Lowered insulin signalling ameliorates age-related sleep fragmentation in Drosophila

Abstract

Sleep fragmentation, particularly reduced and interrupted night sleep, impairs the quality of life of older people. Strikingly similar declines in sleep quality are seen during ageing in laboratory animals, including the fruit fly Drosophila. We investigated whether reduced activity of the nutrient- and stress-sensing insulin/insulin-like growth factor (IIS)/TOR signalling network, which ameliorates ageing in diverse organisms, could rescue the sleep fragmentation of ageing Drosophila. Lowered IIS/TOR network activity improved sleep quality, with increased night sleep and day activity and reduced sleep fragmentation. Reduced TOR activity, even when started for the first time late in life, improved sleep quality. The effects of reduced IIS/TOR network activity on day and night phenotypes were mediated through distinct mechanisms: Day activity was induced by adipokinetic hormone, dFOXO, and enhanced octopaminergic signalling. In contrast, night sleep duration and consolidation were dependent on reduced S6K and dopaminergic signalling. Our findings highlight the importance of different IIS/TOR components as potential therapeutic targets for pharmacological treatment of age-related sleep fragmentation in humans.

Figure 1. Reduced IIS affected activity and sleep and ameliorated age-related sleep fragmentation.

(A) Locomotor activity over 9 d of w^Dah control and dilp2-3,5 mutant flies under 12∶12 h LD and constant darkness 12∶12 h DD (n = 12, age 20 d). Mean free running period (τ) in DD ± s.e.m. (B) Average activity count data (30 min bins) under 12∶12 h LD conditions (25 d w^Dah n = 48, dilp2-3,5 n = 31). (C) dilp2-3,5 mutants were more active during the day and less active during the night compared to controls. (D) There was no significant difference in wakefulness (average activity per waking minute). (E) dilp2-3,5 mutants slept more at night and less during the day than controls. (F) Minutes of sleep per 30 min (25 d w^Dah n = 48, dilp2-3,5 n = 31). (G) Day and night sleep of dilp2-3,5 mutants were interrupted by fewer waking periods compared to controls. (H) dilp2-3,5 flies had longer sleep bouts during the night. (I) Longer sleep bouts were more prevalent in dilp2-3,5 mutants (age 25 d). (J) w^Dah control flies, but not dilp2-3,5 mutants, show a significant age-related increase in night sleep bouts (age 10 d, 25 d, 45 d, 55 d, and 65 d). (B–F) w^Dah , n = 31, 43, 46, 26, and 31 for ages 10 d, 25 d, 45 d, 55 d, and 65 d, respectively; dilp2-3,5, n = 31, 31, 32, 29, and 43 for ages 10 d, 25 d, 45 d, 55 d, and 65 d, respectively. Kruskal Wallis test with Dunn's multiple comparison (selected pairs). ***p<0.001, **p<0.01, and *p<0.05. Error bars represent s.e.m.

Figure 2. dfoxo affected daytime activity and sleep phenotypes of INR^DN flies.

Loss of dfoxo in da-Gal4/UAS-INR^DN flies but not in wild-type flies (age 20 d) decreased day activity but had no effect on night activity, had no significant effect on wakefulness (average activity per waking minute), increased day sleep duration but had no effect on night sleep duration, and reverted the low sleep bout phenotype of da-Gal4/UAS-INR^DN flies by day but not at night and increased night sleep bout duration. dfoxo indicates the dfoxo^Δ94allele (n = 35 for all genotypes). Kruskal Wallis test with Dunn's multiple comparison test (selected pairs). ***p<0.001, **p<0.01, and *p<0.05. Error bars represent s.e.m. Independent experiments verifying activity and sleep phenotypes of INR^DN;dfoxo double mutants are shown in Figure S3.

Figure 3. Day hyperactivity of IIS mutants is dependent on the AkhR.

(A) Loss of AkhR abrogated the day activity phenotype of dilp2-3,5 mutants (age 15 d, w^Dah n = 18, dilp2-3,5 n = 15, AkhR n = 18, AkhR dilp-3,5 n = 17). GLM was used to determine significance of genotype by genotype interactions in sleep and activity behaviours on loss of AkhR in controls and dilp2-3,5 mutants. Significant differences were seen in day activity (p = 0.0057) and day bout number (p = 0.044) but not in day sleep (p = 0.14) or night behaviours (activity p = 0.09, sleep p = 0.63, bout number p = 0.22, night bout length p = 0.067). Corresponding nighttime behaviours are shown in Figure S4A. (B) Two-day tolbutamide (1.35 mg/ml) treatment increased day activity of w^Dah flies. Lack of dfoxo, AkhR, or dilp2-3,5 blocked the tolbutamide effect on day activity (nighttime behaviour shown in Figure S4B) (age 15 d, w^Dah n = 51/47, AkhR n = 30/34, dfoxo^Δ94 n = 36/34, dilp2-3,5 n = 22/18, +/− tolbutamide). Analysis of genotype by treatment interactions (GLM) in sleep and activity behaviours on tolbutamide treatment in IIS, AkhR, and dfoxo mutants compared to controls showed day activity (p = 0.049), day sleep (p = <0.0001), and bout number (p = 0.008) were significantly different. However, no differences were seen in night behaviours (activity p = 0.58, sleep p = 0.89, bout number p = 0.28). (C) Mass spectrometry measurement of octopamine levels in head extracts (age 10 d, w^Dah n = 7, dilp2-3,5 n = 6, AkhR n = 6, AkhR,dilp2-3,5 n = 6). (A and B) Kruskal Wallis test with Dunn's multiple comparison test (selected pairs). (C) Mann–Whitney test. (C) One-way ANOVA with Bonferroni's multiple comparison test. ***p<0.001, **p<0.01, and *p<0.05. Error bars represent s.e.m.

