Forgotten memory storage and retrieval in Drosophila

Abstract

Inaccessibility of stored memory in ensemble cells through the forgetting process causes animals to be unable to respond to natural recalling cues. While accumulating evidence has demonstrated that reactivating memory-stored cells can switch cells from an inaccessible state to an accessible form and lead to recall of previously learned information, the underlying cellular and molecular mechanisms remain elusive. The current study used Drosophila as a model to demonstrate that the memory of one-trial aversive olfactory conditioning, although inaccessible within a few hours after learning, is stored in KCαβ and retrievable after mild retraining. One-trial aversive conditioning triggers protein synthesis to form a long-lasting cellular memory trace, approximately 20 days, via creb in KCαβ, and a transient cellular memory trace, approximately one day, via orb in MBON-α3. PPL1-α3 negatively regulates forgotten one-trial conditioning memory retrieval. The current study demonstrated that KCαβ, PPL1-α3, and MBON-α3 collaboratively regulate the formation of forgotten one-cycle aversive conditioning memory formation and retrieval.

Fig. 1. The memory of one-cycle aversive conditioning is retrievable.

a Experimental protocol. b The memory of one-cycle aversive conditioning remained retrievable after mild-stimulation, 30 V training in the pretrained group. There was no significant behavioral performance observed in the unpaired group. There was no retrieved memory found 2 days after 90 V training, 2 days memory. For the 2-day group, statistical comparison was carried out by one-way ANOVA with Dunnett’s post-hoc test. N = 9, 8 and 8, p = 0.0201, 0.0057. For the other groups, statistical comparison was carried out by two-tailed unpaired t-test. N = 6 and 6, p = 0.0346 for the 6-day group. N = 8 and 8, p = 0.0006 for the 8-day group. N = 6 and 6, p = 0.2379 for the 20-day group. c No memory performance was found when the CS+ and CS- odors were reversed during mild retraining. N = 6 for each group. Statistical comparison was carried out by two-tailed unpaired t-test, p = 0.3948. d Output activity inhibition in KCαβ during the fm-retrieval phase inhibited forgotten memory. N = 6, 6, 7, 7, 6, 6; 6, 6, 6, 6, 6, 6; 6, 7, 6, 5, 7, 5. Statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0022, 0.0017, 0.0012; 0.005, 0.0004, 0.0291; 0.0022, 0.0408, 0.5756 (from left to right). e Left: experimental setup. Right: Increased calcium signal from GCaMP7f was found during CS+ exposure in the flies 8 days after 90 V training but not 8 days after unpaired training. The gray bar represents the time of odor delivery and the area for statistical analysis. N = 11 and 9. Calcium imaging data (ΔF/F0) were evaluated by two-tailed paired t-test, p = 0.0005 and 0.1491. ^*p < 0.05. ^**p < 0.01. ^***p < 0.001. In all figures, each value represents the mean ± SEM. Shi^ts, shibire^{temperature-sensitive}. CS +, conditioned stimulus positive. CS-, conditioned stimulus negative.

Fig. 2. A neural circuit is involved in forming and retrieving forgotten memory.

a Output activity inhibition in KCγ and KCαβ during 90 V training affected forgotten memory. N = 6 for each group. Statistical comparison was carried out by one-way ANOVA with Dunnett’s post-hoc test, p = 0.0416, 0.0098. b, d, e, f Output activity inhibition in KCγ, KCαβ, PPL1-α3 (MB630B), MBON-γ1pedc > αβ (MB112C) and MBON-α3 (G0239) during 90 V training and 3 h after 90 V training affected forgotten memory. For b, N = 6, 6, 6, 6, 5, 5, 6, 6, 5, 5, statistical comparison was carried out by one-way ANOVA with Dunnett’s post-hoc test, p = 0.0001, 0.0002. For d, N = 6 for each group, statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0386. For e, N = 6 for each group, statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0006, <0.0001, 0.2125. For f, N = 5 for each group, statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0011, 0.0003, 0.5391. c Output activity inhibition of KCγ during the early stage prevented memory trace expression in the α3 region of KCαβ. N = 7 for each test group. Calcium imaging data (ΔF/F0) were evaluated by two-tailed paired t-test, p = 0.1994 and 0.0193. g Output activity inhibition in MBON-α3 (MB082C) during mild retraining affected forgotten memory. N = 6 for each test group. Statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0264. ^*p < 0.05. ^**p < 0.01. ^***p < 0.001. ^****p < 0.0001. In all figures, each value represents the mean ± SEM. Shi^ts, shibire^{temperature-sensitive}. CS +, conditioned stimulus positive. CS-, conditioned stimulus negative.

