Ca-α1T, a fly T-type Ca2+ channel, negatively modulates sleep

Mammalian T-type Ca2+ channels are encoded by three separate genes (Cav3.1, 3.2, 3.3). These channels are reported to be sleep stabilizers important in the generation of the delta rhythms of deep sleep, but controversy remains. The identification of precise physiological functions for the T-type channels has been hindered, at least in part, by the potential for compensation between the products of these three genes and a lack of specific pharmacological inhibitors. Invertebrates have only one T-type channel gene, but its functions are even less well-studied. We cloned Ca-α1T, the only Cav3 channel gene in Drosophila melanogaster, expressed it in Xenopus oocytes and HEK-293 cells, and confirmed it passes typical T-type currents. Voltage-clamp analysis revealed the biophysical properties of Ca-α1T show mixed similarity, sometimes falling closer to Cav3.1, sometimes to Cav3.2, and sometimes to Cav3.3. We found Ca-α1T is broadly expressed across the adult fly brain in a pattern vaguely reminiscent of mammalian T-type channels. In addition, flies lacking Ca-α1T show an abnormal increase in sleep duration most pronounced during subjective day under continuous dark conditions despite normal oscillations of the circadian clock. Thus, our study suggests invertebrate T-type Ca2+ channels promote wakefulness rather than stabilizing sleep.

Ca-α1T is broadly expressed in the adult brain. After several failed attempts to generate an antibody that works well for immunohistochemistry, we decided to tag the endogenous Ca-α1T with GFP and then visualize its expression pattern in the adult brain. First, we generated a founder line, (Ca-α1T Founder, w+ ), using end-out homologous recombination to facilitate the versatile generation of a variety of different alleles 15 (Fig. 2a). In Ca-α1T Founder, w+ flies, an attP landing site for φC31-mediated DNA integration and a floxed white + marker replace ~2 kb of genomic DNA surrounding the first coding exon of Ca-α1T. Next, we removed the white + marker from the Ca-α1T Founder, w+ line by Cre-mediated recombination to generate Ca-α1T Founder, w− . We then used the φC31 integrase to insert into the attP landing site of Ca-α1T Founder, w− an attB vector (pGE-attB GMR ) containing the deleted genomic region plus an additional GFP coding sequence and linker sequence in-frame after the start codon of Ca-α1T. This produced the GFP::Ca-α1T line, which expresses an N-terminally GFP-tagged Ca-α1T under the control of its own endogenous promoter.
Although w 1118 control flies show no fluorescent background ( Supplementary Fig. S1), these GFP::Ca-α1T flies show GFP fluorescence broadly across the brain (Fig. 2b). GFP::Ca-α1T is expressed in well-structured neuropils like the antennal lobes, the mushroom bodies, the central complex ( Fig. 2c-h), the optic lobes, as well as in some of the less-structured neuropils. The central complex-comprising the fan-shaped body, ellipsoid body, noduli, and protocerebral bridge-shows the strongest expression with the ventral fan-shaped body and ventral noduli particularly prominent (Fig. 2e,g). In mushroom body neurons, there is far more GFP::Ca-α1T in the dendrite-rich calyx of the dorso-posterior brain (Fig. 2h) than the axon-rich lobes of the anterior brain (Fig. 2d). GFP::Ca-α1T is also limited to the posterior mushroom body peduncles, which are the fiber tracks that join the posterior calyces Scientific RepoRts | 5:17893 | DOI: 10.1038/srep17893 with the anterior mushroom body lobes (Fig. 2f). These results suggest strict regulation of the subcellular localization of Ca-α1T channels in the brain.
