INTRODUCTION

Rapid eye movement (REM) sleep is a distinctive sleep stage that alternates periodically with episodes of slow-wave sleep (S) and is normally preceded by periods of S in which pontogeniculooccipital (PGO) waves become apparent during the transition from S to REM sleep. PGO waves are phasic field potentials recorded from the pontine tegmentum, lateral geniculate body, and occipital cortex. These waves constitute one of the distinguishing features of REM sleep. Brain stem cholinergic mechanisms have long been thought to play a major role in the generation of REM sleep, cortical activation, and the genesis of PGO waves (Datta 1995, 1997; Gillin et al. 1985; Jones 1991; Morrison 1979; Vanderwolf 1988). The cholinergic hypothesis of REM sleep generation comes from microinjection studies showing that centrally administered cholinomimetics evoke a state that is behaviorally and electrophysiologically similar to REM sleep (Baghdoyan et al. 1993; Datta et al. 1992; Gnadt and Pegram 1986; Hernandez-Peon et al. 1962). Furthermore, spontaneous release of acetylcholine (ACh) in the pontine reticular formation is greater during REM sleep than it is during waking (W) and S (Kodama et al. 1990; Lydic et al. 1991). The concept of a brain stem cholinergic mechanism in cortical activation was first proposed by Shute and Lewis (1967) and has been supported by many pharmacological studies (reviewed in Datta 1995; Vanderwolf 1988). The hypothesis that the brain stem cholinergic system contributes to neocortical activation receives additional support from the finding that the levels of ACh in the thalamus are higher during the cortically activated states of W and REM sleep than they are during S (Williams et al. 1994). Like REM sleep and cortical activation, the role of cholinergic mechanisms in the generation of PGO wave activity was also demonstrated by studies involving systemic and intracerebral injection of cholinergic agents (Datta et al. 1991–1993; Henriksen et al. 1972; Jacobs et al. 1972; Magherini et al. 1971; Ruch-Monachon et al. 1976). Taken together, these results strongly support the suggestion that ACh is involved in the generation of REM sleep, cortical activation, and the genesis of PGO waves.

Identification of brain stem cholinergic neurons in the pedunculopontine tegmentum (PPT) has generated a new impetus to study the direct involvement of PPT neurons in the generation of REM sleep, cortical activation, and the genesis of PGO waves (Mesulam et al. 1983, 1984, 1989; Shiromani et al. 1988; Vincent and Reiner 1987). Recent brain stem lesion studies have demonstrated that lesions of the PPT cholinergic neurons reduced and/or eliminated REM sleep, cortical activation, muscle atonia during REM sleep, and PGO waves (Datta and Hobson 1995; Shouse and Siegel 1992; Webster and Jones 1988). Electrical stimulation of the PPT causes increased ACh release in the cholinoceptive REM-sleep-generating region of the pontine reticular formation (Lydic and Baghdoyan 1993). Recordings from the PPTs of behaving cats have identified several different classes of neurons whose firing rates correlate with REM sleep, cortical activation, and PGO waves (Datta 1995, 1997; Datta and Hobson 1994; Datta et al. 1989; El-Mansari et al. 1989; McCarley et al. 1978; Nelson et al. 1983; Sakai and Jouvet 1980; Steriade et al. 1990). Some tonically discharging PPT neurons called REM-on cells progressively increase their firing rates as the animal moves from wakefulness to S and then to REM sleep. Other PPT neurons called W-REM-on cells are tonically active during both cortically activated states of W and REM sleep. A small population of REM-on and W-REM-on cells discharges phasically with phasic PGO waves (called PGO-on cells). Collectively these observations suggest that the PPT and its neurotransmitter ACh are involved in the generation of REM sleep, cortical activation, and the genesis of PGO waves. There is no single study that directly shows that the excitation of PPT cholinergic cells is causal to the generation of REM sleep, cortical activation, and the genesis of PGO waves.

In the present study we examined the hypothesis that the activation of PPT cells is causal to the generation of REM sleep and cortical activation. This was tested by microinjection of the excitatory amino acid l-glutamate into the PPT while simultaneously quantifying wake-sleep signs. Three different levels of activation of PPT cells were achieved by varying the concentration of injected l-glutamate to see whether different levels of activation cause different behaviors (Mayer and Westbrook 1987).

METHODS

Animals

Six male cats weighing 2.5–3.5 kg were used for these experiments. The animals were kept under regular lighting conditions (light on at 7:00 h and off at 19:00 h) in a temperature-controlled environment, with free access to food and water. Principles for the care and use of laboratory animals in research, as outlined by the National Institutes of Health (1985), were strictly followed.

