Clock-dependent Regulation of a Homeostatic Sleep Center Maintains Daytime Sleep and Evening Activity

Summary:

Homeostatic sleep centers promote sleep in response to prolonged wakefulness, but their contribution to circadian-regulated daily sleep is still unclear. Do neuronal circuits driving rebound sleep after extended wake also drive circadian-gated sleep, or does rebound sleep differ on a neurophysiological level from daily baseline sleep? We observed in Drosophila that 23E10+ neurons, which include a homeostatic sleep center, the dorsal fan-shaped body (dFSB) ^1–5, promote sleep in a time-of-day dependent manner⎯- the neurons play the strongest role in maintenance of daytime sleep, and this effect on the siesta maps to cholinergic neurons within the dFSB. We asked whether 23E10+ neurons interact with the circadian clock to regulate daily sleep and find their role in maintaining the daytime siesta is at least partially dependent on the period gene. Through in vivo imaging, we show that calcium levels in the dFSB display a circadian rhythm, with a peak coinciding with the daytime siesta. In the absence of period, the 24 hour rhythm is lost but a daytime increase in calcium activity is maintained. Loss of pigment dispersing factor (PDF) signaling causes premature downregulation of calcium activity in the dFSB, coinciding with the earlier truncation of the siesta in pdfr mutants and resulting in an earlier onset of night sleep. Silencing the dFSB is sufficient to rescue the timing of night sleep onset in pdfr mutants. These results indicate that the dFSB, a homeostatic sleep center, relies on the circadian clock to restrict sleep drive to specific times of day.

eTOC blurb:

Hsu et al. show in flies that the circadian clock regulates a homeostatic sleep center such that its activity peaks at midday, controlling siesta duration and the timing of evening sleep onset. This suggests differential regulation of sleep at different times of day.

Graphical Abstract

RESULTS

The dFSB influences sleep with time-of-day specificity

Although many studies have shown that activating 23E10+ neurons increases sleep^1,5–7, effects of these neurons on daily sleep and their regulation by time-of-day influences are under appreciated. We observed that activating dFSB-projecting neurons by expressing TRPA1, a heat-activated cation channel⁸, with 23E10-Gal4 increased nighttime sleep more dramatically than daytime sleep, although daytime sleep increase was significant when normalized to account for baseline differences (Figure 1A and S1A). Importantly, the evening locomotor activity peak, computed with the Harrisingh Index⁹, persisted in spite of the increased sleep. This effect was even more salient when driving TRPA1 with the weaker binary transcription system, LexA/LexAOp¹⁰. Both morning and evening activity peaks were preserved, as was the evening anticipation index (Figure 1B). These results suggest that the ability of 23E10+ neurons to increase sleep is time-of-day dependent, with sleep-promoting effects minimized during morning and evening.

Figure 1 – Thermogenetic activation of dFSB selectively upregulates nighttime sleep while leaving the evening anticipation peak of locomotor activity intact.

(A) Activation of the dFSB using 23E10-Gal4 to drive expression of TRPA1 does not significantly alter the evening anticipation peak of locomotor activity. Top: sleep in 30 min bins during the second baseline day, 24 hours of activation (as indicated by the red bar), and the first recovery day. Middle: locomotor activity in 30 min bins during the second baseline day, 24 hours of activation (as indicated by the red bar), and the first recovery day. Bottom: minutes of daytime sleep from ZT 0 to 12, nighttime sleep from ZT 12 to 24, and evening anticipation index (computed by the Harrisingh Index). (n=21–39). (B) Activation of the dFSB using 23E10-LexA, a weaker transcriptional activator, selectively increases sleep at night but not the day. Top: sleep in 30 min bins during the second baseline day, 24 hours of activation (as indicated by the red bar), and the first recovery day. Middle: locomotor activity in 30 min bins during the second baseline day, 24 hours of activation (as indicated by the red bar), and the first recovery day. Bottom: day time sleep from ZT 0 to 12, nighttime sleep from ZT 12 to 24, and evening anticipation index (computed by the Harrisingh Index) (n=40–43). For all panels, * = p<0.05, ** = p<0.01, *** = p<0.001 by Kruskal-Wallis with Dunn’s multiple comparisons between the experimental and UAS or LexAop control (represented by the black asterisks) and the experimental and the Gal4 or LexA control (represented by the blue asterisks).

Related to Figure S1.

To examine effects of 23E10+ neurons on daily sleep, we asked whether silencing these neurons would reduce sleep with time-of-day specificity. Thus, we used 23E10-Gal4 to drive expression of GtACR1, a chloride channel activated by green light¹¹. Prior to silencing, flies were entrained for three full days with 12 hours red light and 12 hours darkness. When 23E10+ neurons were silenced from ZT0 to 12, flies lost a significant amount of daytime sleep but not nighttime sleep, with daytime sleep loss observed only from ZT6 to 12 (Figure 2A). Interestingly, ZT6 to 12 sleep loss persisted even after flies returned to their original red light entrainment conditions. As with the activation results (Figure 1A), the evening anticipation index was not significantly different from both control genotypes until after three silencing days (on the first recovery day) (Figure S1B).

Figure 2 – Optogenetic silencing of the dFSB selectively inhibits sleep in the latter half of the day.

(A) Silencing 23E10+ neurons from ZT 0 to 12 for three consecutive days results in sleep loss temporally restricted to ZT 6 to 12 on all three days. Top: sleep in 30 min bins during the second baseline day, 3 days of silencing using a continuous green light from ZT0 to 12, and one recovery day. Bottom left: minutes of daytime sleep from ZT 0 to 6. Bottom right: minutes of daytime sleep from ZT 6 to 12 (n=63–69) (B) Silencing cholinergic dFSB neurons using 10s pulses of green light from ZT 0 to 12 results in sleep loss from ZT 6 to 12. Top: sleep in 30 min bins during the second baseline day, 3 days of silencing using a continuous green light from ZT0 to 12, and one recovery day. Bottom left: minutes of daytime sleep from ZT 0 to 6. Bottom right: minutes of daytime sleep from ZT 6 to 12 (n= 39–44) (C) Silencing glutamatergic neurons using 10s pulses of green light from ZT 0 to 12 does not induce sleep loss from ZT 6 to 12. Top: sleep in 30 min bins during the second baseline day, 3 days of silencing using a continuous green light from ZT0 to 12, and one recovery day. Bottom left: minutes of daytime sleep from ZT 0 to 6. Bottom right: minutes of daytime sleep from ZT 6 to 12 (n=40–44) For all panels, * = p<0.05, ** = p<0.01, *** = p<0.001 by Kruskal-Wallis with Dunn’s multiple comparisons between the experimental flies and UAS control (represented by the black asterisks) and the experimental flies and the Gal4 or split-Gal4 control (represented by the blue asterisks).