We next visualized the projections of Ca-α1T-expressing neurons using another knock-in allele, Ca-α1T Gal4 . In Ca-α1T Gal4 , the first coding exon and flanking introns of Ca-α1T are replaced by the Gal4 coding sequence. This The activation (τ act ) and inactivation (τ inact ) time constants for Ca-α1T (○ ) and Ca v 3.1 (☐) obtained by fitting the current traces to double exponentials. (d) Voltage-dependent deactivation of Ca-α1T in HEK-293 cells. Tail currents elicited by step pulses to − 20 mV for 10 ms, followed by re-polarizing potentials (− 120 mV to − 50 mV). Deactivation time constants were obtained by fitting the traces to a single exponential and plotted against re-polarizing potentials. (e) I Ca /I Ba ratios of Ca-α1T and Ca v 3. puts GAL4 expression under the control of the endogenous Ca-α1T promoter (Fig. 3a). Consistent with our results using GFP::Ca-α1T, Ca-α1T Gal4 drives the expression of a membrane-tethered mCherry (UAS-mCD8-ChRFP) broadly across the brain ( Supplementary Fig. S2). The Ca-α1T Gal4 > mCherry and GFP::Ca-α1T signals are strongly co-localized, including in the central complex and mushroom bodies ( Supplementary Fig. S2). This suggests both reagents reflect proper expression from the same endogenous Ca-α1T promoter.  Ca-α1T mutants show increased sleep. Since the Gal4 coding sequence inserted into Ca-α1T Founder to produce the Ca-α1T Gal4 allele included a termination sequence (Fig. 3a), Ca-α1T Gal4 is likely a null allele. As expected, we were unable to detect Ca-α1T expression in the fly head lysates from Ca-α1T Gal4 in western blot analyses using polyclonal Ca-α1T-specific antisera (Fig. 3b). We did, however, detect strong Ca-α1T expression in lysates from w 1118 controls and from a Ca-α1T Rescue allele in which the fragment deleted in both the Ca-α1T Founder and Ca-α1T Gal4 alleles was re-inserted ( Fig. 3a,b). Ca-α1T Gal4 homozygotes are viable and fertile with normal appearance and no obvious movement defects. Two of the mammalian T-type channel subtypes, Ca v 3.1 and Ca v 3.3, have been implicated in the generation of the neural oscillations characteristic of NREM sleep 1,16 . Flies have a well-established sleep-like state that shares some features with mammalian sleep, but it remains unclear whether flies have a stage akin to mammalian NREM sleep. Still, we hypothesized that Ca-α1T-null flies may exhibit sleep defects.
Ca-α1T Gal4 flies show increased total sleep under both 12 h:12 h light-dark (LD) and constant dark (DD) conditions, but this phenotype is particularly prominent during the subjective day under continuous dark (DD) conditions (Fig. 3c,d). The total sleep of Ca-α1T Rescue flies shows a partial rescue in light-dark (LD) conditions and a full rescue to w 1118 levels under continuous darkness (DD) (Fig. 3c,d). Although levels of Ca-α1T protein are grossly normal in Ca-α1T Rescue flies (Fig. 3b), it is possible that the addition of the attR site and the loxP sites in the Ca-α1T Rescue allele (Fig. 3a) subtly reduce expression of Ca-α1T in some small but important neuronal subpopulation preventing a full rescue. By measuring waking locomotor activity, we were able to confirm that the increased sleep of Ca-α1T Gal4 flies is not an artifact of a generalized reduction in movement. In fact, Ca-α1T Gal4 show slightly higher levels of waking activity than their respective controls (Fig. 3e). Normal fly sleep consists of a number of sleep bouts. We, therefore, asked whether the increased sleep of Ca-α1T Gal4 flies is a result of an increased number of sleep bouts, prolonged bout duration, or both. Ca-α1T Gal4 flies do show reduced sleep bout number under LD conditions, but this phenotype is not rescued in Ca-α1T Rescue flies (Fig. 3f). Sleep bout length, on the other hand, is increased under both LD and DD conditions and rescued in Ca-α1T Rescue flies (Fig. 3g).
To confirm that this elevated sleep phenotype is specific to Ca-α1T loss-of-function, we generated three independent deletion mutants via imprecise P-element excision. As expected, all three deletion mutants as well as a trans-heterozygous mutants (Δ 3/Δ 115) show increased sleep, especially in constant darkness ( Supplementary  Fig. S3). In addition, knockdown of Ca-α1T in its own neurons (Ca-α1T Gal4 > UAS-Ca-α1T-IR) increases sleep after the third day of continuous darkness ( Supplementary Fig. S4). Together, these results implicate Ca-α1T as a novel inhibitor of fly sleep.

