Regulatory, Integrative and Comparative Physiology

Systemic injection of a nitric oxide synthase inhibitor suppresses sleep responses to sleep deprivation in rats

Ana Cristina Ribeiro, John G. Gilligan, Levente Kapás


We hypothesized that nitric oxide (NO) may play a role in homeostatic sleep regulation. To test this hypothesis, we studied the sleep deprivation (SD)-induced homeostatic sleep responses after intraperitoneal administration of an NO synthase inhibitor, Nω-nitro-l-arginine methyl ester (l-NAME, a cumulative dose of 100 mg/kg). Amounts and intensity of sleep were increased in response to 8 h of SD in control rats (n = 8). Sleep amounts remained above baseline for 16 h after SD followed by a negative rebound. Rapid eye movement sleep (REMS) and non-REMS (NREMS) intensities were elevated for 16 and 4 h, respectively. l-NAME treatment (n = 8) suppressed the rebound increases in NREMS amount and intensity. REMS rebound was attenuated by l-NAME in the first dark period after SD; however, a second rebound appeared in the subsequent dark period. REMS intensity did not increase after SD in l-NAME-injected rats. The finding that the NO synthase inhibitor suppressed rebound increases in NREMS suggests that NO may play a role as a signaling molecule in homeostatic regulation of NREMS.

  • rapid eye movement sleep
  • fast Fourier analysis
  • N ω-nitro-l-arginine methyl ester
  • electroencephalography
  • slow-wave activity

sleep is under the influence of the circadian and homeostatic regulatory processes (7). The circadian process is a clocklike mechanism that is mainly controlled by the suprachiasmatic nucleus (SCN) and is independent of prior wakefulness. The homeostatic component of non-rapid eye movement sleep (NREMS) regulation is a mechanism that elicits compensatory increases in NREMS after sleep loss. The existence of homeostatic NREMS-promoting mechanisms is evident from sleep deprivation (SD) studies. SD causes compensatory increases in time spent in NREMS and NREMS intensity, as indicated by electroencephalographic (EEG) δ-wave activity in several species, including rats (8). The homeostatic process is independent of the circadian rhythm. In contrast to the circadian modulation of sleep, the specific brain structure where homeostatic sleep-regulating mechanisms reside has not been identified. Homeostatic increases in sleep can be due to increased activity of a putative neuronal network(s) or the accumulation of an endogenous substance during wakefulness. There are mathematical models that describe homeostatic sleep regulation in quantitative terms, simulate homeostatic sleep responses, and predict changes induced by varying the duration of wakefulness (1).

A growing body of evidence suggests that nitric oxide (NO) may play a role in the circadian and homeostatic processes of sleep regulation. The circadian clock is entrained to the environmental light-dark cycle via the retinohypothalamic pathway (reviewed in Ref. 30). Glutamate is the primary neurotransmitter mediating the transmission of photic signals from the retinal afferents to the SCN (12). NO synthase (NOS) inhibitors block the behavioral circadian responses to photic stimuli in hamsters (12) and autonomic responses in rats (3). Also, NOS inhibitors block the glutamate-induced phase shifts of SCN neuron activity in vitro (35). These results suggest that the transduction of photic stimuli in the SCN and the entrainment of the biological clock to photic stimuli require the formation of NO.

Indirect evidence supports the hypothesis that NO is an endogenous sleep-promoting substance involved in homeostatic sleep regulation. First, intracerebroventricular injection of NO donors, such as 3-morpholinosydnonimine andS-nitroso-N-acetylpenicillamine, mimics the effects of prolonged wakefulness, i.e., increases time spent in NREMS as well as NREMS intensity in rats (21). Injection ofS-nitroso-N-acetylpenicillamine in cats (11) or the NO precursor l-arginine in rats (18) into the pedunculopontine region causes increases in NREMS and REMS. Conversely, systemic (14,15, 19), intracerebroventricular (19), or intrapedunculopontine (18,20) injection of NOS inhibitor suppresses sleep in rats. Similarly, systemic and intracerebroventricular injection of NOS inhibitor suppresses sleep in rabbits (22), and intrapedunculopontine injection of NOS inhibitor suppresses sleep in cats (11). The activity of NOS shows diurnal variation. NOS activity increases during the behaviorally active period of the day in the hypothalamus, brain stem, cerebellum, and hippocampus of rats (4). Direct measurement of NO release from the cerebral cortex (9) or thalamus (36) of rats showed that production of NO is higher during wakefulness than during NREMS.

