Introduction

Seasonal affective disorder (SAD), also known as winter depression, is characterized by the onset of depression in fall/winter months and spontaneous remission in the spring/summer (Rosenthal et al., 1984). It is generally agreed that SAD is caused by lack of day light in the fall/winter months due to the short day length (or photoperiod), as bright light therapy is effective and most commonly used for treating the depression associated with SAD (Golden et al., 2005). The prevalence of SAD is ∼1-10% world-wide, with symptoms lasting for approximately 40% of the year (Kurlansik and Ibay, 2012). About 64-80% of SAD patients report a winter increase in sleep, ranging from 30 min to 2 h longer in duration compared to controls, which is considered to be a distinguishing symptom in the characterization and diagnosis of SAD (Wescott et al., 2020). Currently almost nothing is known regarding the underlying mechanism of this winter hypersomnolence.

This phenomenon of winter hypersomnolence in SAD patients implicates distinct mechanisms that regulate sleep under short winter-like photoperiods vs. longer non-winter-like photoperiods, as the sleep of these individuals appear to be selectively perturbed under short photoperiods. However, the mechanisms by which sleep duration is determined under different photoperiods have not been characterized. Since the circadian clock is believed to be important for adaptations to seasonal changes in the environment and in particular, seasonal changes of photoperiod, we hypothesize that the circadian clock may also participate in regulating sleep duration under different photoperiods (Wood and Loudon, 2014).

To address our hypothesis, we tested fruit flies mutant for different circadian clock genes under a range of photoperiods. We found that flies lacking the circadian photoreceptor CRYPTOCRHOME (CRY) display increased sleep duration specifically under short photoperiods, similar to the winter hypersomnolence in SAD. Genetic and pharmacological analysis identified that CRY is functioning in GABAergic neurons and acts upon GABA-A receptor to promote wakefulness. We further narrowed down the neural circuitry mediating the influences of CRY on sleep by demonstrating that cry deficiency increases the GABAergic tone and reduces calcium concentration in the wake-promoting large ventral lateral neurons (l-LNvs) which are known to be GABA-A+ (Hamasaka et al., 2005; Parisky et al., 2008). Consistently, inhibiting these neurons increases sleep while activating them blocks the effects of cry deficiency on sleep. CRY likely functions in part in the GABAergic small ventral lateral neurons (s-LNvs), and lack of cry increases GABA level and the activity of these cells while impairing their GABA transmission suppresses the sleep phenotype of cry mutants (Allada and Chung, 2010). In summary, here we identify a potential role for CRY in down-regulating the GABAergic signaling from the s-LNvs to the l-LNvs specifically under short photoperiod. This in turn enhances the neural activity of the wake-promoting l-LNvs, resulting in increased wakefulness during short winter-like days. These findings reveal a mechanism underlying how sleep duration is determined under winter-like photoperiod, while disruptions of this regulatory system may be related to the winter hypersomnolence associated with SAD.

Results

cry mutation increases sleep duration specifically under short photoperiods

We assessed the sleep of flies mutant for circadian clock gene period (per⁰), timeless (tim⁰), clock (clk^jrk), cycle (cyc⁰) and cry (cry^b) under a range of photoperiods (Figure 1A and B)(Allada, 1998; Konopka and Benzer, 1971; Sehgal et al., 1994; Stanewsky, 1998). We found that cry^bmutation, which is known to be a loss of function or severe hypomorphic allele, leads to increased sleep duration under 4 h light: 20 h dark condition (4L20D) and 8L16D but not under longer photoperiods (Stanewsky, 1998). Since this phenotype recapitulates the winter hypersomnolence of SAD patients, we further characterized the effects of cry deficiency on sleep. We focused on sleep under 4L20D, as the extent of sleep increase is slightly larger than that of 8L16D. We also demonstrated that cry mutation lengthens sleep duration in both male and female flies, while waking activity is not significantly reduced in the mutants (Figure 1C and D). This means the increased sleep in the mutants is not caused by defects of locomotor ability. We next examined the sleep architecture of these flies and found that cry mutation enhances sleep by extending the duration of average sleep bout rather than increasing sleep bout number, indicating that cry deficiency promotes sleep consolidation under short photoperiod (Figure 1E and F). Because CRY exerts influences on sleep/wakefulness in a gender-independent manner, we used male flies for the remainder of the study. In addition, we tested the effects of a cry knock-out allele (cry⁰) on sleep under 4L20D and also observed significantly prolonged sleep duration, similar to cry^b mutation (Figure 1—figure supplement 1) (Dolezelova et al., 2007).

