Serotonin Signals Modulate Mushroom Body Output Neurons for Sustaining Water-Reward Long-Term Memory in Drosophila

Memory consolidation is a time-dependent process through which an unstable learned experience is transformed into a stable long-term memory; however, the circuit and molecular mechanisms underlying this process are poorly understood. The Drosophila mushroom body (MB) is a huge brain neuropil that plays a crucial role in olfactory memory. The MB neurons can be generally classified into three subsets: γ, αβ, and α′β′. Here, we report that water-reward long-term memory (wLTM) consolidation requires activity from α′β′-related mushroom body output neurons (MBONs) in a specific time window. wLTM consolidation requires neurotransmission in MBON-γ3β′1 during the 0–2 h period after training, and neurotransmission in MBON-α′2 is required during the 2–4 h period after training. Moreover, neurotransmission in MBON-α′1α′3 is required during the 0–4 h period after training. Intriguingly, blocking neurotransmission during consolidation or inhibiting serotonin biosynthesis in serotoninergic dorsal paired medial (DPM) neurons also disrupted the wLTM, suggesting that wLTM consolidation requires serotonin signals from DPM neurons. The GFP Reconstitution Across Synaptic Partners (GRASP) data showed the connectivity between DPM neurons and MBON-γ3β′1, MBON-α′2, and MBON-α′1α′3, and RNAi-mediated silencing of serotonin receptors in MBON-γ3β′1, MBON-α′2, or MBON-α′1α′3 disrupted wLTM. Taken together, our results suggest that serotonin released from DPM neurons modulates neuronal activity in MBON-γ3β′1, MBON-α′2, and MBON-α′1α′3 at specific time windows, which is critical for the consolidation of wLTM in Drosophila.


INTRODUCTION
One of the greatest mysteries in the biological field is how unstable labile memory is consolidated into a stable, long-lasting memory. Memory consolidation in mammals initially requires the hippocampus and cortex, but after the memory is consolidated, the hippocampus is no longer required. Hippocampal damage impairs new memory consolidation while leaving the old memories intact, suggesting that unstable labile memory is formed in the hippocampus and transformed into stable longlasting memory in the cortex (Scoville and Milner, 1957). Systems memory consolidation model suggests that interaction between hippocampal and cortex is critical during and after an experience (Frankland and Bontempi, 2005). Recent study shows that hippocampal and cortical engram cells participate in both early and late stages of memory, and the dominant location of engram moves from the hippocampus to the cortex during systems consolidation (Kitamura et al., 2017).
Long-term memory (LTM), especially appetitive LTM, allows hungry or thirsty animals to behave appropriately when facing lifesupporting experienced events again in the future. LTM is thought to transform from short-term and/or intermediate-term memory and requires a "memory consolidation" process to stabilize short-lasting and labile memory into a long-lasting and stable memory (Margulies et al., 2005). A thirsty fruit fly, Drosophila melanogaster, can be trained to associate an odor with water reward to form water-reward memories. The motivational state of fruit flies, such as the thirst for water, induces a more efficient, strong, and robust long-lasting memory compared to the shock-punitive aversive memory (Tully et al., 1994;Shyu et al., 2017). One-session training for pairing of an odor with water induces not only water-reward short-term memory in a few minutes to hours, but also a water-reward LTM (wLTM) that can persist for approximately 2 days (Lin et al., 2014;Shyu et al., 2017;Lee et al., 2020). It has been shown that thirsty-dependent water memory is mediated by neuropeptide leucokinin, which is released from a pair of neurons in the fly brain. Leucokinin controls water memory expression by inhibiting two classes of dopaminergic neurons in the fly brain to permit thirsty-dependent water memory expression (Senapati et al., 2019). wLTM formation in Drosophila requires the expression of several genes associated with learning and memory within the brain, as observed in mammals (Scharf et al., 2002;Cepeda et al., 2006;Shyu et al., 2017;Lee et al., 2020).
The mushroom body (MB) is traditionally viewed as the olfactory memory center. The MB is composed of three conventional lobes including the αβ, α′β′, and γ lobes, and contains approximately 2,200 Kenyon cells (Lin et al., 2007). The MB can be further compartmentalized into 15 functional compartments based on the innervated fiber of mushroom body output neurons (MBONs) and dopaminergic mushroom body input neurons (DANs). Each compartment of the MB is modulated by its DANs and the information is read out by its MBONs (Aso et al., 2014a). Importantly, the postsynaptic region of the MBON has high plasticity induced by olfactory associative training, suggesting that MBONs serve as a part of the memory processor (Hige et al., 2015). Different MBONs encode distinct intrinsic valences, which can drive the approach, avoidance, or neutral behavior of Drosophila. The summarization of valences from MBONs might be the basic playing keys for performing and recording the behavioral output of memory (Aso et al., 2014b). In our previous studies, we identified that a fruit fly simultaneously associates the water-rewarding event from DAN PAM-β′1 and an odor from projection neurons, both of which converge on MB neurons. After memory acquisition, neurotransmission from MB α′β′ neurons is required for wLTM consolidation. Finally, the fly requires neurotransmission from specific subsets of αβ neurons, γ neurons, and their MBONs to recall the wLTM (Shyu et al., 2017;Lee et al., 2020). It remains unclear how wLTM is transferred from α′β′ neurons to αβ neurons and γ neuron subsets. Moreover, the neuronal circuits and molecules involved in this memory process remain unexplored.

