavioural outcomes of DA receptor signalling in the Drosophila brain.

Dopaminergic signalling across species

DA Synthesis

Monoaminergic neurotransmitters such as DA and serotonin (5-HT) are present in Drosophila and regulate similar functions to those in mammals, including locomotion, drug response, circadian rhythms, aggression, attention, reward, and learning and memory [7]. The fly homologues of most, if not all, molecules involved in monoaminergic synaptic transmission, such as synthesis enzymes, receptors, and transporters have been identified. In flies, as in humans and other vertebrates, DA is synthesized from the amino acid tyrosine via two enzymatic steps. The first and rate limiting step is through the enzymatic action of tyrosine hydroxylase (TH), encoded by the pale (ple) gene in flies, which converts tyrosine into l-3,4-dihydroxyphenylalanine (L-DOPA) [9]. In the second step, L-DOPA is converted to DA by the action of the aromatic amino acid decarboxylase (AADC), encoded by the DOPA decarboxylase gene (Ddc) [9]. DA cannot cross the blood brain barrier and is synthesized within dopaminergic neurons in the brains of both flies and humans [10]. It is packaged into vesicles via the vesicular monoamine transporter (encoded by Vmat) and released via exocytosis into the synapse, where it binds to DA receptors on postsynaptic neurons, as well as DA neuron terminal autoreceptors and other terminal heteroreceptors, triggering an array of signalling pathways that ultimately modulate behaviour [9].

DA neurons

In humans, DA neurons make up less than 1% of the brain but are necessary for a range of behaviours. The four axonal pathways that relay DA signals include: (1) nigrostriatal, which degenerates in Parkinson’s disease and is involved in motor control; (2) mesolimbic, which is involved in reward and motivated behaviour; (3) mesocortical, thought to influence learning and memory; and (4) tuberoinfundibular, which participates in hormonal control [11]. The adult Drosophila brain has approximately 130 DA neurons that are subdivided into clusters [12] and, as in mammals, innervate different brain regions and mediate distinct behaviours such as locomotion, odor, appetite and circadian behaviour [13]. Two well-characterized sites of dopaminergic innervation in the Drosophila brain are the mushroom body, noted for its critical role in olfactory learning and memory [14], and the central complex, shown to play a role in motor activity and sleep [15].

DA receptor signalling

DA exerts its function through activation of D1-like or D2-like receptors, which activate or inhibit the adenylyl cyclase/cyclic adenosine monophosphate (cAMP) pathway, respectively. In humans, there are five DA receptors: The D1-like receptors, D1 and D5, and the D2-like receptors, D2, D3, and D4 [4]. In Drosophila, two receptors, Dop1R1 (aka dD1 or DUMB) and Dop1R2 (aka DAMB), are classified as D1-like, whereas Dop2R (aka DD2R) is the only D2-like receptor homolog [16–19]. These receptors share a number of conserved features common to all rhodopsin-like GPCRs, including a disulfide bond between cysteines in the first and second extracellular loop [20]. Like other biogenic amine receptors, they possess an aspartate residue in the TM3 that is necessary for ligand binding. Three serines in TM5 are also conserved; these interact with the meta- and para-OH of the catechol moiety of DA and are critical for binding as well as efficacy [21]. The P-I-F motif, a cluster of residues encompassing a proline in TM5, an isoleucine in TM3 and a phenylalanine in TM6, is also conserved. This motif, also known as the “transmission switch”, is critical for the conformational changes that occur following ligand binding and receptor activation [22]. An aspartate in TM2, the DRY sequence following TM3, and the NPxxY motif in TM7, all of which are important for receptor activation, are also conserved [20]. Additionally, flies have a non-canonical DA/ecdysteroid receptor, (DopEcR), which binds the arthropod ecdysteroids, ecdysone and 20 hydroxyecdysone (20E), in addition to DA [23]. Sequence comparisons show that it contains many of the above features, including the DRY and PIF motifs, and two of the three serines in TM5. The aspartate in TM3 is conservatively substituted as glutamic acid. Notably, the NPxxY motif is not conserved [20, 23].