Related to Figures S1 and S2.

We next sought to distinguish if ZT6 to 12 sleep loss resulted from time-of-day specificity versus a prolonged, 12-hour silencing paradigm. Thus, we tested whether sleep loss would occur if we silenced 23E10+ neurons from ZT6 to 18, reasoning that with this protocol a 12-hour silencing paradigm would shift sleep loss to ZT12 to 18. We found no sleep loss from ZT12 to 18 (note that sleep in the experimental needs to be significantly different from both controls). However, there was also no sleep loss from ZT 6 to 12, suggesting that silencing from ZT0–6 was required (Figure S2A). In addition, 12h of silencing was required as silencing 23E10+ for 6 hours, even for four consecutive days, did not induce sleep loss, regardless of whether silencing occurred from ZT0 to 6 or ZT6 to 12 (Figure S1C and S1D). Finally, no significant sleep loss was observed during the first or the second 6-hour block when 23E10+ neurons were silenced from ZT12 to 24, although sleep significantly increased from ZT18 to 24 on the recovery day (Figure S2B).

Because recent studies suggested that some sleep promoting effects of 23E10+ cells stem from ‘bowtie’ neurons in the ventral nerve cord (VNC)^6,12, rather than the dFSB, we repeated our silencing experiments using teashirt-Gal80 to suppress GtACR1 expression in the VNC¹³. We observed no sleep loss when silencing 23E10+ cells in the presence of teashirt-Gal80 using our original paradigm of continuous green light (Figure S2C). However, given that previous studies have demonstrated that channelrhodopsins adapt to constant light⁸, we tested whether a pulsed stimulation paradigm (10s on, 10s off) would be more effective. When pulsed stimulation was applied to flies expressing GtACR1 in all 23E10+ neurons (including the VNC), sleep loss occurred earlier in the day (before ZT6) than was observed with continuous green light exposure (Figure S2D). Similarly, when we silenced brain-specific 23E10+ neurons with pulse stimulation, sleep loss was observed throughout the day (from ZT0 to 6 and from ZT6 to 12) (Figure S2E). However, significant baseline sleep differences from ZT0 to 6 between the experimental and the tshGal80; 23E10-Gal4>+ control may have contributed to differences in the earlier half of the day (Figure S2E). In contrast, silencing brain-specific 23E10+ neurons at night (from ZT12 to 24) only decreased sleep after two days of silencing (Figure S2F). This suggests that the dFSB alone can regulate sleep with time-of-day specificity, although effects are stronger when both dFSB and VNC populations are manipulated and dFSB-specific silencing may require a pulse protocol.

Recent studies have shown that silencing glutamatergic or cholinergic dFSB neurons separately reduces sleep more effectively than silencing both subsets simultaneously ⁵. Thus, we used a similar approach to silence neurotransmitter-specific subsets from ZT0 to 12. In contrast to when we used 23E10-Gal4 with or without teashirt-Gal80, driving GtACR1 in neurotransmitter-specific dFSB subsets decreased sleep even under our baseline entrainment protocol (12h dim red light) (Figure 2B and 2C). Under dim red light, with expression of GtACR1 in cholinergic neurons, sleep was reduced from ZT0 to 6, but not ZT6 to 12, which only showed sleep reduction when GtACR1 was further activated with bright green light. In glutamatergic dFSB neurons, GtACR1 expression caused a baseline sleep loss from ZT0 to 12, but green light channel activation actually increased sleep from ZT 6 to 12 to levels comparable to control flies. These data suggest that, unlike cholinergic neurons, glutamatergic dFSB neurons are not sleep-promoting in the latter half of the day.

In summary, these data indicate that while activating or silencing the dFSB can increase or decrease sleep, respectively, the effects depend on the time of day during which such manipulations are performed. In particular, these neurons play a role in maintaining siesta duration in the latter half of the day.

The molecular clock downregulates dFSB activity and regulates sleep during the latter half of the day

Based on its time-of-day specific effects on sleep, we sought to determine if neural activity in the dFSB, the subset of 23E10+ cells associated with homeostatic sleep drive^2,4,12, changed with time of day. We imaged calcium activity in vivo for 5 hours per fly, collecting data once every 5 minutes. Imaging start times were staggered such that sessions covered overlapping time blocks, allowing us to differentiate between effects of ZT time versus experiment duration.

We observed that dFSB signal increased during the first half of the day and decreased during the second half of the day (Figures 3A-C). Rhythmicity (computed by MetaCycle¹⁴) was significant with a phase of 6.17 hours (Figure S3B). Thus, peak dFSB calcium activity coincides with when the siesta is observed. The dFSB calcium oscillation also explains the complementary effects of our activation and silencing experiments (Figure 1 and 2) – dFSB silencing is less effective at night while activation is less effective during the day, as night and day are when calcium activity reaches its floor and peak, respectively.

Figure 3 – The molecular clock is dispensable for maintaining elevated dFSB activity and therefore sleep during the latter half of the day.

(A) Top left: schematic created with Bio Render (August 16, 2022). Flies were tethered to a custom chamber such that the posterior surface of their head was exposed towards the objective. The cuticle on the posterior surface was removed, and the fly was suspended over an air-supported ball underneath the objective. Right panel shows a representative example of the fluorescent signal from the dFSB. Dotted white line indicates the manually drawn ROI. (B) In control flies with an intact molecular clock, the dFSB increases calcium activity between ZT 0 to 6 but begins exhibiting a decrease in calcium activity between ZT 6 and ZT 12. Left panel: Representative examples of calcium activity, normalized by the mean of the first hour of data shown. Right panel: same data shown in the left panel, but normalized by overlapping timepoints. (C) Left panel: all data from control flies, normalized by the mean of the first hour of data. Middle panel: all data from control flies normalized by overlapping timepoints. Right panel: median of data from all control flies, normalized by overlapping timepoints (n=36). For accompanying statistics, refer to Figure S3B. (D) In a per0 background, calcium signal in the dFSB, as labeled by 23E10-Gal4>UAS-GCaMP7b, no longer oscillates, suggesting that a functional molecular clock is necessary for rhythms in calcium activity. Left panel: all data, normalized by the mean of the first hour of data (n=19). Middle panel: all data normalized by overlapping timepoints. Right panel: median data from all per0 flies, normalized by overlapping timepoints. (E) Optogenetic silencing of the dFSB in a per⁰ (null) background does not significantly reduce total minutes of sleep relative to per⁰ controls. Top left: sleep in 30 min bins during the second baseline day and first two days of silencing from ZT 0 to 12 (shown by the green bar). Bottom left: sleep on the third day of silencing from ZT0 to 12 (as indicated by the green bar) and during the first two days of recovery. Pink bar indicates the dim red entrainment light from ZT0 to 12. Top right: minutes of sleep from ZT 0 to 6. Middle right: minutes of sleep from ZT 6 to 12. Bottom right: minutes of sleep for ZT 12 to 24. Black asterisks represent significant differences between the experimental animals and the UAS control. Blue asterisks represent significant differences between the experimental animals and the Gal4 control (n=25–29). * = p<0.05, ** = p<0.01, *** = p<0.001 by Kruskal-Wallis with Dunn’s multiple comparisons between the experimental and GAL4 control and the experimental and the UAS control.

