Introduction

Several studies have suggested that the neurosteroids pregnenolone sulfate (3β-hydroxy-5-pregnen-20-one-3 sulfate; PREGS), dehydroepiandrosterone sulfate (5-androstene-3β-ol-17-one sulfate; DHEAS) and allopregnanolone (3α,5α tetrahydroxyprogesterone; 3α,5α THPROG) may play a critical role in age-related neuropsychiatric disorders in humans and animals, and in the disruption of sleep and memory processes in particular (Vallée et al. 1997, 2001; Maurice 2001; Racchi et al. 2001; Weill-Engerer et al. 2002a; Mayo et al. 2003; Schumacher et al. 2003). There is little direct evidence of a pathophysiological relationship between neurosteroid concentrations in specific cerebral structures and age-related sleep and memory impairments (Vallée et al. 1997; Weill-Engerer et al. 2002b), but many studies have demonstrated that these neurosteroids affect sleep and memory processes in young subjects. Indeed, the peripheral or central administration of PREGS, DHEAS and 3α,5α THPROG induces robust changes in memory performances and sleep states (Flood et al. 1992; Frye 1995; Isaacson et al. 1995; Meziane et al. 1996; Lancel et al. 1997; Darnaudery et al. 1999a,b; Ladurelle et al. 2000; Damianisch et al. 2001; Matthews et al. 2002; Johansson et al. 2002; Turkmen et al. 2004). There are several lines of evidence suggesting that cholinergic systems may mediate these effects. Firstly, the integrity of cholinergic systems is critical for sleep and memory processes (Everitt and Robbins 1997), and these systems are known to degenerate during aging (Bartus et al. 1982; Perry et al. 1999; Sarter and Bruno 2004). Secondly, although little is known about the anatomic distribution of neurosteroid enzymes in discrete cerebral structures in adults, the key regulator of neurosteroid synthesis, the steroidogenic acute regulatory protein (StAR) (King et al. 2002; Sierra et al. 2003) and neurosteroid synthesis enzymes are produced in cholinergic systems (Rajkowski et al. 1997; Mellon et al. 2001). Thirdly, cholinergic neurons receive numerous γ-aminobutyric acid(GABA)ergic and glutamatergic modulatory afferences, and it is known that the neurosteroids PREGS, DHEAS and 3α,5α THPROG exert their pharmacological effects by modulating GABA receptors (GABAA) and glutamate receptors [-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA), kainate and N-methyl-d-aspartate (NMDA)].

In this review, we will consider anatomical and functional aspects of cholinergic systems and analyze the effects of administering PREGS, DHEAS and 3α,5α THPROG (mainly in rodents) on cholinergic neurotransmission. We will then describe how specific effects of neurosteroids on cholinergic systems may account for the modulation of sleep and memory processes. Finally, we will discuss the potential pathophysiological implications for age-related sleep and memory impairments and suggest future lines of research to confirm the existence of interactions between neurosteroids and cholinergic systems.

Cholinergic systems

Definitions

Two main neuronal systems have been identified—the basal forebrain cholinergic system (BFCS) and the brainstem cholinergic system (BCS)—based on the nomenclature of central cholinergic pathways proposed by Mesulam et al. 1983a,b for rats and primates (Mesulam et al. 1983a) (Fig. 1). The BFCS includes the medial septal nucleus (MS or Ch1), the vertical (vdB or Ch2) and horizontal (hdB or Ch3) limb nuclei of the diagonal band of Broca and the nucleus basalis magnocellularis (NBM or Ch4), the rodent equivalent of the nucleus basalis of Meynert in primates. The BCS, located in the brainstem and part of the reticular formation, encompasses the pedunculopontine tegmental (PPT) nucleus and the laterodorsal tegmental (LDT) nucleus, corresponding to the Ch5 and Ch6 groups described by Mesulam et al. 1983a,b. In addition to these cholinergic projection neurons, several interneurons have been identified in the striatum and nucleus accumbens, olfactory tubercle and islands of Calleja complex (Woolf and Butcher 1981; Houser et al. 1983; Satoh et al. 1983).

