DISINHIBITION OF VENTROLATERAL PREOPTIC AREA SLEEP-ACTIVE NEURONS BY ADENOSINE: A NEW MECHANISM FOR SLEEP PROMOTION

Transcription

1 Neuroscience 123 (2004) DISINHIBITION OF VENTROLATERAL PREOPTIC AREA SLEEP-ACTIVE NEURONS BY ADENOSINE: A NEW MECHANISM FOR SLEEP PROMOTION S. MORAIRTY, a * D. RAINNIE, b R. MCCARLEY c AND R. GREENE d a SRI International, Menlo Park, CA, USA b Department of Psychiatry, Emory University, Atlanta, GA, USA c Department of Psychiatry, Harvard University, VAMC, Brockton, MA, USA d Department of Psychiatry, University of Texas, Southwestern, VAMC, Dallas, TX, USA Abstract The ventrolateral preoptic area of the hypothalamus (VLPO) contains a population of sleep-active neurons and is hypothesized to be an important part of the somnogenic process. Adenosine (AD) is an endogenous sleep-promoting factor and may play an important role in promoting natural sleep. We hypothesize that AD may promote sleep, in part, by activating the VLPO sleep-active neurons. Although, in the CNS, AD is generally regarded as an inhibitory neuromodulator, it is possible for AD to be directly excitatory via A2 receptors or indirectly via disinhibition. In order to test the hypotheses that AD can excite VLPO neurons we made intracellular recordings from the VLPO in vitro and examined the effects of AD on VLPO neural activity. Whole cell patch-clamp recordings were obtained from rat brain slices and drugs were bath applied. VLPO neurons were electrophysiologically heterogeneous. Depolarizing current steps elicited rhythmic firing (25 of 57), spike frequency adaptation or accommodation (24 of 57), or an unusual burst firing response (eight of 57). Spontaneous synaptic activity was pronounced in most recorded neurons and consisted of either fast excitatory post-synaptic potentials/currents (EPSP/C s) and/or fast inhibitory postsynaptic potentials/currents (IPSP/C s). The IPSC s were fully blocked by 30 M bicuculline suggesting they are GABA A -mediated events, and the EPSC s were blocked by 40 M DNQX suggesting they are mediated by the AMPA subtype of glutamate receptor (five of five). AD ( M) reduced the frequency of spontaneous IPSC s in 11 of 17 VLPO neurons (28 100%; mean reduction 63%) without significant effects on resting membrane potential. IPSC was unaffected in five neurons and one neuron displayed increases in spontaneous IPSC s. In contrast, AD decreased EPSC frequency in seven cells (36 73%; mean 59%), increased frequency in five cells (30 236%; mean 83%) and had no effect in six cells. AD application *Correspondence to: S. Morairty, 333 Ravenswood Avenue, , Menlo Park, CA 94025, USA. Tel: ; fax: address: stephen.morairty@sri.com (S. Morairty). Abbreviations: acsf, artificial cerebrospinal fluid; AD, adenosine; EPSP/C, excitatory post-synaptic potential/current; IPSP/C, inhibitory post-synaptic potential/current; LDT/PPT, the lateral dorsal tegmentum/pedunculopontine; LTS, low-threshold spike; POAH, preoptic/anterior region of the hypothalamus; SFA, spike frequency adaptation; VLPO, ventrolateral preoptic area of the hypothalamus; V m, membrane potential /04$ Published by Elsevier Ltd on behalf of IBRO. doi: /j.neuroscience increased the firing rate in two of four cells tested. These data are consistent with the hypothesis that one mechanism which AD may promote sleep is by blocking inhibitory inputs on VLPO sleep-active neurons Published by Elsevier Ltd on behalf of IBRO. Key words: rat, brain slice, whole-cell patch clamp, GABA. Adenosine (AD) has been implicated as an endogenous fatigue or sleep-promoting factor (Radulovacki et al., 1984; Rainnie et al., 1994; Bennington and Heller, 1995; Strecker et al., 2000). Microinjections of AD into the preoptic area of rats (Ticho and Radulovacki, 1991), and microdialysis into the lateral dorsal tegmentum/pedunculopontine (LDT/PPT) and basal forebrain (Portas et al., 1996, 1997) in cats, produce significant increases in sleep. Endogenous AD increases with sleep deprivation and decreases during sleep in the basal forebrain of cats (Porkka-Heiskanen et al., 1997). In addition, basal forebrain perfusion of nitrobenzylthioionosine, an AD transport blocker, caused increased extracellular AD in conjunction with increased sleep (Porkka-Heiskanen et al., 1997). In vitro, AD inhibits LDT/PPT cholinergic neurons and basal forebrain neurons (Rainnie et al., 