Abstract
Recent research in rats and humans has shown that exogenous progesterone evokes a sleep profile similar to that induced by agonistic modulators of γ-aminobutyric acidAreceptors, such as benzodiazepines. This finding suggests the involvement of the neuroactive metabolite of progesterone, allopregnanolone. In the vehicle-controlled study reported here, we assessed the sleep effects of two doses of allopregnanolone (7.5 and 15 mg/kg), mixed with oil, administered intraperitoneally at light onset in 8 rats. The electroencephalogram (EEG) and electromyogram were recorded during the first 6 postinjection hr. Compared with vehicle, both doses of allopregnanolone reduced the latency to non-rapid eye movement sleep (non-REMS) and 15 mg/kg allopregnanolone significantly increased the time spent in pre-REMS, an intermediate state between non-REMS and REMS. Furthermore, allopregnanolone dose-dependently influenced EEG activity during non-REMS and REMS. In non-REMS, EEG activity was decreased in the lower frequencies (≤7 Hz) and enhanced in the frequencies of ≥13 Hz. In REMS, allopregnanolone enhanced high-frequency EEG activity (≥17 Hz). The effects were most pronounced during the first postinjection hours and gradually diminished thereafter. Analysis of the plasma and brain concentrations of allopregnanolone in 45 rats revealed long-lasting increases, which reached maximal levels during the first postinjection hour. The sleep effects of allopregnanolone are very similar to those elicited by larger doses of progesterone, which produce comparable brain levels of allopregnanolone. These data indicate that the steroid allopregnanolone has benzodiazepine-like effects on sleep.
The hormone progesterone, which is present in both females and males (Corpéchot et al., 1993), is not only produced in peripheral steroidogenic organs but also synthesized de novoin the central nervous system (for a review, see Robel and Baulieu, 1994). Progesterone is known to exert multiple effects on brain functioning, including a rapid depression of neuronal excitability, as reflected by its anesthetic (Korneyev and Costa, 1996; Selye, 1942), anxiolytic (Bitran et al., 1993; Picazo and Fernández-Guasti, 1995), anticonvulsant (Landgren et al., 1987) and antinociceptive (Frey and Duncan, 1994) properties. Several lines of evidence indicate that the central-depressant action of progesterone is due not so much to the binding of progesterone to intracellular steroid receptors but rather is chiefly mediated by the action of its 5α-reduced metabolite 3α-hydroxy-5α-pregnan-20-one (allopregnanolone) at the membrane-bound GABAAreceptors. Both endogenous and exogenous increases in the level of progesterone result in rapid elevations of allopregnanolone concentrations in plasma and brain (Barbaccia et al., 1996;Korneyev and Costa, 1996; Lancel et al., 1996b; Paul and Purdy, 1992). Electrophysiological and biochemical experiments showed that allopregnanolone is a potent allosteric agonistic modulator of GABAA receptors (for a review, see Majewska, 1992), but it may also regulate gene expression via the progesterone receptor after intracellular oxidation (Rupprecht et al., 1993). Previous research revealed that the anxiolytic behavior evoked by progesterone is not affected by the progesterone receptor antagonist RU 38486 but is effectively prevented by picrotoxin, an antagonist of GABAA receptor-associated chloride channels (Bitran et al., 1995). Furthermore, 5α-reductase inhibitors, which block the conversion of progesterone to allopregnanolone, were found to decrease the anxiolytic and anesthetic action of P (Bitran et al., 1995; Korneyev and Costa, 1996). Finally, administration of allopregnanolone induces anesthesia (Korneyev and Costa, 1996; Mok et al., 1993) and anxiolytic behavior (Picazo and Fernández-Guasti, 1995; Wieland et al., 1995) and reduces seizure activity (Belelli et al., 1989; Landgren et al., 1987) and pain sensitivity (Frey and Duncan, 1994) at much lower doses than progesterone.
