Experimental animals

    Adult male Hannover Wistar (I, II, IV) and Sprague Dawley (III) rats, weight 270-390g, total number 158 were used. The rats were bred in the colony of the Department of Physiology of the University of Helsinki (I,II), at the University of Zürich (III) and  at the Experimental Animal Center of the University of Helsinki (IV). The animals were kept in 12/12h light/dark rhythm with free access to standard rat chow and tap water. The animal experiment protocols and animal housing conditions were approved by the provincial administrative board in accordance with the laws of Finland and the European convention for the protection of experimental animals (N:o 1360/1990).

REM sleep deprivation (I, II, IV)

    REM sleep deprivation was achieved by the platform method first described by Jouvet for cats (Vimont-Vicary et al. 1966) and Morden for rats (Morden et al. 1967). The rats were kept on small platforms (diameter 6.5- 7.5 cm) surrounded by water of approximately 5 cm depth on the cage floor. Food and fresh drinking water were freely available during the REM sleep deprivation. Under these conditions REM sleep is suppressed almost totally (Porkka-Heiskanen et al. 1995) (IV) because the animal is not able to maintain its balance on the platform during REM sleep-associated muscle hypotonia. In the in situ histochemistry experiments (I, II) rats (n=6/group) were deprived in the same waterbath-cage system for 24 h (I, II) or 72 h (II) with 7-9 platforms i.e. the rats were able to move from platform to platform but were unable to sleep on any of them. The rats in the REM sleep rebound group (n=6) were first deprived for 72 h, and then they were moved to their usual cage for 24 h. Groups of control animals were kept for corresponding times on large platforms (diameter 11 cm) which allow the same conditions as the small platforms otherwise, except that almost the same amount of REM sleep is produced as in the baseline conditions (Porkka-Heiskanen et al. 1995). An additional control group consisted of animals housed together and taken directly from their regular cage (“home controls”). In plasma GH measurement (II) and i.c.v. injection (IV) experiments deprivation was produced in single rat cages containing two platforms per rat.

Total sleep deprivation (III)

    Total sleep deprivation  was achieved using the gentle handling method (Franken et al. 1993). Rats were kept awake by introducing and removing different objects e.g. pieces of wood and cardboard boxes from the cage. The aim of this method is to achieve sleep deprivation with minimal stress by using the natural curiosity of rats i.e. their ability to forget sleeping when they examine new objects in the cage. The rats were observed continuously, and were woken up by tapping the cage or by touching them lightly if they were falling asleep regardless of the objects. Dim red illumination allowed observation during the dark phase. During total sleep deprivation, the vigilance state was not monitored by on-line EEG, but the criteria for sleep behavior were identical with those used previously (Tobler and Borbély 1990, Franken et al. 1993) and sleep deprivation was produced in the same laboratory (Institute of Pharmacology, University of Zürich).

In situ hybridization protocol (I-III) (Fig. 2)

