Sleep- and wake-dependent changes in neuronalactivity and reactivity demonstrated in fly neuronsusing in vivo calcium imagingDaniel Bushey1, Giulio Tononi, and Chiara Cirelli1

Department of Psychiatry, University of Wisconsin–Madison, Madison, WI 53719

Edited by Joseph S. Takahashi, Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, TX, and approved March 2, 2015(received for review October 11, 2014)

Sleep in Drosophila shares many features with mammalian sleep,but it remains unknownwhether spontaneous and evoked activityof individual neurons change with the sleep/wake cycle in flies asthey do in mammals. Here we used calcium imaging to assess howthe Kenyon cells in the fly mushroom bodies change their activityand reactivity to stimuli during sleep, wake, and after short orlong sleep deprivation. As before, sleep was defined as a periodof immobility of >5 min associated with a reduced behavioral re-sponse to a stimulus. We found that calcium levels in Kenyon cellsdecline when flies fall asleep and increase when they wake up.Moreover, calcium transients in response to two different stimuliare larger in awake flies than in sleeping flies. The activity ofKenyon cells is also affected by sleep/wake history: in awake flies,more cells are spontaneously active and responding to stimuli ifthe last several hours (5–8 h) before imaging were spent awakerather than asleep. By contrast, long wake (≥29 h) reduces bothbaseline and evoked neural activity and decreases the ability ofneurons to respond consistently to the same repeated stimulus.The latter finding may underlie some of the negative effects ofsleep deprivation on cognitive performance and is consistent withthe occurrence of local sleep during wake as described in behavingrats. Thus, calcium imaging uncovers new similarities between flyand mammalian sleep: fly neurons are more active and reactive inwake than in sleep, and their activity tracks sleep/wake history.

Kenyon cells | sleep deprivation | mushroom bodies | GCaMP5

The fundamental features that characterize mammalian sleepalso define Drosophila melanogaster sleep (1–3). Most cru-

cially, in both flies and mammals, sleep is distinguished fromsimple rest (quiet wake) by an increased arousal threshold, i.e.,a reduced ability to respond to external stimuli. Moreover, inflies and mammals sleep is controlled homeostatically by theduration as well by the intensity of prior wake, suggesting basicsimilarities in the mechanisms of sleep regulation across species.Thus, in flies both sleep deprivation and a rich learning experi-ence lead to a sleep rebound characterized by overall increasedsleep time, increased arousal threshold, and longer sleep epi-sodes (4–6). As in mammals, overall neuronal activity in flies isalso high during wake and low during sleep (6–8). Specifically,a seminal study using local field potential (LFP) recordings fromthe Drosophila medial protocerebrum found high spike-likepotentials that disappeared after the block of synaptic trans-mission in the mushroom bodies (MBs) (7). This high spike ac-tivity was present when flies were moving or had been quiescentfor only a few seconds, but disappeared with sleep (i.e., afterperiods of immobility >5 min), when the overall LFP power in allfrequencies also decreased by ∼60% (7). The study concludedthat neural activity in the sleeping fly brain, or at least in a cen-tral region spanning the MBs, resembles that seen in mammals inseveral brainstem cell groups including noradrenergic neurons,whose firing strongly declines or stops completely during sleep(9). A more recent study in tethered flies able to walk on an air-suspended ball also found that periods of immobility >5 min are

associated with increased arousal thresholds and with “flat”LFPs (6). LFPs, however, reflect the activity of thousands of cells(10), and the use of modified stereotrodes to resolve single-unitactivity in flies is still in its infancy (11). Glass or tungstenmicroelectrodes, on the other hand, have recorded one neuronat a time in flies (e.g., refs. 12–14) and, to our knowledge, havenot measured changes across sleep and wake.Our goal was to study many fly neurons simultaneously while

preserving single-cell resolution, to determine how sleep andwake affect spontaneous activity and the ability of neurons toreact to stimuli. It was recently shown in rats that during sleepdeprivation single cortical neurons may go unpredictably “off-line,” as they normally do during sleep, with negative effects onperformance (15). Thus, we also asked whether this phenome-non occurs in flies. We focused on the MBs, large areas of theDrosophila brain involved in olfactory learning (16, 17) and sleepregulation (18, 19), and used in vivo calcium imaging (20, 21) tomeasure spontaneous and evoked neuronal activity in the MBprincipal neurons, the Kenyon cells, during sleep, wake, and inresponse to different periods of sleep deprivation.

ResultsSleep and Wake States During Calcium Imaging. We expressed thegenetically encoded calcium sensor GCaMP5 (22) in the Kenyoncells. First, we measured changes in GCaMP5 fluorescenceacross the sleep/wake cycle within the same fly. Before imaging,flies were allowed to sleep for 5 h during the first part of the 12-hdark period, when they normally sleep, or were sleep deprived to

Significance

Sleep in fruit flies shares all the essential features of mamma-lian sleep. Here, by using in vivo calcium imaging, we show forthe first time, to our knowledge, that neuronal activity andreactivity decline during sleep and increase during wake si-multaneously in many cells of the fly brain. Furthermore, weshow that long wake reduces baseline and evoked neural ac-tivity and causes neurons to respond inconsistently to stimuli.The latter finding is reminiscent of the phenomenon of “localsleep in wake” described in rats, in which single cortical neu-rons unpredictably go “offline” during extended wake, leadingto performance errors. Thus, these findings open the way touse Drosophila to study the molecular mechanisms underlyingthe cognitive deficits caused by sleep loss.

