Reassessment of the structural basis of the ascending arousal system

Animals

Pathogen free adult male Sprague Dawley rats (275–300 grams, Harlan) were housed in individual cages. The cages were housed inside isolation chambers, which provided ventilation, computer-controlled lighting (12:12 light-dark cycle, lights on at 0700; 200 lux), an ambient temperature of 22 ± 1°C, and visual isolation. Care of the rats in the experiment met National Institutes of Health standards, as set forth in the Guide for the Care and Use of Laboratory Animals and all protocols were approved by the BIDMC and Harvard Medical School Institutional Animal Care and Use Committees.

Overall Experimental Design

In the first part of the present study, we determined the effect of cell-body specific lesions of the thalamus or BF on the cortical EEG, including changes in sleep-wake regulation. Briefly, cell-body specific lesions were placed in the thalamus by injecting 50 nl of a 10% solution of ibotenic acid bilaterally (see below). Lesions of the basal forebrain were done by injecting a 0.1% solution of either IgG192-saporin or orexin-saporin at four different sites (see below). EEG/EMG electrodes were implanted at the same surgery, and continuous recordings began on day 7 postoperatively. At the end of the study, animals were behaviorally stimulated (see below) for two hours, then anesthetized and killed by formaldehyde perfusion, and the brains were removed and processed for c-Fos immunohistochemistry. For animals that were spontaneously active, behavioral stimulation consisted of placing two cages together and removing the tops. Animals would jump back and forth between the cages, and maintained spontaneous wakefulness, even during the light period, for two hours from 10:00 to 12:00. In cases where animals were not spontaneously active, they were stimulated by the experimenter by gentle handling and stroking with a paintbrush for the two hour period.

In the second part of the study, we used cholera toxin subunit B (CTB) to retrogradely trace inputs to the BF and thalamus from sites in the brainstem to define the cell groups whose activity is important for maintaining a normal waking state. We combined this with in situ hybridization for the vesicular glutamate 2 transporter (VGLUT2), to determine which of these cells were likely to be glutamatergic.

In the third part of the study, we used local injections of orexin-saporin to ablate neurons in the parabrachial nucleus and precoeruleus region, which had been found to represent a major brainstem source of glutamatergic inputs to the basal forebrain. Again, animals were implanted with EEG/EMG electrodes, and continuous recordings began seven days post-operatively. At the end of the experiment, we also examined c-Fos immunoreactivity in the brains of these animals after two hours of behavioral stimulation, as in the first part of the study.

EEG/EMG and sleep recording

The surgery, EEG/EMG data collection and analyses have been described in detail previously (Lu et al., 2006a and Lu et al., 2006b). In brief, after animals were anesthetized with chloral hydrate (7% in saline, 350 mg/kg), the skulls were exposed. Four EEG screw electrodes were implanted into the skull, in the frontal (two) and parietal bones (two) of each side, and two flexible EMG wire electrodes were placed into the neck muscles. The free ends of the leads were soldered into a socket that was attached to the skull with dental cement, and the incision was then closed by wound clips. On post-surgical day 3, the sockets were connected via flexible recording cables and a commutator to a Grass polygraph and computer. From day 7 until the completion of each experiment, signals were recorded continuously. Signals were digitized using an Apple Macintosh computer running the ICELUS sleep recording system (M. Opp, University of Michigan, Ann Arbor, MI). EEG signals were filtered to exclude frequencies <0.5 or >35 Hz, whereas EMG signals were filtered to exclude frequencies <10 or >35 Hz. Briefly, NREM sleep was identified by a preponderance of high-amplitude, low-frequency (<4 Hz) EEG activity and relatively low and unchanging EMG activity, whereas wakefulness was characterized by a preponderance of low-amplitude, fast EEG activity and highly variable muscle tone on EMG. REM sleep was identified by very low EMG activity and a low-amplitude monotonous EEG containing a predominance of theta range (4–7 Hz) EEG activity. Scoring was done before histological examination and so the scorers were unaware of the extent of the lesions.

Wake–sleep states were manually scored in 12 s epochs on the digitized EEG/EMG. The amount of time spent in wake, NREM sleep, and REM sleep was determined for each hour. Twenty-four hour fast Fourier transform (FFT) power spectrum plots for all animal groups were also generated. Additional FFT analysis was performed on several short representative EEG epochs (ca. 12 sec).