Surgical procedure

Treatment of the animals and surgical procedures were in accordance with an approved animal welfare protocol. Cats were first sedated with acepromazine (0.1 mg/kg im) and then anesthetized with inhaled halothane (2.0–3.0% in O₂). With the use of sterile procedures, a standard set of recording electrodes was chronically implanted to objectively measure states of sleep and wakefulness (Ursin and Sterman 1981). Stainless steel bilateral guide tubes were also implanted to permit directed intracranial drug administration. Bipolar electrodes were implanted in the skull over the frontal and parietal cortices to record the electroencephalogram (EEG) and in the posterior orbital bone to record the electrooculogram (EOG). Multipolar electrodes were implanted stereotaxically in each lateral geniculate body (A6.5, L10.0, H2.0) (Berman 1968) to record the geniculate component of the PGO waves. Monopolar electrodes were implanted bilaterally in the dorsal cervical neck muscles to record the electromyogram (EMG). These four variables (EEG, EMG, EOG, and PGO waves) were used to score W and sleep stages. In addition, bilateral stainless steel guide tubes (26 gauge) were stereotaxically implanted 5 mm above the PPT (AP0.5 to −0.5, L3.5–4.0, H−2.0 to −2.5) (Berman 1968) as described previously (Datta et al. 1992). The tips of the guide tubes were left 5 mm above the targets to minimize cellular damage at the injection sites (Baghdoyan et al. 1993; Datta et al. 1992).

Postsurgical treatment

Cats were treated with butorphanol (0.4 mg/kg im) every 6 h for ≥24 h postsurgery to control any possible pain on recovery from anesthesia. To prevent possible infection due to surgery, the cats were also treated with amoxicillin (10 mg/kg sc) immediately after surgery and for 5 days postoperatively (5 mg/kg po).

Adaptation to recording conditions

Seven to 10 days after recovery from surgery, cats were adapted for 2–3 wk to the free-moving polygraphic recording conditions (size of recording cage: 4 × 2 × 2 ft) before microinjection sessions began. After ∼2–3 wk of adaptation, cats were able to sleep in the laboratory for recording sessions ≥4 h in duration (between 12:00 and 16:00 h). The endpoints of adaptation recording sessions were determined when animals showed a consistent REM sleep latency and total REM sleep duration for at least three consecutive recording sessions.

Intracerebral microinjections and experimental design

After the adaptation recording sessions, microinjection sessions began. The microinjection procedure has been previously described in detail (Baghdoyan et al. 1993; Datta et al. 1992). Briefly, microinjections were always made during wakefulness, and the cats were head restrained for the 3–5 min necessary for microinjection as previously described (Datta and Hobson 1994). Four-hour control recording sessions were begun after a single, unilateral microinjection of 0.25 μl saline into the PPT. All microinjections were made with a 32-gauge stainless steel injector cannula that was 5 mm longer than the implanted guide tube. On completion of the microinjection procedure, cats were released from head restraint and placed in the recording chamber, where they were free to move, eat, and drink. Polygraphic variables were recorded continuously for 4 h beginning at the onset of injection, and behavior was observed on a video monitor. Two days after the control recording, a single dose of monosodium l-glutamate (0.3, 1.0, or 3.0 μg in 0.25 μl sterile saline) was injected unilaterally into the control saline injection site with the use of the same injector cannula. After microinjections of glutamate, cats were placed in the recording chamber and polygraphic variables were recorded continuously for 4 h. Glutamate microinjections and recordings were made during the same 4-h period as control recordings (12:00–16:00 h) to quantify the glutamatergic stimulation effects at the same point in the circadian cycle. Each PPT site received one control saline and three glutamate microinjections in four different recording sessions 48 h apart. In six cats, a total of 12 PPT sites was microinjected (1 in the left and the other in the right side of each cat; for a summary of the sequence of injections see Table 1). In 4 of those 12 sites, glutamate microinjection doses were arranged in the following order: 0.3, 1.0, and 3.0 μg. In another four sites, the sequence of glutamate microinjection doses was 1.0, 3.0, and 0.3 μg. The other four sites received 3.0, 0.3, and 1.0 μg. In 6 of those 12 PPT sites, control saline microinection sessions were given first, with the second, third, and fourth microinjections being glutamate. In the other six PPT sites, the first, second, and third microinjections were glutamate and the fourth was control saline. At the end of all experimental sessions and before perfusion, with the use of the same injector used for glutamate microinjection, 0.25 μl of black ink was microinjected 1 mm dorsal to each injection site for localizing glutamate injection sites. In Fig. 1A, the arrow points to the ink injection mark. Thus the true position of the glutamate injection is 1 mm below the ink mark.

