The Shaker potassium channel may promote sleep by affecting temperature-sensitive calming signals

To understand how ambient temperature affects sleep behaviors in Drosophila, we designed a behavioral test with a temperature shift (Fig. 1a). We first measured baseline sleep in individual flies at 21 °C in 12-h light:12-h dark (LD) cycles. We then elevated the ambient temperature to 29 °C and continuously monitored their sleep behaviors. This experimental setup allowed us to trace the temperature-dependent behavioral plasticity in individual flies. In addition, we could distinguish between immediate arousal responses to the temperature shift at midnight (i.e., 4 h after lights-off) and chronic effects of the high temperature on baseline sleep in the following days.

In wild-type flies, the temperature shift from 21 to 29 °C acutely suppressed nighttime sleep (D sleep); however, the daily amount of baseline sleep subsequently became comparable between 21 and 29 °C (Supplementary Fig. 1a, b). More notable effects of temperature were observed on sleep bout numbers and average sleep bout length (ABL). In particular, high temperature increased sleep bout numbers (Supplementary Fig. 1c) while substantially shortening nighttime ABL (Supplementary Fig. 1d). Accordingly, daily sleep architecture was modulated at 29 °C by long daytime sleep (L sleep) and D sleep fragmentation (Supplementary Fig. 1a, b). Additional change included longer latency to D sleep onset at 29 °C (Supplementary Fig. 1e). The temperature-driven plasticity of sleep behaviors was consistent with nocturnal activities in wild-type flies at high temperature^27–29 (Supplementary Fig. 1f). Nonetheless, our assessment of waking activity (i.e., activity counts per minute awake) suggested that ambient temperatures were less likely to have an effect on locomotion per se (Supplementary Fig. 1g).

We next examined whether mutations in sleep-regulatory genes could affect sleep behaviors differentially at either temperature. Indeed, we found that loss of Sh function caused short sleep phenotypes in a temperature-dependent manner. Male flies hemizygous for the mns mutation (Sh[mns]) displayed short and fragmented sleep at 21 °C (Fig. 1b, c and Supplementary Fig. 2), consistent with the previous observations⁴. Intriguingly, Sh mutants did not display any L sleep phenotypes at 29 °C. Their short D sleep was also partially rescued by high temperature since we detected a significant interaction between temperature and Sh mutation on D sleep amount (Fig. 1c, P < 0.0001 by Aligned ranks transformation ANOVA). These results thus indicate that Sh promotes sleep more potently at low temperature.

We further investigated if sleep mutant phenotypes in Sh-relevant genes were also sensitive to ambient temperature. As described previously^8,9, both Hk and sss mutants showed Sh-like, short sleep phenotypes at 21 °C (Fig. 1b, c and Supplementary Fig. 2). However, these two sleep mutants behaved very differently at 29 °C. Loss of Hk function largely phenocopied Sh mutation, since short sleep in Hk mutants was significantly rescued at 29 °C (P < 0.0001 for temperature x genotype interaction on either L or D sleep by Aligned ranks transformation ANOVA). By contrast, sss mutants showed short L sleep even at 29 °C and additive effects of temperature and sss mutation were detected on L sleep (P = 0.6722 for sss[MI] heterozygous backgrounds by Aligned ranks transformation ANOVA). Given the lack of Sh phenotypes in L sleep at 29 °C, these observations suggest that sss promotes L sleep independently of its regulatory role in Sh function at high temperature. In this regard, transgenic expression of wild-type sss in cholinergic neurons of sss mutants rescued short sleep but not leg-shaking phenotypes relevant to Sh⁶. It has been further demonstrated that sss antagonizes nicotinic acetylcholine receptors and suppresses cholinergic transmission in wake-promoting neurons¹⁰.

