Octopamine in the mushroom body circuitry for learning and memory

Abstract

Octopamine, the functional analog of noradrenaline, modulates many different behaviors and physiological processes in invertebrates. In the central nervous system, a few octopaminergic neurons project throughout the brain and innervate almost all neuropils. The center of memory formation in insects, the mushroom bodies, receive octopaminergic innervations in all insects investigated so far. Different octopamine receptors, either increasing or decreasing cAMP or calcium levels in the cell, are localized in Kenyon cells, further supporting the release of octopamine in the mushroom bodies. In addition, different mushroom body (MB) output neurons, projection neurons, and dopaminergic PAM cells are targets of octopaminergic neurons, enabling the modulation of learning circuits at different neural sites. For some years, the theory persisted that octopamine mediates rewarding stimuli, whereas dopamine (DA) represents aversive stimuli. This simple picture has been challenged by the finding that DA is required for both appetitive and aversive learning. Furthermore, octopamine is also involved in aversive learning and a rather complex interaction between these biogenic amines seems to modulate learning and memory. This review summarizes the role of octopamine in MB function, focusing on the anatomical principles and the role of the biogenic amine in learning and memory.

The biogenic amine octopamine (OA) is a neurohormone, neurotransmitter, and neuromodulator in invertebrates (for review, see Roeder 1999, 2005; Farooqui 2007). OA is synthetized by tyramine (TA), a neuromodulator by itself, which is produced from the amino acid tyrosine (Barker et al. 1972; Livingstone and Tempel 1983; Monastirioti et al. 1996; Cole et al. 2005). OA and TA have been shown to modulate a range of behaviors, including feeding, locomotion, sleep, aggression, courtship, and learning and memory (Hammer 1993; Roeder 2005, 2020; Crocker et al. 2010; Aonuma and Watanabe 2012; Kim et al. 2013; Stevenson and Rillich 2016; Iliadi et al. 2017; Youn et al. 2018; Sabandal et al. 2020). By activating or inhibiting neuronal circuits, they are thought to orchestrate behavior (Sombati and Hoyle 1984; Claßen and Scholz 2018). OA especially seems to bias the decision between behavioral outcomes, like courtship versus aggression or feeding versus locomotion, but also functions as an independent transmitter (Sombati and Hoyle 1984; Certel et al. 2010; Andrews et al. 2014; Sayin et al. 2019; Selcho and Pauls 2019). The mushroom body (MB) has been demonstrated to be involved in behaviors such as locomotion, sleep, courtship, and, most notably, learning and memory, all of which are affected by OA (Martin et al. 1998; McBride et al. 1999; Heisenberg 2003; Popov et al. 2004; Davis 2005; Joiner et al. 2006; Pitman et al. 2006; Fiala 2007; Busto et al. 2010; Bushey and Cirelli 2011; Bushey et al. 2011; Kaun et al. 2011). Furthermore, OA neurons (OANs) innervate the MB, and OA receptor expression has been identified in Kenyon cells in insects (Han et al. 1998; Cayre et al. 1999; Grohmann et al. 2003; Busch et al. 2009; Sinakevitch et al. 2011; Kim et al. 2013; Wu et al. 2013; Shih et al. 2019). The role of OA in the MB is predominantly investigated in the genetically tractable fruit fly; thus, this review focuses on the current knowledge in Drosophila. However, studies in other invertebrates suggest a conserved role of biogenic amines in MB-driven behaviors like learning and memory (for review, see Roeder 2005; Verlinden et al. 2010; Perry and Barron 2013; Van Damme et al. 2021).

