REM SLEEP-ACTIVE PEDUNCULOPONTINE TEGMENTAL NEURONS SUPRESSES REM SLEEP
EXPRESSION AND RESPIRATORY NETWORK ACTIVITY
by
Kevin Patrick Grace
A thesis submitted in conformity with the requirements for the degree of Master’s of Science
Department of Physiology University of Toronto
© Copyright by Kevin Patrick Grace 2010
ii
REM Sleep-Active Pedunculopontine Tegmental Neurons
Suppresses REM Sleep Expression and Respiratory Network
Activity
Kevin Grace
Master’s of Science
Department of Physiology
University of Toronto
2010
Abstract
The mechanisms underlying the generation of rapid eye movement (REM) sleep are poorly
understood. Despite a lack of direct support, neurons maximally active during REM sleep (REM
sleep-active) located in the pedunculopontine tegmental nucleus (PPTn) are hypothesized to
generate this state and its component phenomenology. This hypothesis has never been directly
tested, since the results of selectively inhibiting this cell-group have never been determined.
Using microdialysis, electrophysiology, histochemical and pharmacological methods in freely-
behaving rats (n=22) instrumented for sleep-wake state and respiratory muscle recordings, I
selectively inhibited REM sleep-active PPTn neurons. Contrary to the prevailing hypothesis, I
showed that REM sleep-active PPTn neurons suppress REM sleep by limiting the frequency of
its onset. These neurons also shape the impact of REM sleep on breathing. REM sleep-active
PPTn neurons restrain behavioural activation of upper-airway musculature during REM sleep,
while depressing breathing rate and respiratory activation of the upper-airway musculature across
sleep-wake-states.
iii
Acknowledgments
This work was supported by funds from the Canadian Institutes of Health Research (CIHR,
Grant MT-15563) and from a graduate studentship awarded to me on behalf of the Ontario
Thoracic Society. Special thanks go to Richard Horner, PhD, for his indispensible mentorship,
Mrs Hattie Liu for her technical support, Gaspard Montandon, PhD, for his thoughtful advice,
and Ralph Lydic, PhD, University of Michigan, Ann Arbor, for assistance on histology.
iv
Table of Contents
Chapter 1: Introduction
1.1 Preface p.1
1.2 Criteria for determining the functional role of a neuronal group
In the control of REM sleep p.2
1.3 Conceptual developments in sleep neurobiology p.6
1.4 REM sleep p.12
1.5 The pedunculopontine tegemental nucleus p.18
1.6 The impact of sleep on breathing p.22
1.7 Summary and hypotheses p.23
Chapter 2: Methods
2.1 Animal care p.27
2.2Anesthesia and surgical procedures p.27
2.3 Habituation p.31
2.4 Microdialysis p.32
2.5 Protocol p.33
2.6 Recording procedures p.38
2.7 Data analysis p.38
Chapter 3: Results
3.1 Effects of bilateral delivery of 8-OH-DPAT to the PPTn on sleep
architecture in freely behaving rats p.45
3.2 Effects of bilateral delivery of 8-OH-DPAT to the PPTn on
sleep micro-architecture in freely behaving rats p.52
3.3 Effects of bilateral perfusion of 8-OH-DPAT to the PPTn
on the drive threshold for REM sleep induction p.55
v
3.4 Effects of bilateral delivery of 8-OH-DPAT to the PPTn on
phenomenology within sleep-wake states. p.60
Chapter 4: Discussion
4.1 REM sleep-active PPTn neurons suppress REM sleep p.73
4.2 mechanism of REM sleep suppression by
PPTn REM sleep-active neurons p.74
4.3 connections of PPTn REM sleep-active neurons possibly
mediating REM sleep suppression p.76
4.4 PPTn REM sleep-active neurons suppress upper-airway
muscle activity during REM sleep p.84
4.5 PPTn REM sleep-active neurons depress breathing
across sleep-wake states p.85
4.6 Future directions p.86
vi
List of Tables
Table 1 p.51
vii
List of Figures
Figure 1 p.8
Figure 2 p.15
Figure 3 p.29
Figure 4 p.36
Figure 5 p.46
Figure 6 p.48
Figure 7 p.53
Figure 8 p.56
Figure 9 p.58
Figure 10 p.61
Figure 11 p.63
Figure 12 p.67
Figure 13 p.69
Figure 14 p.82
viii
List of Abbreviations
ACSF artificial cerebral spinal fluid
ANOVA analysis of variance
DRN dorsal raphé nuclei
EEG electroencephalogram
EMG electromyogram
VLPO ventrolateral preoptic area
GABA gamma-aminobutyric acid
LC locus coeruleus
LDTn laterodorsal tegmental nucleus
LPT lateral pontine tegmentum
NIV non-REM/REM sleep transition indicator value
NADPH nicotinamide-adenine-dinucleotide phosphate
NRT non-REM/REM sleep transition
OSA obstructive sleep apnea
PnO nucleus pontis oralis
PnC nucleus pontis caudalis
PPTn pedunculopontine tegmental nucleus
PRF pontine reticular formation
REM rapid eye-movement sleep
SLDn sublaterodorsal nucleus
TMN tuberomammillary nucleus
VLPO ventrolateral preoptic region
1
Chapter 1 INTRODUCTION
2
INTRODUCTION
1 Introduction
1.1 Preface
One of the principal directives of sleep neurobiology is to provide a full description of the
mechanisms underlying the generation of global brain states and how these states impact upon
an organism’s physiology. The three naturally occurring and phenomenologically distinct
states of the mammalian central nervous system are wakefulness, non-rapid eye movement
(REM) sleep, and REM sleep. The scientific work presented in this manuscript broadly aimed
to contribute to the field of sleep neurobiology by further defining; (i) the circuitry
responsible for the generation of REM sleep and, (ii) the mechanisms responsible for the
impact of REM sleep on respiratory physiology. I sought to achieve these aims by defining
the functional role of neurons located in the pedunculopontine tegmental nucleus (PPTn), that
are maximally active immediately prior to and during REM sleep (i.e., REM sleep-active), in
the generation of REM sleep and its impact on respiratory activity. For over 25 years the
prevailing consensus in the field of sleep neurobiology has been that REM sleep-active PPTn
neurons participate in the generation of the REM sleep state, however; the body of evidence
supporting this claim is in many ways characterized by a lack of coherency as well as deficits
in scientific reasoning. Therefore a large proportion of the introductory remarks will be
dedicated to critically evaluating the evidence regarding REM sleep-active PPTn neuronal
involvement in REM sleep generation. Firstly, I outline the criteria that need to be satisfied in
order to define a cell-group’s functional role in regulating sleep-wake state. I then describe
the major historical developments in the field of sleep neurobiology which have led to our
modern conception of brain sleep-wakes states; focusing particularly on the emergence of
early models of REM sleep generation that are ultimately responsible for the prevailing notion
that REM sleep-active PPTn neurons participate in REM sleep generation. The existing
evidence regarding REM sleep-active PPTn neuron involvement in REM sleep generation
will then be presented, and critically analyzed according to the criteria for defining a cell-
group’s functional role. Finally, the impact of sleep on breathing will be described, and
3
hypotheses will be presented regarding the PPTn REM sleep-active neuron involvement in
REM sleep generation and its respiratory phenotype.
1.2 Criteria for Determining the Functional Role of a Neuronal Cell-Group in the Control of REM sleep
The network controlling the state of REM sleep is composed of neuronal groups acting to
both promote and suppress REM sleep. The term modulator will be used to refer to neuronal
groups acting to either promote or suppress REM sleep. A range of techniques are used in the
field of sleep neurobiology to determine the functional role of neuronal groups in the control
of REM sleep. These techniques can be categorized according to those which demonstrate: (i)
the functional capacity of a neuronal group to modulate REM sleep, (ii) the sufficiency of a
neuronal group to modulate REM sleep or, (iii) the necessity of a neuronal group in the
modulation of REM sleep.
1.2.1 Capacity
Having functional capacity to modulate REM sleep is the weakest designation that can be
assigned to a neuronal group. The capacity of group to modulate REM sleep is primarily
demonstrated via neuroanatomical studies. Clearly, neuronal groups that do not innervate
components of the REM sleep control network would, in most cases, be incapable of
influencing the generation of this state. Therefore if a neuronal group innervates other
modulators known to be necessary for the normal occurrence of REM sleep, it can be said that
this group has the capacity to modulate REM via those projections. Identification of the
neurotransmitter phenotype of projecting neurons can be used to indicate whether a neural
group would act in a promoting or suppressive manner. For example, GABAergic neurons
projecting to a region known to be required for the generation of REM sleep would certainly
have the capacity to suppress REM sleep generation. Neuroanatomical connections only
4
establish capacity, because using conventional tracing techniques, it is often only possible to
define a region of tissue innervated by a group of neurons. Such a region would undoubtedly
be functionally heterogeneous, and so it is often not known whether projections to a region
actually innervate functionally relevant neurons. Moreover the physiological relevance of a
projection cannot be inferred from neuroanatomy alone, and therefore further studies are
required to define the functional role of the projecting neuronal group.
Any neuronal group which is not active during REM sleep, clearly cannot actively modulate
this state (as opposed to modulators which passively permit REM sleep generation via their
inactivity during REM sleep – only active modulators will be discussed here). Many
hypothesized modulators of REM sleep are those neuronal groups which are designated REM
sleep-active (i.e., maximally active immediately prior to and during REM sleep). However,
having such an activity profile does not necessitate that a neuronal group is involved in the
modulation of REM sleep. Consider the fact REM sleep is an aroused state of the brain where
brainstem reticular formation neurons often experience increases in membrane potential
exceeding even waking levels (Steriade and McCarley, 2005b); it would therefore be true that
many neuronal pools throughout the central nervous system, including those uninvolved in
REM sleep generation, would exhibit maximal activity during REM sleep.
1.2.2 Sufficiency
Gain-of-function interventions (e.g., electrical or local pharmacological stimulation) are often
used to identify REM sleep modulators by demonstrating that increases in their activity are
associated with some change in REM sleep. For example, an increase in REM sleep
stemming from the stimulation of a particular neuronal group would be consistent with that
group being involved in the generation of REM sleep. The activation of that group would be
considered sufficient for the promotion of REM sleep. The weakness of gain-of-function
interventions demonstrating sufficiency is that, it is never known if the activation of a
5
neuronal group above normal levels produces an abnormal physiological response that is not
representative of the true function of that group.
1.2.3 Necessity
The most effective means by which the function of a neural group can be determined is by a
loss-of-function intervention (e.g., lesioning or local pharmacological inhibition). By
diminishing or silencing the activity of a neuronal group, and observing the ensuing
functional deficit(s) in REM sleep generation, it can be reasonably concluded that the
endogenous activity of the group in question, subserves the function(s) lost by its inactivation.
For example, an increase in REM sleep stemming from the inhibition of a particular neuronal
group would show that that group is necessarily involved in the suppression of REM sleep.
To demonstrate that a neuronal group is necessarily involved in the modulation of REM sleep,
is to preclude the possibility that it is uninvolved, something which is not accomplished by
demonstrations of sufficiency or capacity. Therefore, determining the functional role of a
neuronal group in the control of REM sleep ultimately requires a demonstration of
necessity or a loss of function intervention.
1.2.4 On the Importance of Selectivity
There is obviously limited usefulness in defining the function of a spatial region of the brain,
which itself contains a heterogeneous mixture of functionally disparate neuronal groups.
However, many of the methodological approaches used in sleep neurobiology are designed to
do exactly that. Such techniques include electrolytic/chemical lesioning, electrical
stimulation, and the local application of non-discriminating drugs like GABA and glutamate
receptor agonists. Although these techniques have proven very useful for focusing in on the
regions of the brain underlying the control of REM sleep; in order to map the network
responsible for REM sleep generation we need to selectively study the actions of functionally
6
homogenous cell groups, as they are the fundamental functional units of information
processing in the brain (Bullmore and Sporns, 2009). For example, use of the above
mentioned techniques to manipulate the PPTn, has shown that neurons in this region are
involved in the control of sleep-wake state. However, there is little meaning in any function(s)
assigned to the PPTn en masse, given that there are different PPTn sub-populations, defined
by their activity profiles across sleep-wake states, that are hypothesized to exert different
influences on sleep-wake state (see sections 1.4.1, 1.3.2, & 1.4.2 for more detail). It is these
subgroups that independently operate as functional units in the circuitry governing sleep wake
state. In other words, close physical proximity is a poor basis for defining a neuronal group
involved in the control of REM sleep; other characteristics like activity profile, connectivity
and neurotransmitter phenotype need to be taken into consideration. Therefore when
determining the function of neuronal groups in the control of REM sleep, interventions
should be used which target not only physical location, but also other functional relevant
parameters.
1.3 Conceptual Developments in Sleep Neurobiology
1.3.1 Seminal Developments
Much of our understanding regarding the mechanisms of sleep wake states can be traced back
to early studies surrounding the pandemic of lethargic encephalitis during the First World
War. Individuals inflicted with lethargic encephalitis were primarily characterized by an
inability to maintain wakefulness. Pioneering studies by the Viennese neurologist, Baron
Constantin von Economo, determined that the loss of wakefulness was associated with
neuronal cell death in regions of the brainstem (Saper et al., 2001). This observation became
the basis for the important conceptual development that the brainstem is a source of ascending
inputs to the forebrain that are required for the maintenance of the waking state. In
7
conjunction with studies of lethargic encephalitis, Frédéric Bremer, preformed a series of
transection experiments in cats confirming the necessity of the brainstem in the generation of
wakefulness (Tretiakoff and Bremer, 1920; Saper et al., 2001). Removal of the entire
encephalon from the spinal cord did not disrupt the occurrence of electroencephalographic
rhythms consistent with wakefulness and sleep, while mesencephalic transections, separating
the forebrain from the midbrain, resulted in forebrain electrographic changes that were
consistent with prolonged sleepiness (i.e. large amplitude and low frequency waveforms).
Therefore Bremer postulated that ascending inputs from the spinal neuraxis and brainstem
were the source of continuous facilitation of forebrain activity (Steriade and McCarley,
2005b). A student of Bremer, G. Moruzzi, in collaboration with H.W. Magoun, demonstrated
that brainstem activation was also sufficient for the maintenance of forebrain activation, by
demonstrating that electrical stimulation of the brainstem transformed the low frequency /
high amplitude electroencephalographic waves typical of their anesthetized animals to a high
frequency / low amplitude waveform indicative of forebrain activation (Moruzzi and Magoun,
1949; Steriade and McCarley, 2005b). By modern standards, these pioneering studies utilized
crude techniques in crude preparations, and the conclusions drawn from these studies were by
no means definitive. In fact, it may be a gross exaggeration to claim, that the work by Bremer,
Moruzzi, and Mangoun demonstrated the sufficiency and necessity of the brainstem reticular
formation in the activation of the forebrain and the maintenance of wakefulness. But
nevertheless, these studies have formed the basis for the modern notion that the so called
brainstem “ascending reticular activating system” is primarily responsible for producing
forebrain arousal.
1.3.2 Ascending Reticular Activating System
More selective transection studies in the years following the work of Bremer, Moruzzi, and
Mangoun localized the roots of the ascending reticular activating system to the rostral pons
(Saper et al., 2001). A schematic of the reticular activating system is shown in figure. 1A.The
use of modern neuroanatomical tracing methods identified pathways from this region into the
8
9
Figure 1. A schematic of the key components of the brainstem circuitry controlling
wakefulness and sleep. (A) Shows hypothesized components of the ascending reticular
activating system producing forebrain arousal and wakefulness. Cholinergic neurons in the
pedunculopontine and laterodorsal tegmental nuclei (PPTn/LDTn) promote arousal and
corticothalamic transmission via a thalamic pathway (orange). Neurons of the noradrenergic
locus coeruleus (LC), serotonergic dorsal/median raphé nuclei, histamtergic
tuberomammillary nucleus (TMN), cholinergic basal forebrain, as well as orexinergic
hypothalamic neurons form an additional arousal pathway coursing through the hypothalamus
(red). (B) Shows the hypothesized action of sleep-active inhibitory (via release of the
inhibitory neurotransmitters GABA and galanin) ventrolateral preoptic nucleus neurons, on
the components of the ascending reticular activating system. Such inhibition actively prevents
forebrain arousal and generates sleep. GABA, γ-aminobutyric acid. Modified version of a
figure by Saper et. al. (2005).