One approach to study the involvement of NO in homeostatic sleep regulation is to test the effects of inhibiting NOS on homeostatic sleep responses. In the present experiments we studied the homeostatic sleep responses to SD in control rats and rats injected withN ω-nitro-l-arginine methyl ester (l-NAME), an inhibitor of NOS. The results indicate that administration of NOS inhibitor decreased NREMS amounts and intensity after SD, suggesting that NO plays a key role in homeostatic sleep regulation.



Male Sprague-Dawley rats (290–440 g) were used. With use of combined ketamine (87 mg/kg) and xylazine (13 mg/kg) anesthesia, animals were implanted with cortical EEG electrodes, nuchal electromyographic (EMG) electrodes, and a brain thermistor. The EEG electrodes were placed over the frontal, parietal, and occipital cortices, and the thermistor was placed on the dura over the parietal cortex. After a 1-wk recovery period, the animals were placed into sound-attenuated individual sleep-recording cages for habituation to the experimental conditions. During this adaptive period, the animals were connected to the recording cables and injected daily with isotonic NaCl solution intraperitoneally at light onset. Animals were kept on a 12:12-h light-dark cycle (light onset at 0600) and an ambient temperature of 26 ± 1°C for ≥2 wk before surgeries, during the recovery period, and throughout the experimental procedure. Water and food were provided ad libitum during this time.


EEG, EMG, and brain temperature (Tbr) were recorded by computer. EMG activity served the sole purpose of aiding in determining the vigilance states of the animals and was not further quantified. EEG was filtered below 0.1 Hz and above 40 Hz. The amplified signals were digitized at the frequency of 128 Hz for EEG and EMG and 2 Hz for Tbr. Single Tbr samples were saved on the hard disk in 10-s intervals. Average Tbr was calculated in 1-h time blocks. The vigilance states were determined offline in 10-s epochs. EEG, EMG, and Tbr were displayed in the computer in 10-s epochs and also simultaneously in a more condensed form in 12-min epochs. Wakefulness, NREMS, and REMS were distinguished as described in detail previously (23). Time spent in NREMS and REMS was calculated in 2-h time blocks. Online fast Fourier analysis of the EEG was also performed in 10-s intervals on 2-s segments of the EEG in 0.5-Hz bands of the 0.5- to 16-Hz frequency range. The EEG power density values were summed in four frequency bands for each 10-s epoch; δ (0.5–4 Hz)-, θ (4.5–8 Hz)-, α (8.5–12 Hz)-, and β (12.5–30 Hz)-activities were calculated. The spectral data were paired with the vigilance states, and EEG power was computed in each of the four bands separately for each vigilance state. Hourly average δ-, θ-, α-, and β-activities (μV2) were then calculated for the NREMS, REMS, and wakefulness epochs. On the baseline day, average power densities were computed across the entire 23 h for each rat to obtain a reference value for each animal. Power densities in 2-h blocks on the baseline day and the test days were then expressed as a percentage of that reference value. Group averages were also calculated for the entire light and dark periods (experiment III) or the last 4 h of the first light period and the subsequent entire dark and light periods (experiments I and II).

Experimental Protocol

Experiment I: effect of SD on sleep, EEG power density, and Tbr.

On the control day, sleep was recorded for 24 h starting at 0600. On the next day, all the animals were sleep deprived by gentle handling for 8 h, starting at light onset. Food and water were provided ad libitum during the SD. Animals received isotonic NaCl (2 ml/kg ip) at 0700 and 1100. They were returned to their home cages at 1400, and sleep was recorded for 48 h.

Experiment II: effect of l-NAME on SD-induced changes in sleep, EEG power density, and Tbr.