Scoring and analysis of behavioral states

Polygraphic data were scored for W, S state 1 (S1), S state 2 (S2), S with PGO waves (SP), and REM sleep according to standard criteria (Datta 1995; Datta and Hobson 1994, 1995; Ursin and Sterman 1981). All states were scored in 15-s epochs, yielding a total of 960 bins after each microinjection trial. The polygraphic and behavioral measures provided the following dependent variables that were quantified for each microinjection trial: 1) percentage of recording time spent in W, S1, S2, SP, and REM sleep per hour and 2) latency to onset of the first episode of REM sleep after the onset of injection. Statistical analyses were performed with the use of StatView statistical software (Abacus Concepts, Berkeley, CA, 1994). Analyses of variance and paired t-tests were used to examine the effects of microinjection of glutamate versus control saline on behavioral states of W, S1, S2, SP, and REM sleep.

Histological localization of injection sites

At the conclusion of the microinjection experiments, cats were deeply reanesthetized with pentobarbital sodium (50 mg/kg iv) and perfused transcardially with heparinized cold phosphate buffer (0.1 M, pH 7.4) followed by 4% paraformaldehyde in 0.1 M phosphate buffer. The brains were removed, and trimmed brain stem blocks were postfixed for 3–4 h in the paraformaldehyde phosphate buffer perfusion fluid. Trimmed brain stem blocks were then stored overnight in a solution of 0.1 M phosphate buffer and 5% sucrose as a cryoprotectant. The brains were then cut in the transverse plane in 40-μm-thick sections with the use of a freezing microtome. To visualize the cholinergic neurons in the PPT, sections werestained for nicotinamide-adenine dinucleotide phosphate (NADPH)diaphorase. Free floating sections were then incubated at 37°C in a solution made up of 50 ml of 0.1 M phosphate buffer (pH 7.4) containing 100 mg of NADPH, 45 mg Nitro Blue Tetrazolium, and 6 ml dimethyl sulfoxide (Sandell et al. 1986). The reaction was terminated after 30 min of incubation. After termination of the reaction, sections were washed in 0.1 M phosphate buffer and mounted on gelatin-coated glass slides. NADPH-diaphorase-stained sections were then counterstained with a 1% neutral red staining solution, dehydrated, cleared, mounted, and coverslipped with the use of Permount. All histological sections were examined under the microscope to localize the injection sites and the NADPH-diaphorase-positive PPT cells.

RESULTS

A total of 48 microinjections was made in 12 PPT injection sites. Histological identification showed that 8 of the 12 sites were placed within NADPH-diaphorase-positive cell compartments (Figs. 1 and 2), and thus were considered to be within the cholinergic cell compartments of the PPT. The remaining four injection sites were placed into NADPH-diaphorase-negative, noncholinergic compartments in the PPT. Only glutamate injections placed in the cholinergic compartment of PPT increased the cortical activated states of W and REM sleep. Because the four glutamate injections into noncholinergic PPT regions did not cause changes in W or REM sleep, those results were not included in the analyses of the data. The analysis below quantifies the effects of differing concentrations of glutamate injected into the PPT cholinergic cell compartment on the W, S, and REM sleep states.

Effects of glutamate on W state

After glutamate microinjections, behavioral and polygraphic signs of W (Fig. 3A) did not show any significant qualitative differences compared with behavioral and polygraphic signs of W following control saline injections (Fig. 3B). Quantitatively, 3.0-μg doses of glutamate applied into the cholinergic cell compartments of the PPT produced short, robust changes in the percentage of time spent in W compared with microinjections of control saline. Analysis across the entire duration of the recording sessions provides only a limited picture because of glutamate's short duration of effect. Therefore analyses were performed with the recording sessions divided into four 1-h periods. Because all microinjections were made during W, there was no measure of W latency. As shown in Fig. 4, microinjection of 0.3 and 1.0 μg glutamate into the cholinergic cell compartments of the PPT did not change the percentage of time in W compared with control microinjection of saline into the same sites. When the concentration of injected glutamate was increased to 3.0 μg, animals spent the first 2 h in W. Compared with the control microinjection, the glutamate-induced changes were highly significant both during the 1st h (df = 7; t = −13.62; P = 0.0001) and the 2nd h (df = 7; t = −42.48; P = 0.0001) of postinjection recording. During the 3rd h, W was 46.00 ± 5.45% (mean ± SD), but these changes were still significantly (df = 7; t = −6.69; P = 0.0003) higher than the 3rd h postcontrol saline injection value (27.50 ± 3.96%). Four hours after glutamate application, the W percentage (24.86 ± 3.18%) came back to its control level (28.75 ± 2.92%). Because for the first 2 h the threshold dose of glutamate (3.0 μg) saturated the percentage of time spent in W, the data on W were not subjected to further analysis of frequency and duration of individual episode.