Tracer and neurotoxin injections

Under chloral hydrate anesthesia (7% in saline, 350 mg/kg), a burr hole was made and a fine glass pipette (1 mm glass stock, tapering slowly to a 10–20 μm tip) containing 1.0 % CTB (List Biological), 10% ibotenic acid, or 0.1% orexin saporin (OX-SAP) or IgG192-saporin (192-SAP) (Advanced Targeting Systems) was lowered to the pre-determined targets (Paxinos and Watson, 1986). Coordinates for the BF include: magnoceullar preoptic nucleus (MCPO): AP=−0.36 mm, DV = 7.5mm, ML = ±3.0 mm; substantia innominata (SI; corticopetal neurons within the SI are also called the nucleus basalis): AP = −1.08 mm, DV = 7.4 mm, ML = ±3.0 mm; medial septal-diagonal band complex (MS): AP= 0.84 mm, DV= 6.6 mm, ML= 0 mm. Coordinates for the thalamus: AP = 3.15 mm, DV = 5.4 mm, ML = ± 1.5 mm. Injections were made slowly (over five minutes) by gentle puffs of air pressure, and the pipette was left in place for five minutes after the injection was completed. Incisions were closed with wound clips and animals were allowed to recover for 7 days.

In the present study, cell-body specific neural lesions were generated using either an excitotoxin (ibotenic acid) or saporin-based immunotoxins (orexin-saporin or IgG192-saporin). While ibotenic is an excellent toxin for areas of the brain that receive a lot of glutamatergic inputs, such as the thalamus, it has two important drawbacks in other areas of the brain, such as the basal forebrain, hypothalamus and brainstem. First, ibotenic acid does not necessarily kill all cell groups (e.g., the tuberomammillary nucleus is resistant to it), and second it may cause excitatory autonomic responses that can compromise circulation or breathing. Orexin-saporin on the other hand has the advantage of causing a slower, non-excitotoxic cell death, allowing animals to survive lesions that might otherwise have lethal side effects if done by ibotenic acid. In addition, neurons are differentially susceptible to orexin-saporin (for example, suprachiasmatic neurons are highly resistant to it; Lu and Saper, unpublished observations). In our experiments, orexin-saporin differentially killed non-cholinergic basal forebrain cells at lower concentrations, but killed all basal forebrain cells at higher concentrations; see Results. By contrast, IgG192-saporin kills only cholinergic cells in the basal forebrain (Smith and Pang, 2005).

Histology

Animals were killed by deep anesthesia with 500 mg/kg of chloral hydrate, followed by perfusion through the heart with 50 ml saline, followed by 250 ml of neutral phosphate buffered formalin (Fischer Scientific Co.). Brains were removed, incubated in 20% sucrose at 4°C until they sank, and then sectioned at 40 μm on a freezing microtome in four series. Sections were washed in phosphate buffered saline (PBS, pH 7.4), and incubated in primary antiserum (antibody characterization below) diluted in PBS containing 0.3% Triton X-100 and 0.2% sodium azide for one day at room temperature. Sections were then washed in PBS and incubated in biotinylated secondary antiserum (against appropriate species IgG, 1:1000, Vector) in PBS for one hour, washed in PBS and incubated in ABC reagents for 1 hour. Sections were then washed again and incubated in a solution of 0.06% 3,3-diaminobenzidine tetrahydrochloride (DAB, Sigma) and 0.02% H₂O₂. The sections were stained brown with DAB (ChAT) or black (c-Fos) by DAB plus 0.05% cobalt chloride and 0.01% nickel ammonium sulfate.

Antibody characterization

Table 1 provides a list of all antibodies used in the present study. The rabbit polyclonal Fos antibody was raised against a synthetic peptide including residues 4–17 from human c-Fos. This antibody stained a single band of 55 kD m.w. on Western blots from rat brain (manufacturer’s technical information). c-Fos staining with the Ab5 antiserum is found in many CNS structures (Gooley et al., 2006; Lu et al., 2006) only when neurons within these structures have recently been physiologically stimulated.

