Neuronal machinery of sleep homeostasis in Drosophila

Abstract

Sleep is under homeostatic control, but the mechanisms that sense sleep need and correct sleep deficits remain unknown. Here, we report that sleep-promoting neurons with projections to the dorsal fan-shaped body (FB) form the output arm of Drosophila's sleep homeostat. Homeostatic sleep control requires the Rho-GTPase-activating protein encoded by the crossveinless-c (cv-c) gene in order to transduce sleep pressure into increased electrical excitability of dorsal FB neurons. cv-c mutants exhibit decreased sleep time, diminished sleep rebound, and memory deficits comparable to those after sleep loss. Targeted ablation and rescue of Cv-c in sleep-control neurons of the dorsal FB impair and restore, respectively, normal sleep patterns. Sleep deprivation increases the excitability of dorsal FB neurons, but this homeostatic adjustment is disrupted in short-sleeping cv-c mutants. Sleep pressure thus shifts the input-output function of sleep-promoting neurons toward heightened activity by modulating ion channel function in a mechanism dependent on Cv-c.

Figure 1

Mutations in cv-c Reduce Sleep (A) The P element insertion site in the C5-GAL4 line (red triangle), as determined by splinkerette PCR, maps to an intronic region of the cv-c locus. The sites affected by mutation in the four cv-c alleles used in this study (black triangles for transposon insertions, vertical arrow for nonsense mutation) are indicated with respect to the exon-intron structure of three annotated cv-c transcripts. (B) cv-c^C524/cv-c^MB03717 mutant flies sleep less than cv-c^C524/+ or cv-c^MB03717/+ heterozygous controls during a 24-hr day (mean ± SEM, n = 48 flies per group). Repeated-measures ANOVA detected a significant time × genotype interaction (F_{[46, 3384]} = 4.22, p < 0.0001). One-way ANOVA detected a significant genotype effect on total sleep time per day (F_[2,141] = 70.03, p < 0.0001). (C) Flies carrying heteroallelic combinations of four independent cv-c mutant alleles (cv-c^DG20401, cv-c^MB01956, cv-c^MB03717, and cv-c^C524; heteroallelic genotypes shown in black or red) exhibit reduced sleep time in comparison to heterozygous controls (gray; mean ± SEM, n = 15–48 flies per group). One-way ANOVA detected a significant genotype effect (F_[9,180] = 24.97, p < 0.0001). Asterisks denote significant differences from heterozygous controls in pairwise post hoc comparisons. (D) cv-c^C524/cv-c^MB03717 mutants (red) exhibit reduced daytime sleep bout duration (top), reduced nighttime sleep bout duration (center), and prolonged latency to sleep onset after the fall of darkness (bottom) relative to cv-c^C524/+ or cv-c^MB03717/+ controls (gray; mean ± SEM, n = 47–48 flies per group). One-way ANOVA detected significant genotype effects for day-time sleep bout duration (F_[2,140] = 29.64, p < 0.0001), nighttime sleep bout duration (F_[2,141] = 8.00, p = 0.0005), and latency to sleep onset (F_[2,141] = 6.49, p = 0.002). Asterisks denote significant differences from heterozygous controls in pairwise post hoc comparisons. (E) cv-c^C524/cv-c^MB03717 mutants (red) show no evidence of hyperactivity, as measured by locomotor counts per waking minute, relative to cv-c^C524/+ and cv-c^MB03717/+ controls (gray; mean ± SEM, n = 48 flies per group). One-way ANOVA failed to detect a significant genotype effect (F_[2,141] = 2.89, p = 0.06). (F) cv-c^C524/cv-c^MB03717 mutants (red) show no evidence of altered arousal thresholds during sleep relative to cv-c^C524/+ or cv-c^MB03717/+ controls (gray; mean ± SEM, n = 5–6 trials per group). Two-way ANOVA detected a significant effect of stimulation intensity (F_[2,42] = 76.79, p < 0.0001) but failed to detect a significant genotype effect (F_[2,42] = 1.057, p = 0.36). See also Figure S1.