High temperature suppresses D sleep consolidation, which may lead to the lengthening of L sleep duration as a consequence of sleep homeostasis^32,33. We reasoned that a similar mechanism could underlie temperature-sensitive L sleep phenotypes in Sh mutants. Accordingly, we compared wild-type and Sh mutant sleep in constant light (LL) or constant dark (DD). LL abolishes circadian rhythms in Drosophila³⁴, thus disrupting daily sleep-wake cycles. Conversely, DD allows circadian gene expression and locomotor rhythms in the free-running condition while avoiding any direct behavioral responses to light transitions.

Sh mutants displayed short sleep phenotypes in LL at 21 °C (Fig. 2a, b), indicating that sleep-promoting effects of Sh require neither circadian clocks nor daily light cycles. By contrast, Sh mutant phenotypes disappeared in LL at 29 °C, consistent with the lack of Sh effects on L sleep in LD cycles at high temperature. Further sleep analyses revealed that Sh mutants showed short DD sleep at either temperature (Fig. 2c, d) and no significant interactions between temperature and Sh mutation were detected on DD sleep (P = 0.1002 for subjective L; P = 0.0918 for subjective D by Aligned ranks transformation ANOVA). These observations could be explained by a model that Sh promotes sleep at either temperature but light masks Sh effects more strongly at high temperature. In either constant condition, high temperature suppressed Hk mutant sleep (Supplementary Fig. 3a, b) whereas ambient temperature did not significantly affect sss mutant sleep (Supplementary Fig. 3c, d). Although sleep behaviors are likely governed by more complex mechanisms in LD cycles, our results demonstrate that light and temperature cooperatively gate the sleep-promoting effects of Sh in a manner independent of circadian rhythms or sleep homeostasis. The interplay between light and temperature has also been documented on heat-induced sleep in circadian clock mutants³¹.

Sh mutation leads to the hyperexcitability of affected neurons, delaying repolarization by potassium efflux and possibly elevating intracellular Ca²⁺ levels^35,36. Given the genetic evidence that Sh promotes sleep, a simple hypothesis would be that loss of Sh function in wake-promoting neurons excites their neural activity and thereby suppresses sleep. In fact, a previous study has shown that RNA interference (RNAi)-mediated depletion of the SH complex, including HK and SSS, in sleep-promoting dFSB neurons suppresses sleep²¹. This puzzling observation indicates a more elaborate mechanism for Sh function in dFSB neurons that sustains their tonic firing during sleep²⁵. Nonetheless, it does not exclude the possibility that Sh may also act in brain regions other than dFSB neurons⁶, contributing to temperature-sensitive L sleep in Sh mutants. Accordingly, we employed a similar RNAi strategy to deplete Sh expression in a specific group of neurons and examined its effects on sleep behaviors.

We first confirmed that overexpression of the Sh RNAi transgene by the pan-neuronal galactose-responsive transcription factor 4 driver (ELAV-Gal4) reduced Sh transcript levels in adult fly heads (Supplementary Fig. 4). The pan-neuronal silencing of Sh expression indeed suppressed sleep only at 21 °C, partially phenocopying Sh mutant sleep (Fig. 3a, b). Similarly, pan-neuronal disruption of the Sh locus by CRISPR-mediated targeting³⁷ suppressed sleep at 21 °C, but not at 29 °C (Supplementary Fig. 5). The stronger transgenic phenotypes at low temperature cannot be explained by the intrinsic temperature dependency of the Gal4 activity since high temperature actually enhances the Gal4-driven expression of Drosophila transgenes in general. These results further support that loss of neuronal Sh function, but not the temperature-sensitive nature of the Sh[mns] allele, may underpin temperature-sensitive sleep phenotypes in Sh mutants.

