Availability of food determines the need for sleep in memory consolidation

Abstract

Sleep remains a major mystery of biology, with little understood about its basic function. One of the most commonly proposed functions of sleep is the consolidation of memory^1-3. However, as conditions such as starvation require the organism to be awake and active⁴, the ability to switch to a memory consolidation mechanism that is not contingent on sleep may confer an evolutionary advantage. Here we identify an adaptive circuit-based mechanism that enables Drosophila to form sleep-dependent and sleep-independent memory. Flies fed after appetitive conditioning needed increased sleep for memory consolidation, but flies starved after training did not require sleep to form memories. Memory in fed flies is mediated by the anterior-posterior α'/β' neurons of the mushroom body, while memory under starvation is mediated by medial α'/β' neurons. Sleep-dependent and sleep-independent memory rely on distinct dopaminergic neurons and corresponding mushroom body output neurons. However, sleep and memory are coupled such that mushroom body neurons required for sleep-dependent memory also promote sleep. Flies lacking Neuropeptide F display sleep-dependent memory even when starved, suggesting that circuit selection is determined by hunger. This plasticity in memory circuits enables flies to retain essential information in changing environments.

Extended Data Fig. 1.. Sleep increases in flies fed after appetitive training

(a) Flies starved for 6h prior to training show no difference in sleep between trained and untrained groups. However, moving trained flies into sucrose tubes post-training resulted in a significant increase in sleep compared to untrained controls despite only 6h of pre-training starvation. Sleep was quantified for the ZT8–12 interval (Two-sided t-tests were performed for each condition to compare trained and untrained groups, followed by Bonferroni correction; n=32). (b) and (c) Training increases sleep bout length in fed flies but not in starved flies. Flies were trained after 18h (b) and 6h (c) starvation (Two-sided Mann-Whitney tests were performed for each condition to compare trained and untrained groups; n=32). Data are represented as mean ± s.e.m. Each data point depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. ***p<0.001; **p<0.01; *p<0.05.

Extended Data Fig. 2.. Memory in flies fed after training is sleep and protein synthesis-dependent but independent of light cycles

(a) Long-term memory in fed flies is sensitive to cycloheximide based inhibition of protein-synthesis (Two-sided t-test; n≥6). (b) Flies demonstrate substantial rebound sleep when sleep-deprived in a group of about 100 flies in a vial in both fed and starved conditions. Flies were sleep-deprived from ZT12-ZT24 and then introduced individually into locomotor tubes (Two-sided Mann-Whitney tests were performed for each condition to compare undisturbed and sleep-deprived groups; n=32). (c) Starved flies were effectively sleep-deprived when exposed to a mechanical stimulus post-training (n≥31). Flies were starved for 6h and then trained at ZT6 and subsequently introduced into agar locomotor tubes. A mechanical stimulus was applied for 6h after training. A rebound is evident after sleep deprivation. (d) Flies starved for only 6h, as opposed to 18h, before training and then allowed to feed showed impaired memory performance when sleep-deprived for 6h post-training (Two-sided t-test; n=8). Sleep post-training was comparable to flies depicted in Extended Data Fig. 1(a). (e) 6h sleep deprivation had no effect on long-term memory in flies kept starved after training. Here, flies were starved for 6h before training (Two-sided t-test; n≥6). Sleep post-training was comparable to flies depicted in Extended Data Fig. 1(a). (f) Sleep deprivation initiated 6h after training had no effect on memory in fed and trained flies (Two-sided t-test; n=8). (g) Long-term memory was resistant to sleep deprivation in flies that were starved after conditioning but then tested after a feeding and re-starvation period (Two-sided t-test; n≥6). Flies were starved (and sleep-deprived) for 6h post-training and then allowed to feed for 18h before 30h restarvation for memory tests. (h) and (i) 6h sleep deprivation of flies maintained in constant light affected appetitive long-term memory when they were fed, but not starved, post-training (Two-sided t-test; n≥6). Data are represented as mean ± s.e.m. Each data point in a memory experiment represents a group of flies and in a sleep experiment it depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. ***p<0.001; **p<0.01; *p<0.05.

Extended Data Fig. 3.. The rye mutation affects sleep-dependent memory

(a) Long-term memory is substantially lower in satiated short-sleeping rye mutants. Background iso line was used as control (Two-sided t-test; n≥8). (b) rye mutants form robust appetitive 24h memory, similar to controls when kept starved (Two-sided t-test; n≥6). (c) Satiated rye mutants demonstrate no difference in sleep between trained and untrained groups. Total sleep in the ZT8–12 interval is depicted (Two-sided t-tests were performed for each genotype to compare trained and untrained groups, followed by Bonferroni correction; n≥31). Data are represented as mean ± s.e.m. Each data point in a memory experiment represents a group of flies and in a sleep experiment it depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. **p<0.01; *p<0.05.