10
diencephalon. This pathway branches into two parts, one innervating the thalamus and the
other the hypothalamus. One of the main pontine inputs to the thalamus is the laterodorsal
tegemental nucleus (LDT) and to a lesser degree the PPTn (Edley and Graybiel, 1983; Rye et
al., 1987; Hallanger and Wainer, 1988). Both these regions contain neurons that are
maximally active during wakefulness and REM sleep, and minimally active during non-REM
sleep (wake/REM sleep-active)(Thakkar et al., 1998). Given that a subset of PPTn/LDTn
neurons are cholinergic (Rye et al., 1987; Shiromani et al., 1988) and that cholinergic activity
is considered to be crucially involved in activating thalamocortical transmission, it is thought
that cholinergic innervation of the thalamus by PPTn/LDTn wake/REM sleep-active neurons
plays a major role in maintaining forebrain arousal (Hallanger and Wainer, 1988; Saper et al.,
2001; Pace-Schott and Hobson, 2002). The hypothalamic arm of the reticular activating
system has also been well characterized. Many cell groups, which are maximally active
during wakefulness and minimally active during sleep, diffusely innervate and activate the
cortex via a pathway coursing through the region of the lateral hypothalamus (Saper et al.,
2001). Examples include noradrenergic and serotonergic neurons located in the locus
coeruleus (LC) and dorsal/median raphé nuclei respectively. In the hypothalamic area,
projections from these cell-groups are joined by histaminergic projections from the
tuberomammillary nucleus (TMN), cholinergic projections from the basal forebrain, and
hypothalamic orexinergic projections. The activation of the forebrain, via the combined
influences of the two arms of the reticular activating system are responsible for maintaining
wakefulness. The inhibition of this system is ultimately responsible for the generation of
sleep.
1.3.3 The Emergence of Active Theories of Sleep
The end of the last section alluded to the notion that sleep occurs as a result of the active
inhibition of wake-generating circuitry. Indeed, the idea that sleep is an actively generated
process is the modern view, however; this has not always been the case. The prevailing
hypothesis regarding the mechanism of sleep generation, from the last century B.C. until the
11
mid 1900’s, was that sleep is a passively generated (Steriade et al., 2005b). Moruzzi and
Mangoun, suggested that sleep might occur due to the passive withdrawal of activity in
brainstem activating circuitry and/or withdrawal of reticular formation stimulation by afferent
sensory systems (Steriade et al., 2005a). This was arguably the most parsimonious
explanation of sleep generation at the time since little evidence existed testifying to the
existence of hypnogenic neural structures capable of generating sleep. So, the notion that
sleep is actively generated emerged from electrical stimulation studies in cats demonstrating
the sufficiency of regions like the medial thalamic area and the lateral preoptic area of the
hypothalamus, to induce sleep or sleep-like states, characterized by electrographic
synchronization (i.e., low frequency / high amplitude waveform) (Sterman and Clemente,
1962a, b). Moreover, exitotoxic lesions of the preoptic area were capable of producing long-
lasting bouts of insomnia demonstrating the necessity of this region for normal sleep
induction (Nauta, 1946; Sallanon et al., 1989).
Since these initial land mark studies, the use of more discriminating lesioning techniques have
confirmed the role of the ventrolateral preoptic nucleus (VLPO) in the generation of sleep and
so solidified the notion that sleep is actively generated via the inhibition of reticular activating
system (Lu et al., 2000). Anterograde tracing methods have shown that VLPO neurons
innervate the main components of the reticular activating system (i.e., dorsal/median raphe,
basal forebrain, LC, TMN, LDT, PPT) (Sherin et al., 1996; Sherin et al., 1998; Steininger et
al., 2001). These projections are argued to be likely inhibitory given that 80% of retrogradely
labeled neurons in the region release the inhibitory neurotransmitters GABA and galanin
(Sherin et al., 1998).
1.3.4 The Dual Nature of Sleep
In the 1950’s, Kleitman, Aserinsky and Dement were the first to describe a unique sleep
stage, which would come to be known as REM sleep (Aserinsky and Kleitman, 1953, 1955;
Dement and Kleitman, 1957b, a). In stark contrast to EEG synchronized sleep (i.e. non-REM
12
sleep), REM sleep was defined by the presence of a wake-like cortical EEG activity (i.e., low
voltage/high frequency), limb twitching, rapid eye movements, and complete loss of muscle
tone. Despite demonstration of the obvious differences compared to EEG synchronized sleep;
REM sleep was not regarded as qualitatively different from other stages of sleep. Instead
sleep was considered a unitary phenomenon having stages of varying depth or intensity, and
REM sleep was regarded as its deepest stage. The roots of our modern conception of REM
sleep as a phenomenologically distinct from non-REM sleep, can be traced back to Dement
and Jouvet, who contrary to popular thinking at the time, publicly emphasized the numerous
differences between REM sleep and EEG synchronized sleep (Steriade and McCarley,
2005b). Their advocacy prompted a shift in conceptual thinking away from the notion that
sleep is a unitary phenomenon, and towards the idea that sleep is an aggregate of two states
which have very little in common at the level of the brain. This conceptual shift was
extremely important because it prompted the field of sleep neurobiology to consider questions
of sleep function and generative mechanisms, in terms of non-REM and REM sleep
individually. In the following decades, it was indeed shown that the circuitry underlying the
generation and maintenance of REM sleep is very different from that generating wakefulness
or non-REM sleep. The following section will detail the efforts to describe the circuitry
responsible for the generation of REM sleep.
1.4 REM sleep
1.4.1 Determining the Basic Circuitry
The circuitry responsible for the generation of REM sleep, was initially defined using
transection and lesioning studies. Attributes of REM sleep were shown to persist following
decortification and transections of the brainstem rostral to the pons, while transections along
the posterior border of the pons, eliminated signs of REM caudal to the transection (Jouvet,
1962; Carli and Zanchetti, 1965). Therefore, the pons was shown to be necessary and
sufficient for the generation of REM sleep, although that is not to say that extrapontine
13
structures do not contribute to the elaboration of REM sleep phenomenology (Webster et al.,
1986). Lesioning studies were used to further pin-point the pontine regions critically
necessary for the generation and maintenance of this state. Electrolytic and chemical lesions
in the dorsal segment of oral and caudal portions of the pontine reticular formation (PRF)
were the most successful in producing suppression of REM sleep, thereby implicating the
neurons in this region in the generation of REM sleep (Jouvet and Valatx, 1962; Carli and
Zanchetti, 1965; Jouvet et al., 1965; Sastre et al., 1981; Webster and Jones, 1988).
Cholinergic influences in the PRF figured centrally in the early models of REM sleep
generation. The fixation on cholinergic influences can be traced back to studies performed by
Jouvet’s group demonstrating that systemic administration of atropine, an antagonist of
muscarinic acetylcholine receptors, suppressed REM sleep (Jouvet, 1962). Moreover,
increasing endogenous acetylcholine levels via systemic administration of an
acetylcholinesterase inhibitor had a promoting effect on REM sleep (Jouvet, 1962). Of course,
few if any meaningful conclusions can be drawn, regarding complex neural mechanisms,
from systemic drug administration, but these findings formed the rationale for subsequent
studies showing that local application of the mixed cholinergic receptor agonist, carbachol, in
the PRF of cats, can induce REM sleep (George et al., 1964). The most effective region for
cholinergic induction of REM sleep in cats was later shown to be a small region within the
dorsal PRF containing REM sleep-active neurons, named the peri-locus coeruleus
(Baghdoyan et al., 1984; Baghdoyan et al., 1987; Vanni-Mercier et al., 1989; Yamamoto et
al., 1990)[corresponds to the sublaterodorsal nucleus (SLDn) in the rat]. These studies
demonstrate: (i) the sufficiency of acetycholine to promote REM sleep generation, and (ii) the
capacity of cholinergic neurons innervating the SLDn to promote REM sleep generation.
There is ample evidence to implicate neurons located in the SLDn, as perhaps the most
critical component of the REM sleep control network. Apart from lesioning experiments that
have established the necessity of neurons in this region for normal REM sleep expression,
additional evidence indicates that SLDn neurons are responsible for producing the
phenomenological components of REM sleep. The SLDn, particularly its ventral portion, is
14
necessarily involved in generating the motor atonia of REM sleep since lesions of this region
can reliably block REM sleep atonia (Sastre and Jouvet, 1979; Sanford et al., 1994; Plazzi et
al., 1996; Lu et al., 2006). The induction of REM sleep atonia is likely mediated by
projections of REM sleep-active SLDn neurons directly to regions of the spinal cord
containing motor neurons (Lu et al., 2006) and/or indirectly via the gigantocellular nucleus of
the ventromedial medullary reticular formation: an important relay in the pathway responsible
for REM sleep atonia (Sastre et al., 1981; Rye et al., 1988). Ascending projections from the
SLDn are implicated in generating the electrographic phenotype of REM sleep (Lu et al.,
2006). Electrographically, REM sleep in rodents is primarily characterized by a prominent
hippocampal theta rhythm generated by neurons projecting to the hippocampus from the
septum. SLDn neurons project to areas of the septum; the lesioning of which can obliterate
theta activity during REM sleep (Lu et al., 2006).
1.4.2 The Reciprocal Interaction Model
Building on early studies by Jouvet and others, Hobson and McCarley (Hobson et al., 1975;
McCarley and Hobson, 1975a) proposed the reciprocal interaction model; a
structural/mathematical hypothesis meant to provide an explanation for the cyclical
generation of REM sleep. Over the last 35 years this model has stood as the most widely
accepted explanation for the mechanism of REM sleep generation (Pace-Schott and Hobson,
2002). The original model proposed that REM sleep-inactive aminergic neurons in the LC
and DRN interact reciprocally with REM sleep-active cholinergic PRF neurons, and that this
interaction drives the rhythmic cycling between non-REM and REM sleep (Hobson et al.,
1975). During wakefulness aminergic neurons would be maximally active and producing
suppression of REM sleep-active PRF neurons. During non-REM sleep the activity of
aminergic REM sleep-inactive neurons would diminish, thereby relieving inhibition of
cholinergic REM sleep-active PRF neurons, thus allowing their activity to steadily rise until
REM sleep is induced. During REM sleep, aminergic neurons are almost completely silent,
but would become active again and terminate the REM sleep episode due to an excitatory
15
16
Figure 2. A schematic diagram of the main circuitry responsible for REM sleep
generation according to the reciprocal interaction model. Shows that REM sleep in-active
monaminergic neurons located in the dorsal/median raphé nuclei (raphe) and locus coeruleus
(LC) interact in a reciprocal manner with presumed cholinergic neurons of the
pedunculopontine and laterodorsal tegmental nuclei (PPTn/LDTn). As LC/raphe neurons
relieve the PPTn/LDTn from inhibition, REM sleep is induced via cholinergic excitation of
sublaterodorsal nucleus (SLDn) neurons: a process which is sustained by a mutual excitatory
relationship between the neurons of the SLDn and PPTn/LDTn. This circuitry is hypothesized
to drive the cyclical generation of REM sleep. This model is responsible for the prevailing
hypothesis that PPTn REM sleep-active neurons generate REM sleep. 5HT, serotonin; NA,
noradrenaline; ACh, acetylcholine. Modified version of figure by Pace-Schott and Hobson
(2002).
17
influence of cholinergic PRF neurons. This arrangement was modeled mathematically using
equations of the Lotka-Volterra type derived from population models of predator-prey
interactions. The time course of neuronal activity in the REM sleep-active and REM sleep in-
active cell groups predicted by the mathematical model, coincide with the actual long-term
recordings of PRF and LC/DRN neurons (McCarley and Hobson, 1975a).
The reciprocal interaction model was postulated prior to the development of immunolabelling
for cholinergic neurons, but once these techniques were made available, it was determined
that the presumed cholinergic REM sleep-active PRF neurons were in fact not cholinergic
(Steriade et al., 2005b). In order to salvage the model, it was instead suggested that the major
source of REM sleep-active cholinergic neurons were the PPTn and LDTn (Sakai et al.,
1981). Moreover, it was suggested that PPTn and LDTn cholinergic neurons act to induce
REM sleep onset via innervation and activation of cholinoceptive REM sleep-active SLDn
neurons (Sakai et al., 1981). Several lines of evidence support the reciprocal interaction
model. The supporting evidence includes but is not limited to the following: (i) exogenous
cholinergic stimulation of the SLDn and other region in the PRF induces REM sleep (see
sec.1.4.1), (ii) microdialysis studies show that endogenous levels of acetylcholine in the PRF
increase during REM sleep (Kodama et al., 1990; Lydic et al., 1991), (iii) The SLDn is
innervated by cholinergic LDTn and PPTn neurons (Datta et al., 1999), (iv) the LDTn and
PPTn contain neuronal sub-groups exhibiting maximal activity during REM sleep (Thakkar et
al., 1998), (v) Electrical stimulation of the LDTn evokes excitatory post synaptic potentials in
PRF neurons that can be blocked by scopolamine , (vi) electrical stimulation of the LDTn
promotes REM sleep (Thakkar et al., 1996), (vii) serotonergic neurons of the DRN have been
shown to project to the LDTn and PPTn (Honda and Semba, 1995), and (viii) serotonin type
1A receptor agonism selectively inhibits activity of REM sleep-active PPTn and LDTn
neurons (Thakkar et al., 1998).
Based on the reciprocal interaction model, the PPTn REM sleep-active cell group is
commonly regarded as an important component of the REM sleep control network,
18
participating in the generation of this state. In the following sections I detail the efforts to
validate the hypothesized involvement of PPTn neurons in REM sleep generation and
critically analyze the evidence supporting this hypothesis according to the criteria defined in,
section 1.2, of this manuscript.
1.5 The Pedunculopontine Tegmental Nucleus
1.5.1 PPTn Neuroanatomy & Heterogeneity
The PPTn was first defined on cytoarchitectonic grounds in the human brain. It consists of a
collection of large neurons that extend from the caudal pole of the red nucleus to the
parabrachial nucleus in close association with the ascending limb of the superior cerebellar
peduncle (Rye et al., 1987). The PPTn is typically considered as having two major
subdivisions differing on the basis of cell density. The caudal half of the nucleus, known as
the subnucleus compactus, has the greatest cell density. Cell density is sparser in the rostral
half of the nucleus, known as the subnucleus dissipatus. Approximately 80% of the large
neurons characterizing the PPTn are cholinergic (Rye et al., 1988). Despite defining the
PPTn, large cholinergic neurons are outnumbered by intermixed glutamatergic and
GABAergic neurons. Glutamatergic and GABAeric PPTn cell groups account for 40 and 35%
respectively of the total number of PPTn neurons (Wang and Morales, 2009). I have
previously mentioned the two functionally distinct PPTn cell groups involved in the control of
sleep-wake states, but I will recap them here. Firstly, wake/REM sleep-active PPTn neurons
(i.e., neurons maximally active during the aroused brain states of wakefulness and REM
sleep) are a hypothesized component of the reticular activating system responsible for
maintaining wakefulness (Saper et al., 2001). Secondly, REM sleep-active PPTn neurons (i.e.,
neurons maximally active immediately prior to and during REM sleep, and minimally active
during wakefulness and non-REM sleep) are hypothesized to participate in the generation of
REM sleep (Pace-Schott and Hobson, 2002).
19
1.5.2 Critique of Evidence Supporting PPTn-Mediated REM Sleep Generation
REM sleep-active PPTn neurons were first incorporated into the reciprocal interaction model
of REM sleep generation on the basis that they are a potential source of REM sleep-specific
release of acetylcholine in the PRF. Therefore, I begin by examining the evidence supporting
cholinergic PRF mechanisms in REM sleep generation. Given the evidence already discussed,
there is little doubt that exogenously applied cholinergic receptor agonists have the capacity
to induce REM sleep in the PRF. Therefore, by proxy, cholinergic neurons projecting to the
PRF, like those located in the PPTn, also have the capacity to induce REM sleep. However,
the necessity of pontine acetylcholine release for the induction of REM sleep has never been
demonstrated. In the nearly 46 years since the seminal demonstration of the capacity of
carbachol to induce REM sleep in the PRF (George et al., 1964), not a single published
account exists demonstrating that antagonism of cholinergic receptors in the PRF produces
REM sleep suppression (Luppi et al., 2006). Even more troubling, is the fact that REM sleep
cannot be reliably induced by PRF application of carbachol, particularly in rodents
(Deurveilher et al., 1997). PRF sites at which carbachol can successfully induce REM sleep,
are intermingled amongst sites at which cholinergic stimulation produces either wakefulness
or no effect. Cholinergic receptor agonism of SLDn neurons has been shown to produce
prolonged insomnia characterized by heightened muscle tone (Bourgin et al., 1995; Boissard
et al., 2002). Similarly, increasing endogenous acetylcholine in the PRF of mice using
acetylcholinesterase inhibition has been shown to produce prolonged bouts of wakefulness
characterized by freezing behavior and elevated muscle tone (Lydic et al., 2002; Pollock and
Mistlberger, 2005). In conclusion, although acetylcholine acting in the PRF is sufficient to
induce REM sleep, it is also sufficient to induce wakefulness. The unreliability of REM sleep
induction by carbachol, and the absence of evidence testifying to the necessity of PRF
acetylcholine for REM sleep generation, casts doubt over the capacity of cholinergic PPTn
REM sleep-active neurons to generate REM sleep.