Author contributions: D.B., G.T., and C.C. designed research; D.B. performed research;D.B. analyzed data; and D.B., G.T., and C.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. Email: ccirelli@wisc.edu or bushey@wisc.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419603112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1419603112 PNAS | April 14, 2015 | vol. 112 | no. 15 | 4785–4790

NEU

ROSC

IENCE

increase their chance to sleep under the microscope. Fliesallowed to sleep were awake only 23 ± 1.7% (mean ± SE) of the5 h before imaging, whereas sleep-deprived flies were awake 99 ±0.0% of the time. Imaging occurred at the same circadian time inall flies. Under the two-photon microscope, flies were placed ina holder that fixed the head in place while legs and abdomenremained free to move and were recorded using an infrared (IR)digital microscope (Fig. 1A; details inMaterials and Methods). MBcells expressing GCaMP5 are located in the periphery on theposterior side of the head (Fig. 1B). Imaging took place every10 min over a 2-h period, with each stack of images taken at 0.7-μmintervals throughout the MB region over a 67-s time period.Sleep in Drosophila is defined as any period of immobility of

more than 5 min, because in many experimental conditions, in-cluding in tethered flies (6), a quiescent period of that length isassociated with a stable increase in arousal threshold. Accordingto this definition, the first minute counted as sleep correspondsto the sixth minute of inactivity, because in previous experimentswe found that arousal thresholds progressively increase duringthe first 5 min of quiescence (23). This period is therefore viewedas a transitional stage and counted as wake. Under the two-photon microscope, we used legs and abdomen movements toidentify putative periods of sleep (>5 min of immobility). Thus,a putatively sleeping fly had to show no movements for morethan 5 min before and during imaging, whereas a putativelyawake fly had to move at least once during that time. The ab-sence of movements alone, however, is not sufficient to definesleep unless associated with an increased arousal threshold.

Thus, we measured whether putatively sleeping flies were lessable to respond to a stimulus, namely the switching on of the IRlaser light when imaging started. In the holder under the mi-croscope, awake flies are not continuously in motion but tend tomove in intense bouts of activity of 1–2 min over a 5-min period.We found that switching the laser on often triggered movementsor, less frequently, caused the movements already present tobecome more intense. Specifically, across all imaging sessions,flies moved in 59% of cases (n = 191 sessions) when the laserlight went on, but only in 36.3% of cases (n = 160) during controlperiods when the laser light was off, which were selected to occurhalf way (at the fifth min) between two stimulation periods (χ2test, P = 2.31E-5) (Fig. 1C, Left). More importantly, awake fliesresponded more frequently to IR stimulation than those puta-tively asleep. Specifically, in flies that were already moving be-fore imaging, movements occurred in 81.3% of the episodes (n =101) when the laser was switched on (Fig. 1C, Right). By contrast,flies that were putatively asleep (quiescent) before imagingmoved only 39.2% of the time (n = 90), and the difference in theresponse rate between the two groups was greater than chance(χ2 test, P = 4.00E-9) (Fig. 1C, Right). Thus, the IR laser light hasa stimulatory effect on the great majority of awake flies, whereasin most cases (60.8%) flies presumed asleep did not move inresponse to the laser. We conclude that the arousal threshold isincreased in the state that we previously called putative sleep,and from now on we will refer to that state as “sleep.”Another essential feature of sleep is its homeostatic regula-

tion, i.e., quality and quantity of sleep increase after sleep dep-rivation. Consistent with our conclusion that >5 min of qui-escence qualifies as sleep in our experimental conditions, wefound that during the 2-h imaging session flies that had pre-viously been sleep-deprived spent more time asleep (Fig. 1D)and had more (Fig. 1E) and longer sleep episodes (Fig. 1F) thanflies that slept before imaging. Moreover, quiescent flies thathad previously been sleep-deprived for 29–34 h had a lower re-sponse rate to IR stimulation than quiescent flies that had pre-viously been sleep-deprived for only 5–8 h (response rate: 13% ±7.2 vs. 32% ± 9.1; χ2 test, P < 0.015; n of flies sleep-deprived(SD) 5–8 = 22, SD 29–34 = 21; Fig. S1B). Spontaneous motoractivity during control periods with laser off (2 min before thelaser was switched on) did not differ between the two SD groups(χ2 test, P = 0.2).The mechanism underlying the arousal response due to the

laser is unknown, but thermal stimulation appears a more likelycandidate relative to acoustic or visual stimulation. Indeed, wefound that the response rate increases with increasing laserpower (Fig. S1A), whereas the noise produced during imaging bythe galvo mirror system and shutters does not change with dif-ferent laser intensities. Moreover, D. melanogaster cannot senseIR wavelengths directly, and insects that sense IR do so throughan organ that uses thermoreception (24). By using a no-contactthermometer centered below the fly, we found that temperatureincreased by 0.18 ± 0.01 °C during stimulation (baseline = 22.16 ±0.047 °C; during stimulation = 22.34 ± 0.046 °C, Mann–Whitneytest P = 0.011, three flies, each fly tested five times) (Fig. S1C).The quick rise in temperature is consistent with the fact that theresponse to the laser is quick. Indeed, by dividing the stimula-tion period into 5-s epochs and scoring each epoch for activitywe found that most responding flies (81.5%) moved within thefirst 5 s after the laser was turned on (14 awake flies, 79 trials;100% laser power tested every 15 min for 60 s). Finally, Trp1A1

null flies, which have an impaired response to temperaturegradients (25, 26), failed to show a significant difference inresponse rate when the laser was switched on relative to when itwas off (Fig. S1D).The laser-induced increase in temperature that we measured

is small, but was detected across a region that included bothhead and body. It is likely that the temperature increased more