Extended Data Fig. 4.. Flies on arabinose demonstrate a significant increase in post-training sleep

(a) Trained flies show a substantial increase in sleep relative to untrained flies when kept on arabinose after appetitive conditioning. Sleep was quantified for the 0–4h interval post-training (Two-sided t-test; n≥31). (b) Bout length was considerably higher in trained flies compared to untrained flies when moved to arabinose after training (Two-sided Mann-Whitney test; n≥31). Data are represented as mean ± s.e.m. Each data point depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data ***p<0.001; **p<0.01.

Extended Data Fig. 5.. npf signaling is essential for sleep-independent memory in starved flies

(a) Starved npfr mutant flies show a substantial increase in sleep post-training compared to untrained flies. npfr/+ was used as control. Total sleep in the 0–4h interval post-training is depicted (Two-sided t-tests were performed for each genotype to compare trained and untrained groups, followed by Bonferroni correction; n≥32). (b) Bout length was considerably higher in trained and starved npfr mutant flies compared to untrained flies. npfr/+ was used as control (Two-sided Mann-Whitney tests were performed for each genotype to compare trained and untrained groups; n≥32). (c) RNAi knockdown of npfr pan-neuronally results in sleep-dependent memory formation in hungry flies. 6h sleep disruption post-training resulted in impaired memory performance in UAS-npfr-RNAi/n-syb-Gal4 flies (Two-sided t-tests were performed for each genotype to compare undisturbed and sleep-deprived groups, followed by Bonferroni correction; n≥6). (d) Starved UAS-npf-RNAi/NPF-Gal4 flies show lower long-term memory when sleep-deprived for 6h post-training (Two-sided t-tests were performed for each genotype to compare undisturbed and sleep-deprived groups, followed by Bonferroni correction; n≥6). Data are represented as mean ± s.e.m. Each data point in a memory experiment represents a group of flies and in a sleep experiment it depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. ***p<0.001; *p<0.05.

Extended Data Fig. 6.. α’/β’ neurotransmission is essential for long-term memory under both fed and starved conditions

(a) Starved UAS-shibire^ts1/MB461B flies kept at restrictive settings for 4h immediately after training show impaired long-term memory (n≥6). Restrictive temperature 8–12 hours after training had no effect (b) (n=6) (One-factor ANOVA with Tukey post-hoc test). (c) Long-term memory remained unchanged in experimental and control flies when kept at 25°C (One-factor ANOVA with Tukey post-hoc test; n=6). (d) and (e) Silencing α’/β’ neurons immediately after conditioning, but not at hours 8–12, affects long-term memory in fed flies (One-factor ANOVA with Tukey post-hoc test; n≥6). (f) Long-term memory remained intact in UAS-shibire^ts1/MB461B flies fed after training but maintained at the permissive temperature (One-factor ANOVA with Tukey post-hoc test; n≥6). Data are represented as mean ± s.e.m. Each data point represents a group of flies. Precise ‘n’ and ‘p’ values are in the Source data. **p<0.01. Asterisks in (a, d) indicate a significant difference between experimental and genetic controls

Extended Data Fig. 7.. Effects of manipulating the activity of α’/β’ subset specific neurons on long-term memory

(a) and (b) Neurotransmission from α’/β’m neurons (UAS-shibire^ts1/R26E01 and UAS-shibire^ts1/MB370B) is dispensable for long-term memory in fed flies (n≥6). Temperature controls are depicted in (b) (n≥6) (One-factor ANOVA with Tukey post-hoc test). (c) and (d) Blocking the activity of α’/β’ap neurons (UAS-shibire^ts1/R35B12 and UAS-shibire^ts1/VT50658) for 4h after conditioning in starved flies has no effect on long-term memory (n=6). Long-term memory in experimental and control flies at the permissive temperature, 25°C, is shown in (d) (n=6) (One-factor ANOVA with Tukey post-hoc test). (e) shibire^ts does not affect memory in flies maintained under starvation conditions at the permissive temperature (One-factor ANOVA with Tukey post-hoc test; n=6). Controls related to Fig 2(a). (f) shibire^ts has no effect on memory in flies maintained on food at the permissive temperature (One-factor ANOVA with Tukey post-hoc test; n≥6). Controls related to Fig. 2(b). (g) Hyperactivation of α’/β’ap neurons (UAS-TrpA1/R35B12) for 4h post-training does not affect long-term memory formation in starved flies (One-factor ANOVA with Tukey post-hoc test; n=6). (h) Memory was not affected in UAS-TrpA1/R35B12 flies at permissive settings (One-factor ANOVA with Tukey post-hoc test; n=6). Data are represented as mean ± s.e.m. Each data point in a memory experiment represents a group of flies. Precise ‘n’ and ‘p’ values are in the Source data.