20
REM sleep active PPTn neurons, can also be considered to have the capacity to generate
REM sleep on the basis that they are maximally active prior to and during REM sleep. The
hypothesis that REM sleep-active PPTn neurons generate REM sleep via the cholinergic
innervation and activation of the SLDn is of course predicated on the assumption that REM
sleep-active PPTn neurons are predominately cholinergic. The activity profiles of neurons
across states of sleep and wakefulness are predominately determined using extracellular
recording techniques (Steriade et al., 2005b). Using such methods, the neurotransmitter
phenotype of recorded neurons cannot be determined. Immunolabeling for c-Fos combined
with REM sleep deprivation/recovery protocols is the only method in common use to define:
(i) whether or not REM sleep-active PPTn neurons are cholinergic, and (ii) the relative
proportions of cholinergic versus non-cholinergic REM sleep-active neurons in the PPTn. c-
Fos protein expression increases along with increasing cellular activity, and so REM sleep
active PPTn neurons would be expected to display heightened c-Fos expression in periods of
REM sleep rebound (i.e., the homeostatic increase in REM sleep following deprivation)
relative to controls (Maloney et al., 2000). Using c-Fos immunostaining protocols, reported
proportions of PPTn neurons that are both REM-sleep active and cholinergic, range from only
3-14% (Maloney et al., 1999; Verret et al., 2005), while the proportion of GABAergic REM
sleep-active PPTn neurons is reported to be within 44-82% (Maloney et al., 2000; Torterolo et
al., 2001; Sapin et al., 2009). It could be argued that a deficiency in cholinergic cell labelling
led to an under representation of the proportion of REM sleep-active PPTn neurons that are
cholinergic. However, this is not likely the case, given that the same studies reported
relatively high numbers of cholinergic REM sleep-active neurons in other brain regions. For
example, Verret and colleagues reported that in the predominately serotonergic raphe
obscurus nucleus; 42.7% of neurons identified as REM sleep active were cholinergic. The
relative scarcity of cholinergic REM sleep-active PPTn neurons is significant because it
means that we cannot reasonably define the role of these neurons in REM sleep generation
according to the actions of acetylcholine in the PRF. In other words, evidence supporting a
role for cholinergic mechanisms in the generation of REM sleep does not implicate the PPTn
REM sleep-active cell-group in the promotion of REM sleep generation.
21
Studies reporting reduced REM sleep following PPTn lesioning are often regarded as
demonstrations of the necessity of PPTn neurons for the generation of REM sleep (Webster
and Jones, 1988; Shouse and Siegel, 1992). However these studies do not report the effects of
lesions of the PPTn per se. The lesions created by these studies were expansive,
encompassing not only the PPTn, but also large portions of the surrounding pontine
tegmentum. The lesions produced by Webster et.al., included the SLDn and so the resulting
reductions in REM sleep could be explained by the loss of SLDn neurons, rather than ablation
of the PPTn. Recognizing this fatal criticism, Shouse et. al., effectively repeated this study
while making sure that lesions spared the region of the SLDn. Despite leaving the SLDn
intact, lesions were by no means spatially selective for the PPTn. Also lesioned were
structures such as the lateral pontine tegmentum and the ventrolateral periaqueductal gray;
two structures which have been recently identified as important REM sleep modulators (Lu et
al., 2006). All that can be reasonably concluded from these studies is that the region of the
pontine tegmentum, containing the PPTn as well as other components of the REM sleep
control network, is necessarily involved in the promotion of REM sleep generation. In other
words, these studies are of little use in defining the functional role of PPTn neurons
specifically in REM sleep generation, due to their lack of spatial specificity.
Nevertheless, more spatially selective ablation of PPTn neurons may still have been expected
to produce reductions in REM sleep, thereby demonstrating the necessity of this region in the
generation of this state. However, Lu et. al. (2006), reported that more spatially specific
lesioning of the of the PPTn resulted in increased REM sleep, indicating that PPTn neurons
act to suppress rather than promote REM sleep generation. The notion that the PPTn exerts an
inhibitory influence on REM sleep generation, is consistent with the results of other studies
producing PPTn inactivation. GABAA receptor agonism at the PPTn has also been shown to
increase REM sleep (Torterolo et al., 2002; Pal and Mallick, 2004, 2009). The increased REM
sleep reported by these studies came at the expense of time spent awake, and so it has been
argued that the observed REM sleep increase, was just a by-product of a wakefulness
suppression, stemming from the inhibition of wake promoting wake/REM sleep active PPTn
neurons (Torterolo et al., 2002). However, this explanation does not account for the
22
disproportionate increases in REM sleep reported by these studies, which would suggest that
some PPTn neurons are necessarily involved in the active suppression of the REM sleep state.
Based on the body of evidence surrounding PPTn neuron involvement in REM sleep
generation, we can conclude that: (i) REM sleep-active PPTn neurons likely do not generate
REM sleep via the cholinergic innervations and activation of PRF neurons, (ii) A population
of PPTn neurons may be actively suppressing REM sleep, (iii) Determination of the necessary
functional role fulfilled by REM sleep-active PPTn neurons in REM sleep generation has
been prevented by the use of techniques which do not discriminate between the multiple
functional cell groups of the PPTn (i.e., REM sleep-active vs. wake/REM sleep active).
Therefore, to determine the functional role of the REM sleep-active PPTn neurons in the
generation of REM sleep, an intervention is needed that selectively inhibits this cell group.
1.6 The Impact of Sleep on Breathing
Sleep results in fundamental changes in respiratory muscle activity and control mechanisms,
changes that can predispose individuals to disordered breathing during sleep: a class of
conditions designated as a major public health burden (Orem et al., 1977; Colten and
Altevogt, 2006). Disturbances of normal breathing are most significant during rapid eye
movement (REM) sleep, when respiratory rate becomes heightened and irregular while the
activity of certain respiratory muscles is completely suppressed (Orem et al., 1977; Horner,
2009). The susceptibility of a given respiratory muscle to suppression by sleep mechanisms
seems to be dependent upon the level of non-respiratory or behavioural input that muscle
receives (Phillipson and Bowes, 1886; Orem, 1990; Horner, 2009). For example, consider the
primary respiratory muscle responsible for drawing air into the lungs: the diaphragm. This
muscle receives very little behavioural input and is spared from the atonia which affects
postural musculature. In contrast, consider the genioglossus muscle of the tongue; it receives
a rhythmic respiratory input that is important for maintaining patency in the collapsible
portion of the upper airway amidst negative pressures generated during inspiration that tend to
23
pull the airway closed. The genioglossus muscle also receives behavioural inputs related to its
accessory functions such as speech and swallowing (Remmers et al., 1978; Horner, 2009).
Activity in this muscle is potently suppressed during REM sleep, an effect which precipitates
obstructive sleep apnea (OSA) (Remmers et al., 1978). OSA is perhaps the most prevalent
example of sleep disordered breathing, and is characterized by repetitive airway collapse and
cessation of breathing during sleep. Determining the neural underpinnings of the impact of
sleep on breathing will permit a greater understanding of the pathogenesis of sleep-related
breathing disorders.
PPTn REM sleep-active neurons have the functional capacity to fulfill a role in the
modulation of breathing during REM sleep. In addition to connections with other components
of the REM sleep control network already described; PPTn neurons innervate critical
components of the respiratory network. PPTn neurons project to areas of the rostral
ventrolateral medulla (Yasui et al., 1990) that contain respiratory neurons critical to the
generation of respiratory rhythm and pattern (Feldman and Del Negro, 2006), and to relevant
motor pools such as the hypoglossal motor nucleus (Woolf and Butcher, 1989; Fay and
Norgren, 1997; Rukhadze and Kubin, 2007), which innervates the genioglossus muscle of the
tongue.
1.7 Summary and Hypotheses
Rapid eye movement (REM) sleep is a naturally recurring and phenomenologically distinct
state of the central nervous system, yet the mechanisms underlying its generation are
unresolved. PPTn REM sleep-active neurons are thought to be significant to the generation of
this state; the prevailing hypothesis being this cell-group promotes REM sleep generation via
cholinergic innervation and activation of SLDn neurons, which themselves gate entry into
REM sleep (Hobson et al., 2000; Pace-Schott and Hobson, 2002; lydic and Baghdoyan, 2003;
Lydic and Baghdoyan, 2005; Steriade and McCarley, 2005b). This hypothesis stems,
primarily, from the body of evidence showing that pontine cholinergic mechanisms have the
24
capacity to influence REM sleep expression. Evidence includes enhancement of REM sleep
following cholinergic stimulation of the PRF and elevated PRF acetylcholine concentrations
during REM sleep and following electrical stimulation of the PPTn (ibid.). Additionally,
PPTn neurons project to the PRF and the activity profile of PPTn REM sleep-active neurons
is consistent with their causal involvement in REM sleep generation (ibid.).
These and other data, however, are not sufficient to establish that REM sleep-active PPTn
neurons as being necessarily involved in REM sleep generation for two reasons. Firstly, these
data do not preclude the alternative hypotheses that REM sleep-active PPTn neurons either
have no effect on, or even suppress, REM sleep. Secondly, determination of the functional
role of REM sleep-active PPTn in REM sleep generation requires identification of the effects
of selective inhibition of this population of PPTn neurons, with the expectation that this
intervention would suppress REM sleep according to the prevailing hypothesis. In each case,
data contrary to the prevailing hypothesis of REM sleep generation would require significant
revision of the current view.
Despite studies showing that pontine cholinergic mechanisms have the capacity to influence
REM sleep expression, there is accumulating evidence that this is not the mechanism by
which REM sleep-active PPTn neurons regulate REM sleep. For example, cholinergic
stimulation of the PRF in rodents does not reliably enhance REM sleep, inducing wakefulness
or having no effect in many cases (Bourgin et al., 1995; Deurveilher et al., 1997; Boissard et
al., 2002; Pollock and Mistlberger, 2005) and endogenous acetylcholine in the PRF has not
yet been shown to be necessary for REM sleep generation as would be tested by focal
application of acetylcholine receptor antagonists. REM sleep-active PPTn neurons are also
predominately non-cholinergic (Maloney et al., 1999; Verret et al., 2005). Reductions in
REM sleep following PPTn lesions are often taken as direct evidence of PPTn involvement in
REM sleep generation (Webster and Jones, 1988; Shouse and Siegel, 1992). However, these
results do not reflect the loss of PPTn neurons per se, given that those lesions were expansive,
encompassing other regions identified as important in REM sleep regulation (e.g., peri-locus
coeruleus alpha). More spatially restricted PPTn lesions result in increased REM sleep rather
25
than the expected decreases (Lu et al., 2006). Likewise, GABAA receptor-mediated inhibition
of PPTn neurons results in increased REM sleep (Torterolo et al., 2002; Pal and Mallick,
2004, 2009). Overall, these data are inconsistent with the hypothesis that PPTn neurons
promote REM sleep generation. However, given that the increases in REM sleep following
PPTn inhibition were accompanied by decreased wakefulness, it has been proposed that
increases in REM sleep occurred as a by-product of simultaneous suppression of the wake
promoting influence of PPTn neurons active during both wakefulness and REM sleep (i.e.,
Wake/REM sleep active) (Torterolo et al., 2002). This argument however, does not account
for the disproportionate changes in REM sleep reported by these studies, which can be taken
to suggest that some population of PPTn neurons actively suppress REM sleep. Ultimately
definitive interpretation of data from past studies is made difficult by the utilization of non-
selective interventions which neglect the potential functional heterogeneity of the PPTn (i.e.,
wakefulness vs. REM sleep promoting influences). In light of the above findings that
contradict the prevailing hypothesis regarding involvement of REM sleep-active PPTn
neurons in the generation of REM sleep, I hypothesize that selective inhibition of REM sleep-
active PPTn neurons (Thakkar et al., 1998) will show that this cell group functions to
suppress REM sleep and its phenomenological components. For the latter I measured indices
of respiratory network activity, given the projections of PPTn to the critical brainstem sites
modulating respiratory rhythm and motor activities (Woolf and Butcher, 1989; Yasui et al.,
1990; Fay and Norgren, 1997; Rukhadze and Kubin, 2007) and the essential physiological
function of respiration and clinical relevance of sleep-disordered breathing.
26
Chapter 2 METHODS
27
METHODS
2 Methods
2.1 Animal Care
Experiments were preformed on 22 male Wistar rats (Charles River) (mean body weight =
283.6g ± 2.18g, range 270-304g). Procedures conformed to the recommendations of the
Canadian Council on Animal Care and the University of Toronto Animal Care Committee
approved the protocols. Rats were housed individually, maintained on a 12-12h light/dark
cycle (lights on at 0700 h), and had free access to food and water.
2.2 Anesthesia and Surgical Procedures
Sterile surgery was performed under general anesthesia induced with isoflurane (4%). Rats
were intraperitoneally injected with buprenorphine (0.03 mg.kg-1) to minimize post-operative
pain, atropine sulphate (1 mg.kg-1) to minimize airway secretions, and saline (3 ml, 0.9%) for
fluid loading. A surgical plane of anesthesia, as judged by abolition of the pedal withdrawal
and corneal blink reflexes, was maintained with isoflurane (2-2.5%) administered with an
anesthesia mask placed over the snout. The rats spontaneously breathed a 50:50 mixture of air
and oxygen for the duration of the surgery. The abdomen, neck, and head regions were shaved
and cleaned with 70% alcohol and the antiseptic/germicide solution triadine (10% Providone-
Iodine, Triad disposables Inc., Brookfield, WI, USA). Sterile 1% chloramphenicol ointment
(Vetcom Inc., Upton, PQ, Canada) was applied to the cornea to prevent drying. With the rats
supine, the ventral surface of the genioglossus muscle was exposed via a submental incision
and dissection of the overlying geniohyoid and mylohyoid muscles. Two insulated, multi-
stranded stainless steel wires (AS636; Cooner Wire, Chatsworth, CA) were implanted
bilaterally into the genioglossus muscle and secured with sutures and tissue glue. Observation
28
of tongue protrusion in response to electrical stimulation (0.4 - 0.5V) during surgery was used
to confirm correct electrode placement. It has been shown previously using sections of the
medial branches of the hypoglossal nerves that genioglossus activity is recorded with such
electrode placements (Morrison et al., 2002). To measure diaphragm electromyogram (EMG)
activity, two insulated, multi-stranded stainless steel wires (AS636; Cooner Wire) were
sutured onto the costal diaphragm via an abdominal approach. The diaphragm and
genioglossus wires were tunneled subcutaneously to a small incision on the skull, and then the
sub-mental and abdominal incisions were closed with absorbable sutures (Polysorb 4-0,
Covidien, Norwalk, CT, USA).
The rats were then placed in a stereotaxic apparatus (Kopf Model 962, Tujunga, CA, USA)
with blunt ear bars. To ensure consistent positioning between animals the flat skull position
was achieved with an alignment tool (Kopf Model 944). Two multi-stranded stainless steel
wires were sutured onto the dorsal neck muscles to record the neck EMG. To record the
cortical electroencephalogram (EEG), two stainless steel screws (1.5mm diameter) attached to
insulated wires (30 gauge) were implanted in the skull. An additional screw implanted in the
same manner served as a ground electrode. Microdialysis guides (CMA/11, Chromatography
Sciences Company Inc. St. Laurent, QC, CA) were inserted bilaterally through small holes
(roughly 1.5mm in diameter) drilled 2mm lateral to the midline on both sides, and 0.57mm
rostral to true lambda (the point of intersection of the sagittal and lambdoid sutures) (Paxinos
and Watson, 1998). Both microdialysis guides (shortened by 3mm using a calliper and sharp
cutting tool) were implanted at the aforementioned coordinates and lowered to a depth of
3.75mm ventral to lambda, thus positioning the guide tips 4mm above the caudal-most part of
the PPTn pars compacta region (Rye et al., 1987; Paxinos and Watson, 1998).
