Dynamics of two distinct memory interactions during water seeking in Drosophila

Abstract

Forming and forgetting memories shape our self-awareness and help us face future challenges. Therefore, understanding how memories are formed and how different memories interact in the brain is important. Previous studies have shown that thirsty flies sense humidity through ionotropic receptors, which help them locate water sources. Here, we showed that thirsty flies can be trained to associate specific odors with humidity to form a humidity memory that lasts for 30 min after association. Humidity memory formation requires the Ir93a and Ir40a ionotropic receptors, which are essential for environmental humidity sensing. Water memory takes precedence, leading to the forgetting of humidity memory by activating a small subset of dopaminergic neurons called protocerebral anterior medial (PAM)-γ4, that project to the restricted region of the mushroom body (MB) γ lobes. Adult-stage-specific silencing of Dop2R dopaminergic receptors in MB γ neurons prolongs humidity memory for 3 h. Live-brain calcium imaging and dopamine sensor studies revealed significantly increased PAM-γ4 neural activity after odor/humidity association, suggesting its role in forgetting the humidity memory. Our results suggest that overlapping neural circuits are responsible for the acquisition of water memory and forgetting humidity memory in thirsty flies.

Keywords: Drosophila; humidity memory; memory interactions; neural circuits; water seeking.

Fig. 1.

Thirsty flies form humidity memory. (A) Illustration depicting experimental design for humidity memory condition in Drosophila. (B) Formation of humidity memory depends on thirst state. Each value represents mean ± SEM (N = 18 and 20 from left to right bars). (C) rut or dnc (rut²⁰⁸⁰ and dnc¹) mutant flies show significant humidity memory defects compared to wild-type flies (+/+). Each value represents mean ± SEM (N = 14, 12, and 16 from left to right bars). P-values: 0.0007 and 0.0004; unpaired two-tailed t test. (D) Memory curve after humidity memory conditioning in wild-type flies. After a single training session, different fly populations were tested once for humidity memory at 3 min, 7 min, 15 min, 22 min, 30 min, 1 h, 3 h, 6 h, 12 h, and 24 h after training. Each value represents mean ± SEM (N = 9, 8, 9, 13, 10, 9, 11, 16, and 27 from left to right points). (D′) Enlarged display of 3 to 30 min memory performances in graph (D). P-values: <0.0001, <0.0001, <0.0001, 0.0475, and 0.4481; *significantly different from zero; unpaired two-tailed t test. (E) Water drinking after training but before testing significantly reduced humidity memory performance in water-deprived flies. Each value represents mean ± SEM (N = 13 for each bar). P = 0.0024; unpaired two-tailed t test. (F) Ir93a and Ir40a mutant flies fail to form humidity memory but ppk28 mutant flies show normal humidity memory compared to wild-type flies (+/+). Each value represents mean ± SEM (N = 10, 11, 10, 11, 8, and 7 from left to right bars). P-values: <0.0001, 0.6241, 0.0636, 0.0004, and 0.9286; unpaired two-tailed t test. (G) Flies show the same odor avoidance for OCT and MCH in both high (98%) and low (30%) humidity environments. Each value represents mean ± SEM (N = 8, 8, 8, and 8 from left to right bars). P-values: 0.9322 and 0.9822; unpaired two-tailed t test.

Fig. 2.

