Neurocalcin regulates nighttime sleep and arousal in Drosophila

Sleep-like states in diverse organisms can be separated into distinct stages, each with a characteristic arousal threshold. However, the molecular pathways underlying different sleep stages remain unclear. The fruit fly, Drosophila melanogaster, exhibits consolidated sleep during both day and night, with night sleep associated with higher arousal thresholds compared to day sleep. Here we identify a role for the neuronal calcium sensor protein Neurocalcin (NCA) in promoting sleep during the night but not the day by suppressing nocturnal arousal and hyperactivity. We show that both circadian and light-sensing pathways define the temporal window in which NCA promotes sleep. Furthermore, we find that NCA promotes sleep by suppressing synaptic release from a dispersed wake-promoting neural network and demonstrate that the mushroom bodies, a sleep-regulatory center, are a module within this network. Our results advance the understanding of how sleep stages are genetically defined.


Introduction
Sleep is a widely conserved behavior that influences numerous aspects of brain function, including neuronal development (Kayser et al., 2014), clearance of metabolic waste (Xie et al., 2013), synaptic plasticity (Havekes et al., 2016;Kuhn et al., 2016;Li et al., 2017;Yang et al., 2014), and complex behaviors (Kayser et al., 2015;Kayser et al., 2014). The fruit fly, Drosophila, exhibits a sleeplike state characterized by immobility, altered posture and elevated arousal threshold during both day and night (Hendricks et al., 2000;Shaw et al., 2000). Similarly to mammals, sleep in Drosophila is regulated by circadian and homeostatic processes (Huber et al., 2004;Liu et al., 2014). Furthermore, just as human sleep can be separated into stages of differing arousal thresholds (REM and three non-REM sleep stages) (Rechtschaffen et al., 1966), sleep in Drosophila also varies in intensity throughout the day/night cycle, with night sleep having a higher arousal threshold relative to day sleep (Faville et al., 2015;van Alphen et al., 2013).
The molecular mechanisms by which sleep is partitioned into stages remain poorly understood. In Drosophila, mutations in a select number of genes modulate either day or night sleep, suggesting that distinct genetic pathways may promote or inhibit these sleep stages (Ishimoto et al., 2012;Tomita et al., 2015). Yet it is still unclear which properties of sleep/wake these genes are influencing, and how the timing of when they affect sleep is controlled.
The identification of new genes selectively impacting day or night sleep will help address such questions. Previously, large-scale screens of EMS-mutagenized (Cirelli et al., 2005;Stavropoulos and Young, 2011), P-element insertion (Koh et al., 2008), or transgenic RNAi knockdown lines (Rogulja and Young, 2012) have been used to identify Drosophila sleep mutants. However, such approaches are highly laborious, requiring screening of thousands of fly lines to identify a limited number of bona fide sleep genes. Thus, targeted screening strategies of higher efficiency may represent a useful complement to unbiased high-throughput, yet low yield, methodologies.
We uncovered a novel sleep-relevant gene in Drosophila using a guilt-by-association strategy. Our approach was based on comparative phenotyping of human and Drosophila mutants of homologous genes, KCTD17/insomniac, both of which encode a Cullin-3 adaptor protein involved in the ubiquination pathway (Mencacci et al., 2015;Pfeiffenberger and Allada, 2012;Stavropoulos and Young, 2011). In humans, a KCTD17 mutation has been associated with myoclonus dystonia, a disorder characterized by repetitive movements, contorted postures and non-epileptic myoclonic jerks in the upper body (Mencacci et al., 2015). In Drosophila, null or hypomorphic mutations in the KCTD17 homolog insomniac result in profound reductions in sleep (Pfeiffenberger and Allada, 2012;Stavropoulos and Young, 2011).
Genotype-to-phenotype relationships arising from conserved cellular pathways can differ substantially between divergent species such as Drosophila and humans (Lehner, 2013;McGary et al., 2010;Wangler et al., 2017). In this context, it is interesting to note that dystonia in humans and sleep in Drosophila are linked by a common cellular mechanism: synaptic downscaling. This process occurs during sleep in both mammals and Drosophila, and is suppressed at cortico-striatal synapses in murine dystonia models (Bushey et al., 2011;Calabresi et al., 2016;Gilestro et al., 2009;Martella et al., 2009;Tononi and Cirelli, 2014). Thus, we hypothesized that homologs of other human dystonia-associated genes might also influence sleep in Drosophila.
To test this hypothesis, we examined whether homologs of dystonia-associated genes influenced sleep in Drosophila. Through this strategy we identified a previously unappreciated role for the HPCA/Hippocalcin homolog Neurocalcin (Nca) in regulating night sleep. Hippocalcin and NCA are neuronal calcium sensors, cytoplasmic proteins that bind calcium via EF hand domains and translocate to lipid membranes via a calcium-dependent myristoylation switch. This in turn alters interactions with membrane-bound proteins such as ion channels and receptors (Braunewell et al., 2009;Burgoyne and Haynes, 2012). In murine hippocampal neurons, Hippocalcin facilitates the slow afterhyperpolarisation (a calcium-dependent potassium current) (Tzingounis et al., 2007), and glutamate receptor endocytosis during LTD Palmer et al., 2005). In humans, rare missense and null mutations in HPCA have been linked to DYT2 primary isolated dystonia, a hyperkinetic movement disorder affecting the upper limbs, cervical and cranial regions (Atasu et al., 2018;Carecchio et al., 2017;Charlesworth et al., 2015). Drosophila NCA has been shown to be expressed in synaptic regions throughout the fly brain (Teng et al., 1994). However, the neuronal and organismal functions of NCA have remained elusive. Here, we demonstrate a role for NCA in suppressing nocturnal arousal and locomotor activity in Drosophila, thus facilitating nighttime sleep.

