Inducing Sleep by Remote Control Facilitates Memory Consolidation in Drosophila

See allHide authors and affiliations

Science  24 Jun 2011:
Vol. 332, Issue 6037, pp. 1571-1576
DOI: 10.1126/science.1202249


Sleep is believed to play an important role in memory consolidation. We induced sleep on demand by expressing the temperature-gated nonspecific cation channel Transient receptor potential cation channel (UAS-TrpA1) in neurons, including those with projections to the dorsal fan-shaped body (FB). When the temperature was raised to 31°C, flies entered a quiescent state that meets the criteria for identifying sleep. When sleep was induced for 4 hours after a massed-training protocol for courtship conditioning that is not capable of inducing long-term memory (LTM) by itself, flies develop an LTM. Activating the dorsal FB in the absence of sleep did not result in the formation of LTM after massed training.

Although the functions of sleep remain unknown, sleep is believed to be important for maintaining optimal performance in a large and diverse number of biological processes (1, 2). Historically, the importance of sleep has been most convincingly established by demonstrating negative consequences that accrue in its absence (3). In contrast, methods that allow an experimenter to induce sleep on demand are lacking. Thus, it has been difficult to demonstrate that sleep serves a beneficial role per se. Studies in humans indicate sleep may play an active role in the strengthening or stabilizing of new memories (4, 5). With this in mind, we conducted experiments in Drosophila to determine whether it is possible to induce sleep on demand and thereby influence memory (6).

We expressed the bacterial sodium channel NaChBac (7, 8) under the control of several GAL4 (galactosidase 4) drivers to constitutively activate specific neuronal circuits. Flies that displayed periods of quiescence lasting ≥5 min were further evaluated to determine whether they met the traditional criterion for identifying sleep (9, 10). Quiescence was dramatically increased when UAS-NaChBac (NaChBac; UAS indicates upstream activating sequence) was expressed under the control of three independent GAL4 drivers: C5-GAL4, 104y-GAL4, and C205-GAL4 (Fig. 1, A to C, and fig. S1) (11). The expression pattern for each of these GAL4 drivers overlaps in neurons that strongly resemble ExFl2 cells, which are known to project to the dorsal Fan-Shaped Body (FB) (Fig. 1C and fig. S2) (11). In contrast, the expression pattern of these three drivers did not show prominent overlap in neurons in other brain regions (fig. S2) (12). Thus, although we cannot formally exclude a role for brain regions outside the dorsal FB, it is likely that the dorsal FB plays a role in regulating quiescence.

Fig. 1

Chronic sleep induction. (A and B) Expressing UAS-NaChBac using C5-GAL4 or 104y-GAL4 increased sleep. Data presented as sleep in min/hour (n = 12 to 16 per group). Error bars indicate SEM; ZT, zeitgeber time. (C) C5/+>UAS-CD8::GFP and 104y/+>UAS-CD8::GFP expression. (D) Total sleep was increased in C5/+>UAS-NaChBac/+ and 104y/+>UAS-NaChBac/+ flies; *P < 0.05, modified Bonferroni test. (E and F) Neuronal activation with C5 and 104y increased daytime and nighttime sleep bout duration. (G) Intensity of waking activity was increased in C5/+>UAS-NaChBac/+ and 104y/+>UAS-NaChBac/+ flies. (H) A light pulse during the night (gray) awakened all genotypes compared with undisturbed siblings. χ2, n = 23 to 29 each group, *P < 0.05; Mann-Whitney U test comparing each group to untreated siblings. (I) All genotypes exhibited a sleep rebound after 12 hours of sleep deprivation. (J and K) Expressing UAS-NaChBac using NP6510-GAL4 or C232-GAL4 did not change sleep. (L) NP6510/+>UAS-CD8::GFP/+ labeled the ventral FB, whereas C232/+>UAS-CD8::GFP/+ was expressed in the EB.