A review of the literature shows that this region of the PPTn (i.e., from its caudal border with
the parabrachial complex to the nearest coronal section at which the decussation of the
superior cerebellar penduncle is fully elaborated) has the necessary connectivity to modulate
REM sleep and its respiratory phenotype (Figure 3). Specifically, unlike more rostral PPTn
locations, neurons in this area are REM sleep-active and project to the PRF
29
30
Figure 3. Schematic diagram of the efferent projections of the PPTn to key components
of the REM sleep and respiratory control networks in the rat. (Left) Coronal sections of
the rat pons containing the PPTn are shown in caudal to rostral order (front to back). Only the
caudal-most half of the PPTn is shown (-8.0mm to -7.3mm w.r.t. bregma). The rostral portion
of the PPTn from -7.04 to -6.72 w.r.t. bregma was omitted for clarity. No projections to
centers of sleep and breathing have been reported to originate from this region. An example of
NADPH-diaphorase histochemical labelling of PPTn cholinergic neurons, shows the position
of these neurons within the rat pons in actual tissue. (Right) Magnified portions of the PPTn
region taken from the immediately adjacent pontine sections, showing the distributions of
PPTn REM sleep-active neurons (purple), and PPTn neurons projecting to: (i) the hypoglossal
motor nucleus (orange), (ii) the rostroventrolateral medulla (yellow), and (iii) the pontine
reticular formation (blue). Distributions were taken from numerous previous studies (see
methods section 2.2 for references). All distributions overlap at the extreme caudal pole of the
PPTn (-8.0mm w.r.t. bregma). This site was therefore targeted for microdialysis delivery of 8-
OH-DPAT.
31
(Jones, 1990; Semba et al., 1990; Semba and Fibiger, 1992; Kohlmeier et al., 2002) and
thalamus (Semba et al., 1990; Semba and Fibiger, 1992; Bevan and Bolam, 1995), and so are
thought to modulate REM sleep and the associated electrocortical activation (Steriade and
McCarley, 2005a; Steriade et al., 2005a). Neurons in this sub-region of the PPTn also project
to the areas of the rostral ventrolateral medulla (Yasui et al., 1990) that contain respiratory
neurons critical to the generation of respiratory rhythm and pattern (Feldman and Del Negro,
2006), and to relevant motor pools such as the hypoglossal motor nucleus which innervates
the genioglossus muscle of the tongue (Woolf and Butcher, 1989; Fay and Norgren, 1997;
Rukhadze and Kubin, 2007), relaxation of which is instrumental to the pathogenesis of
obstructive sleep apnea (Remmers et al., 1978).
An internal cannula was placed inside each guide to keep it free of debris until the day of the
experiment. At the end of the surgery, all the electrodes were connected to pins and inserted
into a miniature plug (STC-89PI-220ABS, Carleton University, Ottawa, ON, Canada). The
plug and microdialysis guides were affixed to the skull with dental acrylic and anchor screws.
Upon completion of the surgery, rats were transferred to a clean cage and kept warm under a
heating lamp until fully recovered as judged by normal locomotor activity, grooming,
drinking and eating. The rats were given soft food for the first day after surgery and were
housed individually for a recovery period of 7-8 days prior to any experiments being
preformed.
2.3 Habituation
The evening before the experiment (~1900 hrs, i.e., the beginning of the active period), rats
were introduced to their recording environment for the purpose of habituation. The rats were
placed in a large open-topped bowl (Rodent Bowl, MD-1514, BAS Inc, West Lafayette, IN,
USA) mounted on a modified stand-alone turntable (Rat Turn MD-1404, BAS Inc), all of
which was housed within a noise-attenuated, electrically-shielded cubicle (EPC-010,
BRS/LVE Inc. Laurel, MD, USA). A light weight recording cable was connected to the plug
on the head of the rat, and this exited the chamber and was connected to the recording
32
apparatus. The rat turn device detects rotational movement of the animal via the recording
cable, and then rotates the rodent bowl in the opposite direction, so preventing >360○ rotation
of the animal and twisting of the cable and microdialysis tubing. Overall, this device
eliminated the need for an electrical commutator and liquid swivel.
Rats were provided with fresh bedding, food and water throughout the study. A video camera
located within the cubicle allowed for continuous visual monitoring without disturbing the
animal. The start time of the habituation procedure was kept constant between animals, the
reason for which was two-fold. Commencing habituation at the beginning of the dark phase /
active period was chosen to minimize the potential loss of sleep stemming from any potential
habituation associated stress (which may have led to changes in sleep on the day of the
experiment if the habituation was performed in the light phase / rest period the previous day),
while standardizing the time of habituation onset ensured that any rebound in sleep quantity
occurring on the day of the experiment would be comparable between animals.
2.4 Microdialysis
On the morning of the experiments at approximately 1000 hrs, the internal cannulae were
removed from the guides and microdialysis probes (CMA/11 14/01) were inserted without
handling the rats, since handling may excessively disrupt the rats and potentially produce
prolonged changes in sleep architecture. The probes projected 4mm from the tip of the guide
and were so targeted to the caudal-most part of the PPTn pars compacta region bilaterally
(described above – section 2.2). The probes were 240μm in diameter with a 1mm cuprophane
membrane and a 6,000 Dalton cut-off. Each probe was connected to FEP Teflon tubing
(inside diameter, 0.12 mm) with this tubing connected to 1.0 ml syringes via a zero dead
space switch (Uniswitch, BAS, West Lafayette, IN, USA). The probes were continuously
flushed with artificial cerebrospinal fluid (aCSF) at a flow rate of 2.1μl·min-1
using a syringe
pump and controller (MD-1001 and MD1020, BAS). The composition (mM) of aCSF was:
NaCl (125), KCl (3), KH2PO4 (1), CaCl2 (2), MgSO4 (1), NaHCO3 (25) and D-glucose (30).
33
The aCSF was warmed to 37°C and bubbled with CO2 to a pH of 7.38±0.005. The CaCl2 was
added after adjusting the temperature and pH.
2.5 Protocol
All experiments were performed during the day when rats normally sleep. As anticipated
probe insertion resulted in an immediate transient disruption of normal sleep architecture,
which stabilized (as judged by the disappearance of prolonged bouts of wakefulness, absence
of sleep fragmentation, and the regular occurrence of REM sleep) within 45-60min of probe
insertion. Nevertheless, to help ensure that sleep architecture had normalized prior to the
recording of baseline values, data obtained in the first two hours following probe insertion
were excluded from the analyses in all rats. In 3 of 22 rats disruption of sleep was still evident
in the second hour following probe insertion as judged by wakefulness accounting for > 63%
of the total recording time (TRT) during perfusion of aCSF, i.e., three standard deviations
greater than the mean quantity of wakefulness. Accordingly, these 3 animals were excluded
from analysis of sleep architecture per se, which was a-priori decision made regardless of the
sleep/wake architecture responses to subsequent manipulation of the PPTn. Nevertheless, the
data from these three rats were still included in the analysis of state phenomenology (i.e., the
postural and respiratory muscle activities, respiratory rate, electrocortical activity that
occurred within each sleep-wake state) because these state-specific phenomena depend on the
presence of the state rather than the amount of its occurrence.
The entire experimental protocol occurred over a 5 hour period (1200-1700 hrs). Two hours
of data were collected for each experimental condition (i.e., aCSF and drug intervention). To
ensure that all the data included in analysis were obtained during periods of time during
which an appreciable quantity of drug had accumulated in the tissue surrounding the
microdialysis probe, data obtained in the hour following switching the microdialysis perfusate
was excluded from all analyses (i.e., state architecture and phenomenology).
34
Since transitioning between behavioural states is sometimes associated with ambiguity in
sleep-wake state determination and unstable breathing, the following exclusion criteria were
adopted to help ensure that all data included in the analysis of state phenomenology were
obtained from temporally enduring periods within unequivocally defined sleep-wake states:
(i) all bouts of REM sleep and wakefulness lasting < 30s, and all bouts of non-REM sleep
lasting < 60s were excluded from the analysis of state phenomenology, (ii) if a state transition
occurred during a scoring epoch then that epoch was also excluded from the analysis of state
phenomenology.
The rats were visually monitored via a video camera for the duration of the experiments and
any changes in body posture were noted. Posture was characterized as either being extended,
partially-curled or the fetal position (Megirian et al., 1985). Since rats preferentially adopted
the partially-curled posture, all data obtained with the rats in other postures were excluded
from the analysis of state phenomenology in order control for any potential posture-dependent
effects on the magnitude of genioglossus muscle activity (Megirian et al., 1985) or any other
respiratory variables.
Following baseline recordings, during which time the PPTn was perfused with ACSF, the
perfusion medium was either maintained as ACSF without drug (i.e., time control, n=11 rats)
or switched to perfusion of the type 1A serotonin (5-HT1A) receptor agonist 8-hydroxy-2-(di-
n-propylamino) tetralin (8-OH-DPAT, Sigma, St. Louis, MO, USA) dissolved in ACSF at a
concentration of 10μM (n=12 rats). The time control group was included in order to
determine if any changes in sleep occurring during 8-OH-DPAT perfusion stemmed from
circadian influences on sleep-wake state architecture independent of an 8-OH-DPAT effect.
8-OH-DPAT at 10μM was chosen for this study because Thakkar and colleagues (1998),
using the same technique of local reverse microdialysis of the PPTn and simultaneous unit
recording of PPTn neurons, showed that this concentration of 8-OH-DPAT selectively
silenced the activity of REM sleep-active PPTn neurons while the discharge of wake/REM
35
sleep-active PPTn neurons was unaffected (figure 4). 5-HT1A receptor expression is thought to
be the principal characteristic distinguishing PPTn wake/REM sleep-active neurons from
REM sleep-active neurons (Steriade and McCarley, 2005b; Steriade et al., 2005a).
At the end of each experiment, as a positive control, rats were re-anesthetized with isoflurane
(as described above) and the genioglossus muscle electrodes were stimulated, as during
surgery, to verify the same responses. The rats were then euthanized with an overdose of
isoflurane (5% administered for approximately 20min) and perfused intracardially with 500
ml of 0.1M phosphate buffered saline pH 7.4) and 500 ml of 4% paraformaldehyde solution.
The brains were removed and fixed in a 1:1 solution of 4% paraformaldehyde and phosphate-
buffered saline overnight. Brains were then transferred to a 30% sucrose solution for at least
24 hours, after which they were rapidly frozen and cut into 50μM coronal sections with a
cryostat (Leica, CM 1850, Nussloch, Germany).
The PPTn is neuroanatomically defined by its cholinergic cell population (Rye et al., 1987)
and so in order to verify the correct probe placement, PPTn cholinergic cells were identified
using nicotinamide adenine dinucleotide phosphate (NADPH)-diaphorase histochemistry.
PPTn neurons that label positive for NADPH-diaphorase activity also contain the enzyme
required for synthesis of acetylcholine (Vincent et al., 1983). Although described in more
detail elsewhere (Vincent et al., 1983), slide-mounted sections were incubated in a solution
composed of; 0.1% NADPH (Sigma N-6505), 0.01% nitro blue tetrazolium (Sigma N-6876),
and 0.3% Triton X-100 (Sigma) in 0.05M tris buffered saline (pH of 7.4) and heated to 37○C.
Slides were lightly shaken during the approximately 30 min of incubation. The reaction was
terminated by submerging the slides in 0.1M phosphate-buffered saline (pH of 7.4). Sections
were then counterstained with neutral red, dehydrated, cleared, and cover-slipped. The
location of the microdialysis probe sites were determined from the stained sections and
marked on standard brain maps (Paxinos and Watson, 1998).
36
37
Figure 4. Effects of 8-OH-DPAT on the discharge rates of (Top) PPTn REM sleep-active
neurons and (bottom) PPTn wake/REM sleep-active PPTn neurons. Grand mean (± SEM) of
discharge rate in each behavioural state before (open circles, ACSF) and after (closed circles)
10µM 8-OH-DPAT was added to the microdialysis perfusate. Note that 8-OH-DPAT
produces a significant suppression of activity only in the case of REM sleep-active neurons.
The discharge of wake/REM sleep-active neurons is not affected by 8-OH-DPAT. Figure
adapted from (Thakkar et al., 1998).
38
As an additional positive control to affirm that any effects of 8-OH-DPAT on REM sleep
were indeed a drug effect on PPTn neurons, the distances between microdialysis lesion sites
and the caudal-most region of the PPTn pars compacta (as defined in section 2.2) for each
animal were correlated with the magnitude of changes in REM sleep time (as a percentage of
total sleep time) during 8-OH-DPAT perfusion. A significant negative correlation is expected
given that the concentration of drug acting at the PPTn would decrease the farther
microdialysis probes are located from the PPTn. The absence of a significant correlation
would indicate that changes in REM sleep during 8-OH-DPAT perfusion are the product of 8-
OH-DPAT acting elsewhere than the PPTn.
2.6 Recording Procedures
The electrical signals were amplified and filtered (Super-Z head-stage amplifiers and BMA-
400 amplifiers/filters, CWE Inc., Ardmore, PA, USA). The EEG was filtered between 1 and
100 Hz, whereas neck, genioglossus and diaphragm EMGs were filtered between 100 and
1000 Hz. The electrocardiogram was removed from the diaphragm EMG using an
oscilloscope and electronic blanker (Model SB-1, CWE Inc.). The moving-time averages
(time constant = 200ms) of the EMGs were also obtained (Model MA-821, CWE Inc.). The
raw EEG and genioglossus signals, along with the moving-time averages of the genioglossus,
diaphragm and neck EMGs were digitized and recorded on computer (Spike 2 software, 1401
interface, CED Ltd, Cambridge, UK). The moving-time averages of the EMG signals and
EEG were sampled at 2000Hz.
2.7 Data analysis
2.7.1 EMG signals
39
The data were analyzed in consecutive 5s time-bins. The genioglossus, diaphragm and neck
EMG signals were analyzed from the respective moving-time average signals (above
electrical zero) and were quantified in arbitrary units. Electrical zero was the voltage recorded
with the amplifier inputs grounded. The genioglossus and diaphragm signals were analyzed
on a breath-by-breath basis which corresponded to approximately 7 to 10 breaths for each 5s
epoch. For each breath, the analysis of the genioglossus EMG was time-locked to breathing as
defined by the peak and trough of the diaphragm signal. Genioglossus activity was quantified
as mean tonic activity (i.e. basal activity in expiration) and respiratory-related activity (i.e.
peak inspiratory activity – tonic activity), and average values for these measures of
genioglossus activity were calculated. Mean neck muscle activity, diaphragm amplitude and
respiratory rate were also calculated in the same consecutive 5s time-bins for all the periods
of sleep and wakefulness in each rat. One rat was excluded from the analysis of genioglossus
muscle activity as no respiratory-related activity was recorded under any circumstances
during the experiment, which is highly atypical of the rats in this and other studies from our
laboratory (Sood et al., 2005; Chan et al., 2006; Younes et al., 2007; Steenland et al., 2008),
and as such was a possible indicator that the electrodes were in a different tongue muscle in
that animal. In addition, in the potential event of there being little effect of 8-OH-DPAT on
the amplitude of respiratory-related genioglossus, such a minimal effect could not be reliably
attributed to a lack of drug effect per se and may simply have been due to the fact that there
was little/no signal to suppress.
Additional analyses of genioglossus activity were performed in REM sleep because
the rhythmic respiratory modulation of the genioglossus which typically persists throughout
wakefulness and non-REM sleep is absent during REM sleep and sporadic muscle twitching,
not obviously related to diaphragm activation, predominates. Accordingly, these REM sleep-
specific muscle twitching events in the genioglossus as well as the non-respiratory muscle of
the neck were analyzed by an additional procedure. Muscle twitching activity was analyzed
from the rectified raw genioglossus and neck muscle signals followed by derivation of the
moving time average signal using a shorter time constant of 30 ms. Using an impulse function
the background tonic activity was filtered out thereby isolating muscle twitches from the other
40
components of the signal. Upon isolating the muscle twitches, their frequency and peak
amplitude could be calculated.
2.7.2 EEG signals
The EEG was sampled at 2000Hz then analyzed on overlapping segments of 1024 samples,
windowed using a raised cosine (Hamming) function and subjected to fast Fourier transform
to yield the power spectrum. The window was advanced in steps of 512 samples, and the
mean power spectrum of the EEG signal over each 5s epoch was calculated. The power
contained within six frequency bands was recorded as absolute power and also as a
percentage of the total power of the signal. The band limits were δ2 (0.5-2 Hz), δ1 (2-4 Hz), θ
(4-7.5 Hz), α (7.5-13.5 Hz), β1 (13.5-20 Hz), β2 (20-30 Hz).
2.7.3 Identification of Sleep-Wake States
Sleep-wake states were identified by visual inspection and classified into wakefulness, non-
REM, and REM sleep according to standard scoring criteria (Horner et al., 1998).
Determination of sleep-wake states was made with reference to EEG and neck muscle EMG
recordings only, i.e., without reference to the phenomenological variables such as
genioglossus activity and respiratory rate that also change in distinct patterns across sleep-
wake states but which were hypothesized to change in response to the interventions. All
periods of wakefulness in which rats were eating, drinking, grooming or engaged in some
overt behaviour were classified as active wakefulness. Periods of wakefulness being
characterized by relatively little or no behavioural activity were classified as quiet
wakefulness. During active wakefulness, the diaphragm EMG recordings can become
contaminated by movement-related artifacts and so periods of active wakefulness were
excluded from analysis of respiratory variables since an inability to identify the peak and
41
trough of each diaphragm breath prevents a meaningful analysis of diaphragm amplitude,
respiratory rate, and respiratory-related genioglossus activity.
