1. Neuroscience
Download icon

Acute control of the sleep switch in Drosophila reveals a role for gap junctions in regulating behavioral responsiveness

  1. Michael Troup
  2. Melvyn HW Yap
  3. Chelsie Rohrscheib
  4. Martyna J Grabowska
  5. Deniz Ertekin
  6. Roshini Randeniya
  7. Benjamin Kottler
  8. Aoife Larkin
  9. Kelly Munro
  10. Paul J Shaw
  11. Bruno van Swinderen  Is a corresponding author
  1. The University of Queensland, Australia
  2. King's College London, United Kingdom
  3. University of Cambridge, United Kingdom
  4. Washington University School of Medicine, United States
Research Article
  • Cited 4
  • Views 13,099
  • Annotations
Cite this article as: eLife 2018;7:e37105 doi: 10.7554/eLife.37105

Abstract

Sleep is a dynamic process in most animals, involving distinct stages that probably perform multiple functions for the brain. Before sleep functions can be initiated, it is likely that behavioral responsiveness to the outside world needs to be reduced, even while the animal is still awake. Recent work in Drosophila has uncovered a sleep switch in the dorsal fan-shaped body (dFB) of the fly’s central brain, but it is not known whether these sleep-promoting neurons also govern the acute need to ignore salient stimuli in the environment during sleep transitions. We found that optogenetic activation of the sleep switch suppressed behavioral responsiveness to mechanical stimuli, even in awake flies, indicating a broader role for these neurons in regulating arousal. The dFB-mediated suppression mechanism and its associated neural correlates requires innexin6 expression, suggesting that the acute need to reduce sensory perception when flies fall asleep is mediated in part by electrical synapses.

https://doi.org/10.7554/eLife.37105.001

Introduction

Most animals sleep, and recent research suggests that some proposed sleep functions may be deeply conserved across all animals (Cirelli, 2009; Kirszenblat and van Swinderen, 2015; Zimmerman et al., 2008). These functions range from cellular stress regulation in nematodes (Hill et al., 2014; Nelson et al., 2014) to synaptic homeostasis and memory consolidation in mammals (Rasch and Born, 2013; Tononi and Cirelli, 2006; Tononi and Cirelli, 2014). Other proposed functions of sleep include maintaining cellular health, by clearing of protein debris for example (Xie et al., 2013). These conserved cellular processes might have been integrated through evolution into distinct stages of the mammalian sleep cycle, to accommodate multiple more recent sleep functions, such as synaptic scaling and memory consolidation, simultaneously (Kirszenblat and van Swinderen, 2015). Although sleep may serve many diverse functions, and although many different molecular processes might be involved, in most animals sleep is understood to result largely from widespread changes in electrical activity throughout the brain. That is, the rapidly reversible switch that allows the aforementioned potential functions to take place probably requires important changes in electrical activity in the brain: initially to decrease behavioral responsiveness and thereby allow some of these processes to proceed without ongoing behavioral interference (Mednick et al., 2011).

In mammals, dis-inhibition of GABAergic sleep-promoting neurons in the vasolateral preoptic nucleus (VLPO) is thought to comprise part of a sleep switch (Morairty et al., 2004; Saper et al., 2010). Interestingly, even the smallest brains appear to have a sleep switch: in the fruit fly Drosophila melanogaster, the dorsal fan-shaped body (dFB) has been suggested to play a role that is analogous to that of the mammalian VLPO (Donlea et al., 2014, 2018; Pimentel et al., 2016). Indeed, activation of these neurons in the fly’s central brain promotes sleep and achieves key sleep functions (Dissel et al., 2015; Donlea et al., 2011). Recently, we have shown that changes in electrical activity in the fly brain are associated with spontaneous sleep transitions, and that activation of dFB neurons causes specific oscillatory signatures in local field potential (LFP) activity recorded from the central brain (Yap et al., 2017). It is not clear, however, how dFB neurons might communicate this activity to the rest of the fly brain to promote sleep rapidly. In addition, dFB-induced sleep has primarily been studied by observing increased periods of behavioral quiescence, rather than behavioral responsiveness to external stimuli. Decreased behavioral responsiveness would appear to be a prerequisite for sleep induction, but it remains unclear whether the same neurons could be involved in both acute loss of responsiveness and promotion of sleep functions (Lebestky et al., 2009), and if so, how the dFB might control both processes simultaneously. While it is understood that sleep is associated with decreased behavioral responsiveness, it is unclear whether the Drosophila dFB plays a larger role in regulating arousal more generally, even in awake animals. Recent research identified a neurochemical mechanism for promoting sleep in Drosophila (Donlea et al., 2018), but it is unknown whether other pathways are employed to link effectively internal sleep pressure signals in the dFB with the acute need to suppress behavioral responsiveness when flies need to sleep.

In addition to neurochemical signaling from sleep centers such as the VLPO, mammalian sleep processes might also involve electrical signaling that is mediated via gap junctions (Coulon and Landisman, 2017; Franco-Pérez and Paz, 2009). Gap-junction-mediated communication appears to be important for the rapid recruitment of much of the mammalian brain into synchronously firing networks (Bennett and Zukin, 2004; Buzsaki, 2006), but it remains unknown whether this is an important aspect of sleep physiology and function. In vertebrates, connexin genes code for a variety of gap junction subtypes (Söhl and Willecke, 2004). Invertebrates such as flies express a family of gap-junction genes that encode proteins called innexins, and Drosophila has eight innexin-encoding loci that have been implicated extensively in the development of the brain and other tissues (Bauer et al., 2005; Stebbings et al., 2002). In the adult fly, there is limited understanding of the role of gap-junction signaling in behavior, but studies have found a role for this signaling in visual processing (Cuntz et al., 2007; Liu et al., 2016a), in escape behavior (Phelan et al., 1996) and in learning and memory (Wu et al., 2011). In this study, we use optogenetics and electrophysiology to investigate the role of the dFB neurons in regulating behavioral responsiveness alongside sleep in the Drosophila model. We then examine how electrical and behavioral readouts of our sleep switch manipulations are affected when we remove innexin6 gap junctions from the dFB.

Results

Correlating sleep duration and behavioral responsiveness

Flies were filmed in the Drosophila ARousal Tracking (DART) platform (Figure 1A) (Faville et al., 2015) to monitor sleep duration and behavioral responsiveness simultaneously (Figure 1B). Sleep duration is measured by well-established inactivity criteria based on >5 min inactivity (Figure 1B, upper panel; Figure 1C) (Shaw et al., 2000), whereas behavioral responsiveness can be measured by tracking how flies respond to a mechanical stimulus, during both sleep or while they are awake (van Alphen et al., 2013). Following a vibration stimulus, responding flies typically increase their locomotion (Figure 1B, lower panel). A stimulus that is delivered hourly provides an estimate of behavioral responsiveness throughout the circadian cycle (Figure 1D), and average responses (mean peak responsiveness, see 'Materials and methods') are typically stronger during the day than during the night (Figure 1E, left and middle panel). Responsiveness metrics therefore complement sleep duration measures (Figure 1E, right): animals are by definition less responsive when they are asleep (Campbell and Tobler, 1984).

Figure 1 with 1 supplement see all
The sleep switch modulates behavioral responsiveness.

(A) Flies in glass tubes were filmed from above and DART was used to track activity and to test behavioral responsiveness using a mechanical vibration. (B) Sleep duration was measured using 5 min inactivity criteria (top panel). Behavioral responsiveness was tested by quantifying the change in fly activity following a vibration stimulus. Following the stimulus (orange line), flies increase their locomotion speed as shown by their displacement in the tube (bottom panel). (C) Mean sleep duration (min/hr) is tracked over a circadian cycle. (D) Fly activity (speed, mm/sec) is plotted for a 24-hr day/night (white and grey, respectively) cycle during which a five-pulse 0.2 s 2.4 g vibration is delivered once per hour. Spikes in activity show timing of the stimuli, and the orange lines highlight three examples. (E) The mean response (speed, mm/s) for all stimuli during the day or night (left panel, black line). Shown in grey is a fitted curve for this average response (see 'Materials and methods'), the peak of which is a measure of the magnitude of response to the stimulus (middle panel). Responsiveness is greater during the day and lower during the night, whereas sleep duration is decreased during the day and increased at night (right panel). (F) Correlation between the peak response speed (mm/s) and sleep duration (min/hr) for wildtype (w2202) flies (n = 225) during the day (yellow R = 0.012, p=0.84) and the night (grey R = −0.372, p<0.0001). (G) R23E10-Gal4 neurons were chronically activated by expressing NaChBac, a bacterial sodium channel. Scale bar = 50 μm, the genotype in this image is R23E10-Gal4/+;UAS-2eGFP/+. (H) Correlation between responsiveness and sleep duration following activation of the dFB (R23E10-Gal4/UAS-NaChBac, n = 51) during daytimes (yellow R = −0.410, p<0.001) and nighttimes (grey R = −0.554, p<0.0001). See also Figure 1—figure supplement 1.

https://doi.org/10.7554/eLife.37105.002

Sleep duration and behavioral responsiveness should be negatively correlated. However, reduced behavioral responsiveness is also a feature of wakefulness, and animals that sleep more are not necessarily less responsive in general, even while awake. To better understand the relationship between these distinct arousal measures, we tracked average sleep duration and average responsiveness in 250 inbred wild-type flies (w2202 [Faville et al., 2015]) over multiple days and nights. Locomotion immediately following the hourly vibrations (in awake or sleeping flies) contributed to the responsiveness metrics, and sleep duration was averaged from unstimulated epochs in the intervening hours. In previous work, we have shown that such hourly probing is not sleep depriving in wildtype flies (van Alphen et al., 2013). We found a surprisingly large amount of individual variability for these responsiveness and sleep metrics (Figure 1F). Sleep duration and behavioral responsiveness were only negatively correlated at night, but not during the day in the wildtype flies (Figure 1F). While this could reflect a qualitative difference between daytime sleep and nighttime sleep (Faville et al., 2015), or might indicate that sleep bouts are generally longer at night (Shaw et al., 2000), this also suggests that sleep duration and responsiveness might be separately controlled, as has been proposed by the authors of a previous study (Lebestky et al., 2009). Sleep-promoting circuits might also regulate responsiveness to external stimuli.

The dFB (Figure 1G) has been identified as a ‘sleep switch’ in the fly brain (Donlea et al., 2011, 2014; Pimentel et al., 2016), but its role in controlling behavioral responsiveness has not been well studied. dFB activation is associated with increased arousal thresholds (Donlea et al., 2011, 2014; Pimentel et al., 2016), but it is unclear whether this is a direct feature of increased dFB activity or a consequence of other processes that result from increased dFB activity. We first examined whether sleep duration and behavioral responsiveness might be simultaneously modulated by these neurons. Previous work has focussed on three Gal4 drivers that induce dFB-driven sleep in flies: 104y-Gal4, C5-Gal4, and R23E10-Gal4 (Donlea et al., 2011, 2014; Pimentel et al., 2016). We activated these sleep-promoting circuits by expressing a bacterial sodium channel, NaChBac (Nitabach et al., 2006). Chronic activation of the R23E10 neurons produced a strong negative correlation between sleep duration and behavioral responsiveness, even for daytime sleep (Figure 1H): flies that slept more tended to respond less to stimuli, day or night. Genetic controls showed only a night-time correlation (Figure 1—figure supplement 1A), as seen in the wildtype background strain (Figure 1F). In contrast to data from R23E10, results with C5 and 104y were less clear: C5 displayed a daytime correlation but not the nighttime correlation, and 104y did not display a daytime correlation (Figure 1—figure supplement 1B,C). This suggests that R23E10, which is a cleaner dFB driver (Jenett et al., 2012), may be more effective as a sleep switch. The R23E10 result also suggests that the dFB might regulate behavioral responsiveness as well as sleep duration, although it is difficult to separate these arousal measures in chronic manipulations.