Identification of neurocalcin as a sleep-promoting factor
Drosophila NCA is highly homologous to the mammalian neuronal calcium sensor Hippocalcin, sharing >90% amino-acid identity (Figure 1-figure supplement 1). To test whether Nca influences sleep or wakefulness we initially used transgenic RNAi. Using the pan-neuronal driver elav-Gal4, we found that neuronal expression of three independent RNAi lines targeting Nca mRNA (kk108825, hmj21533 and jf03398; termed kk, hmj and jf respectively) reduced night sleep but not day sleep in adult male flies housed under 12 hr light: 12 hr dark conditions (12L: 12D) at 25˚C (Figure 1-figure supplement 2A-E), as measured by the Drosophila Activity Monitoring (DAM) system (Pfeiffenberger et al., 2010). In this work we define a Drosophila sleep bout as 5 min of inactivity, the common standard in the field (Pfeiffenberger et al., 2010).
We performed a series of experiments to further validate a specific role of NCA in promoting night sleep. Sleep loss in flies expressing Nca RNAi correlated with significant reductions in Nca expression ( Figure 1-figure supplement 2F). In contrast, expression of the cg7646 locus, which shares 5' regulatory elements with Nca and encodes a neuronal calcium sensor more closely related to mammalian Recoverin than Hippocalcin, was unaffected by Nca knockdown (Figure 1 Sleep architecture in Drosophila is generally studied in 12L: 12D conditions. Interestingly, we found that night sleep in Nca knockdown males appeared even further reduced under short photoperiod conditions (8L: 16D) ( Figure 1A). Similarly to 12L: 12D, in 8L: 16D day sleep was unaffected whilst night sleep was reduced ( Figure 1A-C), due to fragmentation of consolidated sleep bouts during the middle of the night (Figure 1-figure supplement 3). Nocturnal sleep loss in 8L: 16D was again observed in flies expressing the independent hmj and jf Nca RNAi lines in neurons ( Given the limited spatial resolution of the DAM system, which measures activity via a single infrared beam, we undertook a higher resolution analysis of sleep using a video-tracking method -the DART (Drosophila ARousal Tracking) system (Faville et al., 2015). DART recordings confirmed night-specific sleep loss in in Nca knockdown flies housed under 8L: 16D ( Figure 1D-F).
To test whether sleep loss caused by neuronal Nca knockdown flies was due to an indirect effect on the circadian clock, we examined whether Nca knockdown altered circadian patterns of locomotor activity in constant dark (DD) conditions. Importantly, knockdown of Nca in neurons did not alter circadian rhythmicity (Figure 1-figure supplement 5). Furthermore, knockdown of Nca specifically in clock neurons did not affect night sleep (see below). Thus, it is unlikely that sleep loss in Nca knockdown flies is due to circadian clock dysfunction.
To provide further genetic evidence that NCA is a sleep-regulatory factor, we generated three independent Nca null alleles by replacing the entire Nca locus (including 5' and 3' UTRs) with a miniwhite + sequence using ends-out homologous recombination (Baena-Lopez et al., 2013). The miniwhite + is flanked by loxP sites, allowing removal by Cre recombinase and leaving single attP and loxP sites in place of the Nca locus (Figure 1-figure supplement 6A-C). As expected, no Nca mRNA expression was detected in homozygotes for the deleted Nca locus (Figure 1-figure supplement 6D-E). Thus, we term these alleles Nca KO1-3 (Nca knockouts 1-3). Following outcrossing into an isogenic iso31 control background, male homozygotes and transheterozygotes for the three Nca knockout alleles were viable to the adult stage and exhibited normal day sleep but reduced night sleep, as measured by both DAM and DART systems ( Figure 1G-L, Figure 1-figure supplement 7), similarly to Nca knockdown flies.
By examining locomotor patterns in individual flies using the DART system, we found that Nca KO1 males consistently displayed prolonged activity relative to controls following lights-off and frequent bouts of movement even in the middle of the night -a period of quiescence in iso31 controls ( Figure 1M,N). Video-based analysis of waking locomotor velocities revealed that loss of NCA led to a reduction in average locomotor velocity across 24 hr (Figure 1-figure supplement 8A,B). This was primarily driven by reduced locomotor activity during the evening activity peak and following lights-off, suggesting loss of NCA mildly reduces peak levels of activity (Figure 1-figure supplement 8C). In contrast, locomotor velocities during normally quiescent periods of the night were greatly enhanced in Nca KO1 males compared to iso31 controls (Figure 1-figure supplement 8D), consistent with a perturbed sleep state.
Collectively, the above data demonstrate that NCA promotes night sleep in Drosophila and does so by acting in neurons. For simplicity, we use the kk Nca RNAi line and the Nca KO1 knockout line for all subsequent experiments, and refer to these flies as Nca KD (Nca knockdown) and Nca KO (Nca knockout) respectively.

NCA suppresses nighttime arousal
Sleep is characterized by a reduced responsiveness to stimuli (Campbell and Tobler, 1984). Recent studies have shown that responsiveness during sleep stages in Drosophila is dynamically regulated, with night sleep exhibiting a higher arousal threshold relative to day sleep (Faville et al., 2015;van Alphen et al., 2013). Since knockout or knockdown of NCA specifically impacted night sleep, we were interested to test whether NCA might also influence the arousal threshold during the night. To do so, we used the DART system to subject Nca KD flies and respective controls to a mechanical stimulus consisting of five consecutive 50 Hz vibrations of 200 ms duration, each separated by 800 ms, at either Zeitgeber Time (ZT) 4 (the middle of the day) or ZT16 (the middle of the night) in 8L: 16D (see Methods). This paradigm has previously been shown to induce startle responses the majority of white mutant flies sleeping during the day, and a correspondingly smaller proportion when applied during the night (Faville et al., 2015).   At ZT4, we found that the majority of adult males from both control lines exhibited startle responses in response to vibration stimuli, and that Nca knockdown did not significantly alter the arousal threshold at this time point (Figure 2A,B). In contrast, the percentage of Nca KD flies responding to vibration stimulus was significantly higher at ZT16 relative to both control lines ( Figure 2C,D). Furthermore, whereas the percentage of control flies responding to vibration stimulus was significantly higher during the day compared to the night (elav > + and + > kk: p<0.0005, Binomial test with Bonferonni correction for multiple comparisons), there was no significant day/night difference in responsiveness in Nca KD flies (p=0.1). Similar results were also observed in Nca KO flies ( Figure 2E,F). These data suggest that NCA is a molecular regulator of nighttime arousal in Drosophila.