Fig. 1
figure 1

Schematic diagram of cholinergic systems in a rodent brain, adapted from Woolf (1991). The two major cholinergic systems composed of projection neurons are represented, the basal forebrain cholinergic system (BFCS) and the brainstem cholinergic system (BCS), together with the cholinergic interneurons of the striatum. The BFCS and the BCS encompass the Ch1-4 and Ch5–6 groups, respectively

Full size image

Basal forebrain cholinergic system

Anatomical organization

In rat, BFCS cholinergic neurons form a constellation of neurons ranging from the anterior medial septal nucleus rostrally to the lateral hypothalamus caudally (Butcher et al. 1992; Oh et al. 1992). Most, if not all, of these cholinergic cells are projection neurons. The medial septum-vertical limb of the diagonal band of Broca (Ch1 and Ch2) innervates the hippocampus, dentate gyrus, entorhinal, perirhinal and retrosplenial cortex and interpeduncular nucleus (Woolf 1991). These neurons receive excitatory/inhibitory afferents from the hippocampus and enthorinal cortex (Dutar et al. 1985; Jakab and Leranth 2005). They are also innervated by afferents from the hypothalamus, ventral tegmental area, LDT, dorsal and median raphe and locus coeruleus (Woolf 1991). The horizontal limb of the diagonal band of Broca (Ch3) is responsible for cholinergic innervation of the olfactory bulb.

The nucleus basalis cholinergic projection (Ch4) is the single most substantial regulatory afferent system of the cerebral cortex in rat (Mesulam 1995). These cortical projections are topographically organized (Mesulam et al. 1983a; Rye et al. 1984; Saper and Chelimsky 1984; Saper 1984; Woolf et al. 1984). NBM neurons also send projections to the basolateral nucleus of the amygdala (Carlsen et al. 1985), the thalamus and the hypothalamus (Mesulam et al. 1983a). Cholinergic neurons of the NBM are interspersed with other non-cholinergic magnocellular corticopetal neurons, mainly GABAergic (Brashear et al. 1986; Zaborszky et al. 1986; Fisher et al. 1988). These GABAergic neurons are twice as numerous as the cholinergic cells (Gritti et al. 1993). Major afferences to the NBM include projections from the amygdala, nucleus accumbens, hypothalamus, ventral tegmentum, PPT, locus coeruleus and raphe nucleus. Terminals containing GABA (Perez et al. 1981), serotonin (Steinbusch and Nieuwenhuys 1981), glutamate (Davies et al. 1984), aspartate (Fuller et al. 1987) and substance P (Haber and Elde 1981) have been described among these afferences.

Functional role

The functional role of the BFCS has been extensively investigated over the last 20 years because this system is profoundly modified in neurodegenerative diseases classically associated with cognitive and sleep disorders (for review, see Everitt and Robbins 1997; Perry et al. 1999). As the main projection of medial septal neurons is the hippocampus, we would expect the functions of these neurons to be closely related to those of the hippocampus (Everitt and Robbins 1997). Hippocampal lesions are known to induce spatial learning and memory defects, and excitotoxic or electrolytic lesions of the medial septum have been reported to induce severe deficits in spatial memory tasks in rats (Hagan et al. 1988; Marston et al. 1993; Kelsey and Vargas 1993). Indeed, the cholinergic projection from the medial septum to the hippocampus may be involved in short-term memory, as suggested by reported deficits in contextual stimulus trace conditioning (McAlonan et al. 1995). Discrete excitotoxic lesions of the vertical limb of the diagonal band of Broca have been reported to lead to small but significant cholinergic depletions in the cingulate cortex and to impair the delayed retention of conditional discrimination performance in rats (Muir et al. 1996). Many experiments on excitotoxic/electrolytic lesions in rodents carried out from the 1980s onwards revealed that the NBM was involved in the regulation of cognitive processes, including attention, learning and memory, in particular, and in the regulation of sleep states (for review, see Everitt and Robbins 1997; Wenk 1997).

The precise role of cholinergic neurons remains a matter of debate. Cholinergic neurons are intermingled with a significant population of GABAergic neurons, and the lack of a selective toxin targeting cholinergic neurons was a recurrent problem in studies. Experiments involving the use of a specific cholinergic toxin (192 immunoglobulin G (IgG)–saporin) for lesioning basal forebrain cholinergic neurons have suggested that even if cholinergic neurons are involved in some memory processes, they seem to play a more important role in attentional processes (Muir et al. 1992, 1995; Berger-Sweeney et al. 1994; Torres et al. 1994; Baxter et al. 1995; Sarter and Bruno 1997). These results suggest that cholinergic projections from the BFCS (MS, vdB and NBM) are primarily involved in the modulation of attentional and memory processes.