Octopaminergic neurons involved in learning and memory

The first hints that OA in the MB is involved in learning and memory came from experiments performed in honeybees (Mercer and Menzel 1982). OA injection in the peduncle next to the calyx enhanced memory formation and recall of the rewarded stimulus and injection of OA into the calyces could substitute the sugar reward in olfactory conditioning (Bicker and Menzel 1989; Hammer and Menzel 1998). The VUMmx1 neuron of the honeybee was suggested to mediate the sugar reward for olfactory learning, as depolarization substitute for the sugar reward in olfactory learning (Hammer 1993; Hammer and Menzel 1998). Indeed, optogenetic and thermogenetic activation, respectively, of almost all OA/TA neurons in Drosophila was sufficient to substitute the sugar reward during olfactory learning (Schroll et al. 2006; Burke et al. 2012). OA was supposed to signal rewarding stimuli, whereas dopamine (DA) was suggested to mediate aversive stimuli to the MB where coincidence detection of an odor leads to olfactory learning and memory (Farooqui et al. 2003; Schwaerzel et al. 2003; Mizunami et al. 2009; Tomchik and Davis 2009). This simple model was challenged by the finding that DA neurons mediate the appetitive and aversive stimuli in olfactory learning and the observation that OA represents sweet taste rather than the nutrient value of the sugar reward in olfactory learning, potentially signaling the perception/presence of reward (Kim et al. 2007; Selcho et al. 2009, 2014; Burke et al. 2012; Mizunami et al. 2015). Different sets of DA cells modulate aversive or appetitive learning via the DA receptor dDA1R in the MB (Kim et al. 2007; Claridge-Chang et al. 2009; Aso et al. 2010; Liu et al. 2012). In addition, thermogenetic activation of OANs paired with an odor introduced an appetitive memory in flies, only if DA signaling was intact (Burke et al. 2012). This further suggested that OA plays a minor role in learning and memory.

On the other hand, recent publications demonstrate the role of OA in appetitive as well as aversive learning and indicate a rather complex action of both biogenic amines in learning and memory. Tyramine β hydroxylase (TβH) mutants, lacking OA, showed impaired appetitive and aversive learning that could be rescued by TβH expression in OA/TA neurons (Iliadi et al. 2017). A specific knockdown of TβH, and thereby OA, in OANs leads to impaired aversive learning (Wu et al. 2013; Sabandal et al. 2020). Appetitive learning depends on intact OA signaling to α/β and γ KCs, and OA signaling to the MB also mediates aversive stimuli (Zhou et al. 2012; Kim et al. 2013). Blocking the transmitter release from OANs in the training phase leads to impaired courtship learning, whereas thermogenetic activation of these cells introduced artificial courtship behavior (Zhou et al. 2012). Therefore, octopamine signaling in the MB is clearly involved in learning and memory. The following sections will discuss the anatomical basis for the functional role of OA in the MB circuitry, focusing on the morphology of the OANs and localization of the receptors.