    At the end of REM (I, II) or total sleep deprivation (III) the rats were decapitated with a guillotine, their brains  immediately removed and frozen on a piece of solid carbon-dioxide. 20 µm coronal sections were cut through the hypothalamus with a cryotome. Sections were thaw-mounted on gelatin coated slides and stored at -70 °C. Before the hybridization the sections were post-fixed in 4% phosphate-buffered paraformaldehyde (pH 7.4) for 5 min at +4°, washed in cold PBS (0.1 M Na2HPO4, 0.15 M NaCl, pH 7.4) for 2 min, acetylated 10 min in 0.15% triethanolamine containing 2% acetic anhydride, dehydrated with graded ethanols and delipided for 5 min in chloroform. 48 base oligonucleotide probes were made by oligonucleotide synthesizer (Northwestern University, Evanston, IL, U.S.A.). The galanin probe consisted of bases 221-268 of galanin mRNA (Vrontakis et al. 1987), the somatostatin probe consisted of bases 361-408 of the rat somatostatin mRNA (Goodman et al. 1985), and the GHRH probe contained the bases 323-370 of the rat GHRH mRNA (Gonzales-Cresbo and Boronat 1991). The probes were labeled at the 3’ end with 35S-dATP (Amersham, Buckinghamshire, UK) using terminal transferase (TdT, Pharmacia Biotech Europe GmbH). The purified probe was mixed with 1% yeast tRNA (Boehringer Mannheim GmbH, Germany) solution and TED (10 mM Tris, 1 mM EDTA, 10 mM dithiotreitol, pH 8.0) buffer, heated to 70 °C for 3 min and cooled on ice. The final concentration of the probe in the hybridization buffer (50% formamide, 10% dextran, 0.3 M NaCl, 10 mM Tris, 1 mM EDTA, 1x Denhardts and 10 mM dithiotreitol) was 3 pmol/ml, and the specific activity was 3500 Ci/mmol for galanin (I), 1220-1610 Ci/mmol for somatostatin (II, III) and 1250-1750 Ci/mmol for GHRH (II, III). Sections were covered with 45 µl of hybridization mixture, covered with silanized cover slips (50x24 mm), and incubated overnight at 37 - 40 °C. The slides were washed in 1xSSC (0.15 M NaCl, 0.015 M sodium citrate, pH 7.0) for 4 x 15 min at 55 °C, followed by 2 x 1 h washes at room temperature. After dehydration with graded ethanols containing 6% ammonium acetate, slides were dipped in Kodak NTB-2 emulsion diluted 1:1 with 0.6 M ammonium acetate. The slides were exposed in light-proof boxes at +4 °C for 21 days (galanin), 15-18 days (somatostatin) and 29-30 days (GHRH). After exposure the emulsion was developed with D-16 developer (+16 °C diluted 1:1) and Kodak T-max fixer, and the sections were lightly stained with cresyl violet acetate for visualization in bright and dark field photo microscopy (Nikon Labophot, magnification 40 - 400 x). The sections were matched according to the rat brain atlas of Paxinos and Watson (1986) for analysis. The cells containing mRNA of somatostatin, GHRH or galanin were identified in the dark field as clusters of bright silver granules around stained blue nuclei on dark red background. If necessary, the cluster was checked in the bright field as dark granules around a stained violet nucleus. The specifity of the hybridization was tested by treatment of the sections with ribonuclease (10 µg/ml, Pharmacia Biotech) and the addition of 100 x the amount of unlabeled probe into the hybridization mixture. Both of these treatments extinguished the signal in the sections.


Fig. 2  Schematic presentation of the  sequence of the consecutive steps in the in situ hybridization.

Cell count and computer densitometry

    The amount of mRNA was measured by counting the number of mRNA expressing cells in specific nuclei (I-III) or by computer image-analysis densitometry (III). For the cell count, all cells in the area of the specific nucleus which expressed more than five times the background granule density were included and counted visually by the microscoper. Both measurements (cell count and densitometry) were made in 4-6 sections at regular intervals along the rostrocaudal projection of the studied nucleus and the mean of the measured sections was calculated.

    For the computer densitometry, a Sony CCD video camera mounted in a Nikon Labophot microscope with a combined dark/light field condenser was used. The video signal was A/D converted by a LG-3 frame grabber card (Scion Corporation, Frederic, MD, U.S.A.) and image analysis was done by a NIH Image program running in a Macintosh Centris 650. During the measurements the analyzer was unaware of the group status of an individual rat. The sections were analyzed in dark field illumination. The nucleus was encircled by free-hand drawing of a border around the outermost mRNA-expressing cells in the nucleus. The mean density and the area of the selection were measured. The mean density of the background was measured from a rectangular area of corresponding size in the immediate vicinity of the nucleus. The mean density of the background was subtracted from the mean density of the nucleus and the resulting difference was multiplied by the area of the nucleus. Multiplication of the signal density with the area compensates for the influence of the background between the cells to the signal value.

    Linearity of the densitometry was tested by regression analysis of measured density values (n=10) compared with the number of silver granules counted in the same areas. For comparing the slope of regression values between different sections these measurements were done in four sections and the slope values were tested by analysis of variance (ANOVA). In the regression analysis the r2 values were between 0.94 and 0.77 and no significant difference was detected between the slope values of the different sections (F(3,36)=1.27, P > 0.1).