We then tested a number of Gal4 drivers to deplete Sh expression in individual subsets of neurons (Fig. 3a and Supplementary Data 1). Sh depletion in wake-promoting neurons that release DA (TH-Gal4), OA (TDC2-Gal4), or acetylcholine (ChAT-Gal4)³⁸ negligibly affected sleep (Supplementary Fig. 6). We also did not detect any sleep suppression by Sh depletion in dFSB neurons (23E10-Gal4) (Fig. 3a and Supplementary Fig. 7), possibly due to some differences in experimental conditions (e.g., temperature, gender, age). On the other hand, we found that two Gal4 drivers (i.e., 121y-Gal4 and 30y-Gal4) expressed broadly in adult brain, including the mushroom body, significantly suppressed L sleep at either temperature while causing short D sleep only at 21 °C (Supplementary Fig. 8). Accordingly, these data suggest that a sub-group of Sh-expressing neurons may suppress sleep regardless of ambient temperature. We also reason that sleep phenotypes observed in Sh mutants are caused by the net effects of its loss-of-function in various subtypes of Sh-expressing neurons (e.g., wake-promoting vs. sleep-promoting, constitutive vs. thermosensitive).

Our genetic screen further identified a vesicular γ-aminobutyric acid transporter (VGAT)-Gal4 driver as one of the strongest hits that mimicked L sleep phenotypes in Sh mutants (Fig. 3a, b). Sh depletion in VGAT-expressing neurons suppressed L sleep at 21 °C, but not at 29 °C, whereas it did not cause any D sleep phenotypes at either temperature. Similar results were obtained using independent transgenic lines of VGAT-Gal4 or Sh RNAi (Supplementary Fig. 9), excluding their off-target effects. The lack of D sleep phenotypes by VGAT-Gal4 may indicate that VGAT-expressing neurons contribute specifically to short L sleep in Sh mutants. Alternatively, the degree of Sh depletion by VGAT-Gal4 may not be sufficient to suppress D sleep. VGAT-Gal4 contains a transgenic promoter from the VGAT locus and is expressed in neurons that release the inhibitory neurotransmitter GABA³⁹. To validate if Sh expression in GABAergic neurons was responsible for Sh mutant sleep, we genetically combined a glutamic acid decarboxylase 1 (GAD1)-Gal80 repressor transgene⁴⁰ with the pan-neuronal driver (ELAV-Gal4) or VGAT-Gal4 to overexpress the Sh RNAi transgene. Since GAD1 is a key enzyme for GABA synthesis, this genetic manipulation would block the RNAi-mediated depletion of Sh selectively in GABAergic neurons. Indeed, GAD1-Gal80 rescued the sleep phenotypes caused by the RNAi-mediated depletion of Sh (Fig. 3b). These data strongly implicate GABAergic neurons as a neural locus important for temperature-sensitive effects of Sh on sleep behaviors, although we do not rule out the possibility that VGAT-Gal4 or GAD1-Gal80 transgene may also be expressed in some non-GABAergic cells.

To validate Sh-dependent neural activity in GABAergic neuron, we expressed a Ca²⁺-sensitive transcriptional fluorescence reporter CaLexA⁴¹ in VGAT-expressing neurons. Their fluorescence signals were then compared between wild-type or Sh mutant flies entrained at either 21 °C or 29 °C. Similar approaches have been employed in previous studies to measure long-term changes in intracellular Ca²⁺ levels as a proxy for neural activity^22,42,43. The quantification of our Ca²⁺ imaging revealed that Sh mutation elevated intracellular Ca²⁺ levels in subsets of VGAT-expressing neurons (Supplementary Fig. 10). This was consistent with previous observations that loss of Sh function increases neuronal excitability^35,36. Significant effects of Sh on Ca²⁺ levels were detected in mushroom body and ellipsoid body (Supplementary Fig. 10b, P < 0.0001 and P = 0.0165, respectively, by Aligned ranks transformation ANOVA). We further detected statistically significant interaction between Sh mutation and temperature on Ca²⁺ levels in mushroom body (P = 0.0148 by Aligned ranks transformation ANOVA), but not in ellipsoid body (P = 0.2180 by Aligned ranks transformation ANOVA). In fact, high temperature generally weakened the CaLexA-driven fluorescence signals in either wild-type or Sh mutant brains (Supplementary Fig. 10b, P < 0.0001 for mushroom body and antenna lobe; P < 0.001 for ellipsoid body and pars intercerebralis by Aligned ranks transformation ANOVA). This observation possibly points to the intrinsic temperature-sensitivity of the CaLexA transgene. Nonetheless, these data support that Sh mutation likely increases the excitability of GABAergic neurons and thereby sustains GABA transmission.