The experimental design was similar to experiment I, except during SD, the animals received l-NAME (50 mg/kg ip; Sigma Chemical, St. Louis, MO) at 0700 and 1100. l-NAME was dissolved in isotonic NaCl at a concentration of 25 mg/ml. Recordings for experiments I and II were performed on the same animals (n = 8). On a given test day, one-half of the animals received saline, and the others received l-NAME. One week later, the treatments were reversed and the recordings were repeated.

Experiment III: effect of l-NAME on spontaneous sleep, EEG power density, and Tbr.

On the control day, the rats (n = 8) received isotonic NaCl (2 ml/kg ip) at 0700 and 1100. Sleep was recorded for 23 h starting at 0600. On the test day, all the rats received l-NAME (50 mg/kg ip) at 0700 and 1100. Sleep was recorded continuously for 56 h from 0600.

Statistical Analysis

Statistical comparisons were made between the baseline days and the test days. In experiments I and II, ANOVA was done separately on five segments of the recordings: the first 4 h after the end of SD (i.e., the last 4 h of the light period), hours 5–16 after the end of SD (the 1st dark period), hours 17–28 (the 2nd light period), hours 29–40 (the 2nd dark period), and hours 41–48 (the 1st 8 h of the 3rd light period, i.e., the last 8 h of the recording). In experiment III the first segment of the recordings analyzed by ANOVA was the entire light period of the first experimental day. Two-way ANOVA for repeated measures were performed on 2-h time blocks for sleep and 1-h averages for Tbr. EEG spectra values were analyzed by two-way ANOVA on 2-h averages. When ANOVA indicated significant treatment effects, paired t-tests were performed a posteriori.


Experiment I: Effect of SD on Sleep, EEG Power Density, and Tbr in Control Rats

NREMS and REMS amounts and NREMS intensity were increased after SD in saline-treated rats (Fig. 1, Table1). Time spent in NREMS was 196.0 ± 29.9 min (∼58%) above baseline in the 16-h period after SD (hours 9–24). In the subsequent light period, hours 25–36, there was a negative rebound in NREMS, then NREMS returned to baseline levels by the beginning of the second dark period. In the last 8 h of the recordings, there was again a small, but statistically significant, decrease in the amount of NREMS.

Fig. 1.

Effect of sleep deprivation (SD) in saline-treated rats on sleep, electroencephalogram (EEG) power spectra, and brain temperature (Tbr). Filled horizontal bars, dark phases of day; crosshatched bar, period of SD. ○, Control day; ●, SD day. Amounts of non-rapid eye movement sleep (NREMS), rapid eye movement sleep (REMS), wakefulness, EEG δ-activity during NREMS, and EEG θ-activity in REMS were calculated in 2-h time blocks. Tbr values were averaged in 1-h intervals. Error bars, SE. Arrows, time of injections. * Significant differences between control and experimental treatments (P < 0.05, pairedt-test).

View this table:
Table 1.

Effects of SD in saline- and l-NAME-injected animals on sleep amounts, EEG δ-activity during NREMS, EEG θ-activity during REMS, and Tbr: statistical results

REMS did not change significantly in the 2 h immediately after SD (Fig.1). Subsequently, time spent in REMS increased for 14 h by 77.3 ± 9.7 min (∼165%). From the beginning of the second light period (hour 25), REMS remained at baseline levels. Amount of REMS was increased by 73.2 ± 10.9 min across the 48-h recording period.

Time spent in wakefulness decreased by 47% in hours 9–24and increased in the second light period, corresponding to the changes in sleep. Amount of wakefulness returned to baseline levels from the second dark period (Fig. 1).

SD induced significant changes in the EEG δ-, θ-, and α-frequency ranges in NREMS (Table 1). The δ-activity during NREMS, a measure of NREMS intensity, increased by 116% in the first 4 h (Figs. 1 and2) and then exhibited a negative rebound that lasted from hours 19 to 26 (Fig. 1). The θ- and α-activities were also increased in the first 4-h period, and there was a tendency toward increased α- and β-activities for the remainder of the recording period in NREMS. In fact, α-activity was significantly elevated during the second dark period (Fig. 2). SD induced an increase in EEG θ-activity during REMS (Figs. 1 and 2). It increased by 27.2% in the first 4 h and by 20.1% during the subsequent dark period. These increases gradually diminished throughout the rest of the recording period. SD did not affect the δ-activity during REMS, but it increased β- and α-activities during the first and the first and second dark periods, respectively. Changes in EEG power spectra during wakefulness followed a pattern similar to that seen in NREMS (Fig. 2).