Humidity memory formation requires VP1m+VP5 ilPN mALT neurons. (A) Blocking synaptic output in mALT neurons (VP2++ lvPN and VP1m+VP5 ilPN) via VT40053-Gal4 driven UAS-shi^ts expression at 32 °C significantly impairs humidity memory. Each value represents mean ± SEM (N = 7, 8, and 8 from left to right bars). P = 0.0233; one-way ANOVA followed by Tukey’s test. The humidity memory in VT40053/shi^ts flies was normal at 23 °C. Each value represents mean ± SEM (N = 8, 8, and 9 from left to right bars). P = 0.7752; one-way ANOVA followed by Tukey’s test. (B) Blocking synaptic output in mALT (VP1m+VP5 ilPN) neurons using shi^ts significantly impairs humidity memory. Each value represents mean ± SEM (N = 12, 14, and 12 from left to right bars). P = 0.0278; one-way ANOVA followed by Tukey’s test. The humidity memory in R24G07/shi^ts flies is normal at 23 °C. Each value represents mean ± SEM (N = 8, 8, and 7 from left to right bars). P = 0.5439; one-way ANOVA followed by Tukey’s test. (C) Silencing mALT (VP1m+VP5 ilPN) neural activity via the GtACR2 transgene with blue light irradiation during acquisition disrupts humidity memory (Left). Silencing mALT (VP1m+VP5 ilPN) neural activity during retrieval does not affect humidity memory (Right). Each value represents mean ± SEM (N = 8 for each bar). P-values: 0.005 and 0.3974; one-way ANOVA followed by Tukey’s test. (D) Temporal activation of mALT (VP1m+VP5 ilPN) neurons via TrpA1 transgene, simultaneously paired with odor, led to the formation of artificial humidity memory in thirsty flies (Left). TrpA1 activation of mALT neurons (VP1m+VP5 ilPN) paired with odor does not result in the formation of artificial humidity memory in water-sated flies (Right). Each value represents mean ± SEM (N = 9, 9, 9, 16, 12, and 11 from left to right bars). P-values: 0.0045 and 0.8542; one-way ANOVA followed by Tukey’s test. (E) The colocalization of mALT (VP1m+VP5 ilPN) and Ir40a-positive neurons in antenna lobes. The morphology of mALT (VP1m+VP5 ilPN) neurons in the fly brain visualized using R24G07-LexA (green) whereas the morphology of humidity-sensing neurons in the fly brain visualized using Ir40a-Gal4 (red). The arrow indicates VP1m+VP5 ilPN mALT neural projections. (Scale bar, 25 μm.) (F) The morphology of humidity-sensing neurons in the fly brain visualized using Ir40a-Gal4 (green). The morphology of the region downstream of humidity-sensing neurons in the fly brain visualized using trans-Tango transgene (red). The arrow indicates the trans-Tango signals of VP1m+VP5 ilPN mALT neural projections. (Scale bar, 25 μm.)

Fig. 3.

Water memory concurrently inhibits humidity memory. (A) A schematic diagram of different training protocols. (B) Training flies with CS+ odor paired separately with water and humidity results in significantly reduced memory performance (Left, green and red bars) as compared to CS+ odor paired with humidity alone (Left, white bar). Ir40a and Ir93a mutant flies that cannot form humidity memory (two Middle panels, white bars) exhibit significant memory performance in CS+ odors paired separately with water and humidity groups (two Middle panels, green and red bars); ppk28 mutant flies that cannot form water memory do not show any change in memory performance when trained with CS+ odor paired with humidity (Right, white bar) or CS+ odor paired separately with water and humidity (Right, green and red bars). Each value represents mean ± SEM (N = 7, 8, 8, 7, 9, 8, 10, 8, 8, 8, 8, and 8 from left to right bars). P-values: < 0.0001, < 0.0001, < 0.0001, and 0.5525; one-way ANOVA followed by Tukey’s test. (C) Training flies with different odors (OCT or MCH) paired with water and humidity separately causes flies to approach the odor that was paired with water during training. Each value represents mean ± SEM (N = 7 and 8 from left to right bars). (D) The formation of water memory causes the forgetting of humidity memory. In trial 1, thirsty flies were trained in OCT paired with humidity and MCH paired with water, and tested in OCT vs. BEN or MCH vs. BEN. In trial 2, thirsty flies were trained in MCH paired with humidity and OCT paired with water, and tested in OCT vs. BEN or MCH vs. BEN. Flies only form water memory when experienced to odor paired with humidity association. Each value represents mean ± SEM (N = 10, 10, 8, and 8 from left to right bars). P-values: 0.0014 and 0.0006; unpaired two-tailed t test.

Fig. 4.