Fig. 1. Identification of sleep and wake states during calcium imaging.(A) Experimental setup. The fly is positioned between the two-photonmicroscope and the infrared digital microscope, with the head attached toa chamber holding oxygenated saline solution (blue). Airflow passes througha hose ending in front of the head while a vacuum in the back expels theairflow. (B) Diagram detailing MB position in the fly brain. Cell bodies andcalyx region (gray box) were imaged from the back of the head either asa stack of images throughout the entire cell body and calyx region (graybox) or as a time series from a single plane (pink). (C) Probability of move-ment in response to the switching on of the laser light vs. control stimulation(Left) with laser light on in previously moving flies vs. previously quiescentflies (Right). Laser power was set at 100%, and the intensity was used toassess GCaMP5 fluorescence. χ = P < 0.05 between groups (Pearson’s χ2 test).(D–F) Quantity and quality of sleep during testing in flies that either slept (S)or were sleep-deprived (SD) before testing. *P < 0.05 between sleep andsleep-deprived groups (Mann–Whitney test). Number of flies = 17 (sleeping)and 18 (awake).

4786 | www.pnas.org/cgi/doi/10.1073/pnas.1419603112 Bushey et al.

over the head region, where the laser was focused and heat-sensitive neurons are located (26). Moreover, at least in larvae,Trp1A1 is necessary to show preference even within a narrowtemperature range, 18 °C over 19 °C (25). Thus, although notconclusive, these results are consistent with heat being the mostlikely candidate to mediate the arousal response to the laserstimulation.

Calcium Levels in Individual MB Cells Change When Flies Fall Asleepand Wake Up. Having established that flies can sleep under thetwo-photon microscope, calcium levels were measured at thewake-to-sleep and sleep-to-wake transitions. We focused onconsolidated bouts of sleep and wake (>5 min) to measurechanges starting from a stable baseline and used flies previouslysleep-deprived for 8 h to increase their chance of sleep. Asshown in Fig. 2A, individual brightly fluorescing cells could easilybe identified in the stacked images and were defined as thoseneurons whose fluorescence intensity exceeded backgroundfluorescence by at least two times. These bright cells were likelyresponding to environmental conditions such as the constantoxygen-enriched flow passed through the chamber during the 2-himaging session. GCaMP5 fluorescence decreased in these cellsat the wake-to-sleep transition. Specifically, a significant declinein GCaMP5 signal was seen between the last 5 min of wakerelative to the first 10 min of sleep, as well as between the last5 min of wake relative to the end of the sleep period (Fig. 2B). Bycontrast, GCaMP5 fluorescence increased in these cells at thesleep-to-wake transition (Fig. 2C). Of note, no changes in ΔF/Fsignal occurred from the beginning to the end of a 10-min periodof either continuous sleep (Wilcoxon signed-rank test, P = 0.471)or wake (P = 0.082), demonstrating that the changes in calciumlevels in these cells reflect changes in behavioral state.

Sleep/Wake Changes in Calcium Levels in Response to Two DifferentStimuli.We then studied how sleep and wake affect the neuronalresponse evoked by acute exposure to oxygen and then to anoxygen–vinegar mix (Fig. S2). These stimuli were selected to testresponsiveness to a weak stimulus (oxygen) and a strong stimulus(vinegar). The chances of measuring the response to two dif-ferent stimuli in the same fly during both sleep and wake are low.Thus, we used two independent experimental groups, comparingthe evoked response between flies that remained awake or sleptfrom the start of the imaging session to when the stimuli weredelivered. Analysis of single cells was done using stacked imagesacquired as in the previous experiment, but the testing periodwas shortened to 14 min to increase the number of flies thatcould be tested. In the stacks, cells were defined as “responders”if their fluorescence in response to oxygen or vinegar increasedby at least 50% relative to baseline. The number of individualcells responding to oxygen and vinegar was larger in awake fliesthan in sleeping flies (Fig. 3A). Moreover, for both stimuli, themean increase in fluorescence in each responding cell was largerin awake flies relative to sleeping flies (Fig. 3B). When the entireregion of cell bodies was considered, the average response wasalso larger during wake than during sleep (Fig. 3C). In fact,sleeping flies showed no response to oxygen (Wilcoxon signed-rank test, P = 0.192), in contrast to awake flies. In response tovinegar instead, sleeping flies also showed a response (Wilcoxonsigned-rank test P < 0.05), but smaller than that of awake flies(Fig. 3C). The overall weaker response to oxygen relative tovinegar is expected because prior studies have shown that thereare no oxygen-specific sensors in the antennal lobe and the re-sponse to oxygen results from either the airflow itself (27) ora decrease in carbon dioxide concentration (28–30). Finally, wealso measured the response to oxygen and vinegar using a timeseries optimized for rapid imaging of a single plane spanning thecell body region and the calyx (see Materials and Methods fordetails). The peak regional ΔF/F response to both stimuli tendedto be larger in awake flies than in sleeping flies (Fig. 3D). In thebody region, the latency to the peak response to vinegar was alsofaster in awake animals (14 ± 1.5 vs. 9 ± 1 s, P = 0.04), resultingin a significantly larger ΔF/F at the beginning of the stimulationin awake animals compared with sleeping animals (Fig. 3D,green circles). Fluorescence decreased at the end of the timeseries in the absence of stimulation, but this decline did not occurin the calyx or in the presence of stimuli, suggesting that it isunlikely due to photobleaching. Overall, these results using twodifferent stimuli find that sleep is associated with reduced re-activity of Kenyon cells.