Fig. 3. One cycle of aversive conditioning triggers protein synthesis for forgotten memory formation.

a Flies fed CHX 12 h before and 24 h after 90 V training showed disruption of forgotten memory. N = 6 for each test group, statistical comparison was carried out by one-way ANOVA with Dunnett’s post-hoc test, p = 0.0009, 0.0048. b, c, d, and e) Transgenic flies were transferred to a 30 °C environment 12 h before and 24 h after 90 V training to abolish protein synthesis. Disruption of forgotten memory was found only in VT29246-Gal4>Ricin^cs and MB082C-Gal4>Ricin^cs flies. For b, N = 5 for each group, statistical comparison was carried out by two-tailed unpaired t-test. For c, N = 5, 5, 5, 5, 5, 5, 5, 4, 4, 4, statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0015. For d, N = 6, 7, 5, 6, 6, 5, 6, 6, 6, statistical comparison was carried out by two-tailed unpaired t-test. For e, N = 6, 6, 4, 6, statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0042. f and g Knockdown of CrebB and Orb in adult flies in KCαβ and MBON-α3 abolished forgotten memory, respectively. For f, N = 6 for each group, statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0001, 0.0003, 0.1556. For g, N = 6 for each group, statistical comparison was carried out done by one-way ANOVA with Dunnett’s post-hoc test, p = 0.0001, 0.0011. ^*p < 0.05. ^**p < 0.01. ^***p < 0.001. ^****p < 0.0001. In all figures, each value represents the mean ± SEM. Shi^ts, shibire^{temperature-sensitive}. Orb, oo18 RNA-binding protein. CHX cycloheximide, Per period, Cry cryptochrome, Ser Serrate.

Fig. 4. Unconsolidated memory is the main component that forms forgotten memory.

a 90 V trained flies experienced 2 min of cold-shock 30 min after 90 V training abolished the forgotten memory. N = 6, 6, 5. Statistical comparison was carried out by one-way ANOVA with Dunnett’s post-hoc test, p = 0.0376. b 90 V trained flies that experienced 2 min of cold shock 2 h after 90 V training prevented the formation of a cellular memory trace. N = 9 and 8. Calcium imaging data (ΔF/F0) were evaluated by two-tailed paired t-test, p = 0.0002 and 0.2456. c Knocking down radish in adult MBNs for 8 days before and 4 days after 90 V training did not affect forgotten memory. Mild retraining was performed 4 days after 90 V training. N = 6 for each group. Statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0001, 0.0011, 0.0002, 0.0012. d Knockdown of radish in adult MBNs for 8 days reduced ARM expression. N = 6 for each group. Statistical comparison was carried out by one-way ANOVA with Dunnett’s post-hoc test, p = 0.0127. e, f Overexpressed Rac1 V12 and cdc42 V12 in MBNs disruption and enhanced forgotten memory, respectively. Rac1 V12 was expressed for 8 days before 90 V training. cdc42 V12 was expressed for 4 days before 90 V training. N = 6 for each group. For e, statistical comparison was carried out by two-tailed unpaired t-test, p < 0.0001, p = 0.001, 0.6874. For f, statistical comparison was carried out by one-way ANOVA with Dunnett’s post-hoc test, p = 0.002, 0.0301, 0.0387, p < 0.0001. ^*p < 0.05. ^**p < 0.01. ^***p < 0.001. ^****p < 0.0001. In all figures, each value represents the mean ± SEM. Rad radish, Tub tubulin.

Fig. 5. Cellular memory trace formed in the α3 region of KCαβ and MBON-α3 after training.

Increased calcium signal during CS+ odor exposure in the α3 region of KCαβ 9–10 h (b) but not 3–4 h (a) after 90 V training. For both a and b, N = 14 in each group, and the statistical comparisons were carried out by two-tailed paired t-test, p = 0.3683, 0.3412, 0.0424, 0.2406. (from top to bottom) c There was no increased calcium signal during CS+ odor exposure in MBON-α3 8 days after training. N = 10. Statistical comparison was carried out by two-tailed paired t-test, p = 0.4782. d Increased calcium signal during CS+ odor exposure in MBON-α3 3–4 h after 90 V-training. N = 16 for the paired group, and N = 15 for the unpaired group. Statistical comparisons were carried out by two-tailed paired t-test, p = 0.0036 and 0.2621. e Increased calcium signal during CS+ odor exposure in MBON-α3 after mild retraining in the pretrained group. N = 10 for each group. Statistical comparisons were carried out by two-tailed paired t-test, p = 0.0005 and 0.6045. f Flies fed CHX but actinomycin D (act.) (5 µg/ml) 24 h before mild retraining abolished forgotten memory. For the before 90 V training group, act. was applied 12 h before and 24 h after 90 V training. N = 6 for each group. Statistical comparisons were carried out by two-tailed paired t-test, p = 0.0002, 0.1999, 0.0007, 0.4316. g CHX treatment 24 h before mild retraining decreases the calcium response during CS+ exposure. The data were reanalyzed from Figure S5f, in order to emphasize the difference between groups; experiments were performed at the same time. N = 12 in each group. Statistical comparison was carried out by two-tailed unpaired t-test, p = 0.0085. ^*p < 0.05. ^**p < 0.01. ^***p < 0.001. In all figures, each value represents the mean ± SEM. The N values in a, b, c, d, e, and g) represent the number of flies recorded in each experiment, whereas the N value in f) represents the batches of flies trained and tested in the behavioral assay in each group. CHX, cycloheximide. Act, actin. CS + conditioned stimulus positive, CS- conditioned stimulus negative.

