Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Recent studies demonstrated that selective agonists of -aminobutyric acid (GABA)_A receptors share hypnotic properties. In the rat, systemic administration of muscimol (0.2-0.4 mg/kg) as well as 4,5,6,7-tetrahydroisoxazolo[4,5-c]pyridin-3-ol (THIP; 2-4 mg/kg) at the beginning of the light period dose dependently increases non-rapid eye movement (non-REM) sleep, which is particularly associated with a lengthening of non-REM episodes and augments slow wave activity (SWA) in the electroencephalogram (EEG) within non-REM sleep (Lancel et al., 1996; Lancel and Faulhaber, 1996). When given at the beginning of the dark period, coinciding with the daily activity phase in the rat, THIP exerts qualitatively similar but more pronounced effects on sleep (Lancel, 1997). In young, healthy human subjects, a single oral bedtime dose of THIP has been shown to increase sleep efficiency (i.e., percentage of time in bed spent asleep), to promote slow wave sleep (stages 3 and 4), and to enhance low-frequency activity while attenuating activity in the frequency range of sleep spindles within non-REM sleep (Faulhaber et al., 1997). Intriguingly, these findings indicate that GABA_A agonists are able to increase the maintenance as well as the intensity of non-REM sleep. Drugs with such actions may be beneficial in treatment of insomnia characterized by frequent nocturnal awakenings and/or shallow, nonrefreshing sleep.

The effects of muscimol and THIP on sleep differ substantially from those evoked by agonistic modulators of GABA_A receptors. For example, benzodiazepine hypnotics administered to rats generally shorten non-REM sleep latency, increase non-REM sleep time (specifically the substate pre-REM sleep), depress low-frequency activity, and augment spindling as well as higher frequency activity in the EEG within non-REM sleep, and inhibit REM sleep (for review, see Lancel, 1999). The naturally occurring neurosteroids pregnanolone and allopregnanolone, both potent agonistic modulators of GABA_A receptors, influence non-REM sleep in a similar fashion (Edgar et al., 1997; Lancel et al., 1997). In humans, benzodiazepines consistently increase sleep efficiency; promote non-REM sleep, which is related to a reduction in sleep onset latency and an increase in time in stage 2; decrease slow wave sleep as well as power in the lower frequencies in the EEG within non-REM sleep; facilitate spindling; and suppress REM sleep (for review, see Lancel, 1999). A further common characteristic of agonistic modulators of GABA_A receptors, especially of the short-acting ones, is the rapid development of tolerance to their hypnotic effect. For example, mice treated with the benzodiazepine temazepam or the synthetic neuroactive steroid minaxolone, which upon acute administration produce a reduction in locomotor activity caused by sedation, did not exhibit changes in locomotor behavior when tested after a 7-day chronic treatment (Marshall et al., 1997). Rats treated with the benzodiazepines lorazepam or triazolam appear to develop tolerance to the drug-induced reduction in locomotor activity and exploration already after 3 days of chronic dosing (File, 1981). Polysomnographic recordings in rats showed that the promotion of non-REM sleep evoked by triazolam and zaleplon, a novel hypnotic that binds to benzodiazepine receptors, completely disappears within 5 days of daily drug treatment (Depoortere et al., 1998; Crespi, 1999). Similarly, benzodiazepine hypnotics lose their sleep-promoting action in humans after a few days to weeks of chronic use (for review, see Ashton, 1994; Dingemanse, 1995).

Because THIP also acts at the GABA_A receptor and has a short elimination half-life of approximately 2 h (Schultz et al., 1981), we were interested in determining the tolerance potential of THIP to its effects on sleep. Therefore, we assessed sleep in a group of rats before, during, and after chronic administration of THIP and compared their sleep profiles with those of vehicle-treated animals. The substances were administered at the beginning of the dark period because sleep-wake behavior of rats during the dark period is often used as a physiological model for insomnia in humans and is more sensitive to the hypnotic effects of drugs.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

The experiment was approved by the local commission for animal welfare. While under deep anesthesia, 17 adult male Wistar rats (Charles River Laboratories, Sulzfeld, Germany), weighing 170 to 270 g, were implanted with EEG and electromyogram (EMG) electrodes as previously described (Lancel et al., 1996). The animals were housed individually in a ventilated, sound-attenuated Faraday room under a 12-h light/dark cycle (lights on from 8:30 AM) at an ambient temperature of 20-22°C, with free access to food and water. At least 3 weeks were allowed for recovery from surgery and 4 days to adapt to the recording conditions.