If greater neural activity in GABAergic neurons was responsible for Sh mutant sleep, we reasoned that a down-scaling of the presynaptic GABA transmission should suppress Sh mutant phenotypes. To this end, we genetically silenced GABA transmission by heterozygous loss-of-function mutations in VGAT, which incorporates GABA into the synaptic vesicles of GABAergic neurons^39,44, or in GABA-synthesizing GAD1⁴⁵. Their effects on sleep behaviors were then compared between wild-type and Sh mutant backgrounds. The heterozygosity of either VGAT or GAD1 had negligible effects on wild-type sleep at either 21 °C or 29 °C (Fig. 3c). However, their trans-heterozygosity increased the amount of wild-type L sleep specifically at 21 °C (Fig. 3c, P < 0.0001 for temperature x genotype interaction on L sleep by Aligned ranks transformation ANOVA), thereby dampening temperature-dependent changes in L sleep duration. More importantly, the trans-heterozygosity of VGAT and GAD1 rescued Sh mutant sleep at 21 °C (Fig. 3c, P < 0.0001 for trans-heterozygosity of VGAT and GAD1 × Sh[mns] interaction on either L or D sleep by Aligned ranks transformation ANOVA) while it had additive effects with Sh mutation on D sleep at 29 °C (P > 0.05 for trans-heterozygosity of VGAT and GAD1 × Sh[mns] interaction on D sleep by Aligned ranks transformation ANOVA). The heterozygosity of VGAT similarly suppressed L sleep phenotypes caused by Sh depletion in VGAT-expressing neurons (Supplementary Fig. 9b). Collectively, these genetic interaction data provide convincing evidence that a GABA-dependent mechanism contributes to temperature-dependent effects of Sh on sleep behaviors.

GABA receptor agonists potently enhance sleep in both flies and mammals^11,12,46. By contrast, we observed that genetic suppression of GABA transmission actually promoted L sleep at 21 °C in wild-types and rescued short sleep phenotypes in Sh mutants. GABA receptor genes are broadly expressed in the adult fly brain^47,48. Several lines of evidence indicate that the sleep-modulatory effects of each GABA receptor are mediated by a specific group of neurons involved in the regulation of sleep-wake cycles^12–16. Given the inhibitory nature of GABA transmission onto postsynaptic neurons, we reasoned that GABA receptors may act in sleep-promoting neurons to mediate Sh mutant phenotypes. To elucidate this sleep-suppressing role of GABA transmission, we depleted endogenous expression of individual GABA receptors in different groups of sleep-regulatory neurons and examined their effects on sleep behaviors.

The most striking effects were observed when we silenced the expression of the ionotropic GABA receptor, Resistant to dieldrin (Rdl), in sleep-promoting dFSB neurons⁴⁹. While Rdl expression in dFSB neurons has been reported previously^50–52, we found that dFSB-specific Rdl depletion lengthened wild-type L sleep at 21 °C, but not at 29 °C (Fig. 4). These sleep phenotypes were in stark contrast to Sh mutant sleep (Fig. 1b, c), suggesting the opposing effects of presynaptic Sh in GABAergic neurons and postsynaptic Rdl in dFSB neurons on sleep behaviors. Indeed, Rdl depletion in dFSB neurons non-additively suppressed Sh mutant sleep only at 21 °C (Fig. 4, P < 0.001 or P < 0.01 for Rdl depletion × Sh[mns] interaction on L or D sleep, respectively, by Aligned ranks transformation ANOVA). These results support that Rdl in dFSB neurons acts genetically downstream of GABAergic Sh and antagonizes its effects on sleep at low temperature. Accordingly, Sh effects on sleep likely converge on a specific group of sleep-promoting neurons despite our mapping of the Sh RNAi phenotypes broadly to GABAergic neurons (Fig. 3b).