Effects of glutamate on S

The results show that microinjection of glutamate into the cholinergic cell compartment of the PPT decreased and/or eliminated S for 2–3 h of postinjection recording. To demonstrate glutamate-induced changes in S, the percentage of time after glutamate application spent in S was compared with control measurements obtained after saline injection.

S1.

The beginning of S1 is recognized by the presence of spindling in the cortical EEG. The effects of microinjection of different doses of glutamate on the percentage of time spent in S1 are summarized in Fig. 5. A dose-dependent reduction in the percentage of time spent in S1 after glutamate microinjection was only evident during the 1st h of recording. The mean percentages of time spent in S1 after control injections of saline were 24.75 ± 5.82% (mean ± SD), 4.75 ± 1.83%, 26.25 ± 5.73%, and 23.50 ± 6.67% during postinjection recordings of the 1st, 2nd, 3rd, and 4th h, respectively. After the application of 0.3 μg glutamate, the S1 percentage during the 1st h dropped to 20.00 ± 4.90%, and this reduction was slightly significant (df = 7; t = 2.04; P = 0.08) compared with its control value. The percentage of S1 values during the 2nd, 3rd, and 4th h of recording were not significantly different when compared with their control values. Similarly, 1.0 μg glutamate significantly reduced the percentage of time spent in S1 (5.23 ± 1.28%; df = 7; t = 10.17; P = 0.0001) compared with the 1st h control value. Again, postinjection values for the percentage of S1 during the 2nd, 3rd, and 4th h after application of 1.0 μg glutamate were not significantly different when compared with the control values. As the dose of glutamate microinjection was increased to 3.0 μg, S1 sleep was completely eliminated during the first 2 h of recording. But, during the 3rd and 4th h of postinjection recording, S1 sleep returned to its control level.

S2.

The deep stage of S in the cat is called S2. During this stage cortical EEG shows high-amplitude, low-frequency (synchronized) waves. The effects of microinjection of different doses of glutamate on the percentage of time spent in S2 are summarized in Fig. 6. During the control recording period, the percentage of time spent in S2 increased slowly throughout the 4 h of recordings. The control percentages of S2 for the 1st, 2nd, 3rd, and 4th h of recording were 4.13 ± 1.64%, 18.00 ± 2.88%, 20.88 ± 3.36%, and 28.75 ± 2.92%, respectively. After microinjection of 0.3 μg glutamate, the percentage of time spent in S2 was reduced during the 1st h (0.63 ± 1.77%) and 2nd h (11.38 ± 1.69%) of recording. The reductions of S2 values were significantly lower than their control values during the 1st h (df = 7; t = 5.58; P = 0.0008) and 2nd h (df = 7; t = 4.96; P = 0.0016) of recording. During the 3rd and 4th h of recording, the percentages of time spent in S2 after application of 0.3 μg glutamate were similar to the control values. As the glutamate dosage increased to 1.0 μg, the percentage of time spent in S2 was reduced to zero for the 1st and 2nd h of recordings. During the 3rd h, the percentage of time in S2 (13.75 ± 2.32%) was also significantly (df = 7; t = 5.31; P = 0.0011) less than its control value. But during the 4th h, the S2 percentage was comparable with its control value. When the glutamate dose was further increased to 3.0 μg, the percent of time spent in S2 was nil for the first 2 h. During the 3rd h (15.25 ± 3.24%) and 4th h (22.75 ± 3.73%) of recording, the percentages of time spent in S2 were reduced compared with the control values. These reductions in S2 were significant both during the 3rd h (df = 7; t = 3.35; P = 0.012) and 4th h (df = 7; t = 4.09; P = 0.0046) of recording compared with control values.

SP.

