M ATERIALS AND METHODS

Animals and housing

In the present study, we used young male Wistar rats (n=92) weighing ± 250 g at the start of the experiments (purchased from Harlan, Horst, The Netherlands). Animals were housed under a 12h light/12h dark cycle, with lights on from 09.00 h to 21.00 h. The temperature of the animal room was maintained at 21 ± 1^oC. Rats were provided with food and water ad libitum throughout the experiments. The experiments were approved by the Animal Experimentation Committee of the University of Groningen.

Sleep restriction and forced activity

The sleep restriction protocol allowed the rats to sleep 4h per day at the beginning of the light phase (09.00-13.00 h) in their home cage (Meerlo et al., 2002; Roman et al., 2005). The remainder of the time, animals were kept awake by placing them in slowly rotating wheels (40 cm in diameter) driven by an engine at constant speed (0.4 m/min). Since the sleep deprivation procedure includes mild forced locomotion, we used forced activity control rats to test whether effects of sleep restriction might be due to forced activity rather then sleep loss per se. Animals of the forced activity group were placed in the same plastic drums as the ones that were used for sleep restriction. However, these wheels rotated at double speed (0.8 m/min) for half the time (10h). With this protocol, rats walked the same distance as sleep restricted ones, but had sufficient time for sleep (14h). Animals were subjected to forced activity during the last 10 h of the dark phase which is their circadian activity phase. Before starting the experiments, rats were habituated to the experimental apparatus by placing them in the wheels for 1 h over 3 days. The two treatments, sleep restriction and forced activity, are similar in the sense that both involve forced locomotion, however, they differ in the time available for sleep and the amount of stress (Roman et al., 2005). In other words, the sleep restriction protocol is characterized by major sleep loss accompanied by little stress, and the forced activity schedule by little sleep loss but higher stress levels.

Radiotelemetry of body temperature

It has been shown that central serotonin-1A and muscarinic cholinergic neurotransmission can be assessed in vivo by measuring hypothermic body temperature responses after the administration of agonists of the serotonin (Hjorth, 1985) and acetylcholine receptor (Lomax and Jenden, 1966; Ryan et al., 1996), respectively. In the present study, we applied radio telemetry to record body temperature. For the implantation of radio transmitters (model TA10TA-F40; Data Sciences Inc., St. Paul, USA), anaesthesia was induced and maintained by inhalation of a mixture of N2O, O2, and isoflurane. Transmitters were placed in the abdominal cavity. After surgery, animals had at least 10 days to recover. The transmitters measured core body temperature and transformed temperature values into frequency coded radio signals. Receivers (model RPC-1; Data Sciences) placed under home cages relayed the radio signals to a PC. Body temperature data were sampled for 5 sec once every 5 min with the Dataquest Labpro^TM system (Data Sciences).

Adrenalectomy

In order to eliminate endogenous adrenal stress hormone production, bilateral adrenalectomy was performed (Meyer et al., 1979). Surgeries were carried out under anaesthesia induced and maintained by inhalation of a mixture of N2O, O2, and isoflurane. After removing the adrenals, a 100 mg pellet containing 25 mg corticosterone and 75 mg cholesterol (both from Sigma, St. Louis, MO, USA) was implanted under the skin.

These slow-release pellets result in low and stable levels of corticosterone for several weeks (Meyer et al., 1979). The pellet was implanted to assure basal levels of corticosterone and avoid the negative biological effects that have been reported for a total lack of endogenous corticosterone (De Kloet et al., 1986; Le Corre et al., 1997; MacLennan et al., 1998; Wossink et al., 2001). Next to ad libitum tap water, rats also had free access to a 0.9% physiological saline solution, allowing them to compensate for reduced salt retention that sometimes occurs in adrenalectomized animals (Wright, 1973).

Pharmacological challenges

To test serotonin-1A receptor sensitivity, we measured the hypothermic response to a 1A agonist (Hjorth, 1985). Animals received subcutaneous (s.c.) injections of the agonist 8-OH-DPAT (0.25 mg/kg body weight;

Sigma), which has been shown to induce a hypothermic response by acting on postsynaptic serotonin-1A receptors (Bill et al., 1991). The pharmacological challenges were performed between 11.30 h and 12.00 h, during the third hour of the light phase, when animals were in their home cage after 1 and 7 days of partial sleep deprivation. The sensitivity to the drug was determined by recording the acute hypothermic response to the injection (Roman et al., 2005).