The experiment lasted 9 consecutive days, consisting of 2 baseline (BAS) days, 5 treatment days (T1 to T5), and 2 drug withdrawal days (W1 and W2). On each day, the animals received an i.p. injection given 5 min before dark onset. The rats of the placebo group (n = 8) received vehicle (pyrogen-free saline) during all days. The rats of the THIP group (n = 9) received vehicle during the BAS days, 3 mg/kg THIP (Biotrend Chemikalien GmbH, Köln, Germany) dissolved in pyrogen-free saline during each treatment day, and vehicle during both withdrawal days.

EEG and EMG recordings were made during the first 6 h after injection. The signals were amplified and filtered (EEG: high-pass 0.3 Hz and low-pass 29 Hz, 49 dB/octave; EMG: high-pass 16 Hz and low-pass 3000 Hz, 6 dB/octave). Both the EEG and the rectified and integrated EMG were digitized with a sampling rate of 64 Hz. The EEG signal was subjected to an on-line fast Fourier transform routine (cosine taper). A power spectrum was computed for 2-s windows in 0.5-Hz bins for the frequencies between 0.5 and 4.5 Hz and in 1-Hz bins for the frequencies between 5 and 25 Hz. Power spectra were averaged over 10-s epochs. An off-line program displayed the 10-s epochs of raw EEG and of the rectified and integrated EMG on screen for the manual scoring of the states wakefulness, non-REM, pre-REM, and REM sleep (for scoring criteria, see Neckelmann and Ursin, 1993).

For each recording period, the latency to non-REM and REM sleep (arbitrarily defined as the 20th epoch of non-REM and the 3rd epoch of REM sleep) and the number and average duration of the non-REM (pre-REM sleep included) and REM sleep episodes were determined. Furthermore, time in each vigilance state, average SWA (1-4 Hz), and average EEG power densities within non-REM sleep were computed for 6- and 2-h intervals. To analyze changes in the dynamics of SWA during the non-REM episodes, all non-REM episodes that were preceded by at least two 10-s epochs of wakefulness and that lasted at least six 10-s epochs were selected from the first 3 postinjection hours. Average SWA was computed for the last epoch of wakefulness and for the first five epochs of non-REM sleep. For normalization, EEG power densities were expressed as percentage of the average EEG power density in the same frequency range within non-REM sleep during the BAS condition and were then log transformed. The data of the 2 BAS days were averaged for each animal. Statistical analysis was performed with a multiple-factor ANOVA with repeated-measures design (Greenhouse Geisser correction), where group (placebo versus THIP) was a between-subjects factor and day (BAS, treatment, and withdrawal days) and, for the evaluation of intraepisodic dynamics of SWA, epoch (10-s epochs) were within-subjects factors. ANOVAs were followed by tests with contrasts where appropriate.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Influence of THIP on Time Spent in Each Vigilance State. For the states wakefulness and non-REM sleep, ANOVA revealed a significant effect of the factor group (F_1,15 = 4.7, P = .05 and F = 5.4, P = .04, respectively), whereas the effects of day and group by day were not significant. Although marked differences between the groups were only present during the treatment days, the THIP group generally exhibited less wakefulness and more non-REM sleep than the placebo group (Fig. 1, A and B). ANOVA showed no effects for pre-REM (Fig. 1C), whereas for REM sleep a significant day effect was found (F_7,105 = 3.5, P = .02), reflecting a slight group-independent decrease across the experimental days (Fig. 1D).

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Percentage of time spent in wakefulness (A), non-REM sleep (B), pre-REM sleep (C), and REM sleep (D) during the 6-h recording period of BAS, treatment days (T1 to T5), and drug withdrawal days (W1 and W2). Curves connect mean ± S.E. values (placebo group: n = 8 and THIP group: n = 9).

Influence of THIP on Sleep Architecture. Analysis of non-REM sleep latency revealed no significant effects (Fig. 2A). For the number of non-REM episodes, ANOVA showed a significant interaction between the factors group and day (F_7,105 = 3.3, P = .008). Post hoc testing revealed that the THIP group had a similar number of non-REM episodes as the placebo group on the BAS and withdrawal days, but fewer on all treatment days, significantly on T2 and T4 and tendentially on T3 (P = .09) and T5 (P = .07; Fig. 2B). For the duration of the non-REM episodes, ANOVA revealed a significant effect of group (F_1,15 = 11.7, P = .004), day (F_7,105 = 3.6, P = .007), and group by day (F_7,105 = 4.8, P = .001]. Compared with the placebo group, the THIP group had significantly longer non-REM episodes, which was limited to the treatment days (Fig. 2C). To examine whether the THIP-induced decrease in number and increase in duration of non-REM episodes varied across the treatment days, a separate ANOVA was applied to the data of the 5 treatment days. Significant effects of group (number of episodes: F_1,15 = 13.8, P = .002; duration of episodes: F = 23.4, P = .0002) but not day or group by day were found, which indicates that the effects of THIP on both variables did not decline across the treatment period. Analysis of REM sleep latency and REM episode duration revealed no significant effects (Fig. 2, D and F) but a significant day effect emerged for the number of REM episodes (F_7,105 = 2.7, P = .04). Irrespective of the group, the number of REM episodes was slightly reduced in the course of the experiment (Fig. 2E).