1994). The decrease in excitability is mediated by AD A1 receptors acting at both post-synaptic (Rainnie et al., 1994) and pre-synaptic sites (Arrigoni et al., 2001). These data suggest that AD may affect behavioral state by inhibiting arousal systems. The preoptic/anterior region of the hypothalamus (POAH) has been hypothesized to be an important somnogenic center in the CNS (von Economo, 1930; Nauta, 1946; McGinty and Szymusiak, 1990). The POAH contains a population of sleep-active neurons whose firing rates are lowest during active waking and highest during NREM sleep (Findlay and Hayward, 1969; Kaitin, 1984; Szymusiak and McGinty, 1986, 1989; Osaka and Hayaishi, 1995; Koyama and Hayaishi, 1994; Alam et al., 1995, 1996). These neurons comprise about 25% of the POAH cells seen in vivo. These neurons are dispersed throughout a heterogeneous population of cells, and the neurotransmitter content of these sleep-active neurons is not known. Consequently, the electrophysiological and neurochemical properties of these neurons had not been specifically studied. Recently, a cluster of neurons in the ventrolateral preoptic area of the hypothalamus (VLPO) has been found to express c-fos in association with sleep (Sherin et al., 1996). Expression of c-fos is often associated with in- 451

2 452 S. Morairty et al. / Neuroscience 123 (2004) creased neuronal activity. Therefore, the VLPO has been hypothesized to contain a cluster of sleep-active neurons. This hypothesis has been confirmed using in vivo unit recordings (Szymusiak et al., 1998). In addition, ibotenic acid induced lesions of the VLPO produced significant decreases in non-rapid eye movement sleep where the extent of the sleep loss was linearly related to the number of remaining Fos immunoreactive cells (Lu et al., 2000). A recent in vitro electrophysiological study in the VLPO has reported that 68% of neurons display a low-threshold spike (LTS) in response to depolarizing current injection (Gallopin et al., 2000). Moreover, the majority of LTS cells are inhibited by arousal related neurotransmitters such as noradrenalin, serotonin, and acetylcholine. These LTS cells were also more likely to contain the inhibitory neurotransmitter GABA. In this study we investigated the effects of the sleeppromoting factor AD on the excitability of VLPO neurons. We hypothesized that if AD were to promote sleep by an action in the VLPO sleep center, it would be more likely to be associated with an increased excitability of the sleep active VLPO neurons. However, in the CNS, AD is generally regarded as an inhibitory neuromodulator, primarily through activation of A1 receptors (Dunwiddie, 1985; Greene and Haas, 1991). Nonetheless, A2 receptor activation is linked to increased intracellular camp concentrations (Dunwiddie, 1985; Olah and Stiles, 1995), so a direct excitatory effect remained a possibility. In addition, AD has been reported to inhibit the release of GABA in SCN and arcuate nucleus cell cultures (Chen and van den Pol, 1997). Thus, a disinhibition of VLPO neurons by AD may also be a possible mechanism for sleep promotion. In order to test these hypotheses we made intracellular recordings from VLPO neurons in vitro and examined the effects of AD on VLPO neuron activity. EXPERIMENTAL PROCEDURES All experimental protocols were approved by the Harvard Medical School and the Brockton Veterans Administration Medical Center animal care and use committees. All efforts were made to minimize animal suffering or discomfort and to reduce the number of animals used. Six to eight week old ( g) hooded Long-Evans rats were lightly anesthetized with isoflurane and decapitated. The brains were rapidly removed and placed in ice cold artificial cerebrospinal fluid (acsf) bubbled with 95% O 2 /5% CO 2. Horizontal sections (400 m) were cut on a vibrating tissue slicer (model OTS 3000; Electron Microscopy Sciences, Fort Washington, PA, USA) and incubated at room temperature for 1 h. A slice containing the VLPO was placed in a recording chamber and perfused with acsf at a rate of 1.5 ml/min. The temperature of the acsf bathing the slice was gradually raised to 36.5 ( 0.5) o C. The acsf contained (in mm) NaCl 124, KCl 2, KH 2 PO 4 3, MgCl 2 1.3, CaCl 2 2.5, glucose 10, and NaHCO The patch electrode solution contained (in mm) K-gluconate 120, KCl 10, MgCl 2 3, HEPES 10, K 2 ATP 2, Na 