Related to Figure S3.

Hypothesizing that the molecular clock may be required to establish the rhythm in dFSB calcium levels, we also imaged the dFSB in flies lacking a functional period gene (per⁰). Surprisingly, without a molecular clock, the dFSB increased activity during the day, but a decline in activity was not observed, leading to no significant rhythm (computed by MetaCycle, Figure 3D).

Given the lack of a calcium rhythm in the dFSB of per⁰ flies, we next asked if the time-of-day effect of the dFSB on sleep was lost in this background. To test this, we optogenetically silenced 23E10+ cells in a period null (per⁰) mutant¹⁵ (Figure 3E). Under weak (12hr dim red light) cues, per⁰ flies showed blunted day:night differences, and the siesta lasted the entire length of the day. Silencing 23E10+ neurons did not obviously reduce sleep in a per⁰ background. When sleep was normalized to account for significant baseline differences, per⁰ flies with silenced 23E10+ cells slept significantly less than per⁰ controls with functional 23E10+ signaling from ZT6 to 12 after two to three days of green light exposure (Figure S3A), but this was because of increased sleep in per⁰ controls. Thus, green light exposure increases sleep in per⁰ flies, and silencing 23E10+ cells prevents this light-mediated sleep increase, but without decreasing sleep below baseline. per⁰ flies expressing GtACR1 in 23E10+ cells also slept more at night than per⁰ controls, although these were baseline differences present even before green light exposure. These results suggest that the time-of-day specificity through which 23E10+ cells regulate daytime sleep requires a functional period gene, especially on the first day.

PDF signaling maintains dFSB activity and daytime sleep and prolongs latency to sleep at night

Because a molecular clock was necessary for downregulating dFSB activity from ZT0 to 6, we next tested whether subsets of clock neurons contributed to dFSB calcium rhythms. We first tested the role of Pigment Dispersing Factor (PDF), the major signaling molecule released by the ventral lateral neurons, LNvs, which are major regulators of circadian behavior^16–20. The small LNvs are responsible for the morning locomotor activity peak and are thereby termed morning cells. In flies with a loss-of-function mutation for the PDF receptor²¹, pdfr ^han5304, the dFSB calcium signal was downregulated earlier in the day, peaking at ZT4.2 instead of ZT6.2 (Figure 4A and S2B). We also compared the change in dFSB activity of control flies to that of pdfr ^han5304 mutants in the 2.5 hours following the peak. Although there was no significant difference between control and pdfr ^han5304 mutants from ZT5.75 to 6.75 (Figure S2C), dFSB activity in mutants significantly decreases between ZT7.75 and 8.75 (Figure 4B). This is consistent with behavior in pdfr ^han5304 mutants, which shows a truncated siesta leading to decreased sleep from ZT6 to 12 (Figure 4C). We also observed a phase advance in their evening locomotor activity peak (computed with the PHASE program²²), consistent with previous studies where PDF signaling was impaired^17,23. This caused earlier sleep onset (latency to fall asleep) following lights off, significantly increasing sleep from ZT12 to 18 (Figure 4C). As the earlier sleep onset could be related to the advance of dFSB calcium activity, we hypothesized that silencing the 23E10+ cells in this mutant background would delay activation of 23E10+ cells and cause a phase delay in, thereby rescuing, nighttime sleep onset. Silencing rescued (delayed) nighttime sleep onset and ZT12 to 18 sleep amount, although sleep from ZT6 to 12 and the timing of the evening locomotor activity peak were unaffected (Figure 4D). These results suggest that in wild-type animals, PDF signaling delays the offset of dFSB activity, both prolonging the siesta and shifting nighttime sleep onset until the appropriate time.

Figure 4 – PDF-mediated downregulation of the dFSB times the onset of sleep at night.

(A) In a null pdfr background, calcium signal in the dFSB, as labeled by 23E10-Gal4>UAS-GCaMP7b, shows an increased amplitude and remains elevated for a longer period of time than in control flies (n=29). Left: data for individual flies normalized by overlapping timepoints. Right: median data for all pdfr^han5304, normalized by overlapping timepoints. Statistics shown in Figure S4A. (B) Comparison of dFSB activity in the background of a null pdfr mutant (pdfr^han5304) to the dFSB activity of control flies when activity is normalized by the baseline value of the median signal from ZT 7.25 to 7.75. Left panel shows traces of calcium activity from individual flies following normalization by the baseline and transformation by log base 2. Grey vertical lines indicate the time over which F₀ and the area under the curve (AUC) were computed. Right panel shows the sum of the values over the range indicated by the “AUC” label between the grey vertical lines in the left panel. * = p<0.05 by Kruskal-Wallis Test with Dunn’s Multiple Comparisons. (C) pdfr^han5304 mutants (orange) showed a phase advance in their evening locomotor activity peak, resulting in an earlier end to the siesta and an earlier sleep onset (latency to fall asleep) following lights off relative to the control iso31 genetic background (n=52). Far left: average sleep in three days distributed in 30 min bins across 24 hours. Second left: average sleep across the three days during ZT 6 to 12. Third left: average sleep across the three days during ZT 12 to 18. Third right: latency to fall asleep following lights off. Second right: average activity in three days distributed in 30 min bins across 24 hours. Far right: peak phase of normalized activity. *** = p<0.001 by Mann Whitney test. (D) Optogenetic silencing of the dFSB in a pdfr^han5304 (null) background causes an increase in latency to fall asleep following lights off (green), relative to pdfr^han5304; + > UAS-GtACR1 control (orange) and pdfr^han5304; 23E10-Gal4 > + flies (brown). All flies shown are in the pdfr^han5304 mutant background. Top left: sleep in 30 min bins, distributed across 72 hours (three days). Green light (as indicated by the green rectangles) was applied to silence the dFSB from ZT 0 to 12 on all three days. Top middle: sleep from ZT 6 to 12 on the first day when dFSB was silenced from ZT 0 to 12. Top right: average sleep from ZT 12 to 18 on the first day during which the dFSB was silenced from ZT 0 to 12. Bottom left: activity in 30 min bins, distributed across 72 hours (three days). Green light (as indicated by the green rectangles) was applied to silence the dFSB from ZT 0 to 12 on all three days. Bottom middle: peak phase of normalized activity on the first day of silencing. Bottom right: latency to sleep following lights off is increased in flies that have their dFSB silenced. Orange asterisks represent significant differences between the experimental animals and the UAS control. Brown asterisks represent significant differences between the experimental animals and the Gal4 control (n=25–29). * = p<0.05, ** = p<0.01, *** = p<0.001 by Kruskal-Wallis with Dunn’s multiple comparisons between the experimental and GAL4 control and the experimental and the UAS control.