The NBM may also be seen as a ventral extrathalamic relay from the brainstem reticular activating system to the cerebral cortex for the modulation of sleep/wake states (Shute and Lewis 1963; Krnjevic and Silver 1965; Moruzzi and Magoun 1995). Briefly, sleep/wake states include (1) wakefulness, characterized by a low-amplitude desynchronized electroencephalogram (EEG) and high muscular tone; (2) non-rapid eye movement (non-REM) sleep, characterized by a high-amplitude synchronized EEG mainly in the delta (0–4 Hz) and spindle (12–15 Hz) bands; and (3) REM sleep, characterized by a low-amplitude desynchronized EEG without muscular tone. More acetylcholine (ACh) is released in the neocortex during waking and REM sleep than during non-REM sleep (Phillis 1968; Jasper and Tessier 1971). Consistent with this finding, the discharge rates of NBM neurons have been shown to be maximal during waking and REM sleep in cats (Detari et al. 1984), and the basal forebrain has been shown to be deactivated during non-REM sleep in humans (Maquet 2000). Furthermore, lesioning of the NBM abolishes both REM and non-REM sleep (Szymusiak and McGinty 1986) and alters EEG synchronization (Buzsaki et al. 1988; Riekkinen et al. 1991). The state-dependent pattern of activity of NBM neurons may be partly regulated by changes in GABA-mediated inhibition. Indeed, cholinergic NBM neurons receive substantial GABAergic inputs, including local interneurons (Zaborszky et al. 1986; Ingham et al. 1988), and the infusion of GABA agonists or antagonists into the NBM alters the sleep-/wakefulness-related pattern of discharge of these neurons (Szymusiak et al. 2000). Interestingly, some authors have suggested that much of the variation in discharge of NBM neurons during waking and sleep may reflect changes in the activity of brainstem afferents (Szymusiak et al. 2000).

The BFCS therefore (1) modulates sleep/wake states, favoring arousal, via its projections to the neocortex and (2) enhances memory consolidation during wakefulness and/or sleep via projections to the amygdala and hippocampus.

Brainstem cholinergic system

Anatomical organization

The BCS comprises mostly large cholinergic neurons and, as for the cholinergic groups of the basal forebrain, brainstem cholinergic neurons form an integral part of the BCS (Armstrong et al. 1983; Rye et al. 1987; Jones 1990; Steininger et al. 1997) and are intermingled with a variety of other neurons such as GABAergic (Ford et al. 1995; Bevan and Bolam 1995; Torterolo et al. 2001) and glutamatergic neurons (Clements and Grant 1990; Charara et al. 1996). The BCS receives numerous projections from the surrounding area, including the ventral tegmental area, substantia nigra, raphe nucleus and locus coeruleus, and from other distal structures such as the hypothalamus, subthalamic nucleus and amygdala (Rye et al. 1987; Steininger et al. 1997). The main ascending projections from the BCS are the thalamic nuclei, basal forebrain, and brainstem nuclei (ventral tegmental area, raphe nucleus and locus coeruleus) (Fig. 1; Rye et al. 1987; Steininger et al. 1997). There are also some descending projections to the deep cerebellar nuclei, medioventral medulla and pontomedullary reticular nuclei (Fig. 1). BCS neurons are regulated primarily by serotonergic, noradrenergic, GABAergic and glutamatergic inputs originating from the pontomesen cephalic reticular formation and the PPT itself (Steckler et al. 1994; Rye 1997). These cholinergic neurons are also regulated by the autocrine/paracrine release of nitric oxide (NO) (Datta et al. 1997; Leonard and Lydic 1997).