RESULTS
Neurotransmitter Output from α9β9-related MBONs during the First 4-h Period after Training Is Necessary for wLTM It has been shown that neurotransmitter output from α′β′ neurons is critical for wLTM consolidation but not acquisition or retrieval (Shyu et al., 2017), which led us to ask whether there are time-dependent requirements of neurotransmission in α′β′-related neurons during wLTM consolidation. To address this, we performed acute manipulation of neurotransmitter output by expressing temperature-sensitive shibire (shi ts ) allele in specific neurons that inhibits dynamin-mediated endocytosis for recycling of neurotransmitters at restrictive temperatures (Dubnau et al., 2001;Kitamoto, 2001;McGuire et al., 2001). Results showed that blocking neurotransmitter output from α′β′ neurons (via VT30604-GAL4) during the first 4-h period right after training disrupts wLTM ( Figures 1A,B). However, blocking neurotransmitter output in α′β′ neurons at-4-12 or 12-20 h after training did not affect wLTM ( Figure 1B).
Frontiers in Cell and Developmental Biology | www.frontiersin.org November 2021 | Volume 9 | Article 755574 assays. All flies were raised and trained at 23°C, maintained at 32°C for 2 h immediately after the training, and shifted back to 23°C for another 22 h and tested. We observed that blocking neurotransmitter output in MBON-γ3β′1 and MBON-α′1α′3 during the 0-2 h period after training disrupted wLTM ( Figure 3A). We further investigated the 2-4 h period posttraining. The flies were trained at 23°C, maintained at 23°C for 2 h immediately after the training, shifted back to 32°C for 2 h, and again shifted back to 23°C for another 20 h and tested. We observed that blocking neurotransmitter output in MBON-α′2 and MBON-α′1α′3 during the 2-4 h period after training disrupted wLTM ( Figure 3B).

Serotonin Release From DPM Neurons is Necessary for wLTM Consolidation
MBON-γ3β′1, MBON-α′2, and MBON-α′1α′3 receive inputs not only from α′β′ neurons but may also receive inputs from MB modulatory neurons. It has been shown that MB modulatory neurons, the DPM neurons that project to and innervate the MB lobes, play a role in mediating the synaptic activity of MBONs. Previous studies have suggested that DPM neurons are critical for olfactory associative memories, including shock-punishment and sugar-reward conditioning (Waddell et al., 2000;Keene et al., 2004;Yu et al., 2005;Keene et al., 2006;Wu et al., 2011). It has also been reported that DPM neurons release the neurotransmitter 5HT (Lee et al., 2011), which led us to ask whether 5HT synthesis in DPM neurons affects wLTM. In the 5HT biosynthesis pathway, L-tryptophan is first hydroxylated by the enzyme tryptophan hydroxylase (Trh) to 5-hydroxytryptophan, which is then decarboxylated by the enzyme dopa decarboxylase (Ddc) to 5HT (Blenau and Thamm, 2011).
Our results showed that RNAi-mediated silencing of Trh and Ddc using the GAL4/UAS binary system inhibits 5HT biosynthesis in DPM neurons (via VT64246-GAL4) and disrupts wLTM ( Figures 4A,B). However, inhibiting 5HT biosynthesis in MB neurons did not affect wLTM (Supplementary Figure S1). We further analyzed the knockdown efficiency of UAS-Trh RNAi ; UAS-Ddc RNAi in VT64246-GAL4 line. We observed that VT64246-GAL4 > UAS-Ddc RNAi ;;UAS-Trh RNAi flies showed over 50% reduction of anti-5HT antibody immunoreactive signals in both soma and fibers of DPM neurons as compared to the control groups ( Figures 4C-E). To exclude the developmental effects of inhibiting 5HT biosynthesis in DPM neurons, we next tested the flies with inducible knockdown of Ddc and Trh genes using tub-GAL80 ts in the VT64246-GAL4 line. Our results showed that adult-stage-specific knockdown of Ddc and Trh in DPM neurons also impaired wLTM ( Figure 4F).

