Clock and cycle limit starvation-induced sleep loss in Drosophila

Summary

Neural systems controlling the vital functions of sleep and feeding in mammals are tightly inter-connected: sleep deprivation promotes feeding, while starvation suppresses sleep. Here we show that starvation in Drosophila potently suppresses sleep suggesting that these two homeostatically regulated behaviors are also integrated in flies. The sleep suppressing effect of starvation is independent of the mushroom bodies, a previously identified sleep locus in the fly brain, and therefore is regulated by distinct neural circuitry. The circadian clock genes Clock (Clk) and cycle (cyc) are critical for proper sleep suppression during starvation. However, the sleep suppression is independent of light cues and of circadian rhythms because starved period mutants sleep like wild type flies. By selectively targeting subpopulations of Clk-expressing neurons we localize the observed sleep phenotype to the dorsally located circadian neurons. These findings show that Clk and cyc act during starvation to modulate the conflict of whether flies sleep or search for food.

Results and Discussion

Studies in mammals suggest that the homeostatic regulation of feeding and sleep involves functionally interconnected mechanisms [1]. This overlap is evident at the clinical, physiological, cellular and molecular levels. Longitudinal studies in humans reveal a strong correlation between sleep patterns and Body Mass Index, leptin and ghrelin levels [2]. In mammals, long-term sleep deprivation stimulates appetite [3], while food deprivation suppresses sleep [4]. Furthermore, the hypothalamic neuropeptides orexin/hypocretin and neuropeptide Y promote both food intake and wakefulness [5, 6].

Fruit flies exhibit all the behavioral hallmarks of sleep including prolonged periods of behavioral quiescence, a species-specific postural change, increased arousal threshold and rebound following sleep-deprivation [7, 8]. To determine whether flies suppress sleep during starvation, we measured their locomotor activity in the Drosophila Activity Monitor System (DAMS) and analyzed sleep as previously described [9]. Using a 3-day protocol we monitored changes in sleep during 24 hours of food deprivation (Figure 1A). We found that similar to mammals, both male (Figure 1B and C) and female (Figure 1D) flies dramatically suppressed their sleep during starvation compared to ad libitum fed control flies. Determining the percentage change in sleep within individual animals revealed that starved flies robustly suppressed sleep compared to baseline control levels (Figure 1E). Flies housed with sucrose did not suppress their sleep indicating that caloric intake with no amino acids is sufficient to support normal levels of sleep. Flies housed with the non-metabolizable sweetner sucralose suppressed sleep (Figure S1A and B) confirming that the effects of starvation on sleep are due to a caloric deficit rather than sensory systems detecting the absence of food in an environment. Sleep suppression during starvation is a generalizable phenomenon in Drosophilidae as we observed it in multiple D. melanogaster laboratory lines and Drosophila species when tested using the same assay (Figure S1C and D).

Figure 1. Food deprivation suppresses sleep.

A. Paradigm for measuring the effects of starvation on sleep. Experiments took place over a 3-day period that began 1 day following the loading of flies into standard DAMS tubes. Baseline (Day 1): all groups were kept on food. Experiment (Day 2): flies were transferred to fresh food (control, yellow), 150mM sucrose (red), or agar (food deprivation; blue). Recovery (Day 3): All groups were transferred to fresh food. B. Sleep profile of male flies reveals that starvation (agar group, blue line) suppresses sleep compared to food (black line) and sucrose (red line) fed flies during the experimental day (grey bar). Days on food (1 and 3) are depicted by yellow bars. White/black bars indicate lights on/lights off. C, D. Quantification of day (white bars) and night (black bars) sleep over 24hrs reveals decreased sleep during the night in male (C) and female (D) flies in the agar group compared to food (male, N≥49; for all groups; P<0.001: female, N≥42; for all groups: P<0.001) and sucrose-fed (male: P<0.001; female: P<0.001) fed controls. Male flies fed agar did not differ in daytime sleep compared to baseline (P>0.146) while the female agar group suppressed sleep during the day (P<0.001). Sucrose is sufficient to support normal sleep in daytime and nighttime (male, P>0.35; female, P>0.084). E. Percentage change from baseline sleep reveals that male and female flies on agar suppress sleep compared the food and sucrose groups (P<0.001, all groups). F. Starvation starting at ZT0 (N≥26) suppresses sleep during the following night (ZT12–24; P<0.0001), while flies starved at ZT12 suppress sleep during the following day (ZT0–12; P<0.001). In both cases, no statistical difference is observed for hours 0–12 of starvation (P>0.171, P>0.089). G. Male flies were tracked in square arenas (N≥31; left) or tubes (N≥25; right). Daytime and nighttime sleep was significantly suppressed in the agar groups (square arena, P<0.01; tube, P<0.01). Sucralose fed flies in tubes suppressed both daytime and nighttime sleep compared to food and sucrose fed controls (P<0.001). Daytime sleep (ZT0–12) did not differ between the agar and sucralose groups (P>0.63). Asterisk denotes significant difference (P<0.01, ANOVA) from control groups. Data are mean ± SEM. See also Figure S1.