To functionally validate GABAergic synapses onto dFSB neurons, we examined synaptic GFP reconstitution across synaptic partners (GRASP) between GAD1-expressing neurons and dFSB neurons. Unlike the original GRASP^53,54, synaptobrevin-GRASP (syb:GRASP) requires the fusion of synaptic vesicles with the presynaptic membrane to display trans-synaptic fluorescence at the synaptic cleft⁵⁵. Accordingly, the presence of the syb:GRASP signals verifies the directionality of a given synapse between two groups of neurons. In addition, the relative intensity of their fluorescence is proportional to the neural activity of presynaptic neurons. With this experimental design, we detected the syb:GRASP signals enriched in the dendritic regions of dFSB neurons^56–59 (Fig. 5a, b), validating the directionality of functional synapses from GAD1-expressing neurons to dFSB neurons. Quantitative image analyses further revealed that the syb:GRASP signals from GAD1-expressing neurons were more evident in flies that had been entrained in LD cycles at 21 °C than in those entrained at 29 °C (Fig. 5c, P < 0.0001 by Mann–Whitney U test). These data suggest that GABAergic neurons display higher synaptic transmission onto dFSB neurons at low temperature.

Since dFSB neurons also receive inhibitory dopaminergic inputs^18,19,21, we examined whether ambient temperature also affected dopaminergic synapses onto dFSB neurons. In contrast to GABAergic synapses, the syb:GRASP signals from dopaminergic neurons were comparable between two groups of flies that had been entrained at either 21 °C or 29 °C, respectively (Fig. 5c, P = 0.1777 by Mann–Whitney U test). Taken together, our results indicate that high temperature specifically weakens GABAergic synapses onto dFSB neurons. We reason that this temperature-dependent neural plasticity likely explains stronger sleep phenotypes in Sh mutants or Sh-depleted flies at low temperature.

DA transmission down-regulates the excitability of dFSB neurons, switches them to an electrically quiescent status, and suppress sleep²¹. On the other hand, our data indicate that GABAergic synapse onto dFSB neurons displays temperature-sensitive activity (Fig. 5) and supports temperature-dependent plasticity of wild-type and Sh mutant sleep (Figs. 3c and 4b). Given the inhibitory nature of GABA transmission to postsynaptic neurons, we asked if the excitability of dFSB neurons is differentially modulated by GABA and DA. To this end, we expressed an ATP-gated cation channel P2X₂ along with a synaptic calcium sensor sytGCaMP in dFSB neurons. The transgenic combination allowed us to cell-autonomously excite dFSB neurons by ATP application and measure their excitability using fluorescent Ca²⁺ imaging in live brains⁶⁰. We first found that ATP-induced elevation of intracellular Ca²⁺ levels in dFSB neurons were comparable between transgenic flies entrained at 21 °C and 29 °C (Fig. 6). However, pre-incubation of either GABA or DA suppressed the excitability of dFSB neurons in a temperature-sensitive manner. In particular, GABA potently suppressed the excitability of dFSB neurons only at 21 °C (Fig. 6a) whereas DA silenced it only at 29 °C (Fig. 6b). These results indicate that inhibitory effects of GABA and DA on dFSB neurons are differentially gated by temperature. Considering our GRASP data, we reasoned that temperature might modify the neural property of postsynaptic dFSB neurons, thereby governing their excitability via temperature-specific neurotransmitter signaling.