In order to examine muscarinic acetylcholine receptor sensitivity, we measured the hypothermic response to the muscarinic agonist oxotremorine (Lomax and Jenden, 1966; Ryan et al., 1996). Rats received s.c. cocktail injections of oxotremorine (1-[4-(1-pyrrolidinyl)-2-butynyl]-2-pyrrolidinone, sesquifumarate-rate salt; 0.25 mg/kg body weight; Sigma) and methyl scopolamine (hyoscine methyl nitrate; 1.0 mg/kg body weight; Sigma). Methyl scopolamine is a potent muscarinic antagonist which, due to its methyl group, does not pass the blood brain barrier, and which was added to selectively inhibit peripheral muscarinic receptor-mediated responses (Smith et al., 2001). Methyl scopolamine by itself does not significantly change body temperature (Smith et al., 2001; Meerlo et al., unpublished results). Therefore, the physiological response to a cocktail injection of oxotremorine and methyl scopolamine represents the central effects of oxotremorine alone. Similar to the serotonergic challenges, muscarinic challenges were performed between 11.30 h and 12.00 h, during the third hour of the light phase, when the rats were in their home cage after 2 and 8 days of partial sleep deprivation.

Blood sampling and corticosterone measurements

To establish the effect of sleep restriction and forced activity as well as the effect of adrenalectomy on plasma corticosterone levels, blood samples were collected by tail bleeding (Meerlo et al., 2002). Samples were taken before starting the experimental treatment (13.00 h), after the first 20h of sleep deprivation (09.00 h) and after the 4h rest that followed this (13.00 h), as well as after 6 d of partial sleep deprivation at the end of the sleep deprivation period (09.00 h) and after the daily rest period (13.00 h). Blood samples were collected in pre-chilled Eppendorf tubes containing EDTA as anticoagulant. Then, blood was centrifuged at 2600 g for 15 min and the supernatant was stored at -80^oC until radioimmunoassay measurements of corticosterone (ICN Biomedicals, Costa Mesa, CA, USA).

Data analysis and statistics

In the case of the serotonergic challenges, body temperature data from 30 min preceding the injection until 90 min after the injection were used. For the more prolonged oxotremorine-induced hypothermic response data from 30 min preceding the injection until 240 min after the injection were processed. For each individual rat, delta body temperature values were calculated by expressing body temperature changes after the injection relative to baseline, that is, the average body temperature of the 30-min period preceding the injection. Delta body temperature data of the serotonergic challenges and muscarinic challenges, as well as plasma corticosterone data were subjected to analysis of variance with repeated measures (ANOVA) and post hoc Tukey test when appropriate.

R

Serotonergic challenges

In order to estimate the sensitivity of the serotonin-1A receptor system, the 1A receptor agonist 8-OH DPAT was injected in sleep restricted and control rats. The injection of 8-8-OH-DPAT induced an acute hypothermic response with a duration of approximately 90 min. In line with our previous study, the results show that one day of sleep restriction did not significantly change this hypothermic response (Fig. 1A), but seven days of sleep restriction significantly attenuated the drop in body temperature (F(36,378)=2.512, p<0.001) (Fig. 1B).

In order to test whether this reduction of serotonin-1A receptor sensitivity is mediated by adrenal stress hormones, 8-OH-DPAT was injected in sleep restricted and home cage control rats with or without adrenalectomy (Fig. 2). In this experiment, forced activity control rats were omitted since they did not exhibit a change in serotonin-1A receptor sensitivity (Roman et al., 2005). Two-way ANOVA with repeated measures showed that seven days of sleep restriction significantly desensitized the 1A receptors (factor sleep restriction: F(18, 468)=23.811; p<0.001), which was not affected by adrenalectomy (factor adrenalectomy: not significant).