Circadian rhythms and sleep homeostasis of Ca-α1T mutants. As in other animals, sleep in
Drosophila is regulated by the circadian clock, meaning clock mutants generally show altered sleep phenotypes 17,18 . We therefore asked whether the increased sleep observed in Ca-α1T-null flies can be attributed to a disruption of the circadian clock. After monitoring locomotor activity over seven days of continuous darkness, we found that most Ca-α1T Gal4 flies have a slightly elongated circadian period length (24.3 ± 0.6 vs. 23.9 ± 0.2), a significantly reduced power of rhythmicity (22.3 ± 2.9 vs. 53.4 ± 5.1), and a reduced overall percentage of rhythmic flies (70.3% vs. 92.6%) when compared to w 1118 controls (Fig. 4a). This circadian phenotype of Ca-α1T Gal4 flies is unlikely due to problems in the core circadian clock, however, as transcriptional oscillation of period is normal (Fig. 4b). This means Ca-α1T must act downstream of the core circadian clock to affect rhythmic behaviors, perhaps affecting the firing of important clock-related neurons.
In addition to being controlled by the circadian clock, sleep is also associated with a homeostatic drive proportional to the time an animal spends awake. Thus, we next examined this homeostatic sleep drive in Ca-α1T Gal4 flies by depriving them of sleep for 24 hours and measuring the resulting sleep rebound. Ca-α1T Gal4 flies do recover slightly more of their lost sleep than w 1118 controls, but the difference is not statistically significant (Fig. 4c).

Pan-neuronal knock-down of Ca-α1T increases sleep.
We next asked whether the increased sleep phenotype of Ca-α1T Gal4 flies can be attributed to the function of Ca-α1T in the brain. Pan-neuronal knockdown of Ca-α1T (elav-Gal4 > UAS-Ca-α1T-IR) increases sleep beyond that of heterozygous controls under both LD and DD conditions (Fig. 5). Using the drug-inducible GeneSwitch-Gal4 technique 19 , we asked whether Ca-α1T's influence on sleep occurs during development or whether it is limited to its expression in the adult brain. Ca-α1T knock-down using elav-GeneSwitch(GS)-Gal4 increases sleep in continuous darkness when compared to non-induced controls ( Supplementary Fig. S5). This suggests the sleep phenotype of Ca-α1T-null mutants are unlikely due to developmental defects.
Finally, we sought to narrow down the sleep-regulating role of Ca-α1T to a specific brain region or circuit. We used a range of neuronal Gal4 drivers that cover known sleep centers to knockdown Ca-α1T, but none of them were capable of significantly altering sleep (Fig. 6).

Discussion
In this study, we cloned the only voltage-gated T-type Ca 2+ channel from Drosophila, Ca-α1T. Ca-α1T is the largest T-type channel cloned to date, measuring 3205 amino acids 20 . Electrophysiological characterization of Ca-α1T in Xenopus oocytes showed that Ca-α1T has the hallmark properties of a T-type channel: low-threshold activation at around − 60 mV, a maximal current output at − 20 mV, transient current kinetics elicited by a step-pulse protocol producing a "criss-crossing" pattern, and slow deactivation of tail currents (Fig. 1). These biophysical properties are also consistent with previous studies that implicated Ca-α1T in low-voltage-activated (LVA) currents in both the central and peripheral nervous systems of the fly 5,21 .
Mammalian genomes contain three T-type Ca 2+ channel genes (i.e., Ca v 3.1-3.3), while the fly genome contains only one. We therefore measured Ca-α1T for some of the characteristics that distinguish the three mammalian channels. In terms of current kinetics, Ca-α1T is more similar to mammalian Ca v 3.1 and Ca v 3.2 than Ca v 3.3, which exhibits considerably slower kinetics. In terms of both its relative permeability to Ba 2+ over Ca 2+ and its sensitivity to nickel inhibition, Ca-α1T is most similar to Ca v 3.2 22,23 .