2.7.4 Analysis of Sleep Architecture
For each condition (i.e. ACSF or 8-OH-DPAT) the total quantity of wakefulness, non-
REM, and REM sleep were calculated as a percentage of the total recording time (2hr; see
above for data collection inclusion criteria). Amounts of non-REM and REM sleep were also
calculated as a percentage of total sleep time, since changes in non-REM and REM sleep
quantities relative to one another (i.e., independent of wakefulness) are indicative of changes
in the mechanisms of non-REM/REM sleep cycling. The mean bout frequency and duration
for each sleep-wake state was also calculated. Distributions of wakefulness and non-REM
sleep bout length are often non-normal and so to supplement the mean data; bout frequency
histograms were generated. For both wakefulness and non-REM sleep bouts were pooled
across animals according to condition (i.e., aCSF or 8-OH-DPAT) and sorted into bins
according to the following duration ranges: <20s, 21-80s, 81-140s, 141-200s, 201-260s,
>260s. For each state, the bout number for each bin during the control period was compared
to the bout number of the equivalent bin during 8-OH-DPAT delivery in order to determine
changes in the prevalence of bouts of varying length.
We aimed to identify whether or not REM sleep-active PPTn neurons exert control
over two important factors in REM sleep initiation, namely, the strength of REM sleep drive
which builds prior to and is responsible for REM sleep onset, as well as the threshold level of
the drive which need be breached to trigger REM sleep onset. We used an algorithm
developed by Benington et al, (1994) to quantify the strength of REM sleep drive during non-
REM to REM sleep transitionary periods (NRTs) by determining of the magnitude of
stereotypical electrographic changes known to herald the onset of REM sleep. More
specifically, the algorithm identifies periods of non-REM sleep lasting at least 40s, in which
EEG delta power is declining while theta and alpha power is high. From these electrographic
variables, a score (non-REM/REM sleep transition indicator value, NIV) can be calculated for
42
each NRT using the following equation: NIV= (max. theta power)(max. alpha power)(max.
change in delta power). We consider NIV as a valid indicator of relative REM sleep drive
since high NIV is associated with a high likelihood of successful transitioning from non-REM
to REM sleep (Benington et al., 1994).
We defined an NIV threshold to demarcate between periods of high and low REM
sleep drive. The high/low REM sleep demarcation threshold is determined as follows. Each
NIV score was converted to a percentage of the maximum NIV score obtained under control
conditions in the same animal. NIV scores from all animals were then pooled into bins of
equal width (i.e., 0-20% of maximum NIV, 20-40% of maximum NIV, and so on). The
percentage of successful REM sleep onset [i.e., (the number NRT’s immediately followed by
a bout of REM sleep)/the total number NRT’s) was calculated for each bin. The histogram of
successful REM sleep onset percentage versus normalized NIV describes a sigmoid curve;
high normalized NIV’s being associated with a high likelihood of REM sleep onset. The
demarcation threshold is equivalent to the normalized NIV level at which the slope of the
NIV versus successful REM sleep onset curve is maximal. By applying this threshold to the
pool of REM sleep bouts from each animal the percentage of REM sleep bouts issuing from
periods of low versus high REM sleep drive was calculated. Any changes in the relative level
of REM sleep drive required to successfully initiate REM sleep may be taken as an indication
of changes in REM sleep induction threshold.
In some cases bouts of REM sleep can be immediately followed by another REM sleep bout
usually of very short duration. This phenomenon is known as REM sleep clustering (Amici et
al., 2000). We defined REM sleep clusters as two or more REM sleep bouts being separated
by less than 60sec. For the purposes of this analysis, we considered REM sleep a cluster as
single, although fragmented, episode of REM sleep.
43
2.7.5 Averaging data within and between rats
Each rat served as its own control with all interventions performed in one experiment,
therefore allowing for consistent effects of experimental condition (e.g. ACSF followed by 8-
OH-DPAT or ACSF) to be observed across sleep-wake states within and between rats. Data
collected during wakefulness, non-REM and REM sleep were analyzed for each experimental
condition in each rat. Then for each animal a grand mean was calculated for each variable, for
each sleep-wake state, and for each drug delivered to the PPTn.
2.7.6 Statistical Analysis
The analyses performed for each statistical test are included in the text where appropriate. For
all comparisons, differences were considered significant if the null hypothesis was rejected at
P < 0.05 using a two-tailed test. Analyses were performed using Sigmastat (SPSS, Chicago,
IL).
44
Chapter 3 RESULTS
45
RESULTS
3 Results
3.1 Effects of Bilateral Delivery of 8-OH-DPAT to the PPTn on Sleep Architecture in Freely Behaving Rats
In total, for sleep architecture analysis in 20 rats, 76 hours of data were included. 18 hours of
data were recorded during 8-OH-DPAT perfusion of the PPTn, while 58 hours of data were
obtained during administration of ACSF between the within-animal controls (i.e., rats also
administered 8-OH-DPAT) and the time controls (i.e., rats in which ACSF perfusion was
maintained for the duration of the study). Figure 5A shows an example of lesions left by
bilateral microdialysis probes implanted into the PPTn as defined by the presence of NADPH-
diaphorase histochemically labeled cholinergic neurons. The distribution of microdialysis
sites from all the experiments are shown in figure 5B-C; all sites being either in or
immediately adjacent to the PPTn.
Based upon the prevailing hypothesis that REM sleep-active PPTn neurons promote the
generation of REM sleep, it would be expected that bilateral microdialysis delivery of 8-OH-
DPAT to the PPTn for the selective inhibition of REM sleep-active neurons (Thakkar et al.,
1998) would result in suppression of REM sleep. Fig.6 and table 1, show that in direct
opposition to this hypothesis, 8-OH-DPAT delivery to the PPTn increased REM sleep as a
%TRT; %TRT occupied by REM sleep during 8-OH-DPAT perfusion was significantly
higher than the within-animal ACSF control (P=0.004, post-hoc Bonferroni t-test). REM
sleep as a %TRT did not change significantly in the time control group (P=0.14, post-hoc
Bonferroni t-test) and therefore an effect of time cannot account for the increased REM sleep
recorded during 8-OH-DPAT perfusion of the PPTn. Despite being in opposition with the
prevailing hypothesis stated above, these findings are consistent with the previously reported
46
47
Figure 5. Example and group data showing the location of microdialysis probes from all
experiments with 8-OH-DPAT delivery at the PPTn and ACSF time-controls. (A) Example of
a coronal section of tissue from a single experimental animal containing bilateral
microdialysis probe lesion sites located immediately adjacent to the PPTn, as defined using
NADPH-diaphorase histochemical labelling of cholinergic neurons. (B) Coronal diagrams of
the rat pons showing the locations of all microdialysis probe sites from all rats administered
8-OH-DPAT (n=12). (C) Coronal diagrams showing the locations of microdialysis probe sites
from time-control animals (n=11). Grey rectangles represent the space occupied by the semi-
permeable membrane portion of the microdialysis probes. Red squares represent the
epicentres of drug diffusion. (D) Graphs showing the correlation between the magnitude of
changes in REM sleep time as a %TRT during 8-OH-DPAT perfusion and distances from
either the left-side lesion sites (top) or right-sided lesion sites (bottom) to the caudal pole of
the ispilateral PPTn. Abbreviations: Aq, aqueduct; CnF, cuneform nucleus; DRV, ventral part
of dorsal raphe nucleus; xscp, decussation superior cerebellar peduncle; PnO, pontine
reticular nucleus, oral part; PPTn, pedunculopontine tegmental nucleus (marked in yellow).
48
49
Figure 6. (A) Shown are two separate hypnograms depicting sleep-wake state architecture
over a 2 hour period in the presence of ACSF or 8-OH-DPAT at the PPTn in a single rat. (B)
Group data showing the effects of 8-OH-DPAT versus time (i.e., ACSF in both periods A and
B) on wakefulness, non-REM, and REM sleep as a percentage of the total the total recording
time. Effects are also shown on non-REM and REM sleep as a percentage of total sleep time.
For each condition, the total recording time was 2 hours. Values are means ± SEM (n=20
rats). *Significantly different (P< 0.05) from within-animal ACSF control (period A).
‡Significantly different (P< 0.05) from time-matched control (period B). Data are consistent
with PPTn REM sleep-active neurons acting to suppress REM sleep because their inhibition
with 8-OH-DPAT increases REM sleep.
50
51
effects of non-selective PPTn inactivation (Torterolo et al., 2002; Pal and Mallick, 2004; Lu
et al., 2006; Pal and Mallick, 2009). Also consistent with these previous studies is our
observation that the increase in REM sleep as a %TRT was accompanied by a decrease in
wakefulness during 8-OH-DPAT administration. The %TRT occupied by wakefulness in the
presence of 8-OH-DPAT was significantly less than at baseline (P=0.034, post-hoc
Bonferroni t-test) despite the fact that as a function of time, in the absence of 8-OH DPAT,
wakefulness as a %TRT increased relative to baseline (P<0.001, post-hoc Bonferroni t-test).
In the 8-OH-DPAT group, irrespective of the time control group, the decrease in wakefulness
as a %TRT due to 8-OH-DPAT, was completely accounted for by the increase in REM sleep
since non-REM sleep as a %TRT did not change significantly from the within-animal control
(P=0.439, post-hoc Bonferroni t-test). However, when comparing to the time control group,
the decrease in wakefulness is not entirely accounted for by the increase in REM sleep, since
non-REM sleep as a %TRT increases significantly compared to the time-matched control
(i.e., ACSF administered in the same time period as 8-OH-DPAT) (P<0.001, post-hoc
Bonferroni t-test).
In the case of previous studies (Torterolo et al., 2002; Pal and Mallick, 2004; Lu et al., 2006;
Pal and Mallick, 2009) it has been argued that increased REM sleep following en masse
suppression of PPTn activity is a secondary by-product of a primary suppression of
wakefulness (Torterolo et al., 2002). It is highly unlikely that the proposed mechanism of
wakefulness suppression is operative in our case, since it involves the loss of the wake
promoting influence of wake/REM sleep-active neurons which are not responsive to 8-OH-
DPAT treatment (Thakkar et al., 1998), but nevertheless I sought to determine if the primary
effect of 8-OH-DPAT was wakefulness suppression. If the increase in REM sleep was a
passive by-product of an increase in TST stemming from active wakefulness suppression, it
would be expected that that non-REM and REM sleep would increase in proportion to one
another. In other words, independent of changes in wakefulness, the proportion of the TST
occupied by non-REM and REM sleep ought not to change during 8-OH-DPAT perfusion
relative to control. However, this was not the case since the increase in the total sleep time
during 8-OH-DPAT perfusion was disproportionately comprised of REM sleep. During 8-
OH-DPAT delivery, REM sleep (as a %TST) increased relative to the within-animal control
52
(P=0.003, post-hoc Bonferroni t-test) and the time-matched control (P=0.045, post-hoc
Bonferroni t-test).
The distances between microdialysis lesion sites and the PPTn was significantly correlated
with the magnitude of changes in REM sleep time (as a %TST) during 8-OH-DPAT perfusion
[figure 5D, r2=0.659 and p=0.008 (left-sided lesion sites), r
2=0.465and p=0.043 (right-sided
lesion sites)]. Since the concentration of 8-OH-DPAT acting at the PPTn will decline with
increasing distance of microdialysis probes sites, the significant correlation between
microdialysis probe distance and the magnitude of drug effect, represents a dose dependency
of the PPTn response to 8-OH-DPAT. Ultimately the most parsimonious explanation of the
above findings is that 8-OH-DPAT responsive REM sleep-active PPTn neurons act to
suppress REM sleep in opposition to the prevailing hypothesis that these cells generate this
state.
3.2 Effects of Bilateral Delivery of 8-OH-DPAT to the PPTn on Sleep Micro-architecture in Freely Behaving Rats
As shown in Figure 7C and table 1, the increase in REM sleep during 8-OH-DPAT
delivery to the PPT resulted entirely from an increase in the mean bout frequency of REM
sleep episodes (P<0.001, post-hoc Bonferroni t-test) since no significant effect on REM sleep
bout duration was observed (Figure 7F, F1,17= 0.0114, P=0.916, 2-way ANOVA). The
reduction in wakefulness as a %TRT accompanying the increased REM sleep occurred
despite an increase in the frequency of wakefulness bouts (Figure 7A, P=0.02, post-hoc
Bonferroni t-test). Reduced wakefulness as a %TRT was driven by the reduction in the mean
wakefulness bout duration (Figure 7D, P=0.003, post-hoc Bonferroni t-test). The reduction in
the average duration of wakefulness bouts stemmed from an increased frequency of short
awakenings from sleep. The frequency of wakefulness bouts lasting <20s increased during 8-
OH-DPAT perfusion (P<0.001, post-hoc Bonferroni t-test) while the frequency longer
53
54
Figure 7. The effects of 8-OH-DPAT on the frequencies (A-C), and durations of wakefulness,
non-REM, and REM sleep bouts, relative to within-animal and ACSF time controls (D-F).
Also shown are the frequencies of bouts belonging to different duration ranges for
wakefulness and non-REM sleep (G-H). For each treatment, values were obtained from data
collected over a 2 hour period. Values are means ± SEM (n=20 rats). *Significantly different
(P< 0.05) from within-animal ACSF control. #Significant (P< 0.05) effect of time
independent of drug. These data are consistent with PPTn REM sleep-active neurons
suppressing REM sleep via a suppression of REM sleep initiation rather than maintenance.
55
duration (>20s) bouts was reduced (Figure 7G, P=0.022, post-hoc Bonferroni t-test) during 8-
OH-DPAT administration. Non-REM sleep as a %TRT did not change relative to baseline
during 8-OH-DPAT perfusion despite, changes to non-REM sleep micro-architecture. Non-
REM sleep bout frequency increased (Figure 7E, P=0.015, post-hoc Bonferroni t-test) while
bout duration decreased (Figure 7E), offsetting one another and producing no net effect on
non-REM sleep time. The decrease in non-REM sleep bout duration occurred independently
of 8-OH-DPAT treatment (F1,17= 2.993, P=0.102, 2-way ANOVA) and instead occurred as a
function of time (F1,17= 28.870, P<0.001, 2-way ANOVA). Therefore, the increase in non-
REM sleep time due to 8-OH-DPAT relative to the time-matched control was driven by the
increase in the non-REM sleep bout frequency, particularly the frequency of short bouts
lasting <20s (Figure 7H P=0.049, post-hoc Bonferroni t-test), or lasting between 20-80s
(P<0.001, post-hoc Bonferroni t-test).
3.3 Effects of Bilateral Perfusion of 8-OH-DPAT to the PPTn on the Drive Threshold for REM Sleep Induction
As mentioned above, the increase in time spent in REM sleep during delivery of 8-OH-DPAT
to the PPTn stemmed from an increase in the frequency rather than the duration of REM sleep
bouts implicating REM sleep-active neuron in mechanisms of REM sleep initiation rather
than maintenance. REM sleep-active PPTn neurons could restrain initiation of REM sleep via
suppression of REM sleep drive processes or by elevating the threshold of REM sleep drive
required to induce this state. I sought to further define the functional role of this cell
population in REM sleep initiation by determining whether the increased REM sleep during
8-OH-DPAT delivery to the PPTn resulted from increased REM sleep drive or a decrease in
the drive threshold for REM sleep onset.
Figure 9B. shows that under control conditions, the relationship between the NIV value for
non-REM/REM sleep transitions and the percentage of transitions culminating in bona fide
56
57
Figure 8. The effects of 8-OH-DPAT at the PPTn on EEG activity. Shown are the group data
for power in frequency bands ranging from 0.5-30Hz, expressed as the percentage of the total
power. #Significant difference (P < 0.05) between respective sleep-wake states. Values are
means ± SEM (n=12 rats). Mean values for each individual rat, were first calculated from the
population of values for all 5-sec epochs during each sleep-wake, in the 2 hours of either
ACSF or 8-OH-DPAT administration. 8-OH-DPAT had no significant effects on EEG power
in any frequency band.
58
59
Figure 9. (A) The same hypnograms shown in figure 6A, depicting sleep-wake state
architecture in the presence of ACSF or 8-OH-DPAT at the PPTn, but with colour-coding of
the non-REM sleep records according to the prevailing non-REM/REM sleep transition
indicator value (NIV). (B) Graph showing the relationship between NIV value for non-REM /
REM sleep transitionary periods (NRT’s) and the incidence of successful transitioning into
REM sleep. Each point represents the percentage of NRTs from all animals under control
conditions (n=9), having NIVs belonging to a certain range (0-20,20-40,....,80-100), that
result in a successful transition to REM sleep. The sigmoidal curve fit to the data points was
used to define the threshold between NIVs indicative of low and high REM sleep transition
propensity (see methods section 2.7.4). I interpret high REM sleep transition propensity as
being reflective of high REM sleep drive. (C) Graph showing the effect of 8-OH-DPAT
delivery at the PPTn on the percentage of REM sleep bouts issuing from periods of high and
low REM sleep drive. Values are averages (± SEM) of individual means calculated for each
animal. *Significantly different (P< 0.05) from within-animal ACSF control. Data are
consistent with PPTn REM sleep-active neurons suppressing REM sleep initiation via the
elevation of the drive threshold for REM sleep onset.