Figure 2

Mutations in cv-c Impair Homeostatic Sleep Regulation (A and B) cv-c^C524/cv-c^MB03717 mutants show no change in circadian locomotor rhythmicity during 7 days in constant darkness. One-way ANOVA failed to detect a significant genotype effect on circadian period (F_[2,78] = 2.618, p = 0.08). (C) After overnight sleep deprivation for 12 hr, cv-c^C524/cv-c^MB03717 mutants regain a significantly lower percentage of their lost sleep than heterozygous controls (mean ± SEM, n = 67–85 flies per group). Two-way repeated-measures ANOVA of sleep percentage regained detected a significant genotype × time interaction (F_[70,7875] = 5.68, p < 0.0001). Asterisks denote time points at which sleep percentage regained differs significantly between cv-c^C524/cv-c^MB03717 mutants and both heterozygous controls in pairwise post hoc comparisons. A single data point on the right (black) indicates the amount of sleep regained by Canton-S (CS) flies at the end of a comparable experiment (mean ± SEM, n = 242 flies). (D) Short-term memory is impaired in sleep-deprived CS flies (black, left) in comparison to rested siblings (white; mean ± SEM, n = 29–30 flies per group; two-tailed Student’s t test, p = 0.02). Rested cv-c^C524/cv-c^MB03717 mutants (red, right) exhibit similarly reduced short-term memory in comparison to heterozygous controls (gray, right). Overnight sleep deprivation decreases short-term memory in heterozygous controls (compare gray and black columns) but not in cv-c^C524/cv-c^MB03717 mutants (compare red and black columns) (mean ± SEM, n = 37–58 flies per group). Two-way ANOVA detected a significant genotype × sleep history interaction (F_[2,273] = 5.35, p = 0.0052). Asterisks denote significant differences in pairwise post hoc comparisons. See also Figure S2.

Figure 3

Cv-c Functions in the Dorsal Fan-Shaped Body to Regulate Sleep (A) C5-GAL4 drives transgene expression in the dorsal FB (arrow). (B) cv-c^C5-GAL4/cv-c^MB03717 mutants (gray) sleep less than a cv-c^C5-GAL4/cv-c^MB03717, UAS–cv-c rescue strain during a 24-hr day (mean ± SEM, n = 16 flies per group). Two-way repeated-measures ANOVA detected a significant genotype × time interaction (F_[23,690] = 1.86, p = 0.0087). (C) Total sleep time is decreased in cv-c^C5-GAL4/cv-c^MB03717 mutants (gray) in comparison to heterozygous controls (white) and restored to control levels in a cv-c^C5-GAL4/cv-c^MB03717, UAS–cv-c rescue strain (black) (mean ± SEM, n = 16–32 flies per group). One-way ANOVA detected a significant genotype effect (F_[3,75] = 20.61, p < 0.0001). The asterisk denotes a significant difference from both heterozygous controls in pairwise post hoc comparisons. (D) Expression of cv-c^RNAi under the control of C5-GAL4 (black) reduces sleep compared to parental controls (white; mean ± SEM, n = 14–44 flies per group). One-way ANOVA detected a significant genotype effect (F_[2,71] = 24.87, p < 0.0001). The asterisk denotes a significant difference from parental controls in pairwise post hoc comparisons. (E) 104y-GAL4 drives transgene expression in the dorsal FB (arrow). (F) Rescue of cv-c under the control of 104y-GAL4 (black) increases sleep in an otherwise mutant cv-c^C524/cv-c^MB03717 background (gray; mean ± SEM, n = 26–30 flies per group). Two-way repeated measures ANOVA detected a significant genotype × time interaction (F_[23,1242] = 4.60, p < 0.0001). (G) Rescue of cv-c under the control of 104y-GAL4 (black) increases sleep time from mutant (gray) to control levels (white; mean ± SEM, n = 23–48 flies per group). One-way ANOVA detected a significant genotype effect (F_[5,192] = 31.15, p < 0.0001). Asterisks denote significant differences from heterozygous controls in pairwise post hoc comparisons. (H) Expression of cv-c^RNAi under the control of 104y-GAL4 (black) reduces sleep in comparison to parental controls (white; mean ± SEM, n = 16–44 flies per group). One-way ANOVA detected a significant genotype effect (F_[2,73] = 21.09, p < 0.0001). The asterisk denotes a significant difference from both parental controls in pairwise post hoc comparisons. (I) 23E10-GAL4 drives transgene expression exclusively in the dorsal FB (arrow). (J) Rescue of cv-c under the control of 23E10-GAL4 (black) increases sleep in an otherwise mutant cv-c^C524/cv-c^MB03717 background (gray; mean ± SEM, n = 30–31 flies per group). Two-way repeated-measures ANOVA detected a significant genotype × time interaction (F_[23,1357] = 6.89, p < 0.0001). (K) Rescue of cv-c under the control of 23E10-GAL4 (black) increases sleep time from mutant (gray) to control levels (white; mean ± SEM, n = 30–48 flies per group). One-way ANOVA detected a significant genotype effect (F_[5,212] = 29.52, p < 0.0001). Asterisks denote significant differences from heterozygous controls in pairwise post hoc comparisons. (L) Expression of cv-c^RNAi under the control of 23E10-GAL4 (black) reduces sleep in comparison to parental controls (white; mean ± SEM, n = 22–44 flies per group). One-way ANOVA detected a significant genotype effect (F_[2,94] = 38.17, p < 0.0001). Thermally induced expression of cv-c^RNAi in UAS-cv-c^RNAi/+; 23E10/tub-GAL80^ts flies at 31°C reduces sleep in comparison to siblings maintained at 21°C (solid red bar); temperature shifts have no effect on parental controls (open red bars; mean ± SEM, n = 43–50 flies per group). One-way ANOVA detected a significant genotype effect (F_[2,138] = 5.06, p = 0.0076). Asterisks denote significant differences from both parental controls in pairwise post hoc comparisons.