Figure 1.

CRY functions in GABAergic neurons and promotes wakefulness via GABA/GABA-A

Upon light activation, CRY binds to the clock protein TIM and results in its degradation, which is followed subsequently by degradation of CRY itself (Busza et al., 2004; Ceriani et al., 1999; Lin et al., 2001). Therefore, we first tested whether the influences of CRY on sleep/wakefulness requires TIM. We monitored sleep in flies mutant for both cry and tim, and found that the sleep duration of these double mutants are comparable to that of cry mutants (Figure 2—figure supplement 1). This indicates that CRY promotes wakefulness in a TIM-independent manner.

To identify anatomical substrates that mediate the effects of CRY on wakefulness, we employed the UAS/GAL4 system to knock down cry in different brain structures and cell types and verified that cry is indeed knocked down by assessing its mRNA level (Figure 2—figure supplement 2). We found that knocking down cry in GABAergic neurons using vesicular GABA transporter (VGAT)GAL4 or glutamic acid decarboxylase 1 (Gad1)GAL4 results in increased sleep duration under 4L20D but not 12L12D, similar to the cry mutant phenotype (Figure 2A-C; Figure 2—figure supplement 3A and B)(Deng et al., 2019). Given that Gad1GAL4 generated a more prominent sleep phenotype, we used this driver for the remainder of this study.

Figure 2.

Since CRY appears to act in GABAergic neurons to promote wakefulness, we tested whether GABA is involved in this regulatory process. We first took a pharmacological approach and fed flies with drugs to inhibit GABA transaminase (ethanolamine-O-sulphate, EOS) or GABA transporter (nipecotic acid, NipA). Both of these treatments are known to increase GABA level, and indeed lengthens the sleep duration in WT flies (Ki and Lim, 2019; Leal and Neckameyer, 2002). cry mutation fails to enhance sleep when treated with NipA, while the extent of sleep increase caused by cry mutation is smaller when treated with EOS compared to the control (Figure 2D; Figure 2—figure supplement 3C and D). Similarly, knocking down cry in GABAergic neurons no longer lengthens sleep duration after NipA treatment (Figure 2E; Figure 2—figure supplement 3E). For further validation, we adopted a genetic approach. We found that knocking down Gad1, the synthetic enzyme for GABA, suppresses the long sleep phenotype of cry RNAi flies (Figure 2F; Figure 2—figure supplement 3F) (Jackson et al., 1990). In addition, we knocked down VGAT which encodes an essential transporter responsible for packing GABA into synaptic vesicles (Enell et al., 2007). cry deficiency fails to lengthen sleep duration when VGAT is knocked down (Figure 2G and H; Figure 2—figure supplement 3G and H). We verified that Gad1 and VGAT are indeed knocked down by measuring their mRNA levels (Figure 2—figure supplement 3I and J). As a control for the genetic interaction experiments, we co-expressed a GFP RNAi and found this does not significantly alter the sleep duration of cry RNAi flies (Figure 2—figure supplement 3K). These series of results indicate that CRY acts via GABA to promote wakefulness.