Neurotransmission in DPM Neurons Is Specifically Required for wLTM Consolidation but Not Acquisition and Retrieval
Since wLTM formation requires 5HT biosynthesis in DPM neurons, we asked that neurotransmission in DPM neurons is involved in which memory phases. We genetically expressed shi ts transgene in DPM neurons and all transgenic flies were maintained at 23°C and shifted to 32°C to inhibit neurotransmitter output during training (acquisition), trial interval (consolidation), or testing (retrieval). We found that blocking neurotransmission in DPM neurons during consolidation impaired wLTM ( Figure 5A). However, blocking neurotransmission in DPM neurons during acquisition or retrieval did not affect wLTM ( Figure 5A).
Frontiers in Cell and Developmental Biology | www.frontiersin.org November 2021 | Volume 9 | Article 755574 DPM neurons are involved. The flies were trained at 23°C and then maintained at 32°C for 0-2 h or 2-4 h right after the training, and shifted back to 23°C for the resting interval and tested. We found that blocking neurotransmitter output in DPM neurons during the 0-2 h or 2-4 h period disrupted wLTM ( Figure 5B). Taken these data together suggest that 5HT neurotransmitter output from DPM neurons is critical for wLTM consolidation (Figures 4, 5).
Considering the gap junctions between DPM and anterior paired lateral (APL) neurons are critical for aversive anesthesiasensitive memory in Drosophila (Wu et al., 2011), which led us to ask whether inhibiting neurotransmission during memory consolidation in APL neurons affects wLTM. Blocking neurotransmission in APL neurons during the 0-4 h period after training did not affect the wLTM, implying that neurotransmission in APL neurons is not critical for wLTM consolidation (Supplementary Figure S2). It is possible that APL neurons are involved in other memory phases rather than wLTM consolidation.
Frontiers in Cell and Developmental Biology | www.frontiersin.org November 2021 | Volume 9 | Article 755574 6 impairment was not caused by the deficiency of sensory inputs (Supplementary Figure S4).
Neurotransmission in α′β′ neurons is essential for wLTM only during the initial 4 h after training, but not at the later stage during memory consolidation, suggesting that wLTM consolidation is complete within the first 4-h after training. Furthermore, neurotransmission in the three α′β′-related MBONs is required at different periods during consolidation. wLTM formation requires neurotransmission from MBON-γ3β′1 during the 0-2 h period after training, whereas neurotransmission from MBON-α′2 is essential during the 2-4 h period after training. These results suggest the sequential role of MBON-γ3β′1 and MBON-α′2 during wLTM consolidation. In addition, wLTM formation requires neurotransmission from MBON-α′1α′3 during the 0-4 h period after training. Intriguingly, it was shown that MBON-γ3β′1 plays the sleep-promoting role by which activating of MBON-γ3β′1 via TrpA1 transgene prompts sleep (Aso et al.,FIGURE 5 | Neurotransmission in DPM neurons is specifically required for wLTM consolidation but not acquisition or retrieval. (A) Blocking neurotransmission in DPM neurons via shi ts during memory consolidation, but not during acquisition or retrieval, disrupted wLTM. Each value represents mean ± SEM (N 10, 9, 8, 10, 11, 11, 8, 8, 8, 8, 8, and 8 from left to right bars). *p < 0.05; one-way ANOVA followed by Tukey's test. (B) Blocking neurotransmission in DPM neurons via shi ts at 0-2 h period (left panel) or at 2-4 h period (right panel) after training disrupted wLTM. Each value represents mean ± SEM (N 10, 10, 10, 9, 9, and 8 from left to right bars). *p < 0.05; one-way ANOVA followed by Tukey's test.
Frontiers in Cell and Developmental Biology | www.frontiersin.org November 2021 | Volume 9 | Article 755574 2014b). Contrary to this result, a recently study showed that activating MBON-γ3β′1 via TrpA1 transgene cause the mean sleep amount decrease, whereas inhibiting MBON-γ3β′1 via shi ts increase the mean sleep amount (French et al., 2021). Our behavioral result in MBON-γ3β′1 is unlikely caused by sleep loss since blocking MBON-γ3β′1 via shi ts did not reduce sleep.
Frontiers in Cell and Developmental Biology | www.frontiersin.org November 2021 | Volume 9 | Article 755574 8 neurons is inhibited. However, in about 30 min after training, the MVP2 neurons are activated and terminate the period of MP1 neuronal activity suggesting that neuronal activity in specific time period is critical for consolidation process (Pavlowsky et al., 2018).
In fruit flies, the 5HT system participates in olfactory learning and memory (Johnson et al., 2011;Sitaraman et al., 2012;Scheunemann et al., 2018;Ganguly et al., 2020), and a previous study suggested that 5HT signals from DPM neurons modulate the activity of the MB αβ neurons, which play a crucial role in olfactory aversive anesthesia-resistant memory formation (Lee et al., 2011). The axons of DPM neurons innervate the MB lobe, in which the MBON dendrites are distributed. Our GRASP data showed the connectivity between DPM neurons and MBON-γ3β′1, MBON-α′1α′3, and MBON-α′2, suggesting that these three MBONs receive inputs from DPM neurons. DPM neurons release the neurotransmitter 5HT and adult-stagespecific inhibition of 5HT biosynthesis in DPM neurons disrupts wLTM. Blocking neurotransmission in DPM neurons during the 0-2 or 2-4 h period after training also disrupts wLTM, suggesting that neurotransmitters from DPM neurons regulate the activity of the downstream MBON-γ3β′1, MBON-α′1α′3, and MBON-α′2. In addition, blocking 5HT biosynthesis in MB neurons by genetic manipulation did not affect wLTM, further confirming that DPM neurons are the source of 5HT rather than MBs.
Hippocampal serotonin receptors participate in the consolidation and reconsolidation of fear conditioning memory in mammals (Schmidt et al., 2017). Fear memory consolidation requires 5HT 5A , 5HT 6 , and 5HT 7 receptors in the CA1 region of the hippocampus (Schmidt et al., 2017). In Drosophila, there are five different types of 5HT receptors, and our behavioral screening showed that blocking specific 5HT receptors in these three MBONs disrupts wLTM formation. Through genetic manipulation, blocking of 5HT1A, 5HT2A, and 5HT7 in MBON-γ3β′1 or 5HT1A, 5HT1B, 5HT2A, and 5HT2B in MBON-α′1α′3 or 5HT1B and 5HT2A in MBON-α′2 disrupts wLTM. All these genetically manipulated flies showed normal odor acuity and water preference in the thirsty state, which suggests that RNAi-mediated manipulation of 5HT receptor expression in MBONs did not affect sensory inputs. Our results suggest that 5HT release from DPM neurons modulate the activity of MBONs, and this modulation is critical for wLTM consolidation.
In this study, we propose that 5HT release from single paired DPM neurons mediates the neuronal activity of MBON-γ3β′1, MBON-α′2, and MBON-α′1α′3 via different 5HT receptors, and this modulation is critical for sustaining the consolidation of wLTM. Investigating the neuronal circuits that receive inputs from MBON-γ3β′1, MBON-α′2, and MBON-α′1α′3 and identifying the associated GABA and acetylcholine receptors would be an interesting topic for future research.