It has recently been shown in the rat that intraperitoneal administration of progesterone has dose-dependent hypnotic effects: it shortens sleep latency, decreases time spent in wakefulness and REMS and selectively promotes pre-REMS, also called intermediate-stage (Gandolfo et al., 1994) or transition-type sleep (Neckelmann and Ursin, 1993). Moreover, spectral analysis of the EEG signals between 0.5 and 25 Hz revealed that progesterone also affects the EEG activity during non-REMS and REMS in a dose-related manner. During non-REMS progesterone decreases low-frequency (≤7 Hz) EEG activity and enhances EEG activity in the spindle (∼11–16 Hz) and higher-frequency bands. During REMS, it shifts the dominant theta frequency from 8 Hz to 5 to 6 Hz and elicits a general increase in the higher frequencies (Lancel et al., 1996b). In agreement with these findings in the rat, a dose of 300 mg of micronized progesterone given orally to male subjects has been shown to reduce non-REMS latency, to promote stage 2 sleep, to slightly suppress slow-wave sleep and to decrease EEG activity in the lower frequencies and enhance activity in the frequencies >15 Hz during non-REMS (Friess et al., 1997). Two observations suggest that allopregnanolone is implicated in the influence of progesterone on sleep. First, in both studies cited, the temporal development of the sleep alterations induced by progesterone is highly correlated with the time course of the elevations in brain and/or plasma levels of allopregnanolone (Friess et al., 1997; Lancel et al., 1996b). Second, the changes in the time spent in each vigilance state as well as in sleep state-specific EEG activity after administration of progesterone are reminiscent of those evoked by benzodiazepine agonistic modulators of GABAA receptors (Borbély et al., 1985; Gandolfo et al., 1994; Lancel et al., 1996a). However, an earlier study in which two doses (5 and 10 mg/kg) of allopregnanolone were administered intraperitoneal to male rats failed to show significant effects on sleep latency and time spent in the sleep states (Mendelson et al., 1987).
In the present study, we investigated the influence of allopregnanolone on sleep by injecting male rats with two doses of allopregnanolone, which were mixed with oil to achieve a slower release from the injection site, and by assessing the effects both on sleep duration and on the sleep EEG. To compare the magnitude of the changes in allopregnanolone concentrations with those observed previously after the administration of progesterone, we determined the plasma and brain levels of allopregnanolone at several time points after the administration of vehicle and the two doses of allopregnanolone in a separate group of male rats.
Methods
Influence of allopregnanolone administration on sleep.
The experiments were approved by the local commission for animal welfare. Allopregnanolone (Sigma; Deisenhofen, Germany) was dissolved in 35% hydroxypropyl-β-cyclodextrin (1.5 ml/kg b.wt.), which was kindly provided by Besins Iscovesco Laboratories (Paris), and thereafter mixed with corn oil (3 ml/kg b.wt.).
While under deep anesthesia, 8 adult male Wistar rats (Charles River Laboratories; Sulzfeld, Germany) weighing 290 to 400 g were implanted with EEG and EMG electrodes as previously described (Lancelet al., 1996a). The animals were housed individually in a ventilated, sound-attenuated Faraday room under a 12-hr light/dark schedule (lights on from 8:30 a.m.) at a regulated ambient temperature of 20° to 21°C, with free access to food and water. After 2 weeks to recover from surgery and 4 days to adapt to the recording conditions, each rat was subjected to three randomized treatments, which were separated by ≥3 days. The treatments consisted of an intraperitoneal injection of vehicle (35% hydroxypropyl-β-cyclodextrin and corn oil) and 7.5 mg/kg and 15 mg/kg allopregnanolone at the beginning of the light period.
