According to earlier studies there are several endogenous and synthetic compounds that affect sleep when injected systemically or directly into the CNS. The sleep modulatory mechanism of these substances can be either the direct effect on sleep regulatory brain mechanisms or an indirect effect mediated by e.g. metabolism, thermoregulation or the immune system. When studying brain mechanisms that regulate sleep, it is important to get knowledge about not only the sleep modulatory effects of pharmacological manipulation of a neuronal system, but also about the reciprocal effects of the sleep-wake rhythm and sleep deprivation on the activity of the studied neuronal system. In the case of the GH-axis and its effects on sleep there was already established knowledge about the synchronization of GH-secretion and sleep-wake rhythm, and about the sleep modulatory effects of pharmacological manipulation of the GH-axis, while the effects of sleep deprivation on the activity on the GH-regulating neuropeptides in the brain were mostly unknown.
Our results showed that REM sleep deprivation increased somatostatin and
decreased GHRH mRNA in the hypothalamus, while total sleep deprivation
(combined lack of slow wave and REM sleep) increased both GHRH and somatostatin
mRNA in the hypothalamus. Earlier injection studies have revealed that
injection of GHRH was capable to induce SWS, suggesting that GHRH is a
possible SWS facilitatory neuropeptide (Obál
et al. 1996). Our finding that GHRH mRNA increases during SWS loss
but not during REM sleep deprivation when SWS is possible, is generally
in accordance with the findings of the injection studies. According to
the injection studies, somatostatin could facilitate REM sleep but these
results obtained from animal and human studies are not as consistent as
the results of the effect of GHRH on SWS. Our results that the lack of
REM sleep achieved either by REM or total sleep deprivation will lead to
increases in somatostatin mRNA in the hypothalamus, and the finding that
i.c.v. or local injection of somatostatin antagonist will reduce REM sleep
give further support to the hypothesis that somatostatin facilitates REM
sleep in the brain. We also found that REM sleep deprivation increases
the expression of galanin mRNA in the anterior hypothalamus although total
sleep deprivation did not affect galanin mRNA. I.c.v. or local microinjection
of galanin into the locus coeruleus did not affect the amount of sleep.
This may indicate that galanin could mediate some of the effects of REM
sleep deprivation in the brain, but does not itself affect sleep.
The GHRH mRNA expression varied essentially between the hypothalamic GHRH mRNA expressing nuclei after total or REM sleep deprivation. In the paraventricular nucleus total sleep deprivation increased while REM sleep deprivation decreased GHRH mRNA. On the other hand in the arcuate nucleus there were no significant changes after either total or REM sleep deprivation. In the periventromedial hypothalamic area there was a difference in the amount of GHRH mRNAs between morning and afternoon samples. It is of interest that sleep deprivation was ineffective in this respect. Thus it is possible that the GHRH nuclei in the hypothalamus have partly different functions (Fig. 14). In the following chapters the effects of sleep deprivation on different GHRH nuclei in the hypothalamus are discussed separately.
After 24 h REM sleep deprivation the amount of GHRH mRNA was reduced in the paraventricular nucleus and this effect was also found after 72 h deprivation. However, at the 72 h time point the amount of GHRH mRNA was similarly decreased in the large platform controls. This may be due to the possibility that rats staying on the large platforms also experienced partial reduction of sleep although this was not not found in an earlier study from our laboratory (Porkka-Heiskanen et al. 1995). The effects of sleep deprivation may occur more slowly during staying on the large platforms than during staying on the small platforms. In the latter case 24 h REM sleep deprivation was already enough to reduce the amount of GHRH mRNA.
Total sleep deprivation for 6 h during the light phase increased the amount of GHRH mRNA in the paraventricular nucleus, which was an opposite effect to that seen in the REM sleep deprivation experiment. Total sleep deprivation for 12 hours during the dark phase did not affect GHRH mRNA in the paraventricular nucleus when compared to controls. The higher effect of the shorter deprivation may be due to the fact that rats will mostly sleep during the first half of the light phase, when the 6 h deprivation was performed, and thus more sleep was lost when compared to situation in the 12 h deprivation during the dark phase, when rats normally are mostly awake (Franken et al. 1991, Tobler and Borbély 1990). In a recent study the effects of total sleep deprivation on GHRH mRNA in the hypothalamus were measured by a reverse transcription polyclonal chain reaction (RT-PCR) method, and an increase in the GHRH mRNA in the whole hypothalamus was observed after 8-12 h total sleep deprivation (Zhang et al. 1998). This finding agrees with our present findings in the specific hypothalamic nucleus.