Functional role

Several studies have demonstrated that the electrical activity of BCS neurons depends on sleep–wake state (Saito et al. 1977; el Mansari et al. 1989; Steriade et al. 1990; Kayama et al. 1992; Datta and Siwek 2002). It has been shown that the BCS contains at least two classes of neurons in animals: the REM-ON neurons, which display preferential discharge activity during REM sleep, and the wake/REM-ON neurons, which show preferential discharge activity during wakefulness and REM sleep. Wake/REM-ON neuronal activity increases the release of ACh into the pons and thalamus, leading to the electroencephalographic desynchronization seen in the wake and REM sleep states (Dingledine and Kelly 1977; Steriade et al. 1990; Datta 1997; Datta and Siwek 1997). As some BCS neurons are also strictly involved in the initiation of REM sleep, the BCS is now considered to be a key structure in the control of REM–nonREM sleep transitions (Hobson and Pace-Schott 2002; Pace-Schott and Hobson 2002).

Although the role of the BCS in sleep regulation has been extensively investigated, the role of this system in cognitive processes, and particularly in memory, has received little attention. However, the data reported suggest that the BCS may be involved in memory processes. In rodents, excitotoxic/electrolytic lesions of the BCS have been shown to impair contextual memory acquisition in numerous tasks (Fujimoto et al. 1989, 1992; Dellu et al. 1991; Satorra-Marin et al. 2001; Keating et al. 2002; Mitchell et al. 2002; Taylor et al. 2004). However, it is impossible to conclude that cholinergic neurons of the BCS are entirely responsible for these memory effects due to the lack of specific cholinergic-targeted toxins for the induction of lesions in the BCS.

Sleep and memory processes may be seen as independent functions, but recent studies suggest that these two functions may be linked, with the BCS underlying both. Indeed, the cholinergic activation of phasic pontine-wave generator cells in the brainstem improves memory consolidation and prevents REM sleep-deprivation-induced memory impairment in the active avoidance task (Mavanji and Datta 2003; Datta et al. 2004), suggesting that the BCS may play a key role in sleep-dependent memory consolidation.

Modulation of cholinergic transmission by PREGS, DHEAS and THPROG

Septo-hippocampal projection neurons

Several studies based on microdialysis coupled with high-performance liquid chromatography (HPLC) have shown that neurosteroids modulate the release of ACh in freely moving rats. The intraperitoneal (i.p.) administration of various doses of DHEAS (25–250 μmol/kg) increases ACh release in the hippocampus (Rhodes et al. 1996). The highest dose was found to increase ACh release by a factor of more than 4 with respect to treatment with saline. Similarly, the intracerebroventricular (i.c.v.) administration (12–192 nmol/5 μl) of PREGS induces a dose-dependent increase in ACh release into the hippocampus (Vallée et al. 1997; Darnaudery et al. 2000). The administration of 12 or 48 nmol of PREGS induces a transient (20 min) increase in the release of ACh, with a maximum around 120% above baseline. The administration of 96- and 192-nmol doses induced a longer-lasting (80 min) increase that peaked around 300% above baseline. This DHEAS-/PREGS-induced hippocampal ACh release was also observed indirectly, following daily i.p. administration of the non-steroidal steroid sulfatase inhibitor (p-O-sulfamoyl)-N-tetradecanoyl tyramine (DU-14) for 15 days (Rhodes et al. 1997), at a dose sufficient to increase plasma sulfated steroid concentrations by up to 88%. Consistent with these results, local infusion of PREGS (12 pmol/0.5 μl) into the MS induces a transient (30 min) 50% increase in ACh release in the hippocampus over baseline levels (Darnaudery et al. 2002). This result, obtained following local infusion into the MS, is similar to that obtained after i.c.v. administration of 12 and 48 nmol of PREGS, suggesting that the PREGS-induced release of ACh in the hippocampus depends primarily on a specific effect on the MS (Fig. 2). In contrast to the increase in ACh release observed after the administration of DHEAS and PREGS, the i.c.v. administration of 3α,5α THPROG (15–45 nmol/5 μl) decreases basal ACh release in the hippocampus in a dose-dependent manner (Dazzi et al. 1996). Doses of 30 and 45 nmol decrease ACh release by 20 and 55%, respectively, with respect to baseline (Fig. 4a). At a dose of 30 nmol, 3α,5α THPROG completely prevented the increase in ACh release induced by footshock stress, suggesting that the neurosteroid modulation of cholinergic systems could be of physiological importance. Thus, the i.p. or i.c.v. infusion of neurosteroids can modulate hippocampal ACh release, probably through a specific effect on MS cholinergic neurons. However, the regional specificity of neurosteroids remains to be clearly demonstrated using control infusions in other afferent structures to the hippocampus and the hippocampus itself. These results also highlight the opposite effects of DHEAS/PREGS and 3α,5α THPROG on ACh release in the hippocampus (Fig. 4a).