Since starvation has been reported to induce hyperactivity [10] we examined whether increased activity during starvation could account for the observed lack of sleep. We found that food deprivation both suppresses sleep and induces hyperactivity (Figure S1E and F). We also asked whether flies compensated for lost sleep during food deprivation with a homeostatic rebound by increasing their sleep on recover. Analysis in the DAMS and visual inspection revealed male and female flies rebound in the 4 hours following food-deprivation (data not shown).

We speculate that changes in sleep and activity result from the initiation of foraging behavior. A polymorphism in the foraging (for) cGMP-dependent protein kinase (PKG) gene is one determinant of foraging strategy in both larvae and adult Drosophila [11, 12]. We found that for^rover flies enhanced sleep suppression compared to the for^sitter, strain (Figure S1G) supporting the notion that flies suppress sleep during starvation in order to search for food.

Sleep profiles revealed that male flies sleep relatively normally during the initial 10–12hr of starvation (Figure 1B). To test if the lag in sleep suppression is due to light cues or to the metabolic effects of starvation, we shifted the start time of the 24hr food deprivation period to ZT12 (end of lights-on) (Figure 1F). No significant differences in sleep during the initial 12hrs of starvation were apparent between the ZT0 and ZT12 groups and fed control flies. However, sleep was significantly suppressed over hours 12–24 of starvation in both the ZT0 and ZT12 groups. These results indicate sleep suppression is independent of light cues and is likely caused by the amount of time starved.

We repeated our experiments using the recently developed video tracking software pySolo [13]. We assayed male flies in the same tubes used in DAMS and in 1” × 1” square × 0.5” deep arenas containing either agar or food. Male flies also robustly suppressed their sleep upon starvation in these experiments (Figure 1G). Therefore, the sleep suppressing effects of starvation can be measured using independent analysis, are generalizable to a second environment and are not an artifact of the activity monitoring system.

We next asked which neural populations mediate this behavior. The mushroom bodies are involved in regulation of sleep and locomotor activity [9, 14, 15]. They are also a site for the integration of multimodal sensory information and for the gating of behavioral responses involving olfactory memory and complex visually guided behavior [16–18], and thus are a candidate for a neural locus regulating sleep suppression during starvation. We chemically ablated the mushroom bodies with hydroxyurea (HU) treatment (Figure 2A) to test their role in promoting wakefulness upon starvation. Consistent with previous findings [9, 14, 15] HU treated (+HU) male and female flies are more active and sleep less than untreated control flies (−HU) (Figure 2B and C). However, both +HU and −HU flies slept less when housed with agar than with food (Figure 2C and D) demonstrating that flies with or without mushroom bodies suppress sleep equally in response to starvation. These data suggest that while the mushroom bodies promote sleep, they are not involved in starvation-induced sleep suppression.

Figure 2. Mushroom bodies are dispensable for sleep-feeding interactions.

