Modulation of rhythmic brain activity by diazepam: GABA_Areceptor subtype and state specificity

Abstract

The inhibitory neurotransmitter γ-aminobutyric acid (GABA) is involved in the generation of various brain rhythmic activities that can be modulated by benzodiazepines. Here, we assessed the contribution of α₂GABA type A (GABA_A) receptors to the effects of benzodiazepines on sleep and waking oscillatory patterns by combining pharmacological and genetic tools. The effects of diazepam on the electroencephalogram were compared between α₂(H101R) knock-in mice in which the α₂GABA_A receptor was rendered diazepam-insensitive, and their wild-type controls. The suppression of delta activity typically induced by diazepam in non-rapid eye movement (REM) sleep was significantly stronger in wild-type control mice than in α₂(H101R) mice. Moreover, electroencephalogram frequency activity above 16-18 Hz was enhanced in wild-type mice both in non-REM sleep and waking. This effect was absent in α₂(H101R) mice. Theta activity was enhanced after diazepam both in REM sleep and in waking in wild-type mice. In α₂(H101R) mice, this effect was markedly reduced in REM sleep whereas it persisted in waking. These findings suggest that α₂GABA_A receptors, which are expressed in hypothalamic and pontine nuclei and in the hippocampus, are localized in distinct neural circuits relevant for the modulation of rhythmic brain activities by benzodiazepines.

The main inhibitory neurotransmitter γ-aminobutyric acid (GABA) is involved in the generation of various rhythmic activities in the brain (1-3). By potentiating the GABAergic neurotransmission through an allosteric modulation of GABA type A (GABA_A) receptors, benzodiazepines and analogs modify sleep and waking electroencephalogram (EEG) patterns. In humans and rodents, these compounds typically reduce EEG delta activity in non-rapid eye movement (NREM) sleep (4-14). In addition, in humans, benzodiazepines enhance NREM sleep EEG power in the spindle frequency range (7). In rodents, EEG activity was increased in a broader band encompassing frequencies above the spindle range both in NREM sleep and in waking (9-11, 13). Recently, an increase of theta activity in rapid eye movement (REM) sleep and waking was reported in mice injected with diazepam (9, 13).

The mechanisms underlying the effects of benzodiazepines on the sleep EEG have not yet been identified. Benzodiazepines potentiate the GABA-induced chloride influx by increasing the affinity of GABA to its binding site on GABA_A receptors containing α₁, α₂, α₃ or α₅ subunits, hereafter called α₁-, α₂-, α₃-, and α₅GABA_A receptors (15). These receptor subtypes are differentially distributed in the brain (16, 17) and mediate selective psycho-pharmacological properties of benzodiazepines (18-24).

The α₁- and α₃GABA_A receptor subtypes are predominant in the corticothalamic network (16, 25), which is responsible for the generation of delta oscillations and spindle activity in NREM sleep (1, 26-29). Surprisingly, the typical suppression of delta activity in NREM sleep after diazepam was largely present in point-mutated mice carrying diazepam-insensitive α₁- or α₃GABA_A receptors (9, 13). These findings suggested that the corticothalamic system might not be the main target for the effects of diazepam on the NREM sleep EEG.

The prevalence of thalamic delta oscillations is reduced during REM sleep and waking due to the depolarization of the corticothalamic system by cholinergic, monoaminergic, and histaminergic projections from the midbrain, the pons, and the hypothalamus (30). During NREM sleep, these ascending projections are inhibited by GABAergic neurons present locally or projecting from the basal forebrain and the preoptic region (30-33). This inhibition leads to the hyperpolarization of the corticothalamic system, facilitating the generation of the NREM sleep EEG oscillations (1, 27, 30, 34).

Although the α₂GABA_A receptors are practically absent in the thalamus, they are expressed in the ascending system that modulates the generation of NREM sleep oscillations by the corticothalamic system, particularly in hypothalamic and pontine nuclei (16). α₂GABA_A receptors could therefore mediate the effects of diazepam on the NREM sleep EEG, especially the reduction of delta activity.

Moreover, α₂GABA_A receptors are abundant in the hippocampal formation (16) where theta oscillations in REM sleep and in exploratory behaviors are generated (35). Because GABAergic hippocampal interneurons and septo-hippocampal projections are involved in theta activity (36), we further hypothesized that α₂GABA_A receptors might mediate the enhancement of theta activity in REM sleep and waking previously described in mice treated with diazepam (9, 13).