Optogenetic activation of the sleep switch decreases behavioral responsiveness

Falling asleep is an acute event that happens within seconds in most animals (Campbell and Tobler, 1984), and even flies can be significantly less responsive after only one minute of spontaneous quiescence (van Alphen et al., 2013; Faville et al., 2015). Optogenetic tools in Drosophila provide one way to induce sleep on demand experimentally. We used the R23E10-Gal4 driver to express CsChrimson, a red-light shifted channelrhodopsin (Klapoetke et al., 2014), in order to activate dFB neurons transiently in fly populations monitored by DART. As sleep is necessarily measured over prolonged time periods, we first examined whether prolonged optogenetic activation of the dFB was associated with decreased behavioral responsiveness. We therefore measured sleep duration and behavioral responsiveness in R23E10-Gal4 > UAS CsChrimson flies before, during, and after 12 hr of daytime dFB activation (Figure 2A,C). All-trans retinal (ATR) was fed to flies to enable channelrhodopsin function, and genetically identical but non-ATR-fed flies were used as controls (Figure 2A, black). Flies that were fed ATR significantly increased their sleep duration in response to dFB activation with prolonged red light exposure (Figure 2A,B, blue). Behavioral responsiveness was robust in these flies at baseline (Figure 2C, left panel), but optogenetic activation of the dFB significantly decreased responsiveness to the vibration stimulus (Figure 2C, right panel; Figure 2D, left panel). Responsiveness was not, however, abolished by activation of the dFB, and even appeared to recover somewhat after several hours of red-light exposure (Figure 2C, right panel).

Figure 2 with 3 supplements see all
Behavioral effects of acutely activating R23E10 neurons.

(AD) Effects on sleep and responsiveness following the activation of R23E10 neurons (blue, UAS-CsChrimson/+;R23E10-Gal4/+ with ATR, n = 102) compared to those in control flies (black, no ATR feeding, n = 118). Error bars and shading indicate standard errors of the mean (SEMs) and asterisks indicate significance (****p<0.0001, ns = not significant, t-tests). (A) Mean sleep duration (min/hr) during the period 24 hr before red light activation (left), then during the next 24 hr when red light is delivered for 12 hr during the day (pink shading, right). (B) Comparison of the 12-hr day period without red light (baseline) to the period of red-light activation in terms of sleep duration and bout length. (C) Mean activity (mm/s) for R23E10 activation for the time periods in (A). Spikes in activity show timing of hourly vibration stimuli. (D) Comparison between the 12-hr day period without red light (baseline) and the period of red-light activation for peak responsiveness and sleep intensity. (E) Left, average stimulus response for UAS-CsChrimson/+;R23E10-Gal4/+ with ATR (blue, n = 50) compared to control flies (black, no ATR feeding, n = 48) during red-light activation (Figure 2A–D red shading) in flies that moved in the minute prior to the stimulus (i.e. awake flies). Right, summary histogram (average ± SEM). **p<0.01, t-test. (F) Example activity trace of flies responding to stimuli 15 min apart (gray dashed lines): 1 min CsChrimson activation (red shading) prior to the stimulus is alternated with trials without red light (left panel). One minute of dFB activation is sufficient to decrease responsiveness (right panel, UAS-CsChrimson/+;R23E10-Gal4/+ with ATR n = 83, control n = 80, *p<0.05, t-test). See also Figure 2—figure supplements 1, 2 and 3.

https://doi.org/10.7554/eLife.37105.004

So far, our behavioral responsiveness metric does not differentiate between waking or sleep in flies. When responsiveness is analyzed only for sleeping flies, expressed relative to the proportion of immobile flies that respond (at any level) to the stimulus, this provides a measure of sleep intensity (Figure 2—figure supplement 1A). Wildtype flies sleep more deeply at night than during the day (Figure 2—figure supplement 1B,C), and sleep intensity is not necessarily correlated to sleep duration (van Alphen et al., 2013; Faville et al., 2015). Yet, as might be expected, optogenetic activation of the ‘sleep switch’ increases sleep intensity (Figure 2D, right panel; Figure 2—figure supplement 1D) as well as sleep duration. However, dFB-activated flies are not always asleep, they are often awake. We therefore next investigated whether responsiveness was affected in awake (i.e. walking) flies only (see 'Materials and methods' and Figure 2—figure supplement 1A for explanations on how wakeful responsiveness was determined). We were surprised to find that dFB activation decreases behavioral responsiveness in awake flies as well as in sleeping flies (Figure 2E, right panel). Thus, decreased behavioral responsiveness appears to be a feature of prolonged dFB activation, even when sleep reverts to wakefulness. Decreased responsiveness was not due to sluggish locomotion in awake animals: the walking speed of awake dFB-activated flies was not different from that of controls immediately preceding the vibration stimulus (Figure 2E, left panel, green shading). dFB activation therefore reduces behavioral responsiveness in awake as well as in sleeping flies.

To confirm that this loss of behavioral responsiveness in awake flies is an acute effect of dFB activation and not just an indirect effect of prolonged sleep induction, we repeated these experiments with shorter time periods (5 min) of dFB activation. We probed for responsiveness after only 1 min of red-light exposure. In doing so, we found that behavioral responsiveness was significantly reduced after 1 min of red light, and that responsiveness was rapidly restored after the red light was turned off (Figure 2F), that is we observed no evidence of sleep inertia following 5 min of dFB activation. To further confirm that this effect was distinct from sleep, we also analyzed only flies that were walking in the minute preceding the vibration stimulus, and observed the same significant effect (Figure 2—figure supplement 2). Together with our prolonged activation experiments, this suggests a dissociation between dFB’s role in sleep promotion and behavioral responsiveness. While more prolonged dFB activation certainly increased sleep duration and sleep intensity (Figure 2B,D; Figure 2—figure supplement 1D), behavioral responsiveness remains suppressed when dFB-activated flies are awake and can be acutely suppressed in walking flies. As a constant light activation might be an unnatural stimulation regime for the dFB neurons, we repeated our experiments with a sparser activation stimulus, pulsed 5 ms red light (1 Hz), and found that behavioral responsiveness was similarly suppressed (Figure 2—figure supplement 3). This shows that different stimulation regimes produce a similar effect and confirms that activation of the sleep switch suppresses behavioral responsiveness.

Optogenetic activation of the sleep switch causes a rapid and reversible increase in membrane potential of dFB neurons

To better understand the mechanisms underlying these acute effects, we investigated the electrophysiological properties of the dFB neurons that contributed to the observed change in behavior. We used whole-cell patch-clamp electrophysiology in current clamp mode to record the membrane potential of the dFB cell bodies in behaving flies (Figure 3A) (see 'Materials and methods'). dFB neurons were genetically labeled with green fluorescent protein (GFP) to guide recordings from the cell soma visually (Figure 3B, top). As shown previously (Donlea et al., 2014; Pimentel et al., 2016), the spiking activity of dFB soma responded to positive current steps by increasing their firing rate (Figure 3B, bottom left). However, not all neurons shared the same spontaneous firing properties, with ~20% of the recorded cells identified as non-spiking (even following current injection), whereas the remaining cells that did spike were observed to exhibit various combinations of single spikes and burst spiking patterns (Figure 3B, bottom right).

Figure 3 with 2 supplements see all
Electrophysiological effects of acutely activating R23E10 neurons.

(A) Setup for recording in vivo adult Drosophila electrophysiology with whole-cell patch clamp (orange). (B) dFB neuron schematic showing whole-cell recordings targeted to R23E10-Gal4 cell bodies. Injecting current in a stepwise manner causes firing in these neurons (bottom left panel), which are heterogeneous in their endogenous firing patterns (bottom right panel). Scale bars indicate 10 mV and 100 ms. (C) Example traces (left) of a CsChrimson-expressing (UAS-CsChrimson/+;R23E10-Gal4/+ with ATR) spiking cell (top) and non-spiking cell (middle), and of a non-CsChrimson-expressing (R23E10-Gal4/+ with ATR) cell (bottom) when exposed to constant red light (red shading). Superimposed traces of the corresponding types from multiple cell recordings (right panel, top to bottom: n = 10, n = 6, n = 6). Solid red lines indicate mean values. (D) Boxplots show median membrane potentials for CsChrimson-expressing cells (blue) and non-CsChrimson expressing cells (gray, n = 6) before, during and after constant light stimulation (***p<0.001, Friedman test with Dunn’s multiple comparisons to pre-stimulus condition). See also Figure 3—figure supplements 1 and 2.

https://doi.org/10.7554/eLife.37105.008

Despite the heterogeneity in the firing pattern of the dFB neurons, CsChrimson-expressing cells all responded to a red-light stimulus in the same manner, by rapidly increasing their membrane potential upon activation, followed by a gradual return to its resting membrane state after the light was turned off (Figure 3C,D). Induced spike bursts occurred 18.75 ± 3.07 ms apart in a subset of cells. Exposure to 1 Hz light pulses (5 ms each) instead of continuous exposure produced corresponding action potentials in some spiking cells (Figure 3—figure supplement 1A). On average, 1 Hz pulsed activation also increased the membrane potential in all CsChrimson-expressing dFB cells (Figure 3—figure supplement 1B). The robustness of this effect, despite the heterogeneity in the spontaneous firing pattern across cells and the different stimulus regimes, suggests that the acute dFB activation effects on behavioral responsiveness could be associated with increased membrane potential rather than with any specific increased spiking activity, although it remains possible that the subset of spiking cells is key here. Notably, the post-stimulation membrane potential (recorded only 1 min after the red light was turned off) was not significantly different from baseline (Figure 3D; Figure 3—figure supplement 1B), indicating that the acute effect on the membrane potential was not sustained for long after the stimulation. Red light had no effect on spiking or the membrane potential in R23E10-Gal4/+ animals that had been fed ATR but that lacked channelrhodopsin (Figure 3D, Figure 3—figure supplement 1B, gray boxplots), or in control R23E10-Gal4/UAS-CsChrimson animals that had not been fed ATR (Figure 3—figure supplement 2). In a subset of recordings, 1 Hz light pulses reliably evoked secondary and even tertiary spikes between 100 and 500 ms after the first evoked spike (Figure 3—figure supplement 2), suggesting a reverberation in the circuit.

Gap-junction localization in the dFB

An increased input resistance and membrane time constant in dFB cells has been shown to be associated with increased sleep pressure (Donlea et al., 2014), so our optogenetic results are consistent with the idea that these cells form part of a sleep homeostat that is regulated by changing membrane potential levels (Pimentel et al., 2016). Recent work shows that output from the dFB cells regulates sleep duration by inhibiting a number of other systems in the central brain (Donlea et al., 2018; Liu et al., 2016b). These communication channels are probably chemical in nature, but it is not clear whether behavioral responsiveness is also controlled through these downstream synaptic circuits. We therefore next acutely manipulated synaptic release from the dFB. Interestingly, transiently increasing synaptic activity in dFB neurons (measured using temperature-sensitive syntaxin3-69 [Lagow et al., 2007; Kottler et al., 2013]) had no significant effect on behavioral responsiveness, or on sleep (Figure 4—figure supplement 1A). Transiently decreasing synaptic activity (by using the temperature-sensitive shibire [Kitamoto, 2001]) in the dFB decreased responsiveness, but equally so in the genetic controls (Figure 4—figure supplement 1B). A lack of clear effects using these opposing synaptic manipulations suggests that (fast) chemical neurotransmission might not be a relevant mechanism employed by these neurons to control behavior acutely, or at least that these acute synaptic manipulations do not affect our behavioral readouts. This prompted us to explore whether electrical synapses might be involved instead. A previous study has found expression of the gap junction gene, inx6, in the fan-shaped body (Wu et al., 2011), suggesting that these neurons might communicate electrically with other neurons potentially to regulate behavioral responsiveness.

We used an antibody for the INX6 protein to first visualize its location in the fly brain. Expression was seen most strongly in a dorsal layer of the fan-shaped body and in large cells in the pars intercerebralis (PI) (Figure 4A, top and middle panels, Figure 4—figure supplement 2A). Co-labeling with GFP driven by R23E10-Gal4 revealed a significant overlap in the relevant dFB neurons (Figure 4B, bottom panel; a co-localization of 47.5% was found, see 'Materials and methods'). However, we did not find any evidence of co-localization in the dFB cell soma (Figure 4—figure supplement 2B). Dye-fill experiments confirmed that the dFB neurons are coupled to other cells via gap junctions (Figure 4—figure supplement 2C; Video 1), although the extent of this electrically coupled network remains unclear as only the large PI cells above the dFB revealed any reliable dye coupling in our experiments.

Figure 4 with 2 supplements see all
INX6 dFB localization.