We found that total sleep deprivation increases, while REM sleep deprivation
decreases, the amount of GHRH mRNA in the paraventricular nucleus. This
may be due to the different degree of sleep deprivation obtained by the
total and the REM sleep deprivation methods. During the total sleep deprivation
both non-REM and REM sleep are inhibited, but during REM sleep deprivation
non-REM sleep was reduced only partly. A decrease of 38% in non-REM sleep
during the first 24 h was observed, while REM sleep was inhibited almost
et al. 1995). Our finding suggests that total inhibition of non-REM
sleep activates the GHRH cells in the paraventricular nucleus. Reduction
of GHRH mRNA in the paraventricular nucleus during REM sleep deprivation
may be due to the inhibition which is mediated by somatostatin or by other
transmitters which mediate the effects of REM sleep deprivation or stress.
Unfortunately the afferent connections of the paraventricular GHRH cells
are poorly known. Nevertheless, it is possible that this subgroup of GHRH
cells mediates the SWS facilitatory effect of GHRH as suggested by the
pharmacological studies (Fig. 14). The paraventricular
GHRH cells are hypophysiotropic (Liposits
1993), but the nucleus has also projections to other brain areas such
as preoptic area and basal forebrain (Sawchenko
et al. 1985), where cholinergic and GABAergic cells regulate wakefulness
and sleep (Szymusiak 1995) (Fig.14).
Some cells in the preoptic area express GHRH receptors (Takahashi
et al. 1995), but the transmitter identities of these cells are
unknown. In a recent study it was found that local microinjections of GHRH
into the preoptic area increased sleep and that microinjections of GHRH
antagonist decreased spontaneous sleep and rebound sleep after short term
sleep deprivation (Zhang et al.
1999). This supports the hypothesis that the preoptic area may be the
target area for the SWS facilitatory effect of the endogenous GHRH.
In our study neither REM nor total sleep deprivation affected the amount of GHRH mRNA in the arcuate nucleus. It is known that the arcuate GHRH cells are mostly hypophysiotropic and that there are very few projections to other areas than the median eminence among these cells (Merchenthaler et al. 1985). Thus, it is possible that regulation of GH secretion is the main function of the arcuate GHRH cells and that the mRNA expression of these cells is thus not particularly sensitive to sleep loss (Fig. 14). Because secretion of GH is reduced during both REM (II) and total sleep deprivation (Kimura and Tsai 1984), it is probable that the arcuate GHRH cells are somehow inhibited by hypothalamic somatostatin cells, which increase mRNA expression during sleep deprivation. Increased somatostatin activity may reduce the release of GHRH into the pituitary portal circulation, while the synthesis of GHRH remains unaltered, which is indicated by the unchanged amount of the arcuate GHRH mRNA after both total and REM sleep deprivation.
In the periventromedial hypothalamic area the amount of GHRH mRNA was higher in the rats decapitated in the morning (9:00 am) than in the afternoon (3:00 pm), while there was no difference between the sleep deprivation and control groups. The amount of GHRH mRNA in the whole of the hypothalamus samples has in another study using Northern blot analysis also been found to be higher in the morning (Bredow et al. 1996). One possibility is that this variation of GHRH mRNA may be associated with the daily sleep-wake distribution of the rat. Rats typically have the highest amount of non-REM sleep during the first half of the light phase (Borbély and Neuhaus 1978), at the same time when the amount of the periventromedial GHRH mRNA was found to be high. The periventromedial GHRH cells are not hypophysiotropic, but have projections to the preoptic area (Sawchenko et al. 1985), which could mediate the sleep association of these cells similarly as in the case of the paraventricular GHRH cells (Fig.14). The afferent connections of the periventromedial GHRH cells are unclear. It is possible, though at this point only hypothetical, that the activity of these cells is controlled by the suprachiasmatic or ventromedial nuclei, which activity oscillates in an endogenous circadian rhythm. The circadian association of sleep distribution may be partly mediated by this neuronal pathway.
Fig. 14 Schematic
illustration of the possible roles of the hypothalamic GHRH cell populations.
PaV: paraventricular nucleus, pVMH: periventromedial hypothalamic area.