The three mammalian T-type Ca 2+ channels, each with their own distinct biophysical properties, are expressed in largely complementary patterns of neurons throughout the brain, conferring considerable functional diversity. Areas of particularly strong expression include those important for the gating and processing of sensory inputs, motor control, learning and memory, as well as reward circuits 4 . Using a GFP-tagged knock-in allele, we report in this study that Ca-α1T is expressed broadly across the adult fly brain in structures reminiscent of the mammalian T-type Ca 2+ channels. These include sensory neuropils (i.e., the optic and antennal lobes, the antennal mechanosensory and motor centers, the anterior ventrolateral protocerebrum, and the subesophageal zone), motor-associated neuropils (i.e., the central complex), and those associated with learning, memory, and reward (i.e., the mushroom bodies). It is still unclear, however, whether the different isoforms predicted to originate from the Ca-α1T locus will have different biophysical properties or different distributions around the brain.
Considering their broad expression, T-type knockout mice appear healthy and subtle mutant phenotypes emerge only upon close inspection. Sleep, in particular, has become a focal point in the search for a physiological function for the T-type channels. Mammalian T-type Ca 2+ channels may act as sleep stabilizers and may help generate the burst firing necessary for the sleep oscillations of deep NREM sleep. Unfortunately, the three separate mammalian T-type genes all undergo alternative splicing to produce various channel isoforms that each have specific biophysical properties, neuroanatomical and subcellular localizations, and varying abilities to interact with other ion channels. All these variables and more combine to make it difficult if not impossible to define a Scientific RepoRts | 5:17893 | DOI: 10.1038/srep17893 precise physiological role in sleep for T-type channels as a group. Although Ca v 3.1 knockout mice lack the delta oscillations characteristic of deep sleep and show reduced total sleep 1 , when the knockout is limited to the rostral midline thalamus, sleep is still reduced, but delta waves are mildly increased 24 . Another more recent study showed that treatment with the T-type-specific channel blocker TTA-A2 enhances sleep and delta rhythms in wild type mice but not Ca v 3.1/Ca v 3.3 double knockout mice 25 . In other words, manipulation of T-type channels can both enhance and reduce total sleep and deep delta-wave sleep depending on the experimental context.
Although perhaps an underestimate of the actual complexity of the situation, the subtlety of the phenotypes of the homozygous viable Ca v 3 mutant mice are often ascribed to functional compensation among the various Ca v 3.1-3 isoforms 26 . We, therefore, expected that a behavioral investigation of the one and only fly T-type channel, Ca-α1T, would uncover less subtle sleep phenotypes. We were thus surprised to find, that despite its broad and relatively strong expression across adult fly brains, Ca-α1T-null mutants, like the Ca v 3.1-null mice, are homozygous viable and lack any overt phenotypes. Upon closer examination, however, we observed that Ca-α1T-null mutants sleep more than controls, especially in constant darkness.
The reason for this relative specificity in the sleep phenotype caused by Ca-α1T loss-of-function to constant darkness is still unclear. Flies exhibit a burst of activity upon exposure to the early morning light but then sleep , and from throughout the experiment (2 LD + 7 DD, right). In the left and middle panels, data are presented as means ± s.e.m. In the right panel, white, black, and grey bars indicate light phase, dark phase, and subjective light phase, respectively. The dotted line indicates the beginning of constant darkness. The number of flies measured, their rhythmic period, their power of rhythmicity (P-S), and the percentage of rhythmic flies (Rhythmicity) are indicated. a.u., arbitrary unit. The Mann-Whitney U test was used to determine the significance of the period changes (*p < 0.05), while Welch's t-test was used for rhythmic power (***p < 0.001). (b) Transcriptional oscillation of the period gene in Ca-α1T Gal4 under DD conditions. Black and red lines denote w 1118 and Ca-α1T Gal4 , respectively. rp49 was used for normalization. a.u., arbitrary unit. (c) Percentage of lost sleep recovered (% Δ Sleep) over a 12 hr period after 24 hours of mechanically-induced sleep deprivation. w 1118 (n = 35) and Ca-α1T Gal4 (n = 33). Statistical significance was determined using the Student's t-test. ns, not significant. Data are presented as means ± s.e.m.