Fig. 6. PPL1-α3 negatively regulates forgotten memory retrieval.

a Output activity inhibition of PPL1-α3 (MB630B) during mild retraining promoted forgotten memory retrieval. N = 6 for each group. Statistical comparisons were carried out by one-way ANOVA with Dunnett’s post-hoc test, p = 0.001, 0.0001. b, c Output activity inhibition of KCαβ and MBON-α3 during the retraining and retrieval phases disrupted forgotten memory retrieval, while inhibition of PPL1-α3 only during the retraining phase promoted forgotten memory performance. For b N = 7, 8, 7, 7 (from left to right). Statistical comparisons were carried out by two-tailed unpaired t-test, p = 0.0269. For c N = 7, 6, 6, 6(from left to right). Statistical comparisons were carried out by two-tailed unpaired t-test, p = 0.0284, 0.0423. ^*p < 0.05. ^**p < 0.01. ^***p < 0.001. In all figures, each value represents the mean ± SEM. The N values in this figure represent the batches of flies trained and tested in the behavioral assay in each group. Shi^ts, shibire^{temperature-sensitive}.

Fig. 7. Interference training promotes the decay of cellular memory traces.

a Experimental protocol. The forgotten memory could be retrieved with mild retraining 3 days (c) but not 8 days (b) after interference training. For b, N = 7 in each group. Statistical comparisons were carried out by two-tailed unpaired t-test, p = 0.0047 and 0.7805. For c, N = 7 in each group. Statistical comparisons were carried out by two-tailed unpaired t-test, p = 0.0003 and 0.0162. d There was no observed cellular memory trace 8 days after interference training. N = 11 and N = 12(top to bottom). Statistical comparisons were carried out by two-tailed paired t-test, p = 0.2192 and 0.0190. e There was no observed cellular memory trace in Rac1 V12 overexpression flies 8 days after 90 V training. Rac1 V12 was expressed 8 days before 90 V training. N = 11 and N = 10(top to bottom). Statistical comparisons were carried out by two-tailed paired t-test, p = 0.0681 and 0.0054. f There was no observed cellular memory trace 20 days after 90 V training in the pretrained group. N = 18 and N = 15(top to bottom). Statistical comparisons were carried out by two-tailed paired t-test, p = 0.8426 and 0.2104. ^*p < 0.05. ^**p < 0.01. ^***p < 0.001. In all figures, each value represents the mean ± SEM. The N values in b and c represent the batches of flies trained and tested in the behavioral assay in each group, whereas the N values in d, e, and f) represent the number of flies recorded in each experiment. CS +, conditioned stimulus positive. CS- conditioned stimulus negative, Ctrl control.

Fig. 8. Proposed forgotten memory formation and retrieval.

During acquisition and consolidation, 90 V training activates the neural circuit, including KCαβ, MBON-α3, and PPL1-α3. The cellular pathways of CREB and orb are activated in KCαβ and MBON-α3, respectively. The corresponding cellular memory trace is formed in KCαβ and MBON-α3 with different time courses and decay rates. Therefore, the memory of one-cycle aversive conditioning is stored between these two neurons in the early stage. We speculate, dot line, that activated MBON-α3 activates PPL1-α3 to release dopamine to negatively regulate KCαβ and MBON-α3 activity. Forgetting process: released dopamine from PPL1-α3 via dop1R2 to affect the synaptic connection between KCαβ and MBON-α3. Since the cellular memory trace has vanished in MBON-α3 within 24 h, the memory of one-cycle aversive conditioning is only stored in KCαβ. However, the mark of one-cycle conditioning was left in MBON-α3 as mRNA accumulation. During retraining, mild retraining 1) restores and enhances the synaptic connectivity between KCαβ and MBON-α3 and 2) reduces the activity of PPL1-α3. Activated MBON-α3 translates accumulated mRNA to form stable cellular memory. During retrieval, since the positive force, the synaptic connectivity between KCαβ and MBON-α3 is stronger than the negative force, inhibition from PPL1-α3, the memory is retrieved. Passive decay: if there is no similar stimulation in the future to strengthen the activity of KCαβ, the memory would eventually decay, and the memory disappears. Active decay: if there is interference activity from the environment to interfere with memory formation and storage, memory decay will be facilitated. The experience of interference would not affect the formation of memory traces but facilitate memory trace decay. Part of images were Adapted from “Synaptic Cleft with Signaling Molecules (Layout, Horizontal)”, by BioRender.com (2022). Retrieved from https://app.biorender.com/biorender-templates.