Morphological description of OANs innervating the mushroom body

The morphology of the OANs has been studied in great detail in a diverse group of insects, with a focus on dorsal unpaired median (DUM) and ventral unpaired median (VUM) neurons, a group of symmetrically projecting cells located at the dorsal or ventral midline in the subesophageal, thoracic, and abdominal ganglia (Stevenson and Spörhase-Eichmann 1995; Monastirioti 1999; Stern 1999). The broad innervation of central brain neuropils arises from an overall small number of VUM/DUM neurons in the subesophageal ganglion and a few cells located in the brain itself. Interestingly, all insects analyzed for OA immunoreactivity showed staining in the MBs. OA labeling was observed in the MBs of fruit flies and blowflies (Monastirioti et al. 1995; Sinakevitch and Strausfeld 2006), honeybees (Kreissl et al. 1994; Sinakevitch et al. 2005), locusts (Konings et al. 1988; Bräunig 1991; Stevenson et al. 1992; Kononenko et al. 2009), crickets (Spörhase-Eichmann et al. 1992), cockroaches (Sinakevitch et al. 2005), and hawkmoths (Dacks et al. 2005). More specifically, all of these insects possess OA-positive neurites in the calyces. The peduncles (locusts, moths), vertical lobes (bees, cockroaches, blowflies), and medial lobes (Drosophila, blowflies, bees, moths) contain restrictive octopaminergic fibers (Konings et al. 1988; Kreissl et al. 1994; Dacks et al. 2005; Sinakevitch et al. 2005; Sinakevitch and Strausfeld 2006). Single OANs with a similar innervation pattern have been described in locusts, bees, and fruit flies, allowing several groups to speculate that there is a conserved function of OA in insects (Bräunig 1991; Hammer 1993; Bräunig and Burrows 2004; Schröter et al. 2007; Busch et al. 2009; Wu et al. 2013; Selcho et al. 2014). In the following, I will concentrate on the description of the single cells identified in Drosophila melanogaster. Here, five different octopaminergic cell types have been reported to innervate the MB: VUMa2, VPM3, VPM4, VPM5, and the anterior paired lateral (APL) neuron (Fig. 1A–F; Table 1; Busch et al. 2009; Wu et al. 2013). The cell bodies of the VM cluster neurons are distributed along the midline in the subesophageal ganglion and have been divided into three different subclusters depending on the location of the cell bodies in the mandibular (md), maxillary (mx), or labial (lb) neuromere. The VUM cells of the md and mx clusters are duplicated (Busch and Tanimoto 2010). One VUM cell type (VUMa2) projecting into the MB located in the md and mx neuromere has been identified. In addition, three different paired VM neurons (VPM3, VPM4, and VPM5) innervate the MB (Fig. 1C–F; Busch et al. 2009). The two VUMa2 neurons ramify within the calyces (Fig. 1C,F; Busch et al. 2009), while the VPM5 cell type mimics the arborization of the VUMa2 but projects only unilaterally. The two VPM3 cells innervate the γ-lobe, α′β′-lobes, and the calyx (Fig. 1D,F; Busch et al. 2009; Aso et al. 2014). The γ-lobe is also innervated by the two VPM4 neurons (Fig. 1E,F; Busch et al. 2009; Aso et al. 2014). Beside the ascending cells of the VM cluster, another paired OAN with its cell body in the anterior lateral brain arborizes extensively in the MB. The APL neuron projects to the calyx, accessory calyx, peduncle, and all lobes with few arborizations outside the MB (Fig. 1B,F; Tanaka et al. 2008; Liu and Davis 2009).

Figure 1.

Octopaminergic neurons innervate the mushroom body. (A–E) Four octopaminergic cell types that arborize in the MB identified on an electron microscopic level using FlyWire neuroglancer. The VPM5 cell type has not yet been identified in the FlyWire data set. (B) The APL neuron innervates all parts of the MB. (C) Two VUMa2 neurons project from the antennal lobe via the iACT to both calyces and the lateral horn. (D) The paired VPM3 cells innervate the contralateral calyx and lobes, the fan-shaped body, parts of the dorsolateral protocerebrum, and the SEZ. (E) The VPM4 cell type arborizes in the contralateral γ-lobe and innervates the dorsomedial and -lateral protocerebrum as well as the SEZ. (F) Overview of the MB (gray) and the appropriate OA release sites (colored) of the four different octopaminergic cell types shown in B–E and the VPM5 cell type. Based on the data from Busch et al. (2009), Aso et al. (2014), and the analysis of the whole brain EM volume using FlyWire neuroglancer (Zheng et al. 2018; Dorkenwald et al. 2022, 2023). (APL) Anterior paired lateral, (Ca) calyx, (iACT) inner antennocerebral tract, (SEZ) subesophageal zone, (VPM) ventral paired median, (VUM) ventral unpaired median.

Table 1.

Anatomical information on MB-innervating OANs

OA cell typea	MB compartment innervatedb	Vesiclesc	Co-transmitterd
OA-VUMa2	ca (bs)	DCV, SV	TA, Glut
OA-VPM3	ca, γ1,3–5, α′3, β′2 (co)	DCV, SV	TA, Glut, (NO)
OA-VPM4	γ1–5 (co)	DCV, SV	TA, Glut, (NO)
OA-VPM5	ca (ip)	?	?
APL	ca, ped, α1–3, α′1–3, β1–2, β′1–2, γ1–5 (ip)	DCV, SV	TA, GABA

(α,β,α′,β′,γ) Appropriate MB lobe, (ca) calyx, (ped) peduncle, (bs) both sites, (co) contralateral, (ip) ipsilateral, (DCV) dense-core vesicle, (SV) synaptic vesicle, (TA) tyramine, (Glut) glutamate, (NO) nitric oxide, (?) not identified.