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Latency to non-REM sleep (A), number (B) and average duration of non-REM sleep episodes (C), latency to REM sleep (D), and the number (E) and average duration of REM sleep episodes (F) during the 6-h recording period of BAS, treatment days (T1 to T5), and drug withdrawal days (W1 and W2). Curves connect mean ± S.E. values (placebo group: n = 8 and THIP group: n = 9). Differences between the placebo and the THIP group are indicated by *P = .02 and **P = .003 (tests with contrasts).

Influence of THIP on SWA within Non-REM Sleep. Analysis of average SWA within non-REM sleep revealed a significant effect of group (F_1,15 = 7.4, P = .02), day (F_7,105 = 10.6, P < .0001), and group by day (F_7,105 = 8.1, P < .0001). Post hoc analysis showed that SWA did not differ between the groups during BAS and drug withdrawal days, but that SWA in the THIP group significantly exceeded that in the placebo group on all treatment days (see legend of Fig. 3 for the 6-h mean values and results of tests with contrasts). Pairwise comparisons per 2-h interval revealed that the THIP-evoked enhancements mainly occurred during the first 2-h interval (Fig. 3). A separate ANOVA performed on the SWA values obtained during the 5 treatment days yielded a significant effect of group (F_1,15 = 18.2, P = .0007) and day (F_4,60 = 4.4, P = .02), the latter reflecting a moderate, group-independent decrease in SWA across the treatment days. The interaction between the factors group and day was not significant, which indicates that the differences between the groups barely changed in the course of the treatment period. We also analyzed average SWA within the first 2 postinjection hours. Similar to the results of the 6-h values, a significant effect of group (F_1,15 = 36.8, P < .0001), day (F_7,105 = 13.3, P < .0001), and group by day (F_7,105 = 11.8, P < .0001) was found. Significant differences between the groups were limited to the treatment days. ANOVA of the 5 treatment days revealed a significant effect of group (F_1,15 = 54.2, P < .0001) but not day or group by day.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Average SWA within non-REM sleep over 2-h intervals of BAS, treatment days (T1 to T5), and drug withdrawal days (W1 and W2). Curves connect mean ± S.E. values (placebo group: n = 8 and THIP group: n = 9). For plotting purposes, the data are expressed as percentage of the average SWA within non-REM sleep during the 6-h BAS period. Differences between the placebo and the THIP group are indicated by *P = .05 and **P = .0004 (tests with contrasts). Six-hour mean values of SWA (expressed as percentage of average SWA during BAS) and results of tests with contrasts follow. T1: placebo 102.1 ± 8.8 and THIP 116.2 ± 10.9, P = .01; T2: placebo 97.9 ± 7.7 and THIP 112.7 ± 10.6, P = .005; T3: placebo 95.9 ± 5.1 and THIP 107.1 ± 10.6, P = .02; T4: placebo 93.9 ± 6.2 and THIP 110.1 ± 10.9, P = .002; T5: placebo 97.0 ± 5.4 and THIP 106.3 ± 7.8, P = .02; W1: placebo 97.0 ± 10.5 and THIP 90.0 ± 7.0, not significant; W2: placebo 96.5 ± 8.9 and THIP 92.2 ± 5.9, not significant.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Time course of SWA across the last 10-s epoch of wakefulness and the first five epochs of non-REM sleep in non-REM episodes selected from the first 3 postinjection hours of BAS, treatment (T1 to T5), and drug withdrawal days (W1 and W2). Curves connect mean ± S.E. values (placebo group: n = 8 and THIP group: n = 9). For plotting purposes, the data are expressed as percentage of the average SWA within non-REM sleep during the 6-h BAS period. Differences between the placebo and the THIP group are indicated by *P < .05 and **P < .01 (tests with contrasts run on normalized and log-transformed values).