DA receptors are GPCRs that transmit their signals by activating heterotrimeric G protein complexes inside the cell. The Drosophila genome encodes a comprehensive yet reduced repertoire of Gα protein subunits: 8 genes in flies (6 characterized and 2 predicted), compared to 18 in humans [24, 25]. Still, all human subgroups are represented in flies. Gαi and Gαo belong to the Gαi/o subgroup, Gαq belongs to the Gαq/11 subgroup, Gαs and Gαf belong to the Gαs subgroup and concertina (cta) belongs to the Gα12/13 subgroup. The β- and γ-subunit repertoire is also reduced compared to that in mammals; only 3 Gβ and 2 Gγ genes are present in flies as compared to 5 Gβ and 12 Gγ genes in vertebrates [24, 25]. The downstream effectors essential for G protein-mediated signalling, including homologs of adenylyl cyclases which activate cAMP, phosphodiesterases (PDE) which degrade cAMP, and Rho guanine nucleotide exchange factors (RhoGEFs), have also been identified in flies [26]. The adenylyl cyclase gene rutabaga and PDE gene dunce are the most well-studied regulators of cAMP signalling in flies, in addition to the downstream components protein kinase A (PKA) and the transcription factor cAMP response element-binding protein (dCREB) [26].

Other components of GPCR-mediated signal transduction are also conserved in flies, including the phosphorylation of activated receptors by GPCR kinases (GRKs), and the subsequent recruitment of arrestin proteins, resulting in desensitization and receptor internalization. The Drosophila genome encodes 2 GRKs, Gprk1 and Gprk2 (compared to 7 in humans) [27, 28], 2 visual arrestins, arrestin 1 and arrestin 2, and a single non-visual arrestin, Kurtz [29, 30]. In HEK-293 cells transfected with Dop1R1, DA application resulted in translocation of a GFP-Kurtz fusion protein to the membrane [31]. Another study that adapted the TANGO assay to flies took advantage of Dop1R1’s capacity in vivo to recruit arrestin 1 upon DA binding, and used it to map sites of DA release [32]. Still, Gprk phosphorylation of DA receptors and Kurtz-mediated receptor internalization have not been explored in flies and, as yet, there have been no studies of arrestin signalling downstream of DA receptors in this model system.

D1-like receptors

Of the four DA receptors cloned and characterized in Drosophila, two have been classified as D1-like, based on their demonstrated ability to stimulate an increase in intracellular cAMP in the presence of DA: Dop1R1 and Dop1R2 [9]. Dop1R1 is more closely related to the vertebrate D1-like receptors, with the highest degree of sequence identity observed in the TMs (~ 50%) [17, 33]. Interestingly, the gene encoding Dop1R1 varies among different strains of Drosophila. The allele isolated from the wildtype Berlin strain [17] was shown to encode a longer N-terminus (142 vs. 16 amino acids) than that isolated from the wildtype strain Canton S [33]. However, the biological significance of the different isoforms has yet to be determined.

Sequence comparison between Dop1R1 and vertebrate D1-like receptors indicate that the 3rd cytoplasmic loop is not well conserved, and while some motifs known to be required for G protein coupling are present, others are not [33]. Studies show that in insect SF9 cells transiently expressing Dop1R1, treatment with DA or the DA receptor agonist ADTN (2-amino-6,7-dihydroxy-1,2,3,4-tetrahydronaphthalene) stimulates the production of cAMP with high affinity, relative to norepinephrine (NE) and 5-HT with DA>NE>>5-HT [33]. DA agonists such as SKF-82526, SKF 38393, N-n-Propylnorapomorphine (NPA), and apomorphine, also stimulated cAMP production, but to considerably (2–5 fold) lower levels than DA. Known vertebrate D1-targeting antagonists such as butaclamol, SCH-23390, and flupentixol inhibited DA-induced increases in cAMP, albeit with very poor affinity [33]. Other DA agonists and antagonists also failed to behave as expected [33], suggesting that while the DA binding site is conserved, mutations that occurred after the divergence of invertebrates and vertebrates may have led to pharmacological divergence. Recent bioluminescence resonance energy transfer (BRET) analyses using Dop1R1 expressed in human embryonic kidney (HEK293) cells in culture revealed that Dop1R1 can couple to all Gαi/o subunits, G15, and Gs/olf proteins, but the fastest kinetics were observed with Gs and Golf [34]. This was similar to the selectivity demonstrated by human D1 receptor in the study, which was used as a reference. Taken together, these data indicate that Dop1R1 is functionally related to the family of D1-like receptors, despite its divergence structurally and pharmacologically.