Cholinergic challenges

To measure the sensitivity of the acetylcholine muscarinic receptor system, the muscarinic agonist oxotremorine was injected in sleep restricted and control rats. The injection of oxotremorine resulted in a hypothermic response, which lasted for approximately 3h. There were no significant differences between the groups after 2 days of restricted sleep or forced activity (Fig 3A). However, ANOVA revealed a significant treatment effect after 8 days (F(96,432)=4.11, p<0.001) (Fig. 3B). The Tukey test showed that the hypothermic response of the forced activity group was significantly stronger than that of the home cage controls (P=0.017). The sensitivity of the muscarinic system in the sleep restriction group after the 8-day treatment was not significantly different from either that of the home cage controls or the forced activity controls. This suggests that sleep loss per se does not have an effect but that forced activity at a higher intensity causes a hypersensitization of the cholinergic muscarinic receptor system.

Figure 1. Serotonin-1A receptor agonist-induced hypothermic response after 1 and 7 days of sleep restriction. Two-and-a-half hours after the last sleep deprivation episode, at time 0, rats (n=6 per group) received an injection of 8-OH-DPAT (0.25 mg/kg), which resulted in an immediate drop in body temperature. [A] One day of sleep restriction did not significantly change the sensitivity of the serotonin-1A receptor system. [B]

After 7 days of sleep restriction, the sensitivity of the serotonin-1A receptor system was significantly reduced. Neither 1 nor 7 days of forced locomotion changed the sensitivity of the receptor system.

Figure 2. Serotonin-1A receptor agonist-induced hypothermic response after 7 days of sleep restriction in adrenalectomized and sham-operated animals (n=8 per group). Seven days of sleep restriction resulted in a significantly attenuated serotonin-1A receptor function.

This significant decrease in serotonergic sensitivity was not abolished by adrenalectomy, demonstrating the stress hormone-independence of the desensitization.

Receptor sensitivity was unaffected by adrenalectomy in home cage controls.

To examine whether the hypersensitivity of the muscarinic receptor system is related to the stressful nature of forced activity, oxotremorine was injected in adrenalectomized and intact rats, which were exposed to a forced locomotion schedule or were left in their home cage undisturbed.

ANOVA with repeated measures revealed a significant interaction between forced physical activity and adrenalectomy (F(48,1584)=4.301; p<0.001). The post hoc Tukey test showed a significant difference between the hypothermic response of intact home cage controls and intact forced activity rats (p=0.019); and a strong tendency for significant difference between intact and adrenalectomized forced activity rats (p=0.06) (Fig. 4).

Figure 3. The muscarinic cholinergic agonist oxotremorine-induced hypothermic response after 2 and 8 days of sleep restriction and forced activity.

Oxotremorine (0.25 mg/kg) was injected in the rats (n=6 per group) 2½ h after the end of the last sleep deprivation/forced activity block at time 0. [A] After 2 days of sleep restriction, hypothermic responses are the same in the three experimental groups. [B] Eight days of forced activity significantly increased the sensitivity of muscarinic receptors. Eight days of sleep restriction resulted in a slightly, but non-significantly enhanced drop in body temperature.

Figure 4. The muscarinic acetylcholine receptor agonist-induced hypothermic response after 8 days of forced locomotion in adrenalectomized and sham-operated animals (n=7-12 per group). The increased sensitivity of the muscarinic receptors caused by forced activity disappeared in adrenalectomized animals. Receptor sensitivity was unaffected by adrenalectomy in home cage controls.

Plasma corticosterone levels

In order to assess stress levels and to validate the surgical removal of the adrenal glands, corticosterone was measured in plasma samples taken at baseline, and at further time points during the experimental treatment, that is, after 1 and 6 days of treatment at the end of the sleep deprivation and forced activity period as well as after the 4h rest that followed it (Fig. 5). Two-way ANOVA with repeated measures showed a significant effect of experimental treatment (F(8,200)=18.98; p<0.001) and adrenalectomy (F(4,200)=43.632; p<0.001). Both after 1 and 6 days, corticosterone levels immediately after the daily treatment were significantly elevated in the forced activity control group, both compared to home cage controls and sleep restricted animals (day 1, p<0.001 and p<0.001, respectively; day 6, p=0.001 and p=0.003, respectively). Corticosterone levels in sleep restricted animals were slightly but still significantly higher than levels in the home cage control rats on both days (day 1, p=0.005; day 6, p<0.001). Adrenalectomy prevented the elevation of corticosterone and the replacement pellet resulted in low basal corticosterone levels that were indistinguishable from the levels in undisturbed home cage controls.