View larger version (23K): [in this window] [in a new window]   Fig. 5.   SWA accumulated over all non-REM sleep epochs during the 6-h periods of BAS, treatment (T1 to T5), and drug withdrawal days (W1 and W2). Curves connect mean ± S.E. values (placebo group: n = 8 and THIP group: n = 9). For plotting purposes, the data are expressed as percentage of the accumulated SWA during BAS. Differences between the placebo and the THIP group are indicated by *P < .05 and **P < .003 (tests with contrasts run on normalized and log-transformed values).

Influence of THIP on EEG Power Densities within Non-REM Sleep. Analysis of the EEG power densities within non-REM sleep revealed a significant effect (P < .05) of group for the frequencies from 1.5 to 5 Hz, of day for all frequencies 8 Hz, and of group by day for the frequencies between 0.5 and 7 Hz and between 9 and 12 Hz. On all treatment days, the THIP group exhibited higher power in the frequencies <13 Hz, most prominently in the frequencies between 1.5 and 5 Hz (Fig. 6). These effects were mainly caused by large differences during the first 2-h interval (data not shown). Except for power in the 0.5-Hz frequency band during W1, no significant differences between the groups were found on the withdrawal days. A separate ANOVA performed on the EEG power densities obtained during the 5 treatment days revealed a significant effect of group for the frequencies 12 Hz and of day for the frequencies 4.5 Hz, whereas the interaction between group and day was not significant for any of the frequency bands.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   EEG power densities within non-REM sleep during the 6-h periods of the treatment (T1 to T5) and drug withdrawal days (W1 and W2).Curves connect mean ± S.E. values (placebo group: n = 8 and THIP group: n = 9). For plotting purposes, the data are expressed as a percentage of the corresponding BAS value. Dots at the bottom of the graphs denote frequency bands. Lines through the dots indicate frequencies for which significant differences between the placebo and the THIP group were found (P < .05, tests with contrasts run on normalized and log-transformed values).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The results of the present study show that the acute effects of the GABA_A agonist THIP on sleep in the rat, i.e., an increase in non-REM sleep time, a lengthening of the non-REM sleep episodes, an augmentation of average SWA within non-REM sleep, and an increase in the maximal levels of SWA attained in the course of the non-REM sleep episodes, are sustained during chronic dosing (daily administration during 5 days). The persistence of these effects may indicate that no rapid development of tolerance to the hypnotic action of THIP takes place. Moreover, the present data demonstrate that discontinuation of THIP treatment has no effect on sleep-wake behavior, suggesting that abrupt drug withdrawal is not associated with negative rebound effects.

Under BAS conditions, the animals of the placebo and the THIP group exhibited similar sleep patterns. In agreement with the literature on sleep in the rat during the first half of the dark period (Borbély and Neuhaus, 1979; Lancel and Kerkhof, 1989), our rats spent little time in non-REM and REM sleep, the sleep episodes that occurred were short, and average SWA within non-REM sleep was relatively low and did not fluctuate over the 6-h recording period. In the placebo group, the sleep parameters hardly changed across the experimental days, which underlines the stability of sleep-wake behavior. The only exceptions were a moderate decrease in REM sleep, due to a reduction in the number of REM episodes, and a slight decrease of low-frequency activity in the EEG within non-REM sleep, which occurred in both groups. The latter effect may be caused by the buildup of connective tissue around the EEG electrodes, which attenuates the amplitude of the EEG signals as a function of time.

The sleep effects resulting after the initial administration of THIP are in complete accordance with previous reports (Lancel and Faulhaber, 1996; Lancel, 1997). Compared with the placebo group rats, the THIP-treated rats spend more time in non-REM sleep (Fig. 1). This effect was not related to a shorter non-REM sleep latency or to a higher number of non-REM episodes, but rather to a prominent increase in the duration of non-REM episodes (Fig. 2). This finding supports the notion that THIP exerts minimal effects on the initiation of non-REM sleep, but significantly increases non-REM sleep continuity. Spectral analysis of the non-REM sleep-specific EEG showed that THIP significantly enhanced power in the frequencies <13 Hz, particularly in the delta bands (Fig. 6). The elevations in SWA were most pronounced during the first 2 h after injection and gradually subsided thereafter (Fig. 3). The enhancement of average SWA was not only due to the lengthening of the non-REM episodes but also to intraepisodic changes in the dynamics of SWA; after THIP administration, SWA attained higher levels in the course of the non-REM episodes (Fig. 4). Because awakening thresholds increase in parallel with SWA (Williams et al., 1964; Frederickson and Rechtschaffen, 1978; Grahnstedt and Ursin, 1980), this observation indicates that THIP increases the intensity of non-REM sleep. Due to the increases in both time in non-REM sleep and SWA within non-REM sleep, THIP markedly elevated SWA integrated over all non-REM epochs, on average by 40% (Fig. 5). Accumulated low-frequency power has been used as a parameter in various sleep deprivation experiments and appeared to constitute a reliable index of sleep homeostasis (Åkerstedt and Gillberg, 1986; Lancel and Kerkhof, 1989; Lancel et al., 1991). In contrast to non-REM sleep, THIP did not influence time in pre-REM or REM sleep (Fig. 1), or the temporal distribution of REM sleep (Fig. 2). This result extends the existing information, which suggests that THIP selectively affects non-REM sleep-related processes.