Fig. 2
figure 2

Effects of intracerebroventricular (icv; 48 nmol) and intra-medial septum (intra-MS; 12 pmol) PREGS infusions on the release of acetylcholine in the hippocampus, as assessed by microdialysis coupled to HPLC in freely moving rats. Note that the enhancement of cholinergic release in the hippocampus observed after icv infusion is reproduced following local infusion into the MS. Data represent the mean expressed as fold changes vs baseline. Adapted from Vallée et al. (1997) and Darnaudery et al. (2002)

Full size image

Basalo-cortical and basalo-amygdalar projection neurons

Most of the cholinergic sources of the frontal cortex and amygdala come from the BFCS and the NBM. The release of ACh in the frontal cortex induced by i.c.v. administration of PREGS (0, 12, 48, 96 and 192 nmol/5 μl) was investigated by Darnaudery et al. (1998) using intracerebral microdialysis in freely moving rats. Extracellular ACh levels in the cortex were found to increase in a dose-dependent manner. The highest doses (96 and 192 nmol) tripled ACh release, the intermediate dose of 48 nmol doubled ACh release, and 12 nmol PREGS had no effect on extracellular levels of ACh. The increase in cortical ACh concentration was maximal 30 min after administration for all active doses. These data suggest that the cholinergic projections of the NBM require higher concentrations of PREGS (48 nmol) than the cholinergic projections of the MS (12 nmol) to increase ACh release in the corresponding terminals. Pallares et al. (1998) confirmed the effect of PREGS on the cholinergic projection neurons of the NBM (Fig. 3). They showed that the infusion of PREGS (12 pmol/0.5 μl) directly into the NBM induces a long-lasting release (130 min) of ACh in the frontal cortex and amygdala. In contrast, i.c.v. administrations (30–45 nmol/5 μl) of 3α,5α THPROG decrease ACh release in the frontal cortex of rats (Dazzi et al. 1996) (Fig. 4b). Thus, as for septo-hippocampal projection neurons, PREGS and 3α,5α THPROG act in opposite manners on basalo-cortical and basalo-amygdalar projection neurons, suggesting that basal cholinergic release may be controlled by a balance between these two steroids. Similarly, as for septo-hippocampal projection neurons, the regional specificity of neurosteroids remains to be clearly demonstrated using control infusions in other afferent structures to the cortex and amygdala.

Fig. 3
figure 3

Effects of intracerebroventricular (icv; 48 nmol) and intra-nucleus basalis magnocellular (intra-NBM; 12 pmol) PREGS infusions on the release of acetylcholine in the frontal cortex, as assessed by microdialysis coupled to HPLC in freely moving rats. Note that the enhancement of cholinergic release in the frontal cortex observed after icv infusion is reproduced following local infusion into the NBM. Data represent the mean expressed as fold changes vs baseline. Adapted from Darnaudery et al. (1998) and Pallares et al. (1998)

Full size image
Fig. 4
figure 4

Summary of the changes in cholinergic transmission observed following icv infusions of PREGS and THPROG, as assessed by microdialysis coupled to HPLC in freely moving rats. Note that PREGS (black) and THPROG (gray) exert opposite effects on the release of ACh in the hippocampus (a) and frontal cortex (b) but have no effect on striatal cholinergic release (c). Data represent the mean expressed as fold changes vs baseline against time after injection. Adapted from Dazzi et al. (1996), Vallée et al. (1997) and Darnaudery et al. (1998)

Full size image

Striatal interneurons

Cholinergic interneurons of the striatum account for only 1–3% of all neurons in the striatum, but their extensive axonal/dendritic arborization in the striatum is critical for the integration of information in the striatum. Neurosteroids seem to have no effect on striatal cholinergic interneurons in rat. Indeed, the i.c.v. administration of PREGS at the doses (0, 12, 48, 96 and 192 nmol/5 μl) active in other cholinergic systems has absolutely no effect on striatal ACh release as measured by microdialysis in freely moving rats (Darnaudery et al. 1998) (Fig. 4c). Similarly, i.c.v. administration (15–45 nmol/5 μl) of 3α,5α THPROG (Dazzi et al. 1996) has no effect on striatal cholinergic transmission (Fig. 4c). Thus, either striatal cholinergic interneurons require higher concentrations of neurosteroids for changes in ACh release or neurosteroids are not involved in the physiological regulation of cholinergic release in the striatum.