Because expressing UAS-NaChBac with C5-GAL4 and 104y-GAL4 resulted in the largest changes in both quiescence and in the consolidation of quiescent episodes, we focused on these two drivers (Fig. 1, D to F, and fig. S1B). Both C5/+>NaChBac/+ and 104y/+>NaChBac/+ flies behaved normally and spontaneously aroused to engage in goal-directed behaviors. Moreover, the number of quiescent bouts did not differ between experimental lines and both of their parental controls, suggesting that the quiescent bouts are regulated (fig. S3). To determine whether the activation of the FB inhibited locomotion, we examined the intensity of waking activity. C5/+>NaChBac/+ and 104y/+>NaChBac/+ flies exhibited an increase in waking activity compared with controls (Fig. 1G). To determine whether quiescent episodes were rapidly reversible, we exposed flies to a light pulse during the night; all genotypes exhibited a significant increase in the percentage of arousals (Fig. 1H). Similarly, a shorter light pulse was less effective in inducing wakefulness, indicating that quiescent episodes are associated with increased arousal thresholds (fig. S4). Moreover, all genotypes showed a rebound in quiescent episodes when kept awake for 12 hours, indicating the presence of homeostatic regulation. (Fig. 1I) (13). Lastly, we examined quiescence in response to caffeine and during the first days of adulthood to determine whether it would exhibit properties previously described for sleep. We found that caffeine disrupted quiescence in all genotypes (fig. S5). Moreover, the ontogenetic changes previously described for sleep were also observed for all genotypes (fig. S6). Thus, the observed quiescence episodes meet the historical criteria for sleep.

The central complex, comprising the FB and Ellipsoid Body (EB), is associated with locomotion and memory (12, 14, 15). To determine whether exciting neurons in the ventral FB or EB would also alter sleep, we expressed NaChBac in the ventral FB by using NP6510-GAL4 and NP6561-GAL4 and in the EB by using C232-GAL4 and 78y-GAL4. No sleep changes were observed (Fig. 1, J to L, and fig. S7). To test whether previously identified sleep-regulatory genes or known FB neuroanatomical pathways overlap with 104y- and C5-GAL4–expressing neurons, we conducted immunohistochemical and genetic experiments (figs. S8 to S12) (14, 1620). These experiments failed to identify candidate sleep-regulatory genes.

The ability to acutely control the timing and duration of sleep induction may be critical for elucidating the interaction between sleep and other processes such as memory formation. Therefore, we asked whether the transient receptor potential constructs TrpVr1 and TrpA1 could be used to induce sleep on demand (21, 22). Although 104y/+>UAS-TrpVr1/+ flies showed an increase in sleep after being fed 100 mM capsaicin, the effect was not immediate (Fig. 2A). However, when the temperature-sensitive TrpA1 allele (21) was expressed with C5 and 104y-GAL4, sleep and sleep consolidation were significantly increased when flies were moved to 31°C (Fig. 2, B to E). Although overall motor activity temporarily decreased during sleep induction (fig. S13), the intensity of waking activity was also increased, indicating that the acute activation did not impair locomotion (Fig. 2F). Moreover, exposing sleeping flies to increasing amounts of mechanical perturbation induced greater amounts of wakefulness, indicating the presence of increased arousal thresholds (Fig. 2G). To determine whether induced sleep could be antagonized by caffeine, 104y/+>UAS-TrpA1/+ flies were fed caffeine for 1 hour before being placed at 31°C; caffeine attenuated the increase in sleep (Fig. 2H). Lastly, inducing sleep in 104y/+>UAS-TrpA1/+ flies for 4 hours transcriptionally down-regulates the wake-associated genes fatty acid synthase (8) and homer (23), suggesting shared molecular characteristics with spontaneous sleep (fig. S14).