Effects of Sleep/Wake History on Calcium Levels and Response toStimuli. In mammals, overall levels of neuronal activity dependon the behavioral state and also, albeit to a smaller extent, onsleep/wake history. In rats, for example, average firing rates ineach behavioral state (sleep or wake) are higher or lowerdepending on whether the previous several hours were spentawake or asleep, respectively (31). Thus, we asked whethera previous history of sleep, short wake or long wake, affectscalcium levels. For this analysis we used only flies that wereawake during the entire imaging protocol, subdivided into threegroups—sleep, short wake (5–8 h), and long wake (29–34 h)—based on their sleep/wake behavior in the previous 5–34 h (seeMaterials and Methods for details).Analysis of stack images found that, compared with flies that

slept before testing, flies previously awake for 5–8 h showeda larger number of bright cells during baseline, i.e., before anyodor was presented (Fig. 4A), and more activated cells in re-sponse to both oxygen and vinegar (Fig. 4B). By contrast, fliespreviously awake for many hours (long-wake group) did notdiffer from flies previously asleep, neither in the number ofbright cells at baseline nor in the number of cells responding to

Fig. 2. Sleep/wake changes in calcium levels during baseline. (A) Rep-resentative images showing changes in GCaMP5 fluorescence across sleep andwake. Mean intensity levels are shown over the cell body region (Reg.)and in two individual cells. Images show the sum intensity from multi-ple layers primarily containing cell bodies. Individual cell bodies can beseen fluorescing above background level (arrows point to two individualcell bodies). Time line spanning 1 h of the 2-h protocol shows wake(green) and sleep (red) periods. (B) Decline in GCaMP5 fluorescence insingle cells at the wake-to-sleep transition. The change in fluorescence[ΔF/mean(F)] was calculated by subtracting the mean fluorescence of thelast 5 min of wake from either the mean fluorescence of the first 10 minof sleep (+10) or the mean fluorescence of the entire sleep bout (end),divided by the mean fluorescence of both wake and sleep periods (13 flies,257 cells, 23 sleep episodes). (C ) Increase in GCaMP5 fluorescence insingle cells at the sleep-to-wake transition. The change in fluorescence[ΔF/mean(F)] was calculated by subtracting the mean fluorescence of thelast 5 min of sleep from either the mean fluorescence of the first 10 minof wake (+10) or the mean fluorescence of the entire wake bout (end),divided by the mean fluorescence of both wake and sleep periods(15 flies, 267 cells, 26 wake episodes). In B and C, #P < 0.05 (Wilcoxon signed-rank test).

Bushey et al. PNAS | April 14, 2015 | vol. 112 | no. 15 | 4787

NEU

ROSC

IENCE

odors, in both cases due to increased variability (Fig. 4 A and B).Overall fluorescent intensity in the calyx or cell body region wasnot affected by the sleep/wake history, neither in the time seriesnor in stack images, although a trend was present, consistentwith the changes seen in single cells (short wake > long wake >sleep). Overall, these results suggest that, as in rodents, theeffects of sleep/wake history on neuronal activity are subtle andthus can be seen when measuring calcium levels in the fewsingle cells that are active at baseline, but not when averagingGCaMP5 fluorescence across the entire region, where most cellsare not active.We then studied the effects of sleep/wake history on the re-

sponse to repeated stimulation. Image stacks were taken duringexposure to oxygen followed by oxygen–vinegar, and this se-quence was repeated three times to assess the consistency ofthe evoked responses. Cells that responded to oxygen and/or oxy-gen–vinegar (>50% ΔF/F) during any of the three trials wereidentified, and the correlation across the activation patterns of alltrials was compared. Fig. 4C shows raw data for representativeexperiments: each of the 12 panels refers to one fly, with eachrow representing the response of one individual cell to the threetrials for oxygen and for oxygen–vinegar (OVOVOV). Acrossindividual flies, the responding patterns across trials correlatedsignificantly after sleep and after short wake (Wilcoxon signed-rank test, P < 0.05) (Fig. 4D). After long wake, however, acti-vation patterns were no longer correlated (Fig. 4D) due toa reduced number of high-intensity responses compared withsleep (Fig. 4E). As shown in Fig. 4C, this decline could occur inany of the trials, not necessarily in the last one. Indeed, the lackof correlation after long wake was also present when trials 1–2,2–3, and 1–3 were considered separately, and mean raw intensityfor the responders did not decrease from the first to the last trial,indicating that habituation does not account for this result.

