Go Signaling in Mushroom Bodies Regulates Sleep in Drosophila

Drosophila Strains

All flies were cultured at 25°C and 60% relative humidity on cornmeal, sugar, yeast, and agar medium, under 12h: 12h light: dark (L/D) conditions. The genetic background of all strains is the w¹¹¹⁸ stain, except MB247, 7B-Gal4 and UAS-PKAmc* with y w background. The fly stains were kindly gifted as follows: UAS-Gi^GTP by Dr. Knoblich (Institute of Molecular Biotechnology, Vienna, Austria); UAS-Go^WT and UAS-Go^GTP by Dr. Tomlinson (Columbia University, New York, NY); GAD-GAL4 by Dr. Griffith (Brandeis University, Waltham, MA); p{MBswitch}, ElavGeneSwitch, c739, 201y, c772, and rut²⁰⁸⁰by Dr. Davis (Baylor College of Medicine, Houston, Texas); UAS-PTX, UAS-PKAmc*, and UAS-PKAr by Dr. Jan (University of California, San Francisco, CA); MB-GAL80, and UAS-CREB-2b by Dr. Waddell (University of Massachusetts Medical School, Worcester, MA); 30Y, 238Y, c309, and DDC-GAL4 by Dr. Sehgal (University of Pennsylvania, Philadelphia, PA); Pdf-GAL4 by Dr. Rosbash (Brandeis University, Waltham, MA); MB247 by Dr. Zars (University of Missouri-Columbia, Columbia, MO). UAS-Go RNAi (#19124), UAS-Drcr2 (#60008), and UAS-DopR RNAi (#7206) were obtained from the Vienna Drosophila RNAi Center at Vienna, Austria. Other flies were obtained from the Bloomington Stock Center at Indiana University, Bloomington, IN.

Sleep Assays

Female flies aged 3–7 days were transferred into monitor tubes (5 mm × 65 mm) containing 5% sucrose/2% agar. Locomotor activity was recorded by the Drosophila Activity Monitoring System (DAMS; Trikinetics, Waltham, MA). Sleep was defined as periods of 5 min without recorded activity. After 2 days of environmental adaptation in the monitors, data were collected for 4 days. Experiments were performed in an incubator at a temperature of 25 ± 1°C and a relative humidity of 60% ± 5%. Lights were turned on at ZT0 (circadian time 06:00) and off at ZT12 (circadian time 18:00).

Heat-Shock Treatment

GAL80^ts experimental and control groups were raised at 18°C. Three- to five-day old flies were monitored at 18°C for 4 days. Sleep from the third day and the fourth day was used to calculate baseline sleep levels, and then the flies were exposed to a 12-h 32°C heat-shock from ZT 12 on the fifth day to ZT 0 on the sixth day. After a 24-h recovery period, behavioral data from the seventh day and the eighth day were collected for analysis.

Drug Treatment

For the RU486 treatment, flies were maintained in food containing 500 μM RU486 (Sigma) as the treatment group, or in food containing vehicle alone as the control group during both the induction period (first 2 days) and the data collection period (the following 4 days). For hydroxyurea treatment, newly hatched larvae (< 1 h) were collected and fed with hydroxyurea (50 mg/mL) or vehicle for 4 hours. Animals were then washed with water 3 times and transferred to vials with fresh food. Adult flies tested in sleep assays were dissected to examine ablation of the MBs.

Mechanical Stimulation

Mechanical stimulation was applied at ZT 14 (2 h after lights off) when most flies showed quiescence. A 28 mm thick cushion pad was placed between the monitor and table. Monitors were dropped once by quickly removing the pad to simulate the moderate stimulation, and monitors were dropped consecutively 6 times as the strong stimulation. Locomotor activity was recorded continuously, in 30-s bins. The locomotion of flies having no recorded activity in the 5 min before the stimulus was examined. Only flies which started to move within 1 min after the stimulus and continued to move for at least 1 min were considered to respond positively. Otherwise, flies were considered to be nonresponsive.