Light-sensing and circadian pathways define when NCA promotes sleep
The night-specificity of sleep loss and heightened arousal in Nca KO and Nca KD flies prompted us to test whether circadian and/or light-sensing pathways determine when NCA impacts sleep. Initially, we examined sleep patterns in Nca KD flies under DD conditions, in which the circadian clock alone distinguishes subjective day from night. Interestingly, in DD robust sleep loss in Nca KD flies was still restricted to the subjective night ( Figure 3A,B). These data suggest that the circadian clock intersects with NCA and demonstrate that sleep loss in Nca KD flies is not simply due to darkness-induced hyperactivity.
To confirm that the circadian clock influences NCA's sleep-promoting role, we analyzed sleep in Nca KD flies under arrhythmic conditions. In a timeless knockout (tim KO ) background in DD (Lamaze et al., 2017), where the clock no longer demarcates subjective day from night, sleep loss in Nca KD flies was observed throughout the 24 hr dark period ( Figure 3C,D). Intriguingly, in a tim KO background in 8L: 16D conditions, significant sleep loss in Nca KD flies was observed during the night ( Figure 3E,F), but not during the day (elav > kk, tim KO vs elav > +, tim KO : p=0.75, Kruskal-Wallis test with Dunn's post-hoc test). Thus, light is also capable of inhibiting sleep loss in Nca KD flies. Consistent with this finding, under constant light (LL) conditions, in which the circadian clock becomes arrhythmic due to light-dependent degradation of Timeless (Hunter-Ensor et al., 1996;Koh et al., 2006;Peschel et al., 2006), sleep loss in Nca KD flies was completely suppressed ( Figure 3G,H). From the above data we draw two conclusions. Firstly, that the circadian clock is not required for NCA to regulate sleep per se, but instead defines when NCA promotes sleep. Secondly, that light-sensing pathways suppress enhanced wakefulness resulting from reduced NCA expression.  We thus sought to determine which light-sensing pathways restrict sleep loss in Nca KD flies to the night. We reasoned that removing relevant photoreceptive molecules, cells or transduction pathways might restore sleep loss in Nca KD flies during LL. Ablation of photoreceptor cells through expression of the pro-apoptotic gene hid (gmr-hid) did not alter sleep in Nca KD flies in LL ( Figure 3I,J). In contrast, using a loss of function allele of cry (cry 02 ), we found that loss of CRY in LL resulted in a small but significant loss of sleep in Nca KD flies ( Figure     dual roles in synchronization of the circadian clock by light and light-dependent regulation of clock cell excitability (Fogle et al., 2011;Stanewsky et al., 1998). One or both of these pathways may therefore modulate the timing of sleep loss in Nca KD flies. However, the reduction in sleep in Nca KD , cry 02 flies in LL is lower in magnitude compared to Nca KD flies in DD or 8L: 16D ( Figures 1B and  3A), suggesting that additional light-sensing pathways act in concert with CRY to inhibit wakefulness in Nca KD flies in the presence of light. The restoration of clock function in cry 02 homozygotes in LL may also contribute to sleep loss in Nca KD , cry 02 flies under LL (Stanewsky et al., 1998).

NCA acts in two neuronal subpopulations to promote night sleep
Does NCA act in restricted neuropil regions to promote night sleep? To address this question, we used transgenic RNAi to knock down Nca expression in sleep relevant neuronal subpopulations defined by numerous promoter-Gal4 driver lines (Figure 4-figure supplement 1). These include clock neurons, neurotransmitter-and receptor-specific subtypes, fan-shaped body, mushroom body (MB), mechano-sensory, and visual pathway neurons (Donlea et al., 2011;Guo et al., 2018;Jenett et al., 2012;Jiang et al., 2016;Joiner et al., 2006;Lamaze et al., 2018;Lamaze et al., 2017;Liu et al., 2014;Pitman et al., 2006;Seidner et al., 2015;Sitaraman et al., 2015). However, in contrast to broadly expressed drivers (elav-, nsyb-and inc-Gal4), Nca knockdown in neurotransmitter-or neuropil-specific subsets was insufficient to significantly reduce night sleep ( These results suggested that NCA might act in multiple neuropil regions to modulate sleep. Consistent with this hypothesis, we generated a series of driver line combinations and found that Nca knockdown using two enhancer-Gal4 lines (R72C01 -an enhancer in the Dop1R1 locus, and R14A05 -an enhancer in the single-minded locus; we refer to these drivers as C01 and A05 respectively) strongly reduced night sleep in 8L: 16D conditions ( Figure 4A The above data indicate that NCA expression in both C01-and A05-neurons is necessary for normal levels of night sleep. Similarly to pan-neuronal Nca KD flies, knockdown of Nca in C01-and A05neurons also resulted in sleep loss during the subjective night in DD ( Figure 4C,D), no sleep loss in LL ( Figure 4E,F), no alteration in daytime arousal threshold ( Figure 4G), and a reduced arousal threshold during the night ( Figure 4H). Thus, we were able to recapitulate the sleep/arousal phenotypes of Nca KD flies by combinatorial Nca knockdown in C01-and A05-neurons.
The A05 enhancer drives expression in approximately 70 neurons (70.3 ± 4.7, n = 3), as quantified using a fluorescent nuclear marker (Barolo et al., 2004). These include a subset of MB Kenyon cells (MB-KCs), a cluster of cell bodies adjacent to the anterior ventrolateral protocerebrum (AVP), and two visual domains: the optic lobe (OL) and anterior optic tubercle (AOTU) ( Figure 5A). The C01 enhancer drives expression in approximately 240 neurons (239.7 ± 7.8, n = 3) which encompass MB-KCs as well as neurons projecting to the MB g-lobes, the antennal mechanosensory and motor center (AMMC), and the superior medial protocerebrum (SMP) ( Figure 5B). Both drivers label additional cells of unknown identity.
The shared expression of C01 and A05 within the MBs raised the possibility that sleep loss in C01/A05 >Nca RNAi flies was due to strong NCA knockdown in neurons labelled by both enhancers. If so, driving Nca RNAi with two copies of either C01 or A05 should mimic sleep loss in C01/ A05 > Nca RNAi flies. However, this was not the case ( Figure 5-figure supplement 1). Thus, NCA is required in two non-overlapping neuronal populations defined by the C01 and A05 enhancers to promote night sleep. Furthermore, since sleep-promoting NCA activity can largely be mapped to approximately 310 neurons but not to wider populations such as cholinergic or GABAergic neurons (Figure 4-figure supplement 1), these results argue that sleep loss in Nca knockdown and knockout flies is not simply due to broad neuronal dysfunction.