Dop1R2, the second Drosophila DA receptor to be cloned, is more closely related to the invertebrate octopamine receptors than to the members of the vertebrate D1-like receptors [16, 18]. Despite these structural differences, studies in SF9 and HEK cells have shown that Dop1R2 can increase cAMP levels when stimulated with DA, similar to D1-like receptors [16, 18, 35]. This effect is specific to DA; 5-HT, octopamine and tyramine application all failed to increase cAMP levels above background, whereas stimulation with NE led to an increase but with ~100 fold lower potency than DA [18].

In addition to elevating cAMP levels, agonist stimulation of Dop1R2 was also found to increase intracellular calcium (Ca²⁺) levels, monitored as changes in an endogenous inward Ca²⁺-dependent chloride current in Xenopus oocytes [16] or changes in BRET signalling using a Ca²⁺ sensor in HEK293 cells [34]. The elevation of Ca²⁺ levels mediated by Dop1R2 appears to be rapid and biphasic. DA treatment of oocytes expressing Dop1R2 initiates a transient inward current that peaks within 20 sec of DA application. This is followed by a second, more variable, slower component of the response that declines to basal levels with a time course of 10–15 min. In contrast, the increase in cAMP levels in response to Dop1R2 activation occurs over a slower time course. The response appears after a lag period of at least 3 min after DA application, it then peaks after 10 min of exposure and remains elevated for up to 30 min [18, 35].

The different responses to stimulation of Dop1R2 appear to be mediated through distinct G protein signalling pathways. The DA-induced Dop1R2-mediated increase in cAMP levels in Xenopus oocytes was blocked by pertussis toxin (PTX), suggesting that this response is mediated by Gαi/o-like G proteins [35]. These data are consistent with findings that exogenous expression of PTX in the fly brain blocks DA-induced Dop1R2-mediated effects on arousal (see below). Although the fly Gαi is highly homologous to its mammalian counterparts, it lacks a cysteine within the C-terminus that acts as a substrate for PTX-catalyzed ADP-ribosylation [36]. As such, Gαo is the only predicted substrate for PTX in flies, and this has been confirmed in a particulate preparation from fly heads [37]. This suggests that Dop1R2 couples to Gαo to increase cAMP, an unusual transduction mechanism for a D1-like receptor, which typically activates cAMP activity via a PTX-insensitive stimulatory G protein (Gαs). Interestingly, injections of a βARK1-CT GST fusion protein, which blocks Gβγ signalling, also inhibited the DA-mediated increases of cAMP levels in oocytes [35]. These data indicate that the Gβγ subunits play a role in the stimulatory effects of the Dop1R2 receptor on adenylyl cyclase (AC) activity, likely through activation of AC2, AC4 and/or AC7, which have been shown to be activated by Gβγ released from the Gαoβγ heterotrimer [38].

In contrast, the inward Ca²⁺-dependent chloride currents were not altered by pre-exposure to PTX or βARK1-CT GST [35]. More recently, BRET analyses of HEK cells expressing Dop1R2 revealed that the receptor exhibits a much different coupling profile compared to other D1-like receptors [34]. It activated Gαo, Gαz, Gαq, Gα11, Gα14, Gα15 and Gαs/olf proteins to differing extents, but exhibited the most rapid activation kinetics for Gαq and Gα11, with Dop1R2 activating Gαq at a 100-fold lower concentration of DA than Gαs. Given that Gαq has been shown to stimulate phospholipase Cβ resulting in Ca²⁺ release from internal stores, it is possible that it is the G protein that mediates the increases in intracellular Ca²⁺ triggered by Dop1R2 activation. Consistent with this, knockdown of Gαq in the fly brain inhibits Dop1R2-mediated forgetting.[34]. Of note, a pharmacological study in Drosophila neurons in primary culture showed that the action of DA agonists varied depending on the cell type used. Specifically, SKF38393 and 6-chloro-APB mimicked DA-mediated suppression of cholinergic synaptic transmission with similar potency to that reported for vertebrate D1-like receptors [39]. This is different from what was observed in heterologous cell systems expressing Drosophila D1-like receptors, where these same agonists exhibited poor potency relative to DA [16, 18, 39]. It is important to note though that the studies in heterologous cells were mostly measuring receptor-induced increase in cAMP, whereas the effect of agonists in primary culture appear to be mediated through a cAMP-independent pathway. These data highlight the importance of studying Drosophila receptors in their native environment to fully uncover their signalling and pharmacological properties.