A. Hydroxyurea ablation of the mushroom bodies. Anti-FAS-II staining labels the mushroom bodies and central complex of a representative brain from an untreated –HU fly (left). While the central complex is intact, the mushroom bodies are not detectable in brains from +HU flies (right). B. Total 24hr activity reveals that both −HU and +HU flies are more active on agar than on food (N≥33 for all groups; Male, −HU; P<0.001, +HU; P<0.004: Female, −HU; P<0.002, +HU; P<0.001). −HU flies are also less active than +HU flies (Male; P<0.001; Female; P<0.001). C. Percentage of sleep per 24hr reveals that both −HU and +HU flies suppress sleep on agar compared to flies on food (Male, −HU; P<0.001, +HU; P<0.001: Female, −HU; P<0.001, +HU; P<0.003). D. Suppression of sleep calculated as mean percentage change/fly reveals that in both male and female flies sleep suppression during starvation is not significantly affected by HU treatment (Male; P>0.84; Female; P>0.54). E. Schematic of the heat and feeding protocol for silencing the mushroom body with Shi^TS1. Flies in the agar group are starved beginning at ZT18, 6hr prior to temperature shift. F. Flies with silenced mushroom body neurons suppress sleep on agar at 30°C compared to flies on food at 30°C. Data depict the percentage of time spent sleeping from ZT0–ZT12 (N≥16 for all groups; Shi^TS1/+ =P<0.006; H24= P<0.001; OK107=P<0.01; c747; P<0.01). Asterisk denotes significant difference (P<0.01, ANOVA) from control groups. Data are mean ± SEM.

To further test for a role of the mushroom bodies in starvation-induced suppression of sleep we used a neural silencing method and expressed a dominant-negative temperature sensitive UAS-Shibire^TS1 (Shi^TS1) in the mushroom bodies. Expressing Shi^TS1 in mushroom body neurons reversibly blocks their neurotransmission above the restrictive temperature of 29°C [19]. Since flies did not tolerate 24hrs of starvation at 30°C, we developed a more restricted starvation protocol to allow us to silence neurons during starvation. Following 18hr of baseline activity measurement, flies were transferred to agar tubes for food deprivation (ZT18) and maintained at 22°C through ZT24/0 (Figure 2E). At ZT24/0 the temperature was increased to 30°C and sleep was recorded through ZT12. To determine the effects of neural silencing on sleep, daytime sleep (ZT0–12) of male flies on agar was compared to controls of the same genotype kept on food, and to the baseline recording at the permissive temperature (22°C). Starved flies expressing UAS-Shi^TS1 under control of the mushroom body GAL4 drivers OK107, c747, and H24 all suppressed sleep compared to fed flies of the same genotype (Figure 2F) confirming that the mushroom bodies are dispensable for suppression of sleep during starvation.

In an effort to examine the genetic components modulating sleep during starvation, we identified the circadian rhythm defective cyc⁰, Clk^Jrk, and Clk^ar mutant fly strains exhibited enhanced suppression of sleep when starved (Figure 3A and B). Importantly, while the period⁰¹ (per⁰¹), Pigment dispersing factor (Pdf⁰¹), Clk^Jrk, Clk^ar and cyc⁰ mutant fly strains are all arrhythmic in constant darkness, per and Pdf mutant flies suppressed their sleep similar to wild type flies, suggesting that the sleep phenotypes observed in Clk and cyc mutants can be differentiated from defects in circadian behavior (Figure S2A and B). We were unable to test timeless⁰¹ mutants due to high levels of lethality during the 24hr starvation protocol. To confirm that the enhanced sleep suppression observed in Clk and cyc mutants is not due to hyperactivity we analyzed waking activity in Clk^Jrk, Clk^ar, and cyc⁰ mutants and found that all mutants tested have waking activity that is comparable or less than wild type flies during both fed and starved states (Table S1). In agreement with previous reports of circadian phenotypes, we found the sleep phenotype of the Clk^Jrk allele to be dominant while Clk^ar and cyc⁰ phenotypes were recessive [20, 21] (Fig 3A and B). Because Clk and cyc encode transcription factors that heterodimerize to activate the circadian genes per and tim [22] we tested Clk^ar/+,cyc⁰/+ trans-heterozygous flies for sleep suppression. We found that Clk^ar/+,cyc⁰/+ flies suppressed sleep to a greater degree than Clk^ar/+ and cyc⁰/+ suggesting these genes functionally interact to promote sleep during starvation (Figure 3A and B). These data are consistent with a novel role for Clk and cyc in promoting sleep during starvation that is independent of their effect on circadian regulation.