We next sought to identify GABA receptor that mediates the effects of CRY on sleep/wakefulness. We fed flies with agonists of GABA-A (THIP) and GABA-B receptor (SKF-97541) (Ki and Lim, 2019). Both drugs enhance sleep in WT, while cry mutation can increase sleep in flies fed with SKF-97541 but not THIP, implicating that CRY acts through GABA-A to promote wake (Figure 3A-C). Consistently, GABA-A receptor antagonist carbamezapine (CBZ) reduces sleep in WT flies while cry mutation fails to lengthen sleep duration after CBZ treatment (Figure 3D and E) (Agosto et al., 2008). We also treated cry RNAi flies with THIP or CBZ and found this treatment abolished the long sleep phenotype as well (Figure 3F-I). To validate that CRY modulates sleep/wakefulness by acting upon GABA-A receptor, we tested for genetic interaction between cry and Resistant to dieldrin (Rdl), the gene that encodes GABA-A receptor in flies (ffrench-Constant et al., 1993). We found that Rdl mutation (Rdl^MDRR) blocks the sleep-enhancing effect of cry RNAi (Figure 3J-L)(Agosto et al., 2008). These findings demonstrate that GABA-A receptor mediates the wake-promoting function of CRY.

Figure 3.

In short, pharmacological and genetic approaches reveal that CRY acts in GABAergic neurons and impinges on GABA/GABA-A signaling to promote wakefulness.

CRY acts upon l-LNvs to promote wakefulness

Previous studies reported that GABA enhances sleep at least in part by activating RDL on the l-LNvs, thus inhibiting the activities of these cells and reducing the release of the arousal-promoting neuropeptide pigment dispersing factor (PDF) (Chung et al., 2009; Parisky et al., 2008). Therefore, we tested whether the l-LNvs mediate the effects of CRY on sleep/wakefulness. We first verified that the l-LNvs indeed receive projections from GABAergic neurons. We expressed a GFP-tagged synaptotagmin (syt-GFP) which labels axon terminals using Gad1GAL4, and GFP signal was observed at the l-LNv soma (Figure 4—figure supplement 1A) (Zhang et al., 2002). We employed the GFP Across Synaptic Partners (GRASP) method with enhanced specificity (t-GRASP) to check for synaptic connections between GABAergic neurons and l-LNvs (Shearin et al., 2018). Pre-mGRASP was expressed in the axon terminal of GABAergic neurons (using Gad1GAL4), whereas its t-GRASP partner post-mGRASP was expressed in the l-LNvs (using the driver PdfLexA) (Shang et al., 2008). GFP signal can be detected at the l-LNv soma, which means GABAergic neurons send synaptic projections to the l-LNvs (Figure 4—figure supplement 1B). We further validated this connection using the trans-Tango technique which labels down-stream synaptic targets of GABAergic neurons with the HA tag, and we were able to observe HA expression in the l-LNvs (Figure 4—figure supplement 1C) (Talay et al., 2017). All in all, these results demonstrate that GABAergic neurons project to the l-LNvs and form synaptic connections.

We next examined the effects of cry mutation on GABA level at the l-LNv soma by immunostaining. As expected, GABA signal is significantly increased in cry mutants under 4L20D but not 12L12D (Figure 4A-D). Consistent with elevated GABA, GCaMP6m signal, an indicator of intracellular calcium concentration, is significantly reduced in the l-LNvs under 4L20D (Figure 4E and F) (Chen et al., 2013). This implies that the neural activity of these cells is down-regulated, possibly due to increased GABA signaling.

Figure 4.

To test whether the long sleep phenotype in cry mutants is due to reduced activity of the l-LNvs, we electrically silenced these neurons by expressing an inward rectifying potassium channel Kir2.1 (Baines et al., 2001). This results in lengthened sleep duration under 4L20D, mimicking the sleep phenotype of cry mutants (Figure 4G; Figure 4—figure supplement 2A). On the other hand, activating the l-LNvs by expressing the temperature-gated depolarizing cation channel TrpA1 or the bacterial depolarization-activated sodium channel NachBac blocks the sleep enhancing effect of cry mutation (Figure 4H and I; Figure 4—figure supplement 2B and C) (Hamada et al., 2008; Nitabach et al., 2006). These findings support the notion that cry mutation lengthens sleep duration by down-regulating the neural activity of the l-LNvs. Since it has been shown that the l-LNvs promote arousal by releasing PDF, we assessed whether PDF or its receptor PDFR is required by CRY to exert influence on sleep/wakefulness (Chung et al., 2009; Parisky et al., 2008). cry mutation fails to lengthen sleep duration on Pdf or Pdfr mutant background, indicating the necessity of PDF signaling in mediating the arousal-promoting function of CRY (Figure 4J and K; Figure 4—figure supplement 2D and E) (Hyun et al., 2005; Lear et al., 2005; Mertens et al., 2005; Renn et al., 1999).