D2-like receptor

Sequence comparison of the Drosophila DD2R protein with known D1-like and D2-like receptors suggests that it falls within the D2 subfamily. Paired alignments with D2-like receptors (D2, D3, and D4) revealed amino acid identity ranging from 29% to 32%, in addition to structural features that distinguish D2-like receptors from D1-like receptors, including a relatively long 3rd intracellular loop and a very short carboxyl terminus [19]. While the initial characterization suggested that DD2R was similar in length to its vertebrate homologs, more recent data based on RNA sequencing suggest that the fly receptor includes a much longer N-terminal sequence (~300 aa) [40, 41]. In addition, alternative splicing of the 7^th and 8^th exons results in a structurally diverse subfamily of D2-like receptors that differ in the size and composition of the 3rd intracellular loop [19]. This is similar to what has been shown in the corresponding human homologs, where a short, a long, and an extra-long form of the D2 receptor have been identified and characterized [42]. In both flies and humans, pharmacological and signalling studies have not identified major signalling differences amongst the different variants. It is well established that alternative splicing of GPCRs may be associated with tissue- or cell-specific expression. With the mammalian D2R, the short isoform is expressed predominantly in dopaminergic neurons and functions as an autoreceptor, whereas the long isoform is localized to postsynaptic sites, where it mediates locomotor as well as other physiological functions [43]. Whether Drosophila D2-like receptor splice variants demonstrated the same tissue-specific distribution remains to be determined. Recently, repeated administration of alcohol in flies was shown to result in a switch in expression of two DD2R isoforms that differ in a single codon resulting in the inclusion/exclusion of a serine located in the 3rd cytoplasmic loop [41]. Although the functional relevance of this switch was not identified, the relevant serine is conserved in human D2R and is located in the receptor’s 3^rd cytoplasmic loop, where it could conceivably be a target of phosphorylation by modulatory kinases.

DA significantly stimulated DD2R variants expressed in HEK293 cells [19]. Functional characterization revealed that, like their mammalian homologs, the DD2R isoforms couple to inhibitory G proteins. Stimulation of HEK293 cells expressing DD2R with the adenylate cyclase activator forskolin results in an increase in cAMP that is markedly inhibited by addition of DA. This DA-induced effect was inhibited by pretreatment of these cells with PTX, suggesting it is mediated by Gαi/o signalling [19]. Several known D2-like receptor agonists were able to stimulate the receptors. Notably, bromocriptine fully stimulated each of the Drosophila D2R isoforms tested, with nanomolar potency [19]. Other agonists, including the D2-like receptor agonist NPA as well as several less D2-selective agonists also triggered significant levels of ligand-induced signalling [19]. Other D2-like receptor agonists, such as quinpirole, failed to significantly stimulate DD2R-mediated signalling in vitro [19]. Quinpirole application was, however, shown to induce reflexive behavioural responses in decapitated flies [44] and to decrease evoked DA release in an in vivo preparation [45]. The reason for this paradoxical behaviour of quinpirole is not clear but again points to the importance of studying receptor pharmacology in a native context. Of the tested human DA receptor antagonists, only butaclamol and flupentixol showed weak activity at the fly DD2Rs in HEK293 cells, whereas raclopride, eticlopride, spiperone and haloperidol showed no effect [19]. Interestingly, substitution of 3 amino acids in human D2R with the homologous DD2R residue led to marked loss of potency and efficacy of the D2 agonist pergolide [46]. Conversely, substitution of the same human residues into the DD2R significantly enhanced pergolide efficacy and potency at the fly receptor. The substitution also enhanced the potency of the agonists piribedil and ropinirole, which did not show appreciable activity at the wild-type DD2R but exhibited partial agonist activity at the mutated receptor [46]. These findings illustrate the value of such comparative studies for uncovering important pharmacological insights into the structure-function relationships of DA receptors and GPCRs in general.