Related to Figure S4 and Data S1.

To confirm that the dFSB was responsible for delaying nighttime sleep onset when 23E10+ neurons were silenced in pdfr ^han5304 mutants, we coupled 23E10+ silencing in the pdfr ^han5304 mutant background with tshGal80. Lack of daytime sleep in one of our genetic controls in this experiment made the data difficult to interpret (Figure S4A), but connectome data support a link between PDF neurons and the dFSB. Path length measurements between different clusters revealed that small PDF+ LNvs are connected to dFSB neurons through fewer interneurons (as few as one) relative to their connections to the VNC sleep-promoting bowtie neurons (which involve at least four interneurons) (Figure S4B-E and Data S1). The connectome data also indicates that small LNvs are more closely connected to the dFSB than large LNvs. These results suggest that the effect of PDF on sleep behavior most likely occurs through the dFSB and is more likely to be mediated by small, rather than large, LNvs.

We also tested whether circadian function in evening cells contributed to the dFSB rhythm in calcium activity by using the MB122B-splitGAL4 driver²⁴ to drive expression of dominant negative CLOCK²⁵ in the evening cells. We observed an increased period but decreased amplitude (Figure S3B and S3D). These changes were not significant enough to disrupt baseline sleep (Figure S3E).

DISCUSSION

Here, we present evidence suggesting that the dFSB, previously implicated in rebound sleep^{1– 4,12}, has two roles in baseline sleep. First, downregulation of dFSB activity, which occurs only in the presence of a functional period gene, drives siesta offset. This finding may also explain results of the recent Andreani et al study that identified stronger rebound sleep during morning than evening²⁶, as inducing rebound sleep would be difficult if the circadian clock is downregulating the dFSB close to dusk. In contrast, the morning sleep increase observed in a sleep deprivation scenario likely reflects cumulative activation of homeostatic sleep neurons throughout the night. While our study and that of Andreani et al report that the effectiveness of silencing homeostatic sleep neurons varies depending on time of day, several manipulations of other sleep promoting populations show uniform day and night reductions in sleep when silenced^2,27–29. These differences suggest that some sleep promoting neurons, such as the dFSB that we describe here and the ellipsoid body R5 neurons described by Andreani et al, are more subject to circadian regulation than others.

The second role of the dFSB in baseline sleep is in nighttime sleep onset, which is driven by PDF maintaining cumulative diurnal excitation of the dFSB. When PDF signaling is functional, dFSB silencing does not alter nighttime sleep onset. This may be because PDF also acts through other pathways, such as evening cells or the ellipsoid body^30–33, that may play a role in nighttime sleep. Without PDF signaling, premature downregulation of the dFSB homeostat induces an earlier daytime siesta offset as well as an earlier nighttime sleep onset. This shift in sleep timing is along the lines of previous mammalian work showing that dysregulation of circadian-gated arousal centers such as orexinergic neurons leads to impaired wake consolidation in diseases such as narcolepsy^34,35.

The role of the dFSB versus the entire population of 23E10+ cells in sleep has been intensely scrutinized in recent years^2,6. Silencing all 23E10+ neurons was recently shown to prevent homeostatic sleep drive but not baseline sleep, while silencing only the cholinergic VNC subset of 23E10+ neurons caused strong sleep loss². However, we observed baseline changes in sleep, restricted primarily to the latter half of the day, when we silenced 23E10+ cells with GtACR1, and this effect was mapped to the dFSB using tshGal80. Subsequently, using a neurotransmitter-specific split-Gal4, we found that cholinergic, but not glutamatergic, neurons in the dFSB play a significant role in maintaining siesta duration. These results are consistent with the recently published study from Jones et al⁵ reporting stronger effects when activating cholinergic, rather than glutamatergic, neurons. However, the Jones et al study, unlike ours, found that silencing either glutamatergic or cholinergic dFSB neurons by driving a temperature-sensitive dynamin (shibire^ts) decreased sleep specifically at night⁵. The discrepancy in time-of-day effects between our studies may result from temperature shifts, known to promote daytime sleep, or may suggest a role of gap junctions, as electrical synapses are affected by membrane potential changes (such as expression of the GtACR1 chloride channel) but not by dynamin loss-of-function^18,36. We note too that our inhibition was restricted to specific times of days whereas Jones et al did it around the clock. Another interesting result of our study is that sleep loss occurs from ZT6–12 when glutamatergic dFSB neurons expressing GtACR1 are exposed to dim red light (baseline entrainment conditions) but not when they are exposed to bright green light. Given that green light has a stronger effect on the GtACR1 channel than red light^11,37, the different levels of GtACR1 activation induced by these wavelengths may differentially affect the rate and timing of neuronal firing, such as through tonic versus burst versus firing, and therefore sleep. Burst firing during slow wave activity was recently shown to encode sleep need in sleep-deprived flies⁴, and our behavioral data suggest that the role of different firing patterns in different phases of the circadian day during baseline (non-deprived) conditions may be a topic of future study.

Another question our study raises is how per⁰ flies avoid an indefinite increase in dFSB activity, given that the molecular clock is necessary for downregulating itI. We speculate that when a molecular clock is absent, calcium activity may reflect the presence of markers of homeostatic sleep need, such as mitochondrial reactive oxygen species³⁸, autophagosomes^39–41, or lipid species⁴². In this scenario, dFSB calcium may be restored to baseline levels following sleep and therefore dissipation of sleep need. Our inability to observe downregulated calcium activity may reflect the fly’s limited ability to experience restorative sleep in an immobilized imaging preparation.