In the present study, we determined the involvement of α₂GABA_A receptors in the effects of diazepam on the EEG by using a point-mutated mouse model in which the α₂GABA_A receptor was rendered diazepam-insensitive while its physiological function was preserved (20).

Materials and Methods

Animals. The experiments were approved by the government office in charge of animal research. Male mice homozygous for a histidine to arginine point mutation at position 101 of the GABA_A receptor α₂ subunit [α₂(H101R)] and homozygous wild-type controls [α₂(H101H), 129SvJ background, F₅] (20) were used. The mice were kept individually, with food and water available ad libitum, and maintained on a 12-h light-12-h dark cycle (light from 0700 to 1900 hours; 7-W OSRAM Dulux EL, 30 lux) at 22-24°C. Electrodes for recording the occipital and the frontal EEG and the electromyogram (EMG) were implanted under deep anesthesia (56 mg/kg pentobarbital sodium, i.p.) as described (37, 38). The mice were 11-13 weeks old and weighed 26-32 g before surgery.

Data Acquisition. The EEG and EMG signals were amplified (amplification factor ≈ 2,000), conditioned by analog filters (high-pass filter, -3 dB at 0.016 Hz; low-pass filter, -3 dB at 40 Hz, < -35 dB at 128 Hz), sampled with 256 Hz, digitally filtered [EEG, low-pass finite impulse response (FIR) filter 25 Hz; EMG, band-pass FIR filter 20-50 Hz], and stored with a resolution of 128 Hz. Consecutive 4-s epochs were subjected to a Fast Fourier Transform routine, and EEG power density was computed for 4-s epochs in the frequency range of 0.25-25.0 Hz (37). Three vigilance states were visually identified and scored for 4-s epochs based on the amplitudes of the EMG and occipital EEG: waking, high EMG concomitant with low EEG; NREM sleep, low EMG and high EEG; and REM sleep, low EMG and low EEG, as described (37). The vigilance states of all epochs could be identified. Epochs containing EEG artifacts were visually identified and excluded from spectral analysis (11.4 ± 0.7% of total recording time). Sleep onset was defined as the time elapsed between light onset and the first NREM sleep episode lasting at least 1 min and not interrupted by more than six 4-s epochs not scored as NREM sleep. Peak frequency in the theta band (6.0-11.0 Hz) was determined per 4-s epochs in REM sleep or in waking with a resolution of 0.25 Hz, as described (39).

Experimental Protocol. Diazepam (3 mg/kg; Hoffmann-LaRoche) and vehicle were injected in wild-type and α2(H101R) mice (n = 8 and n = 7, respectively) in a crossover design (3 days between treatments). Diazepam was dissolved in saline with 0.3% Tween 80 (vehicle) and injected in a volume of 10 ml/kg of body weight. Injections were performed at light onset, and continuous recordings were obtained throughout the 12-h light period.

Data Analysis and Statistics. The effects of diazepam within genotype were assessed by two-way ANOVAs for repeated measures with factors “condition” (vehicle and diazepam) and “4-h interval” (interval 1-3), and by two-tailed paired t tests. The effects of diazepam were expressed as the difference between diazepam and vehicle and compared between the genotypes by using two-way ANOVAs for repeated measures with factors “genotype” and “2-h interval” or “4-h interval” (interval 1-6 or 1-3) and two-tailed unpaired t tests.

Results

Diazepam reduced sleep latency and led to a minor reduction of REM sleep in both mutant and wild-type mice (Table 1; Fig. 1). Neither effect differed significantly between the genotypes. The lack of significance in the genotype effect on sleep latency (Table 1) is related to the large inter-individual variability that is also reflected in the low power (20%) of the t test to detect the genotype difference. An increase of NREM sleep at the cost of waking in the 12 h posttreatment was observed after diazepam in the mutants only (Table 1). Sleep continuity reflected in the number of brief awakenings (waking episodes lasting <16 s) in NREM sleep was not significantly modified by diazepam (two-tailed paired t test; 12-h mean values ± SEM: wild-type vehicle 58.9 ± 4.2, diazepam 53.4 ± 4.4; mutant vehicle 59.8 ± 3.2, diazepam 57.3 ± 4.3).

Table 1. Effect of diazepam on sleep.