The EEG and EMG were continuously recorded during the 12-hr dark period preceding each treatment and during the first 6 postinjection hr. The signals were amplified and filtered (EEG: high-pass 0.3 Hz and low-pass 59.5 Hz, 49 dB/oct; EMG: high-pass 16 Hz and low-pass 3000 Hz, 6 dB/oct). Both the EEG and the rectified and integrated EMG were digitized with a sampling rate of 128 Hz. The EEG recordings were subjected to an on-line fast Fourier transform routine (cosine taper). A power spectrum was computed for 2-sec 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 40.5 Hz. Power spectra were averaged over 10-sec epochs. An off-line program displayed the 10-sec epochs of raw EEG and of the rectified and integrated EMG on screen for the manual scoring of the vigilance states wakefulness, non-REMS, pre-REMS and REMS (for scoring criteria, see Neckelmann and Ursin, 1993).
For each treatment, the latency to non-REMS and REMS (arbitrarily defined as the 20th epoch of non-REMS and the third epoch of REMS) and the number and average duration of the non-REMS and REMS episodes were determined. For each 2-hr interval, the time spent in each vigilance state and average EEG power densities during non-REMS and REMS were computed. Because the amplitude of the EEG signals decreases as a function of time, EEG power densities were normalized by expressing them as a percentage of the EEG power density in the same frequency band and sleep state during the entire preceding dark period and then log-transformed. Statistical analysis was performed with a one- or two-factor repeated-measures ANOVA (Greenhouse Geisser correction) with the factors treatment (vehicle and 7.5 and 15 mg/kg allopregnanolone) and time (2-hr intervals). Post hoc testing was done by means of a two-sided, paired test.
Influence of allopregnanolone administration on its plasma and brain levels.
Forty-five adult male Wistar rats weighing 310 to 380 g were housed in groups of 5 animals under conditions as described. Each animal was injected intraperitoneally with vehicle (n = 5) or 7.5 (n = 20) or 15 (n = 20) mg/kg allopregnanolone at light onset. Under deep inhalation halothane (Hoechst AG, Frankfurt am Main, Germany) anesthesia, trunk blood was collected in heparinized tubes and centrifuged over ice. After decapitation, the entire brain was quickly removed, divided sagittally into two parts and quick-frozen. Plasma and brain samples were stored at −80°C. Samples were taken 15 min after the administration of vehicle (n = 5) and 15 min and 1, 3 and 5 hr after the administration of 7.5 and 15 mg/kg allopregnanolone (n = 5 each). The values for one plasma and one brain sample taken 3 hr after the injection of 7.5 mg/kg allopregnanolone and for one plasma sample taken 3 hr after the administration of 15 mg/kg allopregnanolone are missing.
Plasma and brain samples were homogenized in 5 volumes of ice-cold distilled water and were extracted three times with 3 volumes of ethyl acetate. Approximately 7000 dpm of 3H progesterone was added to the homogenate to monitor recovery. The extracts were dried under vacuum, taken up in 0.2 ml of ethyl acetate and applied to silica-gel F254 thin-layer plates (Merck, Darmstadt, Germany). Authentic reference allopregnanolone was run on separate lanes. The plates were subjected to a combination of systems consisting of carbon-tetrachloride/methanol (99:1 v/v) and cyclohexane/ethyl acetate (3:2 v/v) (Campbell and Karavolas, 1989). Standard allopregnanolone was located by spraying the reference zones with methanol/sulfuric acid (1:1) and heating at 100°C for 5 min. The zone of samples corresponding to reference allopregnanolone was scraped and eluted three times with 2 ml of ethyl acetate. Then, 10 pmol of progesterone was added to the eluate as an internal standard. The eluate was evaporated to dryness under a stream of nitrogen, and the residue was dissolved in 100 μl of acetone. We added 25 μl of HFBAA as a derivatizing reagent, and the samples were allowed to stand at room temperature for 1 hr. The derivatizing reagent was evaporated to dryness under a stream of nitrogen, and the HFBAA derivatives were dissolved in 10 to 20 μl of hexane. Samples were analyzed by gas chromatography-mass spectrometry using a Hewlett-Packard 5971 mass selective detector coupled to a Hewlett-Packard 5890A gas chromatograph equipped with a Hewlett-Packard capillary column (HP-5 ms; length, 30 m; i.d., 0.25 mm; film thickness, 0.25 mm) as previously described (Romeo et al., 1994). Briefly, the derivatized allopregnanolone and progesterone had a unique retention time and a unique fragmentographic pattern. Sensitivity and selectivity were optimized by focusing on one specific m/z value: progesterone 510 m/z and allopregnanolone 496m/z.