Figure 5. Plasma corticosterone levels (μg/dl) at baseline, after 1 and 6 days of treatment at the end of sleep deprivation/forced locomotion (09.00 h, indicated as 9 on the graph) and 4 h later, at the end of rest (13.00 h, indicated as 13 on the graph). Hormone levels were determined by radioimmunoassay. Cortico-sterone levels were significantly higher in sleep restricted rats (n=8) than in adrenalectomized sleep restricted (n=8) as well as adrenalectomized and sham-operated home cage controls (n=12 each) at the end of sleep deprivation on day 1 and 6. Corticosterone plasma levels were also significantly higher in the forced activity group (n=8) in comparison to sleep restricted rats and adrenalectomized animals in the forced activity group (n=8).

Adrenalectomized animals show moderate levels of corticosterone indicating that indeed the s.c.

corticosterone pellets ensured constant plasma levels of the hormone throughout the experiment.

Abbreviations: ADX, adrenalectomy; B, baseline. Significant differences: ADX vs. sham operated (*); sleep restriction vs. home cage (a); sleep restriction vs. forced activity (b); forced activity vs. home cage (c).

D

The present study confirms that chronic partial sleep deprivation gradually desensitizes the serotonin-1A system. This effect was still present in adrenalectomized animals, indicating that it is not mediated by adrenal stress hormones. Also, when animals were subjected to forced activity of a higher intensity with more time to sleep, these changes were not seen, supporting the conclusion that the sleep restriction-induced serotonin-1A desensitization was mediated neither by forced activity nor by stress. In contrast, not sleep restriction but forced locomotion was associated with a gradual hypersensitization of the muscarinic cholinergic receptor system. This effect may be a consequence of stress associated with the forced locomotion since it was not seen in animals after removal of the adrenals. Taken together, these data suggest that sleep restriction leads to a down-regulation of serotonin-1A signalling, while stress, on the other hand, results in an up-down-regulation of muscarinic cholinergic signalling.

The present study supports our previous finding of a gradually desensitizing serotonin-1A receptor system in the condition of chronic partial sleep deprivation (Roman et al., 2005). It thus seems that sleep, one way or another, helps to maintain the sensitivity of the serotonin-1A receptor system. These data are in line with the hypothesis that sleep serves to up-regulate monoaminergic systems that are tonically active during waking. Indeed, hypothetically, the sleep restriction-induced desensitization might be explained by a direct effect of serotonin itself, i.e. a sleep loss-induced chronically and intermittently higher serotonin load onto the serotonin-1A receptors. This hypothesis is supported by microdialysis studies which have shown that the release of serotonin during wakefulness and sleep deprivation is higher than during sleep (Lopez-Rodriguez et al., 2003; Penalva et al., 2003). It is a common phenomenon that continuous stimulation of receptors gradually reduces their functional sensitivity. Indeed, it has been shown that in serotonin transporter knock-out mice with tonically increased extracellular serotonin levels, the serotonin-1A receptor-mediated temperature and neuroendocrine responses are reduced (Li et al., 1999). Thus, chronic sleep restriction may be a condition with elevated levels of serotonin which, in the long run, may be responsible for postsynaptic receptor desensitization. A loss of functional receptor sensitivity can be the result of various mechanisms, including receptor phosphorylation-induced G-protein uncoupling, receptor internalization and down-regulation by increased G-protein degradation, or reduced receptor mRNA and protein synthesis (Ferguson and Caron, 1998). In addition, it may also be that the attenuated serotonin-1A response found in our study is a consequence of changes in elements of signal transduction cascades downstream to the receptor itself.