Whole-Mount DLG Immunostaining
The fly brains were dissected in phosphate-buffered saline (PBS) solution and immediately transferred to 4% paraformaldehyde (PFA) for fixation for 20 min. Fixed brain samples were incubated in penetration and blocking buffer (PBS containing 2% Triton X-100 and 10% normal goat serum) for 2 h. During the penetration and blocking period, the brain samples were subjected to a degassing procedure. Brains were then incubated in the dilution buffer (PBS Frontiers in Cell and Developmental Biology | www.frontiersin.org November 2021 | Volume 9 | Article 755574 9 containing 0.25% Triton X-100, 1% normal goat serum) containing mouse 4F3 anti-discs large (DLG) monoclonal antibody (1:10, AB 528203, Developmental Studies Hybridoma Bank, University of Iowa) at 25°C for 24 h. After incubation, the tissues were washed in PBS containing 1% Triton X-100 (PBS-T) three times (each time 15 min), and incubated with biotinylated goat antimouse IgG (1:200, 31,800, Thermo Fisher Scientific) at 25°C for 24 h. Next, the brain samples were washed and incubated with Alexa Fluor 635 streptavidin (1:500, S32364, Thermo Fisher Scientific) at 25°C for 24 h. After intensive washing in PBS-T for three times, the brain samples were cleared and mounted in FocusClear (FC-101, CelExplorer).