NCA functions in the mushroom bodies to regulate sleep and arousal
Detailed examination within MB-KCs using standardized confocal images from the Virtual Fly Brain indicated that the C01 and A05 enhancers label non-overlapping regions of the MB, with C01 expressed in the ab-KCs, and A05 expressed in a'b'-KCs ( Figure 5 and  We were also interested to examine whether NCA might act in the MB-KCs to regulate nighttime arousal threshold as well as sleep. Using the DART system, we found that Nca knockdown in either C01-neurons or in the MB-KCs (using ok107-Gal4) significantly increased the number of flies aroused by mechanical stimuli during the night but not the day ( Figure 6A-D). Since the MB ab-KCs are labelled by both the ok107-Gal4 and C01-Gal4 drivers, the above data collectively suggest that NCA acts within the MB ab-KCs to suppress nocturnal arousal, and that additional circuits within the A05-positive domain are required in concert with C01-neurons (including MB ab-KCs) to drive nocturnal hyperactivity when Nca expression is reduced.

NCA inhibits synaptic output in a dark-dependent manner
We next examined whether NCA influences the excitability of C01-and A05-neurons. To do so, we expressed a genetically encoded fluorescent indicator of neurotransmitter release, UAS-synapto-pHluorin (spH), in C01-and A05-neurons with or without Nca RNAi. spH localizes to synaptic vesicles and increases in fluorescence in a pH-dependent manner upon vesicle fusion with the presynaptic membrane, providing an optical read-out of neurotransmitter release (Miesenbö ck, 2012). We measured spH fluorescence in four neuropil regions prominently labelled by the C01-and A05-drivers: the MB ab-lobes, the antennal mechanosensory and motor center (AMMC), presynaptic innervations of the MB g-lobes, and the superior medial protocerebrum (SMP). At ZT9-11 in 8L: 16D conditions, Nca knockdown in C01-and A05-neurons resulted in significantly enhanced synaptic release from the MB ab-lobes and the AMMC ( Figure 7A,B) but not the MB g-lobe region or the SMP ( Figure 7C,D), demonstrating that NCA inhibits synaptic release from a subset of C01-and A05-neurons and supporting a physiological role for NCA in the MB ab-lobes.
Since Nca knockdown in C01-and A05-neurons reduces night sleep in 8L: 16D but not in LL conditions ( Figure 4A-B,E-F), we were interested to test whether the above increases in synaptic release were suppressed in LL. Indeed, at Circadian Time (CT) 9-11 in LL conditions, Nca knockdown in C01-and A05-neurons did not enhance synaptic release from the MB ab-lobes, the MB g-lobe region or the SMP, and surprisingly, reduced synaptic release from the AMMC ( Figure 7E-H). Thus, light-sensing pathways suppress both sleep loss ( Figure 4E,F) and elevated synaptic release in the MB ab-lobes and AMMC following Nca knockdown in C01-and A05-neurons.

NCA acts in wake-promoting neurons
Our results suggested a model in which loss of NCA causes aberrant excitation of a neural network that promotes wakefulness in the absence of light. This model yields two predictions. Firstly, that artificial activation of C01-and A05-neurons should promote wakefulness. Secondly, that reducing excitability of C01-and A05-neurons should suppress sleep loss in Nca knockdown flies.
To test our first prediction, we stimulated C01-and A05-neurons by expressing the temperaturesensitive cation channel TrpA1 in either neuronal subset or both and shifting flies from a non-activating temperature (22˚C) to an activating temperature (27˚C) (Hamada et al., 2008) (Figure 8A). At  Figure 5 continued on next page the non-activating temperature, over-expression of TrpA1 in either neuronal population or both did not affect sleep ( Figure 8B). At the activating temperature, excitation of A05-neurons did not alter night sleep ( Figure 8C,D). In contrast, excitation of C01-neurons profoundly reduced night sleep ( Figure 8C,D) as well as day sleep ( Figure 8C). Interestingly, simultaneous activation of C01-and A05-neurons further reduced night but not day sleep relative to activation of C01-neurons alone, despite activation of A05-neurons alone having no impact on sleep in 8L: 16D ( Figure 8C,D). C01and A05-neurons thus synergistically interact to modulate night sleep. To test our second prediction, we over-expressed a non-inactivating outward rectifying potassium channel (dORKDC2) in C01-and A05-neurons with and without Nca knockdown via RNAi. Here, expression of dORKDC2 is predicted to suppress neuronal firing by hyperpolarizing the resting membrane potential (Nitabach et al., 2002;Park and Griffith, 2006). Silencing C01-and A05-neurons with dORKDC2 in an otherwise wild type background did not alter day or night sleep levels ( Figure 8E,F; p>0.99 compared to dORKDC2/+controls, Kruskal-Wallis test with Dunn's post-hoc test). However, consistent with the above prediction, expression of dORKDC2 in concert with Nca RNAi significantly suppressed night sleep loss relative to male flies expressing Nca RNAi alone or alongside an innocuous transgene (UAS-FRT-stop-FRT-GFP) (p<0.0005). Thus, NCA promotes night sleep by limiting synaptic output from arousal-and wake-promoting neurons within the C01-and A05-Gal4 domains that include the MB ab-KCs.