Western Immunoblotting

The heads of adult flies were collected and homogenized in lysis buffer with a protease inhibitor cocktail (11873580001, Roche). After centrifuging, the supernatants were loaded and separated by SDS-PAGE gel and then transferred to PVDF membranes (0.45 mm, Millipore). The membranes were incubated with a mouse polyclonal antibody against rabbit Goα (4737-1010, AbD Serotec. 1:125) overnight at 4°C and HRP-conjugated goat-anti-rabbit lgG secondary antibody at a 1:5000 dilution (111-035-003, Jackson). We used the chemiluminescent HRP substrate (Millipore) to detect horseradish peroxidase and exposed membranes to Kodak MR film. After stripping the membranes, we reprobed them with rabbit anti-actin antibody (Santa Cruz: sc-1616-R 1:2500). Actin was used as a loading control.

Imaging

Brains were dissected from 3-day-old adult female flies expressing UAS-mCD8:GFP and fixed in 4% paraformaldehyde in phosphate buffer solution (PBS) (1.86 mM NaH₂PO₄, 8.41 mM Na₂HPO₄, and 175 mM NaCl) for 30 min and washed 3 times in PBS. Images were acquired as a z-stack using a leica TCS sp5confocal microscope.

Data Analysis

All statistical analysis was conducted using SPSS. The Wilks-Shapiro test was used to determine normality of data. Normally distributed data were analyzed with 2-tailed, unpaired Student t-tests or one-way analysis of variance (ANOVA) followed by the Tukey-Kramer HSD Test as the post hoc test. Nonparametrically distributed data were assessed using the Mann-Whitney U test. Data were presented as mean behavioral responses and error bars represent the standard error of the mean (SEM). Differences between groups were considered significant if the probability of error was less than 0.05 (P < 0.05).

Pan-Neuronal Activation of Go Signaling Induces Sleep Enhancement

To determine whether Go signaling is involved in sleep regulation, we utilized the conditional RU486-inducible pan-neuronal driver (ElavGeneSwitch)²⁷ to temporally express the wild-type form (Go^WT) or constitutively active form (Go^GTP)²⁸ of the Goα subunit in the adult nervous system. The Go^GTP flies (with pan-neuronal expression of Go^GTP) exhibited a robust sleep increase compared to controls with an identical genotype (Figure 1A). Similarly, flies overexpressing Go^WT also showed increased sleep, indicating that exogenous Go^WT can be activated by endogenous Go-coupled receptors. In contrast, flies treated with vehicle displayed no significant sleep increase compared to untreated parental controls (Figure 1A). These results suggest that pan-neuronal activation of Go signaling enhances fly sleep.

Next, we asked whether the sleep phenotype of the Go^GTP flies is simply a result of impaired locomotor activity. In line with previous reports,^9,15 we assessed 2 parameters—the peak activity (maximal activity in a 30-min period) and the waking activity (total activity divided by the length of the waking period)—as an index of locomotor activity. We found that although their periods of immobility were extended, the experimental flies still showed normal peak activity levels (see supplementary material, Figure S1A). In addition, the Go^GTP flies even exhibited elevated waking activity (Figure S1B), a phenomenon also reported in some other flies with a long-sleep phenotype.^9,29 Therefore, these data indicate that flies were not paralyzed and that sleep increment does not arise from hypoactivity.

In Drosophila, sleep is a physiological state associated with an increased arousal threshold,² so we investigated whether our long-sleep flies exhibit an increased arousal threshold during the sleep period. We measured the arousal level by calculating the percentage of flies responding to mechanical stimuli at ZT 14. Strong stimulation aroused the Go^GTP flies to the same extent (96%) as those without RU486 induction (98%), further indicating that expression of Go^GTP does not affect mobility. Compared to the 85% of vehicle-treated flies aroused by moderate stimulation, only 29% of the Go^GTP flies responded to the stimuli of same intensity (Figure 1B). The higher arousal thresholds found in the Go^GTP flies suggest that activation of Go signaling consolidates fly sleep.