Fig. 2.

Effects of SD in saline (A)- andN ω-nitro-l-arginine methyl ester (l-NAME)-treated rats (B) and effect ofl-NAME in rats not subjected to SD (C) on EEG power spectra. Values are means ± SE (baseline = 100%). Top row: δ-activity; 2nd row from top: θ-activity; 3rd row: α-activity; bottom row: β-activity. W, wakefulness. EEG power spectra were averaged in 4- or 12-h time blocks: last 4 h of 1st light period (hours 9–12, I), 1st dark period (hours 13–24, II), 2nd light period (hours 25–36, III), and 2nd dark period (hours 37–48, IV). * Significant differences between control and treatment conditions (P < 0.05, pairedt-test).

ANOVA for repeated measures indicated significant changes in Tbr during two light periods (hours 25–36 and49–56) after SD (Table 1). Post hoc analysis, however, did not reveal any time point when the differences between the control and SD day were statistically significant.

Experiment II: Effect of SD on Sleep, EEG Power Density, and Tbr in l-NAME-Treated Rats

The pattern of NREMS responses to SD in l-NAME-treated animals was similar to that in saline-treated rats but was shorter in duration and lower in amplitude (Fig. 3, Table 1). The initial positive NREMS rebound lasted only for 6 h. Time spent in NREMS was increased in hours 9–24 by 154.8 ± 15.6 min (40% above baseline), a significantly smaller change than inexperiment I. These increases were followed by a negative rebound in the subsequent light period, and NREMS returned to baseline values thereafter.

Fig. 3.

Effect of SD in l-NAME-treated rats on sleep, EEG power spectra, and Tbr. ○, Control day; ●, SD withl-NAME. See Fig. 1 legend for details.

Time spent in REMS showed biphasic increases after SD inl-NAME-treated rats (Fig. 3). Values were below baseline in the first 2-h time block and then exhibited a significant positive rebound in hours 13–24. This increase was 45.9 ± 16.5 min, 107% above baseline, a significantly attenuated response compared with that in saline-treated animals after SD. REMS amount returned to baseline levels in the following light phase. In the second dark period, there was a second positive rebound; time spent in REMS increased again by 35.4 ± 12.2 min (78% above baseline), a response that was absent in saline-treated rats. For the rest of the recording period, REMS returned to baseline. Time spent in REMS for the entire 48-h recording period in l-NAME-treated rats was increased by 77.2 ± 41.2 min.

Wakefulness remained below baseline for the first 16 h after SD, then there was a tendency toward a positive rebound lasting for 10 h (Fig.2). In the second dark period, coinciding with the second increase in REMS, the amount of wakefulness was again below baseline.

EEG δ-activity during NREMS was increased only in the first 2-h time block after SD, by 56% above baseline (Fig. 3). This is a smaller response than that seen in saline-treated rats, which showed an increase of 198% during the first 2 h. When EEG power spectra were averaged across the first 4 h, there was no significant change in NREMS δ-activity, in contrast to the significant response in saline-treated animals (Fig. 2). From hour 13, δ-activity in NREMS was significantly below baseline until the end of the recording. Similar tendencies were observed for θ-activity during NREMS. The α- and β-frequencies showed no significant changes (Fig. 2). During REMS, there was a gradual increase in EEG δ-activity, which reached statistical significance in the last dark period (Fig. 2).l-NAME-treated rats showed suppressed θ-activity in REMS during the second light period after SD. REMS α-activities were increased during the second and β-activities during the first and second dark periods. There were no significant changes in EEG power spectra during wakefulness (Fig. 2).

Sleep-deprived l-NAME-treated rats showed a statistically significant decrease in Tbr during the first dark phase after SD and an increase in Tbr in the following light period (Fig. 3).

