Dopamine is required for learning and forgetting in Drosophila

Abstract

Psychological studies in humans and behavioral studies of model organisms suggest that forgetting is a common and biologically regulated process, but the molecular, cellular, and circuit mechanisms underlying forgetting are poorly understood. Here we show that the bidirectional modulation of a small subset of dopamine neurons (DANs) after olfactory learning regulates the rate of forgetting of both punishing (aversive) and rewarding (appetitive) memories. Two of these DANs, MP1 and MV1, exhibit synchronized ongoing activity in the mushroom body neuropil in alive and awake flies before and after learning, as revealed by functional cellular imaging. Furthermore, while the mushroom-body-expressed dDA1 dopamine receptor is essential for the acquisition of memory, we show that the dopamine receptor DAMB, also highly expressed in mushroom body neurons, is required for forgetting. We propose a dual role for dopamine: memory acquisition through dDA1 signaling and forgetting through DAMB signaling in the mushroom body neurons.

Figure 1. Bidirectional Modulation of Dopamine Neuron Activity after Learning Regulates Forgetting

The experimental design above each graph shows the time of acquisition (A) and retrieval (R) relative to the time of elevated temperature at 32°C (black) from the 23°C (gray) baseline. (A) Three hour performance index (PI) was significantly enhanced after a 40 min or longer elevated temperature window for TH-gal4/UAS-shi^ts1 versus all other groups (*p < 0.02, n ≥ 7). (B) An 80 min treatment window that varied in the time of onset significantly enhanced PIs for TH-gal4/UAS-shi^ts1 versus all other groups at all time windows (*p < 0.002, n ≥ 8). The enhanced PIs for all time windows were not statistically different (p > 0.4). (C) Three hour PI was significantly reduced after a 5 min or longer elevated temperature window for UAS-trpA1/+; TH-gal4/+ versus all other groups (*p < 0.0001, n ≥ 8). PIs for UAS-trpA1/+;TH-gal4/+ after a 20 min treatment were reduced to a level not significantly different from zero (p = 0.8). (D) A 20 min treatment window significantly reduced PIs for UAS-trpA1/+;TH-gal4/+ versus all other groups at all time windows (*p < 0.0001, n ≥ 8). Scores for the treated UAS-trpA1/+;TH-gal4/+ flies were not significantly different from each other (p > 0.8) or from zero (p ≥ 0.178). Results are presented as means ± SEM.

Figure 2. Blocking Synaptic Release from a Subset of DANs Reduces Forgetting and Stimulation Accelerates Forgetting

(A) Schematic diagram illustrating putative connectivity (black dots) between the mushroom body (MB) neuron classes (α′/β′, α/α, and γ) and the PPL1 DAN class (Tanaka et al., 2008; Mao and Davis, 2009). The cell bodies, corresponding innervation of the MB lobes, and representative gal4 lines used in this study for each DAN class are color coded (yellow, heel/peduncle; green, lower stalk/junction; blue, upper stalk; red, tip of α lobe; purple, tip of α′ lobe). Axis labels: A, anterior; P, posterior; D, dorsal; V, ventral; M, medial; L, lateral. (B) Three hour PIs were significantly enhanced after a 0–80 min elevated temperature window only for flies that carried TH-gal4 or c150-gal4 along with MBgal80/+; UAS-shi^ts1/+ when compared to normal temperature treatment (*p < 0.05, n ≥ 7). The 80 min elevated temperature treatment caused a significant decrease in PI for MBgal80; UAS-shi^ts1 and NP7135-gal4/+; MBgal80/+; UAS-shi^ts1/+ in this experiment compared to normal temperature (*p < 0.05, n ≥ 7). (C) Three hour PIs were significantly reduced after a 20 min elevated temperature exposure immediately after training for flies carrying UAS-trpA1 with TH-gal4, c150-gal4, or c061-gal4 compared to normal temperature treatment (**p < 0.0001, *p < 0.0005, n ≥ 8). Results are presented as means ± SEM.