PAM-γ4 neural activity induces humidity memory forgetting. (A) Flies were trained for odor paired with humidity and tested at 3 h after training. Inhibiting neurotransmission in the majority of PAM (DDC-Gal4) or PAM-γ4 (MB312B-Gal4) neurons by shi^ts significantly prolongs humidity memory. Each value represents mean ± SEM (N = 89, 10, 10, 9, 8, 9, 8, 9, 8, 12, 10, 7, 8, 8, 9, 10, 10, 11, 13, 11, and 11 from left to right bars). P-values: 0.0002, 0.5188, 0.5509, 0.9783, 0.3029, 0.3394, 0.2048, 0.5928, 0.0002, and 0.6502; shi^ts/+, Gal4/+, and Gal4/shi^ts were compared; one-way ANOVA followed by Tukey’s test. (B) Artificial activation of the majority of PAM (DDC-Gal4) or PAM-β′2,γ4,γ5(R48B04-Gal4) or specific PAM-γ4 (MB312B-Gal4) neurons by TrpA1 significantly reduced humidity memory. Artificial activation of PAM-β′1 (VT8167-Gal4) neurons by TrpA1 did not affect humidity memory. Each value represents mean ± SEM (N = 29, 9, 10, 7, 7, 8, 8, 12, and 12 from left to right bars). P-values: < 0.0001, 0.4679, < 0.0001, and 0.0073; TrpA1/+, Gal4/+, and Gal4/TrpA1 were compared; one-way ANOVA followed by Tukey’s test. (C) Live calcium brain imaging shows that PAM-γ4 neurons are responsive to water but not humidity stimuli. Fluorescence signals were recorded from the axons of PAM-γ4 neurons. PAM-γ4 neural responses to water drinking are shown in the Top figure. PAM-γ4 neural responses to humidity are shown in the Bottom figure. (N = 9 and 7 for water and humidity stimuli, respectively). (D) The increased slow calcium oscillations in PAM-γ4 neurons were observed after odor/humidity-paired training and neural activity recording during the 0 to 30 min interval, compared to naïve and unpaired training groups or paired training groups with neural activity recorded during the 30 to 60 min or 60 to 90 min intervals. Fluorescence signals were recorded from the axons of PAM-γ4 neurons. (E) Quantification of the oscillatory responses by frequency. Each value represents mean ± SEM (N = 10, 13, 11, 11, and 6 from left to right bars). P = 0.0015; one-way ANOVA followed by Tukey’s test. (F) Quantification of the oscillatory responses by amplitude. Each value represents mean ± SEM (N = 10, 13, 11, 11, and 10 from left to right bars). P = 0.0022; one-way ANOVA followed by Tukey’s test.

Fig. 5.

Dop2R in γ MBn is critical for humidity memory forgetting. (A) Inhibiting synaptic output in γ MBn using shi^ts significantly decreased humidity memory (Right). Each value represents mean ± SEM (N = 8, 11, and 8 from left to right bars). P < 0.0001; one-way ANOVA followed by Tukey’s test. The permissive temperature (23 °C) control experiment (Left). Each value represents mean ± SEM (N = 8, 8, and 8 from left to right bars). P = 0.3240; one-way ANOVA followed by Tukey’s test. (B) Blocking synaptic output in γ dorsal and γ main MBn subsets impaired humidity memory. Each value represents mean ± SEM (N = 35, 9, 9, 9, 11, 8, 8, 8, and 8 from left to right bars). P-values: < 0.0001, 0.0004, < 0.0001, and < 0.0001; shi^ts/+, Gal4/+, and Gal4/shi^ts were compared; one-way ANOVA followed by Tukey’s test. (C–E) Adult-stage-specific knockdown of Dop1R1, DopEcR, and Dop1R2 in γ do not affect humidity memory (C–E). Each value represents mean ± SEM (N = 8, 8, 8, 10, 10, 9, 13, 12, and 8 from left to right bars). P-values: 0.9928, 0.7401, and 0.2772; one-way ANOVA followed by Tukey’s test. (F) Adult-stage-specific knockdown of Dop2R in γ extended humidity memory to 3 h after training. Each value represents mean ± SEM (N = 21, 14, and 18 from left to right bars for Left). P < 0.0001; one-way ANOVA followed by Tukey’s test. N = 9, 9, and 8 from left to right bars for Left, P = 0.6097. (G) Genetic knockdown of rut in γ MBn significantly disrupted humidity memory. Each value represents mean ± SEM (N = 9, 11, and 12 from left to right bars). P < 0.0001; one-way ANOVA followed by Tukey’s test.