DopEcR Receptor

As mentioned above, binding studies have shown that the DopEcR receptor binds both DA and the hormones ecdysone and 20E with high affinity [23]. Sequence comparison with known vertebrate GPCRs shows that DopEcR is most highly homologous to the β-adrenergic receptors (26% identity and 45% similarity to the human receptor), with higher homology observed in the TM domain. In contrast, it exhibits 15–24% identity and 31–37% identity when compared to either Drosophila or human DA receptors [23]. Studies in heterologous expression systems using Chinese hamster ovary (CHO) cells and SF9 cells showed that, of a wide range of biogenic amines tested, only DA (at 10 µM) was able to produce a significant increase in cAMP levels. The action of DA was both more potent and more efficacious in SF9 cells, and the increase in cAMP levels was higher when DA was applied to forskolin-treated cells. Of the known DA agonists tested, only PD 128,907 and 6-Chloro-APB showed activity. The DA receptor antagonists spiperone and flupentixol, as well as (+)- butaclamol to a lesser degree, were all capable of inhibiting DA-induced increase in cAMP levels in SF9 cells expressing DopEcR, whereas R(+) SCH23390 had no effect [23].

Treatment of cells in culture with ecdysteroids alone did not lead to significant changes in cAMP levels in cells expressing DopEcR [23]. Instead, addition of ecdysone or 20E was shown to inhibit DA-induced, DopEcR-mediated accumulation of cAMP in a dose-dependent manner. This effect appeared to be mediated by direct binding of ecdysone to DopEcR rather than through non-specific effects on the cAMP pathway, as ecdysone did not inhibit cAMP accumulation mediated by the activation of either Dop1R2 or the octopamine receptor [23]. Still, these studies may not have captured the full range of signalling events downstream of DopEcR activation in vivo, as flies that are acutely fed 20E exhibit an increase in cAMP levels in the MB, in a DopEcR-dependent manner [47]. Acute 20E feeding or overexpression of DopEcR has also been shown to rescue a courtship memory defect caused by mutation of the rutabaga gene [47]. These data suggest that binding of ecdysteroids to DopEcR can indeed activate cAMP signalling, given the appropriate cellular milieu. Alternatively, it is possible that 20E feeding stimulates DA release, which in turn acts at the DopEcR receptor to stimulate cAMP production.

The relationship between the binding sites of DA and ecdysone also remains to be determined. Binding studies have shown that ecdysone can displace bound DA from the receptor, but not the other way around [23]. It is therefore possible that their binding sites partially overlap, given that both molecules contain dihydroxylated ring structures. Alternatively, it is also possible that the sites are independent but interact allosterically.

The coupling profile of DopEcR to G proteins has not been described. Nevertheless, studies aimed at identifying downstream second messenger pathways have demonstrated that DopEcR exhibits agonist-specific coupling. Specifically, activation of DopEcR by DA leads to the phosphorylation of Akt (Protein Kinase B) in both CHO and SF9 cells but has no effect on the phosphorylation levels of the extracellular signal–regulated kinases ERK1 and ERK2. Conversely, activation of DopEcR by ecdysone leads to phosphorylation of ERK½ but had no effect on the phosphorylation of Akt. These data indicate that DA-induced signalling is mediated by the phosphatidylinositol 3-kinase (PI3K) pathway, whereas ecdysteroid binding activates the mitogen activated protein kinase (MAP Kinase) pathway [23]. Furthermore, while in vitro assays did not detect any changes in levels of intracellular Ca²⁺ in response to either DA or ecdysteroids, studies in vivo using the Ca²⁺ sensor GCaMP have demonstrated both DA- and ecdysone-induced, DopEcR-dependent increases in Ca²⁺ influx. Application of ecdysone increases nicotine-induced Ca²⁺ in the calyx and cell body region of the MB, whereas application of DA increases Ca²⁺ influx in response to sugar in gustatory receptor neurons. In both cases, the increase in Ca²⁺ influx was dependent on DopEcR [48].