Figure 3. Activity of c150-gal4 DA Neurons after Training Is Necessary for Forgetting Early Labile Memories

(A–A″) PI was significantly enhanced after a 0–80 min elevated temperature exposure for MBgal80/+; c150-gal4/UAS-shi^ts1 flies at 6 hr (A) but not at 16 (A′) or 24 hr (A″) compared to normal temperature treatment (**p < 0.0005, n ≥ 8). (B and B′) As in Figure 2B, 3 hr PIs were significantly enhanced when blocking c150 neurons for 0–80 min, while there was a significant decrease for control MBgal80/+; UAS-shi^ts1/+ flies in this experiment compared to normal temperature (*p < 0.003, n ≥ 7). However, there was no significant enhancement after blocking c150 neurons for 0–80 min with an additional cold shock (cs) at 2 hr (p = 0.777, n = 12). (C) Acquisition and immediate retrieval of aversive memory at 23°C or 32°C. PIs immediately after conditioning were significantly reduced at elevated temperature for MBgal80/+; TH-gal4/UAS-shi^ts1 and MBgal80/+; c150-gal4/UAS-shi^ts1 flies compared to normal temperature (*p < 0.0001, n = 8). A significantly lower PI was observed with TH-gal4 compared to c150-gal4 at 32°C (*p < 0.0001, n = 8). (D) PI immediately after conditioning was significantly reduced, as in (C), at elevated temperature for flies carrying c150-gal4, MBgal80, and UAS-shi^ts1 (*p < 0.0001, n = 8), but not for those that also contained THgal80 (p = 1.00) when compared to normal temperature. Results are presented as means ± SEM.

Figure 4. Stimulation of c150 DANs Causes Forgetting of Consolidated Aversive and Appetitive Odor Memories

(A) Six hour aversive PIs were significantly reduced after a 20 min elevated temperature window 15 min prior to retrieval for MBgal80/UAS-trpA1; c150-gal4/+ flies compared to normal temperature treatment (*p < 0.05, n = 8). (B) Three hour aversive PI was significantly reduced after a cold shock at 2 hr followed by a 20 min elevated temperature window 15 min prior to retrieval for MBgal80/UAS-trpA1; c150-gal4/+ flies compared to normal temperature (*p < 0.05, n = 8). The reduced PI was not significantly different from zero (p = 0.751). (C and C′) Appetitive 3 hr PI was significantly reduced after a 20 or 80 min elevated temperature window for UAS-trpA1/+; TH-gal4/+ flies compared to normal temperature (*p < 0.05, n = 8). (D) Appetitive 3 hr PI was significantly reduced after a 0–80 min elevated temperature window for MBgal80/UAS-trpA1; c150-gal4/+ flies compared to normal temperature (*p < 0.05, n = 8). (E) Appetitive 6 hr PI was significantly reduced after an 80 min heat exposure presented 15 min prior to retrieval for MBgal80/UAS-trpA1; c150-gal4/+ flies when compared to normal temperatures (*p < 0.05, n = 8). Results are presented as means ± SEM.

Figure 5. MV1 and MP1 DANs Exhibit Ongoing Synchronized Ca²⁺-Based Activity before and after Training in Awake, Living Animals