Results
Vigilance states.
ANOVA yielded a significant treatment effect on non-REMS latency [F(2,14) = 4.8, P < .04]. Compared with vehicle, both 7.5 and 15 mg/kg allopregnanolone reduced non-REMS latency, the lower dose did so significantly and the higher dose as a tendency (table 1). Allopregnanolone did not significantly influence the frequency or average duration of the non-REMS and REMS episodes.
Effect of allopregnanolone on sleep latency and sleep episode frequency and duration
ANOVA run on the time spent in each vigilance state revealed a significant effect of the factor treatment on the amount of pre-REMS [F(2,14) = 5.3, P < .04]. Over the 6-hr recording period, 7.5 mg/kg allopregnanolone tended to increase pre-REMS, whereas 15 mg/kg significantly promoted pre-REMS, caused by a significant enhancement during the first 2-hr interval (table 2).
Effect of allopregnanolone on percentage of recording time spent in different vigilance states
EEG power densities during non-REMS.
ANOVA run on the normalized and log-transformed non-REMS-specific EEG power densities revealed a significant treatment effect for the frequencies between 1 and 7 Hz and for almost all frequencies ≥ 20 Hz (see bars below the top plot of fig. 1 for results of the ANOVA). Over the 6-hr recording period, allopregnanolone dose-dependently decreased low-frequency EEG activity and enhanced EEG activity in the higher-frequency region (fig. 1, top). Post hoc testing showed that these changes were most prominent during the first 2-hr interval and gradually diminished thereafter (fig.2). After 15, mg/kg allopregnanolone decreases in low-frequency EEG activity persisted throughout the entire recording period (fig. 2, bottom). ANOVA also yielded significant interaction effects between the factors treatment and time for the frequencies 1.5 to 7, 13 to 17 and 19 to 40 Hz. Due to reductions in low-frequency EEG activity and short-lasting elevations of EEG activity in the spindle and higher-frequency bands, both doses of allopregnanolone affected the time course of EEG power density in the respective frequency ranges.
EEG power densities in non-REMS (top) and REMS (bottom) during the first 6 postinjection hr after administration of 7.5 and 15 mg/kg allopregnanolone. Curves connect mean values (± S.E.M., n = 8). For plotting purposes, the data were expressed as a percentage of the corresponding vehicle value. Dots below the graphs denote frequency bands and serve as visual aids. Lines through the dots indicate frequencies for which ANOVA revealed significant (P < .05) treatment, time and interaction effects (run on normalized and log-transformed 2-hr values).
EEG power densities in non-REMS during the first three 2-hr intervals after administration of 7.5 mg/kg (top) and 15 mg/kg (bottom) allopregnanolone. Curves connect mean values (n = 8). For plotting purposes, the data were expressed as a percentage of the corresponding vehicle value. Dots below the graphs refer to frequency bands. Lines through the dots indicate frequencies for which significant differences from vehicle were found (P < .05, two-sided, paired t test run on log-transformed values).
EEG power densities during REMS.
Analysis of the EEG power densities during REMS showed significant treatment and interaction effects for most frequencies of ≥17 Hz (see bars below the bottom of fig. 1 for results of the ANOVA). Over the 6-hr recording period, 7.5 mg/kg allopregnanolone slightly enhanced EEG activity in some of these frequency bands, whereas 15 mg/kg allopregnanolone markedly elevated EEG power density in all frequencies of ≥17 Hz (fig. 1, bottom). The elevations were maximal during the first 2 hr postinjection but were still present during the second 2-hr interval after the higher dose of allopregnanolone (fig. 3).