(A) INX6 antibody staining (red) in wildtype Canton-S flies at 20x (top) and 60x (middle). White arrow indicates the location of the dFB and white dashes outline it. Staining using the secondary antibody alone (bottom, 20x) shows no detectable reactivity in the dFB. Scale bars = 50 μm. (B) INX6 antibody staining (red, middle) with GFP antibody staining in flies expressing GFP in R23E10 neurons (UAS-GFP/+;R23E10-Gal4/+, green, top). Overlap between these regions (bottom) indicates the presence of INX6 in these neurons (47.5% co-localization, see 'Materials and methods'). Scale bars = 25 μm. See also Figure 4—figure supplements 1 and 2.

https://doi.org/10.7554/eLife.37105.011
Video 1
3D reconstruction of dye coupling experiment.
https://doi.org/10.7554/eLife.37105.014

Gap junctions in the dFB regulate both sleep duration and behavioral responsiveness

To test the role of INX6 in sleep, we used a working RNAi construct (Figure 5—figure supplement 1A,B) to knock down INX6 in dFB neurons. Flies expressing the INX6 RNAi construct in the dFB showed a decrease in both day and night sleep (Figure 5A). We next looked at responsiveness to mechanical stimuli in these flies. Flies slept less deeply and were more responsive to stimuli when INX6 was downregulated in the dFB (Figure 5B). Closer examination of sleep intensity as a function of time asleep (van Alphen et al., 2013) showed that INX6-knockdown flies could still achieve deeper sleep stages (e.g. after 16–20 min immobility, Figure 5C), although these flies slept more lightly in general. This suggests that INX6 plays a role in suppressing behavioral responsiveness during lighter sleep stages, whereas during deeper sleep, the role of INX6 might be less important. Circadian influences also clearly play a major role: responsiveness is always lower at night than during the day. In addition, INX6 expression in different subsets of dFB neurons could also contribute to the behavior, as we observed only partial INX6 overlap with R23E10 (Figure 4B, bottom). To confirm that these behavioral effects were not a consequence of altered developmental pathways, we downregulated INX6 in adult flies by using a temperature-sensitive suppressor of Gal4, Gal80TS (McGuire et al., 2004). Knocking down INX6 in the adult dFB significantly increased daytime behavioral responsiveness and decreased daytime sleep intensity, but had no significant effect on sleep duration (Figure 5—figure supplement 2). These results suggest that INX6 is important for communicating dFB activity levels to regulate behavioral responsiveness. We therefore next questioned whether INX6 knockdown altered electrical readouts associated with acute dFB activation.

Figure 5 with 2 supplements see all
INX6 effects on behavior.

(A) Mean sleep duration (min/hr) and (B) mean peak responsiveness (mm/s) and sleep intensity (% responding) for R23E10-Gal4/+;UAS-INX6-RNAi/+ (red, n = 85) compared to R23E10-Gal4/+ (gray, n = 83) and UAS-INX6-RNAi/+ (black, n = 84) controls. (C) Sleep intensity for inactivity bins of 5 min for the first 30 min of inactivity. See also Figure 2—figure supplement 1. Shading and error bars indicate SEM, asterisks indicate significance (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, two-way ANOVA, adjusted for multiple comparison (Dunnett). See also Figure 5—figure supplements 1 and 2.

https://doi.org/10.7554/eLife.37105.015

Gap junctions promote LFP activity in the dFB

We recorded neural activity from INX6 knockdown flies whose dFB could be optogenetically activated. A dual-electrode setup was used to record intracellularly from the dFB cell body while a second electrode simultaneously positioned within in the dFB structure recorded extracellular local field potentials (LFPs) (Figure 6A,B). Whole-cell recordings revealed that reduced INX6 expression in the dFB (Figure 5—figure supplement 1B) did not impair CsChrimson-mediated cell activation: a significant increase in membrane potential was still observed with constant red light as well as with 1 Hz pulses (Figure 6C,D). This shows that the increase of the membrane potential in activated dFB neurons does not depend upon INX6 expression in these cells. Nevertheless, we did observe some intracellular differences: whereas 40% of wildtype dFB cells exhibited reliable secondary spikes 100–500 ms after a light-evoked action potential (Figure 3—figure supplement 2), none of the INX6 knocked-down cells displayed secondary spikes within that timeframe (Figure 6—figure supplement 1). Also, we noted considerably less variability in the membrane potential for dFB cells lacking INX6 (Figure 6—figure supplement 1 and Figure 6C,D right panels, compared with Figure 3D and Figure 3—figure supplement 1B).

Figure 6 with 1 supplement see all
Electrophysiological effects of INX6 knockdown in activated R23E10 neurons.

(A) In vivo adult Drosophila electrophysiology recording setup with a whole-cell patch clamp (orange) targeting R23E10 cell bodies and a local field potential (blue) electrode targeting the dFB. (B) Local field-potential (LFP) recordings were obtained within the presynaptic arborizations of the R23E10 neurons in the fan-shaped body, while whole-cell recordings were obtained from the cell bodies of R23E10 neurons. (C) Example trace (left) of a CsChrimson-expressing (UAS-CsChrimson/+; UAS-INX6 RNAi/+; R23E10-Gal4/+ with ATR) cell with INX6 knockdown when exposed to constant red light (red shading). Superimposed traces from multiple cell recordings (middle). The solid red line indicates mean values. Boxplots show median membrane potential for CsChrimson-expressing cells with INX6 knockdown before, during and after constant light stimulation (n = 6, **p<0.01, Friedman test with Dunn’s multiple comparisons to the pre-stimulus condition). (D) Example trace (left) of a CsChrimson-expressing cell with INX6 knockdown when exposed to 1 Hz light pulse with 5 ms exposure per pulse (orange shading). Superimposed traces from all 120 trials (middle) for the representative cell shown on the left. Solid yellow line indicates mean value. Boxplots show median membrane potential for CsChrimson-expressing cells with INX6 knockdown before, during, and after 1 Hz light stimulation (n = 6, **p<0.01, Friedman test with Dunn’s multiple comparisons to pre-stimulus condition). (E) Local field potential recordings from the dorsal fan-shaped body represented as power within the 1–15 Hz frequency range, in response to a constant light stimulus (pink bar) with one representative fly shown from each strain. (F) Boxplots show median 1–15 Hz local field potential power (±SEM) for the duration of the constant light stimulus relative to pre-stimulation power (normalized to zero) in UAS-CsChrimson/+;R23E10-Gal4/+ with ATR (blue, n = 7), UAS-CsChrimson/+; UAS-INX6 RNAi/+; R23E10-Gal4/+ with ATR (purple, n = 6), and R23E10-Gal4/+ with ATR (gray, n = 4). Only UAS-CsChrimson/+;R23E10-Gal4/+ showed a significant increase in 1–15 Hz activity when exposed to constant red light (*p<0.05, ns = not significant, by Wilcoxon signed rank test). (G) Peak amplitude of the average LFP response (±SEM) to 1 Hz light pulse stimulus is significantly reduced for UAS-CsChrimson/+; UAS-INX6 RNAi/+; R23E10-Gal4/+ (purple, n = 6), compared to UAS-CsChrimson/+;R23E10-Gal4/+ (blue, n = 7) (**p<0.01, by Mann-Whitney test). (H) Peak LFP amplitude (±SEM) in response to 1 Hz stimulus in the presence of bath-applied (n = 6) or locally injected (n = 4) carbenoxolone (CBX, orange) compared to baseline (gray) or vehicle (green, n = 5). ***p<0.001,**p<0.01, by paired t-test. See also Figure 6—figure supplement 1.

https://doi.org/10.7554/eLife.37105.018

We next asked whether extracellular LFP activity in the dFB might be affected by INX6 knockdown in the R23E10 neurons. We have shown previously that optogenetic dFB activation using the same R23E10-Gal4 driver produces increased LFP activity in the fly brain, especially in low (1–15 Hz) frequencies, and that oscillations within this range are also seen during transitions to spontaneous sleep in flies (Yap et al., 2017). We find this LFP signature of dFB activation again here (Figure 6E, top panel; Figure 6F, blue). In flies with INX6 knocked down in the dFB, however, there was no increased LFP activity (Figure 6E, bottom panel; Figure 6F, purple), as was also the case for control flies lacking the channelrhodopsin (Figure 6F, grey). As this constant light stimulus effect was quite subtle, we decided to drive the system with a pulsed activation regime to further test whether field potentials in the dFB were compromised following INX6 knockdown. 1 Hz stimulation of R23E10-Gal4 neurons also produced an LFP signature in the dFB, evident as an event-related potential (ERP) (Figure 6G, blue). Consistent with our results using a constant red-light activation regime, INX6 knockdown in R23E10 neurons significantly blunted the amplitude of the 1 Hz ERP (Figure 6G, purple). Together, these results suggest that one important consequence of INX6 knockdown in the sleep-promoting R23E10 cells is decreased synchronous activity in the dFB (detected as LFPs), whereas the activation effects on the membrane potential measured at the soma remain robust. To further test whether gap junctions are involved, we applied the gap-junction blocker carbenoxolone, or CBX (Cao and Nitabach, 2008). Perfusion of CBX onto the brain blunted the ERP recorded from the dFB, as did locally injected CBX (Figure 6H). This adds further evidence that gap junctions are required to produce synchronized activity in the dFB.

Gap junctions in the dFB are required for controlling behavioral responsiveness

Given that INX6 knockdown appears to disrupt dFB function at an electrophysiological level, we hypothesized that removing INX6 from these neurons would impair acute effects on behavior, such as those we obtained by optogenetic approaches earlier (Figure 2). To test this, we compared three groups of flies in DART (Figure 7A): (1) Positive control flies where dFB neurons could be activated and had wildtype INX6 expression (ATR-fed, blue), (2) negative control flies with INX6 knocked down but without dFB activation (non-ATR-fed, black), and (3) genetically identical flies where dFB neurons could be optogenetically activated but with INX6 knocked down (ATR-fed, purple).

Behavioral effects of INX6 knockdown in activated R23E10 neurons.

(A–D) Three different behavioral conditions comparing flies with R23E10 neurons that can be activated (UAS-CsChrimson/+; R23E10-Gal4/+with ATR, blue, n = 50) with flies where R23E10 neurons cannot be activated but have INX6 knocked down (UAS CsChrimson/+; UAS-INX6 RNAi/R23E10-Gal4 no ATR, black, n = 51) and flies with R23E10 neurons that can be activated and have INX6 knocked down (UAS-CsChrimson/+; UAS-INX6 RNAi/R23E10-Gal4 with ATR, purple, n = 51) for effects on sleep and responsiveness following red light activation. (A) Mean sleep duration (min/hr, shading indicates SEM) during the 24 hr before red light activation, then during the next 24 hr during which red light is delivered for 12 hr during the day (pink shading). (B) Comparison between the 12-hr day period without red light (baseline) and the period of red-light activation in terms of sleep duration and bout length. (C) Mean activity (mm/s) for UAS-CSChrimson/+; UAS-INX6 RNAi/R23E10-Gal4 with ATR for the same time periods as (A). UAS-CsChrimson/+; R23E10-Gal4/+ with ATR is overlaid (blue) during the 'Light On' period to show differences in responses to hourly stimuli. Other traces (black) not shown for clarity; see (D) for summarized control data. (D) Comparison between the 12-hr day period without red light (baseline) and the period of red-light activation in terms of peak responsiveness and sleep intensity. Error bars indicate SEM and asterisks indicate significance (***p<0.001, ****p<0.0001, t-tests).

https://doi.org/10.7554/eLife.37105.020

Optogenetically activating the dFB in cells where INX6 was knocked down still caused an increase in sleep duration and bout length compared to those in negative control flies (Figure 7A,B, purple compared to black). This effect was, however, not as strong as the activation of the dFB circuit with functional INX6 in these cells (Figure 7A,B, blue). This shows that INX6 knockdown blunts the sleep-promoting effect of dFB activation. Correspondingly, behavioral responsiveness was significantly decreased when the dFB was activated whereas INX6 expression was intact (Figure 7C,D blue). We did not see differences in peak responsiveness at baseline in the knockdown strains (Figure 7D, left), although the baseline sleep intensity data were consistent with our previous findings that inx6 knockdown in the dFB increases responsiveness (Figure 7D, right). It is possible that CsChrimson activity is leaky in ambient light conditions. However, clear differences were evident upon red-light activation: in contrast to the partial effects on sleep duration, INX6 knockdown completely blocked the effect of dFB activation on behavioral responsiveness (Figure 7C, purple) measured in terms of either mean peak responsiveness and sleep intensity (Figure 7D, purple), making the knockdown indistinguishable from negative controls (Figure 7D, black; not shown in Figure 7C for clarity). The knockdown was also indistinguishable from negative controls when we only examined data from awake flies, confirming that this effect is not restricted to sleep (p=0.94, t-test comparing mean responsiveness, n = 51 flies). These activation experiments therefore uncover a dissociation between behavioral responsiveness and sleep duration: INX6 expression is required for the sleep switch to downregulate behavioral responsiveness acutely, but INX6 is not required for these neurons to be sleep promoting. Presumably, dFB neurons are still able to affect sleep duration via other mechanisms (Donlea et al., 2018; Liu et al., 2016b).