To determine if ionotropic GABA transmission in postsynaptic dFSB neurons is affected by temperature, we examined GABA-dependent changes in intracellular chloride levels using a transgenic fluorescence resonance energy transfer (FRET) sensor SuperClomeleon^14,61. Bath application of GABA robustly elevated chloride levels in dFSB neurons, but their FRET responses were comparable at 21 °C and 29 °C (Supplementary Fig. 11). It is thus unlikely that chloride influx via ionotropic GABA receptors is limiting at 29 °C to explain lack of GABA effects on the excitability of dFSB neurons at high temperature. We further quantified postsynaptic DA signaling in dFSB neurons using Epac1-camps, another transgenic FRET sensor for cyclic adenosine monophosphate (cAMP)⁶² (Fig. 7a). cAMP is a signaling molecule downstream of D1-like DA receptors implicated in Drosophila sleep^18,19,21. Live-brain imaging of the transgenic Epac1-camps, expressed in dFSB neurons, allowed us to measure the relative changes in intracellular cAMP levels by bath application of DA¹⁹.

Interestingly, we observed that DA-induced cAMP elevation was more robust in dFSB neurons of transgenic flies entrained at 29 °C than those entrained at 21 °C (Fig. 7b). While these observations were consistent with temperature-sensitive effects of DA on the excitability of dFSB neurons, we reasoned that higher activity of GABAergic synapses at 21 °C might have a negative impact on DA signaling in dFSB neurons. Indeed, dFSB-specific depletion of Rdl reversed the repression of the intracellular cAMP response to DA at 21 °C, thus blunting its temperature-dependency (Fig. 7b). Moreover, pre-incubation with GABA repressed DA-induced cAMP elevation in dFSB neurons of control flies, while the transgenic depletion of Rdl masked GABA effects. The weaker suppression by exogenous GABA at 21 °C likely reflected a floor effect caused by the stronger transmission of endogenous GABA to dFSB neurons at low temperature. Finally, we observed that an agonist of ionotropic GABA receptors (THIP), but not that of metabotropic GABA receptors (SKF-97541), suppressed DA-induced cAMP elevation (Fig. 7c), consistent with the Rdl RNAi effects. Collectively, our data demonstrate that ambient temperature tunes the postsynaptic signaling of DA transmission to dFSB neurons via temperature-sensitive GABA synapses. This neural mechanism may confer temperature-dependent plasticity on the sleep-regulatory function of dFSB neurons.

All flies were maintained in standard cornmeal–yeast–agar medium at 25 °C. w[1118] (BL5905; wild-type control), Sh[mns] (BL24149), Sh[TKO] (BL76968), sss[P1] (BL16588), sss[MIC] (BL34309), Gad1[L352F] (BL6295), VGAT[MB01219] (BL23022), TrpA1[1] (BL26504), UAS-Sh RNAi #1 (BL53347), UAS-Rdl RNAi (BL52903), ELAV-Gal4 (BL458), ELAV-Gal4; UAS-Cas9 (BL67073), VGAT-Gal4 (II) (BL58980), VGAT-Gal4 (III) (BL58409), 23E10-Gal4 (BL49032), TH-Gal4 (BL8848), TDC2-Gal4 (BL9313), ChAT-Gal4 (BL6798), 121y-Gal4 (BL30815), 30y-Gal4 (BL30818), GAD1-LexA (BL60324), UAS-mLexA-VP16-NFAT (BL66542; CaLexA), UAS-DenMark, UAS-syt.eGFP (BL33065), lexAop-nSyb-spGFP1-10, UAS-CD4-spGFP11 (BL64315; syb:GRASP), UAS-SuperClomeleon (BL59847), and UAS-Epac1-camps (BL25407) were obtained from Bloomington Drosophila Stock Center. Hk[1] (101-119) was obtained from Kyoto Stock Center. UAS-Sh RNAi #2 (v104474) was obtained from Vienna Drosophila Resource Center. GAD1-Gal80, TH-LexA, and UAS-sytGCaMP6s; UAS-P2X₂ were gifts from Y. Li (Chinese Academy of Sciences), H. Tanimoto (Tohoku University), and R. Allada (Northwestern University), respectively.