Analysis of somatostatin cells (II, III)

    Expression of somatostatin mRNA was measured in the arcuate and periventricular nuclei of the hypothalamus. In the arcuate nucleus the somatostatin cells were identified as a densely packed cell group situated just laterally of the ventral part of the third ventricle (Fig. 3b). In the REM sleep deprivation study (II) somatostatin mRNA was measured bilaterally from 6 sections between stereotactical levels -2.1 and -3.6 mm from the bregma. In the total sleep deprivation study (III) 4 sections were analyzed between -3.1 and -3.5. In the periventricular nucleus the somatostatin-expressing cells were grouped bilaterally beside the third ventricle (Fig. 3a). Expression of somatostatin mRNA was analyzed in 4 sections between -1.4 and -1.8.

Analysis of GHRH cells (II, III)

    Expression of GHRH mRNA was measured in the paraventricular and the arcuate nuclei. In the total sleep deprivation study (III) the GHRH cells of the periventromedial hypothalamic area were also analyzed. The paraventricular nucleus was identified as the area where the GHRH mRNA- expressing cells were clustered lateral of the third ventricle (Fig. 3a). GHRH cells were analyzed bilaterally in 4 sections between the levels -1.8 and -2.2 from the bregma. GHRH cells in the arcuate nucleus were identified as a GHRH mRNA-expressing group of cells situated ventrolaterally of the third ventricle (Fig. 3b). Analysis was made bilaterally of 6 sections between the levels -1.9 and -3.4 (REM sleep deprivation) (II) and 4 sections between -3.05 and -3.45 (total sleep deprivation) (III). GHRH cells in the periventromedial hypothalamic area were a sparsely distributed group of cells situated dorsolaterally from the arcuate GHRH cells near the border of the ventromedial hypothalamic nucleus (Fig. 3b). Analysis was made bilaterally in 4 sections between levels -3.2 and -3.6.

Fig. 3  Schematic drawing of the rat hypothalamus approximately from the coronal levels -1.8 (a) and -3.3 (b) from the bregma. a) location of somatostatin cells in the periventricular nucleus (·) and GHRH cells in the paraventricular nucleus (°). b) location of the somatostatin cells  (·) and  the GHRH cells (°) in the arcuate nucleus and the GHRH in the periventromedial hypothalamic area (*). 3V: third ventricle, F: fornix, ME: median eminence, OT: optic tract, VMH: ventromedial nucleus.

Analysis of galanin cells (I)

    Galanin mRNA expressing cells were counted in the medial preoptic area and the periventricular nucleus in the anterior hypothalamus. In the medial preoptic area galanin expressing cells were scattered in the anterior hypothalamus between anterior commissure and optic chiasm (Fig. 4). Cell counts were made between the stereotactical coronal levels -0.04 and -1.04 mm from the bregma. Galanin cells in the periventricular nucleus were clustered bilaterally beside the third ventricle (Fig. 4). Counts were performed in 5 sections between -0.56 and -1.04 mm from the bregma. Cells within a distance of 1 mm lateral from the third ventricle were included.

Fig. 4  Schematic drawing of the rat anterior hypothalamus approximately from the coronal level -0.92 from the bregma. Location of galanin cells in the periventricular nucleus (PE) (·) and in the medial preoptic area (MPA)  (°). F: fornix, OX: optic chiasm, SCN: suprachiasmatic nucleus, 3V: third ventricle.

Plasma GH measurement during REM sleep deprivation (II)

    Rats were provided with a silastic jugular cannula under combined medetomidine (DomitorR, Orion-yhtymä, Espoo, Finland, 75 µg/kg s.c.) and pentobarbital (MebunatR, Orion, 45mg/kg i.p.) anesthesia. After the operation the cannula was heparinized, and the rats were allowed 48 h of recovery before the experiment. Five rats were deprived of REM sleep for 24 h by the platform method. A control group consisted of six rats kept on the large platforms. During deprivation hours 25-31 samples (200 µl)were collected from the cannula at 20 minute intervals. Blood was similarly collected from rats (n=4+4) that stayed in their home cages. REM sleep deprivation and large platform groups were compared to respective home control groups in two separate radioimmunoassays for GH. In the analysis laboratory the detection limit for GH-assay was 2.5 ng/ml and the intra- and inter assay coefficient of variations were 4.5% (intra) and 15.2% (inter) (Laartz et al. 1994). Reagents for the GH assay were provided by the National Institute of Diabetes, and Digestive and Kidney Diseases (NIDDK, Bethesda, Maryland, U.S.A.) and RP-2 was used as a GH reference.