2 GTP 0.2, and ph adjusted to 7.2 with 1 N KOH. Electrode placement for VLPO centered around the stereotaxic coordinates (from bregma) 0.5 mm AP, 1.0 mm L, and 9.0 mm VD. We considered VLPO to extend around this coordinate by a mm cube. Easily recognizable landmarks included the third ventricle and the optic chiasm. Whole cell patch-clamp recordings were made using an Axopatch 1D amplifier (Axon Instruments, Foster City, CA, USA). Patch pipettes were pulled on a Flamming/Brown horizontal puller (model P97; Sutter Instruments, Novato, CA, USA) to a resistance of 6 10 Mohms. Data were collected and analyzed using pclamp 6.2 software (Axon Instruments). AD, bicuculline and DNQX (Sigma, St. Louis, MO, USA) were bath applied. Typically, cells were patch-clamped in current clamp mode. For each neuron recorded, a series of hyperpolarizing, depolarizing and single spike protocols were performed to examine the basic membrane properties. Then, we switched to voltage clamp to study synaptic activity and the effects of drugs. Occasionally, we switched back to current clamp to observe the effects of AD on the firing rates of a neuron. Counts of spontaneous synaptic activity met a minimum criterion of at least a 2:1 signal to noise ration. A significant affect by AD met a minimum criterion of a 25% change in frequency of spontaneous synaptic events. Student s t-tests were performed where indicated. RESULTS Stable recordings were obtained from of 57 VLPO neurons. In agreement with immunohistochemical data, we found the VLPO to be an electrophysiologically heterogeneous population of neurons. Fourteen percent (eight of 57) of these neurons displayed an unusual burst firing response to depolarizing current steps from membrane potentials (V m ) just below the threshold for action potential generation ( 60 to 65 mv). The current responsible for this burst firing response is not likely to be I T (low threshold Ca -current) since 60 to 65 mv V m is not hyperpolarized enough to activate this current. The remaining 86% of the cells (49 of 57) displayed heterogeneous responses to depolarizing current steps including rhythmic firing (25 of 57), and spike frequency adaptation (SFA) or accommodation (24 of 57). For simplicity of description, neurons that displayed burst firing in respond to depolarizing current steps from V m just below firing potential will be referred to as type 1 neurons and all the other neurons will be referred to as type 2 neurons. Type 1 neurons displayed a significantly lower input resistance than type 2 neurons ( vs Mohms, respectively, P 0.04). Conversely, no significant difference was observed in the time constant for membrane charging ( vs ms, P 0.14). In many areas of the CNS inhibitory interneurons are typified by the rapid kinetics of their action potentials, when compared with projection neurons. However, no significant differences were observed in the spike properties of type 1 and type 2 neurons (rise time vs ms.; decay time vs ms.; half width vs ms). Most VLPO neurons (89%) displayed I h like currents (hyperpolarization-activated, time-dependent, inward rectifying current) and /or I t -like currents. Thus, these neurons could produce LTS when released from hyperpolarizing current steps ( 20 to 40 mv steps below a holding potential of 60 mv). Only eight of 56 units tested did not display either I t -like or I h -like currents. Of these neurons, four were units that displayed SFA in response to depolarizing current steps and two displayed rhythmic firing patterns in response to depolar-

3 S. Morairty et al. / Neuroscience 123 (2004) Fig. 1. The effects of bicuculline and DNQX on spontaneous PSC s. A. A voltage clamp recording form a neuron held at 55 mv under (Ai) baseline, (Aii) 40 M DNQX and (Aiii) under 40 M DNQX plus 30 M bicuculline. The traces on the right are enlargements of the indicated sections of the traces on the left. These data show that the GABA A antagonist bicuculline blocks spontaneous ipsc s and that the AMPA/kainate type glutamatergic antagonist DNQX blocks spontaneous epsc s. B. Voltage ramp protocols ( 100 to 40 mv) under voltage clamp conditions. Bi. 30 M bicuculline blocks all spontaneous ipsc s. Bii. 40 M DNQX blocks all spontaneous epsc s. izing current steps. All type 1 units displayed I t -like and/or I h -like currents. Spontaneous synaptic activity was pronounced