Discussion
Human sleep can be partitioned into stages characterized by unique electroencephalographic signatures and differing arousal thresholds (Rechtschaffen et al., 1966;Rechtschaffen and Kales, 1968). Across the day/night cycle, Drosophila sleep is similarly characterized by dynamic alterations in arousal threshold, with day sleep associated with lower arousal thresholds relative to night sleep (Faville et al., 2015;van Alphen et al., 2013). However, molecular pathways underlying distinct sleep stages are poorly defined. Here we demonstrate a role for the neuronal calcium sensor NCA as a regulator of nocturnal sleep and arousal, thus providing a novel entry point to address this issue.
Previous genetic screens have identified an array of sleep-promoting factors in Drosophila (Tomita et al., 2017). However, despite extensive circuit analyses, the complete neural substrates in which these factors function have yet to be determined (Afonso et al., 2015;Rogulja and Young, 2012;Shi et al., 2014;Stavropoulos and Young, 2011;Tomita et al., 2015;Wu et al., 2014). Our results are consistent with these findings and offer a tentative explanation for the difficulties in defining circuit requirements for sleep-relevant proteins in Drosophila. We show that NCA is not required within a single cell-type or neuropil region to inhibit nighttime arousal and wakefulness. Instead, sleep-relevant NCA activity is necessary within two distinct domains of the Drosophila nervous system defined by the A05-and C01-Gal4 drivers (Jenett et al., 2012).
Ex vivo imaging demonstrates that Nca knockdown enhances synaptic output from subsets of C01-and A05-neurons innervating the MB ab-lobes and the AMMC. Reversing this effect via dORKDC2-mediated electrical silencing suppresses sleep loss in Nca knockdown flies, suggesting that enhanced synaptic output from C01-and A05-neurons via drives nighttime wakefulness. We note that while dORKDC2 expression does not grossly effect the development or axonal guidance of particular clock neurons in Drosophila (Nitabach et al., 2002), prior work has shown that potassium channel overexpression can reduce the viability of mammalian hippocampal neurons (Nadeau et al., 2000). Thus, we cannot entirely rule out an effect of dORKDC2 expression that is secondary to electrical silencing. However, adult-stage excitation via heat-activated TrpA1 channels reveals a clear capacity of C01-neurons to promote wake during both day and night, whereas A05-neurons promote nighttime wakefulness only when C01-neurons are concurrently activated. Since this thermogenetic approach avoids unforeseen effects of chronic alterations in excitability on cellular processes , the above data collectively support a model in which reduced NCA activity in C01-and A05-neurons causes a mild elevation in neurotransmitter release from neuronal subsets within the C01-and A05-domains. Reduced NCA activity in C01-or A05-neurons alone is insufficient to promote wakefulness. Yet when NCA expression is inhibited in C01-and A05-neurons simultaneously, the resulting enhancement of synaptic output within this wider network is sufficient to reduce night sleep. While the precise identities of the wake-promoting circuits within the C01-and A05-domains remain enigmatic, our data suggests a role for NCA in the MB ab-lobes in suppressing arousal during the night. The MB-KCs have been shown to exert a multifaceted influence on Drosophila sleep (Joiner et al., 2006;Pitman et al., 2006;Sitaraman et al., 2015). Recent data has shown that thermo-genetic activation of MB ab-lobes does not affect sleep levels (Sitaraman et al., 2015). Similarly, we find that Nca knockdown in MB-KCs or in C01-neurons (which overlap in the MB ab-lobes) does not impact sleep in 8L: 16D. Nonetheless, either manipulation is sufficient to reduce the arousal threshold in the context of a mechanical stimulus. Thus, NCA plays dual functions in modulating arousal and wakefulness, likely by acting in distinct circuits within the fly brain.
Two questions arise from these results. Firstly, how might NCA inhibit synaptic output? The mammalian NCA homolog Hippocalcin modulates neuronal excitability and plasticity through multiple pathways. Hippocalcin facilitates NMDA receptor endocytosis during LTD and gates the slow afterhyperpolarisation, a calcium-activated potassium current controlling spike frequency adaptation (Andrade et al., 2012;Jo et al., 2010;Tzingounis et al., 2007). Recent data suggest that Hippocalcin also inhibits calcium influx through N-and P/Q-type voltage-gated calcium channels (Helassa et al., 2017). Given the strong homology between Hippocalcin and NCA, it will be intriguing to test whether NCA limits excitatory synaptic input and reduces spike frequency and/or neurotransmitter release through similar pathways in Drosophila. Indeed, cell-type-specific expression of homologous NCA-binding proteins may explain why synaptic output is enhanced in only a subset of C01-and A05-neurons following Nca knockdown, despite previous results showing that NCA is broadly expressed in the Drosophila brain (Teng et al., 1994).
Secondly, how is the sleep-promoting role of NCA limited to the night? Our results show that both internal and external cues regulate when NCA impacts sleep. Nca knockdown reduces sleep solely during the subjective night in DD, but throughout 24 hr in DD when the circadian clock is disrupted. Thus, our data demonstrate a role for the clock in timing when NCA promotes sleep. However, light also acts in parallel as an environmental signal capable of suppressing enhanced wakefulness when NCA activity is reduced, in part through the CRY photoreceptor. At the circuitlevel, our results suggest that constant light suppresses increased neurotransmitter release from neurons in the MB ab-lobes and AMMC following Nca knockdown, further supporting a role for the MB ab-KCs as a component of the neural network through which NCA influences sleep and suggesting a potential contribution from neurons innervating the AMMC. Elucidating the identity of clock-and light-regulated circuits (including CRY-expressing neurons) that gate when and whether NCA promotes sleep will prove a fruitful avenue of future research. More broadly, our work provides a framework to study how complex interactions between genes, neural circuits and the environment influence a critical behavior such as sleep. A05 > spH, kk). Flies were housed in LL conditions, in which Nca knockdown in C01-and A05-neurons has no effect on sleep levels. In each panel, representative confocal images of spH fluorescence (left) and mean fluorescent intensity (right, normalized to the mean of C01/A05 > spH controls) are shown. Dots within dot plots represent individual brain hemisphere measurements. A-D: n = 22-24. E-H: n = 15-18. Neuropil regions are noted. MB: mushroom body. AMMC: antennal mechanosensory motor center. SMP: superior medial protocerebrum. ns -p>0.05, *p<0.05, **p<0.01, ***p<0.001, Mann-Whitney U-test. DOI: https://doi.org/10.7554/eLife.38114.026 The following source data is available for figure 7: Source data 1. Normalized synaptopHluorin fluorescence in specified neuropil regions (see Figure 7) in a wild-type background or following Nca knockdown in C01-and A05-neurons, in either 8L: 16D or in constant light (LL  Night sleep (mins) The following source data is available for figure 8: Source data 1. Sleep levels following excitation or inhibition of C01-and A05-neurons (simultaneously or in isolation), either in a wild type background or in parallel to Nca knockdown. Fly husbandry  (16130), cg7646 RNAi 1 (7646R-1) and cg7646 RNAi 2 (7646R-2). The remaining lines obtained from the Bloomington stock center are part of the Janelia Flylight collection with identifiable prefixes: R23E10-Gal4, R55B01-Gal4, R52H12-Gal4, Hdc-Gal4 (R17F12-Gal4), R14A05-Gal4, R72B05-Gal4, R72B07-Gal4, R72B08-Gal4, R72B11-Gal4, R72C01-Gal4, R72C02-Gal4, R78B07-Gal4, R91A07-Gal4, and R88A06-Gal4. The following lines were generous gifts from Kyunghee Koh: elav-Gal4, nsyb-Gal4, tim-Gal4, TUG-Gal4 and cry-Gal4:16;Joerg Albert: nompC-Gal4 (Kamikouchi et al., 2009) and Nicolas Stavropouplos: inc-Gal4:2 (Stavropoulos and Young, 2011). ppk-Gal4 and TrpA1-CD-Gal4 were described previously (Zhong et al., 2012). GMR-hid, tim KO and cry 02 were previously described in Lamaze et al. (2017). Except for Ddc-Gal4, Trh.1-Gal4, Tdc2-Gal4, nompC-Gal4 and Hdc-Gal4, all Drosophila strains above were either outcrossed five times into an isogenic control background (iso31) or insertion-free chromosomes were exchanged with the iso31 line (hmj21533 and jf03398) before testing for sleep-wake activity behavior. Note: R14A05-Gal4 was initially mislabelled as R21G01-Gal4 in the Bloomington shipment. The mismatch between the image of R21G01 > GFP in the FlyLight database and our immuno-staining data (A05, Figure 5A) led us to clarify the actual identity of the line as R14A05-Gal4 by sequencing genomic PCR product using the following primers pair: pBPGw_ampF: agggttattgtctcatgagcgg and pBPGw_Gal4R: ggcgcacttcggtttttctt.