	WT		MUT		Difference Dz - Veh
	Veh	Dz	Veh	Dz	WT	MUT
Sleep latency	33.1 (2.7)	24.6 (3.9)*	30.1 (2.5)	15.7 (2.8)*	−8.5 (3.3)	−14.4 (3.7)
Waking	30.5 (1.8)	30.2 (2.8)	38.3 (1.9)	31.0 (2.1)*	−3.3 (2.4)	−7.2 (1.4)†
NREM sleep	57.7 (1.3)	59.4 (2.5)	52.3 (1.9)	60.8 (1.8)*	1.6 (2.5)	8.4 (1.5)†
REM sleep	11.8 (0.6)	10.5 (0.6)*	9.4 (0.6)	8.2 (0.5)	−1.3 (0.3)	−1.2 (0.5)
REM sleep/TST	16.9 (0.6)	15.0 (0.7)	15.3 (0.1)	11.8 (0.1)*	−1.9 (0.1)	−3.4 (1.0)

Mean sleep latency (minutes), vigilance states (% of total recording time), and REM sleep per total sleep time (REM sleep/TST). Mean 12-h values (SEM) after vehicle (Veh; 0.3% Tween 80) and diazepam (Dz; 3 mg/kg i.p.) in wild-type mice (WT; n = 8) and α₂ (H101R) mutants (MUT; n = 7). *, P < 0.05, Dz vs. Veh within genotype, two-tailed paired t test. †, P < 0.05, MUT vs. WT, two-tailed t test.

Fig. 1.

Effect of diazepam (3 mg/kg i.p.) on waking, non-REM sleep (NREMS), REM sleep (REMS), and REMS per total sleep time (REMS/TST), expressed as a difference from the corresponding vehicle interval in α₂(H101R) mutants (MUT, n = 7) and wild-type controls (WT, n = 8). Mean values ± SEM for 2-h intervals. Two-way ANOVA for repeated measures with factors “genotype” (WT, MUT) and “2-h interval” (interval 1-6). Waking: genotype, F_1,13 = 5.74, P < 0.05; interaction genotype × 2-h interval, F_5,65 = 0.87, nonsignificant (n.s.). NREMS: genotype, F_1,13 = 4.96, P < 0.05; interaction genotype × 2-h interval, F_5,65 = 0.79, n.s. REMS: genotype, F_1,13 = 0.02, n.s.; interaction genotype × 2-h interval, F_5,65 = 0.87, n.s. REMS/TST: genotype, F_1,13 = 0.24, n.s.; interaction genotype × 2-h interval, F_5,65 = 0.587, n.s.

After vehicle, absolute EEG power density in each vigilance state did not differ between the genotypes in occipital and frontal derivations (not shown). Fig. 2 illustrates the effects of diazepam on the EEG in NREM sleep. In both derivations, EEG power in the delta band (0.75-4 Hz) was markedly reduced in wild-type mice. In the occipital derivation the power decrease was smaller in the mutants than in the wild-type controls over the entire delta band. In the frontal derivation, the reduction of delta power was restricted to 3 and 4 Hz in the mutant mice and was smaller than in the wild-type mice. A reduction of EEG power was also observed in frequencies above the delta range up to 9 Hz in wild-type mice and 14 Hz in the mutants. This effect did not differ significantly between the genotypes.

Fig. 2.

EEG power density of the occipital and frontal derivations in non-REM sleep over the 12-h recording after diazepam (Dz; 3.0 mg/kg i.p.) for α₂(H101R) mutants (MUT, n = 7) and wild-type mice (WT, n = 8). Means of relative power density of each frequency bin are expressed as a percentage of the corresponding bin in the 12-h period after vehicle treatment (Veh = 100%). Values are plotted at the upper limit of each bin. Horizontal lines above the abscissa indicate frequency bins that differed significantly from the corresponding bins after Veh within each genotype (P < 0.05, two-tailed paired t test). (Lower) Horizontal lines indicate frequency bins where the effect of Dz (difference Dz - Veh) differed between the genotypes (P < 0.05, two-tailed unpaired t test).

After diazepam, EEG power in NREM sleep above 16 or 18 Hz was enhanced in wild-type mice (Fig. 2). This effect was not specific for NREM sleep because a similar effect was evident in the waking EEG (Fig. 3). In contrast, in mutant mice high-frequency EEG activity in NREM sleep and in waking was not affected by diazepam (Figs. 2 and 3).

Fig. 3.