in most recorded neurons and consisted of either fast excitatory post-synaptic potentials/currents (EPSP/C s) and/or fast inhibitory post-synaptic potentials/currents (IPSP/C s). The IPSC s were fully blocked by 30 M bicuculline suggesting they are GABA A -mediated events, and the EPSC s were blocked by 40 M DNQX suggesting they are mediated by the AMPA subtype of glutamate receptor (five of 5 s neuron tested; Fig. 1). Type 2 neurons displayed higher rates of spontaneous synaptic events than type 1 neurons. Spontaneous events for type 2 neurons were either primarily IPSP/C s or mixed IPSP/C s and EPSP/C s. The low level spontaneous synaptic events observed in type 1 neurons were primarily EPSP/C s. These results are similar to reports from recorded neurons in adjacent preoptic areas (Hoffman et al., 1994a,b). AD ( M) was applied to 22 VLPO neurons. AD reduces the frequency of spontaneous IPSC s in the majority of VLPO neurons in the absence post-synaptic effects. Hence, in those neurons that displayed spontaneous IPSC s (17 of 22) exogenous AD caused a significant reduction (28 100%; mean 63%) in the number of spontaneous IPSC s (Fig. 2). However, in one cell AD increased the frequency of IPSC s by 68%. In contrast, the effect of exogenous AD application on EPSC

4 454 S. Morairty et al. / Neuroscience 123 (2004) Fig. 2. The effects of AD on spontaneous PSC s. A. 30 M AD reduced the spontaneous IPSC s by 48% and increased the spontaneous EPSC s by 59%. The traces on the right are enlargements of the indicated sections of the traces on the left. B. AD (50 M) reduced the spontaneous IPSC s by 56% and did not affect the frequency of spontaneous EPSC s. The traces on the right are enlargements of the indicated sections of the traces on the left. C. AD (100 M) completely blocks the spontaneous IPSC s in this cell. This cell s unusual rhythmic pattern of spontaneous IPSC s was displayed during the entire recording period. D. Voltage ramps ( 100 to 40 mv) showing 50 M AD eliminating spontaneous IPSC s without affecting the frequency of spontaneous EPSC s.

5 S. Morairty et al. / Neuroscience 123 (2004) VLPO neurons. Consistent with the observed dis-inhibitory effects, AD application increased the firing rate in two of four cells tested (Fig. 3). One neuron was not spontaneously active under baseline conditions but became active with application of AD. The other neuron s spontaneous activity was increased almost two-fold with AD. In both cases the activity returned to baseline levels after washout. DISCUSSION AD may promote sleep by disinhibiting the sleepactive VLPO neurons Fig. 3. Top. Current clamp recording where 50 M AD increased firing rate. Bottom. Our working model on how AD helps promote sleep by disinhibiting the sleep-active VLPO neurons. frequency was much less predictable. Eighteen of the 22 neurons that received AD also displayed spontaneous EPSC s. However, of these 18 neurons the EPSC frequency was decreased in seven cells (36 73%; mean 59%), increased in five cells (30 236%; mean 83%) and was unaffected in six cells. Interestingly, all five of the above cells that expressed spontaneous EP- SCs but did not display spontaneous IPSC s were type 1 neurons. In addition, no relationship was found between the effect of AD on IPSC and EPSC frequency. For instance, of the 12 cells that displayed a decreased frequency of IPSC s in the presence of AD, the EPSC frequency was decreased in three cells, increased in three cells, unaffected in three cells, and were not expressed in the remaining three cells. Thus the most consistent effect of AD application in the VLPO is a reduction of a tonic inhibitory drive, which would tend to increase the basal level of excitability of the majority of The data presented here suggest that AD disinhibits a significant population of VLPO neurons by decreasing spontaneous IPSP/C s. By decreasing a tonic inhibitory tone, AD may allow for increases in neuronal activity. We found this to be the case in two of four neurons that we studied in both current clamp and voltage clamp conditions. If the sleep-active neurons of the VLPO are an important part of a somnogenic pathway, then AD may help facilitate sleep by disinhibiting these neurons. Although there is not a direct excitatory effect on V m, increasing AD levels during waking may help activate the VLPO sleep-active