Experiment III: Effect of l-NAME on Sleep, EEG Power Density, and Tbr in Rats Not Subjected to SD

l-NAME treatment alone decreased REMS and NREMS during the light periods of the day, thus resulting in increased wakefulness at these times (Fig. 4, Table2). NREMS amounts were significantly suppressed during the second light period, hours 25–36, by 63.2 ± 18.4 min (∼17% below baseline).

Fig. 4.

Effect of l-NAME on sleep, EEG power spectra, and Tbr in rats not subjected to SD. ○, Control day; ●,l-NAME treatment. Amounts of NREMS, REMS, wakefulness, EEG δ-activity during NREMS, and EEG θ-activity in REMS were calculated in 2-h time blocks with exception of last time block in light period on control day, which is a 1-h time block. See Fig. 1 legend for details.

View this table:
Table 2.

Effect of l-NAME on sleep amounts, EEG δ-activity during NREMS, EEG θ-activity during REMS, and Tbr in rats not subjected to SD: statistical results

Time spent in REMS was more affected by l-NAME treatment than NREMS (Fig. 4). REMS responses were significantly suppressed in all light periods, i.e., hours 1–12, 25–36, and49–56, by 26.9 ± 6.7 min (∼38%), 25.4 ± 8.8 min (∼37%), and 23.6 ± 5.1 min (∼43%), respectively. During the dark periods, the amount of REMS remained at baseline levels.

Time spent in wakefulness was significantly increased in all the light periods and remained at baseline levels during the dark periods (Fig.4). In the first 12-h period, wakefulness increased by 51.3 ± 17.2 min (∼21%) compared with the baseline day. In the second light period, hours 25–36, the amount of wakefulness increased by 88.6 ± 16.2 min (∼38%) and in the last 8-h period of the recording period, hours 49–56, by 73.2 ± 16.7 min (∼57%).

EEG δ-activity during NREMS in l-NAME-treated non-SD rats was significantly below baseline levels for the entire recording period (Figs. 2 and 4, Table 2). NREMS θ-activity was suppressed inhours 13–36. There was also a tendency toward decreased α- and β-frequencies during NREMS. During REMS, EEG δ-, α-, and β-activities were not affected by l-NAME treatment. REMS θ-activity decreased gradually, reaching statistical significance in the second light period. EEG δ-activity in wakefulness was significantly suppressed in l-NAME-treated rats inhours 13–36. In wakefulness, α- and β-frequencies were not affected by l-NAME. θ-Activity in wakefulness was significantly suppressed during the second light period.

In the first 12 h, Tbr was decreased inl-NAME-treated non-SD rats. There was still a strong tendency toward suppressed Tbr in the first dark period. Thereafter, Tbr returned to baseline levels (Fig. 4).


The present study confirms our previous findings that systemic administration of the NOS inhibitor l-NAME suppresses spontaneous NREMS, REMS, and EEG δ-activity during NREMS and lowers Tbr in rats (19). To study the role of NO and NOS in the brain, it is important that l-NAME be administered at doses high enough to suppress brain NOS activities. We did not measure brain NOS activity in the present study, but previous experiments support our assumption that NOS activities in the brain were significantly suppressed after l-NAME administration. In the present experiments we injected l-NAME (50 mg/kg ip) twice, at 4-h intervals. Intraperitoneal bolus injection of 50 mg/kgl-NAME causes a ∼30% decrease in the NOS activity of the hypothalamus and ∼60% suppression in the brain stem 330 min after the treatments in rats (5).

The sleep-suppressive and EEG effects of l-NAME were relatively long lasting. This is consistent with our previous observations and the finding that l-NAME is an irreversible inhibitor of NOS (13). The effects of l-NAME were restricted to the lower-frequency, δ- and θ-, range of the EEG. The most robust EEG effects occurred during NREMS; δ- and θ-activities of the EEG were suppressed for 36–48 h after l-NAME injection. Also, there were decreases in slow-wave activities during wakefulness. Another NOS inhibitor, 3-bromo-7-nitroindazole, also suppresses EEG power in wakefulness and NREMS in rats (14). However, unlike 3-bromo-7-nitroindazole, which suppresses EEG in a wide (0.5–20 Hz) frequency range, the effect of l-NAME was restricted to the δ- and θ-bands.