60
REM sleep bouts describes a sigmoidal curve, in agreement with the initial findings of
Bennigton and Heller (1994) (see section 2.7.4 of the methods). Hence, high NIV is
associated with a high likelihood of REM sleep onset, and since REM sleep episodes are
necessarily the product of local maximums in REM sleep drive, high NIV may act as an index
of high REM sleep drive. The demarcation threshold between high and low drive NIVs was
determined to be a NIV of 40 (see methods section 2.7.4 for threshold determination
procedure). This threshold is considered a valid predictor of high and low REM sleep drive
since under control conditions an average of 91.3 ± 3.5% of REM sleep episodes were
preceded by NRTs assigned NIVs above the demarcation threshold. Importantly, as depicted
in figure 8, 8-OH-DPAT treatment did not result in significant changes to EEG power.
Changes in EEG power (particularly in the α, δ, and θ bands) could theoretically produce
artifactual changes in NIVs during 8-OH-DPAT delivery relative to baseline which would not
be necessarily representative of changes in REM sleep drive.
The mean proportion of REM sleep episodes resulting from periods of high REM sleep-drive
decreased from 91.3±3.5% to 66.1±10.6%, (Figure 9C, p=0.043, paired t-test) during 8-OH-
DPAT treatment. The hypnogram depicted in figure 9A shows that in the presence of 8-OH-
DPAT that bouts of REM sleep indicated in red, are more often preceded by periods of non-
REM sleep coded by cooler colours (i.e., low NIV) indicative of low REM sleep drive (8.7%
- 37.2%, P=0.019, paired t-test). These findings suggest that PPTn REM sleep-active neurons
act to increase REM sleep drive threshold, thus preventing premature transitioning into REM
sleep.
3.4 Effects of Bilateral Delivery of 8-OH-DPAT to the PPTn on Phenomenology within Sleep-Wake States.
In the analysis of state phenomenology, a total of 23,246 5-sec epochs (i.e., a total of 32.3
hours of data, 89.7% of all epochs recorded) were included, of which 5,768 epochs were from
61
62
Figure 10. Representative example from single rat showing the effect of ACSF (A) versus 8-
OH-DPAT (B) perfusion into the PPTn, on sleep-wake state indicator variables and
respiratory muscle activity across states of sleep and wakefulness. Apart from breaks during
non-REM sleep, the tracings are continuous records of activity from wakefulness, to non-
REM, to REM sleep. Changes in state are indicated by the hypnograms at the top of the
tracings and by alternation between shaded and non-shaded areas. Breathing rate is displayed,
with each point representing the mean respiratory rate over a 5-sec epoch. Electromyograms
for the diaphragm, genioglossus, and nuchal musculature are displayed as their moving-time
averages (MTA) in arbitrary units. The raw signals of the genioglossus and nuchal muscles
are also shown. Representative examples of the electroencephalogram (EEG) from each
sleep-wake state are shown on a larger time-scale than the other tracings for the purposes of
resolvability.
63
64
Figure 11. The effects of 8-OH-DPAT at the PPTn on (A) the amplitude of respiratory-
related genioglossus muscle activity (A), respiratory rate (B), and diaphragm amplitude (C),
across sleep-wake states. Values are means ± SEM (n=12 rats). Mean values for each
individual rat were first calculated from the population of values for all 5-sec epochs, during
each sleep-wake sate, in the 2 hours of either ACSF or 8-OH-DPAT administration at the
PPTn. *Significant effect of drug, relative to ACSF control, independent of sleep-wake state
(P< 0.05). These data are consistent with REM sleep-active PPTn neurons acting to depress
breathing in a state-independent manner.
65
periods of quiet wakefulness, 13,574 were from non-REM sleep, and 3,904 epochs were from
REM sleep. Of these epochs, 11,820 and 11,426 were obtained during ACSF and 8-OH-
DPAT perfusion of the PPTn respectively.
I sought to determine if REM sleep-active PPTn neurons contribute to the elaboration of the
respiratory phenotype of REM sleep given that innervation of respiratory network by PPTn
neurons has been previously demonstrated (Yasui et al., 1990; Rukhadze and Kubin, 2007).
The amplitude of diaphragm muscle activity was not affected by 8-OH-DPAT delivery to the
PPTn (Figure 11C, F1,11= 0.223, P = 0.646, 2-way ANOVA-RM). As depicted in figures 10
and 11, 8-OH-DPAT did produce an increase in respiratory rate (F1,11= 14.00, P = 0.003, 2-
way ANOVA-RM) that occurred independently of the prevailing sleep-wake state (F2,11 =
0.560, P =0.579).
Normally, the magnitude of the respiratory component of genioglossus muscle activity is
progressively suppressed across the sleep wake cycle, being highest in wakefulness,
moderately suppressed during non-REM sleep, and almost completely abrogated in REM
sleep (see figure 10A). In the presence of 8-OH-DPAT at the PPTn, an increase in
respiratory-related genioglossus activity is evident across all states of sleep and wakefulness
(figure 10B). This observation is reflected by the group data (figure 11A), which shows that
the 8-OH-DPAT mediated increase in respiratory-related genioglossus activity (F1,20=6.69,
P=0.027, 2-way ANOVA-RM) occurred independently of the prevailing sleep-wake state
(F2,10 =0.19, P =0.828). Even though the normal decline of genioglossus activity from non-
REM to REM sleep was still present during 8-OH-DPAT perfusion, notice in figure 10B, that
in the presence of 8-OH-DPAT, respiratory activation of the genioglossus muscle, albeit
diminished in magnitude, persists for a period after REM sleep onset. This is significant on
the basis that the complete loss of muscle activity in the genioglossus muscle, as with most
skeletal muscles, is normally coincident with REM sleep onset owing to the influence of
powerful REM sleep-specific inhibitory processes.
66
There was no significant effect of 8-OH-DPAT delivery at the PPTn on tonic (non-
respiratory) activation of genioglossus muscle activity across states of sleep and wakefulness
(F1,11=0.717, P = 0.415, 2-way ANOVA-RM). Figure 12A shows that despite this
insignificance, there is a noticeable increase in tonic levels of genioglossus muscle activity
during REM sleep. This increase stems from the fact that tonic levels of genioglossus muscle
activity during REM sleep are a combination of the background level of tonus and high
amplitude REM sleep-related muscle twitching (see figure 10A for reference). Figure 13 A-B
shows that the amplitude (P=0.037, paired t-test) and frequency (P=0.019, paired t-test) of
REM sleep-related muscle twitching, punctuating motor atonia, is increased with 8-OH-
DPAT perfusion of the PPTn relative to ACSF control. In contrast to the respiratory
genioglossus muscle, the frequency (figure 13A, p=0.871, paired t-test) and amplitude (figure
12B, p=0.094, paired t-test) of REM sleep-related muscle twitching in the non-respiratory
nuchal musculature was not significantly affected by 8-OH-DPAT treatment.
67
68
Figure 12. The effects of 8-OH-DPAT delivery at the PPTn on tonic muscle activation across
sleep-wake states. (A) Effect of 8-OH-DPAT at the PPTn on tonic muscle activity in the
genioglossus muscle during active and quiet wakefulness, as well as non-REM and REM
sleep. Active wakefulness is defined by the presence of overt behaviour and movement, while
quiet wakefulness is defined by behavioural inactivity. During active wakefulness levels of
tonic activity represent a mixture of a background tone and behavioural activation. Likewise,
included in levels of tonic muscle activation during REM sleep is twitching activity, arising in
part, from intense activation of behavioural pre-motor pathways. The increase in tonic levels
of genioglossus muscle activity during REM sleep reflects the increase in REM sleep-related
muscle twitch frequency and amplitude (Fig. 12A). The lack of behavioural activation during
active wakefulness compared to REM sleep may reflect a REM sleep-specific restraint of
behavioural activation of the genioglossus muscle.(B) Effect of 8-OH-DPAT at the PPTn on
tonic muscle activity in the nuchal musculature during active and quiet wakefulness as well as
non-REM and REM sleep. Values are means ± SEM (n=12 rats). Mean values for each
individual rat, were first calculated from the population of values for all 5-sec epochs during
each sleep-wake state, in the 2 hours of either ACSF or 8-OH-DPAT administration. n.s.= no
significant effect of drug independent of sleep-wake state.
69
70
Figure 13. Effects of 8-OH-DPAT delivery at the PPTn on the, (A) frequency and (B)
amplitude of REM sleep-related twitch events in the respiratory genioglossus muscle versus
the non-respiratory nuchal musculature (shaded section). Values are means ± SEM (n=12
rats). *Significantly different (P< 0.05) ACSF control. These data are consistent with PPTn
REM sleep-active neurons acting to specifically suppress activation of respiratory
musculature during REM sleep.
71
Chapter 3 Discussion
72
DISCUSSION
4 Discussion
The neuroscience of brain states aims to describe how global states of the brain are generated
and how these states impact the physiology of an organism. This study makes two important
contributions to this field. This study generates new knowledge of the mechanisms underlying
the generation of the REM sleep state and its impact on respiratory network activity by
characterizing the functional role of a critical component in these mechanisms: the PPTn
REM sleep-active cell-group. Since early formulations of the reciprocal interaction model of
REM sleep generation (Hobson et al., 1975; McCarley and Hobson, 1975a), PPTn REM
sleep-active neurons have been considered to be causally involved in the generation of REM
sleep. However, causal linkage between REM sleep active PPTn neurons and REM sleep
generation has never been explicitly established. This study is the first to directly test this
hypothesis because I am the first to demonstrate the effect of selectively inhibiting 8-OH-
DPAT responsive REM sleep-active PPTn neurons on the state of REM sleep. I showed that
selective inhibition of REM sleep-active PPTn neurons increased REM sleep, indicating that
this cell group normally acts to suppress the state of REM sleep rather than generate it as the
prevailing hypothesis would suggest. This study is the first to determine the involvement of
REM sleep-active PPTn neurons in the phenomenological changes in respiratory activity that
occur during REM sleep. The study of breathing during sleep is important, on the basis that
sleep results in fundamental changes in respiratory muscle activity and control mechanisms.
Disturbances of normal breathing are most significant during REM sleep, when respiratory
rate becomes heightened and irregular while the activity of certain respiratory muscles is
completely suppressed, such as occurs in the genioglossus muscle of the tongue, an effect
which precipitates obstructive sleep apnea (OSA)(Remmers et al., 1978). Our results indicate
that the REM sleep-active PPTn cell-group is an important modulator of upper-airway
patency, especially during REM sleep when this cell group produces a combined suppression
of the behavioural and respiratory components of genioglossus muscle activity. Moreover,
PPTn REM sleep-active cells produce a state-independent depression of respiratory activity
73
via suppressive effects on breathing rate and respiratory genioglossus activation. Therefore
from a clinical perspective, the actions of REM sleep-active PPTn neurons likely contribute to
the precipitation of sleep disordered breathing.
4.1 REM Sleep-active PPTn Neurons Suppress REM Sleep
The observed increase in REM sleep, while selectively inhibiting PPTn REM sleep-active
neurons, is consistent with findings of previous studies reporting enhanced REM sleep
following non-selective inactivation of the PPTn. Increased REM sleep has been observed
following spatially restricted chemical lesioning of the PPTn (Lu et al., 2006), and
microinjection of GABAA receptor agonists at the PPTn in both cats (Torterolo et al., 2006)
and rats (Pal and Mallick, 2004, 2009). These findings could certainly be taken to suggest that
PPTn neurons suppress REM sleep. However, in order to square the observation that PPTn
inactivation increases REM sleep with the notion that PPTn REM sleep-active neurons
generate REM sleep, it has been argued that the primary effect of PPTn inactivation is
wakefulness suppression, stemming from inhibition of wake/REM sleep-active PPTn neurons
(Torterolo et al., 2002; Steriade et al., 2005a). Consistent with the idea of wakefulness
suppression, increases in REM sleep following PPTn inactivation were often accompanied by
reduced wakefulness. Suppression of wakefulness necessarily results in an increase in total
sleep time. The increase in REM sleep stemming from wakefulness suppression could mask
the decrease in REM sleep that would follow from the simultaneous loss of REM sleep active
neuron activity, according to the prevailing hypothesis that REM sleep active neurons
generate REM sleep. Glutamateric stimulation of the PPTn produces prolonged bouts of
wakefulness (Datta and Siwek, 1997; Datta et al., 2001b; Datta et al., 2001a), showing that
when manipulating the PPTn en masse, that effects on wakefulness can predominate over any
effects on REM sleep. The predominance of wakefulness effects is not surprising given that
wake/REM sleep active PPTn neurons are more abundant than their REM sleep active
counterparts (Rye et al., 1987). The fundamental difference between these two subpopulations
of the PPTn is that REM sleep-active neurons are inactive during wakefulness due to a type-
1A serotonergic receptor (5HT1AR) mediated inhibition that is not shared by wake/REM sleep
74
active neurons (Thakkar et al., 1998). The dramatic increase in the discharge of PPTn REM
sleep-active neurons immediately prior to and during REM sleep appears to stem from the
withdrawal of serotonergic inhibition at this time. Using the 5HT1AR agonist, 8-OH-DPAT, I
was able to discriminate between the disparate functional influences of the PPTn. Although I
cannot possibly account for the effects of 8-OH-DPAT on every neuronal subtype in the
PPTn, what is important, is that I can account for the effects of microdialized 8-OH-DPAT on
the two PPTn subpopulations implicated in sleep-wake state regulation (Thakkar et al., 1998).
Given that 8-OH-DPAT has no effect on the activity of wake/REM sleep-active neurons, the
observed increase in REM sleep during 8-OH-DPAT perfusion cannot be explained by
inadvertent suppression of the wake/REM sleep active cell group. Nevertheless, I provide
evidence that the increase in REM sleep during 8-OH-DPAT perfusion is not a by-product of
wakefulness suppression. I have shown that during 8-OH-DPAT treatment that the increase in
total sleep was disproportionately comprised of REM sleep (i.e., REM sleep increased as a
%TST at the expense of non-REM sleep during 8-OH-DPAT perfusion). This is significant
because if the increase in REM sleep were a passive compensation for wakefulness
suppression, it would be expected that independent of changes in wakefulness, the proportion
of the TST occupied by non-REM and REM sleep ought not to change during 8-OH-DPAT
perfusion. The observed increase in REM sleep during 8-OH-DPAT perfusion is highly
inconsistent with the notion that any subset of PPT neurons generates REM sleep. This is
because, even if the predominant effect of PPTn inactivation were wakefulness suppression,
any underlying REM sleep suppression ought to manifest as a decrease in REM sleep as a
%TST (i.e., independent of changes in wakefulness). Therefore based on the finding that
REM sleep increases during 8-OH-DPAT perfusion as %TST, I most parsimoniously
conclude that PPTn REM sleep active neurons act to suppress the state of REM sleep.
4.2 Mechanism of REM sleep Suppression by PPTn REM Sleep-Active Neurons
Mechanisms of REM sleep regulation can be broadly divided into mechanisms of REM sleep
initiation and mechanisms of maintenance which sustain REM sleep episodes once initiated. I
75
have shown that while selectively inhibiting PPTn REM sleep-active neurons that the increase
in REM sleep is driven by an increase in REM sleep bout frequency. This result shows that
PPTn REM sleep-active neurons suppress the occurrence of REM sleep by preventing
initiation of this state. Moreover, the lack of an effect on REM sleep bout duration, shows that
PPTn REM sleep-active neurons do not play a role in the maintenance of REM sleep. It
would appear then that in non-REM / REM sleep transitionary periods when PPTn REM
sleep-active neurons become more active (Steriade et al., 1990; Steriade and McCarley,
2005b) that they effectively reduce the probability of REM sleep onset, until REM sleep is
successfully initiated and maintenance mechanisms are engaged at which point these neurons
are rendered powerless to restrain REM sleep despite their continued firing.