neurons by removing GABAergic inhibitory inputs. Fig. 3 shows a schematic of this hypothesis. Although our data support a role for AD in the control of behavioral state, they do not exclude a possible role for AD in the homeostatic regulation of sleep. The processes underlying the homeostatic regulation of sleep could occur in different brain regions than the VLPO or in processes that were not measured by our experimental conditions and techniques. Other studies are needed to investigate the role of AD in the homeostatic regulation of sleep. Recently, Gallopin et al. (2000) has shown that some VLPO neurons that are likely to be GABAergic are inhibited by the arousal related neurotransmitters noradrenalin, serotonin, and acetylcholine. Thus, they suggest that the sleep-active VLPO neurons may be inhibited during waking by these three neurotransmitter systems. Projections from the VLPO have been shown to innervate monoaminergic and cholinergic nuclei (Sherin et al., 1998). The reduction of a tonic GABAergic input by AD may allow for the VLPO sleep-active neurons to inhibit monoaminergic and cholinergic systems thus further disinhibiting them. This positive feedback loop may be an important system for sleep consolidation. The study by Gallopin et al. (2000) also reported finding two populations of cells with one type expressing LTS while the second type did not express LTS. They report no other electrophysiological differences. We have found a more electrophysiologically heterogeneous population of neurons in the VLPO. However, Gallopin et al. s (2000) LTS cell type likely corresponds to our neurons that display I t -like and/or I h -like currents. Both I t and I h can manifest low threshold spiking. The percentage of Gallopin et al. s (2000) LTS cell type is

6 456 S. Morairty et al. / Neuroscience 123 (2004) reasonably similar to our percentage of cells displaying I t -like and/or I h -like currents (68% vs. 89%). In addition, the inhibitory affects of AD on spontaneous IPSP/C s only occurred in cells expressing I t -like and/or I h -like currents. Cells not expressing I t -like and/or I h -like currents were unaffected by AD. It is interesting that all of the cells that displayed an inhibitory affect of AD on spontaneous IPSP/C s also expressed I t -like and/or I h -like currents. These currents may play a role in the transition between wakefulness and sleep. As discussed above, it is hypothesized that the sleep-active VLPO neurons are hyperpolarized during wakefulness through a combination of monoaminergic and GABAergic inputs. As rising AD levels block the GABAergic inputs, the I t -like and/or I h -like currents might allow small bursts of activity from the VLPO neurons. Since the sleep-active VLPO neurons are hypothesized to be GABAergic and have been shown to project to the brainstem monoaminergic and cholinergic nuclei, the bursts of activity from the VLPO neurons could inhibit the monoaminergic neurons that project to the VLPO thus further disinhibiting the sleep-active neurons. AD may be contributing to the control of behavioral state in two ways. First, AD may be directly inhibiting arousal systems such as the LDT/PPT and basal forebrain cholinergic systems. Secondly, AD may promote sleep by blocking inhibitory inputs onto sleep-active neurons. The first mechanism has support both in vivo (Portas et al., 1996, 1997; Porkka-Heiskanen et al., 1997) and in vitro (Rainnie et al., 1994; see Strecker et al., 2000 for review). The current study would support the second mechanism if the population of neurons that are disinhibited by AD could be shown to correspond to the population of VLPO sleepactive neurons. REFERENCES Alam MN, McGinty D, Szymusiak R (1995) Neuronal discharge of preoptic anterior hypothalamic thermoresponsive neurons: relation to nonrapid eye movement sleep. Am J Physiol 269: Alam MN, McGinty D, Szymusiak R (1996) Lateral preoptic area and diagonal band thermosensitive neurons in rats: relation to wakefulness and nonrapid eye movement sleep. Brain Res 718: Arrigoni E, Rainnie DG, McCarley RW, Greene RW (2001) Adenosinemediated presynaptic modulation of glutamatergic transmission in the laterodorsal tegmentum. J Neurosci 21: Bennington JH, Heller HC (1995) Restoration of brain energy metabolism as the function of sleep. Prog Neurobiol 45: Chen G, van den Pol AN (1997) Adenosine