Given that period mutants maintain high dFSB activity throughout the day, while pdfr^han5403 mutants show premature decrease in dFSB calcium signal, other clock factors likely also contribute to the regulation of the dFSB. However, these are not known at this time. While PDF prolongs the siesta via effects on dFSB activity, evening cell clock disruption has no effect on behavior and minimal effects on dFSB activity. However, functional heterogeneity has recently been characterized in evening cells⁴³. Activation of E1 neurons truncates the siesta, while flies lacking the molecular clock in E2 cells show delayed evening anticipation. The MB122B split-Gal4 our study used labels both E1 and E2 classes of evening cells; thus, targeting these classes individually may disassociate potential confounds of manipulating both simultaneously. Another candidate is the lateral posterior clock neurons (LPNs), which are presynaptic to dFSB-projecting neurons^44,45. A recent study demonstrated that ablating LPNs broadens the siesta, so these may contribute to cessation of the siesta.

Although we find that silencing the dFSB during the day delays evening sleep onset in pdfr ^han5403 mutants, the low nighttime activity of this sleep-promoting center raises the question of how daily, circadian-gated sleep (nighttime sleep in diurnal animals such as flies and humans) is maintained. One possibility is that homeostatic sleep pressure builds up during a day of wakefulness and the circadian gating manifests largely as a wake drive that counters the homeostat in the evening and thereby times sleep onset. On the other hand, other sleep promoting regions could mediate circadian-gated sleep. Several studies have established indirect but functional connections between circadian neurons and ellipsoid body ring neurons^26,46,47, and more recently, the mushroom body⁴⁸. Release of small neuropeptide F (sNPF) directly by circadian small ventrolateral neurons, which are also PDF positive, has also been shown to promote nighttime sleep⁴⁹. Thus, our work and that of others highlights how sleep at different times of day and under different conditions reflects a complex interaction of different sleep and wake-promoting regions and both circadian and non-circadian factors.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Amita Sehgal (amita@pennmedicine.upenn.edu)

Materials availability

This study did not generate new unique reagents.

Data and code availability

• All data reported in this paper will be shared by the lead contact upon request.

• All original code has been deposited at Zenodo and is publicly available at https://doi.org/10.5281/zenodo.15486096 as of the date of publication.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR METHODS

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Drosophila melanogaster

Flies were raised on a standard cornmeal-molasses diet, including 64.7g/L cornmeal, 27.1g/L dry yeast, 8g/L agar, 61.6mL/L molasses, 10.2mL/L 20% tegosept, and 2.5mL/L propionic acid. Flies used for thermogenetic experiments (involving TrpA1) were raised at 18–22 °C and transferred to an 18 °C entrainment incubator with 12:12 LD 1–2 days prior to eclosion. Flies used for optogenetics were raised at 25 °C in constant darkness. Flies with the genotypes pdfr^han5043 and MB122B>UAS-dnClk and their respective controls and flies used for imaging were raised at 25 °C in 12:12 LD.

The full genotypes of flies presented in this paper are as follows:

Figure 1A, S1A:

• UAS-dTrpA1 control: iso31/iso31; UAS-dTrpA1/iso31; iso31/iso31. UAS-dTrpA1(II) was originally a gift from Dr. Leslie Griffith and outcrossed to iso31 for seven generations by Dr. Daniel J. Cavanaugh⁵².

• 23E10-Gal4 control: iso31/iso31; iso31/iso31; 23E10-Gal4/iso31. 23E10-Gal4 was outcrossed to iso31 for at least five generations by Dr. Christine Dubowy.

• Experimental (23E10-Gal4 > UAS-dTrpA1): iso31/iso31; UAS-dTrpA1/iso31; 23E10-Gal4/iso31.

Figure 1B:

• LexAop-TrpA1 (2X) control: iso31/iso31; LexAOp-TrpA1/iso31; LexAOp-TrpA1/iso31 LexAop-TrpA1 (chromosome 2) was a gift from Dr. Gerald M. Rubin. LexAop-TrpA1 (chromosome 3) was a gift from Dr. Scott Waddell. The LexAop-TrpA1; LexAop-TrpA1 fly line was created and outcrossed to iso31 for at least five generations by Dr. Hirofumi Toda.

• 23E10-LexA control: 23E10-LexA was outcrossed to iso31 for at least five generations by Dr. Christine Dubowy.

• Experimental (23E10-LexA > LexAOp-TrpA1 (2X)): iso31/iso31; LexAOp-TrpA1/23E10-LexA; LexAOp-TrpA1/iso31.

Figure 2A, S1B-D, S2A-B, S2D:

• UAS-GtACR1 control: iso31/iso31; iso31/iso31; UAS-GtACR1/iso31. UAS-GtACR1 was backcrossed to iso31 for at least five generations by Dr. Hirofumi Toda.

• 23E10-Gal4 control: iso31/iso31; iso31/iso31; 23E10-Gal4/iso31. 23E10-Gal4 was outcrossed to iso31 for at least five generations by Dr. Christine Dubowy.

• Experimental (23E10-Gal4 > UAS-GtACR1): iso31/iso31; UAS-GtACR1/iso31; 23E10-Gal4/iso31.

Figure 2B:

• UAS-GtACR1 control: iso31/iso31; iso31/iso31; UAS-GtACR1/iso31. UAS-GtACR1 was backcrossed to iso31 for at least five generations by Dr. Hirofumi Toda.

• 84C10-AD; ChAT-DBD control: iso31/+; 84C10-AD/iso31; Chat-DBD/iso31. 84C10-AD; ChAT-DBD stock was provided Dr. Stephane Dissel. F1 progeny heterozygous for 84C10-AD and ChAT-DBD were generated by outcrossing to iso31 once.

• Experimental (84C10-AD; ChAT-DBD > UAS-GtACR1): iso31/+; UAS-GtACR1/84C10-AD; ChAT-DBD/iso31.

Figure 2C:

• UAS-GtACR1 control: iso31/iso31; iso31/iso31; UAS-GtACR1/iso31. UAS-GtACR1 was backcrossed to iso31 for at least five generations by Dr. Hirofumi Toda.

• Vglut-AD; 84C10-DBD control: iso31/+; Vglut-AD/iso31; 84C10-DBD/iso31.Vglut-AD; 84C10-DBD stock was provided Dr. Stephane Dissel. F1 progeny heterozygous for Vglut-AD and 84C10-DBD were generated by outcrossing to iso31 once.

• Experimental (Vglut-AD; 84C10-DBD > UAS-GtACR1): iso31/+; UAS-GtACR1/VGlut-AD; 84C10-DBD/iso31.