Discussion

Before sleep can begin to achieve any of its multiple putative functions, it would seem that behavioral responsiveness to the outside world first needs to be reduced. This is a fundamental yet poorly understood aspect of sleep in all animals (Campbell and Tobler, 1984; Cirelli, 2009). Mechanisms that regulate behavioral responsiveness are therefore likely to be involved during transitions from wakefulness to sleep, so it seems parsimonious that overlapping neuronal systems might govern both behavioral states. In this way, sleep onset has some similarity with selective attention in its capacity to gate or suppress perception rapidly (Kirszenblat and van Swinderen, 2015). We show here that gap junctions that are expressed in the sleep-promoting neurons of the dFB in the central fly brain are probably mediators for gating behavioral responsiveness in both awake and sleeping flies, as a consequence of acutely increased membrane potential in these neurons. Recent work has shown that sleep pressure increases the input resistance and membrane time constants in these same neurons (Donlea et al., 2014; Pimentel et al., 2016), so this seems to be a probable mechanism whereby increased sleep pressure impairs behavioral responsiveness via electrical channels, perhaps even before promoting and maintaining sleep — which may then occur via chemical (e.g. peptidergic) signaling from these neurons (Donlea et al., 2018). It is therefore possible that the R23E10 neurons, which have been recently characterized as comprising part of a ‘sleep homeostat’ in the fly brain (Donlea et al., 2018; Pimentel et al., 2016), have two distinct arousal-related functions: to regulate behavioral responsiveness more generally — and thus to promote sleep acutely — and then to maintain sleep for the duration required by homeostatic demands. As increased arousal thresholds are a key criterion for sleep, it is of course difficult to disentangle these processes from each other. However, our data suggest that dFB-mediated suppression of behavioral responsiveness could be a prequel to sleep. First, on the basis of the current definitions of sleep (Campbell and Tobler, 1984; Cirelli, 2009), it does not make sense for sleep to precede changes in responsiveness, so either these processes are simultaneous or sleep functions succeed an initial increase in arousal thresholds that may already be evident during wakefulness.

Our experiments show that behavioral responsiveness can be suppressed by acutely activating the dFB even in awake flies, and that this effect has little inertia, at the levels of both behavior and electrophysiology. Sleep intensity was increased however upon prolonged dFB activation, suggesting a distinct cumulative effect. Indeed, at the deepest stage of sleep (when flies are least responsive to mechanical stimuli), removal of INX6 from the dFB had no effect on behavioral responsiveness (Figure 5C), suggesting that other factors are involved at that stage for maintaining high arousal thresholds. When flies are closer to awakening again, or when they are engaged in lighter sleep stages, it is possible that INX6-mediated mechanisms come back into play to maintain sleep. In recent work, we have shown that transitions in and out of sleep are associated with dFB activity (Yap et al., 2017), which is consistent with the view that a common system might be governing behavioral responsiveness during wakefulness as well as sleep.

It is likely that other dFB neurons outside of the R23E10 circuit are also involved, as suggested by only partial overlap with INX6 labeling (Figure 4B). However, without INX6, activated R23E10 neurons were unable to reduce behavioral responsiveness, suggesting a cell-autonomous effect. We have shown here and elsewhere (Yap et al., 2017) that dFB activation is associated with increased LFP activity in the central brain, which could reflect increased synchronous neural firing. Without expression of INX6 in the R23E10 neurons, these induced LFP effects also disappear, suggesting that gap junctions are important for promoting synchronous activity in this circuit. This suggests a rapid mechanism through which the dFB might suppress behavioral responsiveness upon sleep onset, by for example producing synchronous low frequency activity that could interfere with ongoing sensory processing and integration in the central brain. Although acute activation of the R23E10 neurons alone is unlikely to occur in a natural context, electrical communication to other cells via gap junctions could support subtler levels of behavioral control while the flies are awake, which might consolidate into overall increased arousal thresholds as membrane potentials across the dFB increase with increasing sleep pressure. We have not completely excluded the possibility that a fast-acting neuropeptide or neurotransmitter secreted from R23E10 neurons regulates behavioral responsiveness, but our data favor the view that a parallel channel involving electrical synapses exists. Future studies should uncover the extent of this gap-junction-coupled network, although our antibody-labeling (Figure 4) and dye-coupling experiments (Figure 4—figure supplement 1) suggest that cells in the pars intercerebralis (PI) are probably involved. It also remains possible that dFB neurons are electrically coupled to each other, thereby generating LFP oscillations that are employed in the regulation of behavior. However, in our dye-coupling experiments (see 'Materials and methods'), we did not see any evidence of labeling of another dFB cell, whereas multiple PI cells were labeled.

It is not known whether electrical communication might be employed to promote sleep in other animal brains, although there is increasing evidence that gap junctions play an important role in the regulation of behavioral state and arousal in the mammalian brain (Coulon and Landisman, 2017). More generally, a role for gap junctions in sleep-promoting neurons also suggests a novel plasticity mechanism for regulating behavioral states. Whereas plasticity is typically viewed as a property of chemical synapses, the parallel electrical communication channel afforded by gap junctions could provide an alternate way of regulating sleep pressure, and thereby promoting sleep functions while ensuring that arousal thresholds are tightly linked to sleep need. In recent work, we have shown that transient electrical activation of the dFB neurons during fly brain development permanently impairs behavioral responsiveness in adult animals (Ferguson et al., 2017), further confirming the strong link between these neurons and the regulation of arousal in Drosophila. Different INX6 expression levels in the dFB may provide a mechanism for optimally linking behavioral responsiveness with sleep need. It will be interesting to see whether INX6 expression in the dFB correlates with the striking range of individual differences that we observed for sleep duration and behavioral responsiveness in wildtype flies (Figure 1F), and whether INX6 expression in the dFB might be co-regulated alongside other proteins that have been shown to control neuronal excitability in this arousal circuit (Donlea et al., 2018; Pimentel et al., 2016).

Materials and methods

Key resources table
Reagent type (species)
or resource
DesignationSource or referenceIdentifiersAdditional
information
Genetic reagent
(D. melanogaster)
R23E10-Gal4BloomingtonRRID:BDSC_49032
Genetic reagent
(D. melanogaster)
C5-Gal4doi: 10.1002/
ssscne.22284
Paul Shaw Lab
Genetic reagent
(D. melanogaster)
104y-Gal4Paul Shaw Lab
Genetic reagent
(D. melanogaster)
UAS-CsChrimsondoi: 10.1038/
nmeth.2836
provided by Vivek
Jarayaman Lab
Genetic reagent
(D. melanogaster)
UAS-2eGFPBloomingtonRRID:BDSC_32186
Genetic reagent
(D. melanogaster)
UAS-NaChBacBloomingtonRRID:BDSC_9469
Genetic reagent
(D. melanogaster)
UAS-syntaxin3-69Fly BaseFBal0092503
Genetic reagent
(D. melanogaster)
UAS-shibireTSPaul Shaw LabGene ID: 45928
Genetic reagent
(D. melanogaster)
tubpGAL80tsBloomingtonRRID:BDSC_7108
Genetic reagent
(D. melanogaster)
UAS-INX6 RNAiVDRCv8638Provided by the
Chia-Lin Wu Lab
AntibodyRabbit anti
-INX6
Provided by
the Chia-Lin Wu Lab
1:1,000
AntibodyGoat anti-mouse
AlexaFluor488
InvitrogenCatalog # A-106801:200
AntibodyGoat anti-rabbit
AlexaFluor568
InvitrogenCatalog # A-110111:200
AntibodyGoat anti-rabbit
AlexaFluor647
InvitrogenCatalog # A-212441:200
AntibodyMouse anti-NC82DSHBAB_23148661:10
AntibodyGoat anti-rabbit
AlexaFluor488
InvitrogenCatalog # A-110081:200
Chemical compound, drugNeurobiotinVector LabsCat. No: SP-1120
Software, algorithmDARTbfklabhttp://www.bfklab.com/
Software, algorithmMATLAB codeThis paper142facahttps://github.com/melvynyap/gap-junction-sleep-control
(copy archived at
https://github.com/elifesciences-publications/gap-junction-sleep-control)
Chemical compound, drugAll-trans retinalSIGMA-AldrichSID 24899355
Chemical compound, drugVectashieldVector LabsCat. No: H-1000
Chemical compound, drugStreptavidinInvitrogenCatalog number: S323571:200

Fly stocks and media

Request a detailed protocol

Flies were cultured on standard agar medium under a 12-hr day/night cycle. Flies used for optogenetics were placed on food containing 1 mM all-trans retinal (Sigma) for two days before experiments. All flies were outcrossed six generations to a w2202 genetic background (isoCJ1; (Yin et al., 1994). The INX6-RNAi lines were a gift from Chia-Lin Wu. UAS-CsChrimson was a gift from Vivek Jayaraman. R23E10-Gal4, C5-Gal4, 104y-Gal4, and UAS-shibire were obtained from Bloomington. The INX6 RNAi is v8638, originally obtained from the VDRC (Dietzl et al., 2007). UAS-syx3-69 was obtained from Bing Zhang.

Behavioral experiments

Request a detailed protocol

One-day-old female virgin flies were briefly anesthetized on CO2 for collection. At two days old, flies were transferred to 65 mm glass tubes (Trikinetics, Waltham, MA) sealed with food (containing retinal if necessary) at one end and cotton at the other and placed inside a 25°C incubator. A camera was positioned above for recording activity, and shaftless vibrating motors (Precision Microdrives, 312–101) were positioned underneath for stimulus delivery using DART (Faville et al., 2015). Four 700 mA 617 nm LED lights (Luxeon Star, Canada, SP-01-E4) were placed above the flies and delivered irradiance of 0.03–0.1 mW/mm2. These LEDs were controlled using a custom circuit board and Arduino with custom code.

After a day of acclimatization, fly activity was recorded over three days. Every hour, a stimulus was delivered consisting of five 0.2 s 2.4 g vibrations separated by 0.8 s. For optogenetic experiments, the red LEDs were switched on for the second and third day during ZT 0–12 periods to deliver either constant or 1 Hz (5 ms pulse width) light. For INX6 RNAi, sleep experiments were run for 3–4 days. All experiments were replicated at least three times. For UAS-ShibireTS and UAS-Syntaxin3–69, experiments were run for 3 days and the temperature of the incubator was raised from 23°C to 31°C on days 2 or 3.

For acute R23E10-Gal4 activation experiments, the mechanical vibration was delivered to flies once every 15 min, alternating between periods with and without light exposure for six trials (three light on, three light off). The light was delivered for1 min prior to and 3 min after the stimulus.

Sleep intensity was measured as the proportion of immobile (sleeping) flies that responded (at any level) to these stimuli. Flies were determined to have responded if they moved by a threshold of at least 3 mm (~3 body lengths) within the minute following the stimulus (Faville et al., 2015). To measure mean peak responsiveness, fly activity was averaged for two minutes prior to and the 15 min after each stimulus. This average activity was fitted with a single-inactivation exponential equation and the peak amplitude of activity following the stimulus was measured (Faville et al., 2015). To determine awake responsiveness, we included only flies that had moved within the 1 min prior to the stimulus (i.e. awake flies) in the analysis. Statistical analyses were performed in Prism 7 for Mac (GraphPad) and all tests were corrected for multiple comparisons.

For experiments in which we downregulated INX6 expression in adult flies (using R23E10-Gal4 > UAS-INX6 RNAi GAL80TS), flies were raised at 19°C. A four-day experiment used the same hourly stimulus protocol as above and consisted of 1 d at 19°C, 1 d at 31°C for GAL4 induction, followed by 2 d at 25°C. Data from both days following 31°C heating were combined and averaged.