^aData from Busch et al. (2009) and Wu et al. (2013).

^bData from Busch et al. (2009), Chen et al. (2012), and Aso et al. (2014).

^cObserved in FlyWire neuroglancer (Fig. 2).

^dData from Liu and Davis (2009), Aso et al. (2019), and Sherer et al. (2020).

The arborization pattern of the VM cells has been identified using the Tdc2-GAL4 line (Cole et al. 2005). Tyrosine decarboxylase 2 (Tdc2) is the enzyme necessary for TA production and therefore all GAL4-positive cells should contain TA and potentially also OA. As all Tdc2-GAL4-positive VM cells were stained by an OA antibody and an antibody against TβH, the enzyme for OA synthesis, these cells are indeed octopaminergic (Busch et al. 2009; Schneider et al. 2012). The APL neuron, which is not included in the Tdc2-GAL4 line, is also OA- and TβH-positive (Wu et al. 2013). So far, 10 out of approximately 100 octopaminergic cells in the Drosophila brain have been shown to innervate the MB.

One other paired neuron was also marked by OA antibody labeling. The DAL (dorsal–anterior–lateral) neuron contacts α/β KCs in the calyx and was shown to be involved in long-term memory (LTM) formation but independent of OA production within the cell (Chen et al. 2012). A detailed anatomical description is not available and the DAL neuron has not yet been identified at the EM level. Another OA cell cluster located in the anterior superior medial protocerebrum (ASM) might contain additional MB innervating cells (FlyCircuit) (Chiang et al. 2011; Zhou et al. 2012). One neuron innervating the αβ-lobes was identified, but in contrast to the VM cluster, not all Tdc2-GAL4-positive cells of the ASM cluster are OA-immunoreactive (Busch et al. 2009). Therefore, this MB-innervating cell could be tyraminergic. Further investigations need to show if other so far uncharacterized OANs exist that contribute to OA release into the MB circuitry.

Octopaminergic neurons contain a diverse set of transmitters

OA is synthetized from TA; therefore, all OANs contain TA, which can act as a neuromodulator by itself (Nagaya et al. 2002; Schützler et al. 2019). In addition, OANs are reported to express at least one more transmitter (many glutamate, some GABA, or acetylcholine) and in many cases additional peptides (Schneider et al. 2012; Wu et al. 2013; Croset et al. 2018; Sherer et al. 2020). Certel and colleagues (Sherer et al. 2020) have shown that nearly all OANs of the VM cluster show colocalization of the Drosophila vesicular glutamate transporter (dVGlut) and therefore contain glutamate as a classical co-transmitter. In contrast, the APL neuron has been demonstrated to contain GABA as a co-transmitter (Liu and Davis 2009). Even though co-transmission is a common principle of many neurons across species, the anatomical principles and behavioral consequences of these combinations of transmitters are poorly understood (Hnasko and Edwards 2012; Vaaga et al. 2014; Nusbaum et al. 2017; Nässel 2018). The Drosophila vesicular monoamine transporter mediates the transport of biogenic amines into a large dense-core vesicle (DCV) and small synaptic vesicle (SV) (Greer et al. 2005; Deshpande et al. 2020). Therefore, OA can be released extrasynaptically by fusion of DCVs to the membrane as shown for type II motor neuron endings and potentially also from SVs at presynaptic active zones like classical transmitters, which are only packed into SVs (Grygoruk et al. 2014; Stocker et al. 2018).