modulation of calcium currents and presynaptic inhibition of GABA release in suprachiasmatic and arcuate nucleus neurons. J Neurophysiol 77: Dunwiddie TV (1985) The physiological role of adenosine in the central nervous system. Int Rev Neurobiol 27: Findlay AR, Hayward JN (1969) Spontaneous activity of single neurons in the hypothalamus of rabbits during sleep and waking. J Physiol 201: Gallopin T, Fort P, Eggermann E, Cauli B, Luppi PH, Rossier J, Audinat E, Muhlethaler M, Serafin M (2000) Identification of sleeppromoting neurons in vitro. Nature 404: Greene RW, Haas HL (1991) The electrophysiology of adenosine in the mammalian central nervous system. Prog Neurobiol 36: Hoffman NW, Wuarin JP, Dudek FE (1994a) Whole-cell recordings of spontaneous synaptic currents in medial preoptic neurons from rat hypothalamic slices: mediation by amino acid neurotransmitters. Brain Res 660: Hoffman NW, Kim YI, Gorski RA, Dudek FE (1994b) Homogeneity of intracellular electrophysiological properties in different neuronal subtypes in medial preoptic slices containing the sexually dimorphic nucleus of the rat. J Comp Neurol 345: Kaitin KI (1984) Preoptic area unit activity during sleep and wakefulness in the cat. Exp Neurol 83: Koyama Y, Hayaishi O (1994) Modulation by prostaglandins of activity of sleep-related neurons in the preoptic/anterior hypothalamic areas of rats. Brain Res Bull 33: Lu J, Greco MA, Shiromani P, Saper CB (2000) Effect of lesions of the ventrolateral preoptic nucleus on NREM and REM sleep. J Neurosci 20: McGinty DJ, Szymusiak R (1990) Keeping cool: a hypothesis about the mechanisms and functions of slow wave sleep. Trends Neurosci 13: Nauta WJH (1946) Hypothalamic regulation of sleep in rats: an experimental study. J Neurophysiol 9: Olah ME, Stiles GL (1995) Adenosine receptor subtypes: characterization and therapeutic regulation. Annu Rev Pharmacol Toxicol 35: Osaka T, Hayaishi O (1995) Prostaglandin D2 modulates sleep-related and noradrenaline-induced activity of preoptic and basal forebrain neurons in the rat. Neurosci Res 23: Porkka-Heiskanen T, Strecker RE, Thakkar M, Bjørkum AA, Greene RW, McCarley RW (1997) Adenosine: a mediator of the sleepinducing effects of prolonged wakefulness. Science 276: Portas CM, Thakkar M, Rainnie DG, Greene RW, McCarley RW (1997) Role of adenosine in behavioral state modulation: a microdialysis study in the freely moving cat. Neuroscience 79: Portas CM, Thakkar M, Rainnie DG, McCarley RW (1996) Microdialysis perfusion of 8-OH-DPAT in the dorsal raphe nucleus decreases serotonin release and increases REM sleep in freely moving cat. J Neurosci 16: Radulovacki M, Virus RM, Djuricic-Nedelson M, Green RD (1984) Adenosine analogs and sleep in rats. J Pharmacol Exp Ther 228: Rainnie DG, Grunze HC, McCarley RW, Greene RW (1994) Adenosine inhibition of mesopontine cholinergic neurons: implications for EEG arousal. Science 263: Sherin JE, Elmquist JK, Torrealba F, Saper CB (1998) Innervation of histaminergic tuberomammillary neurons by GABAergic and galaninergic neurons in the ventrolateral preoptic nucleus of the rat. J Neurosci 18: Sherin JE, Shiromani PJ, McCarley RW, Saper CB (1996) Activation of ventrolateral preoptic neurons during sleep. Science 271: Strecker RE, Morairty S, Thakkar MM, Porkka-Heiskanen T, Basheer R, Dauphin LJ, Rainnie DG, Portas CM, Robert W, Greene RW, McCarley RW (2000) Adenosinergic modulation of basal forebrain and preoptic/anterior hypothalamic neuronal activity in the control of behavioral state. Behav Brain Res 115: Szymusiak R, Alam N, Steininger TL, McGinty D (1998) Sleep-wakingdischarge patterns of ventrolateral preoptic/anterior hypothalamic neurons in rats. Brain Res 24 (803): Szymusiak R, Danowski J, McGinty D (1991) Exposure to heat restores sleep in cats with preoptic/anterior hypothalamic cell loss. Brain Res 541: Szymusiak R, McGinty D (1986) Sleep-related neuronal discharge in the basal forebrain of cats. Brain Res 370: Szymusiak R, McGinty D (1989) Sleep-waking discharge of basal forebrain projection neurons in cats. Brain Res Bull 22: Ticho SR, Radulovacki M (1991) Role of adenosine in sleep and

7 S. Morairty et al. / Neuroscience 123 (2004) temperature regulation in the preoptic area of rats. Pharmacol Biochem Behav 40: von Economo C (1930) Sleep as a problem of localization. J Nerv Ment Dis 71: (Accepted 21 August 2003)