We also analyzed sleep parameters. Go^GTP flies had prolonged average sleep episodes and decreased sleep bout numbers within one LD cycle (24 h) compared to controls (Figure 1C-D), suggesting that activation of Go signaling can help flies maintain sleep status by prolonging sleep time and reducing wake time.

The Sleep-Promoting Effect of Go^GTP Is Localized in Mushroom Bodies

After showing that pan-neuronal activation of Go signaling facilitated sleep, we next asked where the sleep-promoting effect of Go signaling occurs. To identify the potential brain areas, we used different GAL4 lines with region-specific expression patterns to drive Go^GTP expression. However, some developmental defects were observed, such as lethality, uncoordinated motility, and wing deformities. To avoid these defects, we introduced the temperature-sensitive GAL4 suppressor GAL80^ts, which inhibits transcriptional activator activity of GAL4 at permissive temperatures (18°C) and loses its suppression at restrictive temperatures (32°C).³⁰ We screened 25 GAL4 lines (see supplementary material, Table S1) whose expression covered major brain regions such as the circadian neurons, antenna lobes (AL) and central complex (CC), and regions recently identified in sleep regulation such as the pars intercerebralis (PIs),¹⁶ dopaminergic neurons,^7,31 GABAergic neurons,³² serotoninergic neurons⁸ and MBs.^14,33 Although GABAergic, serotoninergic and dopaminergic neurons are vital for sleep and arousal modulation,^7,8,31,32 heat-shock induction of Go^GTP with DDC-GAL4 (specifically expressed in serotoninergic and dopaminergic cells) or GAD-GAL4 (specifically expressed in GABAergic cells) had no obvious effect on sleep (Figure 2A), suggesting that these neurons may be dispensable for Go^GTP-induced sleep changes, and that Go^GTP may not function as a general negative modulator of neurotransmission.

Interestingly, we found that expression of Go^GTP with 2 MB-specific drivers (either MB247 or OK107) resulted in a significant sleep increment (Figure 2B), consistent with current opinion that the MB is the main brain region for sleep regulation.^14,33 To drive the expression of Go^GTP in MBs more conveniently, we used the RU486-inducible MB driver P{MBSwitch}³⁴ and observed that sleep was markedly increased in P{MBSwitch} > Go^GTP flies after RU486 induction (Figure 2C-D). In accordance with our pan-neuronal expression (ElavGeneSwitch) data, P{MBSwitch} > Go^GTP flies exhibited a prolonged sleep duration and a decline in sleep bout number (Figure S2A-B).

Go signaling is characterized as a classical inhibitory pathways,²¹ thus we reasoned that inhibiting the activity of P{MBSwitch}-labeled MBs could mimic the effect of Go signaling on sleep regulation. To test this possibility, we first used P{MBSwitch} to express 2 kinds of potassium channel—UAS-Kir and UAS-EKO, whose overexpression has been shown to reduce neural activity.^35,36 We observed that overexpression of Kir and EKO with P{MBSwitch} significantly increased sleep (Figure S3A), similar to a previous report.¹⁴ We next overexpressed DT-A (Diphtheria toxin A-chain), which has been used by others to efficiently ablate cells,³⁷ in a subset of MBs labeled by P{MBSwitch} (Figure S3B). Ablation of these neurons also resulted in sleep increase (Figure S3A), implying that activation of Go signaling might cause inhibition of neural function. Given that irreversible ablation of P{MBSwitch}-labeled MBs also promotes sleep, we wondered whether Go^GTP-induced sleep change is just a consequence of the toxic effect of Go^GTP. Thus we investigated whether the sleep increase caused by Go^GTP is reversible. After treating with RU486 for 3 days, when P{MBSwitch} > Go^GTP flies were transferred to tubes with regular food for another 3 days, their sleep reduced to normal levels (Figure S4). Therefore, our results imply that Go signaling in MBs dynamically regulates sleep in a reversible manner.