The results described above demonstrate that neurosteroids have a strong influence on cholinergic transmission. The modulation of this transmission is specific to cholinergic projections (MS, NBM and PPT) because neurosteroids have no effect on striatal cholinergic interneurons. The differential effects of neurosteroids on the BFCS, BCS and on striatal cholinergic systems may result from differences in the composition of GABAA receptor subunits in these systems, as the effects of neurosteroids on GABAA receptors depend on subunit composition. In particular, the presence of the δ subunit has been shown to increase the sensitivity of GABAA receptors to neurosteroids (Lambert et al. 2003). In line with this, neurons in the BFCS and BCS produce mostly the α1, β2, γ1, γ3, ɛ, δ and α3, γ3, δ subunits, respectively, whereas those in the striatum produce mostly the α3, α5, β1 and β2 subunits, but these are devoid of δ subunits (Moragues et al. 2000, 2002; Pirker et al. 2000). For a full evaluation of the effects of neurosteroids on cholinergic systems, further studies are required to demonstrate the effects of neurosteroids on cholinergic transmission from the BCS.

Modulation of cholinergic-related functions by PREGS, DHEAS and THPROG

Memory processes

Many studies have investigated the pharmacological effects of peripheral and central administrations of PREGS, DHEAS and 3α,5α THPROG on memory processes in rats. PREGS and DHEAS have been shown to improve memory, whereas 3α,5α THPROG has been shown to impair memory. These effects have been observed for several memory-related tasks such as the active and passive avoidance task (Flood et al. 1992; Isaacson et al. 1995), appetitively reinforced go–no go visual discrimination task (Meziane et al. 1996), the Y-maze (Ladurelle et al. 2000), a spatial version of the water-maze task (Frye and Lacey 1999; Matthews et al. 2002; Johansson et al. 2002; Turkmen et al. 2004) and a spatial recognition task (Darnaudery et al. 1999b, 2000). The memory-enhancing effect of i.c.v. administration of PREGS is correlated with a parallel increase in ACh release in the hippocampus (Darnaudery et al. 2000), suggesting that PREGS acts on the BFCS. These findings were confirmed after local infusions of PREGS into the BFCS. Local infusion into the MS (12 pmol/0.5 μl) and the NBM (12 pmol/0.5 μl) improves spatial memory and increases ACh release in the corresponding projection structures: hippocampus, amygdala and frontal cortex (Pallares et al. 1998; Darnaudery et al. 2002). Consistent with the opposite actions of PREGS and 3α,5α THPROG on cholinergic transmission from the BFCS, these two neurosteroids have opposite effects on spatial memory performance. Mayo et al. (1993) found that PREGS (12 pmol/0.5 μl) infusion in the NBM enhanced spatial performance in the Y-maze recognition task, whereas the infusion of 3α,5α THPROG (0.6–6 pmol/0.5 μl) in the NBM impaired performance in this task (Fig. 5). PREGS and DHEAS have been shown to enhance memory in physiological conditions in young subjects, and PREGS and DHEAS have been shown to reverse memory impairments in various rodent models of amnesia (Mathis et al. 1994, 1996; Meziane et al. 1996; Urani et al. 1998; Zou et al. 2000), suggesting that these neurosteroids may be of therapeutic value for the memory impairments observed during aging.

