A Paradoxical Kind of Sleep in Drosophila melanogaster

Abstract

The dynamic nature of sleep in many animals suggests distinct stages that serve different functions. Genetic sleep induction methods in animal models provide a powerful way to disambiguate these stages and functions, although behavioral methods alone are insufficient to accurately identify what kind of sleep is being engaged. In Drosophila, activation of the dorsal fan-shaped body (dFB) promotes sleep, but it remains unclear what kind of sleep this is, how the rest of the fly brain is behaving, or if any specific sleep functions are being achieved. Here, we developed a method to record calcium activity from thousands of neurons across a volume of the fly brain during spontaneous sleep and compared this to dFB-induced sleep. We found that spontaneous sleep typically transitions from an active "wake-like" stage to a less active stage. In contrast, optogenetic activation of the dFB promotes sustained wake-like levels of neural activity even though flies become unresponsive to mechanical stimuli. When we probed flies with salient visual stimuli, we found that the activity of visually responsive neurons in the central brain was blocked by transient dFB activation, confirming an acute disconnect from the external environment. Prolonged optogenetic dFB activation nevertheless achieved a key sleep function by correcting visual attention defects brought on by sleep deprivation. These results suggest that dFB activation promotes a distinct form of sleep in Drosophila, where brain activity appears similar to wakefulness, but responsiveness to external sensory stimuli is profoundly suppressed.

Keywords: REM sleep; brain; calcium imaging; dorsal fan-shaped body; optogenetics; sleep stages; two-photon microscopy; visual attention.

Figure 1:. Neural activity during wake and spontaneous sleep stages.

A) Top: microscopy setup (see Figure S1A,B). Bottom: temporally-matched behavior captured from fly legs and abdomen (red box). B) Brain imaging across 18 z-slices, over ~100μm depth from the top of the brain. C) Example GCaMP6f expression. Arrow: a single neuronal soma. D) Activity trace of single neuron indicated in C. Activity events are indicated (red dots, top), alongside the corresponding binary data (bottom; gray traces did not meet criteria. See Figure S1C). E) A collapsed mask from one fly of neurons found to be active (green) in C alongside all identified regions of interest (ROIs, gray). F) Left: adjacency matrix of pairwise correlations between all active neurons in E. Middle: significant correlations (yellow) were identified (See Figure S1D–G). Right: mean degree was calculated by averaging the number of significant correlations among all active neurons. G) Movement data from an example fly. Sleep (green bars) was determined by >5min immobility. H) Corresponding active neurons during the behavioral trace in G. I) Sleep epochs in the same sample fly, corresponding to panels G and H, with temporally segmented sleep bouts: early sleep (0–5 min), mid sleep (5 – 10 min), late sleep (< 10 min), and prior to wake (5 min to wake). J) Neural activity (% neurons active) during 3 sleep bouts indicated in (I), compared to wake and wake prior to sleep. K) Average neural activity (% active neurons ± sem) for successive 5 min epochs of wake and sleep. Blue datapoints indicates average (± sem) for the fly used as an example in G-J. L) Average connectivity (mean degree ± sem) for the data in K. M) Mean degree for the data in L combining all sleep epochs per fly, compared to wake epochs of similar duration. For K and L, statistical tests are one-way ANOVAs with Dunnett’s multiple comparison correction. ns = not significant, * = p < 0.05, *** = p < 0.001. For M, statistical tests is Wilcoxon signed-rank test, ** = p < 0.01. n=7 flies. See also Figure S1 and Video S1.

Figure 2:. Neural identity overlaps during wake and spontaneous sleep stages.

A) Collapsed masks of neurons active during wake, early sleep, and late sleep in neurons from Figure 1E. Active neurons met activity criteria in other epochs. Inactive neurons are ROIs that were identified through segmentation, but never met activity criteria. Arrow: an active neuron tracked across epochs. B) Example neural activity traces for neurons during a sleep bout. Dotted green line: activity threshold (colored dots; see Figure S2A,B). *, same neuron as in C. C) Example neural activity across three successive 5 min epochs of wakefulness. Dotted purple line: activity threshold (colored dots). D) Calculation of the average overlap in neural activity during wake (waking average). Neurons indicated are the same as in B and C. E) Overlaps in neural identities during wake and sleep stages in two example flies. The number of active neurons in each is indicated. % overlap = 100*((#stage 1 ∩ #stage 2) / #stage2). F) Overlap in neural identities compared to the waking average (% overlap ± sem). Flies from E are indicated with red dots. One-way ANOVAs with Dunnett’s multiple comparison correction. ns = not significant, * = p < 0.05, ** = p < 0.01, **** = p < 0.0001. Waking average, n = 9; all other data, n = 7. See also Figure S2.

Figure 3:. Short sleep bouts.