EEG power density of the occipital derivation in REM sleep and waking for three consecutive 4-h intervals (interval 1-3) after diazepam (Dz; 3.0 mg/kg i.p.) for α₂(H101R) mutants (MUT, n = 7), and wild-type mice (WT, n = 8). Means values of power density of each frequency bin are expressed as a percentage of the corresponding bin and interval after vehicle treatment (Veh = 100%). Horizontal lines below the abscissa indicate frequency bins that differed significantly between Dz and the corresponding bins and intervals after Veh within each genotype (P < 0.05, two-tailed paired t test after significance in a two-way ANOVA for repeated measures). (Lower) Horizontal lines indicate frequency bins where the effect of Dz (difference Dz - Veh) differed between the genotypes (P < 0.05, two-tailed t test after significance in a two-way ANOVA for repeated measures).

Diazepam induced changes also in REM sleep in the occipital EEG (Fig. 3). Thus, power was enhanced in the low theta band up to 7 Hz in wild-type mice. The increase was most prominent in the first 4-h posttreatment interval. This effect was remarkably reduced in the mutant mice. Neither genotype showed a consistent effect of diazepam in the frontal EEG (not shown). To examine whether the effect of diazepam on the amplitude of theta activity was accompanied by a change in the EEG theta peak frequency, the frequency with the maximum power value between 6 and 11 Hz was determined. The analysis showed that the theta peak in REM sleep was shifted to a lower frequency in wild-type mice whereas, in the mutants, it remained unchanged (Fig. 4). This effect differed significantly between the genotypes (Fig. 4).

Fig. 4.

(Left) Effects of diazepam (Dz; 3.0 mg/kg i.p.) on EEG theta peak frequency between 6 and 11 Hz in REM sleep and waking in α₂(H101R) mutants (MUT, n = 7) and wild-type controls (WT, n = 8). Mean values ± SEM. ^*, P < 0.05, Dz vs. vehicle (Veh) within genotype, two-tailed paired t test. (Right) Theta peak frequency after Dz expressed as a difference from Veh. ^*, P < 0.05, MUT vs. WT, two-tailed unpaired t test.

In the waking EEG, an increase of power occurred in the low theta range both in the mutant and in the wild-type mice (Fig. 3). In contrast to REM sleep, the effect did not differ significantly between the genotypes. Theta peak frequency during waking was reduced after diazepam in both genotypes, and the difference between diazepam and vehicle was significantly larger in the mutant mice than in the wild-type mice (Fig. 4).

Discussion

Our main findings confirmed the hypothesis that α₂GABA_A receptors are involved in the modulation of EEG patterns by diazepam particularly during sleep. In contrast to the effects on the sleep EEG, the hypnotic effect of diazepam reflected in the reduction of NREM sleep latency was present both in α₂GABA_A receptor point-mutated mice and in wild-type mice, indicating that it was mediated by receptors other than α₂GABA_A receptors. Whereas the reduction of sleep latency by diazepam is unlikely to depend on α₃GABA_A receptors (9), the role of α₁GABA_A receptors cannot be excluded (13). The contribution of α₅GABA_A receptors remains to be examined. The α₂GABA_A receptors might therefore be interesting targets for anxiolytic pharmacological agents (20) that could be free of hypnotic side effects.

The effect of diazepam on NREM sleep and waking (Table 1) was larger in mutant mice than in wild-type controls whereas no change in sleep continuity was observed. Because the genotype differences in the vigilance states were related to the vehicle values, these differences do not allow conclusions about diazepam. It is unlikely that a general difference in the level of motor activity between the genotypes contributed to the results because a previous study showed that motor activity and performance in various behavioral tasks were similar between mutant and wild-type mice treated with vehicle (20).

Delta activity in NREM sleep was suppressed by diazepam in wild-type mice, as reported in rodents for benzodiazepines (9-11, 13). Interestingly, in mutant mice, this effect was smaller or even absent in the frontal EEG up to 3 Hz. These results show a role of α₂GABA_A receptors in mediating the diazepam-induced suppression of delta EEG activity in NREM sleep. In the frontal derivation, diazepam reduced EEG power in frequencies above the delta band (up to 9 or 14 Hz). This effect encompassed a broader frequency range than in the occipital derivation, confirming previous results (9). In the mutant mice, the reduction of EEG power between 5 and 14 Hz did not differ significantly from the wild-type mice. Therefore, this effect is mediated by receptors other than α₂GABA_A receptors and seems to be independent from the decrease of delta activity.