To summarize, these results strongly suggest that CRY promotes wakefulness by reducing GABA signaling and thus increasing the activity of the l-LNvs, which may in turn lead to increased release of PDF.

CRY acts in the s-LNvs to promote wakefulness

To identify the subset of GABAergic neurons in which CRY functions to regulate wakefulness, we first examined the expression pattern of Gad1GAL4 by labeling the GAL4+ cells with a nuclear GFP (nls-GFP) (Shiga et al., 1996). While the l-LNvs are not GFP+, we noticed that the s-LNvs which are close to the l-LNvs and also express PDF appear to be GFP+ (Figure 5A and B). Because the s-LNvs are known to express CRY, we suspected that CRY may be acting in these s-LNvs to regulate the activity of the l-LNvs via GABA signaling (Benito et al., 2008; Yoshii et al., 2008). To test this idea, we knocked down cry in the s-LNvs using R6GAL4 while over-expressing dicer2 (dcr2) to enhance RNAi efficiency and observed a modest but significant lengthening of sleep duration (Figure 5C; Figure 5—figure supplement 1A) (Helfrich-Forster et al., 2007). However, when we adopted a PdfGAL80 to block the actions of GAL4 in the PDF neurons in Gad1GAL4/UAScryRNAi, this does not alter the long-sleep phenotype (Figure 5—figure supplement 1B) (Stoleru et al., 2004). These series of results suggest that cry expression in the s-LNvs are necessary but not sufficient to maintain normal sleep/wakefulness. We were indeed able to detect GABA signal at the s-LNvs, while GABA intensity is enhanced in cry mutants under 4L20D but not 12L12D (Figure 5D-G). To further validate that the s-LNvs are GABAergic, we knocked down VGAT in these cells and observed a decrease of GABA intensity (Figure 5—figure supplement 2A and B).

Figure 5.

Next, we examined the effects of cry deficiency on the activity level of the s-LNvs. We found that cry mutation increases calcium concentration in these cells under 4L20D, in stark contrast to that of the l-LNvs (Figure 5H and I). This strongly suggests that cry deficiency leads to elevated neural activity in the s-LNvs. Consistently, we observed increased sleep when we activated these cells using TrpA1, similar to the long-sleep phenotype of cry mutants (Figure 5J; Figure 5—figure supplement 3A). Because R6GAL4 is also expressed in several other cells in the brain, we combined PdfGAL80 to verify that the sleep-enhancing effect is caused by over-activation of the s-LNvs (Helfrich-Forster et al., 2007). Indeed, we no longer observed the long-sleep phenotype when GAL4 expression is inhibited in the PDF neurons by GAL80 (Figure 5J; Figure 5—figure supplement 3A). Moreover, we found that cry mutation can no longer exert effects on sleep when the s-LNvs are activated, further validating that the s-LNvs mediate the influences of CRY on sleep (Figure 5K; Figure 5—figure supplement 3B). Because CRY has been shown to regulate membrane depolarization and input resistance together with the redox sensor of the voltage-gated potassium channel β-subunit HYPERKINETIC (HK), we tested whether CRY and HK also function together in the s-LNvs to regulate arousal (Agrawal et al., 2017; Fogle et al., 2015). We found that while over-expressing hk in the s-LNvs of WT flies does not alter sleep duration under 4L20D, over-expressing it in cry mutants significantly increases sleep (Figure 5L; Figure 5—figure supplement 3C). This genetic interaction indicates that CRY and HK cooperate to regulate wakefulness, possibly by modulating the electric activity of the s-LNvs.

Next, we expressed syt-GFP in the s-LNvs and observed GFP signal at the soma of the l-LNvs, indicating that the s-LNvs send axonal terminals to the l-LNvs (Figure 6A). trans-Tango technique further demonstrated that the s-LNvs project to form synaptic connections with the l-LNvs (Figure 6B). Moreover, when we disrupted GABA transmission in the s-LNvs by knocking down VGAT or Gad1, this reduced GABA intensity in the l-LNvs, implicating that the s-LNvs release GABA onto the l-LNvs (Figure 6C-E). At the behavioral level, these manipulations supress the long-sleep phenotype of cry mutation (Figure 6F-I).