Fig. 2

Inducing sleep on demand. (A) Capsaicin-fed 104y/+>UAS-TrpVr1/+ flies increased sleep. (B to D) Transferring C5/+>UAS-TrpA1/+ and 104y/+>UAS-TrpA1/+ flies to 31°C for 6 hours increased sleep. *P < 0.05 modified Bonferroni test. (E) Mean sleep bout duration was increased in C5/+>UAS-TrpA1/+ and 104y/+>UAS-TrpA1/+ flies at 31°C. (F) Counts per waking minute increased at 31°C in C5/+>UAS-TrpA1/+ and 104y/+>UAS-TrpA1/+ flies. (G) Awakenings increased with perturbation number. SNAP, sleep-nullifying apparatus. (H) 104y/+>UAS-TrpA1/+ flies fed caffeine (2.5 mg/ml or 7.5 mg/ml) 1 hour before 31°C exposure display weakened sleep induction.

Recent studies have used sleep deprivation protocols to examine the role for sleep in synaptic downscaling and the consolidation of long-term memory (LTM) (20, 2327). Can inducing sleep on demand provide additional insight into the role for sleep in these processes? First, we evaluated the synaptic homeostasis model, which hypothesizes that synaptic connections increase during waking and are downscaled during sleep; in the absence of sleep, neuronal circuits would exceed available space and/or saturate, thereby interfering with individual’s ability to learn (28). Because housing flies in a complex social environment for 5 days increases the number of synaptic terminals in projections from the large ventral lateral neurons (LNVs) (20), we asked whether social enrichment would impair LTM. We used pigment-dispersing factor (pdf)–GAL4 to drive expression of a green fluorescent protein (GFP)–tagged construct of the postsynaptic protein discs-large (UAS-dlgWT-gfp) in flies exposed to increasingly larger social groups. The number of dlg-positive terminals increased with the size of the social group (Fig. 3, A and B). To determine whether the saturation of dlg-positive terminals had functional consequences, male Canton-s (Cs) flies were socially enriched in a group of 90 flies and then exposed to spaced training for courtship conditioning (Fig. 3C). Cs flies that had been maintained in social isolation displayed normal LTM (Fig. 3D). However, enriched flies had no LTM when trained either immediately after enrichment or after 1 day of delay.

Fig. 3

Inducing sleep reverses deficits in LTM after social enrichment. (A) Quantification of dlg-positive terminals expressed as percent of isolated siblings (n = 13 to 16 brains per group). (B) Images of dlgGFP-positive terminals from LNVs in isolated flies and enriched siblings (~95 flies). (C) Schematic for evaluating LTM after social enrichment. (D) Isolated (ISO) flies reduced courtship 48 hours after training (Student’s t test P = 0.008, n = 40 to 42). Enriched siblings lacked LTM (ENR+0 days, Student’s t test P = 0.77, n = 28 or 29). Enriched flies that were isolated for 1 day before training showed no LTM (ENR+1 day, Student’s t test P = 0.61, n = 25 to 29). Enriched flies that were isolated for 3 days exhibited LTM (ENR+3 days, Student’s t test P = 0.029, n = 12 or 13). (E) Schematic for testing LTM recovery after sleep induction. (F) Isolated males displayed LTM. (G) Enriched males showed no LTM when housed at 25°C. (H) 104y/+>UAS-TrpA1/+ flies that were enriched, housed at 31°C for 4 hours, then trained exhibited LTM; *P < 0.05 planned comparisons. Error bars indicate SEM.

The impaired LTM was not attributable to overcrowding: Enrichment using only 45 flies also disrupted LTM, and flies enriched in constant darkness retained LTM (fig. S15). Consistent with our previous results showing that dlg-positive terminals return to baseline after 48 hours (20), LTM was observed in flies trained 3 days after enrichment (Fig. 3D, right).