The amount of sleep under normal, baseline conditions is determined by the interaction of three major factors. First, the duration of prior wakefulness, which is the homeostatic factor (7). Prolonged periods of wakefulness are followed by lengthened sleep periods and increased sleep intensity. Second, a circadian factor also contributes to sleep regulation (7). In nocturnal animals, sleep threshold is lower during the day and higher during the night; therefore, sleep occurs predominantly during the light phase of the day. Finally, several environmental factors, such as ambient temperature, also influence sleep amount. Various neurotransmitters (17), hormones (17), and cytokines, such as interleukin-1β and tumor necrosis factor-α (24), may be involved in some or all aspects of sleep regulation. We hypothesized that NO plays a role in the homeostatic regulation of sleep. The function of homeostatic sleep-regulatory mechanisms can be studied using SD. Our results confirm the well-known findings that total SD elicits rebound increases in the amounts of NREMS and REMS. The magnitude of these increases depends on the duration of the SD, the circadian phase when the SD was performed, and other factors. Under our experimental conditions, NREMS rebound started immediately after the cessation of SD, whereas compensatory REMS increases did not become evident until the second 2-h time block after SD.

It has been proposed that the δ-activity of the EEG during NREMS is an indicator of NREMS intensity (7) and θ-activity in REMS is a measure of REMS intensity (8) in several species, including rats and humans. Our results confirm that δ-activity in NREMS shows a biphasic response to SD (16). After an initial increase for 4 h, NREMS was suppressed in hours 12–18 after SD. In contrast to NREMS δ-activities, θ-activity in REMS exhibited a monophasic increase after SD. These findings indirectly support the hypothesis that δ- and θ-activities may be considered a measure for NREMS and REMS intensities, respectively. It is of particular interest that the increases in θ-activities in REMS preceded the increases in time spent in REMS after SD, indicating that REMS θ-activity (i.e., REMS intensity) is more sensitive to SD than REMS amounts. Similarly, increased sleep intensity without concomitant increase in sleep duration has been reported for NREMS after SD in rats (8). Our results also confirm that in wakefulness, as well as in NREMS, there is an increase in EEG power over nearly the entire frequency range after SD (8). Tbr did not change significantly after SD, although there was a slight tendency toward decreased Tbr inhours 4–6 after SD, which is in agreement with previously reported observations in rats (16).

Our major finding is that l-NAME shortened and decreased the amplitude of NREMS responses to SD. The rebound increases in NREMS amounts and intensities were suppressed by l-NAME. This suggests that the homeostatic increase in NREMS requires an intact NO-generating capacity of the brain. Homeostatic sleep-promoting mechanisms are not only active after SD but also play a role in determining the amount of spontaneous sleep under normal conditions. That is, the amount of sleep during the rest period is, in part, set by the amount of wakefulness in the preceding behaviorally active period.l-NAME suppressed spontaneous sleep predominantly in the light, which is the rest period for rats. This is likely due to the inhibitory effect of l-NAME on the homeostatic component of NREMS regulation. In the dark period, when the activity of homeostatic sleep-promoting mechanisms is low, l-NAME did not affect NREMS.

SD induced increases in the amount of REMS and EEG θ-activity during REMS. In contrast to NREMS, the total rebound increase in REMS after SD was not affected by l-NAME. This suggests that NO is not likely to play a key role in the homeostatic regulation of REMS amounts. The time course of the REMS rebound, however, was greatly affected by l-NAME. Only a partial REMS rebound took place during the first dark period after SD, and it was followed by a second wave of rebound in the second dark phase after SD. The presence of the second REMS rebound indicates that homeostatic REMS pressure is still high 28 h after the end of SD in l-NAME-injected rats. The finding that l-NAME did not affect the total compensatory REMS increase after SD suggests that it is not the homeostatic signaling for REMS that l-NAME suppresses. In addition to homeostatic and circadian signaling, normal sleep requires the intact function of the neuronal mechanisms that generate and maintain sleep. It is possible that l-NAME interferes with the REMS-generating mechanisms, and this is why REMS responses during the first night after SD were decreased. Cholinergic and cholinoceptive neurons, essential for triggering and maintaining REMS, reside in the lateral pontine region (reviewed in Refs. 6 and 31). In rats, NOS is colocalized with ACh in the pons (34), and inhibition of NOS suppresses pontine ACh release in cats (25). This raises the possibility thatl-NAME may attenuate the activity of pontine REMS-generating mechanisms immediately after SD. Because REMS pressure is not completely discharged in the first dark period, a second REMS rebound takes place in the subsequent dark phase. The lack of rebound increases during the light period is likely due to the fact that, during the light period, REMS-promoting mechanisms are already fully activated, and further increases in REMS are difficult to achieve. It has been suggested that EEG θ-activity during REMS indicates the intensity of REMS in rats (8). l-NAME did not affect REMS θ-activities under normal conditions but completely prevented the SD-induced increases in REMS θ-activity. This suggests that although NO is unlikely to have a central role in the homeostatic regulation of the REMS amounts, it may be involved in the induction of increased REMS intensity after SD.