Figure 6.

Taken together, these results suggest that cry deficiency in the s-LNvs results in increased neuronal activity and thus enhanced GABA release at the l-LNvs via direct synaptic projections, ultimately leading to decreased wakefulness and increased sleep.

Short photoperiod reduces GABA level at the l-LNvs

Our findings thus far point to an inhibitory role of CRY on the GABAergic tone, and this in turn removes the inhibition on the neural activities of the l-LNvs. Since CRY is degraded by light, we hypothesized that CRY should exert a stronger effect on GABA under short photoperiod (Emery et al., 1998). Consistently, GABA intensity does appear to be down-regulated at the l-LNvs under 4L20D vs. 12L12D (Figure 7A and B). On the other hand, GABA intensity does not exhibit photoperiod-dependent alteration in the s-LNvs (Figure 7C and D). Presumably, this reduced GABA level at the l-LNvs under 4L20D can result in elevated activation of these cells and increased wakefulness. In line with this, WT flies display shortened sleep duration under 4L20D compared to 12L12D (Figure 7E and F). This sleep reduction is a result of decreased sleep bout length but not bout number, while wake activity is not altered by photoperiod (Figure 7G-I). These observations indicate that short photoperiod hampers sleep maintenance, similar to the effects of CRY.

Figure 7.

Discussion

Previous studies have shown that CRY mediates light-induced electrical activity of the l-LNvs and acute arousal, which indicates that CRY can immediately act to promote wake and terminate sleep in response to light pulse (Fogle et al., 2015; Sheeba et al., 2008). At the molecular level, this is believed to be accomplished by a direct coupling of light-activated CRY with HK in the l-LNvs (Fogle et al., 2015). Here we found that CRY promotes extended wakefulness under short photoperiod by functioning as an inhibitor of GABAergic tone. In contrast to the previously characterized roles of CRY which are activated by light, we believe this novel function we identify here reflects a role for CRY in the dark. Under short photoperiods, CRY acts in the dark to inhibit GABA signaling and thus promote wakefulness, leading to reduced sleep during the dark phase compared to longer photoperiods. Indeed, CRY is an ideal signal for conveying photoperiodic information, for it is degraded by light and only accumulates during darkness (Emery et al., 1998). It can measure the length of the day/night and thus instruct downstream signaling components to modulate photoperiod-dependent processes. Consistent with this idea, the largest increases of sleep in cry mutants occur immediately after lights-off and prior to lights-on (Fig 1B), indicating that CRY functions to promote wakefulness during these time windows which are exactly the time windows that would be affected by photoperiod changes (i.e. there will be light during these time windows under longer photoperiods). In other words, CRY appears to act at the time of the day that is most sensitive to photoperiodic changes, which fits perfectly with the idea that it serves as an instructive signal of day/night length.

While EOS and NipA treatment lengthen sleep duration as previously reported, we are somewhat surprised by the observation that knocking down Gad1 or VGAT in GABAergic neurons also extends sleep duration (Ki and Lim, 2019; Leal and Neckameyer, 2002). We reason this may be due to some sort of over-compensation induced by chronic GABA deficiency to maintain excitation/inhibition balance, as previous studies have reported that the amplitude of glutamatergic current is substantially down-regulated in Gad1 mutant flies and increased in Gad1 over-expression flies (Featherstone et al., 2000), (Featherstone et al., 2002). Consistent with this idea, we noticed when cry is knocked down in Gad1/VGAT RNAi flies, sleep duration is shortened and comparable to that of the controls. This is probably because cry depletion enhances the GABAergic tone, thus normalizing GABA signaling in these flies which results in normal sleep duration.