Indeed, using FlyWire neuroglancer (Dorkenwald et al. 2022) to analyze the morphology of the OANs in the whole brain EM volume (Zheng et al. 2018), t-bars (the structural correlate of SV release sites), SVs, and DCVs can be identified in MB-innervating OA cell types (Fig. 2A–D). DCVs are visible in the arborizations of OANs in the MB calyces and lobes, whereas all cell types contain additional classical synapses. This suggests that OANs exert their function in the MB through locally restricted synaptic connections to specific postsynaptic partners and through volume transmission targeting a broader set of neurons within the MB. The identification of SVs opens in addition the possibility that VM-OANs corelease glutamate in the MBs. In contrast, the APL neuron was discovered as a GABAergic neuron and corelease of OA was identified a few years later (Liu and Davis 2009; Wu et al. 2013). The observation that the APL cell contains DCVs supports the octopaminergic identity of these neurons (Fig. 2A; Takemura et al. 2017). Further investigations are needed to show if the transmitters are released together at the same synapse or if release can occur at distinct sites. Indeed, the first experiments analyzing the distribution of the vesicular monoamine transporter and dVGlut in VPM4 neurons showed potential colocalization of both transporters and, in addition, areas of distinct localization (Sherer et al. 2020). Interestingly, co-transmission has recently been shown in the MB for a subset of DA neurons (Aso et al. 2019). Here, the co-transmitter nitric oxide (NO) seems to activate the soluble guanylate cyclase in KCs to form an inverse memory, acting antagonistically to DA. The effect of NO is slower than the DA effect and seems to shorten memory retention to enable rapid updating of memories (Aso et al. 2019). In addition, Aso and colleagues proposed that OA-VPM3 or -VPM4 neurons also contain NO as they seem to express NO synthase. Whether this observation holds true, and whether a similar function can be achieved by the release of this co-transmitter from OANs, remains to be tested in further experiments.

Figure 2.

Ultrastructure of octopaminergic neurons in the mushroom body. (A–D) Example images of the ultrastructure of the MB-innervating OA cell types show large dense-core vesicles (DCVs, arrowheads), synaptic vesicles (SVs, open arrowheads), and potential SV release sites (t-bars, open arrowheads). FlyWire neuroglancer was used to investigate the morphology of OANs in the whole brain EM volume (Zheng et al. 2018; Dorkenwald et al. 2022). (A) The APL neuron contains mostly SVs and only a few DCVs in all areas. Example boutons from the calyx (A1), α- (A2), β- (A3), α′- (A4), and γ-lobes (A5). (B) DCVs and SV release sites in a VUMa2 cell bouton in the calyx. (C) Ultrastructure of the VPM3 neuron in the γ- (C1) and α′-lobe (C2) and the calyx (C3). (D) The VPM4 cell innervates the complete γ-lobe. DCVs, as well as SVs and their release sites, are visible in the γ1 (D1), γ2 (D2), γ3 (D3), γ4 (D4), and γ5 (D5) compartments. Scale bars, 300 nm.

OA receptors expressed in the mushroom bodies

OA receptors have been studied intensively in order to decipher the function of OA in the MB circuitry. OA receptors are G-protein-coupled and can be classified as α-adrenergic-like OA receptors (Octα1-R and Octα2-R) and β-adrenergic-like OA receptors (Octβ1-R, Octβ2-R, Octβ3-R) according to their structural and functional similarities to vertebrate α- and β-adrenergic receptor subtypes (Balfanz et al. 2005; Evans and Maqueira 2005; Maqueira et al. 2005; Wu et al. 2014; Qi et al. 2017). Octα-R activation leads to an increase in intracellular calcium levels and influences intracellular cAMP concentration (up-regulation in Octα1-R and down-regulation in Octα2-R (in vitro studies: Han et al. 1998; Balfanz et al. 2005; Beggs et al. 2011; Wu et al. 2014; Qi et al. 2017; Nakagawa et al. 2022; in vivo: Lee et al. 2009). In contrast, Octβ-R activation elicits an increase in intracellular cAMP but not in calcium levels (Balfanz et al. 2005; Maqueira et al. 2005; ex vivo: Sabandal et al. 2020; in vivo: Lim et al. 2014). In addition to the OA-affiliated receptors, three TA receptor types have been identified (Tyr1-R, Tyr2-R, and Tyr3-R). Tyr1-R and Tyr3-R respond to TA and OA, whereas Tyr2-Rs are specifically activated by TA. Activation of these receptors leads to either an increase or decrease of intracellular cAMP levels, or to changes in intracellular calcium levels (for review, see Finetti et al. 2021).