DiscussionSeveral groups have recorded GCaMP fluorescence from theMBs in tethered flies (27, 32–35), but none of the previousstudies tested the effects of sleep and wake. To do so, it was firstnecessary to prove that flies can indeed sleep during calciumimaging. We show here that, when flies are immobile under thetwo-photon microscope for >5 min, their ability to behaviorallyrespond to a stimulus—e.g., switching on the laser light—is re-duced. We call this state of reduced arousal threshold “sleep,”applying the same criteria used in previous studies that foundthat tethered flies, when asleep, have reduced neuronal activity(6, 7). We also show, again consistent with previous results (6),that sleep as defined in the current study is homeostaticallyregulated by the duration of previous wake. Thus, we concludethat the state of quiescence as defined in our experimental setupqualifies as sleep and not simply as rest.

State-Dependent Effects. The most stringent condition under whichto study the effects of sleep and wake on calcium levels is to

compare GCaMP5 fluorescence within the same fly while theanimal spontaneously transitions from wake to sleep and back inthe absence of acute stimulation. By doing so, we found thatcalcium levels in individual MB cells increased when the fly wokeup and declined when it fell asleep, consistent with a decrease inspontaneous activity during sleep relative to wake as reportedusing LFP recordings (6, 7) (Fig. 2). In addition to spontaneousactivity, evoked activity in response to oxygen and vinegar wasalso higher in wake than in sleep. Both the number of MB cellsresponding to these stimuli and their mean response were greaterin awake flies relative to sleeping flies (Fig. 3). A decline in theelectrical response of neurons to different kinds of stimuli isa classical marker of sleep, documented in mammals (e.g., ref. 36)as well as in bees (37). Thus, we were able with calcium imaging todetect sleep/wake-dependent changes in both spontaneous andevoked activity of Kenyon cells.In the mammalian cortex, most of sleep [non-rapid eye

movement (NREM) sleep] is associated with a decrease in glu-tamate and acetylcholine levels, an increase in GABA levels(38), and an overall decrease in the level of activity of arousalpromoting neuromodulatory systems such as the noradrenergic,orexinergic, and dopaminergic systems (39, 40). All these factorsare likely to contribute to the decline in neuronal activity andexcitability during sleep. Similarly, it is likely that severalmechanisms exist to make Kenyon cells more active and moreresponsive to stimuli during wake than during sleep. The neuronsprojecting from the antennal lobe to the calyx are cholinergic.In addition to acetycholine, receptors for GABA, glutamate,octopamine, and dopamine are also expressed in the MBs (41).Direct evidence for sleep/wake changes in the brain levels of anyof these neurotransmitters is missing in flies. However, bothdopamine (42, 43) and octopamine (44, 45) promote arousal,although they do so by affecting brain regions other than theMBs (43–45). Moreover, both dopamine and octopamine mod-ulate calcium and cAMP levels (46) in response to cholinergicinputs and play an important role in modulating MB neuronalactivity during associative olfactory learning (47–50), suggestingthat they could affect the excitability of Kenyon cells. Also,suppressing GABA activity reduces sleep time in flies, andmutations in the GABAA receptor confer resistance to GABAantagonists that promote wakefulness (51). As for dopamine andoctopamine, the sleep-promoting effect of GABAA receptors ismediated outside the MBs by the pigment-dispersing factorneurons (52). However, GABAergic inputs through the ante-rior paired lateral neurons modulate the odor-specific activa-tion pattern in the MBs (53). Reducing GABAA receptorexpression increases the number of cells included in the pattern,and suppressing GABAB expression increases calcium influx in theresponding cells. Thus, an increase in GABAergic transmissionmediated by both GABAA and GABAB receptors could explainwhy sleep is associated with a decrease both in the number ofresponders and in the extent of their response to stimuli.

Fig. 3. Evoked calcium responses during sleep and wake.(A) Number of responding cells (ΔF/F > 50%) to the first exposureto oxygen and vinegar. The number of flies tested were sleep = 18and wake = 42. (B) ΔF/F in individual cells responding to oxygenand vinegar. The number of cell bodies tested were 147:654 (sleep:wake) during oxygen and 171:756 during vinegar. (C) ΔF/F inthe cell body region. #P < 0.05 relative to no response (Wilcoxonsigned-rank test). In A–C, *P < 0.05 between sleep and wakegroups (Mann–Whitney test). (D) Mean change in fluorescence(ΔF/F) during the time series. The odor sequence is as follows:no airflow (light gray), oxygen-enriched airflow (gray), vinegar(black), oxygen-enriched airflow (gray), and no airflow (lightgray). Wake points are circled when Mann–Whitney P values are<0.05 compared with sleep. Dotted line indicates SE.

4788 | www.pnas.org/cgi/doi/10.1073/pnas.1419603112 Bushey et al.