Figure S2E, F:

• UAS-GtACR1 control: iso31/iso31; iso31/iso31; UAS-GtACR1/iso31. UAS-GtACR1 was backcrossed to iso31 for at least five generations by Dr. Hirofumi Toda.

• teashirt-Gal80; 23E10-Gal4 control: +/+; teashirt-Gal80/+; 23E10-Gal4/+. Generated from a non-outcrossed teashirt-Gal80 fly and a 23E10-Gal4 fly that had been outcrossed to iso31 for at least five generations by Dr. Christine Dubowy.

• Experimental: +/+; teashirt-Gal80/+; 23E10-Gal4/UAS-GtACR1

Figure 3A-C and S3B: +/+; UAS-GCaMP7b/+; 23E10-Gal4/UAS-tdTomato. Generated from 23E10-Gal4/23E10-Gal4 parent genotype outcrossed to iso31. +/+; UAS-GCaMP7b; UAS-tdTomato genotype was generated from non-outcrossed fly lines.

Figure 3D: per⁰/per⁰; UAS-GCaMP7b/+; 23E10-Gal4/UAS-tdTomato. Generated from per⁰/per⁰; iso31/iso31; 23E10-Gal4/23E10-Gal4 parent genotype. per⁰/per⁰ genotype was outcrossed to iso31 by Dr. William Joiner. +/+; UAS-GCaMP7b; UAS-tdTomato genotype was generated from non-outcrossed fly lines.

Figure 3E and S3A:

• per⁰; 23E10-Gal4 control: per⁰/ per⁰; iso31/iso31; 23E10-Gal4/iso31.

• per⁰; UAS-GtACR1 control: per⁰/ per⁰; iso31/iso31; UAS-GtACR1 /iso31.

• Experimental (per⁰; 23E10-Gal4 > UAS-GtACR1): per⁰/ per⁰; iso31/iso31; UAS-GtACR1/23E10-Gal4

Figure S3D: +/+; MB122B-p65.AD, UAS-dnClk, 23E10-LexA/23E10LexA; MB122B-Gal4.DBD/LexAop-IVS-jGCaMP7b.

Figure S3E:

• UAS-dnClk control: iso31/iso31; UAS-dnClk/iso31; iso31/iso31.

• MB122B control: iso31/+; MB122B-p65.AD/iso31; MB122B-Gal4.DBD/iso31. Parental line was not outcrossed, but control flies used in the experiment were heterozygous for the transgenes and for the iso31 genetic background (crossed once for 50% isogeny).

• Experimental (MB122B > UAS-dnClk): iso31/+; MB122B-p65.AD/UAS-dnClk; MB122B-Gal4.DBD/iso31.

Figure 4A, B and S3B-C: pdfr^han5043/pdfr^han5043; UAS-GCaMP7b/+; 23E10-Gal4/UAS-tdTomato. Generated from pdfr^han5043/pdfr^han5043; iso31/iso31; 23E10-Gal4/23E10-Gal4 parent genotype. pdfr^han5043/pdfr^han5043 genotype was outcrossed to iso31 by Dr. Wenyu Luo. +/+; UAS-GCaMP7b; UAS-tdTomato genotype was generated from non-outcrossed fly lines.

Figure 4C: pdfr^han5043/pdfr^han5043; iso31/iso31; iso31/iso31.

Figure 4D:

• UAS-GtACR1 control: iso31/iso31; iso31/iso31; UAS-GtACR1/iso31. UAS-GtACR1 was backcrossed to iso31 for at least five generations by Dr. Hirofumi Toda.

• 23E10-Gal4 control: iso31/iso31; iso31/iso31; 23E10-Gal4/iso31. 23E10-Gal4 was outcrossed to iso31 for at least five generations by Dr. Christine Dubowy.

• Experimental (23E10-Gal4 > UAS-dTrpA1): iso31/iso31; UAS-dTrpA1/iso31; 23E10-Gal4/iso31.

Figure S4A:

• UAS-GtACR1 control: pdfr^han5043/iso31; iso31/iso31; UAS-GtACR1/iso31. UAS-GtACR1 was backcrossed to iso31 for at least five generations by Dr. Hirofumi Toda.

• pdfr^han5043; tshGal80; 23E10-Gal4 control: pdfr^han5043;/iso31; tshGal80/iso31; 23E10-Gal4/iso31. 23E10-Gal4 was outcrossed to iso31 for at least five generations by Dr. Christine Dubowy. tshGal80 was not outcrossed.

• Experimental (pdfr^han5043; tshGal80; 23E10-Gal4 > UAS-GtACR1): pdfr^han5043/+; tshGal80/+; 23E10-Gal4/UAS-GtACR1.

METHOD DETAILS

Sleep and locomotor activity

Sleep and activity was recorded using the Drosophila Activity Monitoring (DAM) system (TriKinetics, Waltham, MA). Female flies 2–6 days of age were loaded into glass tubes (5 mm diameter and 65 mm length) containing 5% sucrose and 2% agar (and in the case of optogenetic experiments with GtACR1, 1 mM of all-trans-retinal). Yarn was inserted into the open end to prevent flies from escaping and rubber bands were used to stabilize the position of the tubes such that the infrared detection beams were centered over the middle of the locomotor tubes. DAMFileScan112 (TriKinetics, Waltham, MA) was used to convert monitor data into channel files (counts per minute per channel data). Analysis excluded data from the day flies were loaded and the subsequent full 24 hours following lights on (ZT 0); thus ZT0 on “Baseline Day 1” is 36 to 48 hours after the flies have been loaded. Channel files were converted into sleep data using custom software (found at https://github.com/cthsu86/damSleepConverter) written in MATLAB 2014a (Mathworks, Natick, MA) and described previously⁵³. Activity and phase were computed using the PHASE program with its default parameters (filter order 3, filter frame length 241 minutes, and minimum distance between peaks 180 minutes).

In cases where was a significant difference between the experimental fly and at least one of the two controls at baseline, values were reported as “Fraction Change”. To compute “Fraction Change”, the mean baseline value for all flies of a specific genotype was computed over the two days of baseline. The difference between the value for the individual fly and the mean baseline value for the genotype was computed; this difference was then divided by the mean value of the genotype.

Optogenetics

For optogenetic experiments, flies were loaded into glass locomotor tubes containing 5% sucrose, 2% agar, and 1 mM all trans-retinal (ATR). The end of the tube was covered in foil to prevent light from degrading the ATR.

Optogenetic sleep and activity assays were illuminated using LED strips fixed at intervals 8.75 cm apart to a 61 × 61 cm aluminum board. The board was positioned in the incubator on the shelf above the Drosophila Activity Monitors and positioned facing down on the shelf above the monitors. Upon loading, flies were entrained to 12 hours of dim red light (400–700 lux). During GtACR1 silencing experiments, flies were exposed to green light (4000–6000 lux) for the time period indicated.