(A) The left panel illustrates the projections of the V1, MV1, and MP1 DANs to both the MB neuropil (blue, green, and yellow areas) and protocerebrum (dark gray areas) imaged in this study. The PPL1 DAN cell bodies, located adjacent to the calyx (“c”), project a neurite anteriormedially toward the MB lobes. The V1 DAN (blue) neurite branches to innervate both the middle superior medial protocerebrum (msmpr) just posterior to the vertical tips of the MBs and the middle segments of α′ and α MB lobes. The MV1 and MP1 neurites innervate the anterior inferior medial protocerebrum (aimpr) located dorsal and anterior to the horizontal MB lobes. These neurites then travel around the tip of the horizontal lobes to finally innervate distinct regions of the MBs. The MV1 terminates in the anterior face of the γ lobe and the lower-stalk region of the α′ lobe, while MP1 terminates in the heel region of the γ lobe and the α/β layers of the peduncle. The MV1 and MP1 neurons project to other areas of the protocerebrum, which are not shown for graphical simplicity. The middle and right panels show GCaMP^3.0 expression and the corresponding pseudocolored activity (%ΔDF/F, see Experimental Procedures) from an activity event (red arrow in B) occurring in a representative recording. The regions of interest used to measure activity are identified by outlining. (B) Activity traces (%ΔDF/F) from a 600 s recording from the representative animal in (A) are shown for the MV1, MP1, and V1 innervations of the MBs along with the DAN innervations within the aimpr. The red arrows indicate the time point used for collecting the static images in the middle and right panels of (A). (C) Group data of activity per second (see Experimental Procedures for calculation) for recordings from the MV1, MP1, and V1 innervations of the MBs and the DAN innervations of the aimpr as assayed in naive animals and in animals 15 min after forward and backward conditioning. MV1, MP1, and the aimpr displayed robust ongoing activity in naive animals and after both conditioning paradigms (significantly different from zero, p < 0.0001, n ≥ 20), while V1 showed no significant activity (not significantly different from zero, p ≥ 0.245, n ≥ 17). Learning did not significantly alter the activity in any of the regions measured (ns, p ≥ 0.185). The scores for animals tested behaviorally in parallel with optical imaging were PI = 0.586 ± 0.015 (forward conditioning) and PI = −0.052 ± 0.015 (backward conditioning); p < 0.0001, n ≥ 35. (D) Group data of normalized cross-correlations of activity (see Experimental Procedures) between simultaneously recorded regions of interest. In naive animals, simultaneously recorded activities of MV1, MP1, and the aimpr were highly correlated to each other (significantly different from zero, p < 0.0001, n ≥ 9), while V1 was not significantly correlated with MV1 or the aimpr (not significantly different from zero, p ≥ 0.384, n = 18). Forward and backward conditioning did not significantly alter activity correlations in any of the regions measured (ns, p ≥ 0.194, n ≥ 9). Results are presented as means ± SEM.

Figure 6. Dopamine Receptor DAMB Is Required for Forgetting

(A) Memory retention was significantly enhanced at 3 hr (p = 0.0046, n = 8), 6 hr (p = 0.0003, n = 8), and 24 hr (p = 0.0083, n = 8) after training in animals lacking the DAMB receptor despite a slight but significantly reduced immediate memory (p = 0.0453, n = 8). (B) damb mutants showed a significant reduction in a reversal learning paradigm. Performance index is measured with respect to the second “reversal” odor contingency (p = 0.0166, n = 8). (C) Animals lacking the DAMB receptor acquired memory at a similar rate for one (p = 0.3738, n = 8), two (p = 0.2473, n = 8), three (p = 0.1835, n = 8), and six (p = 0.2673, n = 8) shocks but plateaued at a significantly lower level than Canton-S for 12 shocks (p = 0.0355, n = 8). (D) damb mutants displayed significantly reduced shock avoidance at the higher shock intensities of 45V (p = 0.0004, n = 16) and 90V (p < 0.0001, n = 16) relative to the control. (E) damb mutants exhibit odor avoidance indistinguishable from the control to both MCH and OCT at several odor concentrations (*p < 0.05, n = 8). Results are presented as means ± SEM.

Figure 7. Model of Dopamine-Mediated Learning and Forgetting

The left panel illustrates dopamine-mediated learning. Upon electric shock, a US signal is delivered when DANs release large amounts of dopamine (red spheres) onto the MBs. Binding of dopamine to the dDA1 receptor (green) leads to increased cAMP, while the CS odor stimuli results in increased Ca²⁺ signaling. Coincident Ca²⁺ and cAMP signaling leads to strong memory formation within the MBs (Tomchik and Davis, 2009). The right panel illustrates dopamine-mediated forgetting. After learning, a low level of ongoing dopamine release is sensed by the DAMB receptor (red) and weakens the memory over time. Differences in affinity of the two receptors for ligand, compartmentalization, coupling to downstream signaling pathways, or signaling kinetics may account for the dominant action of dDA1 in acquisition and DAMB in forgetting. Results are presented as means ± SEM.