Fig. 5
figure 5

Memory-enhancing and memory-impairing effects in the Y-maze after local infusion of PREGS and THPROG into the BMN. Data are expressed as variation of the recognition index as a percentage of the value for control animals (maximum SEM of control animals indicated by the striped area). Adapted from Mayo et al. (1993)

Full size image

Sleep states

Few studies have investigated the effects of peripheral administration of PREGS, DHEAS and 3α,5α THPROG on sleep/wake states (Mendelson et al. 1987; Lancel et al. 1997; Darnaudery et al. 1999a; Schiffelholz et al. 2000). The peripheral administration of 3α,5α THPROG leads to robust changes in sleep architecture and cortical activities during sleep in rodents. The i.p. administration of 3α,5α THPROG (24–48 μmol/kg) decreases non-REM sleep latency and EEG power in the delta band (0–4Hz), whereas it increases EEG power in the spindles band (12–15Hz). The peripheral administration of 3α,5α THPROG therefore tends to have a benzodiazepine-like effect, increasing the tendency to fall asleep and promoting non-REM sleep at the expense of REM sleep, which may be transiently abolished at the highest doses (Mendelson et al. 1987; Lancel et al. 1997; Schiffelholz et al. 2000; Damianisch et al. 2001). In contrast, the i.p. administration of PREGS (113 μmol/kg) increases the amount of REM sleep without affecting non-REM sleep in rats (Darnaudery et al. 1999a; Schiffelholz et al. 2000). The opposite effects of PREGS and 3α,5α THPROG on REM sleep are consistent with those obtained following local infusions into the NBM. The infusion of PREGS (12 pmol/0.5 μl) in this structure induces a long-lasting (8 h) increase (+30%) in the amount of REM sleep, whereas the infusion of 3α,5α THPROG (6 pmol/0.5 μl) induces a long-lasting decrease (−25%) in the amount of REM sleep (Darnaudery et al. 1999b).

We have also shown that PREGS administration in the PPT affects sleep–wakefulness states in a dose-dependent manner (Darbra et al. 2004). The infusion in the PPT of a low concentration (12 pmol/0.5 μl) of PREGS, similar to that known to induce a slight increase in REM sleep following infusion into the NBM, produces a robust, transient (90 min) increase in the amount of REM sleep (+200%), with no change in non-REM sleep and wakefulness, suggesting that PPT cholinergic neurons are a primary target of PREGS. Moreover, increasing the dose of PREGS (24–48 pmol/0.5 μl) increases REM sleep and non-REM sleep, and also increases delta power and decreases theta power during wakefulness. Thus, depending on the dose used, PREGS can promote REM sleep alone or the global propensity to fall asleep, impairing the quality of wakefulness. The regulation and the functional role of neurosteroids appear to be radically different in the NBM and the PPT, and the effects of 3α,5α THPROG on these two systems differ: 3α,5α THPROG infusion in the NBM decreases the amount of REM sleep, whereas the same infusion in the PPT induces no change (Fig. 6).

Fig. 6
figure 6

Summary of REM sleep results obtained after local infusions of PREGS and THPROG into the BFCS (BMN) and BCS (PPT). REM sleep is assessed by chronic sleep recording and expressed as a percentage of the control in each experiment (maximum SEM of control animals indicated by the striped area). Note that in the BMN, the two neurosteroids have opposite effects, whereas in the PPT, only PREGS has an effect, inducing a robust increase in REM sleep. Adapted from Darnaudery et al. (1999a,b) and Darbra et al. (2004)

Full size image

The data described here provide strong evidence for the differential regulation of the BFCS and BCS by neurosteroids. The effects of PREGS and 3α,5α THPROG on these two systems suggest that (1) cholinergic projection neurons of the NBM are controlled by a physiological balance between PREGS and 3α,5α THPROG because in the NBM, these two neurosteroids affect sleep states, memory and cholinergic transmission in opposite manners; (2) cholinergic projection neurons of the PPT are controlled by strong 3α,5α THPROG tonic inhibition that is reversed by exogenous PREGS because 3α,5α THPROG administration has no effect on sleep state. Given that REM sleep is controlled by the BCS, it could be suggested that the infusion of PREGS (but not of 3α,5α THPROG) into the BCS should enhance cholinergic transmission in the thalamus and pons. Confirmation of the hypothesized local and differential actions of these neurosteroids in the BFCS and BCS will require measurements of the concentration of neurosteroids in these regions and analysis of the various mechanisms of neurosteroidogenesis regulation in these two systems.