In order to understand how REM sleep-active PPTn neurons are involved in REM sleep
initiation I must first describe the characteristics of non-REM / REM sleep transitions. REM
sleep initiation is a progressive process. The potentials of reticular neuronal pools, responsible
for the generation of REM sleep undergo gradual depolarization in advance of transitions into
REM sleep (Hobson et al., 1974; McCarley and Hobson, 1975b; Ito and McCarley, 1984; Ito
et al., 2002). This augmentation in pontoreticular activity is associated with overt
stereotypical changes in electrographic activity which herald the onset of REM sleep, (i.e.,
decreased power in the delta band and increased power in the theta and sigma bands of the
electroencephalogram) (Benington et al., 1994). A change in brain state to REM sleep seems
to occur when sub-threshold REM sleep drive processes build to a level that is sufficient to
overcome opposing processes acting to maintain the current state. If the threshold level of
pontoreticular activity needed to induce REM sleep were to increase, it would stand to reason
that the magnitude of NRT electrographic changes would also increase so long as the
magnitude of those changes are an accurate index of the level of pontoreticular activation. I
showed, in accordance with previous work (Benington et al., 1994), that there is a sigmoidal
relationship between the magnitude of electrographic events associated with REM sleep
induction (quantified as NIV) and the incidence of REM sleep. This relationship is important,
because it means that the incidence of REM sleep increases in association with increasing
NIV. Since increased REM sleep drive most certainly accompanies increased incidence of
REM sleep, NIV can therefore serve as an indicator of REM sleep drive. I submit that NRT
76
associated electrographic changes can be treated as the phenotypic manifestation of sub-
threshold REM sleep drive. I used NIV magnitude to gauge the level of REM sleep drive
responsible for state transitions into REM sleep and to demarcate between REM sleep bouts
issuing from high and low REM sleep drive.
The increased proportion of REM sleep bouts issuing from periods of low REM sleep drive
during 8-OH-DPAT perfusion relative to baseline is consistent with the notion that PPTn
REM sleep-active neurons suppress transitioning into REM sleep by elevating the drive
threshold for REM sleep induction. In other words, REM sleep-active PPTn neurons, act to
prevent premature transitioning into REM sleep and so constrain the occurrence of this state
to narrow temporal windows of high REM sleep drive, by maintaining a high drive threshold
for REM sleep induction.
4.3 Connections of PPTn REM Sleep-Active Neurons Possibly Mediating REM Sleep Suppression
So far, I have discussed the mechanism of REM sleep suppression by REM sleep-active PPTn
neurons, couched in terms of abstract concepts like “drive intensity” and “induction
threshold”. Ultimately the suppression of REM sleep, by REM sleep-active PPTn neurons,
involves the interaction between this cell group and others within the REM sleep control
network. The identification of downstream effector sites of REM sleep-active PPTn neurons
was beyond the scope of this study. Nevertheless, I posit a testable model of REM sleep
generation in which the REM sleep-active PPTn cell group functions as a negative modulator
of REM sleep.
Historically, models of REM sleep generation have always considered REM sleep-active
PPTn neurons to promote REM sleep to one degree or another. The reciprocal interaction
77
hypothesis, originally formulated by Hobson and McCarley (Hobson et al., 1975; McCarley
and Hobson, 1975a), postulates that prior to REM sleep onset, cholinergic REM sleep-active
PPTn neurons become increasingly active, as inhibitory monoaminergic influences on these
cells are withdrawn. Increased PPTn REM sleep-active neuron activity activates
cholinoceptive SLDn neurons in the pontine reticular formation. Activation of pathways
emanating from the SLDn increase thalamocortical / cortical neuronal firing and tonicly
inhibit spinal and medullary motor pools, producing the EEG activation and motor atonia,
characteristic of REM sleep (Pace-Schott and Hobson, 2002). The emphasis on cholinergic
involvement in REM sleep generation by early models is not surprising, given that they were
founded on studies in cats in which induction of REM sleep by cholinergic PRF stimulation is
very robust. The results of these cat studies could not be reliably reproduced in rodents, and
so as rodents have become the preferred model for the study of sleep neurobiology, the role of
acetylcholine, and by proxy the PPTn, have been marginalized in recent iterations of REM
sleep generation models (Lu et al., 2006; Luppi et al., 2006; Fuller et al., 2007). The implicit
assumption has always been that REM sleep-active PPTn neurons are predominately
cholinergic, and so the functional role attributed to this cell group in REM sleep generation,
has always been indirectly defined by the result of cholinergic agonism in the PRF. However,
the functional role that I have defined for the REM sleep-active PPTn cell group as a negative
regulator of REM sleep is highly inconsistent with the notion of cholinergic innervation and
activation of the PRF by REM sleep-active PPTn neurons. The lack of consistency between
this data and the prevailing hypothesis regarding PPTn involvement in REM sleep generation
does not reflect a lack of consistency between this data and that of previous studies, because
the notion that REM sleep-active PPTn neurons are cholinergic is unjustified. A
preponderance of evidence shows that cholinergic REM sleep-active PPTn neurons occupy a
small proportion of the total REM sleep-active neuronal population of the PPTn (Maloney et
al., 1999; Verret et al., 2005). It stands to reason, that due to the scarcity of cholinergic REM
sleep-active PPTn neurons, that the increase in REM sleep that I observed during 8-OH-
DPAT administration, was the product of inhibition of predominately non-cholinergic
neurons. The proportion of REM sleep-active PPTn neurons that are GABAergic is reported
to be between 44 and 82% (Maloney et al., 2002; Sapin et al., 2009), vastly outnumbering the
cholinergic contingent of PPTn REM sleep-active neurons. The increase in REM sleep due to
the selective suppression of PPTn REM sleep-active neurons, is consistent with the expected
78
effect of reducing GABA release at downstream sites implicated in the generation of REM
sleep. GABA receptor antagonism at the SLD increases the discharge of REM sleep-active
SLD neurons (Boissard et al., 2002). Blocking the action of endogenous GABA at the SLD
using GABAA receptor antagonists reliably induces REM sleep, while GABA receptors
agonists suppress the occurrence of REM sleep (Boissard et al., 2002; Pollock and
Mistlberger, 2003; Sanford et al., 2003; Sapin et al., 2009). GABAergic mechanisms at the
SLDn are certainly responsible for restraining the occurrence of REM sleep, and recent
models of REM sleep generation propose that an inactivation of such GABAergic inputs into
SLD REM sleep-active neurons is the principal event responsible for REM sleep onset (Lu et
al., 2006; Luppi et al., 2006; Fuller et al., 2007). If disinhibition of the SLDn is sufficient to
induce REM sleep, then an antagonistic increase in GABAergic drive to the SLDn during
non-REM /REM sleep transitionary periods, would certainly act to suppress initiation of REM
sleep, and limit the overall frequency of its occurrence. Given that the PPTn is a known
source of GABAergic innervation of the SLDn (Boissard et al., 2003; Lu et al., 2006), REM
sleep-active PPTn neurons could therefore suppress the occurrence of REM sleep by
suppressing the excitability of SLDn REM sleep-active neurons via increased release of
GABA at non-REM / REM sleep transitionary periods.
It is known that that the PRF and the neuronal pools contained therein are necessary and
sufficient for the production of REM sleep (Steriade et al., 2005a). These neuronal pools and
the connections between them comprise the REM sleep control network. Therefore, in
accordance with the Sherringtonian view of a neural pool or node (Sherrington, 1961), the
generation of REM sleep comes as a result of a net facilitation of the nodes within the REM
sleep control network. Electrophysiological evidence supports this notion. The transition from
non-REM to REM sleep is marked by stereotypical changes in the membrane potential of
pontoreticular neurons (Ito and McCarley, 1984; Ito et al., 2002). Consistent with non-REM
sleep being a quiescent state of the brain, the membrane potential of PRF neurons are
hyperpolarized relative to REM sleep and little excitatory post synaptic activity occurs (Ito
and McCarley, 1984). Preceding bouts of REM sleep, PRF neuron membrane potential
becomes progressively more depolarized and excitatory postsynaptic events become more
frequent (Ito and McCarley, 1984). This progressive increase in PRF excitability can be
79
regarded as the neural correlate of increasing REM sleep drive. With the onset of REM sleep,
PRF neuron membrane potential plateaus at a level 7-10mv more depolarized than either quiet
wakefulness or non-REM sleep. PRF membrane potential promptly repolarizes with REM
sleep termination and wake onset (Ito and McCarley, 1984). These observations strongly
indicate that increased pontoreticular excitability is inextricably linked with REM sleep
generation.
Steriade and Mccarley (2005a), have suggested that the progressive depolarization of PRF
neurons can be attributed to a process of “recruitment”, whereby through reticuloreticular
excitatory connections increasing numbers of neurons become increasingly active over time.
Given that more than half of all afferent inputs into pontobulbar reticular neurons come from
other excitatory pontobulbar reticular neurons (Shammah-Lagnado et al., 1987), it would
stand to reason that increased activity of even a small number of neurons in the pontoreticular
field would lead to the recruitment of other pontoreticular neurons via a self-augmenting
positive feedback process (Steriade et al., 2005a). This positive feedback mechanism would
act to sustain the gradual depolarization of neuronal pools that comprise the REM sleep
control network (Steriade et al., 2005a).
Now if this process is the neural correlate of REM sleep drive, then what is the neural
correlate of the REM sleep drive induction threshold? Addressing this question requires that
we consider the organizational principles of neural networks. All biological networks,
including those in the brain are thought to exhibit so called, “small world” organization
(Bullmore and Sporns, 2009). Having small world organization means that complex
networks, such as the brain, are organized into functional modules, each module being a set of
densely interconnected nodes. The vast majority of nodes are connected to a small number of
nearby nodes within the same module. Only a select few nodes exhibit a high degree of
connectivity with the nodes of their own module and more importantly with nodes of other
modules. These highly connected nodes, referred to as connecting hubs, are critically
important for the physical transmission of information between modules. This kind of small
world organization is not only theoretical, as it has been empirically demonstrated to be
80
applicable to the vertebrate brainstem (Humphries et al., 2006). So, consider the brain as a
complex network, having small-world organization, in which the REM sleep control network
is a single functional module. REM sleep is a state of this entire brain network, but the origins
of this state are restricted to a single module, i.e., the REM sleep control network. Therefore,
the principal requirement for REM sleep generation must be the transmission of activity
within the REM sleep control network to other functional brain modules thereby allowing the
phenomenology of REM sleep to become fully elaborated. According to the principles of
small-world organization outlined above, the flow of activation from the REM sleep control
module to the rest of the brain will critically depend upon the activation of a select number of
connecting hubs, having high connectivity with nodes in distant modules. Therefore the rate
limiting step in initiating REM sleep is necessarily the activation of these connecting hubs. A
primary candidate for such a connecting hub, in the case of the REM sleep control module, is
the SLD. As previously mentioned, ascending and descending pathways emanating from the
SLD, may be largely responsible for mediating the motor atonia and electrographic changes
indicative of REM sleep (Sastre and Jouvet, 1979; Rye et al., 1988; Sanford et al., 1994;
Plazzi et al., 1996; Lu et al., 2006) (see section 1.4.1 for details), and so the SLDn has the
appropriate connectivity to act as a connecting hub for the REM sleep control network. Given
that REM sleep drive may be thought of as the progressive recruitment of pontoreticular
neurons making up the REM sleep control module, then the threshold for REM sleep
induction is inevitably the level of pontoreticular activity necessary to recruit or activate
connecting hubs such as the SLD. Since it is these hubs that ultimately permit activity in the
REM sleep control module to spread and affect the state of the brain-at-large. This
explanation of REM sleep initiation using the organizational principles of small-world
networks, may be called the small-world network hypothesis of REM sleep initiation (figure
14A). This hypothesis is meant to dovetail with current models of REM sleep generation such
as the reciprocal interaction(Hobson et al., 1975; McCarley and Hobson, 1975a; Pace-Schott
and Hobson, 2002) and flip-flop hypotheses (Lu et al., 2006; Luppi et al., 2006). These
models best describe the physical arrangement of the circuitry underlying REM sleep
generation (network physical topology), while the small-world network hypothesis is meant
to describe the flow of activity throughout the physical network, the so called “logical
topology” of the network, which is responsible for the generation of REM sleep. The
contribution of the small-world network hypothesis or any logical description of the REM
81
sleep control network is important because a unified theory of REM sleep generation will
require descriptions of both the physical and logical topologies of the network producing this
state.
The small-world network hypothesis can be used to explain many of the features of REM
sleep initiation as well as many of our own observations. According to the classical indicator
variables used to define the occurrence of REM sleep, only slight changes in phenomenology
occur prior to REM sleep, relative to the larger and more abrupt changes which mark
transitions from non-REM to REM sleep. In contrast, REM sleep, from the perspective of
those cells which underlie its generation, seems to be a much more gradual process.
According to small-world organizational principles, as pontoreticular neurons become more
and more active, changes in state phenomenology would be expected to be very slight, given
that most of these neurons have very few connections beyond their own module and therefore
have very little capacity to influence phenomenolocial variables. Major shifts in state
phenomenology would be expected to occur upon recruitment of connecting hubs such as the
SLDn, which have much greater capacities to influence state variables. Therefore, the fact that
phenomenologically discrete brain states emerge from neural activity that is highly dynamic
and constantly in flux, may be a consequence of the modular small-world architecture of
networks controlling state.
The mechanism of REM sleep suppression by REM sleep-active PPTn neurons which I have
proposed fits with the small-world network hypothesis of REM sleep initiation. According to
this hypothesis, the GABAergic suppression of SLDn neurons by the PPTn REM sleep-active
cell-group, which increases its discharge in the period preceding an episode of REM sleep,
would be expected to increase the threshold level of intra-modular pontoreticular activity
required to activate the SLDn and induce REM sleep onset via the transmission of activation
beyond the REM sleep control module. By so elevating the threshold of activation for a
connecting hub such as the SLD, a greater amount of time would necessarily be required
82
83
Figure 14. A schematic of the small-world network hypothesis of REM sleep initiation.
(A) Shows the hypothesized mechanism by which individual pontoreticular neuronal pools or
nodes of the REM sleep control module (depicted as numbered circles), are progressively
recruited during a non-REM to REM sleep transition via a series of interlocking positive
feedback loops. Due to mutual excitatory connection between pontoreticular neurons, activity
within the REM sleep control module gradually increases resulting in the recruitment of
nodes having ever higher recruitment thresholds. According to small-world network
organizational principles, the flow of activation is largely constrained to the REM sleep
control module because the vast majority of nodes have a small number of connections with
nearby nodes within the same module. Furthermore, the amount of intramodular activity
required to trigger the spread of activation from the REM-sleep control module to other brain
modules initiating a change in brain state is dictated by the recruitment threshold of
connecting hubs like the SLDn (sublaterodosal nucleus). Due to their increased connectivity,
it is these hubs that ultimately permit activity in the REM sleep control module to spread and
affect the state of the brain-at-large. (B) Shows the hypothesized mechanism by which REM
sleep-active PPTn neurons act to limit the frequency of REM sleep episodes. The
hypothesized GABAergic suppression of SLDn neurons by the PPTn REM sleep-active cell-
group would be expected to increase the threshold level of intra-modular pontoreticular
activity required to activate the SLDn and induce REM sleep onset. Consistent with the
results presented in this thesis is the notion that the 8-OH-DPAT mediated inhibition of REM
sleep-active PPTn neurons removed a source of GABAergic inhibition of the SLDn thereby
lowering the threshold level of REM sleep drive intensity required to initiate a bout of REM
sleep. Subsequently less time was required to accumulate the intra-modular activation
required to breach threshold, thereby increasing the frequency of REM sleep episodes.
84
to accumulate the intra-modular activation required to breach threshold, thereby restraining
the frequency of REM sleep episodes (figure 14B).
4.4 PPTn REM Sleep-Active Neurons Suppress Upper-Airway Muscle Activity During REM Sleep
Our findings indicate that REM sleep-active PPTn neurons suppress initiation of REM sleep,
but despite their continued activity, these neurons do not affect the time-course of individual
episodes once initiated. The continued activity of these neurons is responsible, however, for
regulating respiratory network activity within REM sleep episodes. One of the hallmarks of
the respiratory phenotype of REM sleep is the atonia of the upper-airway musculature,
including the genioglossus muscle of the tongue. The hyperpolarization of hypoglossal
motoneurons producing genioglossus muscle atonia during REM sleep, serves to prevent the
activation of this muscle amidst the bombardment of the hypoglossal motorpool by volleys of
excitatory activity (Orem and Netick, 1986; Orem et al., 2005). Occasionally these volleys of
excitation are able to penetrate the prevailing atonia and produce pronounced genioglossus
muscle twitching (shown in figure 9). The frequency and magnitude of these REM sleep-
related muscle twitches at any given time, undoubtedly reflect the balance of opposing
excitatory and inhibitory inputs into the hypoglossal motor nucleus. I have shown that while
selectively inhibiting REM sleep-active PPTn neurons that the frequency and magnitude of
REM sleep-related genioglossus muscle twitching increased. In other words, the atonia of the
genioglossus muscle during REM sleep was punctuated by excitatory events more often and
to a greater degree in the absence of PPTn REM sleep-active neuron activity. This finding
indicates that REM sleep-active PPTn neurons normally act to shift the balance of activity at
the hypoglossal motor nucleus towards inhibition, thereby buffering against excessive
genioglossus activation during REM sleep. The predominant increase in REM sleep-related
twitching in the respiratory genioglossus muscle versus that of non-respiratory nuchal
musculature, indicates that the PPTn REM sleep-active cell-group does not act to suppress
skeletal muscle activity ubiquitously during REM sleep, but rather acts to preferentially
prevent respiratory muscle activation during REM sleep.