According to our findings, both 24 h REM sleep deprivation and 6-12 h total sleep deprivation increased the amount of somatostatin mRNA in the arcuate nucleus. After the longer 72 h REM sleep deprivation there was no difference in the arcuate somatostatin mRNA between the REM sleep deprivation and control groups. The observed activation of the arcuate somatostatin cells as a consequence of REM sleep loss caused by REM or total sleep deprivation is in accordance with the injection studies suggesting that somatostatin facilitates REM sleep. It is known that the arcuate somatostatin cells are not hypophysiotropic, but project locally in the arcuate nucleus where they inhibit the hypophysiotropic GHRH cells (Merchenthaler et al. 1989, Willoughby et al. 1989). The amount of somatostatin mRNA oscillates in the arcuate nucleus in association with the GH secretion rhythm (Zeitler et al. 1991). By their effect on the hypophysiotropic GHRH cells in the same nucleus, these somatostatin cells may mediate the inhibition of GH secretion which is found during REM (II) and total sleep deprivation (Kimura and Tsai 1984). During the longer REM sleep deprivation the amount of somatostatin mRNA in the arcuate nucleus returned to the level of the control group. This may be a consequence of the feedback regulation caused by the decreased plasma GH, which may inhibit somatostatin cells in the arcuate nucleus.
In the periventricular nucleus the amount of somatostatin mRNA increased after 72 h REM sleep deprivation when compared to large platform control treatment but not when compared to unhandled animals allowed normal sleep. The durations of total sleep deprivation of 6 and 12 h did not affect the amount of the periventricular somatostatin mRNA. The periventricular nucleus is the most important source of hypophysiotropic somatostatin by its projection to the median eminence (Merchenthaler et al. 1989). There are findings that this somatostatin cell group projects also to other brain areas such as arcuate nucleus and locus coeruleus (Dickson et al. 1994, Liposits 1993). Because sleep deprivation decreases GH secretion, it would be logical that the activity of the periventricular somatostatin cells increases during the deprivation. It is possible that the transmitter synthesis rate and the stores are so high in this massive somatostatin cell population that an induction of mRNA expression is not needed during short sleep deprivation. Another possibility is that these cells regulate primarly GH secretion as do the GHRH cells in the arcuate nucleus, but that the cells are not sensitive to short sleep deprivation.
Some somatostatin cells in the hypothalamus are connected to cells in the locus coeruleus (Palkovits et al. 1982, Liposits 1993, Mounier et al. 1996). Somatostatin immunoreactive fibers have been found in the locus coeruleus (Sutin and Jacobowitz 1988). Expression of somatostatin receptors SSR1-3 has been observed in this nucleus in various studies (Señarís et al. 1994, Pérez et al. 1994, Schindler et al. 1997). Somatostatin affects noradrenergic cells in the locus coeruleus by decreasing their electrical activity (Inoue et al. 1988). Noradrenergic locus coeruleus cells have a reciprocal efferent connection to hypothalamic somatostatin cells (Liposits 1993, Mounier et al. 1996), by which the locus coeruleus stimulates somatostatin release. Because somatostatin inhibits the locus coeruleus, and inhibition of the locus coeruleus is needed during REM sleep episodes, it is possible that the locus coeruleus mediates at least partly the effects of somatostatin on REM sleep (Fig. 15).
Fig. 15 Schematic
illustration of the reciprocal connection between hypothalamic somatostatin
cells and noradrenergic (NA) cells in the locus coeruleus (LC) and the
possible connection of somatostatin to the REM sleep regulatory system
in the brainstem (LDT: laterodorsal tegmental nucleus, PPT: pedunculopontine
tegmental nucleus, ACh: acetylcholine).