Figure 4

Deficits in Homeostatic Sleep Control and Memory Reflect the Role of Cv-c in the Dorsal Fan-Shaped Body (A) Rescue of cv-c under the control of 23E10-GAL4 increases sleep rebound after 12 hr of sleep deprivation in an otherwise mutant background (mean ± SEM, n = 52–61 flies per group). Two-way repeated-measures ANOVA detected a significant genotype × time interaction (F_{[72, 6084]} = 3.84, p < 0.0001). Asterisks denote time points at which sleep percentage regained differs significantly between experimental flies and both parental controls in pairwise post hoc comparisons. A single data point on the right (black) indicates the amount of sleep regained by CS flies at the end of a comparable experiment (mean ± SEM, n = 242 flies). (B) Rescue of cv-c under the control of 23E10-GAL4 restores short-term memory in an otherwise mutant background (mean ± SEM, n = 35–91 flies per group). One-way ANOVA detected a significant genotype effect (F_[2,192] = 5.01, p = 0.0076). The asterisk denotes a significant difference from both parental controls in pairwise post hoc comparisons. (C) Expression of cv-c^RNAi under the control of 23E10-GAL4 impairs sleep rebound after overnight sleep deprivation in comparison to parental controls (mean ± SEM, n = 27–110 flies per group). Two-way repeated measures ANOVA detected a significant genotype × time interaction (F_[72, 6804] = 6.90, p < 0.0001). Asterisks denote time points at which sleep percentage regained differs significantly between experimental flies and both parental controls in pairwise post hoc comparisons. A single data point on the right (black) indicates the amount of sleep regained by CS flies at the end of a comparable experiment (mean ± SEM, n = 242 flies). (D) Expression of cv-c^RNAi under the control of 23E10-GAL4 impairs short-term memory in comparison to parental controls (mean ± SEM, n = 40 flies per group). One-way ANOVA detected a significant genotype effect (F_[2,117] = 6.13, p = 0.0029). The asterisk denotes a significant difference from both parental controls in pairwise post hoc comparisons. See also Figure S3.

Figure 5

Sleep-Control Neurons Are Morphologically Homogeneous but Electrophysiologically Diverse (A and B) Example morphologies of dorsal FB neurons expressing UAS–mCD8-GFP under 104y-GAL4 control in WT (A) or cv-c^C524/cv-c^MB03717 mutant (B) flies. Top panels show the innervation of the dorsal FB by the genetically labeled neuronal population; the corresponding bottom panels depict individual biocytin-filled neurons in the same individuals. The axons of the 104y-GAL4-positive neurons span the entire width of the dorsal stratum of the FB (100% ± 0.006% overlap, n = 6 cells). (C and D) Examples of membrane potential changes evoked in dorsal FB neurons by current steps.