Immunochemistry

Request a detailed protocol

Fly heads were removed and brains dissected in cold PBS, before being fixed in 4% paraformaldehyde for 30 min. After washing with PBST (0.2% Triton-X and 0.01% sodium azide) and blocking with bovine serum albumin, brains were incubated with primary antibodies (R23E10-Gal4 > UAS-2eGFP: mouse anti-GFP 1:1,000, rabbit anti-INX6 1:1,000; CS: rabbit anti-INX6 1:1,000) overnight, and with secondary antibodies (R23E10-GAal4 > UAS-2eGFP: goat anti-mouse AlexaFluor488 1:200, goat anti-rabbit AlexFluor568 1:200; CS: AlexFluor687 1:200) for 1 d, then mounted using Vectashield. The INX6 primary antibody was a generous gift from Chia-Lin Wu. Samples were imaged on a spinning-disk confocal system (Marianas; 3I, Inc.) consisting of a Axio Observer Z1 (Carl Zeiss) equipped with a CSU-W1 spinning-disk head (Yokogawa Corporation of America), an ORCA-Flash4.0 v2 sCMOS camera (Hamamatsu Photonics), and 20 × 0.8 NA PlanApo and 100 × 1.4 NA PlanApo objectives. Image acquisition was performed using SlideBook 6.0 (3I, Inc). To quantify the degree of colocalization between INX6 and R23E10-Gal4 dFB neurons, we used ImarisColoc (Imaris, Bitplane Inc). INX6 and GFP channels were selected using automatic threshold detection from pixel intensity histograms. The number of voxels in these channels that colocalized across Z-planes inside of the dFB region of interest (ROI) was then determined. To determine INX6 RNAi efficacy, we compared INX6 expression in the dFB in control brains (Elav-Gal4/+) to brains expressing the RNAi (Elav-Gal4 >UAS-INX6 RNAi) using an INX6 antibody. Brains were incubated with primary antibody (rabbit anti-INX6 1:1,000, mouse anti-NC82 1:10) overnight, and with secondary antibody (goat anti-rabbit AlexaFluor488 1:200, goat anti-rabbit AlexFluor647 1:200) for a day. A ROI was drawn around the dFB using the NC82 staining. Mean INX6 intensity within this ROI was normalized to mean NC82 intensity. Normalized dFB immunoreactivity in control brains was compared to Elav-Gal4 >UAS-INX6 RNAi brains using a t-test.

Quantitative PCR

Request a detailed protocol

A quantitative reverse transcriptase PCR (qRT-PCR) assay was used to conclude whether INX6 knockdown was achieved relative to that of the housekeeping gene Act88F, which was determined to be stably expressed across all experimental conditions. Females of 3–5 d old were collected by CO2 anesthesia, snap frozen, and stored at −80°C. Six pools of five fly heads (30 heads total) were placed into a 1.5 ml Eppendorf tube. Total RNA was purified using TRIzol according to the manufacturer’s protocols (Invitrogen, Carlsbad, California), immediately after dissection. Total RNA was treated with DNase (Sigma-Aldrich, St. Louis, MO) to eliminate contaminant genomic DNA. Approximately 0.5 µg of total RNA was reverse transcribed using random primers (Invitrogen) and reverse transcriptase (Invitrogen) according to the manufacturer’s protocols. Gene expression was estimated with two technical replicates using a standard quantitative PCR (qPCR) assay (McMeniman et al., 2009). Each qPCR mixture contained 12.5 µL of 2X SYBR premix (Invitrogen), 1 µL of forward primer, 1 µL of reverse primer, 100 ng of DNA, and H2O to a final volume of 25 µL. The expression of two genes was estimated relative to Act88F using the CT (where CT is the threshold cycle) method (Pfaffl, 2001). Averages of expression were compared using Student’s t test (in SPSS software). The primers that were used were actinF:ATCGAGCACGGCATCATCAC, actinR:CACGCGCAGCTCGTTGTA, inx6(v8638)F:GAACGGCATGCCCAAGTC, and inx6(v8638)R:ACCAGTCGCCGTATCCAG.

Electrophysiological recordings

Request a detailed protocol

Methods for performing in vivo open brain recordings (Figure 3A) were described elsewhere (Yap et al., 2017). Briefly, to prepare for brain electrical recordings, we secured a 3–7-d-old fly on a custom fly plate (Maimon et al., 2010) (Figure 3A). The bath chamber of the fly plate was filled with oxygenated (bubbled in 95% O2, 5% CO2) extracellular fluid (ECF) containing (in mM): 103 NaCl, 10.5 trehalose, 10 glucose, 26 NaHCO3, 5 C6H15NO6S, 5 MgCl2 (hexa-hydrate), 2 sucrose, 3 KCl, 1.5 CaCl (dihydrate), and 1 NaH2PO4. To enable access to the brain, the cuticle was partially removed using forceps, with the perineural sheath removed either mechanically with forceps or chemically using protease (0.5% collagenase type IV). The fly in this preparation was positioned on an air-supported ball and the brain was continuously superfused with oxygenated ECF.

Whole-cell patch clamp recordings were obtained from the cell bodies of the R23E10 neurons using a fixed-stage upright fluorescence microscope (Olympus BX51WI, U-RFL-T, Olympus, Berlin, Germany) with a 40x water-immersion objective. To achieve high-contrast visuals, an infrared LED (Osram SFH 4232) coupled with an infrared camera (DAGE-MTI IR-1000) was used, whereas visualization of the GFP-labeled neurons was achieved using a mercury short-arc lamp (HBO 103 W/2). Electrode positioning was controlled by a motorized micromanipulator system (Sutter MP-285).

Borosilicate glass capillaries (Harvard Instruments GC150F-25) were used, pulled (Sutter P-97 micropipette puller) to a tip size of 8–12 MΩ, and subsequently filled with an internal solution containing (in mM): 140 potassium aspartate, 10 HEPES, 1 KCl, 4 MgATP, 0.5 Na3GTP, 1 EGTA, and 0.05 Alexa fluor 568, pH 7.3, adjusted to 265 mOsm. Voltage signals were acquired in current-clamp mode, with a CV-7B headstage (Molecular Devices) and a Multiclamp 700B amplifier (Molecular Devices), low-pass filtered at 10 kHz (Bessel), and digitized at 10 kHz using an Axon Digidata 1440A Digitizer controlled by the MultiClamp 700B Commander Software and AxoGraph X 1.4.4 (Axon Instrument) on a computer running Windows XP. In some cells, it was necessary to inject a small constant hyperpolarizing current (15–40 pA) in order to stabilize the resting membrane potential to between −30 and −50 mV.

LFP recordings were performed simultaneously with whole-cell recordings using same-sized micropipettes filled with ECF instead of internal solution. LFP signals were acquired with a FET electrode, amplified and filtered (low: 0.1 Hz, high: 1 kHz) (A-M Systems Model 1700), digitized (Axon Digidata 1440A Digitizer) and sampled at 10 kHz using the data acquisition software AxoGraph X 1.4.4.

Electrophysiology and optogenetic stimulation

Request a detailed protocol

Photostimulation of CsChrimson-expressing neurons was achieved by using an ultra-bright red LED (617 nm Luxeon Rebel LED, Luxeon Star LEDs, Ontario, Canada) directed to the opened section of the fly head, producing 0.1–0.2 mW/mm2 at a distance of 4–5 cm with the aid of concentrator optics (Polymer Optics 6° 15 mm Circular Beam Optic, Luxeon Star LEDs). To prevent overheating of the fly and the immediate environment, the LED was mounted onto a sink pad (SinkPAD-II 20 mm Star Base) that was attached to a small heat sink. The temperature of the solution bath was also kept constant within the range of 22°C to 23°C by using a thermistor and an in-line heater/cooler (Warner Instruments Model SC-20), both driven by a temperature controller (Warner Instruments Model CL-100). Light exposure was triggered after 1 min of baseline recording and lasted for 2 min. Light was delivered either in a continuous or in a pulsatile fashion, the latter of which consists of a 1 Hz train of 5 ms optical pulses. Timing of the light switch was controlled by AxoGraph, which also measured the timing of light exposure from a basic photodiode.

Dye labeling following electrophysiology recordings

Request a detailed protocol

Neurons in the dFB were injected with internal solution supplemented with 0.5% neurobiotin (Vector Labs TM), which has been shown to pass through gap junctions in invertebrates effectively (Fan et al., 2005). Whole-cell configuration was achieved for the purpose of passive flow of dye into the cytoplasm. Dye injection was aided by iontophoresis, with the delivery of depolarizing current pulses (200 pA, 500 ms, 1 Hz, 50% duty cycle) delivered over 5 min, followed by hyperpolarizing current pulses (−200 pA). The preparation was left for a further 40 min to allow the dye to diffuse into the cell passively. After withdrawing the micropipette, the preparation was left untouched for about 1 h to allow sufficient time for dye to diffuse within and across cells.

The fly brain was dissected and fixed at room temperature for 20 min in 4% paraformaldehyde in 0.1M phosphate buffer (pH 7.0) then washed 3 × 10 min in PBST. The brain was incubated for 24–48 hr at 4°C with streptavidin conjugated with Alexa Fluor 647, in 5% bovine serum albumin (BSA) (1:200). Brains were rinsed in PBST (3 × 10 mins) and were mounted on glass slides in Vecta Shield mounting medium. All confocal images were obtained using an inverted spinning disk microscope (Yokogawa W1). All confocal images obtained were analyzed and processed in Fiji/Imaris. Deconvolution of images acquired using the Yokogawa W1 spinning disc confocal microscope was performed with Huygens Professional Plus Deconvolution software (Scientific Volume Imaging, Hilversum, The Netherlands). We used a theoretical point spread function (PSF) that was obtained by the parameters of image acquisition. For the deconvolution, we used a total image change threshold of 0.01, with single block processing on, a maximum iteration value of 40 and a signal to noise ratio (SNR) of 20.

Gap-junction blocker delivery

Request a detailed protocol

For the gap junction blocker experiments, carbenoxolone (CBX, 1 mM) (Cao and Nitabach, 2008) was dissolved into ECF, which was delivered to the fly brain by either perfusing the bath chamber with it or locally injecting it into the dFB region via a micropipette attached to a micromanipulator. For local delivery, constant air pressure was applied to push the CBX solution out of the micropipette, with the flow of solution into the dFB confirmed using a fluorescent dye (Alexa Fluor 568) that was added to the CBX solution. LFP recordings were obtained prior to bath perfusion of CBX, which served as baseline for comparison with the CBX effect. As an additional control for the locally delivered CBX experiments, ECF with no added CBX (vehicle) was injected to observe the effect of applying physical pressure into the central brain.

Electrophysiological analyses

Request a detailed protocol

All analyses were performed offline using custom scripts in MATLAB (2014a) (Yap, 2018) (https://github.com/melvynyap/gap-junction-sleep-control, 142faca; copy archived at https://github.com/elifesciences-publications/gap-junction-sleep-control). Cells were identified as burst-spiking cells when an action potential failed to return to resting membrane potential prior to reaching the next peak. These cells, when hyperpolarized with a constant current, also failed to produce single spiking events. Cells observed to have at least one single spiking event in addition to the burst spiking were identified as having a pattern involving both single and burst-spiking events. Non-spiking cells were identified as such when the cell failed to produce any action potentials despite the injection of depolarizing current in steps of +10 pA up to +150 pA.

We use the change in membrane potential (ΔVm) as the measure of cell response to photostimulation. For the constant light stimulus condition, the mean membrane potential for the pre-stimulation period in each cell was obtained over the length of baseline, which was 1 min. Mean membrane potential during stimulation was calculated over the first 1 min of photostimulation, whereas for post-stimulation membrane potential, a mean was calculated over the last 1 min of recording (1 min after the stimulation ended). The mean membrane potential for each cell for the 1 Hz pulse photostimulation was calculated by first producing a composite membrane potential that represents the average of all 120 pulses, followed by obtaining the average membrane potential for the first 100 ms of this composite data. From these composite membrane potentials, it was visually evident that the 1 Hz light pulse appeared to be having an effect on the spike timing in some cells.

Owing to the unipolar morphology of the recorded cells, the true value of the cells’ resting membrane potentials were probably lower and less variable than those recorded from the cell bodies (Gouwens and Wilson, 2009). Therefore, to better represent the effect of photostimulation of the CsChrimson-expressing cells, the data were displayed as ΔVm instead of absolute values. Cells with various single-spike and burst-spiking characteristics were found to exhibit the same membrane potential response, and were therefore combined for the purpose of statistical analyses. In a subset of cells, we observed a volley of evoked spikes in response to the constant light stimulus that lasted very briefly. Interspike intervals were obtained from these spikes by measuring the time between the first and second peak.