The rabbit polyclonal antibody against orexin A was raised against the orexin peptide Pyr-Pro-Leu-Pro-Asp-Cys-Cys-Arg-Gln-Lys-Thr-Cys-Ser-Cys-Arg-Leu-Tyr-Glu-Leu-Leu-His-Gly-Ala-Gly-Asn-His-Ala-Ala-Gly-Ile-Leu-Thr-Leu-NH2. On immunoblots from rat brain, it recognizes a pair of bands representing the orexin peptides at about 3.5 kD m.w. (manufacturer’s technical information). The pattern of cellular morphology and distribution that is stained is limited to the lateral hypothalamus and is identical to previous reports (Chou et al., 2003). Adsorption controls have been reported previously to result in elimination of staining by the antibodies that we used against orexin-A (Chen et al., 1999).

The goat polyclonal antibody against ChAT was raised against human placental enzyme, and only a single band of appropriate molecular weight was identified by immunoblot in mouse NIH/3T3 cell lysates (manufacturer’s technical information). In the basal forebrain and the mesopontine tegmentum, it stains a pattern of cellular morphology and distribution identical to previous reports (Chou et al., 2002).

The mouse monoclonal antibody against Tyrosine Hydroxylase (TH) was raised against TH isolated and purified from rat PC12 cells. Adsorption controls have been reported previously to result in elimination of staining by the antibodies that we used against TH (Gonzales-Cuello et al., 2004).

The goat polyclonal antibody against Cholera toxin subunit B was raised against CTB and the specificity of immunostaining for CTB was indicated by the lack of detectable immunostaining in uninjected rats.

For all secondary antibody immunohistochemical controls, the primary antibodies were omitted and the tissue showed no immunoreactivity above background.

Double-staining of VGLUT2 mRNA by in situ hybridization and cholera toxin subunit B (CTB) by immunohistochemistry

We used a method we have previously described (Chou et al., 2002; Lu et al., 2006; Lu et al., 2002) to label CTB with an immunoperoxidase brown color and then VGLUT2 mRNA by autoradiography. Briefly, sections (40 μm thickness) were first stained immunohistochemically for CTB (brown DAB reaction), as above, except that all solutions were DEPC treated to destroy RNAase. The sections were then acetylated and hybridized overnight (55 °C) with a ³⁵S-labeled cRNA probe synthesized from a plasmid containing the complete coding sequence of VGLUT2 mRNA. After a succession of one hour washes (2xSSC/1mM DTT, 50 °C; 0.2xSSC/1mMDTT, 55 °C; 0.2xSSC/1mM DTT, 60 °C), the tissue was treated with RNAase-A (Boehringer-Mannheim, Indianapolis) and washed under conditions of increasing stringency, including a 30 min wash at 60°C in 0.1× SSC. The tissue was then dehydrated in alcohol and air-dried. The sections were exposed to X-ray film (Eastman-Kodak) for 2–3 days; the slides were then dipped in Kodak NTB-2 emulsion and exposed for one month. Slides were developed in Kodak D-19, fixed, and then dehydrated and cover-slipped.

Cell profile counting

Cell profile counts were used both to assess completeness of lesions (counts of all remaining neurons in a field in three sections per region), and to measure the number of Fos-immunoreactive neurons in selected brain regions after behavioral stimulation. In all cases, cell profiles were counted only when there was a clearly distinguishable nucleus. In the case of lesions, the nucleus appeared as a clear area in either a Nissl or immunostained neuron. In the Fos studies, the nucleus was stained black. Lesions of BF cholinergic neurons were determined by counting remaining ChAT-ir cells in all the BF containing sections, and comparing results to controls. As the same tissues were Nissl counterstained, we were able to count remaining non-cholinergic cells as well. For counting non-cholinergic neurons, the ChAT-ir field (in corresponding sections from unlesioned brains) was used to mark the counting region. The orexin field in the lateral hypothalamus was similarly localized from immunostained sections. The PC and the PB, as well as the ventrolateral tuberomammillary nucleus (TMN) and locus coeruleus (LC) counting fields were defined by their location in Nissl counterstained sections. Profiles containing Fos-ir nuclei were counted in two consecutive (1:4) sections through the center of the cell group for the PC and four sections for the PB, TMN, lateral hypothalamic orexin neurons (OX) and LC on both sides of the brain. As the level of c-Fos expression was similar across the cerebral cortex and in the same laminar pattern as in non-lesioned rats during normal wakefulness, we chose to quantify Fos expression in the cingulate cortex (CC) in the control animals and those with thalamic lesions (Figure 1). For the CC, a 0.3 × 0.3mm counting box was placed on one side of the brain with one border parallel to and along the edge of layer 1 at the level of the genu of the corpus callosum, which included the full thickness of layers I–VI, where the Fos-immunoreactive neurons were found. The profile counts were then corrected by the Abercrombie equation, where the corrected count N = n×T/(D+T) where n is the uncorrected count, T is thickness of the tissue section, and D is mean nuclear diameter. Brightness and contrast of images were optimized for visualization purposes, and figures were created in Adobe Photoshop and Illustrator.