Scientific RepoRts | 5:17893 | DOI: 10.1038/srep17893 through most of the rest of the day. Since control flies show less sleep during subjective daytime under continuous darkness than under the light phase of light-dark conditions (Fig. 3c), it is clear that light exposure can also have sleep-promoting effects. Through a series of imaging experiments, Shang et al. reported that although dopamine (DA) is potently wake-promoting, light exposure can suppress this action of DA at least partly by causing the up-regulation of the inhibitory DA receptor D2R in PDF neurons, which are themselves wake-promoting 27 . This modulation of the wake-promoting PDF neurons by light may help explain why the Ca-α1T loss-of-function phenotype is biased toward continuous dark conditions if Ca-α1T functions downstream of the PDF neurons. It would mean the responsible Ca-α1T-positive neurons are also modulated by light.
We were able to replicate the increased sleep phenotype of Ca-α1T-null mutants via pan-neuronal knock-down of Ca-α1T, but we were unable to further narrow the cause of this phenotype to a more specific neuronal subpopulation. This was in spite of numerous attempts with neuronal Gal4 driver lines ranging from broadly expressed enhancer traps and neurotransmitter Gal4 drivers to much more narrowly expressed neuropeptide drivers. This difficulty suggests Ca-α1T may function in novel sleep circuits.  . White, grey, and black bars denote UAS-Ca-α1T-IR/+ , Gal4/+ and Gal4 > Ca-α1T-IR, respectively (n = 21-83). PI, pars intercerebralis, MB, mushroom body. Data are presented as means ± s.e.m. Statistical significance was determined using Welch's ANOVA followed by the Games-Howell post hoc test. ns, not significant. *p < 0.05, **p < 0.01, ***p < 0.001.
In addition to their sleep phenotype, Ca-α1T-null mutants also have a circadian phenotype: an elongated circadian period and a reduction in rhythmic power. It is difficult to say, however, whether these altered circadian parameters are independent of or secondary to the sleep phenotype. Rhythmic power is proportional to the magnitude of the changes in activity level and the regularity with which they occur. Since the increased sleep observed in the Ca-α1T-null mutants does reduce the change in overall activity level between subjective day and subjective night, the increased sleep must also cause a reduction in rhythmic power.
The length of time animals spend sleeping is controlled by both the circadian clock and by a homeostatic drive to sleep that is proportional to time spent awake. Thus, most "sleep mutants" described so far have had defects in one or the other-they are either circadian sleep mutants or homeostatic sleep mutants. After 24 hours of mechanically-induced sleep deprivation, we observed that Ca-α1T-null mutants re-gain slightly more of their lost sleep than control flies (Fig. 4c), although the increase was not statistically significant. This suggests that, in addition to their circadian phenotype, Ca-α1T-null mutants may also have a slightly stronger homeostatic drive to sleep than controls. Although neither the circadian phenotype nor the homeostatic phenotype are particularly strong, together they produce a robust increase in sleep.