The first OA receptor identified in the MBs was the Drosophila Octα1-R named OAMB (Han et al. 1998). Using antibody staining, the MBs were highlighted because of the high amount of receptor expression in this densely packed neuropil. El-Kholy et al. (2015) used promoter GAL4 lines to visualize OA and TA receptor expression. All known Drosophila OA and TA receptors seem to be expressed in the MBs. Single-cell type RNA-seq data of α/β and γ KCs revealed expression of Octα1-R (OAMB), Octβ1-R (OA2), Octβ2-R, and TyrR in α/β KCs and OAMB and OA2 in γ KCs (Crocker et al. 2016). RNA-seq data from all KC types revealed expression of OAMB and Octα2-R in α/β and γ KCs and less strong in α′/β′ KCs. Octβ1-R, Octβ2-R, and Octβ3-R are detected in α/β, α′/β′, and γ KCs, respectively, with Octβ3-R exclusively expressed in γd subtypes (Shih et al. 2019). Octopamine receptors (OARs) appear to be broadly expressed within the MB (Table 2), but such a large-scale distribution seems rather unrealistic considering the innervation pattern of the OANs. Therefore, it is necessary to investigate the exact localization of the receptors within a KC.

Table 2.

Octopamine receptors in the MB

OA receptor	Expression in KCsa	IR in MB regionsb	Functional role for learning and memory (OAR localization)
Octα1-R (OAMB)	α/β, α′/β′, γ KCs	α, α′, β, β′, γ, ped, ca	Appetitive olfactory learning (α/β, γ KCs)c; sweet taste reinforcement (PAM DAN)d; courtship learning (αβ lobes)e
Octα2-R	α/β, α′/β′, γ KCs	?	?
Octβ1-R	α/β, α′/β′, γ KCs	?	Aversive olfactory learning (α/β KCs); appetitive olfactory learning (PN)f
Octβ2-R	α/β, α′/β′, γ KCs	α′, β′	sweet taste reinforcement (PPL1 DAN)d; ARM (α′/β′ KCs)g
Octβ3-R	α/β, α′/β′, γ KCs	?	?

(α,β,α′,β′,γ) Appropriate MB lobe, (ARM) anesthesia-resistant memory, (ca) calyx, (DAN) dopaminergic neuron, (IR) immunoreactivity, (KCs) Kenyon cells, (OAR) octopamine receptor, (ped) peduncle, (?) unknown.

^aData from Crocker et al. (2016) and Shih et al. (2019).

^bData from Han et al. (1998), Crittenden et al. (1998), Lee et al. (2003), Wu et al. (2013), and Kim et al. (2013).

^cData from Kim et al. (2013).

^dData from Burke et al. (2012) and Huetteroth et al. (2015).

^eData from Zhou et al. (2012).

^fData from Sabandal et al. (2020).

^gData from Wu et al. (2013).

Behavioral experiments in Drosophila revealed that OA signaling via OAMB in the α/β and γ KCs is sufficient to rescue learning (Kim et al. 2013). Additionally, courtship learning depends on OAMB expression in the MB (Zhou et al. 2012). The sweet taste information in sugar reward learning is mediated by OAMB expressing dopaminergic PAM neurons activating β′2 and γ4 compartments of the MB (Burke et al. 2012; Huetteroth et al. 2015). OA signaling via Octβ1-R in α/β KCs is necessary for aversive olfactory learning, while signaling via Octβ1-R in projection neurons is involved in appetitive olfactory learning (Sabandal et al. 2020). In addition, Octβ2-R in the dopaminergic MB-MP1 PPL1 neurons is needed to modulate negative DA signals for sweet taste learning (Burke et al. 2012). Octβ2-R RNAi in α′/β′ KCs impaired anesthesia-resistant memory (ARM) (Wu et al. 2013). Analyzing receptor function in specific sets of KCs, MB output neurons (MBONs), or MB input neurons (MBINs) is an effective way to demonstrate the role of OA in MB function (Table 2).