Fig. 6.

mALT neurons convey humidity signals and PAM-γ4 neurons convey forgetting signals to γ MBn for humidity memory process. (A) Live calcium brain imaging shows that γ lobe is responsive to humidity stimulus. Genetic inhibition of mALT neurons by Kir2.1 disrupted the GCaMP7 response to humidity in γ lobe. Each value represents mean ± SEM (N = 8 and 10 from left to right bars). P = 0.0005; unpaired two-tailed t test. (B) DA2m, a dopamine sensor, was used to monitor dopamine secretion in flies after different training protocols. In the paired training protocol, CS+ odor and humidity were delivered simultaneously during live brain imaging assays. In the unpaired training protocol, odor and humidity were delivered separately. (C) Recorded DA2m fluorescence changes in the MB γ lobe after paired and unpaired training and the γ4 compartment of MBn was analyzed. The quantification data are shown in the Right panel. Each value represents mean ± SEM (N = 10 and 8 from left to right bars). P < 0.0001; unpaired two-tailed t test. (D) DA2m fluorescence changes in PAM-γ4 neurons after paired and unpaired training. The quantification data are shown in the Right panel. Fluorescence signals were recorded from the axons of PAM-γ4 neurons. Each value represents mean ± SEM (N = 11 and 9 from left to right bars). P = 0.0163; unpaired two-tailed t test.

Fig. 7.

Inhibiting synaptic output from PAM-γ4 neurons protects humidity memory from being inhibited by water memory. (A) Thirsty flies were trained to CS+ odor paired with both water and humidity simultaneously. Inhibiting synaptic output in PAM neurons using shi^ts at restrictive temperatures (32 °C) significantly increases memory (red bar) as compared to the control groups. Each value represents mean ± SEM (N = 9 for each bar). P = 0.0026; one-way ANOVA followed by Tukey’s test. No significant difference is noted in humidity memory at permissive temperatures (23 °C) in DDC/shi^ts flies. Each value represents mean ± SEM (N = 9 for each bar). P = 0.7532; one-way ANOVA followed by Tukey’s test. (B) Thirsty wild-type (+/+) or ppk28 mutant (ppk28) flies were trained to CS+ odors paired with both water and humidity simultaneously. ppk28 mutant flies showed significantly increased memory (red bar) compared to the wild-type flies. Each value represents mean ± SEM (N = 12 for each bar). P = 0.0051; unpaired two-tailed t test. (C) Thirsty flies were trained to CS+ odor paired with humidity, followed by CS+ odor paired with water and then tested. Inhibiting synaptic output in PAM-γ4 neurons significantly increases memory (red bar). Each value represents mean ± SEM (N = 14 for each bar). P = 0.0103; one-way ANOVA followed by Tukey’s test. (D) There was no significant difference in humidity memory at permissive temperatures (23 °C) in MB312B/shi^ts flies compared to the control groups. Each value represents mean ± SEM (N = 9 for each bar). P = 0.9982; one-way ANOVA followed by Tukey’s test. (E) Thirsty flies were trained to CS+ odor paired with both water and humidity simultaneously. Inhibiting synaptic output in PAM-γ4 neurons using shi^ts at restrictive temperatures (32 °C) significantly enhanced memory (red bar) compared to the control groups whereas no effect was observed at permissive temperatures (23 °C). Each value represents mean ± SEM (N = 9, 9, 9, 8, 8, and 8 from left to right bars). P-values: < 0.0001 and 0.5072; one-way ANOVA followed by Tukey’s test. (F) Dop2R mutant flies (Dop2R^−/−; UAS-Dop2R) show significant increase memory compared to wild-type flies (+/+), whereas overexpression of Dop2R in γ MBn (Dop2R^−/−; VT44966/UAS-Dop2R) abolishes this effect. Each value represents mean ± SEM (N = 10 for each bar). P = 0.0002; one-way ANOVA followed by Tukey’s test.

Fig. 8.

A model of humidity memory formation and forgetting in thirsty Drosophila. (A) Drosophila form humidity memory through olfactory circuits (olfactory receptor neurons [ORN]) and humidity circuits (IR40a, IR93a, and humidity sensing neurons [HSN]). Both olfactory and humidity information is converged in the γ MBn via VP1m+VP5 ilPN mALT neurons for humidity memory formation. (B) Flies follow odor/humidity cues in searching for water to quench their thirst. PAM-γ4 neural activity is increased during water consumption, which induces water memory formation via Dop1R1 and inhibits humidity memory via Dop2R. (C) PAM-γ4 neural oscillations are increased after odor/humidity association. However, if thirsty flies do not find water within 30 min, the humidity memory is totally erased by PAM-γ4 via Dop2R modulation in γ MBn, prompting the flies to search for water source in different directions. This mechanism helps thirsty flies to search in the correct direction to find water sources. Once the flies find and drink water, water memory is formed and humidity memory is erased.