According to the synaptic homeostasis model, sleep downscales synapses and restores optimal learning. Thus, we hypothesized that inducing sleep immediately after social enrichment should restore LTM more quickly. Flies were housed in either social isolation or in a socially enriched group of ~95 flies. After 5 days, flies were placed into individual tubes at either 25°C or 31°C for 4 hours. All flies were returned to 25°C and then trained for courtship conditioning the following morning; LTM was tested 2 days after training (Fig. 3E). All isolated flies exhibited LTM (Fig. 3F). No LTM was observed in any genotype when training was preceded by 5 days of enrichment and flies were maintained at 25°C (Fig. 3G). However, LTM was observed in 104y/+>UAS-TrpA1/+ flies when sleep was induced for 4 hours immediately after social enrichment (Fig. 3H). Reduced courtship cannot be attributed to nonspecific effects of FB activation or heat exposure because courtship is not altered in naïve104y/+>UAS-TrpA1/+ flies or controls exposed to 31°C for 4 hours (Fig. 3G-H). Inducing sleep for 4 hours after enrichment not only accelerated LTM recovery but also reduced PDF-positive terminals in 104y/+>UAS-TrpA1/+ flies within 24 hours (fig. S16). Thus, our data are consistent with the synaptic homeostasis model and support a positive role for sleep in reversing LTM deficits induced by social enrichment.

To determine whether sleep plays a positive role in memory consolidation, we induced sleep for 4 hours immediately after a massed training protocol that does not result in the formation of LTM (Fig. 4A). Massed training did not induce LTM in any genotype at 25°C (Fig. 4, B, D, and F). Moreover, activating the EB did not alter LTM, indicating that neither heat exposure nor activating portions of the central complex that do not alter sleep facilitates memory consolidation (Fig. 4C). However, when sleep was induced after massed training by switching flies to 31°C for 4 hours, C5/+>UAS-TrpA1/+ and 104y-GAL4 flies displayed an LTM (Fig. 4, E and G). Thus, when flies were maintained at 25°C after massed training, no LTM was observed in any genotype. However, when sleep was induced for 4 hours, flies display LTM, even though the training protocol is not sufficient to induce memory consolidation by itself.

Fig. 4

Massed-training generates LTM when accompanied by sleep induction. (A) Schematic of experimental design. SD, sleep deprived. (B, D, and F) No LTM was detected in flies at 25°C. Planned comparisons between naïve and trained siblings found no significant differences. CI, confidence interval. (C) Activating the EB did not alter LTM. (E and G) Placing C5/+>UAS-TrpA1/+ and 104y/+>TrpA1/+ flies at 31°C for 4 hours after massed-training induced LTM. (H) Mechanical perturbation prevents sleep in 104y/+>UAS-TrpA1/+ flies at 31°C. (I) 104y/+>UAS-TrpA1/+ males sleep deprived at 31°C after massed training show no LTM (Student’s t test P = 0.54, n = 13 to 15).

An alternative interpretation is that activating neuronal circuits enhances memory consolidation and that LTM would be observed even in the absence of sleep. Therefore, we sleep-deprived 104y/+>UAS-TrpA1/+ flies while they were maintained at 31°C after training. Our sleep deprivation apparatus efficiently kept 104y/+>UAS-TrpA1/+ flies awake, further emphasizing that neuronal activation does not simply result in a paralyzed or comatose-like state (Fig. 4H). Additionally, activation of neuronal activation in the absence of sleep does not result in the formation of LTM after massed training.

We identified a circuit that plays a role in sleep regulation and can be activated on demand to precisely control the timing and duration of sleep. Our data complements previous sleep-deprivation experiments by demonstrating that sleep plays a positive role in both synaptic homeostasis and memory consolidation. Thus, inducing sleep facilitates the formation of LTM even for a training protocol that, by itself, is only able to induce short-term memory. We expect that the ability to acutely control the timing and duration of sleep induction will be critical for elucidating the interaction between sleep and other complex processes and ultimately the function of sleep.

Supporting Online Material

Materials and Methods

Figs. S1 to S16

Table S1

References (2934)

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Acknowledgments: We thank P. Taghert, E. Herzog, P. Gray, L. Seugnet, A. Diantonio, W. Vanderheyden, and B. van Swinderen for discussions. This work was supported by NIH grants R01-NS051305-01A1, NS057105, T32GM008151, and 5F31NS063514-02.
View Abstract

Stay Connected to Science

Navigate This Article