Taken together these data suggest that signalling through DopEcR is agonist-specific and dependent on cellular context. It would be interesting to determine whether the receptor’s interaction with its different ligands leads to differential G protein coupling (biased agonism). Furthermore, it was recently proposed that DopEcR may function similarly to the mammalian GPCR for oestrogen GPER1, which is known to mediate the rapid, non-genomic actions of the steroid hormone oestrogen [49]. GPER1 shares downstream effectors with DopEcR, modulating Akt, EGFR/ERK, and cAMP signalling pathways [50, 51]. When tested in vitro, GPER1 was shown to respond to DA in a dose-dependent manner to increase cAMP similar to DopEcR [49, 52]. GPER1 signalling has also been shown to play a role in dopaminergic neuroprotection [49]. Still, the direct modulation of GPER1 activity by DA in vivo has yet to be demonstrated and merits further investigation.

Behaviours mediated by DA receptors in Drosophila

All 4 Drosophila DA receptors have been shown to play critical roles in behaviours ranging from locomotion, circadian rhythm, reward, motivation, learning, memory, courtship and aggression [34, 47, 53–64]. Below, we highlight recent findings that have aimed at identifying the signalling pathways and effectors downstream of receptor activation that modulate behaviour, with a focus on the role of DA in learning and memory, as well as arousal.

Learning and memory

Several clusters of DA neurons innervate the mushroom body (MB) neuropil, a fly brain structure that is critical for associative learning and memory [8]. The MB receives information from multiple sensory centers in the brain, including the olfactory system. This sensory information is then integrated with dopaminergic information about stimulus valence, and encoded into long-term memory via the activation of MB output neurons. The punishing or reinforcing nature of a stimulus is largely determined by the type of DA neurons relaying the signal and the specific MB regions they innervate. This spatial separation of behavioural outputs presents intriguing parallels to the direct/indirect pathways of midbrain DA projections to the basal ganglia in rodents, whereby activation of D1 or D2 receptors differentially influences reward learning or punishment avoidance, respectively [65]. It remains unclear, however, whether aversive and appetitive input in flies also involves differential DA receptor activation, and therefore understanding the signalling mechanisms downstream of the receptors will be critical to determine whether these circuits are conserved. The complex and highly regulated process of associative memory formation and its modulation by DA have been reviewed in great detail previously [66, 67, 8, 14]. Here, we will focus our discussion on recent studies that have aimed to uncover the signalling pathways that modulate learning and memory downstream of DA receptor activation in vivo.

Research into learning and memory in flies has firmly established a role for cAMP signalling. Disruptions of the G protein α subunits Gs and Go [37, 68], as well as genes encoding downstream effectors such as dunce [69], rutabaga [70], PKA [71], and dCREB2 [72], all cause deficits in learning and/or memory formation. Consistent with a role for DA in this process, fluorescence resonance energy transfer (FRET)-based imaging of brains ex vivo using genetically-encoded biosensors has shown that stimulation of dopaminergic neurons evokes consistent, compartmentalized, and receptor-dependent elevation of cAMP and PKA in the MB [73]. Furthermore, high expression of DA receptors has been reported in the MB. Specifically, single-cell transcriptomic analyses show abundant expression of various combinations of the receptors in different MB cell clusters, with 24% of cells expressing all four receptors and only 5% expressing no DA receptors [74].