Quantification of 5HT Immunostaining
Fly brains were immunostained with rabbit anti-5HT polyclonal antibody (1:200, S5545, Sigma-Aldrich) at 25°C for 24 h. After three washes in PBS-T, the samples were incubated in biotinylated goat anti-rabbit IgG (1:200, 31,822, Thermo Fisher Scientific) at 25°C for 1 day. The brain samples were then washed and incubated in Alexa Fluor 635 streptavidin (1: 500, S32364, Thermo Fisher Scientific) at 25°C for 1 day. After three-time extensive washing, the brain samples were cleared and mounted in FocusClear (FC-101, CelExplorer). Brain images were obtained using a Zeiss LSM 700 confocal microscope under the same confocal settings for each brain sample, and the images were further analyzed using ImageJ. Single optical sections were used to calculate the average intensity values per voxel of the 5HT immunopositive signals in the soma and fibers of DPM neurons, and the signals were normalized to the GFP intensity.

Confocal Microscopy
The fly brain samples were imaged under the Zeiss LSM 700 confocal microscope with a ×63 Plan-Apochromat oil-immersion objective lens (N.A. value, 1.4; working distance, 170 μm) or a ×40 C-Apochromat water-immersion objective lens (N.A. value, 1.2; working distance, 220 μm). The confocal pinhole (optical section) was set at 1.5 μm when using ×63 objective lens or 2 μm when imaging thorough ×40 objective lens. The brain image stacks were processed using the ZEN software.

Behavioral Assay
Flies were deprived of water by keeping them in a glass milk bottle containing a 6 cm × 3 cm piece of dry sucrose-soaked filter paper (the filter paper was soaked in saturated sucrose solution and allowed to dry before use) at 23°C and 20-30% humidity for 16 h before water conditioning. Approximately 50 water-deprived flies were transferred to the plastic training tube of the T-maze (TM-101, CelExplorer) and provided with a stream of relatively odorless room air for 1 min. The flies were exposed to the first odor for 2 min [unconditioned stimulus, CS-: 3-octanol (OCT) or 4methylcyclohexanol (MCH)] in a plastic tube lined with dry filter paper, followed by 1 min of room air. The flies were then transferred to another plastic tube containing a water-soaked filter paper and exposed to a second odor for 2 min (conditioned stimulus, CS+: MCH or OCT). Finally, the flies were transferred to a clean training tube and exposed to fresh room air for 1 min. The trained flies were kept in a plastic vial containing a 1.5 cm × 3 cm piece of dried sucrose-soaked filter paper during the 24 h interval between training and testing. In the testing phase, flies were presented with a choice between CS+ and CS-odors in a T-maze for 2 min. The performance index (PI) was calculated as the number of flies running toward the CS + odor minus the number of flies running toward the CS-odor, divided by the total number of trained flies and multiplied by 100. For calculating individual PI, naive flies were first trained by pairing water with OCT (CS+), and the index (PI O ) was calculated. Next, another group of naive flies was trained by pairing water with MCH (CS+), and the index (PI M ) was calculated. The individual PI was calculated as the average of the single PI O and PI M values.

Odor Avoidance and Water Preference Assays
Groups of approximately 50 naive flies were given a choice between either MCH or OCT versus 'fresh' room air for 2 min in the T-maze for the odor avoidance assay. The odor avoidance index was calculated as the number of flies in the fresh room air tube minus the number of flies in the MCH or OCT tube, divided by the total number of flies, and multiplied by 100. Groups of approximately 50 naive water-deprived flies were given 2 min to choose between tubes containing a water-soaked filter paper or a dry filter paper in T-maze in the water preference assay. The water preference index was calculated as the number of flies in the tube containing the water-soaked filter paper minus the number of flies in the tube containing the dry filter paper, divided by the total number of flies, and multiplied by 100.

Quantitative PCR
The efficiency of gene silencing in each 5HT receptor RNAi line from collections was verified with qPCR. Flies for qPCR experiments were generated by crossing virgin elav-GAL4 flies to either wild-type males or the various male UAS-5HT receptor RNAi flies. The RNA from isolated heads of adult flies was extracted by the TRIzol Reagent (T9424, Sigma-Aldrich). The extracted RNA was used to synthesize first strand of cDNA with SuperScript TM III First-Strand Synthesis SuperMix Kit (18080400, Thermo Fisher Scientific). The expression levels of mRNA were quantified with SYBR Green PCR Master Mix on a StepOnePlus System. Frontiers in Cell and Developmental Biology | www.frontiersin.org November 2021 | Volume 9 | Article 755574