We acknowledge that THIP treatment leads to prominent lengthening of sleep duration, and thus it is possible in this case cry deficiency no longer increases sleep duration due to a ceiling effect rather than epistatic interaction. Nonetheless, considering that CBZ and rdl mutation also block the effect of cry deficiency on sleep duration, it is highly likely that GABA-A receptor mediates the influences of CRY on sleep. One caveat is that the rdl^MD-RR mutation has been shown to diminish the desensitization of GABA-A receptor and is thus believed to be a gain-of-function allele, but here we found that it eliminates the effect of cry RNAi on sleep (Zhang et al., 1994). Since treatment with GABA-A agonist THIP leads to substantially increased sleep while under the same condition rdl^MD-RR mutation does not appear to significantly alter sleep duration, we suspect that similar to Gad1/VGAT RNAi, chronic enhancement of GABA-A function associated with rdl^MD-RR mutation may also trigger some kind of compensatory mechanism that counteracts this increased GABA-A activity. Consequently, the influences of cry RNAi on sleep are blocked. Of course, a lot more in depth characterizations are required to elucidate these issues.

The s-LNvs have been reported to receive GABAergic inputs possibly via GABA-B receptors, but have not been shown to be able to synthesize GABA (Dahdal et al., 2010; Hamasaka et al., 2005). Here we observed that Gad1GAL4 is expressed in these cells, and their GABA intensity is reduced when we use R6GAL4 to knock down VGAT in these cells. R6GAL4 drives prominent expression in the s-LNvs with very little if any expression in the l-LNvs, and weaker (and not very consistent) expression in several other neurons in the protocerebrum, pars intercerebralis and subesophageal area which are all believed to lie outside of the circadian neuron network (Helfrich-Forster et al., 2007). Therefore, we reason that the alteration of GABA signal associated with knocking down VGAT should arise from VGAT deficiency within the s-LNvs. Taken together, these results suggest that the s-LNvs are GABAergic. Although knocking down VGAT or Gad1 in these cells can suppress the long sleep phenotype of cry mutants, knocking down cry in these cells only lead to a modest lengthening of sleep duration. Moreover, inhibiting cry RNAi expression in PDF neurons does not eliminate the long-sleep phenotype of Gad1GAL4/UAScryRNAi flies. Therefore, we suspect that cry deficiency in other GABAergic neurons are also required for the long-sleep phenotype. Given that the s-LNvs are known to express CRY and appear to be GABAergic based on our findings here, we believe that CRY acts at least in part in the s-LNvs to promote wakefulness under short photoperiod.

The molecular mechanism by which CRY down-regulates the GABAergic tone remains unclear. Besides conducting light-driven depolarization via HK, CRY has also been shown to act in synergy with HK to prevent membrane input resistance from falling to a low level in larval salivary glands (Agrawal et al., 2017; Fogle et al., 2015). In contrast to previous studies, here we found that lack of CRY increases the activity of the s-LNvs. Instead of functioning in synergy with HK, CRY appears to act in the opposite direction of HK as over-expressing hk enhances the long-sleep phenotype caused by cry mutation. We reason that the coupling between CRY and HK as well as their influences on the electric activity in the s-LNvs may be different from that of the l-LNvs and larval salivary glands. Nonetheless, our results also support an interaction between CRY and HK to promote arousal. We suspect that CRY acts via HK to inhibit the activity of the s-LNvs, which results in decreased GABA release and dis-inhibition of the l-LNvs. Extensive further investigations will be needed to elucidate the mechanism by which CRY regulates the activity of the s-LNvs.

In conclusion, here we describe a CRY-controlled GABAergic circuitry involving the l-LNvs and the s-LNvs that adjusts sleep duration in adaptation to changes in day length and propose a mechanistic explanation regarding how this circuitry functions (Fig 7J). Under short photoperiods, more CRY accumulates and inhibits the activity of the GABAergic s-LNvs, leading to a dis-inhibition of the l-LNvs which can release more PDF and promote arousal. Under longer photoperiods, on the other hand, less CRY can accumulate and thus the s-LNvs will exert more inhibitory influences on the l-LNvs, leading to decreased release of PDF and wakefulness. Notably, almost all neurons in the mammalian pacemaker, the suprachiasmatic nucleus, are GABAergic, and GABA/GABA-A signaling have been shown to mediate neuronal coupling in response to photoperiod changes (Ono et al., 2021). We believe a similar GABAergic circuitry may exist in the mammalian system that adjusts sleep/wakefulness to photoperiodic changes.