Statistical Analysis

We analyzed the percent time spent in wake and sleep states and the Fos expression in the brain by using an ANOVA. Post-hoc testing (Tukey’s HSD) was used for individual mean comparisons where appropriate. Because all data conformed to the requirements of the ANOVA model, we did not need to take any remedial measures for transforming the response variables. An α< 0.05 was considered significant.

Role of thalamus and basal forebrain in cortical EEG

The parabrachial nucleus (PB) and precoeruleus_area (PC) provide critical neuronal inputs for maintaining cortical arousal

Role of the thalamus vs. BF in the control of cortical arousal

The functional integrity of the thalamus has been widely considered to be necessary for generating the EEG correlates of cortical arousal and a wakeful state (Steriade et al., 1993; Llinas and Steriade, 2006). It is likely that the thalamus, which transmits a wide range of sensory inputs to the cerebral cortex, is necessary for normal cognitive function. On the other hand, a number of previous studies have reported evidence that the integrity of the thalamus is not necessary to maintain an aroused cortical state in rats or cats (Starlz et al., 1951; Villablanca and Salinas-Zeballos, 1972; Vanderwolf and Stewart, 1988; Buzsaki et al., 1988). Other studies showed that infusion of lidocaine into the BF, but not the reticular or intralaminar thalamic nuclei, strongly attenuated the cortical arousal effects (desynchronized EEG) elicited by electrical stimulation in the PPT region or by tail pinch in urethane-anesthetized rats (Dringenberg and Olmstead, 2003). Injection of procaine into the BF caused the behavioral and EEG appearances of NREM sleep, whereas injections of the glutamate agonists, either AMPA or NMDA, caused immediate and prolonged wakefulness (Cape and Jones, 2000). These results strongly argue that activating influences from the brainstem may reach the neocortex via an extra-thalamic, BF route.

Despite this evidence, the role of the thalamus in cortical arousal has remained controversial. For example, Buzsaki and colleagues (1988) proposed that low-frequency cortical EEG observed after lesions of the BF was due to loss of BF projections to both the neocortex and to the thalamus. However, in our experiments the Fos expression (following behavioral stimulation) in the thalamus was comparable in control animals and those with BF lesions, suggesting that the thalamus does not necessarily depend upon the BF for its activation nor was it capable of activating the cortex without the BF input. In this view, the thalamus would transmit specific information that provides the content of the waking state and may therefore contribute to overall arousal if thalamic input demands attention, but the thalamus is neither necessary to produce wakefulness nor sufficient to produce it without the BF.

Because thalamic relay neurons fire in patterns that correlate with cortical EEG (Steriade et al., 1993; Llinas and Steriade, 2006), it is often assumed that the thalamus plays a causal role in entraining the frequency of the EEG. The results of the present study suggest this is not the case for EEG frequencies below 35 Hz (the upper end of the range that our EEG recordings could measure) because elimination of thalamic inputs does not appear to prevent the normal sleep or wake EEG patterns (with the exception of EEG spindles) in the corresponding cortical areas. It is possible that the thalamus may be critical for driving high frequency cortical rhythms, but this will require further investigation. However, our experiments suggest that it is likely that the correlation between thalamic neuronal firing and the cortical EEG in the 1–35 Hz range results from cortical activity driving thalamic firing, rather than vice versa. Another possibility is that the small amount of remaining thalamus in our study (mainly lateral and medial geniculate nuclei) could be entraining the visual and auditory cortex, and that these could in turn entrain the EEG rhythms of the entire remaining cortex. We think this possibility is unlikely because current theories of the role of the cortex and the thalamus in the generation of EEG waveforms rely upon the relationship between cortical neurons and their thalamic relay nuclei, which were clearly eliminated for the fronto-parietal areas from which we recorded EEG. On the other hand, the dialog between the cortex and the thalamus clearly plays a major role in such features as EEG spindles.