History-Dependent Effects. Analysis of stack images found that,compared with flies that slept before testing, flies previouslyawake for 5–8 h showed more bright cell bodies during baseline,i.e., before any odor was presented (Fig. 4A), and more activatedcells in response to both oxygen and vinegar (Fig. 4B). This resultis consistent with the finding in mammals that spontaneousneuronal activity increases with time spent awake and decreasesafter sleep (31).We found that, after long wake, Kenyon cells were no lon-

ger able to respond consistently to repeated exposure to thesame stimulus. It is unlikely that this result is due to habitua-tion because the correlation among activation patterns didnot decline in the last two trials relative to the first two trials.Instead, the results are more in line with recent findings inmammals suggesting that sleep may not be as global as pre-viously thought. For example, intracranial recordings in epi-leptic patients have shown that the slow waves and spindles ofNREM sleep often occur in some cortical areas but not inothers (54) and frequently involve only small groups of neu-rons (55, 56), a finding that was confirmed in rats (15). Re-cordings of evoked responses in the rat barrel cortex afterwhisker stimulation also suggest the possible occurrence of“local wake during sleep”: responses sometimes appear sleep-like in one cortical column but wake-like in another (57). Fi-nally, by using intracortical multiarray recordings, we recently

found that, when rats stay awake longer than usual to exploreand learn, small groups of cortical neurons can go offline whilethe rest of the brain is awake (15), a phenomenon called “localsleep during wake.” Thus, the few Kenyon cells that un-predictably fail to respond to stimuli after long wake may re-flect the occurrence of local sleep in awake flies. In rats, theoccurrence of local sleep during wake leads to performanceerrors (15) and is likely to account for some of the cognitivedeficits associated with sleep deprivation in humans (58). TheMBs have a pivotal function in olfactory memory processing(59, 60), and sleep deprivation either before or after training(4, 61) impairs olfactory memory, whereas blocking neuro-transmission in the MBs during extended wake before trainingcan rescue memory impairments (61). Thus, it is possible thatat least one of the mechanisms by which extended wake im-pairs olfactory memory involves the occurrence of local sleepin Kenyon cells.

Materials and MethodsFly Strains, Husbandry, and Sleep/Wake Recordings Before Calcium Imaging.Flies tested were generated by crossing Canton-S (CS) wild type (BarryGanetzky, Laboratory of Genetics, University of Wisconsin-Madison) to w1118;UAS-GCaMP5.003; P{w[+mW.hs]=GawB}OK107. The P{w[+mW.hs]=GawB}OK107(stock no. 854) transgene and TrpA11 mutant (stock no. 26504) were orderedfrom the Bloomington Drosophila Stock Center. The UAS-GCaMP5 transgenewas provided by Vivek Jayaraman, Howard Hughes Medical Institute,

Fig. 4. Effects of previous sleep or wake on baselinecalcium levels and response to stimuli. (A) Number ofspontaneously active cells during baseline. (B) Num-ber of cells that responded to oxygen or vinegarthroughout the three trials (at least in one of thethree trials; stacked images). In A and B, the numberof flies tested were sleep = 22, short-wake = 31,long-wake = 17. *P < 0.05 relative to sleep (post hocKruskal–Wallis test followed by Mann–Whitneytest). All flies included were awake when tested.(C) Examples of cell response profiles from singleflies that were asleep or awake for a short or longtime before testing. Each row is a single cell, andcolumns indicate the ΔF/F response during the threetrials for oxygen (O) and vinegar (V). Note that, be-cause the number of responding cells varies acrossflies, the size of the panels and the width of the rowsis not the same across flies. (D) Mean Pearson cor-relation (r) comparing the activation pattern in fliesacross trials (1–2, 2–3, 1–3) in response to oxygen(Left) and vinegar (Right). All flies were awake dur-ing testing and divided based on prior behavior:sleep, short wake, long wake. An r score of 1 wouldindicate a perfectly matched pattern. A “#” indicatesthat median values significantly differ from nocorrelation (Wilcoxon signed-rank test, P < 0.05).(E) Frequency plots describing the range in meanfluorescent response for each cell. Responses weresubdivided based on their intensity into bins of 268relative units, and the rate (number of responses/total number of cells) within each bin is plotted forthe three groups. For each bin, an asterisk indicatesa significant difference from the sleep group (Krus-kal–Wallis test followed by Mann–Whitney test,P < 0.05). Note that the frequency of high-intensityresponses is also reduced after short wake relative tosleep, but the overall increased number of responsescompensates for this effect.

Bushey et al. PNAS | April 14, 2015 | vol. 112 | no. 15 | 4789

NEU

ROSC

IENCE

Janelia Research Campus. Flies were reared at 20 °C on standard cornmealmolasses. Virgin females were harvested within 12 h after eclosure andplaced in clear 5- × 65-mm tubes with standard cornmeal molasses food atone end and a cotton-sealed cap in the other. Behavior was monitored inDrosophila activity monitors (DAM5, Trikinetics) at 20 °C and 68% humidity.Sleep deprivation using a mechanical agitator was performed as previouslydescribed (23).

ACKNOWLEDGMENTS. We thank Dr. Vivek Jayaraman (Howard HughesMedical Institute, Janelia Research Campus) for providing the GCAMP5 stockand advice on fly preparation for calcium imaging and Dr. Glenn Turner(Cold Spring Harbor Laboratory) for advice on calcium imaging. This researchwas funded by National Institute of Mental Health Grant R01MH099231 (toC.C. and G.T.) and National Institute of Neurological Disorders andStroke Grant P01NS083514 (to C.C. and G.T.).

1. Cirelli C (2003) Searching for sleep mutants of Drosophila melanogaster. BioEssays25(10):940–949.

2. Shaw P (2003) Awakening to the behavioral analysis of sleep in Drosophila. J BiolRhythms 18(1):4–11.

3. Ho KS, Sehgal A (2005) Drosophila melanogaster: An insect model for fundamentalstudies of sleep. Methods Enzymol 393:772–793.

