Specific dopaminergic neurons for the formation of labile aversive memory

Abstract

A paired presentation of an odor and electric shock induces aversive odor memory in Drosophila melanogaster. Electric shock reinforcement is mediated by dopaminergic neurons, and it converges with the odor signal in the mushroom body (MB). Dopamine is synthesized in approximately 280 neurons that form distinct cell clusters and is involved in a variety of brain functions. Recently, one of the dopaminergic clusters (PPL1) that includes MB-projecting neurons was shown to signal reinforcement for aversive odor memory. As each dopaminergic cluster contains multiple types of neurons with different projections and physiological characteristics, functional understanding of the circuit for aversive memory requires cellular identification. Here, we show that MB-M3, a specific type of dopaminergic neurons in the PAM cluster, is preferentially required for the formation of labile memory. Strikingly, flies formed significant aversive odor memory without electric shock when MB-M3 was selectively stimulated together with odor presentation. In addition, we identified another type of dopaminergic neurons in the PPL1 cluster, MB-MP1, which can induce aversive odor memory. As MB-M3 and MB-MP1 target the distinct subdomains of the MB, these reinforcement circuits might induce different forms of aversive memory in spatially segregated synapses in the MB.

Figure 1. Dopaminergic inputs into the mushroom body via MB-M3 neurons

(A) Projection of MB-M3 neurons visualized in NP5272 UAS-mCD8::GFP. In the MB (light green overlay based on the Synapsin counterstaining), MB-M3 neurons arbors in the medial tip of the β lobe (β₂; arrowhead). Sparse terminals are detected also in the β' lobe ((β’₂; small arrowhead). (B) In UAS-Syt::HA; NP5272 UAS-mCD8::GFP, the arborizations in the anterior and middle inferiormedial protocerebrum (aimpr-mimpr; arrows) are only weakly labeled by presynaptic marker Syt::HA (magenta), if any, relative to mCD8::GFP (green), implying their dendritic nature. Scale bar 20 µm. (C) The GAL4-expressing cells in the PAM cluster are visualized with UAS-Cameleon2.1. and counted. No significant difference is observed between NP5272, NP1528 and the comnbination of them. Also, combining NP5272 and NP1528 does not significantly increase the number of the labeled cells as compared to TH-GAL4 alone. Throughout the paper, bars and error bars represent the mean and s.e.m., respectively. Unless otherwise stated, the most conservative statistical result of multiple pair-wise comparisons is shown throughout the paper. * P<0.05; ** P<0.01; *** P<0.001; n.s. not significant. (D) The diagram illustrates the terminal areas of GAL4-expressing cells of TH-GAL4 (green) and NP5272/NP1528 (orange) in the MB lobes, based on cell counting and a single-cell analysis [11]. The MB lobes are shown as an outline. See also Table S2. (E) Aversive odor memory tested at 30 min after training. The respective dopaminergic neurons are blocked with Shi^ts1 driven by TH-GAL4, NP5272 or NP1528. Block with these drivers significantly impairs aversive memory. n=17–22. (F)Thirty-min appetitive memory of the same genotypes. Learning indices of all the experimental groups (TH-GAL4/UAS- shi^ts1 and NP5272/UAS-shi^ts1) are not significantly different from corresponding control groups (P>0.05; one-way ANOVA). n=13–16. (G) Transient block only during aversive training. The results are essentially the same as in (E). n=20–22.

Figure 2. Preferential requirement of MB-M3 for inducing labile middle-term memory

(A) Requirement of the MB-M3 neurons for different memory phases. Flies were trained and tested immediately at restrictive temperature, or kept for two or nine hours and tested at permissive temperature. Memory is significantly impaired at all three retention intervals by TH-GAL4. Blocking MB-M3 slightly affects immediate and 9-hour memory, but only 2-hour memory is significantly impaired. n=14–20. (B) Total memory (ASM+ARM; the same data set as in a), a consolidated memory component (ARM), and memory induced by unpaired presentation of odors and electric shock tested at 2 hours after training (n=12–18). Although the block with TH-GAL4 impairs both total memory and ARM, only total memory is significantly impaired when MB-M3 neurons are blocked during training. The requirement of the MB-M3 neurons for the total memory and ARM is differential (P<0.05; significant interaction [genotype X cold shock treatment] in two-way ANOVA).

Figure 3. dTrpA1-dependent activation of MB-M3 can induce aversive odor memory

(A) The conditioning protocol for dTrpA1-induced memory (see Experimental Procedures for detail). (B) Immediate aversive odor memory formed by odor presentation and simultaneous thermo-activation of the subsets of dopaminergic neurons expressing dTrpA1. n=18–22. (C) Thermo-activation did not induce aversive odor memory when it is applied 60–120 seconds prior to the presentation of the odor. n=10–12. (D–G) Expression pattern of TH-GAL4 (D) and NP5272 (F) in the brain (left panels; frontal view, dorsal up) and thoracicoabdominal ganglion (right panels; dorsal view, anterior up). TH-GAL80 silences mCD8::GFP expression in MB-M3 (G) and the most of cells labeled by TH-GAL4 (E). Remaining cells are presumably non-dopaminergic cells judging from their size and position [11]. Scale bars: 20 µm. (H) Immediate aversive odor memory induced by dTrpA1-dependent activation is significantly suppressed by TH-GAL80, indicating that the corresponding dopaminergic neurons are responsible. n=15–17.

Figure 4. dTrpA1-induced memory by other dopaminergic neurons

(A, C, E, G) Projection of brain regions including the MB (light green outline; frontal view, dorsal up). Various dopaminergic neurons projecting to the MB are visualized with mCD8::GFP (arrowheads) driven by c061; MB-GAL80 (A), NP2758 (C), MZ840 (E), and NP6510 (G). Scale bar 20 µm. see also Figure S2 and Table S2. (B, D, F, H) Memory induced by dTrpA1-dependent activation of the various types of dopaminergic neurons. (E) With c061;MB-GAL80, robust immediate and 2-hour memory are formed and significantly suppressed by TH-GAL80 and Cha^3.3kb-GAL80. n=18–22. (F) Activation of dTrpA1-expressing cells in NP2758 induces robust aversive odor memory, which is significantly suppressed by TH-GAL80. (G) Despite the tendency, no significant memory is formed with MZ840. (H) With NP6510, the learning index of NP6510/UAS-dTrpA1 is different from NP6510/+ but not from +/UAS-dTrpA1. n=15–18.