I.c.v. injection of somatostatin antagonist decreased the amount of spontaneous daytime REM sleep in rats and also of rebound REM sleep after 24 h REM sleep deprivation. Local microinjection of somatostatin antagonist into the locus coeruleus decreased the amount of spontaneous daytime REM sleep. The cells in the locus coeruleus are also capable of mediating the actions of i.c.v. somatostatin antagonist, because the nucleus is situated in the vicinity of the fourth ventricle. While somatostatin antagonist was effective, an i.c.v. or locus coeruleus microinjection of somatostatin did not affect the amount of daytime REM sleep. There are at least three possible causes which may explain this discrepancy between somatostatin and somatostatin antagonist. First, the half-life of endogenous somatostatin may be much shorter than the half-life of synthetic somatostatin antagonist. The fact that in an earlier study (Danguir 1986) an increased amount of REM sleep was found during continuous long-term infusion of somatostatin supports this hypothesis. However, in another study (Lin et al. 1991) it was found that the duration of the effect of somatostatin and somatostatin antagonist on the baroreceptor reflex was almost equal. The second possibility is the ceiling effect. During the first half of the light phase, when the amount of sleep is highest in the rats, inhibition of the locus coeruleus caused by endogenous somatostatin may be strong enough, so that somatostatin injection is ineffective to cause any further inhibition to increase the amount of REM sleep. In the study of Danguir (1986), the proportions of sleep phases were calculated in 24 h time periods, which does not reveal, whether the enhancement of REM sleep was concentrated in the light or dark phase. The third possibility would be different receptor affinity profiles of somatostatin and somatostatin antagonist. Periventricular somatostatin cells are autoinhibited through SSR1 receptors, which are located in the pericarya and axon terminals of these cells (Helboe et al. 1998). Due to this autoinhibiton, i.c.v. injection of somatostatin paradoxically increases GH secretion (Mounier et al. 1996). I.c.v. injected somatostatin inhibits hypophysiotropic somatostatin release, while the i.c.v. somatostatin is unable to affect somatotropes in the pituitary directly. Somatostatin terminals in the locus coeruleus may also be autoinhibited by SSR1 and thus either i.c.v. or local locus coeruleus microinjection of somatostatin may inhibit the release of endogenous somatostatin, which may blunt the postsynaptic effect of somatostatin mediated by SSR2 or 3. Somatostatin antagonist may have higher affinity on postsynaptic somatostatin receptors in the locus coeruleus. It is as well possible that the somatostatin antagonist which we used has also some antagonist effect on VIP receptors (Chen et al. 1993). Microinjection of VIP into the oral pontine tegmentum, close to the locus coeruleus, has been found to increase REM sleep (Bourgin et al. 1999). As mentioned earlier, i.c.v. injection of somatostatin paradoxically increases GH secretion (Mounier et al. 1996). According to this, one could expect that the effect of i.c.v. injected somatostatin on REM sleep is possibly explained by indirect enhancement of GH secretion. However, this hypothesis is not supported by the finding that i.p. injection of the somatostatin analog octreotide increases REM sleep, while it decreases GH secretion by direct inhibition of pituitary somatotropes (Beranek et al. 1997). Novel, more potent and specific somatostatin antagonists have been synthesized (Bass et al. 1996, Hocart et al. 1998), but they are not yet commercially available. These compounds may reveal the identity of the receptor responsible for the effect somatostatin on REM sleep.
REM sleep deprivation for 24 h increased the amount of galanin mRNA in the periventricular nucleus and the median preoptic area, while 6 or 12 h total sleep deprivation did not affect the amount of galanin mRNA in the same regions (Toppila et al. 1996). The galanin cells in the medial preoptic area, which are sensitive to REM sleep deprivation are intermingled with or in the immediate vicinity of the GABA and cholinergic cells which regulate cortical and thalamic activity and the state of vigilance (Szymusiak 1995). Galanin is usually co-expressed in the cells with another peptide or a non-peptide transmitter (Bartfai et al. 1993). The co-transmitter identity of the REM sleep deprivation sensitive cells is not clear. I.c.v. injection of galanin increases GH secretion (Melander et al. 1987), which is possibly mediated by inhibition of somatostatin cells in the periventricular nucleus, which in turn reverses the inhibition of GHRH and GH release. This may be one mechanism which could mediate the hypothetical effects of galanin on sleep. However, we found that i.c.v. injection of galanin at doses which should affect GH release did not affect day or night-time sleep in the rats. The periventricular nucleus is connected reciprocally with the locus coeruleus (Liposits 1993), where galanin receptors in noradrenergic cells decrease the activity of these cells (Seutin et al. 1989). During REM sleep deprivation continuous and/or increased activity of the locus coeruleus may activate galanin cells in the periventricular nucleus. However, galanin microinjections into the locus coeruleus did not affect the amount of daytime REM sleep in rats either alone or when combined with somatostatin. It is possible that these negative findings are caused by the short half-life of these peptides, or by the ceiling effect during daytime, when the amount of sleep is high in the rats anyway. However, these findings together with the data of i.c.v. injections suggests that galanin does not affect sleep when injected i.c.v. or locally into the locus coeruleus. In a quite recent study a high dose (10 nmol) of galanin i.c.v. decreased locomotor activity in rats, but the same effect was also obtained by an injection of galanin antagonist M-35 (Ericsson and Ahlenius 1999). The increased galanin mRNA in the medial preoptic area and the periventricular nucleus after REM sleep deprivation may mediate the effects of REM sleep deprivation to other neuronal systems, while it possibly has no direct feedback connections to sleep regulating systems. Galanin mRNA has been found to increase in response to various neuronal stimuli such as lesions and blockade of axonal transport by colchicine (Cortés et al. 1990a, Cortés et al. 1990b). REM sleep deprivation could be a stimulus which belongs to this category.