SP is a transitional stage from S to REM sleep. During this stage cortical EEG is less synchronized compared with the S2 phase. The effects of microinjection of different doses of glutamate on the percentage of time spent in SP are summarized in Fig. 7. During control recording periods, the percentages of SP for the 1st, 2nd, 3rd, and 4th h of recording were 1.38 ± 1.18%, 6.63 ± 2.50%, 6.86 ± 1.89%, and 5.88 ± 1.55%, respectively. After microinjection of 0.3 μg glutamate, the percentage of time spent in SP (3.88 ± 1.13%) was increased (df = 7; t = −5.40; P = 0.001) significantly during the 1st h of recordings compared with its control value. But for the next 3 h of recording, the percentages of time spent in SP were comparable with control values. With increased dosages of glutamate (1.0 and 3.0 μg), the percentage of time spent in SP was reduced to zero for the 1st and 2nd h of recording. For 1.0 μg glutamate, the SP percentage during the 3rd h (3.88 ± 1.25%) was significantly (df = 7; t = 6.0; P = 0.0005) less compared with its control value. The SPvalues in hour 4 after injection of 1.0 μg glutamate,and in hours 3 and 4 after injection of 3.0 μg glutamate, were not significantly different compared with controlvalues.

Effects of glutamate on REM sleep

The changes in the percentage of time spent in REM sleep after microinjection of different dosages of glutamate are summarized in Fig. 8. Glutamate, at lower concentrations, enhanced REM sleep during the first 2 h of postinjection recordings, whereas at the higher concentrations it eliminated REM sleep by enhancing wakefulness.

Microinjections of 0.3 μg glutamate facilitated the first episode of REM sleep (latency 38.01 ± 5.85 min, mean ± SD), compared with a REM sleep latency of 52.83 ± 11.56 min in saline controls (df = 7; t = 5.00; P = 0.0016). In addition to reduced REM sleep latency, 0.3 μg glutamate increased the mean duration of REM sleep (4.04 ± 0.30 min), compared with a REM sleep duration of 1.94 ± 0.27 min after control saline injections (df = 7; t = −12.053; P = 0.0001). During the first 2 h postinjection, the numbers of REM sleep episodes after injection of either 0.3 μg glutamate (2.38 ± 0.74) or control saline (2.63 ± 0.74) did not differ significantly (df = 7; t = 0.683; P = 0.517). This reduced latency and increased duration of REM sleep episodes increased the total percentage of time spent in REM sleep during the first 2 h after injection of 0.3 μg glutamate. During the 1st h after application of 0.3 μg glutamate, REM sleep percentages increased to 10.5 ± 2.07% (mean ± SD) compared with control saline (1.38 ± 1.06%; df = 7; t = −9.974; P = 0.0001). In the 2nd h, the increased REM sleep percentage (30.00 ± 2.78%) remained significantly higher (df = 7; t = −12.204; P = 0.0001) compared with control saline (19.75 ± 2.66%). During the 3rd h (18.38 ± 2.33%) and 4th h (18.50 ± 1.51%) of recordings, the REM sleep percentage did not change significantly compared with the control saline.

When the glutamate dosage was increased to 1.0 μg, the latency of REM sleep was further reduced to 24.54 ± 3.55 min. This reduction was highly significant (df = 7; t = 7.623; P = 0.0001) compared with REM sleep latency after control saline injection. Increased glutamate dosages also increased the mean duration of the REM sleep events to 5.91 ± 0.34 min (df = 7; t = −26.553; P = 0.0001). Like 0.3 μg glutamate, 1.0 μg glutamate did not change the frequency of REM sleep episodes (2.5 ± 0.54) compared with control saline (2.63 ± 0.74; df = 7; t = 0.552, P = 0.598). An increased glutamate dosage of 1.0 μg also increased the total percentage of time in REM sleep during the 1st, 2nd, and 3rd h to 24.88 ± 4.67% (df = 7; t = −14.655; P = 0.0001), 34.13 ± 3.48% (df = 7; t = −11.347; P = 0.0001), and 34.25 ± 6.96% (df = 7; t = −5.349; P = 0.0011), respectively. During the 4th h postinjection, the total percentage of time spent in REM sleep (14.13 ± 3.04%) was comparable with control values (18.14 ± 2.03%).

In contrast, a higher dosage of glutamate (3.0 μg) produced the opposite type of REM sleep response compared with lower doses (0.3 and 1.0 μg). Microinjection of 3.0 μg glutamate suppressed REM sleep completely for the first 2 h of the postinjection periods. During the 3rd h postinjection, REM sleep began to reappear, but the total percentage of REM sleep (9.14 ± 5.28%) was significantly less (df = 7; t = 3.935; P = 0.0058) compared with its control saline value. During the 4th h, this REM sleep percentage (20.13 ± 5.44%) returned to its control level.

Figure 9 shows representative polygraphic signs of REM sleep after control saline (Fig. 9A) and glutamate microinjections (Fig. 9B) into the cholinergic cell compartment of the PPT. Behavioral and polygraphic signs of REM sleep after glutamate application resembled those signs during REM sleep after control saline microinjections. During REM sleep, cortical EEG is activated (desynchronized), and is similar to a cortical EEG during W.