Raw LFP signals were transformed into wavelets using the Morlet wavelet transformation function ft_specest_wavelet in the Fieldtrip MATLAB toolbox, with wavelet width value set at 30 and 3 standard deviations. To generate the averaged time-frequency spectrograms, wavelets were normalized to the mean wavelet values of the pre-stimulus segment (baseline) for each fly prior to averaging across all flies. Wavelets differ in magnitudes across fly recordings, and were therefore normalized for each fly prior to averaging. For statistical analyses of 1–15 Hz LFP power, wavelets were also normalized to the mean wavelet values of baseline but were done within the respective frequency bins of 0.1 Hz, producing wavelet ratios (to baseline). The mean wavelet ratios during the stimulation period were calculated for each fly and then averaged across flies. Baseline was subsequently zeroed so that any positive values obtained during the stimulation period denotes an increase in LFP power. Peak amplitude from raw LFP signal represents the LFP response to the pulsative light stimulation and as such was defined as the maximum field potential deviation from zero in the 50 ms after the onset of the 1 Hz 5 ms light pulse. Peak amplitude was averaged across all 120 trials per fly prior to averaging across flies.

Statistical analyses

Request a detailed protocol

All statistical analyses were performed using Prism 7 for Windows (GraphPad). Where dataset failed the Shapiro-Wilk normality test (p<0.05), non-parametric tests were used. Friedman test with Dunn’s post hoc multiple comparisons test were used to compare the membrane potential averages during the stimulation and post-stimulation period to pre-stimulation baseline period. The Wilcoxon signed rank test was used to test for significant difference between the LFP power ratio and a baseline of zero during the stimulation period. The Mann-Whitney test was used to compare the peak LFP amplitudes between the two fly strains. Paired t-tests were used to compare the peak LFP amplitudes between the CBX and baseline conditions. All membrane potential data in the figures are presented as median and range, whereas the LFP data presented in figures are shown as means ± SEM. All tests for significance were two-tailed and confidence levels were set at α = 0.05.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
    Transient dysregulation of dopamine signaling in a developing Drosophila arousal circuit permanently impairs behavioral responsiveness in adults
    1. L Ferguson
    2. A Petty
    3. C Rohrscheib
    4. M Troup
    5. L Kirszenblat
    6. DW Eyles
    7. B van Swinderen
    (2017)
    Frontiers in Psychiatry, 8, 10.3389/fpsyt.2017.00022, 28243212.
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52

Decision letter

  1. Mani Ramaswami
    Reviewing Editor; Trinity College Dublin, Ireland
  2. K VijayRaghavan
    Senior Editor; National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: this article was originally rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]

Thank you for submitting your work entitled "Acute control of the sleep switch in Drosophila reveals a role for gap junctions in regulating behavioral responsiveness" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Mani Ramaswami as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Senior Editor.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

All the reviewers observed the intrinsic interest of the questions being addressed: the control of arousal threshold, potentially independently of sleep; and underlying cellular mechanisms. They also were universally appreciative of the impressive range of techniques employed in this study. However, an overriding concern was the need for a substantial number additional experiments to establish the main premise of the paper that a mechanism for reducing arousal threshold is deployed prior to one involved in sleep induction. Moreover, there also remained some questions on the role of innexin6 function, in particular on its role in baseline sleep and in neurons other that dFB neurons.

Reviewer #1:

Troup and coauthors analyse mechanisms by which R23E10 marked fan-shaped body neuron control sensory responsiveness and sleep states. Previous work has shown that these neurons mediate and drive entry into the sleep state in response to sleep need, but has not described whether they influence sensory responsiveness that must decrease prior to entry into sleep. The authors show that activation of dFB neurons also reduces sensory responsiveness and,by studying this in actively moving animals, present evidence to show that this effect is prior to and independent of entry into the sleep state. Building on this through a variety of technically impressive experiments, the main conclusion from this work is that these dFB neurons act to reduce sensory responsiveness prior to and independent of their role in promoting the sleep state. This independent function is proposed to be mediated by innexin6 gap junctions expressed in these cells that allow coherent activity among connected dFB neurons that is reflected in stimulation-evoked LFP potentials that oscillate in the 1-15 Hz power range.

1) The role for innexin6 in coupling dFB neurons is interesting and solid and the experiments seem well done. Without experiments that hint at differences in downstream mechanisms, one is left with the impression that differences in the innexin6 requirement for sensory responsiveness vs sleep may in some way reflect different thresholds of dFB activity, rather than two distinctive downstream pathways as proposed by the authors. Perhaps the paper could be summarised and considered as a Short Report?

2) Figure 1 sensory responsiveness and sleep are only correlated at night in wild-type flies, but this correlation is seen in the day as well following dFB stimulation. Could this be an artefact of the level of sleep? Another way of interpreting Figure 1F and H is to state that correlations emerge when sleep duration is more than 30 mins/hour. If this is a major point, it needs to be better justified. If minor (as it appears), then it either should not be made or should be duly qualified.

3) While Figure 2 shows that dFB activation results in reduced responsiveness in active flies, it does not clarify whether this is a threshold effect – ie. FB activation and resulting GABA release results in reduced responsiveness at low levels of inhibition, but requires higher levels of inhibition of the same downstream substrates for sleep. Perhaps the levels of GABA release required are different in day and night?

4) The experiments themselves seem nicely done, but some of the figures make points, which though important to establish, are not highly instructive in themselves. E.g. Figure 3 shows that R23E10 neurons are different in their intrinsic properties (though the significance of this is unclear) and that light driven ChR does indeed result in neuronal firing. Figure 4 shows that innexin6 is expressed in at dFB neurons. The authors may consider integrating these points in summary form with one of the other figures and show more comprehensive images in the figure supplements.

Reviewer #2:

This is an interesting paper on fly sleep from the van Swinderen and Shaw labs. The paper follows up recent results from van Swinderen's lab, which reported local field potential oscillations in the fly ellipsoid body/central complex, and these oscillations vaguely resemble the famous thalamo-cortical oscillations characteristic of mammalian sleep. These previously observed oscillations were linked to arousal threshold and fly sleep. Arousal threshold is a fundamental property of sleep in all organisms and is therefore important. The relationship or arousal threshold to other aspects of sleep is also of interest. These authors focus on the dorsal fan shaped body (dFB) an important sleep-promoting sub-structure of the central complex.

There are 3 central results: (1) Optogenetic activation of the dFB increases arousal threshold as well as other sleep parameters. (2) The effect on arousal threshold is faster or more complete than the effect on other sleep parameters, suggesting that the dFB has a more primary or more prominent role in affecting arousal threshold than on other aspects of sleep like sleep duration. (3) Gap Junctions and in particular innexin6 is relevant to this arousal threshold function of the dFB.

1) The R23E10-Gal4 driver is not so clean and has some VNC expression. Even if the authors dispute this statement, it is prudent to use at least two drivers with dFB expression and show that both give comparable results. For example: R84C10-GAL4 is a strong and clean dFB-GAL4 driver.

2) It is important to use inhibition as well as activation to back up the claim that the dFB plays an important role in arousal threshold. (Sufficiency does not ensure necessity, and we should now learn which it is.) The Discussion section is written as if this had been done. This request would have been unfair a couple of years ago, but new optogenetic tools have come on-line with widespread use and endorsement. If inhibition fails to have an effect, I would not recommend rejection of the paper, but I think the authors should find out.

3) In Figure 7, the authors showed the behavioral responsiveness to stimulus when INX6 knockdown in non-activated R23E10 neurons, and it is comparable to control flies. Does this mean that gap junctions in dFB neurons play no role in arousal regulation under basal conditions? If so, what do the authors think about this?

4) The knockdown experiments should be done with tubpGAL80ts to eliminate developmental effects. And why were the MB lobes were not identified with INX6 staining? In Wu et al. (2011), they reported that INX6 was expressed in the MB and EB as well as the FB.

5) Why is daytime baselines sleep so different in Figure 7A than in Figure 5A? This makes me doubt whether the important result in Figure 5A is correct, but perhaps I am missing something. This difference is also true for nighttime sleep, which showed an effect in Figure 5A but no apparent effect in Figure 7A. I understand that in Figure 7A they co-expressed UAS-CsChrimson with INX6 RNAi but CsChrimson should not affect INX6 RNAi-induced less sleep phenotype on the baseline day unless there is a subtle RNAi baseline effect. This should then be tested with GFP RNAi or something like this.

6) I am not convinced the order of events (first arousal threshold, then sleep) is as the authors suggest. The former can be assayed instantly whereas the latter requires some time to qualify as sleep.

7) Lots of control data is missing for the behavior experiments. For example, GAL4 and UAS control are needed in Figure 2 and Figure 7.

8) Error bars are needed in Figure 2—figure supplement 2A and Figure 2—figure supplement 2E. And why is the behavioral response in Figure 2—figure supplement 2G on the baseline day is so different compared to Figure 2—figure supplement 2C on the baseline day? They should be the same genotype and condition. And what is the explanation why for the 10Hz light pulse activation of R23E10 has no effect on sleep duration?

Reviewer #3:

In this manuscript, the authors attempt to define the cellular and molecular control of behavioral responsiveness in flies. The work is well-presented and technically very impressive. However, I have major experimental and conceptual concerns. Most importantly, the novelty of the manuscript is in trying to explain how behavioral responsiveness changes to permit sleep, with the key point being that sleep pressure must alter responsiveness prior to sleep onset itself. This is an intriguing idea but the evidence is lacking. The major concerns are as follows:

1) Figure 2 is essential in making the case that dFB activation reduces behavioral responsiveness separately from (and perhaps prior to) sleep. However, it is perplexing that responses are "abolished" (subsection section “Optogenetic activation of the sleep switch decreases behavioral responsiveness”) in wake flies with dFB activation, while in sleeping flies, the response is only diminished. Is this a function of needing to normalize to wake activity (Figure 2E)? If so, is wake activity (mean speed) affected by dFB activation? How about movement in response to a strong stimulus? This raises the more general possibility that locomotor function is simply different with dFB activation, as opposed to an actual change in responsiveness to the stimulus at a sensory level. Can the authors address this?

2) Also in Figure 2, I am confused that the earlier work (Figure 2C and D, excluding sleep intensity) do not distinguish between wake and sleeping flies. Why not only analyze sleeping flies until shifting focus to wake flies in Figure 2E? If analysis of wake flies requires normalizing the baseline activity (Figure 2E right panel), then shouldn't the authors need to account for that in analyzing mean peak responsiveness (Figure 2D, for example)?

3) Figure 2F is particularly critical, and the authors set it up as if it will answer something about whether the change in responsiveness or sleep comes first (subsection section “Optogenetic activation of the sleep switch decreases behavioral responsiveness” "even in awake flies"). However, it appears to again mix/combine sleeping and awake flies for analysis, which is a major confound. In other words, what if most of these flies are rapidly put to sleep with the 1 minute dFB activation prior to stimulus? Then the figure simply shows that inducing sleep with dFB changes responsiveness, which we already know (from this work, but also from Donlea et al. (2011) and Donlea, Pimentel and Miesenbock (2014) – it is important to note that these published works have already demonstrated a change in arousability with dFB activation, even if not with such detail as shown here).

4) The authors make a point of 1Hz and 10Hz optogenetic activation of dFB both changing responsiveness, but only 1Hz inducing sleep, therefore dissociating the two. However, these data are not convincing. In Figure 2—figure supplement 2G, why is the response so poor at baseline compared to C in this same figure? And why suddenly so much noise? Moreover, the change in responsiveness (H) appears similar +/- ATR: the baselines appear different and the 10Hz light causes a similar decrease in both (same issue in D of this figure). In other words, I am not convinced that there is a change in responsiveness, and therefore the ability to dissociate sleep time from arousal is hindered. The sleep intensity data is more convincing, but this is a comparatively less precise measure.