EEG power densities in REMS during the first three 2-hr intervals after administration of 7.5 mg/kg (top) and 15 mg/kg (bottom) allopregnanolone. For more information, see legend to fig.2.
Plasma and brain concentrations of allopregnanolone.
The levels of allopregnanolone measured in the plasma and brain 15 min after the vehicle injection were generally low, 9.1 ± 7.4 SD pmol/ml and 15.4 ± 9.8 SD pmol/g, respectively. Administration of allopreganolone resulted in dose-dependent elevations in both plasma (computed over all samples: 34.9 ± 26.7 pmol/ml for 7.5 mg/kg and 43.0 ± 29.0 for 15 mg/kg allopregnanolone) and brain (54.3 ± 30.9 pmol/g for 7.5 mg/kg and 64.5 ± 40.3 for 15 mg/kg allopregnanolone). The increases in allopregnanolone concentrations were maximal after 15 min and gradually declined thereafter (fig.4).
Plasma and brain levels of allopregnanolone after administration of vehicle or 7.5 or 15 mg/kg allopregnanolone.
Discussion
The plasma and brain concentrations of allopregnanolone in our vehicle-treated male Wistar rats (fig. 4) are comparable to values observed earlier in male Sprague-Dawley rats (Paul and Purdy, 1992) and ovariectomized female Long-Evans rats (Bitran et al., 1993). The present data show that systemic administration of allopregnanolone mixed with oil produced rapid, long-lasting (>5 hr), seemingly dose-dependent increases in the levels of allopregnanolone in both plasma and brain. Reportedly, allopregnanolone induces a loss of the righting reflex when brain content exceeds 3000 pmol/g (Korneyev and Costa, 1996). Thus, the allopregnanolone levels attained in the present study are far below the anesthetic range.
The present study shows for the first time that exogenous allopregnanolone significantly influences sleep. Both doses tended to reduce non-REMS latency (table 1), which indicates a rapid hypnotic action. The higher dose significantly promoted pre-REMS, which occurred mainly during the first 2 postinjection hr (table 2). Furthermore, spectral analysis of the EEG yielded dose-related changes in non-REMS and REMS. During non-REMS, 15 mg/kg and, to a lesser extent, 7.5 mg/kg allopregnanolone persistently decreased the EEG activity in the lower frequency region (≤7 Hz) and initially enhanced the EEG activity in the spindle frequency range as well as in most higher-frequency bands (fig. 1, top, 2). The lower dose of allopregnanolone had only minor effects on EEG during REMS, whereas the higher dose elicited a significant, overall enhancement of EEG activity in all frequencies of ≥17 Hz (fig. 1, bottom), which was evident during the first two 2-hr intervals (fig. 3, bottom) and, although not statistically significant, transiently elevated EEG activity in the lower theta frequencies (5–6 Hz) and reduced high-frequency theta activity (8–9 Hz).
The most extensively studied agonistic modulators of GABAA receptors, the benzodiazepines, are known to shorten non-REMS latency and selectively increase the time spent in pre-REMS and, at higher doses, tend to inhibit REMS in the rat (Gandolfo et al., 1994; Lancel et al., 1996a,1997). Moreover, the benzodiazepine midazolam has been shown to decrease low-frequency EEG activity and enhance spindling during non-REMS, to shift the dominant theta frequency from 8 Hz to 5 to 6 Hz during REMS and to nonspecifically enhance EEG activity in the higher frequencies (Lancel et al., 1996a, 1997). The influence of allopregnanolone both on the amount of time spent in the different vigilance states and on EEG activity during non-REMS and REMS is reminiscent of the effects induced by benzodiazepines, which suggests that the influence of allopregnanolone on sleep is mediated by GABAA receptors. In almost all previous sleep studies, analysis of EEG activity was limited to the frequency bands between 0.5 and 25 Hz. In the present study, in which we investigated EEG changes up to 40 Hz, we showed that the allopregnanolone-induced increase in high-frequency EEG activity during both sleep states is not limited to 25 Hz but is still evident in 40 Hz (fig. 1). This indicates that agonistic modulators of GABAA receptors induce a general enhancement of fast-frequency EEG signals.