These results imply that the α₂GABA_A receptor represents a molecular marker for distinct neuronal circuits (17) relevant for the modulation of NREM sleep delta activity by diazepam. The α₂GABA_A receptor subtype is expressed in the hypothalamic and pontine arousal systems (16, 17) that modulate the activity of the corticothalamic system (1, 27, 30, 33). Particularly, α₂GABA_A receptors are expressed in the tuberomammillary nucleus and the locus coeruleus (16), which enhance cortical activation through noradrenergic and histaminergic projections to the cortex and the thalamus during waking (30, 33). During NREM sleep these ascending arousal systems are inhibited by GABAergic inputs, especially coming from the ventrolateral preoptic nucleus (30).

The involvement of cortical α₂GABA_A receptors in the suppression of delta activity by diazepam cannot be excluded. This receptor subtype is particularly expressed in the outer cortical layers I-IV (16). The activation of α₂GABA_A receptors in the neocortex could influence the activity of cortical volleys important for thalamic delta oscillations (26, 40). Moreover, the action of diazepam on cortical α₂GABA_A receptors could affect the generation of cortical delta waves that are known to be independent from the thalamus (26).

The interactions between the different neural systems affected by benzodiazepines and the electrophysiological mechanisms leading to the suppression of delta activity remain to be determined. The absence of an effect of diazepam on the NREM sleep and waking EEG power in the high frequencies in the mutant mice suggests that similar mechanisms may account both for the reduced delta activity and for the elevated EEG activity in the high frequencies after diazepam.

Theta EEG activity in REM sleep was markedly enhanced by diazepam in wild-type mice, confirming previous results (9, 13). Such an effect was not visible in the frontal EEG derivation, suggesting that it reflected changes in the hippocampal theta rhythm, better detected over the parieto-occipital cortex than over the frontal one. The change of amplitude in EEG theta activity was accompanied by a lower theta peak frequency. A visual frequency estimation of the REM sleep EEG had also resulted in a reduction of theta activity frequency in rats treated with the benzodiazepine triazolam (41). The decrease or the absence of diazepam effects on the REM sleep EEG in the mutant mice suggests that they are mediated by α₂GABA_A receptors, which is consistent with the presence of this receptor subtype in the hippocampus (16, 36).

The role of α₂GABA_A receptors in the modulation of theta activity in REM sleep might reflect the importance of their synapse-specific distribution in hippocampal neurons (17). This hypothesis is reinforced by the fact that α₁GABA_A receptors are not involved in the enhancement of REM sleep theta activity despite their prominent expression in hippocampal cells (13, 16, 17). These two GABA_A receptor subtypes are differentially localized in synapses formed between the soma of hippocampal pyramidal neurons and two classes of interneurons characterized by different functional properties (42-44). In particular, somatic synapses with input from cholecystokinin- and vasoactive intestinal polypeptide-positive basket cells and axo-axonic synapses with an input from chandelier interneurons contain α₂GABA_A receptors whereas somatic synapses from parvalbumine-positive basket cells contain α₁GABA_A receptors (17). Interestingly, a recent study showed that the firing patterns of GABAergic hippocampal interneurons are predictive for their class and participate differentially in theta oscillations (45).

Theta frequency in REM sleep and in waking slows down as brain or body temperature decreases (46, 47). These temperature-related shifts of theta peak frequency were similar in both vigilance states (47). It is unlikely that the genotype difference in the effects of diazepam on theta activity reflects differences in body or brain temperature because the lack of peak frequency change in α₂GABA_A receptor mutant mice injected with diazepam was REM sleep-specific. In waking, theta peak frequency was lowered by diazepam in mutant and in wild-type mice, and EEG power in the low theta frequencies was enhanced to a similar extent in both genotypes. These results suggest that the α₂GABA_A receptors mediate the effects of diazepam on theta activity in REM sleep but not in waking. They provide further evidence that the theta rhythm in REM sleep and waking depend on different mechanisms (48, 49).

In conclusion, our findings suggest that benzodiazepines affect sleep and waking EEG patterns through their action on α₂GABA_A receptors, implying that these receptors are strategically expressed in distinct neural circuits relevant for these brain oscillatory activities (17). Interestingly, the results indicate that the molecular substrates responsible for the modulation of theta rhythm by benzodiazepines are vigilance state-dependent.