Extended Data Fig. 8.. α’/β’ subsets differentially regulate sleep

(a) Thermogenetic activation of α’/β’ap neurons (UAS-TrpA1/VT50658) results in a considerable enhancement in sleep while flies in which α’/β’m neurons (UAS-TrpA1/MB370B) were activated showed a significant decrease in sleep (One-factor ANOVA with Tukey post-hoc test; n≥30). (b) and (c) Disabling neurotransmission in α’/β’ap neurons (UAS-shibire^ts1/R35B12 and UAS-shibire^ts1/VT50658) or α’/β’m neurons (UAS-shibire^ts1/R26E01 and UAS-shibire^ts1/MB370B) had no effect on sleep (One-factor ANOVA with Tukey post-hoc test; n≥31). Data are represented as mean ± s.e.m. Each data point depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. Asterisks in (a) indicate a significant difference between experimental and genetic controls.

Extended Data Fig. 9.. The activity of α’/β’ap, but not α’/β’m, neurons is relevant for sleep after conditioning

(a) shibire^ts expression in α’/β’ap neurons has no effect on sleep post-training if flies are maintained at the permissive temperature (Two-sided t-tests were performed for each genotype to compare trained and untrained groups, followed by Bonferroni correction; n≥30). Controls related to Fig. 2(g). (b) and (c) A training-dependent increase in sleep bout length was prevented in flies in which α’/β’ap neurons were silenced (n≥31). Temperature controls are shown in (c) (n≥30) (Two-sided Mann-Whitney tests were performed for each genotype to compare trained and untrained groups). (d) and (e) Trained flies expressing shibire^ts1 in α’/β’m neurons showed an enhancement in sleep even when moved to 32°C for 4h post-training. The total amount of sleep in 0–4h interval after training is quantified (n≥32). Post-training sleep in experimental and control flies at the permissive temperature, 25°C, is shown in (e) (n≥16) (Two-sided t-tests were performed for each genotype to compare trained and untrained groups, followed by Bonferroni correction). (f) and (g) Silencing α’/β’m neurons does not prevent an increase in sleep bout length after training (n≥32). Temperature controls are shown in (g) (n≥16) (Two-sided Mann-Whitney tests were performed for each genotype to compare trained and untrained groups). (h) Calcium/GFP signal in α’/β’m neurons was comparable between control and sleep-deprived flies when kept starved post-training (Two-sided Mann-Whitney test; n≥11). Representative images are shown, two independent experiments; Scale bar, 50 μm. Data are represented as mean ± s.e.m. Each data point depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. ***p<0.001; **p<0.01; *p<0.05.

Extended Data Fig. 10.. Effects of manipulating the neurotransmission of PPL1 neurons and MBONs on long-term memory

(a) and (b) Trained starved flies show lower long-term memory performance when the PPL1 cluster neurons (UAS-shibire^ts1/MB504B) are silenced for 4h post-training (n≥8). Temperature controls are shown in (b) (n≥7) (One-factor ANOVA with Tukey post-hoc test). (c) and (d) Silencing PPL1 DANs affects long-term memory performance in flies kept fed after training (n≥7). Temperature controls are shown in (d) (n≥6) (One-factor ANOVA with Tukey post-hoc test). (e) Expression of shibire^ts1 in MP1 and MV1 neurons at permissive temperature does not affect memory in flies starved after training (One-factor ANOVA with Tukey post-hoc test; n≥6). Controls related to Fig. 3(a). (f) Permissive temperature control for Fig. 3(b). Expression of shibire^ts1 in MP1 and MV1 neurons does not affect memory in flies kept on food at 25^°C after training (One-factor ANOVA with Tukey post-hoc test; n≥6). (g) Blocking the activity of MP1 neurons (UAS-shibire^ts1/MB320C at restrictive temperature) for 6h after conditioning has no effect on long-term memory in flies kept on food vials after training (One-factor ANOVA with Tukey post-hoc test; n=6). (h) and (i) Long-term memory in UAS-shibire^ts1/MB077B and UAS-shibire^ts1/MB112C flies was similar to that of genetic controls when kept starved or fed at 25°C (One-factor ANOVA with Tukey post-hoc test; n≥6). Temperature controls related to Fig. 3(c–d). Data are represented as mean ± s.e.m. Each data point represents a group of flies. Precise ‘n’ and ‘p’ values are in the Source data. Asterisks in (a, c) indicate a significant difference between experimental and genetic controls.