To avoid possible effects of undefined genetic backgrounds on sleep behaviors, all mutants and transgenic flies were tested in outcrossed conditions. Sleep behaviors in experimental genotypes were compared to those in control genotypes obtained from appropriate genetic crosses in parallel. w[1118] was set as a wild-type in all genetic crosses to generate control flies heterozygous for mutant alleles or transgenes (e.g., Gal4, Gal80, UAS). Hemizygous male mutants (e.g., Sh[mns], Hk[1]) were generated by crossing mutant virgins to w[1118] males whereas their genetic controls were obtained from parallel crosses in reverse orientation. Sleep behaviors in experimental flies with a specific combination of mutant alleles or transgenes were compared to those in control flies heterozygous for individual ones. Each male fly was transferred into a 65 × 5 mm glass tube containing 5% sucrose and 2% agar food. Locomotor activity in individual flies was then recorded using the Drosophila Activity Monitor system (Trikinetics) and quantified by the number of infrared beam crosses per minute. Behavioral data were collected from more than two independent tests and averaged per genotype. For LD sleep analysis, flies were entrained for 3 days in LD cycle at 21 °C prior to the temperature shift to 29 °C at ZT16 (lights-on at ZT0; lights-off at ZT12) in the last LD cycle at 21 °C. They were further incubated for 3 days in LD cycle at 29 °C. For LL or DD sleep analysis, flies were kept in constant conditions at 21 °C or 29 °C. A sleep bout was defined as a behavioral episode during which flies did not show any activity for 5 min or longer. Sleep parameters were accordingly analyzed with an Excel macro⁷⁶.

Transgenic flies were entrained in LD cycles at either 21 °C or 29 °C for a week. Whole brains were dissected in phosphate-buffered saline (PBS), fixed in PBS containing 3.7% formaldehyde for 28 minutes at room temperature (RT), and then washed three times in PBS containing 0.3% Triton X-100 (PBS-T). For immunostaining, fixed brains were blocked in PBS-T containing 0.5% normal goat serum for 30 min at RT and then incubated with mouse anti-GFP antibody (diluted in PBS-T containing 0.5% NGS and 0.05% sodium azide at 1:1000, NeuroMab) for 2 days at 4 °C. After washing with PBS-T, brains were further incubated with anti-mouse Alexa Flour 488 antibody (diluted at 1:600, Jackson ImmunoResearch) for 1 day at 4 °C, washed with PBS-T, and then mounted in a VECTASHIELD mounting medium (Vector Laboratories). Fluorescence images were acquired using a multi-photon microscope (LSM780NLO, Carl Zeiss) and analyzed using ImageJ software.