Inhibition of Endogenous Go Signaling in Mushroom Bodies Affects Sleep

To test whether endogenous Go signaling in MBs is required for sleep regulation, we used 2 kinds of approaches for inhibiting endogenous Go activity. First we expressed the S1 subunit of pertussis toxin (PTX), which was used as a selective inhibitor of Go signaling in fruit flies in a previous study.³⁸ Activation of Go is dependent on upstream G protein-coupled receptors (GPCRs), which catalyze the exchange of nucleotide binding to G protein. PTX-induced ADP-ribosylation of the G alpha subunit prevents the interaction of Go with GPCRs, leaving Go in an inactive state.³⁹ In vertebrates, PTX can ADP-ribosylate both Gi and Go protein, while in Drosophila, only the Goα protein contains the PTX recognition site and can be specifically inhibited by PTX.^21,38 Secondly, we used in vivo RNA interference (RNAi) to knock down the expression of Goα in targeted cells, and co-expressed Dcr-2 to enhance the effect of transgenic RNAi.⁴⁰ We found that expression of Go RNAi in all cells with Act5C-GAL4 produced severe lethality in embryos, consistent with the lethal phenotype in the G(o)α47A homozygous null alleles Go⁰⁰⁷ and Go⁰⁶¹¹ embryos.²⁸ Narrowing expression of Go RNAi to the nervous system using elav[c155]-Gal4 significantly reduced the viability of adults compared with controls, and surviving adults showed reduced levels of endogenous Go protein (Figure 3A).

Based on the sleep phenotype of P{MBSwitch} > Go^GTP flies, we predicted that blocking Go signaling in MBs would reduce sleep amount. As expected, we found that flies expressing PTX or Go RNAi driven by MB247-GAL4 showed a similar level of sleep loss, significantly lower than controls (Figure 3B). However, expressing another RNAi construct with MB247-GAL4 against the Dopamine receptor DopR did not produce any obvious effects on sleep (data not shown). Taken together, these data show that knockdown of endogenous Go in MBs reduces the amount of sleep in flies.

Inhibition of Go expression has been reported to cause lethality and developmental defects.³⁸ It is possible, therefore, that sleep loss induced by PTX or Go RNAi might be due to impairment of MB development rather than to inhibition of endogenous Go signaling. To exclude this possibility, we used P{MBSwitch} to express PTX only in adult flies during the experimental period. Flies carrying both UAS-PTX and P{MBSwitch} exhibited sleep reduction after RU486 induction compared to flies of the same genotype raised with normal food. Figure 3C shows the sleep of these 2 groups in either the LD (day 4) or DD (day 5) treatment. We observed sleep reduction under both conditions. However, sleep in flies carrying only one transgene was not affected by feeding with RU486 (Figure 3D). Moreover, PTX induction did not cause any detectable morphological changes in MBs ³⁸ (data not shown). Taken together, these results suggest that sleep reduction by expression of PTX in MBs results from inhibition of endogenous Go signaling rather than developmental defects.

Further analysis revealed that sleep decrement in PTX-induced flies was derived from shortened sleep episodes, while there was no significant change in sleep bout number (Figure S5A and Figure S5B). Go loss-of-function and Go overexpression both affected sleep maintenance, but in opposite ways, confirming the role of Go in sleep regulation. Although total locomotor activity of the PTX-induced flies increased, their activity during wake periods was comparable with control groups (Figures S5C and S5D), suggesting that the increment in locomotor activity resulted from extended wake periods, rather than hyperactivity.

We also investigated whether MBs are required for PTX or Go^GTP-driven sleep changes. Since P{MBSwitch} encodes a GAL4-progesterone-receptor fusion protein which cannot be inhibited by GAL80 (data not shown), we used hydroxyurea (HU) to ablate the MBs.⁴¹ As Figure 4A shows, HU treatment efficiently removed most of MBs. RU486-induced expression of PTX in residual MBs did not produce significant sleep reduction compared to control flies with intact MBs. Furthermore, the Go^GTP-induced sleep increment was also abolished by MB ablation (Figure 4C). These results thus demonstrate that these Go-induced sleep changes observed earlier in this study requires functional MBs.