In-vivo imaging

Female flies aged 4–12 days were anesthetized with ice, then inserted into a custom cut hole in a piece of foil mounted in a perfusion chamber. Their head and thorax were fixed with wax (melted to 60 degrees with a dental waxer) and UV-curable glue, with the head tilted forward to expose the posterior surface to the objective as much as possible. Flies were then mounted by head and thorax to a fixed position on a perfusion chamber with cuticle dissection to expose the dorsal brain (Murthy and Turner, 2013). Flies that were removed from the incubator during lights off (between ZT12 to 24) were dissected under dim amber light.

The preparation was mounted over an air-suspended ball (600 mL air/s) and imaged using a Bruker Nanosystems Ultima two-photon microscope in. Images were acquired using a Chameleon ULTRA I laser directed through a resonant scanning galvanometer with a 20X Olympus objective at a digital zoom of 2.5 to 4X. Images were acquired at a resolution of 256 × 256 pixels in Z-stacks slices 5 μm apart. Image stacks were acquired once every five minutes for six to seven hours; however, data acquired in the first hour of the experiment was discarded as an “acclimation period”. Fresh saline was continuously perfused over the brain from the time of dissection to the end of the experiment. Flies that were not responsive to tactile stimulus (removal and reintroduction to the ball) at the end of the six hours were not included in the analysis.

Path length analysis of PDF neurons and bowtie/ dorsal fan shaped neurons

The hemibrain ID of sleep-relevant neurons within the R23E10 driver was identified using the neuronbridger tool^54,55,56,57. A query for the R23E10 and R84C10 drivers was performed and only the common neurons between the drivers were used for further analysis. The hemibrain Neuron Type nomenclature was transformed into the full connectome Flywire Cell type based on Schlegel et al⁵⁸. Ascending bowtie neurons were identified through manual inspection of ascending neurons in Codex^58,59, resulting in the cell IDs 720575940618054981 and 720575940637957446 that are also annotated as “bowtie” neurons; further confirmation was done by querying the R23E10 driver with Neuronbridge selecting the slide code 20180921_64_A6 where the bowtie is identifiable. The hemibrain ID for the bowtie was determined using Color Depth Search resulting in the cell IDs 514272175 and 5813047188. The Cocoglancer tool was used to verify that the hemibrain bowtie neuron and the codex bowtie neuron overlap.

Using the network analysis tool from Codex^59–62, with the FlyWire Brain Dataset FAFB v783, andwith a minimum threshold of 5 synapses between cells, the path between PDF neurons and the selected dorsal fan-shaped body and bowtie neurons was calculated. Sankey plots were made in Python, representing the shortest path between PDF neurons and each subtype of dorsal fan-shaped body neurons.

QUANTIFICATION AND STATISTICAL ANALYSIS

Imaging Data Analysis

For each experiment, the image stack acquired at each 5 minute timepoint was converted into a Z projection prior to analysis. An ROI was manually drawn around the fan. The mean pixel intensity for this area was computed for each Z projection (resulting in a single value once every 5 minutes). Within each experiment, data was normalized by the formula:

$$F_{normalized}=\left(F_{raw}-F_{baseline}\right)/F_{baseline}$$

where $F_{baseline}$ represents the mean value during the first 30 minutes (6 image stacks) of the experiment (excluding the discarded first hour from the acclimation period). Data was then transformed using the log2 transform so that increases and decreases in signal would be represented with the same magnitude.

To normalize by overlapping timepoints, experiments were first sorted in order of the first ZT time point reported. The values of the first experiment $\left(F_{1_{raw}}\right)$ and second experiment $\left(F_{2_{raw}}\right)$ according to this order were then normalized as follows:

$$\begin{gathered} F_{1_{\text{\_}normalized}}=F_{1_{raw}}/\overline{F_{1_{raw}}\left(t_{overlap(1,2)}\right)} \\ {F}_{2_{\text{\_}normalized}}=F_{2_{raw}}\ast \overline{F_{1_{raw}}\left(t_{overlap(1,2)}\right)}÷\overline{F_{2_{raw}}\left(t_{overlap(1,2)}\right)} \\ {F}_{3_{\text{\_}normalized}}=F_{3_{raw}}\ast \overline{F_{2\text{\_}normalized}\left(t_{overlap(2,3)}\right)}÷\overline{F_{3_{raw}}\left(t_{overlap(2,3)}\right)} \\ ⋮ \\ {F}_{n_{\text{\_}normalized}}=F_{n_{raw}}\ast \overline{F_{(n-1)\text{\_}normalized}\left(t_{overlap(n-1,n)}\right)}÷\overline{F_{n_{\text{raw}}}\left(t_{overlap(n-1,n)}\right)} \end{gathered}$$

where $\overline{F_{1_{raw}}\left(t_{overlap(1,2)}\right)}$ represents the mean fluorescence values for the first experiment during the overlapping timepoints between experiments 1 and 2.

Significance of cycling and other parameters were evaluated using MetaCycle. For each experiment, the median normalized value of every 6 timepoints (representing 30 min of data) was entered in as a separate value.

To compute the area under the curve (AUC), experiments that began between ZT 5 and ZT 5.25 and ZT between ZT 7 and 7.25 were considered for Figures 4B and 4S3C, respectively. For the baseline $\left(F_0\right)$, the median value from ZT 5 to 5.25 or ZT7 to 7.25 was computed. The values were then normalized by dividing by the baseline fluorescence, after which a log2transform was applied. The AUC was computed by integrating the normalized and transformed values across the hour following the baseline period.

Statistical Analysis of Behavior

All significance tests of behavior were performed using GraphPad Prism 9 or 10. All data was assumed to be non-parametric. Statistical parameters are listed in the caption of each figure. For 30 minute binned profiles of activity and sleep, data is shown representing means and SEMs. For box and quartile plots, data is shown representing medians and interquartile ranges, with outliers defined as those whose distance from the median is at least 1.5 times the interquartile range.

Supplementary Material

1. Data S1. Identification of sleep-relevant dorsal fan-shaped body neurons in the Hemibrain Database. Related to Figure 4.

A) Identification of neurons labeled by 23E10-Gal4 in the Hemibrain Database.

B) Identification of neurons labeled by 84C10-Gal4 in the Hemibrain Database

C) Cross-reference of neurons labeled by both 84C10-Gal4 and 23E10-Gal4

D) Path length analysis between FB6 and s-LNvs.