Figure 3. Mutations in Clk and cyc enhance sleep-suppression during starvation.

A. Male flies with mutations in Clk and cyc were tested for sleep in tubes containing food or agar. In all groups tested, flies slept less on agar than on food (N≥41; P<0.001 for all groups). B. Calculating percentage change in sleep reveals Clk^ar, Clk^Jrk, and cyc⁰ flies significantly enhanced sleep suppression during starvation compared to wild type controls (P<0.001 for all groups). Sleep suppression in flies heterozygous for Clk^Jrk (Clk^Jrk/+) does not differ from Clk^Jrk homozygous flies (P>0.38) and is greater than wild type controls (P<0.001). Clk^ar/+,cyc⁰/+ heterozygous flies enhance sleep suppression compared to wild type and Clk^ar/+, and cyc⁰/+ heterozygous flies (P<0.01 for all groups). C. The short-sleep mutants fmn and Sh^MNS suppress sleep comparably to wild type flies (P>0.75). D. Whole-brain confocal images of Pdf-GAL4 and cry-Gal4 driving UAS-mCD8:GFP reveals the expression pattern of each driver. E. Male flies expressing UAS-ClkDN with tim-GAL4 and cry-GAL4 show enhanced sleep suppression when starved. All groups tested significantly suppress sleep on agar compared to food (N≥28; P<0.001 for both groups). F. Analyzing data in panel D as percentage of sleep suppression between flies on food or agar reveals that tim-GAL4;Pdf-GAL80;UAS-ClkDN flies suppress sleep significantly more than wild type flies and those harboring UAS-ClkDN or GAL4 transgenes alone (P<0.001 for all groups). Asterisk denotes significant difference (P<0.01, ANOVA) from control groups. Data are mean ± SEM. See also Figure S2 and Table S1.

Clk and cyc have previously been implicated in promoting sleep and flies mutant for these genes are short sleepers [23]. To test if enhanced suppression of sleep during starvation is a general property of short sleeping flies we assayed sleep suppression in the short sleeping mutants fumin (fmn) and shaker^minisleep (Sh^MNS). We were unable to test sleepless mutant flies due to lethality during the 24hr starvation protocol. While Clk and cyc mutants enhance sleep suppression during starvation, both fmn and Sh^MNS suppressed sleep similar to wild type controls (Figure 3C) suggesting that the enhanced sleep suppression observed in Clk and cyc mutants is not due to their short-sleep phenotype.

We next sought to localize Clk and cyc function in starvation-induced sleep suppression. For this, we used a previously described dominant-negative Clk transgene (ClkDN) lacking its basic DNA-binding domain but retaining its protein interaction domains [24]. Disrupting Clk function in perhiphal tissues, sensory neurons, and the glucagon-like AKH neurons had no effect on response to starvation (Figure S2C and D) suggesting Clk function in promoting sleep during starvation is not in peripheral sensory neurons.

Within the fly brain Clk is expressed in circadian oscillator cells, some of which are the small and large ventrolateral neurons (LNvs) that express the neuropeptide Pigment dispersing factor (Pdf). Clk is also expressed in dorsally located circadian neurons and peripheral cells that include populations of dorsal neurons (DNs) and dorsally located LNs (LNds) [25]. Specifically disrupting Clk function in the LNvs abolishes circadian rhythm [24] suggesting that the LNvs are critical for pacemaker function. We first ectopically expressed a dominant negative Clk (ClkDN) isoform in either all circadian neurons or selectively in subpopulations of Clk-expressing cells [24]. Disrupting Clk function in all circadian neurons with tim-GAL4 (tim-GAL4; UAS-ClkDN) enhanced sleep-suppression during starvation compared to flies harboring either tim-GAL4 or UAS-ClkDN single transgenes alone (Figure 3 E and F).