DISCUSSION

The principal findings of this study are that 1) microinjection of low and medium dosages of the excitatory amino acid l-glutamate into the cholinergic cell compartment of the PPT increases the total amount of REM sleep, and 2) a higher dosage of l-glutamate increases the total amount of wakefulness when injected into the same site. As a consequence of this glutamate-induced increase in wakefulness and REM sleep, S is suppressed and/or eliminated. The results show for the first time that the excitation of the cholinergic cells of the PPT causes wakefulness and/or REM sleep depending on levels of activity relative to baseline. The results presented here strengthen and extend the hypothesis that the activity of PPT cholinergic cells is important for the generation of REM sleep as well as wakefulness (Datta 1995, 1997).

Methodological considerations

We chose to study freely moving unanesthetized cats under chronic conditions because this preparation provides for the most physiological approach to a longitudinal analysis of the behavioral and electrographic events related to the wake-sleep cycle (Siegel 1979; Siegel and McGinty 1977). After the habituation period of each experimental condition, the cats demonstrated regular values of W and sleep stages. Thus pharmacological manipulations that affected electrographic signs of wake-sleep patterns could be evaluated accurately.

Current concepts of brain stem cholinergic cells' involvement in behavioral state control are based in large part on spontaneous activity patterns of PPT cells across the wake-sleep cycle (Datta 1995). Modulation of these PPT cholinergic cells' activity correlates highly with changes in behavioral state, but is not definitive evidence of causality. Barring more direct evidence, the correlative behavior of PPT cholinergic neurons across states cannot be considered as causative with complete confidence. In this study, we used local microinjections of l-glutamate to increase the activity of cells located within the cholinergic cell compartments of the PPT. In the study of the structure-function relationship at the system level, cerebral microinjection of the excitatory amino acid l-glutamate has certain advantages over traditional lesion and electrical stimulation techniques, especially in the study of the wake-sleep cycle. Local application of l-glutamate excites mostly cell bodies of that particular diffusion site, but electrolytic lesion and electrical stimulation includes not only a larger area, it also includes processes of the cell body and fibers passing through that particular region (Mayer and Westbrook 1987; Sakai et al. 1976). The level of excitation of a population of neurons (multiunit activity) could be changed in a dose-dependent manner by changing the concentration of injected l-glutamate (Mayer and Westbrook 1987). Thus l-glutamate application to excite particular cell groups that affect wake-sleep patterns could be accurately evaluated for causality.

A major limitation of the l-glutamate microinjection method relates to the diversity in the neurochemical nature of the neuronal population affected by the application (Mayer and Westbrook 1987). l-glutamate is a nonspecific excitatory amino acid that excites all types of neuronal cells with different neurochemical signatures. In this study, we injected glutamate into the part of the PPT where most cells are cholinergic (Jones and Beaudet 1987; Shiromani et al. 1988; Vincent and Reiner 1987). But we acknowledge that if there are noncholinergic cells within those cholinergic cell groups, they would also be excited by the application of glutamate.

Anatomic considerations

With the use of specific monoclonal antibodies to choline acetyltransferase (ChAT), the laterodorsal tegmental nucleus and PPT have been shown to be the major aggregations of cholinergic neurons in the mammalian brain stem (Armstrong et al. 1983; Jones and Beaudet 1987; Kimura et al. 1981; Mesulam et al. 1983, 1984, 1989; Satoh and Fibiger 1985; Shiromani et al. 1988; Vincent and Reiner 1987). All of these ChAT-positive cells also displayed intense NADPH-diaphorase activity (Vincent et al. 1983, 1986). Thus mesopontine NADPH-diaphorase-positive cells are unequivocally considered to be the cholinergic cells (Luebke et al. 1992, 1993; Steriade et al. 1990; Vincent et al. 1983, 1986; Williams and Reiner 1993). All of our eight positive injection sites were located within the cluster of NADPH-diaphorase-positive cells of the PPT at stereotaxic levels A0.5–P0.5. At this stereotaxic level 95% of those cells are cholinergic (Reiner and Vincent 1987; Steriade et al. 1990; Webster and Jones 1988). Four other injection sites were close to these positive sites but were not effective in changing electrographic events related to the wake-sleep cycle. In fact, in these injection sites, very few cells were NADPH-diaphorase positive. These results indicate that electrographic changes in the wake-sleep cycle observed after glutamate microinjection into the PPT were mainly due to the excitation of cholinergic cells.

