Modulation of GABAA receptor desensitization uncouples sleep onset and maintenance in Drosophila

Abstract

Many lines of evidence indicate that GABA and GABA(A) receptors make important contributions to human sleep regulation. Pharmacological manipulation of these receptors has differential effects on sleep onset and sleep maintenance insomnia. Here we show that sleep is regulated by GABA in Drosophila and that a mutant GABA(A) receptor, Rdl(A302S), specifically decreases sleep latency. The drug carbamazepine (CBZ) has the opposite effect on sleep; it increases sleep latency as well as decreasing sleep. Behavioral and physiological experiments indicated that Rdl(A302S) mutant flies are resistant to the effects of CBZ on sleep latency and that mutant RDL(A302S) channels are resistant to the effects of CBZ on desensitization, respectively. These results suggest that this biophysical property of the channel, specifically channel desensitization, underlies the regulation of sleep latency in flies. These experiments uncouple the regulation of sleep latency from that of sleep duration and suggest that the kinetics of GABA(A) receptor signaling dictate sleep latency.

Figure 1

GABAergic neurons control sleep in Drosophila. Sleep parameters are shown for flies expressing a hyperpolarizing potassium channel (UAS-Shaw) under the control of GAD-GAL4. Single transgene sibling controls (UAS only and GAL4 only) are also shown. (a) Conventional sleep plot showing sleep in a 12-h light:dark cycle. (b) Total sleep duration (24 h) and total sleep during the light and dark periods. (c) Maximum sleep-episode duration. (d) Sleep latency after lights off. (e) Mean wake duration for light and dark periods. Values and error bars for b-e indicate mean ± s.e.m. (* indicates P < 0.05, ** indicates P < 0.005 and *** indicates P < 0.0005 for the comparison of wild type and mutant by one-way ANOVA with post hoc test; n = 32 for GAL4 alone, n = 31 for UAS alone and n = 62 for GAD-GAL4;UAS-Shaw).

Figure 2

Rdl^A302S flies show decreased sleep latency and increased sleep-episode duration. Rdl^MDRR and Rdl^CB1 are independently isolated A302S point mutations in the Rdl gene. Wild-type sibling controls for each experimental genotype are shown. (a) Raster plots showing the sleep/wake pattern of 16 individual wild-type (+/+, left) and Rdl^MDRR (right) flies during a 24-h (12 h:12 h dark:light) period. White areas represent periods of wakefulness (periods of movement, or inactivity <5 min), whereas the black areas show sleep periods (inactive periods >5 min). (b) Quantification and comparison of total 24-h sleep between wild-type (n = 135) and Rdl^MDRR(n = 155, left) flies, wild-type (n = 51) and Rdl^MDRR/Rdl^CB1(n = 49, middle) flies, and wild-type (n = 277) and Rdl^MDRR/Rdl¹ (n = 50, right) flies. (c) Sleep latency after lights off for same genotypes. Values and error bars for b and c indicate mean ± s.e.m. (* indicates P < 0.005, ** P < 0.0005 for the comparison of wild type and mutant by two-tailed Student’s t-test).

Figure 3

CBZ markedly decreases fly sleep by increasing sleep latency and decreasing episode duration (a) Conventional sleep plot showing the effect of different concentrations of CBZ on the sleep pattern of wild-type (Canton-S) flies during the first day of drug treatment. Values indicate the average sleep per 30 min of groups of flies fed with either solvent (blue, 0 mg ml^-1, n = 27) or various doses of CBZ (orange, 0.2 mg ml^-1, n = 26; red, 0.4 mg ml^-1, n = 24; turquoise, 0.8 mg ml^-1, n = 28, 1.2 purple, mg ml^-1, n = 27). White and gray areas are the dark and light periods, respectively. Black arrow on the top part of the graph indicates CBZ application. (b-f) Effect of CBZ on total sleep (b), sleep latency after lights off (c), sleep-episode duration (d), number of sleep episodes (e) and locomotion during wake periods (f) on the first night of drug application. Values represent mean ± s.e.m. for all panels. For panels b-f, * indicates P < 0.05 compared with vehicle control using one-way ANOVA factor ‘CBZ concentration’ and Tukey-Kramer post hoc test.

Figure 4

Rdl^A302S flies are resistant to CBZ effects on sleep latency, but not to its effects on sleep-episode duration. (a) Conventional sleep plot showing the effect of different concentrations of CBZ on the sleep pattern of Rdl^A302S mutants during the first day of drug treatment. (b-d) CBZ dose-response curves comparing the different sleep parameters between wild-type (+/+) and Rdl^A302S flies during the fist night of drug treatment. The behavioral response to each CBZ concentration was divided by the average response of the respective control group (0 mg ml^-1) to normalize for genotype differences on the various sleep parameters. Significant differences in CBZ sensitivity were observed between wild type and Rdl^A302S in total sleep (0.4, 0.8 and 1.2 mg ml^-1; b) and sleep latency (0.8 and 1.2 mg ml^-1; c), but not mean sleep-episode duration (d).* indicates P < 0.05 by two-way ANOVA with ‘genotype’ and ‘CBZ concentration’ as factors. In order of increasing dose, n = 27, 26, 24, 28 and 27 for the A302S mutant Rdl^MDRR.

Figure 5

The Rdl^A302S mutation rescues sleep homeostasis. (a) Significant differences in CBZ sensitivity were observed between wild type and Rdl^A302S in the number of sleep episodes at 0.4, 0.8 and 1.2 mg ml^-1, showing that the Rdl^A302S mutant can respond homeostatically to sleep deprivation. * indicates P < 0.05 by two-way ANOVA with ‘genotype’ and ‘CBZ concentration’ as factors. (b,c) Sleep-episode bout length distribution in wild-type (b) and Rdl^A302S (c) flies with increasing amounts of CBZ.

Figure 6

CBZ specifically increases RDL desensitization and the A302S mutation prevents CBZ effects (a) Response to 90-s application of 100 μM GABA with variable doses of CBZ, recorded from oocytes expressing RDL held at -60 mV under voltage clamp. (b) Comparison of current amplitudes with (1 mM CBZ) and without (control) drug perfusion. Change in peak amplitude was not statistically significant (left, P > 0.9, paired t-test), whereas steady state amplitude was significantly decreased (P < 0.005, paired t-test). Steady state amplitude was calculated by normalizing the peak current amplitude to 1 and fitting to a single exponential equation. (c) Current evoked by successive pulses of 50 μM GABA with and without CBZ. (d) Response of oocytes expressing A302S mutant channel to 100 μM GABA under the same conditions as in a. Traces in c and d were normalized to the first peak amplitude.