Animal surgery for i.c.v. and local microinjections (I, IV)

    Two dental screw electrodes for electroencephalography (EEG) were implanted into the scull ipsilaterally above the frontoparietal cortex and two steel wire electrodes for electromyography (EMG) were inserted into neck muscles under combined medetomidine (DomitorR, Orion, 0.1 mg/kg s.c.) and pentobarbital (MebunatR, Orion, 30 mg/kg i.p.) anesthesia. The leads of the electrodes were soldered into a connector, which was secured in place with dental acrylate. For intracerebroventricular injections (I, IV) a guide cannula (outer diameter 0.7 mm) was implanted into the lateral ventricle according to stereotactical coordinates: 0.8 mm posterior from the bregma, 1.4-1.5 mm lateral from the midline, 6.0 mm (I) or 3.8 mm (IV) ventral from the level of the bregma (Paxinos and Watson 1986). After the surgery the rats were allowed to recover for 5-7 days before the recordings. After the operation and during the first post-operative day rats received 20 µg/kg buprenorphine s.c. (TemgesicR, Reckitt & Colman, Hull, England) for analgesia (IV). During the recovery and experiments the animals were kept in single animal boxes. Before the experiments the placement of the cannula was checked by injecting 0.2 µg of angiotensin II (Sigma Chemical Co., St. Louis, MO, U.S.A. product: A9525) into the cannula. Animals that did not show a clear drinking response were not used in the experiments. After the experiments the cannula placement was also verified by microscopy of 20 µm histological cryosections made of the dissected and frozen brains.

Injection procedure  (I, IV)

    In the i.c.v. experiments a steel injection cannula (outer diameter 0.4 mm) extending 0.5 mm beyond the guide was used. The cannula was connected by polyethylene tubing to a Hamilton syringe which was fitted into a CMA/100 microinjection pump (Carnegie Medicin, Stockholm, Sweden). In the microinjection experiments a cannula made of fused silica capillary (Composite Metal Services Ltd., The Chase, Hallow, Worcestershire, England.) (outer diameter 0.12 mm, extending 2 mm beyond the guide) was used. The injection system was first filled with distilled water and then a sufficient amount of injection solution was driven into the tubing separated from the water by an air bubble. In the i.c.v. injections a 2 µl volume was injected in 4 min (0.5 µl/min) and in the locus coeruleus microinjections 0.25 µl in 5 min (0.05 µl/min).

I.c.v. injections of somatostatin and somatostatin antagonist (IV)

    In the i.c.v. somatostatin antagonist experiment, rats (n=5) received in a random order either 2 µl of artificial cerebrospinal fluid (aCSF) or 0.5 nmol of somatostatin antagonist, cyclo-(7-aminoheptanoyl-phe-d-trp-lys-thr(bzl)) (Sigma C4801, Fig. 5) dissolved in 2 µl of artificial CSF or 2 nmol of somatostatin antagonist in the same volume. In the somatostatin experiment 5 rats received artificial CSF, 0.5 nmol or 2 nmol of somatostatin (Sigma S9129) according to the same protocol. Injections were given between 9:30 and 10:30 after sleep onset.


Fig. 5  Structure of somatostatin, octreotide (synthetic somatostatin analog) and somatostatin antagonist (SA) used in this study. Bzl: benzoate.

I.c.v. somatostatin antagonist after REM sleep deprivation (IV)

    24 h REM sleep deprivation was performed by the platform method starting and ending at 9:00 am. After the deprivation, rats were put into their home cages and an injection of artificial CSF or 0.5 nmol of somatostatin antagonist was given to 4 rats after the onset of sleep. A 2 nmol dose of somatostatin antagonist was studied using different rats (n=5) with the same protocol. Every rat received both treatments (artificial CSF and somatostatin antagonist) and there were 3-6 non-experiment days before the same rats were deprived and injected again.

I.c.v. injections of galanin (I)

    In the i.c.v. galanin experiments (I) rats received 0.06, 0.6 or 6 nmol of porcine galanin (Sigma G5773) at 9:00 am or 0.6 nmol at 8:00 pm (5-6 rats / treatment). Control injections consisted of saline vehicle. Every rat received altogether 2 treatments (1 dose of galanin and 1 saline control).