The "three channel" compensation hypothesis in mice may yet turn out to be correct, but our results in flies suggest that other factors-isoform-specific differences, differences related to protein-protein interactions, or even something completely unforeseen-may allow mice and flies lacking these broadly expressed and highly conserved ion channels to still function remarkably well. It will be interesting to see whether future studies focused on the technically demanding study of isoform-specific expression patterns and isoform-specific rescues in both mice and flies will clarify how T-type channels can at various times and in various contexts both enhance and reduce sleep. Cloning Ca-α1T. We generated a full-length Ca-α1T (CG15899) cDNA by piecemeal PCR amplification. Total RNA extracted from adult heads using Trizol reagents (Invitrogen) was reverse transcribed using RevertAid First Strand cDNA Synthesis Kit (Fermentas). Six adjacent DNA fragments that cover the entire Ca-α1T cDNA were obtained by PCR amplification. Primer sets were designed based on the FlyBase (FB2011_07) annotation for Ca-α1T. Hind III and Xba I sites were inserted at the 5′ end of fragment 1 and 3′ end of fragment 6, respectively. Primer sets: fragment 1 (5′ -CGAGATAAGCTTAAAATGCTGCCACAGCCA-3′ , 5′ -GCATCAGACTACATCGCTGTC-3′ ), fragment 2 (5′ -CTGGACACGCTGCCCATGCTG-3′ , 5′ -TTCCAGCTCCTCCACTTGCAC-3′ ), fragment 3 (5′ -CAACGGTGGCTCCAACAGTCG-3′ , 5′ -CCACTGGCGGAAGCTCATGCC-3′ ), fragment 4 (5′ -GCCACGCCTCTCCAAGATCCG-3′ , 5′ -GACGATAAGAGCGTTTGCACG-3′ ), fragment 5 (5′ -TCTGAAACTAGTCGTGCAAAC-3′ , 5′ -TGGAAGTACTGGACGGTCTGC-3′ ), and fragment 6 (5′ -AATCCCAGCCTGACCAGCTCG-3′ , 5′ -TCTAGATTAGTCCATGGAGGATTGGGGTGA-3′ ). Amplified PCR fragments were sequenced and assembled into pBlueScript II KS (+ ) using sequential restriction enzyme digests. Clones 2 and 3 contained isoform-specific exons. Of the combinations that were amplified by PCR, we chose to proceed to assembling the RB and RC isoforms. We observed frequent, but inconsistent mutations and instances of A to G RNA editing in fragments 3 and 5. To achieve a final Ca-α1T cDNA matching the FlyBase annotation, we reverted one edited site in fragment 3 (5′ -AGTTCAGAGC-3′ ) by site-directed mutagenesis. Since fragment 5 had so many inconsistencies and contained no introns, we used genomic DNA as a template for fragment 5 instead of cDNA. The final assembled full-length cDNAs were cut with HindIII/XbaI and subcloned into pcDNA3-HE3 downstream of the 5′ -UTR from the Xenopus laevis β-globin gene to improve expression in Xenopus oocytes.

Chemicals and preparation of solutions.
Most of the chemicals for electrophysiological recordings were purchased from Sigma-Aldrich (St. Louis, MO, USA). A 100 mM nickel-chloride stock solution was made in deionized water. A series of nickel solutions (in μM: 0.3, 1, 3, 10, 30, 100, 300, 1000, and 3000) were prepared by diluting the stock solution with 10 mM Ba 2+ recording solution (in mM: 10 BaOH2, 90 NaOH, 1 KOH, 5 HEPES, pH 7.4 adjusted with methanesulfonic acid) before every nickel inhibition experiment.

Functional expression of T-type channels in Xenopus oocytes.
Linearized cDNAs encoding rat Ca v 3.1 or Ca-α1T were used as templates for the synthesis of capped cRNAs using T7 RNA polymerase (Ambion, Austin, TX, USA). cRNA concentrations were estimated based on spectrophotometric optical density measurements at 260 nm. Oocyte preparation from female Xenopus laevis and injection of cRNAs was performed as previously reported 22 . GenBank accession numbers: rat Ca v 3.1 (α 1 G), AF027984 6 ; Ca-α1T C isoform, NP001096889.
We used whole cell patch clamp recordings from HEK-293 cells transiently transfected with Ca-α1T to measure tail currents. These recordings were obtained at room temperature using an Axopatch 200A amplifier connected to a computer through a Digidata 1300 A/D converter and controlled with the pCLAMP 9.2 software. Tail currents were recorded in a 10 mM Ba 2+ solution containing the following: 140 mM TEACl, 2.5 mM CsCl, 10 mM BaCl 2 , 1 mM MgCl 2 , 10 mM glucose, and 10 mM HEPES (pH 7.3, adjusted with TEAOH). The pipette solution contained the following: 130 mM CsCl, 10 mM EGTA, 5 mM MgATP, 1 mM NaGTP, and 10 mM HEPES (pH 7.4, adjusted with CsOH). Recording pipettes were prepared from TW-150-3 capillaries (World Precision Instruments, Inc., Sarasota, FL). The pipette resistance was 2.0 ~ 3.0 MΩ . Access resistance was compensated by 70-80% using the compensation circuit and series resistance prediction. Tail current data were filtered at 10 kHz and digitized at 20 kHz. Peak currents and exponential fits were analyzed using the Clampfit software package (Axon instruments, Foster City, CA, USA). The activation and inactivation time constants for the T-type currents elicited by step pulse protocols were estimated by fitting individual current traces with double exponential functions: where A 1 and A 2 are the coefficients for the activation and inactivation exponentials, t is time, and τ 1 and τ 2 are the activation and inactivation time constants, respectively. The smooth curves for channel activation and steady-state inactivation were obtained by fitting the data with a Boltzmann equation: , where V 50 is the potential for half-maximal activation and S act is the slope conductance. Dose-response curves for Ni 2+ inhibition of T-type channel currents were derived by fitting the data using a Hill equation: = / ( + / ) , where B is the normalized block, IC 50 is the concentration of Ni 2+ giving half maximal blockade, and n is the Hill coefficient.