Materials and Methods

Fly strains

All strains were obtained from Bloomington Drosophila Stock Center, Vienna Drosophila Resource Center and TsingHua Fly Center or as gifts from colleagues. Except for Gad1GAL4, all neurotransmitter related GAL4 lines were generated in Dr. Yi Rao’s laboratory (Deng et al., 2019). The following fly strains were used in this study: cry^b (BDSC:80921), per⁰ (BDSC:80917), tim⁰(BDSC:80922), clk^jrk (BDSC:80927), cyc⁰ (BDSC:80929), Pdf⁰¹ (BDSC:26654), Pdfr^han5304 (BDSC:33068), UAS-cryRNAi-1 (THU4761), UAS-cryRNAi-2 (THU5148), cry⁰³ (Dolezelova et al., 2007), UAS-gad1RNAi-1 (THU2920), Rdl^MD-RR (BDSC:1675), UAS-VGATRNAi-1 (THU4303), UAS-VGATRNAi-2 (V45916), UAS-hk (BDSC:86270), isogenic w¹¹¹⁸ (BDSC:5905), UAS-NachBac (BDSC:9469), PdfGAL4 (BDSC:6899), Gad1GAL4 (BDSC:51630), VGATGAL4 (BDSC:58980), UAS-TrpA1 (BDSC:26263), UAS-Kir2.1 (BDSC:6596), UAS-dcr2 (V6009), UAS-sytGFP (BDSC:6925), UAS-nlsGFP (BDSC:4775), UAS-myrGFP,QUAS-mtdtomato(3XHA);Trans-tango (BDSC:77480), 13XLexAop2-post-t-GRASP,20XUAS-pre-mGRASP (BDSC:79040), Dop2RGAL4, THGAL4, TrhGAL4, AdoRGAL4, DATGAL4, GABAR1GAL4, GABAR3GAL4, SerTGAL4, GluRIAGAL4, R6GAL4(Helfrich-Forster et al., 2007), c929GAL4(Park et al., 2008), PdfGAL80(Stoleru et al., 2004), UAS-GCaMP6m-Tdtomato (Chen et al., 2013) and Pdf-LexA(Shang et al., 2008). All flies used for sleep monitoring were backcrossed with the isogenic w¹¹¹⁸ strain for at least 5 times except for cry⁰³ which was backcrossed three times. All experiments were conducted in male flies unless otherwise specified.

Fly sleep monitoring and analysis

Flies were raised on standard cornmeal-yeast-sucrose medium and kept in 12L12D at 25°C until behavior monitoring. ∼3 to 4-day-old flies were entrained under different photoperiods at 25℃ for 4 days, and then their activities in the next 3 days were analyzed. Sleep is defined as 5 min consecutive inactivity. Sleep was analyzed with Counting Macro written in Excel (Microsoft) following previously published protocol (Pfeiffenberger et al., 2010). Flies were fed with agar-sucrose food (2% agar, 5% sucrose) during entire sleep monitoring. TrpA1 flies were raised at 21°C and baseline sleep was monitored at 21°C. Temperature was then raised at lights on to 29°C for further sleep monitoring.

Drug treatment

For pharmacological experiments, drugs were mixed in the fly food at the following concentrations. For nipecotic acid (10 mg/ml, Sigma) and EOS (10 mM, Sigma), drugs were fed during the entire sleep monitoring. For THIP (10 ug/ml, Sigma), SKF-97541 (10 ug/ml, Tocris) and CBZ (0.15 mg/ml, Sinopharm Chemical Reagent), drugs were fed for 1 day after baseline sleep monitoring. The same amount of solvent was added into the fly food as vehicle control.