Function of OANs signaling to the mushroom body circuitry

Currently, there is limited information available regarding the function of the individual MB-innervating OANs. Only two cell types have been linked to behavioral function: The VPM4 cell was found to play a role in feeding-related behaviors, whereas APL neurons are involved in learning and memory (Liu and Davis 2009; Pitman et al. 2011; Ren et al. 2012; Wu et al. 2012, 2013; Youn et al. 2018; Sayin et al. 2019; Zhou et al. 2019). The VPM4 cells promote proboscis extension to sugar, the first step in feeding behavior. The modulation of the proboscis extension reflex depends on functional OA synthesis within the VPM4 cells and on OAMB activity in sugar-sensing gustatory receptor neurons, increasing sensory responses to sugar (Youn et al. 2018). Additionally, VPM4 neurons reduce hunger-driven odor tracking by signaling to MBON-γ1pedc > αβ (MBON11) in the γ1 compartment (Sayin et al. 2019). Together with the findings that VPM4 neurons reinforce proboscis extension, it seems that these OANs promote the behavioral switch from food search to feeding (Youn et al. 2018; Sayin et al. 2019). Manjila et al. (2019) demonstrated that VPM4 neurons modulate flight duration through OAMB activation in PAM dopaminergic neurons. Octopaminergic input to PAM neurons innervating the MB β′1 compartment reduces the inhibitory output of MBON-β′1 leading to prolonged flight bouts. This OA–PAM–MBON circuit might influence flight in response to food availability, feeding state, or other sensory input (Manjila et al. 2019).

The APL neurons play a role in learning and memory and enable sparse odor coding by KCs through feedback inhibition in a compartment-specific manner (Liu and Davis 2009; Pitman et al. 2011; Ren et al. 2012; Wu et al. 2012, 2013; Lei et al. 2013; Lin et al. 2014; Zhou et al. 2019; Amin et al. 2020; Prisco et al. 2021). APL neurons suppress aversive olfactory learning, and their activity is reduced by olfactory learning via dopaminergic PPL1 neurons in the vertical lobes (Liu and Davis 2009; Zhou et al. 2019). Furthermore, both forms of intermediate-term memory, ARM and anesthesia-sensitive memory, are modulated by the APL neurons (Wu et al. 2011, 2013). However, only ARM is regulated by OA. Memory formation is modulated by APL cells via Octβ2-R signaling in α′/β′ KCs (Wu et al. 2013). APL neurons corelease OA and GABA. The decrease in OA levels does not affect aversive learning, but impairs ARM formation. Conversely, the decrease in GABA levels within the cells affects learning, but not intermediate-term memory formation (Liu and Davis 2009; Wu et al. 2013). The APL neuron exhibits a separation between co-transmitter functions, possibly due to the different localization of the appropriate receptor.

Outlook

MB function is modulated by OA. Ten OANs that innervate the MB have been described, and all cells have multiple transmitters. To holistically decipher the role of OANs in the MB circuitry, it is essential to gain further insights into the set of co-transmitters released from these neurons. The first step to a functional connectome has been taken by the precise morphological analysis within the openly accessible connectome of the complete fly brain (FlyWire.ai). However, the synaptic connections within the MB are insufficient as they do not cover DCV release. Therefore, to obtain a comprehensive morphological overview of all release sites, it is necessary to identify a marker for DCV “synapses.” Furthermore, fundamental questions remain to be answered: Is OA packed in SVs and released from synapses? Are co-transmitters spatially separated? Is the release of co-transmitters regulated? In addition, it is essential to study the localization of the OA receptors in more detail in order to identify postsynaptic partners. Knockdown of receptor expression in specific sets of KCs, MBONs, or MBINs is an effective way to demonstrate the role of OA in MB function.