Generation of Neurocalcin knockout alleles
Null alleles of Nca were generated using homologous recombination as described previously (Baena-Lopez et al., 2013). Briefly, genomic DNA was extracted from 20 wild type flies (Canton S) using the BDGP buffer A-LiCl/KAc precipitation protocol (http://www.fruitfly.org/about/methods/ inverse.pcr.html). The 5' (Arm 1) and 3' (Arm 2) genomic regions flanking the Nca coding sequence were PCR amplified via high fidelity DNA polymerase (Q5 high-fidelity 2X master mix, M0492S, NEB) with the following primers: NotI_Arm1F1: gcggccgctaatttgcagctctgcatcg, NotI_Arm1R1: gcggccgcatggtaagaagcacgcaacc, AscI_Arm2F1: ggcgcgccttatgaccgttccaaaacacc, AvrII_Arm2R1: cctaggggctaaatacgttgaccaagc. The corresponding Arm1 and Arm2 fragments (~2.5 kb) were gel purified (Wizard SV Gel and PCR Clean-Up System, A9281, Promega) and cloned into pCR-Blunt II-TOPO vector (Zero Blunt TOPO PCR Cloning Kit,450245,ThermoFisher Scientific), and subsequently sub-cloned via NotI (R3189S, NEB) and AscI/AvrII digestion (R0558S and R0174S, NEB) and T4 ligation (M0202S, NEB) into the pTV cherry vector, a P-element construct containing the miniwhite + marker and UAS-reaper flanked by FRT and I-SceI sites (Baena-Lopez et al., 2013). The sequence identifies of Arm one and Arm two fragments within the pTV cherry vector were verified via Sanger sequencing using the following primers: nca1_f: cagctctgcatcgctttttgt, nca1_3_f: ccctcgcgcatggtacttta, nca1_r: agcgtcacataagttctccca, nca1_4_f: tggacgaaaataacgatggtca, nca1_5_f: agactacttagccatgttttcatact, nca1_2_f: tgacgaagccacaattaaagagtg, nca1_1_f: gcaaccctgttcccctttca, nca2_f: gaccgttccaaaacaccca, nca2_3_f: ttgttgtgcgccacgttttc, nca2_r: acgtatgctccatgattcctct nca2_4_f: tgcaggtcggttaatcaatgc, nca2_5_f: tcaatcgatttggggccagg, nca2_2_f: ccttctccaggctcagcaaa, nca2_1_f: actctgcatttcgataagattagcc. Donor lines containing the pTV vector with Arm1 and Arm2 homologous fragments (pTV_nca1 + 2) were then generated via embryonic injection and random P-element mediated genomic insertions (Bestgene inc CA, USA). To initiate homologous recombination between pTV_nca1 + 2 and the endogenous Nca locus, donor lines were crossed to yw; hs-flp, hs-I-SceI/CyO and the resulting larvae were heat shocked at 48 hr and 72 hr after egg laying for 1 hr at 37˚C. Around 200 female offspring with mottled/mosaic red eyes were crossed in pools of three to ubiquitin-Gal4[3xP3-GFP] males to remove nonspecific recombination events (via UAS-reapermediated apoptotic activity). The crossings were flipped once over and the progeny (~12000 adults) was screened for the presence of red-eyed and GFP-positive flies. Three independent GFP + redeyed lines (ko1, ko2, and ko3) were identified. The exchange of endogenous Nca locus with pTV_nca1 + 2 fragments was confirmed by detecting a 2.6 kb PCR product (Figure 1-figure supplement 1C) in the genomic DNA samples of the above three lines (pre-digested by EcoRI/NotI) using the following primer pairs: ncaKO-F2: tgggaattgactgatacagcct; ncaKO-R2: ggcactacggtacctgcat. ncaKO-F2 matches to the region between 24 bp and 2 bp upstream of Arm1 and ncaKO-R2 overlaps with attP site ( Figure 1A). The absence of endogenous Nca mRNA in ko1 flies was confirmed by standard and quantitative RT-PCR (Figure 1-figure supplement 5D,E; also see below). The min-white + cassette and majority of pTV vector sequences were further removed from the ko1 genome via Cre-loxP recombination ( Figure 1A). This 'Cre-out' strain was then backcrossed five times to a Nca A502 line (where A502 is a P-element insertion two kbp upstream of the Nca CDS) that was outcrossed previously into the iso31 background (see Fly husbandry section). Before testing for changes in sleep/wake behaviour, the resulting line, termed Nca knockout (Nca KO1 ), was lastly verified by sequencing a 576 bp genomic PCR product (using primer pair: nca1_5_f and nca2_r), confirming the absence of Nca CDS sequence and the insertion of an attP site in the Nca locus. Two independent 'Cre-out' lines derived from the ko2 and ko3 alleles were also outcrossed to Nca A502 for two generations (Nca KO2 and Nca KO3 ) and tested for sleep-wake behaviour.