In agreement with this conclusion we found a significant sleep reduction when expressing Go RNAi with another MB driver, 30Y-GAL4 (Figure 4D), which was previously used to identify the sleep regulatory function of MBs,³³ and labels several regions including MBs, Pis, and AL neurons. To further demonstrate the necessity of Go signaling in MBs for producing the sleep phenotype, we introduced MB-GAL80 which has been reported to abolish GAL4 activity specifically in MBs.⁴² Using the UAS-GFP reporter, we confirmed that MB-GAL80 inhibited GFP expression in MBs (Figure 4B). With both MB-GAL80 and 30Y-GAL4, we found that fly sleep returned to level of control flies (Figure 4D). Taken together, these data support the hypothesis that expression of endogenous Go within the MBs is necessary for normal sleep.

The Functions of Go Signaling on Sleep Are Independent of the Cyclic-AMP Pathway

Previous studies have shown that cAMP levels are inversely related to sleep amount.^14,15 In addition, constitutive activation of the vertebrate Goα subunit inhibits Ca2+/calmodulin-responsive type I adenylyl cyclase,⁴³ resulting in decreased intracellular cAMP levels. Hence, the effect of Go^GTP on sleep might arise from alteration of the cAMP pathway. To address this possibility, we tested whether cAMP signaling is required for Go-induced sleep changes. We first used Gs^GTP and Gi^GTP, constitutively active forms of stimulatory Gs protein⁴⁴ and inhibitory Gi protein⁴⁵ that directly up-regulate and down-regulate the synthesis of cAMP through affecting the catalytic activity of adenylyl cyclase,⁴³ as positive controls. In keeping with previous studies, pan-neuronal expression of Gs^GTP inhibited sleep, whereas expression of Gi^GTP significantly promoted sleep (Figure 5A). In Drosophila, the rutabaga gene encodes Ca²⁺/calmodulin-responsive type I adenylyl cyclase,⁴⁶ and the dunce gene encodes cAMP-specific phosphodiesterase (PDE).^38,47 Mutations in rutabaga or dunce disrupt intracellular cAMP levels and impair cAMP-dependent memory acquisition.^38,47 Given that activation of the Gs pathway is mediated by adenylyl cyclase, loss of rut activity should inhibit Gs^GTP-driven sleep reduction. As predicted, rut²⁰⁸⁰ hypomorph mutant flies did not display further sleep reduction following Gs^GTP induction (Figure 5A) implying that rutabaga adenylyl cyclase acts downstream of Gs^GTP. In contrast, a Go^GTP-driven sleep-enhancing effect was still observed despite reduction of rut activity in the rut²⁰⁸⁰ mutant or rut¹ loss-of-function mutant (Figure 5B). Furthermore, the homozygous dunce hypomorph mutant dnc¹ that showed increased intracellular cAMP levels did not suppress Go^GTP-induced sleep enhancement either (Figure 5B). rutabaga adenylyl cyclase is enriched in MBs,⁴⁶ and rut¹ mutants still show sleep increase when Go^GTP expression is driven with P{MBSwitch} (Figure 5B), further confirming that the cAMP pathway is not required for the Go^GTP-induced sleep increase in MBs. We also examined whether several other key components in the cAMP pathway, including PKA and CREB, are involved in the Go^GTP-driven sleep phenotype using PKAmc* (a constitutively active catalytic subunit of PKA),⁴⁸ PKAr (Drosophila inhibitory subunit of PKA with a mutated cAMP binding site),⁴⁸ and CREB-2b (dominant-negative CREB blocker).⁴⁹ These proteins have been used in several previous reports to produce sleep changes through activating or blocking the cAMP pathway.^14,15 Although pan-neuronal expression of PKAmc* and PKAr inhibited and promoted sleep, respectively, additional expression of Go^GTP bypassed these sleep phenotypes and yielded remarkable sleep increase (Figure 5C). A similar result was also observed in the CREB-2b data (Figure 5C). Feeding with RU486 did not significantly alter sleep in control groups (Figure S6). These results further support the idea that the cAMP pathway may not be required for Go signaling-mediated sleep changes.