Pathophysiological implications for age-related sleep and memory disorders

Neurosteroid–cholinergic system interactions are thought to be involved in age-related sleep and memory disorders primarily because a marked change in cholinergic systems is observed during aging (Muir 1997; Mesulam 1998). Such changes take place in both the BFCS and BCS, in which decreases in numbers of ACh neurons and in ACh transmission have been described in aged rodents, primates and humans (Fischer et al. 1992; Kobayashi et al. 1994; Lolova et al. 1996, 1997; Martinez-Serrano and Bjorklund 1998; Ransmayr et al. 2000) and in demented patients (Perry et al. 1995; Arendt et al. 1995a,b, 1997; Herholz et al. 2004). Age-related changes in cholinergic-related functions have been also demonstrated. Although human aging is associated with numerous neuropsychiatric changes that affect daily life, a hallmark of these changes is the higher prevalence of memory and sleep disorders in the elderly population. The dysfunctions of episodic and working memories (Grady and Craik 2000; Nyberg et al. 2002) and the decrease in amplitude of the sleep–wake circadian rhythm associated with a fragmentation of non-REM sleep have critical health outcomes (Rosenberg et al. 1979; Ingram et al. 1982; Stone 1989; Myers and Badia 1995; Van Someren 2000; Dagan 2002; Mignot et al. 2002). For instance, a lack of restful sleep at night results in excessive daytime sleepiness, attention and memory problems, depressed mood, falls, and poor quality of life (Young 2004; Foley et al. 2004). The mechanisms underlying age-related sleep and memory disorders are poorly understood, but numerous correlations have been demonstrated between cholinergic dysfunctions and memory impairments in aged subjects. Published results suggest that changes in neurosteroid concentrations in the BFCS and BCS may mediate these dysfunctions.

Extensive but controversial studies have shown that plasma steroid levels change with age and may be associated with memory deficits in aged humans and demented patients (see for review Vallée et al. 2001, 2004). However, there have been few experimental reports of changes in brain neurosteroids during aging or relationships between brain neurosteroid levels and age-related dysfunctions. Overall decreases in PREGS and 3α,5α THPROG levels in the hippocampus and cerebral cortex, respectively, have been reported in aged rats, with young rats used as the reference group (Vallée et al. 1997; Bernardi et al. 1998), whereas 3α,5α THPROG levels in the hypothalamus have been reported to increase with age (Bernardi et al. 1998). However, Barbaccia et al. (1998) reported no change in 3α,5α THPROG concentration in the cortex during aging (Barbaccia et al. 1998). Interestingly, hippocampal PREGS concentrations have been shown to be negatively correlated with memory impairments in aged rats (Vallée et al. 1997), strongly suggesting that hippocampal PREGS is important for memory processing in aged animals. In humans, changes in brain neurosteroid concentrations have been described in demented patients [Alzheimer’s disease (AD)], with age-matched controls used as the reference group (Weill-Engerer et al. 2002a; Kim et al. 2003; Brown et al. 2003). Weill-Engerer et al. (2002a,b) reported decreases in the concentrations of PREGS and DHEAS in the striatum, hypothalamus (DHEAS only) and cerebellum of demented patients, consistent with the decrease in cerebrospinal fluid (CSF) DHEAS levels observed in patients with AD and vascular dementia (Kim et al. 2003). Moreover, DHEA levels have been shown to be high in the hippocampus, hypothalamus, frontal cortex and CSF in AD patients (Brown et al. 2003). Given the small number and conflicting nature of the results obtained to date, further studies are required to determine the involvement of neurosteroids in age-related sleep and memory dysfunctions in normal aging and pathological aging, as in neurodegenerative disorders.

Conclusions

This review deals with the modulation of cholinergic transmission and cholinergic-related function by neurosteroids. We suggest that neurosteroids play a critical role in sleep and memory processes by selectively modulating the BFCS and BCS, and that the dysregulation of neurosteroid synthesis in these regions may play a key role in sleep and memory disorders during aging. Improvements in our understanding of the role of neurosteroids in these processes in young subjects and in pathological conditions during aging will require a demonstration that neurosteroidogenesis is particularly important in cholinergic systems. Further studies are required (1) to demonstrate the specific cerebral regionalization, particularly in cholinergic structures, of neurosteroids and their associated proteins (steroidogenic enzymes, StaR), (2) to determine whether this cerebral regionalization changes during the subject’s lifetime and (3) to study the relationship between neurosteroid concentrations in cholinergic systems and the sleep/memory impairments observed in some subjects during aging.