Dop1R1 signalling in the MB has been shown to encode long-term memory by integrating high DA input with odour-evoked Ca²⁺, leading to an increase in the activity of MB output neurons [57, 75]. In a series of mechanistic studies investigating aversive memory, it was also shown that over time, as released DA levels decrease, Dop1R2 acts in the same neurons as Dop1R1 to mediate forgetting [57]. Given the differential G protein activation profiles these two receptors exhibit, the authors hypothesized that this paradoxical phenomenon is made possible through signalling via differential coupling of Dop1R1 and Dop1R2 to Gαs and Gαq during high or low DA states, respectively [34, 57]. Consistent with this, pan-neuronal or MB-specific RNAi knockdown of Gαq in adulthood led to impaired forgetting and enhanced memory performance in flies [34]. Downstream of Dop1R2 activation, a signalosome scaffolded by the cell polarity determinant scribble (scrb) appears to mediate active forgetting. Scrb was found to interact physically and genetically with the small G protein Ras-related C3 botulinum toxin substrate 1 (Rac1), the p21-activated protein kinase Pak3, and the actin depolymerizing agent Cofilin, and knockdown of all these proteins in the MB impairs normal memory loss [76, 77].

Functional interplay between receptors was also shown to be necessary for the consolidation of appetitive memory, which is dependent on the timed activation of an inhibitory feedback loop regulating presynaptic DA neuron activity. These DA neurons express a GABA receptor and form a circuit with GABAergic MB output neurons that express both DD2R and Dop1R2 [78]. This feedback loop is critical, given that long-term stimulation of DA neurons was shown to mediate aversion [78, 79], and is similar to one that was described in the mouse mesolimbic system, in which feedback from inhibitory neurons prevents the over-activation of DA signalling [80].

Signalling through DD2R in the GABAergic anterior paired lateral (APL) neurons that innervate the MB has also been shown to be critical during aversive conditioning, and functions to restrain GABAergic inhibition. Knockdown of either DD2R or Gαo specifically in APL neurons impairs associative learning [81]. Consistent with this, imaging studies using a novel, exogenously expressed DA sensor showed that DA inhibits APL neuron activity by activation of G protein signalling downstream of DD2R [81]. DD2R has also been postulated to modulate food-seeking behaviour and olfactory associative learning by functioning as an autoreceptor [61, 63, 82]. Consistent with this, of the four receptors only DD2R was found to be expressed in DA neurons [74]. Furthermore, administration of the D2 agonist bromocriptine decreased evoked DA release, as measured by fast-scan cyclic voltammetry in vivo [45].

Sleep and Arousal

The central complex (CC), a major locomotion center in the fly brain, is made up of five neuropils, the fan-shaped body, ellipsoid body, protocerebral bridge and paired noduli [15, 83]. Disruptions of CC circuitry lead to disruptions in both sensorimotor processing and sleep functions [15]. Multiple lines of evidence suggest that the CC is highly homologous to the vertebrate basal ganglia. Both are critical for the selection and maintenance of adaptive behaviour, and many of the same developmental growth factors aid in patterning both brain structures. There is also evidence that the CC, similar to the basal ganglia, uses GABAergic inhibitory projections paired with reciprocal glutamatergic pathways to aid in action selection [15].

Like the basal ganglia, the CC receives dense dopaminergic innervation and exhibits high expression of the various DA receptors [64, 74, 84–87]. FRET-based imaging has shown that application of DA leads to an increase in cAMP in the fan-shaped body that can be blocked using a DA receptor antagonist [88]. Consistent with this, Dop1R1 and Dop1R2 have been shown to mediate arousing signals that are mediated by exposure to stimulant drugs, mechanical stimulation or specific genetic manipulations [58, 59, 86, 88–94].

Details of the signalling events downstream of receptor signalling in the CC complex have also started to emerge. Optogenetic activation of DA neurons or pressure ejection of DA onto dorsal fan-shaped body neuron dendrites were shown to induce hyperpolariziation of fan-shaped body neurons and suppress their spiking, inducing arousal [55]. This DA-induced neuronal hyperpolarization is inhibited by knockdown of Dop1R2 specifically in the fan-shaped body. Exogenous expression of PTX in the fan-shaped body, or its injection in the patch pipette also inhibits hyperpolarization in response to DA, consistent with the response being mediated by coupling of Dop1R2 to Gαo. Furthermore, while a single pulse of DA was shown to transiently inhibit the spiking of fan-shaped body neurons within seconds, prolonged DA application led to lasting suppression of excitability, whereby the neurons remained quiescent even after the DA signal is terminated [55]. Potassium conductance, in the form of voltage-gated A-type currents carried by the Kv potassium channel Shaker, was shown to play a critical role in the initial arousing response. Interestingly, the transition to quiescence was accompanied by a downregulation of these currents and the upregulation of voltage-independent leak currents through a two-pore-domain potassium (K2p) channel [55]. Remakably, further mechanistic analyses recently linked this pathway to sleep need. Fan-shaped body neurons were shown to register sleep loss-induced elevation of reactive oxygen species (ROS) through a nicotinamide adenine dinucleotide phosphate (NADPH) cofactor bound to the oxidoreductase domain of Shaker’s KVβ subunit. The oxidation of the cofactor was shown to slow the inactivation of the A-type current and boosts the frequency of action potentials, thereby promoting sleep [95].