Transgenic flies pre-entrained in LD cycles at either 21 °C or 29 °C for a week were anesthetized in ice. A whole brain was briefly dissected in hemolymph-like HL3 solution (5 mM HEPES pH 7.2, 70 mM NaCl, 5 mM KCl, 1.5 mM CaCl₂, 20 mM MgCl₂, 10 mM NaHCO₃, 5 mM trehalose, 115 mM sucrose) and then placed on a cover glass in a magnetic imaging chamber (Chamlide CMB, Live Cell Instrument) filled with HL3 buffer. Live-brain imaging was recorded at room temperature. For GCaMP-based calcium imaging, brains were equilibrated with HL3 buffer for 5 min and then incubated with 50 mM GABA or 10 mM dopamine for 5 min prior to the batch application of 5 mM ATP. A time series of the fluorescence images was recorded using a photoactivated localization microscopy (ELYRA P.1, Carl Zeiss) with a C-Apochromat 40×/1.20 W Korr M27 at a pixel resolution of 512 × 512. Calcium signals were quantified by background substitution method and analyzed using ZEN software (Carl Zeiss). For SuperClomeleon-based FRET imaging, brains were equilibrated with HL3 buffer for 10 minutes prior to the induction of FRET responses by 50 mM GABA. A time series of the fluorescence images was recorded using a multi-photon microscope (LSM780NLO, Carl Zeiss) with a Plan-Apochromat 20×/0.8 at a pixel resolution of 512 × 512. The power of a 458 nm-laser projection was 10%. Two filter ranges (473–491 nm and 509–535 nm) were set for ECFP and EYFP channels, respectively. Pinhole was set as 13.69 AU to select the region of interest within the axon bundle of dFSB neurons. For Epac1-camps-based FRET imaging, brains were incubated with GABA (1 mM), THIP (4,5,6,7-tetrahydroisoxazolo(5,4-c)pyridin-3-ol, 0.1 mg/ml), or SKF-97541 (20 µM) for 5 min prior to the induction of FRET responses by 10 mM DA. A time series of the fluorescence images was recorded using a multi-photon microscope (LSM780NLO, Carl Zeiss) with a Plan-Apochromat 40×/1.3 oil lens at a pixel resolution of 256 × 256. Each stack contained five slices with ~10 µm of step sizes. Pinhole was fully opened (19.07 AU) to avoid any subtle z-drift during imaging acquirement. FRET image analyses were performed using ZEN software (Carl Zeiss).

Fifty fly heads per genotype were homogenized in TRIzol Reagent and total RNAs were extracted according to the manufacturer’s instructions (Thermo Fisher Scientific). After DNase I digestion, purified RNAs were reverse-transcribed using M-MLV Reverse Transcriptase (Promega) and random hexamers. Quantitative real-time PCR was performed using 2x Prime Q-Mastermix (GeNet Bio) with the diluted cDNA samples and gene-specific sets of primers in LightCycler 480 real-time PCR system (Roche). The primer sequences used in this study were as follows: 5’-CCG GTC AAT GTC CCT TTA GA-3’ (forward) and 5’-CTC GAA GAG CAG CCA GAC TT-3’ (reverse) for Sh; 5’-ATC TCC CAC AGG ACG TCA AC-3’ (forward) and 5’-GCG ACG AAG AGA AGG ATC AC-3’ (reverse) for pabp (internal control).

All statistical analyses were performed using GraphPad Prism 6 or R (version 3.5.3). Datasets that did not pass either Shapiro-Wilk test for normality (P < 0.05) or Brown-Forsythe test for equality of variances (P < 0.05) were analyzed by Aligned ranks transformation ANOVA with ARTool library for multifactorial statistical analyses^77,78. Aligned rank transformation (ART) allowed the assessment of temperature or genotype effects on sleep behaviors using non-parametric datasets to validate their significant interaction by ANOVA. Unless otherwise indicated, all sleep parameters were analyzed by repeated measures one-way or two-way ANOVA. For sleep analyses in LL or DD, ordinary two-way ANOVA was conducted. Interaction comparisons were performed using contrast function from the emmeans package (α = 0.05 with the Bonferroni correction)⁷⁹. Post hoc multiple comparisons were performed by Wilcoxon signed-rank test for repeated measures or by Wilcoxon rank-sum test for unpaired conditions (α = 0.05 with the Bonferroni correction). For statistical analyses of imaging data, significant differences or interactions between experimental conditions were determined by Student’s t-test, one-way ANOVA, or two-way ANOVA. Where normality or homoscedasticity test was not passed, Mann–Whitney U test or Aligned ranks transformation ANOVA was conducted accordingly. The numbers of samples analyzed per condition and P values obtained from individual statistical analyses were indicated in the main text or figure legends.

Further information on research design is available in thelinked to this article. Nature Research Reporting Summary

The data that support the findings of this study are included in the paper or available from the corresponding author upon request.