Interestingly, the acute effects of THIP on sleep closely match those evoked by prolonged wakefulness. Sleep deprivation in the rat is known to increase sleeptime, especially non-REM sleep, and to lengthen the duration of the sleep episodes. Furthermore, sleep deprivation consistently augments slow EEG components within non-REM sleep, elevates the maximal SWA levels attained within the non-REM episodes, and increases accumulated EEG power (Borbély and Neuhaus, 1979; Mistelberger et al., 1983; Borbély et al., 1984; Lancel and Kerkhof, 1989; Trachsel et al., 1989). Thus, except for the absence of a REM sleep-promoting effect, THIP seems to induce a sleep profile with all characteristics of recovery sleep after sleep loss.

The sleep effects of THIP observed after acute administration were evident on all successive treatment days. Compared with the placebo group rats, the THIP rats tended to exhibit less wakefulness and more non-REM sleep, whereas pre-REM and REM sleep remained similar on all treatment days (Fig. 1). Furthermore, the THIP rats had fewer but significantly longer non-REM episodes than the placebo-treated animals during each treatment period (Fig. 2). Moreover, all THIP-induced changes in the EEG within non-REM sleep [enhancement of low-frequency activity (Figs. 3 and 6) and elevation of the highest SWA levels reached within the non-REM episodes (Fig. 4)] as well as the increase in accumulated SWA (Fig. 5) persist over the treatment days. Statistical analysis yielded no evidence for a decline of any of these effects across the treatment days. These data demonstrate that THIP effectively increases non-REM sleep continuity and intensity during chronic dosing and suggest that, in contrast to benzodiazepine hypnotics, tolerance to its effects on sleep may not develop rapidly.

In humans, abrupt withdrawal of benzodiazepine hypnotics often produces rebound insomnia, i.e., a transient deterioration of sleep compared with pretreatment levels, reflected by a prolongation of sleep onset latency, an increase in intermittent wakefulness, and/or a decrease in total sleep time, which may lead to continued use (for review, see Lader, 1992; Ashton, 1994; Dingemanse, 1995). To elucidate whether discontinuation of THIP treatment is associated with sleep disturbances, we also assessed sleep in the rat during the first 2 THIP withdrawal days. On both days, the THIP and the placebo group exhibited practically identical sleep patterns: the latency to non-REM as well as REM sleep, the amount of time spent in each vigilance state, and the number and duration of the non-REM and REM sleep episodes did not differ between the two groups (Figs. 1 and 2). Nevertheless, low-frequency activity in the EEG within non-REM sleep in the THIP group was slightly below placebo levels during both withdrawal days (Figs. 3 and 6), which may indicate that THIP discontinuation is associated with a decrease in sleep intensity. However, the relatively low mean values of the THIP group are mainly due to the fact that the general, moderate decline in low-frequency activity across the experimental days was very prominent in two of the THIP-treated animals. The latter also explains why neither average SWA nor the power densities in the frequencies >0.5 Hz differed significantly between the two groups. Thus, these findings suggest that the abrupt withdrawal of THIP after repeated daily administration exerts no negative effects on sleep.

The present observations confirm that THIP is able to consolidate and intensify non-REM sleep, without interfering with REM sleep, and suggest that it has a relatively low tolerance and rebound potential. This compound thus shows potential for the treatment of sleep complaints with respect to annoyingly frequent or long-lasting nocturnal awakenings and/or too light, nonrefreshing sleep. The prevalence of chronic insomnia is disproportionally high in the elderly (for review, see Silva et al., 1996). Among other factors, this trend is attributed to age-related sleep changes. A wealth of evidence demonstrates that aging is associated with a decrease in sleep efficiency, which is related to an increase in the number and duration of nocturnal awakenings, and with a progressive decline in non-REM sleep intensity, as reflected by a decrease in slow wave sleep as well as by an attenuation of slow-frequency components in the EEG within non-REM sleep (for review, see Miles and Dement, 1980; Bliwise, 1993). Thus, it seems worthwhile to investigate whether insomniacs, particularly elderly subjects, can benefit from THIP medication.