RNA extraction and quantitative real-time PCR (qRT-PCR)

Approximately 50 5-day-old flies were collected and frozen immediately on dry ice. Fly heads were isolated and homogenized in Trizol reagent (Life Technologies). Total RNA was extracted and qRT-PCR conducted following our previously published procedures (Bu et al., 2019). The following primers were used to quantify mRNA level, cry F: TGCAGGTACCAAGAATGTGG, R: GTCCACGTCCATCAGTTGC. Gad1 F: TGCCACCACATTGAAGTACC, R: GGTGAACATGTTGGTGTTCG. VGAT F: ACGGCTTTAGGCAAGGTAGC, R: TTGGAATTCGTCGATTTTGC.

Immunostaining

Male flies were entrained for 3 days under indicated photoperiod and collected on Day 4. flies were anesthetized with CO₂ and dissected and fixed with 4% paraformaldehyde diluted in PBS at the indicated time points. The brains were then fixed with 4% paraformaldehyde for 30 minutes. Samples were washed with PBT (PBS with 0.3% TritonX-100) for 3 × 10min. Then samples were incubated in PBS with 1% TritonX-100 for 20 min at room temperature. Brain samples were then blocked in PBT with 5% fetal bovine serum (Hyclone) for 30 minutes and subsequently incubated with mouse anti-PDF (1:100, DSHB), rabbit anti-GABA (1:200, Sigma), and anti-HA (1:100, DSHB) for 2-4 days at 4℃ (4 days for anti-GABA and 2 days for the other antibodies). After PBT rinses for 3 times, the brains were incubated with donkey anti-mouse Alexa Fluor-594 (1:1000, Life Technologies), donkey anti-rabbit Alexa Fluor-488 (1:1000, Abcam) and donkey anti-rabbit Alexa Fluor-647 (1:1000, Abcam) overnight at 4℃. Then the brains were rinsed 3 times in PBS and mounted and imaged using Olympus FV3000 confocal microscope with a 60× objective lens. The intensity of GABA signal was quantified by ImageJ software. For each cell, the image slice with the strongest signal in the Z stack was selected and average intensity was quantified. A region on the same image slice was then selected as background and its intensity value was subtracted from the average intensity value of the cell. This subtracted value was subsequently normalized to the average intensity of the control group as indicated in figure legends. Sample sizes in the legends indicate number of cells examined per genotype. For trans-Tango experiment, offsprings were raised at 18°C for 3-4 weeks before HA and PDF immunostaining.

Calcium imaging

For live imaging experiments (GCaMP and tdTomato), flies 2-3 days old were collected into tubes with standard food and entrained under LD for 3 days. Flies were anesthetized with CO₂ and brains were dissected at the indicated time points in Drosophila adult hemolymph-like saline solution. The dissected brain samples were put on glass slide and sealed with cover slide. The duration for dissection and microscopy should be completed within 0.5 hours for each time point. Images were captured with Olympus FV3000 confocal microscopy with a 20× objective lens. The intensity of GCaMP and tdTomato signals were quantified by ImageJ software. Sample sizes in the legends indicate number of cells examined per genotype.

Statistical Analysis

For data that fit normal distribution, two-tailed Student’s t-test (Microsoft Excel) was used to compare the difference between two genotypes. For data that do not fit normal distribution, Mann-Whitney test (Prism Graphpad) was used to compare the difference between two genotypes. For multiple comparisons, one-way ANOVA with Bonferroni multiple comparison test (Prism Graphpad) was used. Sample size, statistical test, and significance values are indicated in figure legends. Sample size is determined based on previous studies with similar experimental assays.

Acknowledgements

We would like to thank Drs. Chang Liu, Yi Rao, Liming Wang, Junhai Han, Zhihua Liu and Yi Zhong for kindly providing flies used in this study. We would also like to thank Dr. Chang Liu for helpful advice on GABA immunostaining. This work was supported by grants from the Natural Science Foundation of China (31930021 and 32022035) and the Ministry of Science and Technology of China STI 2030-Major Projects (2021ZD0203200-02) to Luoying Zhang and grant from the Natural Science Foundation of China (32300984) to Lixia Chen.

Author contributions

L.C. and L.Z. designed the experiments. L.C. and D.T. conducted the experiments. L.C., C.S. and L.Z. wrote the paper.

Disclosure and competing interests statement

The authors declare no competing interests.

Data Availability

This study includes no data deposited in external repositories.