Fig. 1.. Flies fed post-training require sleep for memory consolidation

(a) Flies trained at ZT6, and thereafter kept in agar tubes, show sleep comparable to untrained flies. In contrast, feeding post-training increases sleep in trained flies compared to controls. Sleep amount was quantified for the ZT8–12 interval (0–4h time-points on the curve) (Two-sided t-tests were performed for each condition to compare trained and untrained groups, followed by Bonferroni correction, n=32) (b) 6h sleep disruption affects long-term memory in flies fed post-training (Two-sided t-test; n≥8). Sleep post-training was comparable to flies depicted in (a). (c) 6h sleep deprivation does not affect long-term memory in flies starved post-training (Two-sided t-test; n=8). Sleep post-training was comparable to flies depicted in (a). (d) Long-term memory is sensitive to sleep deprivation in flies kept on arabinose post-training (Two-sided t-test; n=8). (e) 6h sleep disruption affects long-term memory in npfr mutant flies kept starved post-training. npfr/+ was used as control (Two-sided t-tests were performed for each genotype to compare undisturbed and sleep-deprived groups, followed by Bonferroni correction; n≥6). Data are represented as mean ± s.e.m. Each data point in a memory experiment represents a group of flies and in a sleep experiment it depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. ***p<0.001; **p<0.01; *p<0.05.

Fig. 2.. Distinct α’/β’ subsets mediate sleep-dependent and sleep-independent memory

(a) Silencing α’/β’m neurons (UAS-shibire^ts1/R26E01 and UAS-shibire^ts1/MB370B) affects long-term memory in starved flies (One-factor ANOVA with Tukey post-hoc test; n=6). (b) In fed flies, long-term memory is reduced by the silencing of α’/β’ap neurons (UAS-shibire^ts1/R35B12 and UAS-shibire^ts1/VT50658) post-training (One-factor ANOVA with Tukey post-hoc test; n≥6). (c) The GFP signal in α’/β’ap neurons was substantially reduced in trained-starved flies compared to both fed-trained flies and untrained controls (Two-sided Mann-Whitney tests; n≥19). (d) Trained-starved flies demonstrated an increase in α’/β’m activity compared to both fed flies and untrained controls. A significant decrease in calcium/GFP was also observed in trained-fed flies compared to untrained fed flies (Two-sided Mann-Whitney tests; n≥14). (e) Thermogenetic activation of α’/β’ap neurons (UAS-TrpA1/R35B12) resulted in a substantial gain in sleep while sleep was reduced significantly when α’/β’m neurons (UAS-TrpA1/R26E01) were activated (One-factor ANOVA with Tukey post-hoc test; n=32). (f) UAS-shibire^ts1/R35B12 flies showed an enhancement in sleep post-training at permissive but not at restrictive settings. Sleep measurements at restrictive settings are shown in (g) (Two-sided t-tests were performed for each genotype to compare trained and untrained groups, followed by Bonferroni correction; n≥32). (h) In trained fed flies, CaLexA-based neuronal activity in α’/β’ap neurons was substantially reduced in sleep-deprived flies compared to controls (Two-sided Mann-Whitney test; n≥11). In (c, d, h), representative images are shown, two independent experiments; Scale bar, 50 μm. Data are represented as mean ± s.e.m. Each data point in a memory experiment represents a group of flies and in a CaLexA imaging and sleep experiment it depicts a single fly. Precise ‘n’ and ‘p’ values are in the Source data. ***p<0.001; **p<0.01; *p<0.05. Asterisks in (a, b, e) indicate a significant difference between experimental and genetic controls.

Fig. 3.. Feeding drives recruitment of different DANs and MBONs for appetitive memory formation

(a) Silencing MB-MP1 (UAS-shibire^ts1/MB320C), but not MB-MV1 (UAS-shibire^ts1/MB296B), neurons affects long-term memory in starved flies (One-factor ANOVA with Tukey post-hoc test; n≥6). (b) Neuronal activity in MB-MV1, but not in MB-MP1, DANs is required for long-term memory in flies fed post-training (One-factor ANOVA with Tukey post-hoc test; n≥7). (c) Trained and starved flies show impaired memory when MBON-γ1pedc (UAS-shibire^ts1/MB112C) neurons are blocked for 4h post-training but remain unaffected if MBON-γ2α’1 (UAS-shibire^ts1/MB077B) neurons are silenced (One-factor ANOVA with Tukey post-hoc test; n≥6). (d) Long-term memory was lower in fed flies in which MBON-γ2α’1 neurons were silenced post-training (One-factor ANOVA with Tukey post-hoc test; n≥6). Data are represented as mean ± s.e.m. Each data point represents a group of flies. Precise ‘n’ and ‘p’ values are in the Source data. Asterisks in (a-d) indicate a significant difference between experimental and genetic controls. (e) Fed flies form sleep-dependent memory which requires activity in α’/β’ap neurons in association with a circuit comprised of MB-MV1 DANs and MBON-γ2α’1. In contrast, α’/β’m neurons with MB-MP1 DANs and MBON-γ1pedc mediate sleep-independent long-term memory in starved flies.