To further refine the population of tim-expressing neurons regulating sleep during starvation, we used GAL4 and GAL80 in combinations to express ClkDN in subpopulations of Clk neurons. Flies with disrupted Clk function in all tim-expressing cells except PDF neurons (tim-GAL4, Pdf-GAL80;UAS-ClkDN) showed enhanced sleep suppression suggesting the primary pacemaker neurons do not modulate sleep during food deprivation. Fortifying this conclusion we found that expressing ClkDN only in LNvs (Pdf-Gal4; UAS-ClkDN) did not affect suppression of sleep (Figure 3 E and F). The cry-GAL4 driver labels the large and small LNvs, the DN1 subpopulation of DNs and the LNds ([26] and Figure 3D). cry-GAL4;UAS-ClkDN flies enhanced sleep suppression suggesting that either a single PDF negative small-LNvs and/or the LNds and/or the DN1s, which are all cry+, Pdf-, promote sleep during starvation. Supporting this notion, expression of ClkDN in all tim cells except those expressing cry (tim-GAL4,cry-GAL80) [27] does not enhance sleep suppression. Taken together, these results suggest a population of dorsally located Clk-expressing neurons promote sleep during starvation. Therefore, we have identified a novel mechanism through which circadian and feeding systems modulate sleep.

To support this anatomical localization of Clk function we selectively silenced or activated different populations of Clk-expressing neurons in adulthood. Expressing UAS-Shi^TS1 in all clock cells with tim-GAL4 or Clk-GAL4 enhanced sleep suppression while selectively silencing LNvs using Pdf-GAL4 did not affect sleep during starvation (Figure 4A, B and C) fortifying the conclusion that a population of Clk-expressing Pdf negative neurons promote sleep during starvation.

Figure 4. Clk/ cyc expressing neurons acutely regulate sleep during starvation.

A. Schematic of the restricted temperature shift and feeding protocol for Shi^TS1 manipulations. Flies on agar were food deprived for 12hrs during the temperature shift (N≥22; ZT0–ZT12). B. Blocking transmission from tim or Clk-GAL4 neurons suppressed sleep in starved flies. All flies harbor UAS-Shi^TS1. Control (UAS-ShiTS1/+) and Pdf-GAL4;UAS-Shi^TS1 flies do not suppress sleep at 30°C in the restricted starvation protocol (P>0.11; P>0.79), while tim-GAL4;UAS-Shi^TS1 and Clk-GAL4;UAS-Shi^TS1 flies sleep significantly less on agar than on food (P<0.001; P<0.001). C. Analyzing data as percentage change from fed flies at 22°C reveals that starved UAS-Shi^TS1/+ and Pdf-GAL4;UAS-Shi^TS1 did not differ from fed flies (P>0.62) while tim-GAL4;UAS-Shi^TS1 and Clk-GAL4;UAS-Shi^TS1 suppress sleep when starved (P>0.001) D. Schematic of the temperature shift and feeding protocol for dTrpA1 manipulations. Flies on agar were food deprived for 18 hrs beginning at ZT18, 6 hrs prior to the temperature shift (N≥23; ZT0–ZT12). E. Acute excitation of cry-GAL4 expressing neurons blocks the effects of starvation on sleep. All flies harbor UAS-dTrpA1. Control flies (UAS-dTrpA1/+) and Pdf-GAL4;UAS -dTrpA1 flies suppress sleep at 28°C (P<0.003; P<0.007), while cry-GAL4;UAS-dTrpA1 flies do not suppress sleep during starvation (P>0.71). F. Analyzing data as percentage change from fed flies at 22°C reveals starved UAS-TrpA1/+ and Pdf-GAL4;UAS-TrpA1 suppress sleep compared to fed controls at 30°C (P<0.001) while starved cry-GAL4;UAS-TrpA1 do not differ from fed controls (P>0.56) Asterisk denotes significant difference (P<0.01, ANOVA) from control groups. Data are mean ± SEM.