5) In Figure 7, there are concerns. First, in contrast to Figure 5A, there is no longer an effect of INX6 RNAi expressed using R23E10-GAL4 (Figure 7A, left). Is there dilution of UAS-driven expression by co-expressing two UAS? Is INX6 knockdown even effective here? If we assume so, and if INX6 is mostly important in the context of dFB activation, then the authors should test these same measures following sleep deprivation. That would be most convincing with regard to a role for INX6 in modulating dFB driven sleep intensity. Overall, it would add a lot to this work for the authors to perform many of the experiments in Figure 2 after sleep deprivation, which is more physiologically relevant and should induce changes in behavioral responsiveness.

https://doi.org/10.7554/eLife.37105.024

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

[…]

1) The role for innexin6 in coupling dFB neurons is interesting and solid and the experiments seem well done. Without experiments that hint at differences in downstream mechanisms, one is left with the impression that differences in the innexin6 requirement for sensory responsiveness vs sleep may in some way reflect different thresholds of dFB activity, rather than two distinctive downstream pathways as proposed by the authors. Perhaps the paper could be summarised and considered as a Short Report?

It is possible that different thresholds of dFB activity are involved for sleep and responsiveness. However, we get similar behavioral and electrophysiological results when we stimulate for 5ms at 1Hz as for when stimulate continuously with red light. As the 1Hz regime involves 99.5% less red light, it seems unlikely that this is a threshold effect. Rather, thresholds could be determined in a different way. Indeed, our 10Hz stimulation data suggests a likely frequency effect: 10Hz suppressed responsiveness but did not induce sleep. This already suggests a dissociation between these processes, which is a key theme throughout our paper. Dissecting this in the frequency domain will be an interesting future research direction, however this is not what this paper is about. To prevent any confusion, we are inclined to remove the 10Hz data from our paper (see replies to other reviewers, below). We believe the 1Hz regime is sufficient to further convince that our activation effects are real.

2) Figure 1 sensory responsiveness and sleep are only correlated at night in wild-type flies, but this correlation is seen in the day as well following dFB stimulation. Could this be an artefact of the level of sleep? Another way of interpreting Figure 1F and H is to state that correlations emerge when sleep duration is more than 30 mins/hour. If this is a major point, it needs to be better justified. If minor (as it appears), then it either should not be made or should be duly qualified.

It is minor, and we have altered the text to qualify this (e.g., Results subsection “Correlating sleep duration and behavioral responsiveness”). The point is simply sleep and responsiveness can be separated (day and night) or made more tightly linked (by dFB activation). However, it is difficult to disentangle responsiveness and sleep duration with chronic manipulations, and we now make this point clear (in the aforementioned subsection). This first result is important to confirm the tight correlation between these processes.

3) While Figure 2 shows that dFB activation results in reduced responsiveness in active flies, it does not clarify whether this is a threshold effect – ie. FB activation and resulting GABA release results in reduced responsiveness at low levels of inhibition, but requires higher levels of inhibition of the same downstream substrates for sleep. Perhaps the levels of GABA release required are different in day and night?

It is unlikely to be a threshold effect, because flies stop responding within a minute of dFB activation (Figure 2F), but then the same stimulus regime if left on will lead to increased sleep duration and intensity (Figure 2B and C). Of course, this could be a cumulative effect, but then that would be exactly the sequence of events we might predict from increased arousal thresholds preceding deeper sleep induction. We now provide some more support for this sequence of events in sleep intensity data for R23E10/Chrimson flies (new Figure 2—figure supplement 1D), which is separated in 5min bins. If further exposed to red light, these flies will sleep more deeply, so there appears to be a sequential effect here of decreased behavioral responsiveness while awake followed by increased sleep intensity.

While the GABA-related questions are thought provoking, we are not investigating GABA pathways in this paper. We are showing that removing INX6 from the sleep switch impairs its ability to block behavioral responsiveness upon activation. GABA could still play a role, but our story is about INX6, and the effects are clear.

We decided to nevertheless probe potential downstream synaptic effects, by expressing the temperature sensitive synaptic effectors UAS-ShibireTS and UAS-Syntaxin3-69 in R23E10 neurons. These manipulations should have opposite effects: UAS-shibire impairs synaptic release in neurons upon exposure to elevated temperatures (31°C), whereas syx3-69 increases synaptic release at elevated temperatures. In neither case did we see convincing evidence of altered behavioral responsiveness (or sleep), compared to genetic controls (Figure 4—figure supplement 1). This is in striking contrast with the evidence of effects with INX6 manipulations (see Author response image 1 below). While it remains possible that these synaptic manipulations might not influence some of the downstream synaptic effects likely to be involved here (e.g., neuropeptides or GABA), and there might also be a confounding effect of temperature, we feel these negative results highlight the importance of the electrical activation and INX6 effects we have discovered, which is what our work is focussed on. We have included these synaptic manipulations as new supplement Figure 4—figure supplement 1 in the revised manuscript, as it also provides a good rationale for why we proceeded to investigate electrical communication channels instead.

4) The experiments themselves seem nicely done, but some of the figures make points, which though important to establish, are not highly instructive in themselves. E.g. Figure 3 shows that R23E10 neurons are different in their intrinsic properties (though the significance of this is unclear) and that light driven ChR does indeed result in neuronal firing. Figure 4 shows that innexin6 is expressed in at dFB neurons. The authors may consider integrating these points in summary form with one of the other figures and show more comprehensive images in the figure supplements.

Showing that INX6 is expressed in the dFB is critical for the paper, and we went through quite a bit of work to quantify the overlap between INX6 labeling and R23E10 expression (in the Methods). We are not aware of any whole-cell patch electrophysiology showing sleep switch cell behavior during optogenetic induction, so feel those are valuable figures. Showing the intrinsic properties of these neurons aligns our findings with what others have shown, and this is often a valuable starting point for any electrophysiological study. We can however reconsider how to present our figures to move more effectively to the questions of greatest interest.

Reviewer #2:

[…]

1) The R23E10-Gal4 driver is not so clean and has some VNC expression. Even if the authors dispute this statement, it is prudent to use at least two drivers with dFB expression and show that both give comparable results. For example: R84C10-GAL4 is a strong and clean dFB-GAL4 driver.

It is true that R23E10 has some VNC expression, but the evidence of multiple studies (including our recent paper, Yap et al., 2017) point to the dFB as relevant for these sleep effects, with the R23E10 strain being accepted by the field as a sleep switch Gal4. We have tested R84C10-Gal4 as suggested, and, interestingly, it is not sleep promoting. So, we did not pursue that Gal4 in this paper. This suggests cellular heterogeneity in the dFB, which is the subject of a future paper.

2) It is important to use inhibition as well as activation to back up the claim that the dFB plays an important role in arousal threshold. (Sufficiency does not ensure necessity, and we should now learn which it is.) The Discussion section is written as if this had been done. This request would have been unfair a couple of years ago, but new optogenetic tools have come on-line with widespread use and endorsement. If inhibition fails to have an effect, I would not recommend rejection of the paper, but I think the authors should find out.

We acquired the inhibition reagents (GtACR) but have had limited success in getting them to work reliably, at least in the context of R23E10 expression. On first blush it appears that expressing GtACR in the dFB increases responsiveness as predicted (see exemplary traces and summary stats in Author response image 1). However, this effect seems independent of green light activation, suggesting a leakiness due to ambient light. Then, it also seemed like acute dFB inhibition suppressed behavioral responsiveness (Author response image 1A), similar to what we saw with acute activation (Figure 2F in the paper). However, the non-ATR fed controls show a similar effect, if not stronger suppression (see Author response image 1B). There are multiple, opposing ways to interpret this experiment. Perhaps the green light (to activate the inhibitory channel) causes the control flies to freeze, and this is suppressed when the dFB is acutely inhibited. This would be consistent with our view of a role for the dFB in regulating behavioral responsiveness. Or, it could be the exact opposite: acute dFB inhibition might be suppressing responsiveness, and controls might be showing this too because the effect is leaky even without ATR feeding. Or, the fact that this reagent might already respond to ambient light could be an issue: this could explain the greater overall responsiveness in ATR-fed flies, even outside of the acute green light exposure, compared to non-ATR controls. We see similar results with prolonged light exposure (Author response image 1C), and, oddly, sleep also increases in ATR-fed flies regardless of light exposure (Author response image 1D). Clearly there is a leakiness issue here, so these experiments are difficult to interpret. Our worry is that including these manipulations in the paper (we have repeated and confirmed this) will not help explain much without a lot more optimisation. Including these results will just raise more questions and will break the flow of our activation-centred narrative. Optimising this paradigm to understand what dFB inhibition is actually doing, and to exclude a large number of alternative explanations, will require almost a paper of its own, with associated electrophysiology. Our suspicion is the acute green light induces freezing behavior in some flies (which is already interesting), that this is suppressed when the dFB is inhibited (which could be really interesting), but that GtACR is responding already to ambient light (which is troubling).

However, this is quite tangential to our current story, so I would rather keep this out of the revised manuscript.

Author response image 1
Acute inhibition of the dFB.

(A) Example average activity trace (speed ± SEM) of ATR-fed R23E10/GtACR-3M flies responding to stimuli 15 minutes apart (gray dashed lines); 5min GtACR activation (green shading), with 1min prior to the stimulus, is alternated with trials without green light. (B) Same acute experiment as in A, for non-ATR fed controls. (C) Average responsiveness ( ± SEM) in same flies as in A&B, for a 12-hour experiment. (D) Average sleep duration ( ± SEM) for same flies as in A&B, for a 12-hour experiment. N=16 flies for all..

https://doi.org/10.7554/eLife.37105.023

3) In Figure 7, the authors showed the behavioral responsiveness to stimulus when INX6 knockdown in non-activated R23E10 neurons, and it is comparable to control flies. Does this mean that gap junctions in dFB neurons play no role in arousal regulation under basal conditions? If so, what do the authors think about this?

Mean peak responsiveness (Figure 7D) is the odd one out here, for baseline; both sleep duration and more importantly, sleep intensity, show the expected difference at baseline. Peak responsiveness refers to fly speed immediately after the mechanical stimulus. It is possible that leaky Chrimson expression in ambient light already attenuated the baseline peak responsiveness. We now acknowledge this one baseline inconsistency in our Results section and suggest the likely explanation. Importantly, the effect during acute dFB activation is really clear: INX6 knockdown in the dFB completely blocks the optogenetic effect on behavioral responsiveness.

4) The knockdown experiments should be done with tubpGAL80ts to eliminate developmental effects. And why were the MB lobes were not identified with INX6 staining? In Wu et al. (2011), they reported that INX6 was expressed in the MB and EB as well as the FB.

It isn’t clear to us why we need to show MB staining here? The antibody is working. We could focus on the MB, and turn up the gain. It’s in there. But the dFB is brighter, and this is a dFB study. For tub-Gal80, we have done those experiments, and now include them in the paper (Figure 5—figure supplement2). While effects on sleep duration are not significant, it is clear that daytime responsiveness and sleep intensity are showing the same significant effects if INX6 is downregulated in adults (pink) as for as the chronic manipulation (red, repeated here). This argues against any developmental effects. The lack of significant effects on sleep duration also further solidifies our view that INX6 is primarily affecting behavioral responsiveness levels.

5) Why is daytime baselines sleep so different in Figure 7A than in Figure 5A? This makes me doubt whether the important result in Figure 5A is correct, but perhaps I am missing something. This difference is also true for nighttime sleep, which showed an effect in Figure 5A but no apparent effect in Figure 7A. I understand that in Figure 7A they co-expressed UAS-CsChrimson with INX6 RNAi but CsChrimson should not affect INX6 RNAi-induced less sleep phenotype on the baseline day unless there is a subtle RNAi baseline effect. This should then be tested with GFP RNAi or something like this.

Both are around 20min per hour, for daytime sleep; this is not so different. We attribute the slightly increased sleep duration in the baseline controls (blue) to leaky Chrimson expression in ambient light. It is true that nighttime sleep in these strains does not replicate our Figure 5 results. We don’t know why, but our sleep induction protocol is during the day, so this does not alter our conclusions. It is possible that some long-term homeostatic processes are affected here in these particular strains, which this study does not focus on. We are not sure what the reviewer means by GFP RNAi. The knockdown clearly works (Figure 5—figure supplement 1); any minor discrepancies the reviewer noticed are probably due to leaky or cumulative Chrimson effects. These do not affect the conclusions we draw from the red light activation conditions.

6) I am not convinced the order of events (first arousal threshold, then sleep) is as the authors suggest. The former can be assayed instantly whereas the latter requires some time to qualify as sleep.