A certain amount of stress is inevitably associated with any kind of sleep deprivation procedure. In order to be able to correct for the stress caused during REM sleep deprivation by the platform method, control groups of rats were kept for the same time on large platforms. It is suggested that this treatment causes approximately the same amount of immobilization than the REM sleep deprivation treatment on the small platforms, while REM sleep is not reduced. The effect of REM sleep deprivation by the platform method on plasma corticosterone has been measured in our laboratory and there was no significant difference between the REM sleep deprivation and the large platform control groups during 8-72 h deprivation (Porkka-Heiskanen et al. 1995). Sleep deprivation by gentle handling has been regarded less stressful than the platform method and it does not affect blood corticosterone levels significantly (Murison et al. 1982)
Stress activates waking mechanisms in the brain, the same ones which can be assumed to be activated during sleep deprivation. The noradrenergic locus coeruleus is one important neuronal system supporting wakefulness and alertness in the brain. Exposure to REM sleep deprivation but also stress increases the amount of tyrosine hydroxylase mRNA in the locus coeruleus (Angulo et al. 1991, Porkka-Heiskanen et al. 1995, Basheer et al. 1998). On the contrary, total sleep deprivation produced by the same protocol as in the somatostatin, GHRH, and galanin mRNA studies did not affect tyrosine hydroxylase mRNA in the locus coeruleus (Alanko et al. 1996). After stress exposure similar REM sleep rebound in the REM sleep amounts has been found as after REM sleep deprivation (Rampin et al. 1991). It is well known that all kinds of stress exposure activate the hypothalamo-pituitary-adrenal axis which leads to increased levels of glucocorticoids in the plasma by increased release of CRH and ACTH. There is possibly a similar association between wakefulness and hypothalamic CRH release as between SWS and GHRH (Steiger and Holsboer 1997, Chang and Opp 1998). I.c.v. injection of CRH increases wakefulness and decreases SWS, while i.c.v CRH antagonist decreases wakefulness, increases SWS and has no effect on REM sleep (Ehlers et al. 1986, Chang and Opp 1998).
Stress affects also the GH axis. Acute and repeated stress caused by immobilization
increases somatostatin release in the median eminence (Benyassi
et al. 1992, Benyassi
et al. 1993) and decreases plasma GH (Marti
et al. 1996, Marti et al.
1993). However, in humans several kinds of stressful conditions have
been found to increase plasma GH (reviewed by Reichlin
1987). During acute stress the GHRH release in the median eminence
is decreased with a slight delay, which possibly indicates that this effect
is indirect and mediated by somatostatin (Aguila
et al. 1991). Sleep deprivation and stressful situations are usually
related to increased wakefulness, when the noradrenergic system in the
locus coeruleus is activated (Fig. 15). This may stimulate
hypothalamic somatostatin cells which in turn inhibit GHRH cells in the
hypothalamus and GH release from the pituitary. This hypothesis is supported
by the finding that the ß-adrenergic antagonist propranolol reverses
the blunted GH response to GHRH caused by glucocorticoid injections (Devesa
et al. 1995). Reciprocally, hypothalamic somatostatin may inhibit
noradrenergic cells in the locus coeruleus by a feedback mechanism, but
if the stimulus supporting wakefulness is strong enough, it will overcome
the inhibitory effect of somatostatin, while the inhibition may make the
maintaining of prolonged wakefulness more difficult. When the stimulus
maintaining wakefulness is over, an activated somatostatin system may promote