We therefore reasoned that activation of these neurons would cause starved flies to sleep as if they were fed. We ectopically expressed the high heat-sensitive cation channel dTrpA1 to activate Clk-expressing neurons with regional and temporal specificity [28]. Activation of all Clk neurons caused lethality in both fed and starved flies at 28°C and we therefore used the more restricted cry-GAL4 driver. Activation of cry-GAL4 expressing neurons during starvation abolished sleep suppression (Figure 4 D, E and F) confirming a role for these neurons in promoting sleep during starvation. Activating LNvs alone (Pdf-GAL4;UAS-dTrpA1) during starvation did not affect sleep suppression compared to wild type controls. Therefore, neurotransmission from cells other than the central pacemaker neurons act acutely during adulthood to promote wakefulness during starvation.

In addition to modulating circadian rhythms, core circadian genes and neurons have been implicated in numerous behaviors including cocaine sensitivity, feeding, courtship and memory [29–33]. We find the DN1s or LNds that express circadian genes and promote sleep during starvation. These neurons are functionally distinct from the LNvs that control behavioral rhythms during constant darkness. These findings are consistent with the existence of a neural mechanism mediating a behavioral conflict that determines whether a fly sleeps or seeks food.

We find that for kinase activity is positively correlated with sleep suppression [34] suggesting that for may counteract Clk and cyc function. In addition to for, feeding-related genes such as neuropeptide F (NPF) and metabolism related genes including Drosophila insulin-like peptides (DILPs), take out, and Drosophila p70/S6 Kinase have been implicated in control of feeding [35–37]. DILPs are expressed in the pars intercerebralis (PI), a brain region previously implicated in sleep regulation and starvation response [38, 39]. Future studies examining these genes and the role of the PI may advance our understanding of genes selectively modulating sleep during food deprivation.

Materials and Methods

Food deprivation experiments

For food deprivation experiments, 2–4 day old males or mated females (unless otherwise stated) were loaded into tubes containing standard brown for acclimation. Following one day of acclimation in DAMS tubes with standard fly food, baseline sleep was measured for 24hrs. Flies were then transferred at ZT0 (start of lights on, Day 2) to a tube containing either standard fly food (ad libitum control), agar supplemented with 150mM sucrose, 1mM sucralose or 1% agar, alone for 24hrs. Flies were then transferred back to food-containing vials, and activity was recorded for an additional 24hr recovery period (Day 3).Tubes were maintained in a 25°C incubator with 12:12 LD cycles. All data presented result from at least 2 independent experiments.

Sleep analysis

For all experiments except those using pySolo, sleep was analyzed with the excel-based ‘Sleep Counting Macro [9] generously supplied by R. Allada (Northwestern University). For video monitoring experiments we used pySolo analysis suite generously supplied by G. Gilestro and C. Chirelli (University of Wisconsin, Madison). The within fly percentage change in sleep was calculated as ((% sleep starved-% sleep baseline) / % sleep baseline) × 100.

Shibire^TS1 and dTrpA1 experiments

Flies were prepared as described for standard food deprivation experiments, except that flies were kept at 22°C for acclimation and for 24hrs of baseline recordings. For UAS-Shi^TS1 experiments flies were then transferred at ZT0 to tubes containing either food (fed group) or 1% agar (control group) and sleep was recorded for 12hrs with the incubator temperature at 30°C. For dTrpA1 experiments flies were transferred to agar at ZT18 and maintained at 22°C. The following day the temperature was increased to 28°C at ZT0 and the experiment proceeded through ZT12. Only the light phase (ZT0–12) was analyzed in both protocols.

Highlights

• Drosophila strike an optimal balance between sleep and food-seeking behavior

• The circadian genes Clk and cyc modulate sleep in response to metabolic needs

• Dorsally located Clk-expressing neurons promote sleep during food deprivation

• Clk regulates sleep and circadian rhythms through independent neural mechanisms.