Microinjections into locus coeruleus  (IV)

    In the locus coeruleus microinjection experiments 8 rats received unilaterally 0.25 nmol of somatostatin, somatostatin antagonist or 0.25 µl of artificial CSF in a random order (Fig. 6). Injections were given in the morning between 9:30 and 10:00am after sleep onset. Another group of rats (n=5) received 0.25 nmol of rat galanin (Bachem AG, Bubendorf, Switzerland, product: H7450) or combined injections of 0.25 nmol of galanin + 0.25 nmol of somatostatin or 0.25 µl of artificial CSF according to the same protocol.

Fig. 6  Schematic illustration of the microinjection sites in the brainstem at the coronal level -9.3 - 9.8 from the Bregma  according to photomicroscopic verification. On the right: somatostatin / somatostatin antagonist injections (n=8),  on the left: galanin / somatostatin (n=5) (for clarity, plotted on the left side, actual injections were performed to the right).  4V: fourth ventricle, LC: locus coeruleus

Sleep polygraphy (I, IV)

    The EEG and EMG signals were amplified with an 8 channel Beckman type T electroencephalograph (Offner division, Chicago, U.S.A.). The amplified signal was plotted on paper (I) or collected and digitized by a CED 1401 plus interface (Cambridge Electronic Design Ltd, Cambridge, England) with a sampling rate of 62 Hz. Data were displayed and stored by a Spike2 program (Cambridge Electronic Design Ltd) running in a desktop PC. During the recordings rats were remote-observed by means of a video camera and monitor. Collected data were categorized visually into wakefulness, non-REM sleep and REM sleep in 30-s epochs according to the standard criteria (Timo-Iaria et al. 1970). Briefly, wakefulness was defined as low amplitude and mixed intermediate/high frequencies in EEG and high amplitude EMG with artifacts caused by body movements. Non-REM sleep was defined as high amplitude slow waves in the EEG and intermediate EMG amplitude, and REM sleep was defined as low/intermediate amplitude of predominant theta frequency in EEG combined with low amplitude EMG with occasional twitches. Transition from non-REM to REM sleep (“pre-REM sleep”, high amplitude theta bursts mixed into slow waves in EEG) was allocated equally between non-REM and REM sleep.

    Proportions of vigilance states per 2 h time intervals were calculated from the 6 h (IV) 8 h (I) recording. In experiment IV the first half hour after the injection was excluded because of the arousal caused by the injection procedure.


    The effect of REM sleep deprivation on mRNA expression (I, II) was evaluated with one-way analysis of variance (ANOVA) between the home cage condition and 24 h or 72 h on small or large platforms. Significant results were further analyzed by pairwise comparisons between the treatment groups with Student-Newman-Keuls method (I) or Duncan’s test (II). The effects of rebound sleep after 72 h REM sleep deprivation (II) were calculated separately by two-way ANOVA for the 72 h deprivation and 72 h deprivation + 24 h rebound groups. The factors were treatment (small or large platform) and length of rebound (0 or 24 h).

    The effects of total sleep deprivation on mRNA expression (III) were tested by two-way ANOVA. The factors were treatment (deprivation or control) and time of decapitation (9:00 after 12 h treatment or 15:00 after 6 h treatment). If interaction between the factors was detected (P > 0.1 for interaction) all four groups were compared by one-way ANOVA followed with paired comparisons by Newman-Keuls method.

    The plasma GH measurements (II) were evaluated by integrating the area under the curve (AUC), and the difference between platform and corresponding home cage control groups was compared using the t-test for independent measurements.

    In the i.c.v. somatostatin, somatostatin antagonist and in the microinjection experiments (IV) differences in the proportions of the vigilance states between the treatments were compared by one-way ANOVA for repeated measurements. Significant results were compared pairwise between the treatments by Newman-Keuls test or by Bonferroni t-test (somatostatin and somatostatin antagonist microinjections). In the i.c.v. galanin injection (I) and i.c.v. somatostatin antagonist after REM sleep deprivation (IV) experiments the treatments were compared by paired t-test. In all statistical tests P < 0.05 was set as a limit of statistical significance.