RNA extraction and quantitative PCR
For RNA extractions, 10-20 fly heads per genotype were collected with liquid nitrogen and dry ice. Total RNA was extracted using TRIzol reagent following manufacturer's manual (Thermo Fisher Scientific). cDNA was reverse transcribed from 250 or 500 ng of DNase I (M0303S, NEB) treated RNA via MMLV RT (M170A, Promega). A set of five or six standards across 3125-fold dilution was prepared from the equally pooled cDNA of all genotypes in each experiment. Triplicated PCR reactions were prepared in 96-well or 384-well plates for standards and the cDNA sample of each genotype (20-to 40-fold dilution) by mixing in Power SYBR Green Master Mix (Thermo Fisher Scientific) and the following primer sets: ncaqF2: acagagttcacagacgctgag, ncaqR2: ttgctagcgtcaccatatggg; cg7646F: gcctttcgaatgtacgatgtcg, cg7646R: cctagcatgtcataaattgcctgaac or rp49F: cgatatgctaagctgtcgcaca, rp49R: cgcttgttcgatccgtaacc. PCR reactions were performed in Applied Biosystems StepOne (96-wells module) or QuantStudio 6Flex instruments (384 wells module) using standard thermocycle protocols. Melting curve analysis was also performed to evaluate the quality of the PCR product and avoid contamination. The Ct values were exported as csv files and a standard curve between Ct values and logarithm of dilution was calculated using the liner regression function in Graphpad Prism. The relative expression level for Nca, cg7646 and rp49 of each sample were estimated by interpolation and anti-logarithm. The expression levels of Nca and cg7646 for each genotype were further normalized to their respective average rp49 expression level. Statistical differences between the normalized expressions levels of each genotype were determined by Mann-Whitney test or Kruskal-Wallis test with Dunn's post-hoc test using Graphpad Prism.