Conclusions

The first Drosophila DA receptor was cloned and characterized in the mid 1990s. The two and a half decades since have seen significant strides in developing the knowledge base and tools necessary to use the power of fly genetics to explicate the mechanisms that modulate dopaminergic neurotransmission. Many of the circuits involved in DA-mediated behaviours, including learning, memory, locomotion, arousal and sleep have been mapped [12, 84, 86, 91, 96]. The efforts have uncovered the complexity and diversity of post-synaptic signalling through DA receptors to modulate these behaviours in vivo. These studies have highlighted several criteria that determine which receptor is activated to relay the DA signal. These include the type of receptor expressed in specific neurons, the levels of neurotransmitter released, as well as the levels of expression, and the different G proteins to which distinct receptors can couple. Findings have revealed several similarities between the fly and vertebrate systems. The fly D1-like receptor Dop1R1 and the DD2R appear to signal via Gαs and Gαi/o, respectively, like their vertebrate counterparts. Dop1R1 further seems to modulate a direct pathway that drives motor activity and arousal in the CC, in a similar manner to that observed in the vertebrate basal gaglia. In mammals, interplay between the direct and indirect pathways results in action selection by disinhibition of a selected motor or reward program and the simultaneous inhibition of other competing actions [97]. Evidence of an indirect pathway regulating motor behaviour in flies has not been established; however, the function of DD2R has not been explored comprehensively and future work that is focused on its role in motor function and arousal may help uncover a comparable interplay in flies. Finally, the signalling properties of Dop1R2 diverge from those of mammalian D1-like receptors, but are intriguing in their diversity in response to various signals and could model novel alternative signalling mechanisms that have yet to be fully explored in mammalian GPCRs.

As these general themes start to emerge, much work needs to be done to delineate the precise signalling mechanisms downstream of DA receptors in flies. Most studies aimed at identifying the G protein activation profiles of the various Drosophila receptors have been done in vertebrate cells that express vertebrate G proteins. Monitoring G protein activation in insect cells using assays based on fluorescence and luminescence are therefore critical. Moreover, genetically-encoded FRET-based sensors are emerging, which will allow researchers to query DA release and signalling in vivo. These include the GPCR-activation-based-DA sensor (GRAB_DA) [98], engineered by inserting a conformationally sensitive circular-permutated EFGP (cpEGFP) into the 3^rd extracellular loop of a DA receptor, and the Tango sensor [32], which couples the arrestin signalling pathway to the expression of a reporter gene. Using the UAS-GAL4 system, these sensors can be expressed in specific tissue to identify neurons that receive DA input in response to various behavioural stimuli. UAS-driven RNAi transgenes can be used to knock down individual G proteins to identify the precise effectors mediating behaviour [99]. Similarly, this approach can be used to study the role of β-arrestin, which is emerging as an important G protein-independent secondary messenger, downstream of DA receptors. Additionally, given data showing that multiple DA receptors are expressed in the same neurons, it will be important to identify the specific receptor(s) required in distinct neuronal populations that modulate behaviour and to characterize the functional interplay between co-expressed receptors. Finally, as genome editing technologies (e.g. TALEN and CRISPR-Cas9) are being increasingly streamlined and optimized [99], it will be possible to engineer endogenously expressed DA receptors and effector molecules fused to FRET sensors to monitor DA release and signalling in brains ex vivo or potentially in living flies.