Role of the PPT in REM sleep generation

The results of the present study demonstrate that the microinjection of l-glutamate into the cholinergic cell compartments of the PPT increases REM sleep in a dose-dependent manner. These results are consistent with the hypothesis that the activity of mesencephalic cholinergic neurons is involved in the generation of REM sleep (Datta 1995).

Histological identifications of glutamate injection sites were shown to be within the cluster of the cholinergic cell compartment of the PPT. We suggest that the glutamatergic stimulation of those cells caused release of ACh. In support of our suggestion, one earlier electrical stimulation study has shown that the stimulation of PPT cells in an anesthetized preparation releases ACh (Lydic and Baghdoyan 1993).

In the last decade, evidence has accumulated indicating that each of the events of REM sleep is generated by distinct cell groups in the brain stem (reviewed in Datta 1995). Although these cell groups are distinct, they are components of a widely distributed network (Datta and Hobson 1994, 1995; Jones 1991; Morrison 1979; Sakai 1985; Vertes 1984). Cortical EEG desynchronization of REM sleep is controlled by the peribrachial area; muscle atonia by nucleus locus coeruleus-alpha; rapid eye movements by the rostral periabducens reticular formation; PGO waves by the caudolateral peribrachial area; the hippocampal theta rhythm by the pontis oralis; muscle twitches by the nucleus giganto cellularis, especially the caudal part; and increases in brain temperature and cardiorespiratory fluctuations by the parabrachial nucleus (Bertrand and Hugelin 1971; Datta 1995, 1996; Datta and Hobson 1994, 1995; Datta et al. 1989; El-Mansari et al. 1989; Gottesmann 1992; Jones 1991; Morrison 1979; Sakai 1985; Shouse and Siegel 1992; Steriade et al. 1990; Vertes 1984). We suggest that the glutamate induced low to medium levels of excitation of the PPT cholinergic cells, resulting in release of ACh into all of the above REM-sleep-sign-generating structures of the brain stem.

Neuroanatomic studies in which retrograde tracers were used have shown that the axons of PPT cholinergic cells project to the REM-sleep-sign-generating structures of the brain stem (Jones 1991; Leichnetz et al. 1989; Mitani et al. 1988; Satoh and Fibiger 1986; Semba 1993; Semba et al. 1990; Shiromani et al. 1988). These anatomic connectivity studies indicate that the PPT cholinergic cells could release ACh to the cholinoceptive REM-sleep-generating structures of the brain stem. The present study also suggests that the glutamatergic stimulation of the PPT cholinergic cells excites REM-sleep-sign-generating cells by releasing ACh and ultimately induces REM sleep. In support of this suggestion, several studies have shown that the local microinjection of cholinomimetics into the many discrete parts of the pons evoke a state that is behaviorally and electrophysiologically similar to REM sleep (Baghdoyan et al. 1993; Baxter 1969; Chase and Morales 1990; Datta et al. 1992; Gnadt and Pegram 1986; Mitler and Dement 1974; Vanni-Mercier et al. 1989; Velazquez-Moctezuma et al. 1989). Also, it has been shown that each of these REM sleep signs could be triggered individually by microinjection of a cholinergic agonist into its respective triggering structure in the brain stem (Datta 1996; Datta et al. 1991, 1992; Magherini et al. 1971; Nunez et al. 1991; Reid et al. 1994; Vertes et al. 1993).

Another line of evidence supporting the notion that the PPT neurons are involved in the generation of REM sleep comes from extracellular single-cell recording studies (Datta 1995; Datta et al. 1989; El-Mansari et al. 1989; Kayama et al. 1992; Steriade et al. 1990). These studies recorded the activity of a group of PPT cholinergic neurons that increased their firing rates during REM sleep. The results of the present study are direct evidence implicating the excitation of PPT cholinergic cells in the generation of REM sleep.

On the basis of single-cell recording studies (Datta 1995; El-Mansari et al. 1989), we expected to see some changes in PGO wave density after glutamatergic excitation of the PPT cells. To our surprise, we did not observe any differences in PGO wave density or any state-independent PGO waves after glutamatergic stimulation. This observation suggests that activation of PPT cells alone is insufficient to induce PGO waves; however, it seems clear that these cells do play an important role in PGO wave generation and modulation (Datta 1997; Datta and Hobson 1994).