Sleep-wake behavioral analysis
Three to five day old male or virgin female flies were collected and loaded into glass tubes containing 4% sucrose and 2% agar (w/v). Sleep-wake behavior was recorded using the Drosophila Activity Monitor (DAM, TriKinetics inc MA, USA) system or Drosophila ARousal Tracking (DART, BFKlab, UK) in the designated LD regime (12L: 12D, 8L: 16D, DD or LL) at 25˚C. Behavioral recordings from the third day of the given LD/DD/LL regime were then analyzed. All flies were entrained to 12L: 12D prior to entering designated LD regimes. For ectopic activation experiments involving UAS-TrpA1, flies were cultured in 18˚C during development and then entrained to 8L: 16D at 22˚C before entering 8L: 16D condition at 27˚C. Drosophila activity (or wake) is measured by infra-red beam crosses in DAM or by direct movement tracking in DART. A sleep bout is defined by 5 min of inactivity (where inactivity is defined as no beam crosses during 1 min in the DAM or less than 3 mm movement in 5 s in the DART). As a readout of the arousal threshold at ZT4 and ZT16, we measured the proportion of immediate movement initiation in sleeping fly populations (flies that had been immobile for >5 mins before stimulus) upon 5 s of vibration stimuli (five 200 ms 50 Hz pulses with 800 ms intervals) provided by the motors installed within DART system. The csv output files with beam crosses (DAM) or velocity data (DART) were processed by a customized Excel calculators (Supplementary file 1) and R-scripts (https://github.com/PatrickKratsch/DAM_analysR) to calculate the following parameters for individual flies: Onset and offset of each sleep bout, sleep bout length, day and night sleep minutes, daily total sleep minutes, and daily sleep profile (30 min interval).

Analysis of circadian rhythm strength
An established MATLAB based tool, Flytoolbox, was used for circadian rhythmicity analysis (Levine et al., 2002a;Levine et al., 2002b). Flies from control and experimental genotypes developed and eclosed under 12L: 12D conditions (25˚C). After 3 days of entrainment in 12L: 12D, adult males were transferred into DAM tubes, and circadian rhythmicity of locomotor activity was assessed over eleven days of constant dark (DD) following one initial day of 12L: 12D within the experimental incubator. The strength of rhythmicity (RI) was estimated using the height of the third peak coefficient in the auto-correlogram calculated for the activity time series of each fly. Rhythmic Statistics values were then obtained from the ratio of the RI value to the 95% confidence interval for the correlogram (2/HN, where N is the number of observations, which correlatively increase with the sampling frequency), in order to determine statistical significance of any identified period (RS is 1).

Immunohistochemistry and confocal microscopy
Adult male flies were anesthetized in 70% ethanol before brains were dissected in PBT (0.1M phosphate buffer with 0.3% Triton-X100) and collected in 4% paraformaldehyde/PBT on ice. The fixation was then performed at room temperature for 15 min before washing 3 times with PBT. The brain samples were blocked using 5% goat serum/PBT for 1 hr at room temperature before incubation with primary antibodies. The samples were washed 6 times with PBT before incubated with Alexa Fluor secondary antibodies in 5% goat serum/PBT at 4˚C over 24 hr. After washing 6 times with PBT, the samples were mounted in SlowFade Gold antifade reagent (S36936, Thermo Fisher Scientific) on microscope slides and stored at 4˚C until imaged using an inverted confocal microscope Zeiss LSM 710. Primary antibody concentrations were as follows: mouse anti-BRP (nc82, Developmental Studies Hybridoma Bank) -1:200; rabbit anti-GFP (Invitrogen) -1:1000; rabbit anti-dsRED (Clontech) -1:2000. Alexa Fluor secondaries (Invitrogen) were used as follows: Alexa Fluor 647 goat anti-mouse IgG -1:500, Alexa Fluor 488 goat anti-rabbit IgG -1:2000, Alexa Fluor 555 goat anti-rabbit IgG -1:2000. For quantification of nuclei number in C01 >red stinger and A05 >red stinger brains, unstained Red-Stinger fluorescence was captured via confocal microscopy. DAPI (Sigma Aldrich) was used to counterstain nuclei (at a dilution of 1:5000). The number of Red-Stinger-positive nuclei in each brain was subsequently quantified using the ImageJ 3D Objects Counter tool, with a variable threshold used to incorporate all of the visible Red-Stinger-positive nuclei. Standardized images from the Virtual Fly Brain can be found in the following files (Milyaev et al., 2012;Manton et al., 2014):

SynaptopHluorin imaging
Synaptic activity of C01/A05 neurons was monitored in ex vivo fly brains using UAS-super-ecliptic synaptopHluorin construct (UAS-spH) (Miesenbö ck, 2012). Adult male C01/A05 > UAS spH or C01/ A05 > UAS spH, kk flies were housed in normal behaviour tubes (see behaviour analysis section) and entrained for 3 days in 8L: 16D or LL conditions at 25˚C. Individual flies of either genotype were carefully captured between ZT/CT9 and ZT/CT11 and fly brains were immediately dissected in HL3 Drosophila saline (70 mM NaCl, 5 mM KCl, 1.5 mM CaCl 2 , 20 mM MgCl 2 , 10 mM NaHCO 3 , 5 mM Trehalose, 115 mM Sucrose and 5 mM HEPES, pH 7.2) at room temperature. Fly brains were transferred into 200 ml HL3 in a poly-lysine treated glass bottom dish (35 mm, 627860, Greiner Bio-One) before imaging using an inverted confocal Zeiss LSM 710 microscope (20x objective with maximum pinhole). Three to five image stacks (12 bits and 16 bits) were taken within two minutes to minimize tissue degradation and to cover the depth of all spH-positive anatomical regions. Z-projections of the image stacks of each brain were generated by ImageJ software before the fluorescent intensity of the indicated neuropil centers was quantified using freely drawn ROIs. Background fluorescence measured by the same ROIs from areas with no brain tissue was then subtracted to obtain the final fluorescent value. Mean fluorescent values of the indicated neuropil regions in each hemisphere were calculated and normalized to the average value of corresponding controls. Medians of the normalized value are compared between genotypes. The statistical difference was determined by Mann-Whitney U-test using Graphpad Prism.

Bioinformatics
Conservation of amino acid residues between Drosophila Neurocalcin and human Hippocalcin was determined using ClustalW2 software for multiple sequence alignment. Amino-acid identity and similarity was visualized using BOXSHADE.