Figure 4. Reduced IIS causes day hyperactivity through increased octopaminergic signalling.

(A) Two days feeding with mianserin hydrochloride (0.2 mg/ml) reverted the day activity phenotype of dilp2-3,5 mutants (age 10 d), but not night activity, sleep, sleep bouts, and sleep bouts length (w^Dah n = 17/24, dilp2-3,5 n = 17/27 +/− mianserin). GLM was used to determine significance of treatment by genotype interactions in sleep and activity behaviours on treatment with mianserin in controls and IIS mutants. Significant differences were seen in day activity (p = 0.0031), in day sleep (p = 0.0148), and day bout number (p = 0.002), but not in night behaviours (activity p = 0.31, sleep p = 0.49, bout number p = 0.72, night bout length p = 0.15). (B) Average activity count data (30 min bins) under 12∶12 h LD. (A) Kruskal Wallis test with Dunn's multiple comparison of selected pairs. ***p<0.001, **p<0.01, and *p<0.05. Error bars represent s.e.m.

Figure 5. Rapamycin rescued age-related night sleep fragmentation in an S6K-dependent manner.

(A) Rapamycin treatment (9 d) did not significantly affect day/night activity or wakefulness, but significantly increased night sleep duration, reduced night sleep fragmentation, and increased the length of night sleep periods (w^Dah, age 10 d, n = 21/20 control/rapamycin). (B) Average activity count data (30 min bins) under 12∶12 h LD. (C) Acute rapamycin treatment of 45-d-old flies for 3 d did not significantly affect day/night activity or wakefulness, but significantly increased night sleep duration, reduced night sleep fragmentation, and increased the length of night sleep periods (w^Dah, n = 64/64). Rapamycin-mediated night sleep, bout number, and bout length were independent of (D) 4E-BP (n = 19/19) and (E) reduced autophagy (a = da-Gal4/UAS-ATG5-RNAi, n = 20/17 or genetic controls c1 = da-Gal4/+, n = 20/21 and c2 = UAS-ATG5-RNAi/+, n = 23/19). Flies with reduced autophagy responded to rapamycin as controls in sleep (p = 0.81), bout number (p = 0.82), and night bout length (p = 0.42) (GLM). (F) Ubiquitous expression of constitutively active S6K blocked the rescue of night sleep fragmentation by rapamycin (c1 = da-Gal4/+, n = 20/21, and c2 = UAS-S6K^STDETE/+, n = 20/18, S6K = da-Gal4/UAS-S6K^STDETE, n = 20/17). Flies expressing a constitutively active form of S6K significantly differed from controls in the response to rapamycin (sleep p = 0.01, bout number p = 0.03, night bout length p = <0.0001, GLM). Kruskal Wallis test with Dunn's multiple comparisons of selected pairs. ***p<0.001, **p<0.01, and *p<0.05. Error bars represent s.e.m. Day behaviours of (D–F) are shown in Figure S6B–D.

Figure 6. Dopamine receptor mutants do not respond to rapamycin treatment.

(A) DopR1 mutants had similar activity and sleep features as rapamycin-fed flies. Behaviour of DopR1 mutants was not affected by rapamycin feeding (age 10 d, n = 64 for all genotypes). Flies were fed with rapamycin for 9 d. (B) dilp2-3,5, DopR1 mutants had similar activity and sleep features as dilp2-3,5 mutants (n = 64 for all genotypes). (C) QRT-PCR analysis of dopamine receptor (DopR1) expression normalized to Rpl32 expression and controls in head extracts of dilp2-3,5 mutants (age 10 d, n = 9) and da-Gal4/UAS/INR^DN flies and da-Gal4/INR^DN;dfoxo mutants (age 10 d, n = 3). (D) Mass spectrometry measurement of dopamine levels in head extracts of female flies (age 10 d, n = 3). (E) QRT-PCR analysis of DAT expression, normalized to Rpl32 expression (n = 9). (F) Behaviour of IIS mutants after short-term exposure (2 d) to the tyrosine hydroxylase inhibitor 3IY (5 mg/ml) (age 35 d, n = 32 for all genotypes). IIS mutants differed from controls in the nighttime activity and sleep response to 3IY treatment, but not in bout or bout length (activity p = 0.044, sleep p = 0.014, bout number p = 0.276, night bout length p = 0.463, GLM). (G) Behaviour of IIS mutants after short-term (12 h) exposure to METH (1 mg/ml) (age 25 d, n = 48 for all genotypes). IIS mutants differed from controls in daytime behaviours after METH treatment (activity p = 0.005, sleep p = <0.0001, bout number p = 0.0002, bout length p = 0.031, GLM) along with nighttime bouts (p = <0.0001) and bout length (p = <0.0001), whereas nighttime activity and sleep did not differ (activity p = 0.796, sleep p = 0.352, GLM). (A, B, G) Kruskal Wallis test with Dunn's multiple comparison of selected pairs. (F) Individual comparisons by Mann–Whitney U test. (C–E) Two-tailed t test. ***p<0.001, **p<0.01, and *p<0.05. Error bars represent s.e.m. (C, E, F) QRT-PCR analysis normalized to RNApolII expression shown in Figure S5D.