Generation of deletion mutants of Ca-α1T using imprecise P-element excision. EP line G1047
from the Genexel collection was crossed to the transposase line (Dr [1]/TMS,Sb,P[Δ 2-3]). Mosaic-eyed progeny were collected and crossed to an X chromosome balancer (phl12/FM6) to obtain candidate excision lines. These candidates were then verified by PCR and backcrossed to w 1118 for more than 8 generations before continuing to the sleep analyses.
Antibody generation and western blotting. A polyclonal antisera against Ca-α1T was generated using antigen derived from the 302 C-terminal amino acids of Ca-α1T (cloning primers: 5′ -GAATTCCAAATTAATCCAATCCGTA-3′ , 5′ -GCGGCCGCTTAGTCCATGGAGGATT-3′ ). His-tagged antigen was expressed in E. coli, purified and injected into rabbits to generate an immune response (YoungIn Frontier, South Korea). Western blot analyses were performed according to standard protocols using rabbit antisera obtained after the third Ca-α1T antigen boost. β-Actin-specific antibodies (Santa Cruz Biotechnology, sc-47778) were used for the loading control.

Sleep and locomotor behavior analysis. Fly sleep and locomotor behavior was measured with the
Drosophila Activity Monitoring system (Trikinetics). For sleep analysis, 3-4 day-old female flies were placed individually into 65 mm X 5 mm glass tubes with one end filled with 2% agar/5% sucrose food and the other end plugged with cotton. We defined periods of activity as periods with a beam break frequency higher than 1 per minute and periods of sleep as periods during which no beam break occurred for at least 5 consecutive minutes 42 . After one day of habituation in an incubator (25 °C, 60% humidity), we used the "Counting Macro" software 43 to measure sleep over the course of 4 days-2 days of 12 hr:12 hr light-dark conditions and 2 days of continuous darkness. For experiments using the GeneSwitch technique, flies were maintained on normal food containing 500 μM RU486 (M8046, Sigma-Aldrich) dissolved in ethanol (1%) for two days prior to the experiment. Control flies were maintained on normal food containing only ethanol (1%). For the GeneSwitch experiments, flies were placed in 2% agar/5% sucrose food with or without 500 μM RU486. For sleep deprivation, activity monitors with 3-5 day-old female flies were placed in a Sleep Nullifying Apparatus (SNAP) 44 designed to rotate and give a swift mechanical stimulus twice per minute. After three days under 12 hr:12 hr light-dark conditions, flies were sleep-deprived for 24 hr and allowed to recover 12 hr. The percentage of lost sleep recovered (% Δ Sleep) was calculated by subtracting the baseline sleep (i.e., sleep during the light phase immediately before the deprivation day) from the amount of sleep during the recovery period and then dividing by sleep lost. We confirmed that each genotype lost 90% of their baseline sleep during the deprivation period and we included only flies with ≥ 70% sleep lost in the following calculations. For the circadian locomotor analyses, we measured the activity of 1-3 day-old male flies in 30 minute bins and analyzed the data using ClockLab (Actimetrics) and the Counting Macro 45 . Significance level for the χ 2 periodogram was set to α = 0.05. Flies with a power of significance (P-S) ≥ 10 were considered rhythmic.