4. Ganguly-Fitzgerald I, Donlea J, Shaw PJ (2006) Waking experience affects sleep needin Drosophila. Science 313(5794):1775–1781.

5. Bushey D, Tononi G, Cirelli C (2011) Sleep and synaptic homeostasis: Structural evi-dence in Drosophila. Science 332(6037):1576–1581.

6. van Alphen B, Yap MH, Kirszenblat L, Kottler B, van Swinderen B (2013) A dynamicdeep sleep stage in Drosophila. J Neurosci 33(16):6917–6927.

7. Nitz DA, van Swinderen B, Tononi G, Greenspan RJ (2002) Electrophysiologicalcorrelates of rest and activity in Drosophila melanogaster. Curr Biol 12(22):1934–1940.

8. van Swinderen B, Nitz DA, Greenspan RJ (2004) Uncoupling of brain activity frommovement defines arousal states in Drosophila. Curr Biol 14(2):81–87.

9. Steriade M, McCarley RW (1990) Brainstem Control of Wakefulness and Sleep (PlenumPress, New York).

10. Buzsáki G, Anastassiou CA, Koch C (2012) The origin of extracellular fields and cur-rents: EEG, ECoG, LFP and spikes. Nat Rev Neurosci 13(6):407–420.

11. Zhong C, et al. (2014) Multi-unit recording with iridium oxide modified stereotrodesin Drosophila melanogaster. J Neurosci Methods 222:218–229.

12. Wilson RI, Turner GC, Laurent G (2004) Transformation of olfactory representations inthe Drosophila antennal lobe. Science 303(5656):366–370.

13. Maimon G, Straw AD, Dickinson MH (2010) Active flight increases the gain of visualmotion processing in Drosophila. Nat Neurosci 13(3):393–399.

14. Chiappe ME, Jayaraman V (2012) Performing electrophysiology and two-photoncalcium imaging in the adult Drosophila central brain during walking behavior. Ge-netically Encoded Functional Indicators (Neuromethods), ed Martin J-R (Springer), Vol72, pp 83–101.

15. Vyazovskiy VV, et al. (2011) Local sleep in awake rats. Nature 472(7344):443–447.16. Heisenberg M (2003) Mushroom body memoir: From maps to models. Nat Rev Neu-

rosci 4(4):266–275.17. Berry J, Krause WC, Davis RL (2008) Olfactory memory traces in Drosophila. Prog Brain

Res 169:293–304.18. Pitman JL, McGill JJ, Keegan KP, Allada R (2006) A dynamic role for the mushroom

bodies in promoting sleep in Drosophila. Nature 441(7094):753–756.19. Joiner WJ, Crocker A, White BH, Sehgal A (2006) Sleep in Drosophila is regulated by

adult mushroom bodies. Nature 441(7094):757–760.20. Tian L, Hires SA, Looger LL (2012) Imaging neuronal activity with genetically encoded

calcium indicators. Cold Spring Harb Protoc 2012 Jun(6):647–656.21. Riemensperger T, Pech U, Dipt S, Fiala A (2012) Optical calcium imaging in the nervous

system of Drosophila melanogaster. Biochim Biophys Acta 1820(8):1169–1178.22. Akerboom J, et al. (2012) Optimization of a GCaMP calcium indicator for neural ac-

tivity imaging. J Neurosci 32(40):13819–13840.23. Huber R, et al. (2004) Sleep homeostasis in Drosophila melanogaster. Sleep 27(4):

628–639.24. Evans G (2005) Infrared radiation sensors of Melanophila acuminata (Coleoptera:

Buprestidae): A thermopneumatic model. Ann Entomol Soc Am 98(5):738–746.25. Kwon Y, Shim HS, Wang X, Montell C (2008) Control of thermotactic behavior via

coupling of a TRP channel to a phospholipase C signaling cascade. Nat Neurosci 11(8):871–873.

26. Hamada FN, et al. (2008) An internal thermal sensor controlling temperature pref-erence in Drosophila. Nature 454(7201):217–220.

27. Mamiya A, Beshel J, Xu C, Zhong Y (2008) Neural representations of airflow in Dro-sophila mushroom body. PLoS ONE 3(12):e4063.

28. Scott K (2011) Out of thin air: Sensory detection of oxygen and carbon dioxide.Neuron 69(2):194–202.

29. Lin HH, Chu LA, Fu TF, Dickson BJ, Chiang AS (2013) Parallel neural pathways mediateCO2 avoidance responses in Drosophila. Science 340(6138):1338–1341.

30. Vermehren-Schmaedick A, Ainsley JA, Johnson WA, Davies SA, Morton DB (2010)Behavioral responses to hypoxia in Drosophila larvae are mediated by atypical solubleguanylyl cyclases. Genetics 186(1):183–196.

31. Vyazovskiy VV, et al. (2009) Cortical firing and sleep homeostasis. Neuron 63(6):865–878.

32. Wang Y, et al. (2004) Stereotyped odor-evoked activity in the mushroom body ofDrosophila revealed by green fluorescent protein-based Ca2+ imaging. J Neurosci24(29):6507–6514.