As expected, the kinetics of the increases in allopregnanolone concentrations after administration of allopregnanolone mixed with oil (fig. 4) and of progesterone differ in that progesterone administration results in larger increases that decline more steeply (Lancel et al., 1996b). Nevertheless, the comparison of the sleep changes induced by allopregnanolone with those observed earlier after the administration of various doses of progesterone (Lancel et al., 1996b) shows that the overall effects of 15 mg/kg allopregnanolone are similar to those evoked by 90 mg/kg progesterone. As with the higher dose of allopregnanolone, 90 mg/kg progesterone shortened sleep latency, significantly increased the time spent in pre-REMS and was too low to suppress REMS or affect the number and average duration of the sleep episodes. The effects on sleep state-specific EEG activity are also similar in that this dose of progesterone evoked a decrease in EEG activity in the frequencies of ≤7 Hz and enhanced EEG activity in the spindle and all higher frequencies during non-REMS and lowered the dominant theta frequency, while elevating high-frequency (≥11 Hz) EEG activity during REMS. Furthermore, in both studies, pre-REMS was increased only shortly after the administration of progesterone and allopregnanolone, when brain levels of allopregnanolone were very high (>100 pmol/g), whereas EEG activity in non-REMS and REMS was affected as long as the brain content exceeded 35 pmol/g. The findings that allopregnanolone influences sleep in a manner similar to progesterone and at comparable brain concentrations of allopregnanolone indicate that the hypnotic effects of progesterone are to a large extent mediated by the positive allosteric interaction of its metabolite allopregnanolone with GABAA receptors.
The role of endogenous allopregnanolone in the modulation of sleep has yet to be investigated. Previous research on physiological increases in the level of allopregnanolone in rats showed that brain concentrations of ∼19 pmol/g are attained during estrus, whereas levels around 40 pmol/g can easily be reached during pregnancy and immediately after acute stress (Paul and Purdy, 1992). The only change in the amount of time spent in the different vigilance states, the increase in pre-REMS after the administration of 15 mg/kg allopregnanolone, was limited to the first 2-hr interval, when the brain levels of allopregnanolone were supranormal. However, non-REMS-specific decreases in low-frequency EEG activity and state-independent enhancements in high-frequency EEG activity were still present after the brain levels of allopregnanolone fell within the physiological range (from the third postinjection hour on). Our results thus suggest that allopregnanolone may modulate sleep under physiological conditions. From the pharmacological point of view, steroids related to allopregnanolone, known as epalons, are likely to affect sleep in a benzodiazepine-like fashion.
Acknowledgments
We are grateful to Arnold Höhne for his technical assistance and to Bettina Hermann and Bianca Abstreiter for their help in preparation of the plasma and brain samples.
Footnotes
-
Send reprint requests to: Dr. Marike Lancel, Max Planck Institute of Psychiatry, Clinical Institute, Kraepelinstrasse 2, 80804 Munich, Germany. E-mail:lancel{at}mpipsykl.mpg.de
-
↵1 This study was supported by grants from the Deutsche Forschungsgemeinschaft (M.L.) and the Gerhard-Heβ-Programm of the Deutsche Forschungsgemeinschaft (R.R.).
- Abbreviations:
- GABA
- γ-aminobutyric acid
- EEG
- electroencephalogram
- EMG
- electromyogram
- REMS
- rapid eye movement sleep
- ANOVA
- analysis of variance
- HFBAA
- heptafluorobutyric acid anhydride
- Received January 21, 1997.
- Accepted May 14, 1997.
- The American Society for Pharmacology and Experimental Therapeutics