Role of the PPT in wakefulness

At present there are only two lines of evidence implicating involvement of the PPT cholinergic cells in wakefulness. The first comes from extracellular single-cell recording studies. One group of PPT cholinergic cells is active both during wakefulness and REM sleep (Datta 1995; Datta et al. 1989; El-Mansari et al. 1989; Saito et al. 1977). These cells have a much higher firing rate during active W than during REM sleep. The second line of evidence comes from local microinjection studies. Microinjection of a cholinergic agonist into the locus coeruleus excites noradrenergic cells and activates cortical EEG (Berridge and Foote 1991). From these studies, we hypothesized that the maximal level of activity of the mesencephalic cholinergic neurons would induce wakefulness (Datta 1995). The results of the present study demonstrate that a higher dose of l-glutamate microinjection into the cholinergic cell compartments of the PPT induces wakefulness. This finding directly supports the hypothesis that the maximal level of activity of the cholinergic neurons induces and sustains wakefulness.

To explain the observation, we suggest that the high dose of l-glutamate maximally activates PPT cholinergic cells. Maximal excitation of cholinergic cells released ACh to the locus coeruleus and other structures involved in wakefulness. It is already known that the locus coeruleus receives afferent inputs from the PPT cholinergic cells (Jones 1991; McBride and Sutin 1976; Sakai et al. 1977). It is also known that the cholinergic agonist directly excites locus coeruleus cells (Berridge and Foote 1991). Excitation of the locus coeruleus noradrenergic cells causes inhibition of REM-sleep-sign-generating cells and ultimately causes wakefulness (Aston-Jones and Bloom 1981; Hobson et al. 1975; Karczmar et al. 1970; McCarley and Hobson 1975). Indeed, noradrenergic agonists suppressed REM sleep by inhibiting REM-sleep-sign-generating cells (Bier and McCarley 1994; Tononi et al. 1989, 1991; Williams and Reiner 1993).

Role of the PPT in cortical activation

Cortical activation is a prerequisite for perceptual processes and purposeful behaviors. The replacement of high-amplitude synchronized EEG waves by low-voltage, high-frequency desynchronized waves in the neocortex is now recognized as a reliable indicator of cortical activation (Berridge and Foote 1991; Buzsaki et al. 1988; Jacobs 1986; Jasper and Carmichael 1935; Moruzzi and Magoun 1949; Rheinberger and Jasper 1937; Steriade et al. 1990; Vanderwolf 1988; Whalen et al. 1994). Cortical activation is a striking characteristic of wakefulness and REM sleep (Aserinsky and Kleitman 1953; Dement and Kleitman 1957). The cellular, molecular, and network mechanisms for cortical activation are different during wakefulness and REM sleep (reviewed and modeled in Datta 1995).

Given the long-term association of ACh with neocortical activation or arousal (Celesia and Jasper 1966; Jasper and Tessier 1971; Longo 1966; Szerb 1967; Wikler 1952), very little research has been directed toward the contribution of the brain stem cholinergic system to neocortical activation (Datta 1995; Steriade et al. 1990; Williams et al. 1994). Single-cell recording studies in behaving cats have shown that a group of PPT cholinergic cells is active both during wakefulness and REM sleep. These cells' firing rates began to increase 20–60 s before the onset of both wakefulness and REM sleep in the transition from S (reviewed in Datta 1995). The increased neuronal activity of the PPT cholinergic cells that heralds cortical (EEG) activated states (wakefulness and REM sleep) implicates these neurons as causal elements for the cortical activation. By demonstrating that glutamatergic stimulation of PPT cholinergic cells increases cortically activated states of wakefulness and REM sleep by suppressing S, the present study supports the idea that PPT cholinergic cells are causal to the cortical activation.

Limitations and future studies

The present data cannot address the involvement of specific glutamate receptor subtypes by which PPT cholinergic cells were stimulated to produce their effects on sleep and wakefulness. The excitatory amino acid glutamate is a mixed agonist for the N-methyl-d-aspartate (NMDA) and non-NMDA receptors (Mayer and Westbrook 1987). The present results point to the need for future studies to specify the receptor subtypes involved in the glutamatergic excitation of the PPT cholinergic cell.

Another limitation, which can be confirmed and quantified by future studies, is the amount of ACh release in the locus coeruleus and REM-sleep-sign-generating structures of the brain stem after glutamate microinjection into the PPT cholinergic cell compartments. The present results also encourage the measurement of norepinephrine levels in the REM-sleep-sign-generating structures of the brain stem after microinjection of a high dose of l-glutamate into the cholinergic cell compartments of the PPT.

Despite the above limitations, our results show for the first time that the increased activity of cholinergic cells of the PPT is causal for the generation of REM sleep as well as for wakefulness. The present study also contributes, currently, the most direct evidence for the role of the brain stem cholinergic system in cortical activation.