We agree that this was poorly framed in the first paper. We have changed the language (e.g., in the Abstract) to frame this differently, without losing any of the impact of our work regarding a distinct role for the dFB here. The reverse order (sleep first, then increased arousal thresholds) obviously makes no sense, so it is either simultaneous or the order we propose. It would be hard to imagine, however, how the engagement of sleep functions would not be preceded by increased arousal thresholds, and why the mechanisms involved in reducing behavioral responsiveness upon sleep initiation would not also operate during the final moments of wakefulness. We now make this more of a discussion point and less of a fait-accompli.

7) Lots of control data is missing for the behavior experiments. For example, GAL4 and UAS control are needed in Figure 2 and Figure 7.

Our controls are appropriate here: the exact same strain with or without ATR. This is accepted for Drosophila optogenetics. This applies to both Figure 2 and Figure 7. Perhaps the referee was confused because we did not show the 3 traces in Figure 7C, which would have been really messy. We show all of the summarised control data in Figure 7D. In other parts of the paper we also show heterozygous genetic controls, and data are consistent. To have to show both genetic controls, both of them with and without ATR, with and without red light, is excessive and would set a strange precedent for future optogenetic work. There is a good reason optogenetics is very appealing: these experiments are blessed with internal controls. For Chrimson, this works very well.

8) Error bars are needed in Figure 2—figure supplement 2A and Figure 2—figure supplement 2E. And why is the behavioral response in Figure 2—figure supplement 2G on the baseline day is so different compared to Figure 2—figure supplement 2C on the baseline day? They should be the same genotype and condition. And what is the explanation why for the 10Hz light pulse activation of R23E10 has no effect on sleep duration?

We apologise for the oversight and have corrected the figure (now Figure 3—figure supplement 3A). We agree that the 10Hz responsiveness data was a little messy and potentially confusing, so we took it out. Our goal for doing the 1Hz 5ms experiments was to provide a more physiological stimulation regime. Since we found similar results, that is sufficient. What other frequencies might or might not be doing is potentially interesting, but beyond the scope of this paper. In future work, we will be looking at a sweep of frequency effects.

Reviewer #3:

[…]

1) Figure 2 is essential in making the case that dFB activation reduces behavioral responsiveness separately from (and perhaps prior to) sleep. However, it is perplexing that responses are "abolished" (subsection section “Optogenetic activation of the sleep switch decreases behavioral responsiveness”) in wake flies with dFB activation, while in sleeping flies, the response is only diminished. Is this a function of needing to normalize to wake activity (Figure 2E)? If so, is wake activity (mean speed) affected by dFB activation? How about movement in response to a strong stimulus? This raises the more general possibility that locomotor function is simply different with dFB activation, as opposed to an actual change in responsiveness to the stimulus at a sensory level. Can the authors address this?

Figure 2 is indeed essential in making the case that dFB activation reduces behavioral responsiveness separately from sleep. We have adjusted the text, added more analyses, and corrected some mistakes to make sure this is absolutely clear. First the mistake: the reviewer was correct in suspecting that responsiveness is not ‘abolished’ in awake flies; rather, it is impaired. In our previous figure, we showed summary stats that looked like responsiveness was abolished (old Figure 2E, right panel). This was a mistake based upon poorly fitted responsiveness curves, where we reported the peaks of fitted curves. Since the awake dFB-activated flies had poor response, this was a poorly fitted curve, leading to what looked like a negative response. We have now corrected this mistake and show the raw data instead (mean peak responsiveness), where it is clear that awake dFB-activated flies respond less than controls (new Figure 2E, right panel).

These data are supported by the acute activation experiment that follows (Figure 2F). This experiment showed that only 1min of red light activation is sufficient to impair responsiveness, even in awake flies. We now provide an additional supplemental figure to address some of the questions raised by the reviewer (Figure 2—figure supplement 2), by focusing only on walking animals. Here we show that (A) Walking speed was not significantly different in the minute immediately preceding the stimulus, for the dFB-activated flies compared to controls; (B), Responsiveness is not significantly different between the experimental groups when the red light is off; (C) and (D), When we only analyse flies that are actively walking, we see a significant decrease in responsiveness in dFB activated flies compared to controls. So, these are both groups of flies that are awake and walking, but only the dFB-activated group is impaired in its behavioral responsiveness to the vibration stimulus.

2) Also in Figure 2, I am confused that the earlier work (Figure 2C and D, excluding sleep intensity) do not distinguish between wake and sleeping flies. Why not only analyze sleeping flies until shifting focus to wake flies in Figure 2E? If analysis of wake flies requires normalizing the baseline activity (Figure 2E right panel), then shouldn't the authors need to account for that in analyzing mean peak responsiveness (Figure 2D, for example)?

Our sleep intensity metric (Figure 2D, right) is exactly what the reviewer is looking for: responsiveness only in sleeping flies. The methodology for this is explained in Figure 2—figure supplement 1. Mean peak responsiveness (Figure 2D, left) indeed combines data from both mobile and immobile flies, as an average. However, we look at only awake flies in panel E and F, and in the new Figure 2—figure supplement 1. We’ve pretty much covered all aspects of the data.

3) Figure 2F is particularly critical, and the authors set it up as if it will answer something about whether the change in responsiveness or sleep comes first (subsection “Optogenetic activation of the sleep switch decreases behavioral responsiveness” "even in awake flies"). However, it appears to again mix/combine sleeping and awake flies for analysis, which is a major confound. In other words, what if most of these flies are rapidly put to sleep with the 1 minute dFB activation prior to stimulus? Then the figure simply shows that inducing sleep with dFB changes responsiveness, which we already know (from this work, but also from Donlea et al. (2011) and Donlea, Pimentel and Miesenbock (2014) – it is important to note that these published works have already demonstrated a change in arousability with dFB activation, even if not with such detail as shown here).

Please see our response to similar points above. Most of these flies are not rapidly put to sleep within the 1min of dFB activation, which is why the average speed is the same as baseline. We now provide analysis on only awake, walking flies as well, and the results are almost identical (this is because most flies were walking in the 1min before the stimulus, even with dFB activated). That said, some flies do become immobile very rapidly, and some gradually, leading to increased sleep duration and intensity after 5min. We could dissect these dynamics a little more, but the key point here is that responsiveness can be abolished in awake flies that are still walking around. This dissociates the responsiveness functions of the dFB from its sleep-inducing functions, at least temporally. This is a very critical observation, as the reviewer has noted. We hope this is now clearer.

Please note that we have also analysed the results from Figure 7 in awake flies only and see similar effects. The reviewer may have missed this as these results are stated in the text (subsection “Optogenetic activation of the sleep switch decreases behavioral responsiveness”). We can provide a figure for that if requested.

We appreciate that others have demonstrated increased arousal thresholds in dFB activated flies. However, the detail whereby we elaborate on the question of behavioral responsiveness is a major point of our paper. The dFB sleep studies that have previously addressed responsiveness have used it only as a ‘sanity check’ for sleep. Without that, it would not have been convincing to call these neurons sleep promoting. What we report here is a broader role for these neurons, to control behavioral responsiveness more generally. Arousal thresholds are more than just a sanity check for sleep studies. This is a necessary and complementary research direction for understanding sleep processes, and how some of these might operate during wakefulness as well. This is a major point of our paper, and we hope that this is clearer now.

4) The authors make a point of 1Hz and 10Hz optogenetic activation of dFB both changing responsiveness, but only 1Hz inducing sleep, therefore dissociating the two. However, these data are not convincing. In Figure 2—figure supplement 2G, why is the response so poor at baseline compared to C in this same figure? And why suddenly so much noise? Moreover, the change in responsiveness (H) appears similar +/- ATR: the baselines appear different and the 10Hz light causes a similar decrease in both (same issue in D of this figure). In other words, I am not convinced that there is a change in responsiveness, and therefore the ability to dissociate sleep time from arousal is hindered. The sleep intensity data is more convincing, but this is a comparatively less precise measure.

We disagree with the last point: sleep intensity is the more precise measure. With sleep intensity, we are looking at how only sleeping (immobile for >5min) flies respond to the mechanical stimulus, rather than an amalgam of awake and sleeping flies. We agree on the first point though, the 10 Hz data in panel G was messy. We’ve taken the 10Hz data out. It isn’t crucial for this study. We’ll redo it and put it use it somewhere else. It is interesting though.

5) In Figure 7, there are concerns. First, in contrast to Figure 5A, there is no longer an effect of INX6 RNAi expressed using R23E10-GAL4 (Figure 7A, left). Is there dilution of UAS-driven expression by co-expressing two UAS? Is INX6 knockdown even effective here? If we assume so, and if INX6 is mostly important in the context of dFB activation, then the authors should test these same measures following sleep deprivation. That would be most convincing with regard to a role for INX6 in modulating dFB driven sleep intensity. Overall, it would add a lot to this work for the authors to perform many of the experiments in Figure 2 after sleep deprivation, which is more physiologically relevant and should induce changes in behavioral responsiveness.

Please see our reply to reviewer #2 above, who had a similar question about these data. The suggestion to do sleep deprivation experiments is interesting: one prediction would be that less dFB activation would be required to achieve a similar level of behavioral inhibition. How to titrate that exactly would be an entirely new study. Rather, our study recurrently returns to electrophysiology to provide a level of understanding on the likely mechanisms involved, which we suggest pertains to oscillations generated by these neurons (as in Yap et al., 2017). We are hesitant at this point to engage in an entirely new manipulation (sleep deprivation) which might then need to be applied to all of our readouts, including electrophysiology. This is an excellent suggestion, but for a different study. That said, we have shown previously (in Yap et al., 2017), that 12 hours of dFB activation essentially produces a sleep-deprived fly (Figure 6E in Yap et al., 2017), consistent with our view that this is activating a distinct (transitional) sleep stage. There are bound to be confounds that will inevitably require a full-blown paper to explore and explain, hence our hesitation to approach this thorny area at the revision stage for this paper. We will be fully exploring these issues (and the unique genomics underlying dFB-induced sleep) in another paper (in preparation).

https://doi.org/10.7554/eLife.37105.025

Article and author information

Author details

  1. Michael Troup

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Melvyn HW Yap
    Competing interests
    No competing interests declared
  2. Melvyn HW Yap

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Conceptualization, Resources, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft
    Contributed equally with
    Michael Troup
    Competing interests
    No competing interests declared
  3. Chelsie Rohrscheib

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Resources, Formal analysis, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
  4. Martyna J Grabowska

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Resources, Software, Formal analysis, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1727-7714
  5. Deniz Ertekin

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Resources, Software, Formal analysis, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
  6. Roshini Randeniya

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Formal analysis, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1340-750X
  7. Benjamin Kottler

    1. Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    2. King's College London, London, United Kingdom
    Contribution
    Conceptualization, Resources
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4551-5791
  8. Aoife Larkin

    1. Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    2. University of Cambridge, Cambridge, United Kingdom
    Contribution
    Data curation, Formal analysis, Visualization, Methodology
    Competing interests
    No competing interests declared
  9. Kelly Munro

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Formal analysis, Investigation, Visualization
    Competing interests
    No competing interests declared
  10. Paul J Shaw

    Washington University School of Medicine, St Louis, United States
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing—original draft, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
  11. Bruno van Swinderen

    Queensland Brain Institute, The University of Queensland, Brisbane, Australia
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    b.vanswinderen@uq.edu.au
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6552-7418

Funding

National Institutes of Health (R01 NS076980)

  • Melvyn HW Yap
  • Paul Shaw
  • Bruno van Swinderen

National Health and Medical Research Council (GNT1065713)

  • Michael Troup
  • Chelsie Rohrscheib
  • Aoife Larkin
  • Bruno van Swinderen

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Leonie Kirszenblat for help and comments on the manuscript. We thank Eleni Notaras for help with behavioral experiments. We also thank Chia-Lin Wu for the INX6 antibody. This work was supported by an NIH grant RO1 NS076980-01 to PJS and BVS, and by an NHMRC grant GNT1065713 to BVS. The authors declare no conflicts of interest.

Senior Editor

  1. K VijayRaghavan, National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

Reviewing Editor

  1. Mani Ramaswami, Trinity College Dublin, Ireland

Publication history

  1. Received: April 22, 2018
  2. Accepted: August 14, 2018
  3. Accepted Manuscript published: August 15, 2018 (version 1)
  4. Version of Record published: August 30, 2018 (version 2)

Copyright

© 2018, Troup et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 13,099
    Page views
  • 475
    Downloads
  • 4
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)