33. Honegger KS, Campbell RA, Turner GC (2011) Cellular-resolution population imagingreveals robust sparse coding in the Drosophila mushroom body. J Neurosci 31(33):11772–11785.

34. Li H, Li Y, Lei Z, Wang K, Guo A (2013) Transformation of odor selectivity from pro-jection neurons to single mushroom body neurons mapped with dual-color calciumimaging. Proc Natl Acad Sci USA 110(29):12084–12089.

35. Lin AC, Bygrave AM, de Calignon A, Lee T, Miesenböck G (2014) Sparse, decorrelatedodor coding in the mushroom body enhances learned odor discrimination. NatNeurosci 17(4):559–568.

36. Livingstone MS, Hubel DH (1981) Effects of sleep and arousal on the processing ofvisual information in the cat. Nature 291(5816):554–561.

37. Kaiser W, Steiner-Kaiser J (1983) Neuronal correlates of sleep, wakefulness andarousal in a diurnal insect. Nature 301(5902):707–709.

38. Vanini G, Lydic R, Baghdoyan HA (2012) GABA-to-ACh ratio in basal forebrain andcerebral cortex varies significantly during sleep. Sleep 35(10):1325–1334.

39. Jones BE (2005) From waking to sleeping: Neuronal and chemical substrates. TrendsPharmacol Sci 26(11):578–586.

40. Brown RE, Basheer R, McKenna JT, Strecker RE, McCarley RW (2012) Control of sleepand wakefulness. Physiol Rev 92(3):1087–1187.

41. Cayre M, Buckingham SD, Yagodin S, Sattelle DB (1999) Cultured insect mushroombody neurons express functional receptors for acetylcholine, GABA, glutamate, oc-topamine, and dopamine. J Neurophysiol 81(1):1–14.

42. Andretic R, van Swinderen B, Greenspan RJ (2005) Dopaminergic modulation ofarousal in Drosophila. Curr Biol 15(13):1165–1175.

43. Lebestky T, et al. (2009) Two different forms of arousal in Drosophila are oppositelyregulated by the dopamine D1 receptor ortholog DopR via distinct neural circuits.Neuron 64(4):522–536.

44. Crocker A, Sehgal A (2008) Octopamine regulates sleep in Drosophila through proteinkinase A-dependent mechanisms. J Neurosci 28(38):9377–9385.

45. Crocker A, Shahidullah M, Levitan IB, Sehgal A (2010) Identification of a neural circuitthat underlies the effects of octopamine on sleep:wake behavior. Neuron 65(5):670–681.

46. Tomchik SM, Davis RL (2009) Dynamics of learning-related cAMP signaling andstimulus integration in the Drosophila olfactory pathway. Neuron 64(4):510–521.

47. Schwaerzel M, et al. (2003) Dopamine and octopamine differentiate betweenaversive and appetitive olfactory memories in Drosophila. J Neurosci 23(33):10495–10502.

48. Riemensperger T, Völler T, Stock P, Buchner E, Fiala A (2005) Punishment predictionby dopaminergic neurons in Drosophila. Curr Biol 15(21):1953–1960.

49. Schroll C, et al. (2006) Light-induced activation of distinct modulatory neurons trig-gers appetitive or aversive learning in Drosophila larvae. Curr Biol 16(17):1741–1747.

50. Liu C, et al. (2012) A subset of dopamine neurons signals reward for odour memory inDrosophila. Nature 488(7412):512–516.

51. Agosto J, et al. (2008) Modulation of GABAA receptor desensitization uncouples sleeponset and maintenance in Drosophila. Nat Neurosci 11(3):354–359.

52. Parisky KM, et al. (2008) PDF cells are a GABA-responsive wake-promoting componentof the Drosophila sleep circuit. Neuron 60(4):672–682.

53. Lei Z, Chen K, Li H, Liu H, Guo A (2013) The GABA system regulates the sparse codingof odors in the mushroom bodies of Drosophila. Biochem Biophys Res Commun436(1):35–40.

54. Nobili L, et al. (2011) Dissociated wake-like and sleep-like electro-cortical activityduring sleep. Neuroimage 58(2):612–619.

55. Nir Y, et al. (2011) Regional slow waves and spindles in human sleep. Neuron 70(1):153–169.

56. Andrillon T, et al. (2011) Sleep spindles in humans: Insights from intracranial EEG andunit recordings. J Neurosci 31(49):17821–17834.

57. Rector DM, Topchiy IA, Carter KM, Rojas MJ (2005) Local functional state differencesbetween rat cortical columns. Brain Res 1047(1):45–55.

58. Hung CS, et al. (2013) Local experience-dependent changes in the wake EEG afterprolonged wakefulness. Sleep 36(1):59–72.

59. de Belle JS, Heisenberg M (1994) Associative odor learning in Drosophila abolished bychemical ablation of mushroom bodies. Science 263(5147):692–695.

60. Dubnau J, Grady L, Kitamoto T, Tully T (2001) Disruption of neurotransmission inDrosophila mushroom body blocks retrieval but not acquisition of memory. Nature411(6836):476–480.

61. Li X, Yu F, Guo A (2009) Sleep deprivation specifically impairs short-term olfactorymemory in Drosophila. Sleep 32(11):1417–1424.

4790 | www.pnas.org/cgi/doi/10.1073/pnas.1419603112 Bushey et al.