E) Path length analysis between FB7 and s-LNvs.

F) Path length analysis between FB8 and s-LNvs.

G) Path length analysis between VNC bowtie neurons and s-LNvs.

H) Path length analysis between FB6 and l-LNvs.

I) Path length analysis between FB7 and l-LNvs.

J) Path length analysis between FB8 and l-LNvs.

K) Path length analysis between VNC bowtie neurons and l-LNvs.

L) Summary of information in Data S4D-K.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Bacterial and virus strains
Biological samples
Chemicals, peptides, and recombinant proteins
Agar, Drosophila Type	Fisher	NC1429200
Corn Meal	LabScientific	FLY-8010–20
Molasses	Genesee Scientific	62–118
Drosophila Dry Active Yeast	LabScientific	FLY-8040–10
All trans-retinal (ATR)	Sigma-Aldrich	R2500
Critical commercial assays
Deposited data
Experimental models: Cell lines
Experimental models: Organisms/strains
D. melanogaster: iso31 control strain	Bloomington Drosophila Stock Center (BDSC), maintained as lab stock	BDSC #5905
D. melanogaster: per 0	Bloomington Drosophila Stock Center (BDSC), maintained as lab stock outcrossed to iso31	BDSC #80928
D. melanogaster: pdfr han5403	Bloomington Drosophila Stock Center (BDSC), maintained as lab stock outcrossed to iso31	BDSC #33068
D. melanogaster: UAS-TrpA1 (II)	Gift from Dr. Leslie Griffith50, maintained as lab stock outcrossed to iso31	Flybase: FBtp0040248
D. melanogaster: LexAOp-TrpA1 (II)	Gift from Dr. Gerald M. Rubin10	N/A
D. melanogaster: LexAOp-TrpA1 (III)	Gift from Dr. Scott Waddell51	N/A
D. melanogaster: P{UAS-GtACR1.d.EYFP}attP2	Gift from Dr. Adam Claridge Chang11	BSDC #92983
D. melanogaster: teashirt-Gal80	Gift from Dr. Rebecca Chung-Hui Yang13	FBti0114123
D. melanogaster: UAS-dnClk	Gift from Paul Hardin25, maintained as lab stock outcrossed to iso31	http://flybase.org/reports/FBtp0051807
D. melanogaster: P{20XUAS-IVS-jGCaMP7b}su(Hw)attP5	Bloomington Drosophila Stock Center (BDSC)	BSDC #80907
D. melanogaster: P{10XUAS-IVS-myr::tdTomato}attP2	Bloomington Drosophila Stock Center (BDSC)	BSDC #32221
D. melanogaster: PBac{13XLexAop-IVS-jGCaMP7b}VK00005 (III)	Bloomington Drosophila Stock Center (BDSC)	BSDC #80915
D. melanogaster: R23E10-Gal4	Bloomington Drosophila Stock Center (BDSC)	BDSC #49032
D. melanogaster: R23E10-LexA	Bloomington Drosophila Stock Center (BDSC)	BDSC #52693
D. melanogaster: MB122B-Gal4	Gift from Drs. Heather Dionne, Aljoscha Nern and Gerald M. Rubin (Janelia Research Center, VA)24	N/A
D. melanogaster: 84C10-AD; ChAT-DBD	Dissel Lab5	N/A
D. melanogaster: Vglut-AD; 84C10-DBD	Dissel Lab5	N/A
Oligonucleotides
p65.AD F primer: GAG CTC GCC CGG GGA TC	IDT DNA	N/A
p65.AD R primer: GTC CAC TGG GGA ACA CCA TCG	IDT DNA	N/A
Gal4-DBD F primer: GGA GGT ACT AGT ATG AAG CTG CTG AG	IDT DNA	N/A
Gal4-DBD R primer: CGA TGG CGC GCC TTA CGA TAC CGT CAG TTG CCG T	IDT DNA	N/A
dnClk F primer: CAAGCTGCGCGTCTGAGTAG	IDT DNA	N/A
dnClk R primer: GACTGAGTATTGTCCCACCCC	IDT DNA	N/A
LexA F primer: GCCGGAGAATAGCGAGTTCA	IDT DNA	N/A
LexA R primer: TTTGTCCACTGGGGAACACC	IDT DNA	N/A
Recombinant DNA
Software and algorithms
GraphPad Prism 9 or 10	GraphPad	Download at graphpad.com
DAMSystem3, DAMFileScan	Trikinetics	Free download at trikinetics.com
Matlab	Mathworks	Download at mathworks.com
Matlab scripts for sleep quantification	Sehgal Lab52	Download at https://github.com/cthsu86/damSleepConverter
Matlab scripts for analysis of 2P data	Mathworks	Deposited on Zenodo (https://doi.org/10.5281/zenodo.15486096).
Phase	Shafer Lab22	Download at https://github.com/ajlopatkin/PHASE
Other
Drosophila Activity Monitor	Trikinetics	DAM2
ADJ High Quality Professional Color Filter Sheet – Amber	StageLightingStore.com	ADJ Z-Progel/SH-AM
Delrin® Acetal Resin Balls, 1/4” Diameter	Mcmaster-Carr	9614K24
LED Strip Lights – Red	superbrightleds.com	NFLS-R300X3-WHT-LC2
LED Strip Lights – Green	superbrightleds.com	NFLS-G300X3-WHT-LC2
LDK-8A 12~24 Volt DC Single Color LED Dimmer - Single Color LED Dimmer	superbrightleds.com	LDK-8A
Mean Well LED Switching Power Supply - LPV Series 20–100W Single Output LED Power Supply - 12V DC - 100 Watt (with Power Cord)	superbrightleds.com	LPV-100–12(POWERCORD)
24”x24” 6061 Aluminum Sheet, Unpolished (Mill) Finish, T6 Temper, Standard Tolerance, Inch, AMS QQ-A-250/11/ASTM B209	Amazon	B003JKJEP0
Bondic UV Glue Kit with Light, Super Glue, Plastic Welding Kit, Curing Light Liquid Plastic Repair, Jewelry Making, Necklace Craft, Glue Adhesive Epoxy Ultraviolet Mold for Glass Epoxy Resin Supplies	Amazon	B018IBEHQU
Electric Waxer Carving Knife Machine Double Pen and 6 Wax Tip Pot for Dental Lab	Amazon	B01DOEOMVG

Highlights:

• Cholinergic dorsal fan-shaped body (dFSB) neurons maintain late afternoon siesta

• Circadian clock dependent Ca2+ oscillations in the dFSB peak at ZT6.

• PDF signaling and the dFSB interact to time the siesta and evening sleep onset