A) Short sleep (shaded, 1–5min) epochs (arrows, from same fly as in Figure 1I). B,C) For every 1–5min epoch identified (green), an immediately preceding ‘prior wake’ epoch of equal length (1–5min) was identified and used for further comparisons (purple). D) Neural activity across all active ROIs for the duration of the experiment in A, with all short sleep and corresponding prior wake epochs indicated. E) Neural activity (% neurons active ± sem) in short sleep (green) compared to wake prior to sleep (purple) and the waking average (lavender). F) Significantly more active neurons are shared between short sleep and early sleep, and between short sleep and prior sleep, than between short sleep and wake. E, Wilcoxon matched-pairs signed rank test, * = p < 0.05; F, one-way ANOVA with Bonferroni multiple comparison correction. ns = not significant, * = p < 0.05, ** = p < 0.01. n=7 flies. G) Average total sleep (± sem) for day and night in freely-walking flies, based on >5min immobility criterion. H) Average total 1–5min sleep (± sem) for day and night in the same flies as in G. I. Average behavioral responsiveness (% ± sem) to a vibration stimulus during day and night as a function of prior immobility duration, for different durations (green) compared to briefly quiescent flies (<10 sec, lavender). Flies are the same as in G and H. <1min* epoch includes <10 sec. For G and H, test = Wilcoxon signed-rank test; for I, test is One-way ANOVA with Dunnett’s multiple comparison correction. ns = not significant, *** = p < 0.001, **** = p < 0.0001.

Figure 4:. dFB-induced sleep resembles wakefulness.

A) R23E10:Gal4-UAS:GFP expression (green). Scalebar=100μm. B) Experimental sequence. C) R23E10:Gal4-UAS:CsChrimson; UAS:GCaMP6f imaging during optogenetic activation. Scalebar=100μm. D) Optogenetic activation (red) led to an increase in GCaMP6f fluorescence in dorsal fan-shaped body (dFB) soma (circled blue in C). E) Corresponding movement trace from fly in D. F) Behavioral responsiveness (% ± sem) of flies to air puffs, before, during (red), and after optogenetic activation of the dFB. G) % neurons active (± SD) in UAS:Chrimson / X ; Nsyb:LexA/+ ; LexOp:nlsGCaMP6f / R23E10:Gal4 flies did not change during optogenetic activation of the dFB (red) in ATR-fed flies, compared to baseline wake and recovery. H) Mean degree did not change during optogenetic activation of the dFB (red), compared to baseline wake and recovery. I) Example collapsed mask of neurons active during waking, dFB sleep, and recovery. J) Overlaps between these three neural groups in two example flies. Numbers indicate active neurons within each condition. K) Analysis of the overlapping neurons between conditions. Red dots indicate flies shown in J. F: 2-way ANOVA with Dunnett’s multiple comparison test. *=p<0.05, **=p<0.01, ***=p<0.001; n=11 flies. G, H, and K: one-way ANOVA with Kruskal-Wallis multiple comparison test, ns = non-significant; n = 9 flies. See also Figure S3 and Video S2.

Figure 5.. dFB-induced sleep corrects visual attention deficits following sleep deprivation.

A) Diagram of the visual attention paradigm. Flies walking central platform are presented with two opposing flickering dark bars (targets) and a moving cyan grating in the background (distractor). Inset: attention is measured by the angle between the fly’s trajectory (black arrow) and the closest target (target deviation). B) Protocol for behavioral testing (all experiments with w+; R23E10-Gal4/+>UAS-CsChrimson/+ flies). Attention (pre & post) indicate when flies were tested for visual attention. C) Target deviation (degrees ± sem) before (pre) and after (post) dFB-induced sleep, for all of the conditions outlined in B. Fly numbers are indicated, pooled across 2–3 experiments each. D) Example traces of walking paths of flies immediately following sleep deprivation (SD) compared to non sleep-deprived controls (top), and after subsequent sleep induction (SD +red light) compared to controls (SD – red light). Statistical tests for C is one-way ANOVA with Dunnett’s multiple comparison correction. * = p<0.05, ** = p<0.01, *** = p<0.001. See also Figure S4.

Figure 6.. Visual responses in the brain are lost during dFB sleep.

A) A 2-second ultraviolet (UV) stimulus (V1) was used to probe visual responses in the brain of w+ ; UAS:CsChrimson / X ; Nsyb:LexA/+ ; LexOp:nlsGCaMP6f / R23E10:Gal4 flies during wakefulness and during dFB sleep (V2). B) A collapsed mask of spontaneous baseline neurons (purple), dFB-responsive neurons (red), spontaneous dFB sleep neurons (green), and visually responsive neurons (blue) in an example fly. C) Top left: visually responsive neurons in an example fly. Top right: spontaneous activity in the same neurons during wake and dFB sleep (red box). Bottom left: data from the same fly for dFB responsive neurons presented with the visual stimulus. Bottom right: the same dFB responsive neurons during wake and dFB sleep (red box). Gray bar: timing of visual stimulus. D) Calcium activity traces of five neurons found to be responsive to the first visual stimulus during wake (V1, blue dot); purple, spontaneous waking activity; green, spontaneous activity during dFB activation; red, response directly driven by activation of the dFB. E) Left: % active neurons during wake (purple), dFB sleep (green/red) and following visual stimulation (cyan). Right: spontaneous activity of visual neurons during both wake and dFB sleep. F) Overlapping active neurons between V1 (blue), wake (purple) and dFB sleep (red/green) in two example flies. The number of active neurons per group is indicated. G) V1 neurons had a significantly higher overlap across dFB sleep-active neurons compared to spontaneous wake neurons. Red dots indicate the flies shown in F. H) Cartoon of dFB sleep effects on spontaneous and evoked activity. (6). Tests = one-way ANOVA with Friedman multiple comparison test. ns = non-significant, * = p<0.05, ** =p<0.01, **** =p<0.0001. n = 9 flies. Error bars in E are SD, error bars in G are sem. See also Figure S5 and S6.