Figure 6

cv-c Regulates the Membrane Properties of Sleep-Control Neurons (A) The input resistances (R_m, left) and membrane time constants (τ_m, right) of dorsal FB neurons differ in WT (gray) and cv-c^C524/cv-c^MB03717 mutant flies (red; mean ± SEM, n = 37–82 cells per group; two-tailed Student’s t test, p = 0.0026 for R_m and p < 0.0001 for τ_m). (B and C) Input resistances (B) and membrane time constants (C) of simultaneously recorded pairs of dorsal FB neurons. The larger value in each pair is by convention assigned to cell A, and all data points therefore lie on or below the dotted identity line. Pairs of two electrically excitable neurons are represented by solid symbols, and pairs of two electrically silent neurons are represented by open symbols. Semifilled symbols denote mixed pairs. Significant correlations exist between members of a pair: R² = 0.5493, p < 0.0001 (R_m) and R² = 0.3306, p < 0.0001 (τ_m). Note that R_m and τ_m tend to be elevated in electrically excitable neurons, and correlations tend to be weak in mixed pairs. (D) The input resistances (R_m, left) and membrane time constants (τ_m, right) of olfactory projection neurons do not differ significantly between WT (gray) and cv-c^C524/cv-c^MB03717 mutant flies (red; mean ± SEM, n = 19–47 cells per group; two-tailed Student’s t test, p = 0.9352 for R_m and p = 0.3205 for τ_m). (E and F) Input resistances (E) and membrane time constants (F) of simultaneously recorded pairs of olfactory projection neurons. The larger value in each pair is by convention assigned to cell A, and all data points therefore lie on or below the dotted identity line. There are no significant correlations between members of a pair. R² = 0.0357, p = 0.4119 (R_m) and R² = 0.1100, p = 0.1653 (τ_m).

Figure 7

Sleep History Alters the Excitability of Sleep-Control Neurons in a Mechanism Dependent on Cv-c (A) Sleep history alters the input resistance of dorsal FB neurons in WT (black), but not cv-c^C524/cv-c^MB03717 mutant, flies (red; mean ± SEM, n = 32–81 cells per group). Two-way ANOVA detected significant effects of sleep history (F_[2,243] = 4.592, p = 0.0110) and genotype (F_[1,243] = 37.93, p < 0.0001) as well as a significant sleep history × genotype interaction (F_[2,243] = 3.109, p = 0.0464). Asterisks denote significant differences in pairwise post hoc comparisons. (B) Sleep history alters the membrane time constant of dorsal FB neurons in WT (black), but not cv-c^C524/cv-c^MB03717 mutant, flies (red; mean ± SEM, n = 29–77 cells per group). Two-way ANOVA detected significant effects of sleep history (F_[2,231] = 6.856, p = 0.0013) and genotype (F_[1,231] = 38.96, p < 0.0001) as well as a significant sleep history × genotype interaction (F_[2,231] = 3.701, p = 0.0262). Asterisks denote significant differences in pairwise post hoc comparisons. (C) Cumulative distribution functions of the percentages of dorsal FB neurons reaching defined spike frequency thresholds (1–50 Hz; see the color code in D) as functions of injected current in WT flies. Left, control flies; center, sleep-deprived flies; right, sleep-deprived flies after 24 hr of recovery sleep. Shaded areas represent the regimes of spike train generation. (D) Superimposition of cumulative distribution functions of the percentages of dorsal FB neurons reaching defined spike frequency thresholds (1–50 Hz) in WT flies. Solid traces that are color coded according to threshold represent the sleep-deprived state, and dashed color-coded traces represent the state after 24 hr of recovery sleep. Dotted black traces indicate flies whose sleep histories have not been experimentally controlled. (E) Cumulative distribution functions of the percentages of dorsal FB neurons reaching defined spike frequency thresholds (1–50 Hz; see the color code in D) as functions of injected current in cv-c^C524/cv-c^MB03717 mutants. Left, control flies; center, sleep-deprived flies; right, sleep-deprived flies after 24 hr of recovery sleep. Shaded areas represent the regimes of spike train generation. See also Figure S4 and Table S1.