In recent years it has been shown that 8-OH-DPAT is an agonist, not only of the serotonin-1A receptor, but the serotonin-7 receptor as well (Hedlund et al., 2004). It is also known that the serotonin-7 receptor is expressed at high levels within the hypothalamus and is involved in the regulation of body temperature (Barnes and Sharp, 1999; Hedlund and Sutcliffe, 2004). A recent paper has shown that the 8-OH-DPAT-induced hypothermia is a result of combined serotonin-1A and serotonin-7 receptor stimulation, while the 8-OH-DPAT-induced ACTH release is purely serotonin-1A receptor-dependent (Faure et al., 2006). Recent studies in our laboratory indicate that

chronic partial sleep deprivation not only reduces the 8-OH-DPAT-induced hypothermic response but also the ACTH response to this agonist (Roman and Meerlo, unpublished data). This finding supports that the sleep loss-related reduction of the hypothermic response presented here is at least partly due to changes at the serotonin-1A receptors. However, the involvement of alterations in serotonin-7 receptors cannot be excluded and further experiments are needed to elucidate the role of this particular receptor subtype.

The body temperature response that we used as a read out for serotonin-1A receptor sensitivity in the present study is known to be mediated by postsynaptic receptors, thus representing the functional sensitivity of 1A receptors in the target areas of the serotonergic system (Bill et al., 1991; O’Connell et al., 1992). We did not examine the presynaptic serotonin-1A autoreceptors but a number of other studies in rodents have shown that sleep deprivation may lead to a desensitization of the 1A autoreceptors as well (Evrard et al., 2006; Prevot et al., 1996).

However, this autoreceptor desensitization is a rapid effect that occurs within one day of sleep deprivation and normalizes within a few hours of recovery sleep (Prevot et al., 1996). This is in contrast to the much slower desensitization of the postsynaptic receptors that only develops gradually over a week of sleep restriction and also persists for many days thereafter, even with unlimited recovery sleep (Roman et al., 2005). Importantly, whereas the rapid desensitization of the presynaptic 1A autoreceptors with acute sleep deprivation appears to be mediated by glucocorticoids (Evrard et al., 2006), the gradual desensitization of the postsynaptic 1A receptors in the present study was independent of adrenal stress hormones. In other words, sleep loss may lead to a desensitization of the presynaptic serotonin-1A autoreceptor rather quickly, but only when sleep deprivation is associated with elevated levels of glucocorticoids (Evrard et al., 2006); while sleep loss may lead to desensitization of the postsynaptic serotonin-1A receptors independent of stress hormones, but only when sleep restriction is chronic (Roman et al., 2005).

These different dynamics of sleep loss-induced changes in pre and postsynaptic serotonin 1A receptors may explain why sleep restriction can have such seemingly paradoxical effects on the regulation of mood. Specifically, while chronically disrupted and restricted sleep is thought to be a factor that contributes to the development of mood disorders (Riemann and Voderholzer, 2003), once an individual is depressed acute sleep deprivation may improve mood and temporarily reverse depression (Wirz-Justice and Van den Hoofdakker, 1999). On the basis of our results, we hypothesize that chronically disrupted or restricted sleep may slowly sensitize an individual to depression by gradually desensitizing the postsynaptic serotonin 1A receptor system and thereby decreasing serotonergic transmission (Roman et al., 2005). In contrast, once an individual is depressed, the therapeutic effect of an acute one-night sleep deprivation may be related to a rapid decrease in the sensitivity of presynaptic serotonin-1A autoreceptors, thereby reducing the auto-inhibition of the serotonergic neurons and increasing serotonin release in the target areas (Evrard et al., 2006; Prevot et al., 1996).

In the present study, the adrenal stress hormone independence of the postsynaptic serotonin-1A receptor desensitization in chronically sleep restricted animals was proven by the fact

desensitization in the forced activity control group, despite the fact that animals in this group had significantly higher corticosterone levels. The latter seems in contrast with a number of other studies, which have shown that stress and high levels of glucocorticoids are capable of desensitizing the serotonin-1A receptor system (Karten et al., 1999; Meijer and De Kloet, 1994).

This apparent contradiction may be due to the fact that even our forced activity control group had only moderate levels of corticosterone (26-28 μg/dl) compared to levels after severe stress and levels in some of the studies that reported serotonin-1A receptor desensitization (e.g., 70-80 μg/dl

This apparent contradiction may be due to the fact that even our forced activity control group had only moderate levels of corticosterone (26-28 μg/dl) compared to levels after severe stress and levels in some of the studies that reported serotonin-1A receptor desensitization (e.g., 70-80 μg/dl