Our results provide strong evidence for the critical role of the BF in sleep-wake behavior and cortical EEG activation. There is an extensive literature demonstrating that neurons in the BF fire in bursts that are time-locked to EEG waves (Lee et al., 2005) and that stimulation of the BF can activate the EEG (Berridge and Foote, 1996; Cape and Jones, 2000), whereas inhibition of BF firing can slow the EEG (Cape and Jones, 2000; Xu et al, 2004). Hence, the BF is a candidate for entraining cortical EEG rhythms in the absence of thalamic activity. Earlier studies had found modest chronic effects on EEG of either limited non-selective BF lesions (Buzsaki et al., 1988; Kaur et al. 2008) or more extensive selective cholinergic BF lesions by using IgG-192-saporin (Berntson et al., 2002; Kapas et al., 1996; Wenk et al., 1994; Blanco-Centurion et al., 2006). In the present study, widespread nonselective cell-specific BF lesions resulted in the loss of EEG components >1Hz EEG, resulting in a rhythm that is similar to isolated cerebral cortex, and the loss of behavioral responsiveness, despite the high state of activity in brainstem, hypothalamic, and thalamic arousal systems as evidenced by Fos activation of key subcortical structures. These findings are in agreement with but extend those of Buzsaki et al (1988) who found unilateral low frequency EEG abnormalities following more limited unilateral BF lesions, and Kaur et al. (2008) who found persistent increases in delta power after relatively small bilateral ibotenic acid lesions killing about 40% of parvalbumin and 20% of cholinergic neurons in the caudal basal forebrain.

The PB-PC as a key component of the ascending arousal system

Many modern reviews (including those we have contributed) emphasize the roles played by the cholinergic and monoaminergic ascending pathways in the arousal system (Datta and Maclean, 2007; Fuller et al, 2006; Saper et al., 2005; Jones, 2003; Siegel, 2004). However, studies in which these different cell groups were selectively damaged with cell-specific toxins have found only minimal changes in wake-sleep amount or EEG activation (see Lu et al., 2006a,b; Blanco-Centurion et al., 2007 for review). The results from the present study identify a single specific and delimited node in the ascending arousal system, the PB-PC, whose participation is absolutely required for maintenance of normal wakefulness. In essence, we propose that the acute loss of wakefulness caused by electrolytic or mechanical lesions in the midbrain reticular formation may be due mainly to disruption of the ascending projections to the BF from the PB-PC. While many other cell groups, including the monoaminergic and cholinergic components of the arousal system, may play specific roles in modulating arousal, particularly under specific physiological circumstances, no other brainstem cell population has yet been found to be absolutely necessary to maintain a waking state.

The PB-PC complex is a major, but previously under-appreciated brainstem source of projections to the BF (Fulwiler and Saper, 1984; Moga et al., 1990). In fact, in rats the PB is the largest source of brainstem afferents to the BF and the PB projection to the BF is one of the largest forebrain projections of the PB (Saper and Loewy, 1980; Grove, 1988). The PB-PC also has substantial projections to the intralaminar thalamus, the lateral hypothalamus, and the amygdala (Krout and Loewy, 2000; Munk et al., 1996; Fulwiler and Saper, 1984). In addition, the medial PB is reciprocally connected with the dorsolateral and medial prefrontal and insular cortices (Zaborszky et al., 2005; Saper, 1982, 1984). Thus, the PB/PC has projections to a wide range of forebrain targets that may contribute to wakefulness.

In cats, the PB overlaps spatially with the PPT, and the entire complex is often called the “peribrachial region.” Several earlier studies examining single unit recordings in freely moving cats showed that many neurons in the “peribrachial” region are wake-active whereas others are REM-active (Gilbert and Lydic, 1990; Sieck and Harper, 1980), suggesting separate populations of neurons in this region controlling brain activation of wakefulness and REM sleep. In rats, the PPT and LDT are spatially differentiated from the PB and PC (respectively) so that the contributions of the two populations can be delineated. Recordings from the PPT and LDT in rats (Datta and Siwek, 2002) support the activity patterns that have been identified in cats (Steriade et al., 1990), but comparable recordings from the PB-PC across wake-sleep states are lacking. Given that PPT and LDT lesions cause minimal effects on wake-sleep activity or EEG in rats (Lu et al., 2006b) except when they extend into the PB-PC, such studies are long overdue.