Our present study has its limitations. First, it does not address the question of which cells or neural circuits mediate the effects ofl-NAME. It is likely that l-NAME acts on cellular elements in the brain that contain NOS. Many brain structures implicated in sleep regulation also contain relatively high concentrations of NOS. For example, neurons in the basal forebrain, various nuclei of the hypothalamus (e.g., the preoptic area, paraventricular nucleus, and lateral hypothalamic area), and the laterodorsal and pedunculopontine tegmental nuclei show strong staining for NOS-like immunoreactivity (33). Microinjections ofl-NAME into these structures are needed to localize the effects of l-NAME in the brain. It is possible that at least some of the effects of l-NAME are mediated by the pedunculopontine tegmental nucleus, since microinjection ofl-NAME into this structure suppresses spontaneous sleep in rats (18, 20) and cats (11). Second, in the present study we measured the effects of an NOS inhibitor on homeostatic sleep responses. Other approaches, such as measuring NOS activities or NO formation after SD, are needed to complement our understanding of the role of NO in homeostatic sleep regulation. Third, we injected l-NAME systemically. After systemic injections, l-NAME exerts its effects on central and peripheral targets. It has been speculated that central and peripheral NO may have opposite roles in regulating sleep (10).

A substance involved in homeostatic NREMS-promoting mechanisms should fulfill several criteria. For example, the injection of the substance should increase NREMS, whereas decreasing the endogenous levels of the substance should suppress spontaneous sleep as well as homeostatic NREMS responses to SD. The production of the substance is expected to increase during wakefulness and decrease during sleep. NO seems to fulfill these criteria. Injection of NO donors stimulates NREMS (11,21), and inhibition of NOS suppresses spontaneous NREMS (11, 14, 15,19, 22) and REMS (11, 14, 15, 19, 22, 26). NO production in the brain is higher during wakefulness than during NREMS (4, 9, 36). The present study indicates that inhibition of NOS also suppresses NREMS rebound after SD. These findings are consistent with the hypothesis that NO may play a role in the homeostatic regulation of NREMS. In addition to its role in homeostatic sleep regulation, NO may have a function in regulatory mechanisms mediating the effects of circadian (12, 35) and environmental influences on sleep.


It is hypothesized that a complex network of endogenous sleep-promoting and -inhibiting substances, also called “sleep factors,” may play a role in sleep regulation (reviewed in Ref. 24). Sleep factors may mediate the circadian and homeostatic signaling of sleep, or they may be involved in executive mechanisms that generate and maintain sleep. Our results support the notion that NO may be a signaling molecule involved in the homeostatic control of NREMS regulation. There are other sleep factors, e.g., interleukin-1β (29), tumor necrosis factor-α (32), and growth hormone-releasing hormone (GHRH) (28), the inhibition of which also attenuates homeostatic sleep responses after SD. Cytokines (27) and GHRH (2) stimulate NO release in the brain. Future experiments are needed to clarify the interactions among homeostatic sleep factors. NO may serve as a common mediator of effects of cytokines and GHRH in sleep homeostasis.


This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-30514.


  • Address for reprint requests and other correspondence: L. Kapás, Dept. of Biological Sciences, Fordham University, 441 E. Fordham Rd., Bronx, NY 10458 (E-mail: kapas{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


View Abstract