Because our PB/PC lesions that caused coma included small but variable amounts of the adjacent LDT, PPT or LC, the question can be raised whether these could have contributed to the loss of wakefulness. However, much more complete lesions of the LC in rats (Lu et al., 2006; Blanco-Centurion et al., 2008) or the PPT and LDT in rats (Lu et al., 2006) or cats (Webster and Jones, 1988; Shouse and Siegel, 1992) have had no major effect on the amount of wakefulness. In the Lu et al. (2006b) study, the small loss of wakefulness with large PPT lesions was seen only in cases that involved the medial PB. We also recognize that other non-cholinergic (i.e., glutamatergic) neurons that are mixed among the cholinergic neurons in the PPT, were damaged by the lesions in the PB/PC that caused coma. However, partial involvement of glutamatergic neurons in the PPT in our experiments is unlikely to account for the loss of wakefulness given that even large lesions of this same cell group, with little damage to the PB, have little or no effect on wakefulness.

In monkeys and humans with high-speed impact head injury, long-lasting coma and persistent vegetative state correlate with hemorrhage in the PB region due to the impact of the free edge of the tentorium on the surface of the pons (Kampf et al., 1998; Adams et al., 1982; Shigermori et al., 1992; Blumbergs et al., 1989; Gennarelli et al., 1982). By contrast, cell-specific lesions confined to the pontine reticular formation in animals never induce coma (Denoyer et al., 1991; Lu et al., 2006b). On the other hand, damage to axons traversing the brainstem along the route of the ascending fibers from the PB-PC (Saper and Loewy, 1980) causes coma in animals and humans (Parvizi and Damasio, 2003; Lindsely et al., 1949), and electrical stimulation along this route desynchronizes cortical EEG (Moruzzi and Magoun, 1949).

Another behavioral state that may be explained by the PB-PC activity profile is peduncular hallucinosis. In this puzzling condition, patients with lesions of the mesopontine tegmentum have dream-like states with intense and very active visual experiences, against a waking background (Manford and Andermann, 1998). Our previous study found that there is intense GABAergic inhibitory input to the PB-PC from the REM-off area in the lateral pontine tegmentum and ventrolateral periaqueductal gray matter (Lu et al., 2006b). We hypothesize that lesions of the REM-off neurons that inhibit the PB-PC may disinhibit patterns of neuronal firing that normally occur only during REM sleep, thus producing dream-like states at inappropriate times. A similar mechanism may explain hypnagogic (while falling asleep) and hypnopompic (while awakening) hallucinations, which occur most commonly in patients with narcolepsy, but are sometimes seen in healthy young adults who have been sleep deprived.

Is the PB-PC a component of the wake-sleep switching system?

We have previously proposed that a mutually inhibitory interaction between sleep-active neurons of the ventrolateral preoptic nucleus (VLPO) and monoaminergic arousal systems regulates sleep-wake behavior (see Saper et al., 2005). The central lateral PB subnucleus is also a major source of inputs to the VLPO (Chou et al., 2002), and the VLPO sends a return projection that innervates the entire PB-PC complex to about the same extent as the adjacent LC (Sherin et al., 1998). Hence, the PB-PC may be an important part of the flip-flop switch that has been proposed to regulate wake-sleep transitions (Saper et al., 2005). On the other hand, PB-PC neurons that project to the BF should be active during both wakefulness and REM sleep. In the presence of concomitant activation of the monoaminergic arousal systems (Saper et al., 2005), the cortical response to PB/PC-BF activation would be wakefulness; in the absence of monoaminergic stimulation, a state of REM sleep would prevail. In this model, monoaminergic and cholinergic systems may modulate the activity of the PB-PC, and can determine (via their cortical and thalamic projections) whether the PB-PC induces wakefulness vs. REM sleep, but they cannot sustain a waking (or REM) state without the PB-PC.