85
4.5 PPTn REM Sleep-Active Neurons Depress Breathing Across Sleep-Wake States
Although discharging maximally prior to and during REM sleep, PPTn REM sleep-active
neurons remain active during wakefulness and non-REM sleep (Steriade et al., 1990; Thakkar
et al., 1998). Therefore, it is not surprising that the influence of these cells on breathing is not
constrained to the state of REM sleep. In contrast to the REM sleep-specific inhibition of the
behavior-related genioglossus muscle activity by PPTn REM sleep-active neurons, I showed
that inhibition of this cell group augmented the respiratory component of genioglossus
activity independently of the prevailing sleep-wake state. Therefore, REM sleep-active PPTn
neurons normally act to suppress respiratory inputs into the hypoglossal motor nucleus, and
this suppression manifests in all states of sleep and wakefulness. PPTn REM sleep-active
neurons may modulate genioglossus activity via known PPTn projections either to the
hypoglossal motor nucleus or to hypoglossal pre-motor sites (Woolf and Butcher, 1989; Fay
and Norgren, 1997; Rukhadze and Kubin, 2007).
In addition to depressing the respiratory activation of the genioglossus muscle, REM sleep-
active PPTn neurons produce a depression of respiratory rate across states of sleep and
wakefulness, as evidenced by the state-independent elevation of respiratory rate during the
selective suppression of REM sleep active PPTn neuron activity. The PPTn has the
appropriate connectivity to mediate this effect, since PPTn neurons have been shown to
project to the rostoventolateral medulla (Yasui et al., 1990), which houses the circuitry
responsible for the generation of respiratory rate (Feldman and Del Negro, 2006). The
capacity of the PPTn to depress respiratory rate has been previously demonstrated using
electrical PPTn stimulation in anesthetized cats causing a reduction in respiratory rate (Lydic
and Baghdoyan, 1993). This study shows for the first time that beyond a mere capacity to
depress breathing, that the input of the PPTn into the respiratory network is actually required
for producing a normal respiratory rate.
86
By acting to suppress respiratory rate and upper-airway muscle activity, particularly during
REM sleep, REM sleep-active PPTn neurons increase the vulnerability of predisposed
individuals to experiencing disordered breathing during sleep. By so identifying the neural
underpinnings of sleep’s impact on breathing, we are better able to understand the
pathogenesis of sleep-related breathing disorders and to develop novel therapeutic strategies.
4.6 Future Directions
This thesis emphasizes the fact that as network phenomena, both the generation of REM sleep
and its respiratory phenotype are the product of complex interactions between multiple,
functionally distinct cell-groups. The evidence presented here served to define the functional
role of the PPTn REM sleep-active cell population. Future studies are needed to determine the
necessary interactions between REM sleep-active PPTn neurons, and other cell-groups which
are responsible for the suppressive influence of REM sleep-active PPTn neurons on REM
sleep initiation, genioglossus activity and respiratory rate. To determine these interactions, I
propose a dual microdialysis strategy, involving local microdialysis delivery of a stimulatory
agent at one site to induce an effect that can be blocked by delivery of an antagonist at the
hypothesized effector site. A microdialysis probe would be implanted at the PPTn, and the
activity of REM sleep-active PPTn neurons would be raised using a 5HT1A receptor
antagonist. The resulting increase in neurotransmitter release at downstream effector sites
would presumably act to suppress REM sleep initiation, genioglossus activity and respiratory
rate. A second microdialysis probe would be located at the hypothesized site of REM sleep-
active PPTn neuron influence, and an antagonist of the neurotransmitter hypothesized to be
mediating the effect would be delivered. For example, I previously hypothesized that REM
sleep-active PPTn neurons may act to suppress REM sleep via GABA-mediated inhibition of
SLDn neurons. If this hypothesis is correct, I would expect that the decrease in REM sleep,
stemming from 5HT1A receptor antagonist-induced disinhibition of REM sleep-active PPTn
neuron activity, would be reversed by the simultaneous delivery of a GABA receptor
antagonist at the SLDn. Therefore, this strategy will permit the determination of the
connections and neurotransmitters underlying the functional impact of PPTn REM sleep-
87
active neurons on the control of REM sleep and its respiratory phenotype. Such
determinations are required to fully map the network underlying the generation of REM sleep
and its phenomenological components.
88
References
Amici R, Domeniconi R, Jones CA, Morales-Cobas G, Perez E, Tavernese L, Torterolo P,
Zamboni G, Parmeggiani PL (2000) Changes in REM sleep occurrence due to
rhythmical auditory stimulation in the rat. Brain Res 868:241-250.
Aserinsky E, Kleitman N (1953) Regularly occurring periods of eye motility, and
concomitant phenomena, during sleep. Science 118:273-274.
Aserinsky E, Kleitman N (1955) Two types of ocular motility occurring in sleep. J Appl
Physiol 8:1-10.
Baghdoyan HA, Rodrigo-Angulo ML, McCarley RW, Hobson JA (1987) A neuroanatomical
gradient in the pontine tegmentum for the cholinoceptive induction of desynchronized
sleep signs. Brain Res 414:245-261.
Baghdoyan HA, Monaco AP, Rodrigo-Angulo ML, Assens F, McCarley RW, Hobson JA
(1984) Microinjection of neostigmine into the pontine reticular formation of cats
enhances desynchronized sleep signs. J Pharmacol Exp Ther 231:173-180.
Benington JH, Kodali SK, Heller HC (1994) Scoring transitions to REM sleep in rats based
on the EEG phenomena of pre-REM sleep: an improved analysis of sleep structure.
Sleep 17:28-36.
Bevan MD, Bolam JP (1995) Cholinergic, GABAergic, and glutamate-enriched inputs from
the mesopontine tegmentum to the subthalamic nucleus in the rat. J Neurosci 15:7105-
7120.
Boissard R, Fort P, Gervasoni D, Barbagli B, Luppi PH (2003) Localization of the
GABAergic and non-GABAergic neurons projecting to the sublaterodorsal nucleus
and potentially gating paradoxical sleep onset. Eur J Neurosci 18:1627-1639.
Boissard R, Gervasoni D, Schmidt MH, Barbagli B, Fort P, Luppi PH (2002) The rat ponto-
medullary network responsible for paradoxical sleep onset and maintenance: a
combined microinjection and functional neuroanatomical study. Eur J Neurosci
16:1959-1973.
Bourgin P, Escourrou P, Gaultier C, Adrien J (1995) Induction of rapid eye movement sleep
by carbachol infusion into the pontine reticular formation in the rat. Neuroreport
6:532-536.
Bullmore E, Sporns O (2009) Complex brain networks: graph theoretical analysis of
structural and functional systems. Nat Rev Neurosci 10:186-198.
Carli G, Zanchetti A (1965) A study of pontine lesions suppressing deep sleep in the cat. Arch
Ital Biol 103:751-788.
Chan E, Steenland HW, Liu H, Horner RL (2006) Endogenous excitatory drive modulating
respiratory muscle activity across sleep-wake states. Am J Respir Crit Care Med
174:1264-1273.
89
Colten HR, Altevogt BM (2006) Sleep disorders and sleep deprivation: an unmet public
health problem. In: (Institute of Medicine. Committee on Sleep Medicine and
Research BoHSP, ed), p 386. Washington, DC: The National Academies Press
Datta S, Siwek DF (1997) Excitation of the brain stem pedunculopontine tegmentum
cholinergic cells induces wakefulness and REM sleep. J Neurophysiol 77:2975-2988.
Datta S, Patterson EH, Siwek DF (1999) Brainstem afferents of the cholinoceptive pontine
wave generation sites in the rat. Sleep Res Online 2:79-82.
Datta S, Spoley EE, Patterson EH (2001a) Microinjection of glutamate into the
pedunculopontine tegmentum induces REM sleep and wakefulness in the rat. Am J
Physiol Regul Integr Comp Physiol 280:R752-759.
Datta S, Patterson EH, Spoley EE (2001b) Excitation of the pedunculopontine tegmental
NMDA receptors induces wakefulness and cortical activation in the rat. J Neurosci
Res 66:109-116.
Dement W, Kleitman N (1957a) The relation of eye movements during sleep to dream
activity: an objective method for the study of dreaming. J Exp Psychol 53:339-346.
Dement W, Kleitman N (1957b) Cyclic variations in EEG during sleep and their relation to
eye movements, body motility, and dreaming. Electroencephalogr Clin Neurophysiol
9:673-690.
Deurveilher S, Hars B, Hennevin E (1997) Pontine microinjection of carbachol does not
reliably enhance paradoxical sleep in rats. Sleep 20:593-607.
Edley SM, Graybiel AM (1983) The afferent and efferent connections of the feline nucleus
tegmenti pedunculopontinus, pars compacta. J Comp Neurol 217:187-215.
Fay RA, Norgren R (1997) Identification of rat brainstem multisynaptic connections to the
oral motor nuclei using pseudorabies virus. III. Lingual muscle motor systems. Brain
Res Brain Res Rev 25:291-311.
Feldman JL, Del Negro CA (2006) Looking for inspiration: new perspectives on respiratory
rhythm. Nat Rev Neurosci 7:232-242.
Fuller PM, Saper CB, Lu J (2007) The pontine REM switch: past and present. J Physiol
584:735-741.
George R, Haslett WL, Jenden DJ (1964) A Cholinergic Mechanism in the Brainstem
Reticular Formation: Induction of Paradoxical Sleep. Int J Neuropharmacol 3:541-
552.
Hallanger AE, Wainer BH (1988) Ascending projections from the pedunculopontine
tegmental nucleus and the adjacent mesopontine tegmentum in the rat. J Comp Neurol
274:483-515.
Hamrahi H, Chan B, Horner RL (2001) On-line detection of sleep-wake states and application
to produce intermittent hypoxia only in sleep in rats. J Appl Physiol 90:2130-2140.
Hobson JA, McCarley RW, Wyzinski PW (1975) Sleep cycle oscillation: reciprocal discharge
by two brainstem neuronal groups. Science 189:55-58.
Hobson JA, Pace-Schott EF, Stickgold R (2000) Dreaming and the brain: toward a cognitive
neuroscience of conscious states. Behav Brain Sci 23:793-842; discussion 904-1121.
90
Hobson JA, McCarley RW, Pivik RT, Freedman R (1974) Selective firing by cat pontine
brain stem neurons in desynchronized sleep. J Neurophysiol 37:497-511.
Honda T, Semba K (1995) An ultrastructural study of cholinergic and non-cholinergic
neurons in the laterodorsal and pedunculopontine tegmental nuclei in the rat.
Neuroscience 68:837-853.
Horner RL (2009) Emerging principles and neural substrates underlying tonic sleep-state-
dependent influences on respiratory motor activity. Philos Trans R Soc Lond B Biol
Sci 364:2553-2564.
Horner RL, Brooks D, Kozar LF, Leung E, Hamrahi H, Render-Teixeira CL, Makino H,
Kimoff RJ, Phillipson EA (1998) Sleep architecture in a canine model of obstructive
sleep apnea. Sleep 21:847-858.
Humphries MD, Gurney K, Prescott TJ (2006) The brainstem reticular formation is a small-
world, not scale-free, network. Proc Biol Sci 273:503-511.
Ito K, McCarley RW (1984) Alterations in membrane potential and excitability of cat medial
pontine reticular formation neurons during changes in naturally occurring sleep-wake
states. Brain Res 292:169-175.
Ito K, Yanagihara M, Imon H, Dauphin L, McCarley RW (2002) Intracellular recordings of
pontine medial gigantocellular tegmental field neurons in the naturally sleeping cat:
behavioral state-related activity and soma size difference in order of recruitment.
Neuroscience 114:23-37.
Jones BE (1990) Immunohistochemical study of choline acetyltransferase-immunoreactive
processes and cells innervating the pontomedullary reticular formation in the rat. J
Comp Neurol 295:485-514.
Jouvet D, Valatx JL (1962) [Polygraphic study of sleep in lambs.]. C R Seances Soc Biol Fil
156:1411-1414.
Jouvet M (1962) [Research on the neural structures and responsible mechanisms in different
phases of physiological sleep.]. Arch Ital Biol 100:125-206.
Jouvet M, Vimont P, Delorme F (1965) [Elective suppression of paradoxal sleep in the cat by
monoamine oxidase inhibitors]. C R Seances Soc Biol Fil 159:1595-1599.
Kodama T, Takahashi Y, Honda Y (1990) Enhancement of acetylcholine release during
paradoxical sleep in the dorsal tegmental field of the cat brain stem. Neurosci Lett
114:277-282.
Kohlmeier KA, Burns J, Reiner PB, Semba K (2002) Substance P in the descending
cholinergic projection to REM sleep-induction regions of the rat pontine reticular
formation: anatomical and electrophysiological analyses. Eur J Neurosci 15:176-196.
Lu J, Greco MA, Shiromani P, Saper CB (2000) Effect of lesions of the ventrolateral preoptic
nucleus on NREM and REM sleep. J Neurosci 20:3830-3842.
Lu J, Sherman D, Devor M, Saper CB (2006) A putative flip-flop switch for control of REM
sleep. Nature 441:589-594.
91
Luppi PH, Gervasoni D, Verret L, Goutagny R, Peyron C, Salvert D, Leger L, Fort P (2006)
Paradoxical (REM) sleep genesis: the switch from an aminergic-cholinergic to a
GABAergic-glutamatergic hypothesis. J Physiol Paris 100:271-283.
Lydic R, Baghdoyan HA (1993) Pedunculopontine stimulation alters respiration and increases
ACh release in the pontine reticular formation. Am J Physiol 264:R544-554.
lydic R, Baghdoyan HA (2003) Neurochemical evidence for the cholinergic modulation of
sleep and breathing. In: In Sleep-related breathing disorders: Experimental models and
therapeutic potential (Carley DW, Radulovacki M, eds), pp 57-91. new york: dekker.
Lydic R, Baghdoyan HA (2005) Sleep, anesthesiology, and the neurobiology of arousal state
control. Anesthesiology 103:1268-1295.
Lydic R, Baghdoyan HA, Lorinc Z (1991) Microdialysis of cat pons reveals enhanced
acetylcholine release during state-dependent respiratory depression. Am J Physiol
261:R766-770.
Lydic R, Douglas CL, Baghdoyan HA (2002) Microinjection of neostigmine into the pontine
reticular formation of C57BL/6J mouse enhances rapid eye movement sleep and
depresses breathing. Sleep 25:835-841.
Maloney KJ, Mainville L, Jones BE (1999) Differential c-Fos expression in cholinergic,
monoaminergic, and GABAergic cell groups of the pontomesencephalic tegmentum
after paradoxical sleep deprivation and recovery. J Neurosci 19:3057-3072.
Maloney KJ, Mainville L, Jones BE (2000) c-Fos expression in GABAergic, serotonergic,
and other neurons of the pontomedullary reticular formation and raphe after
paradoxical sleep deprivation and recovery. J Neurosci 20:4669-4679.
Maloney KJ, Mainville L, Jones BE (2002) c-Fos expression in dopaminergic and
GABAergic neurons of the ventral mesencephalic tegmentum after paradoxical sleep
deprivation and recovery. Eur J Neurosci 15:774-778.
McCarley RW, Hobson JA (1975a) Neuronal excitability modulation over the sleep cycle: a
structural and mathematical model. Science 189:58-60.
McCarley RW, Hobson JA (1975b) Discharge patterns of cat pontine brain stem neurons
during desynchronized sleep. J Neurophysiol 38:751-766.
Megirian D, Hinrichsen CF, Sherrey JH (1985) Respiratory roles of genioglossus,
sternothyroid, and sternohyoid muscles during sleep. Exp Neurol 90:118-128.
Morrison JL, Sood S, Liu X, Liu H, Park E, Nolan P, Horner RL (2002) Glycine at
hypoglossal motor nucleus: genioglossus activity, CO(2) responses, and the additive
effects of GABA. J Appl Physiol 93:1786-1796.
Moruzzi G, Magoun HW (1949) Brain stem reticular formation and activation of the EEG.
Electroencephalogr Clin Neurophysiol 1:455-473.
Nauta WJH (1946) Hpothalamic regulation of sleep in rats. J Neurophysiol 9:285-316.
Orem J (1990) The nature of the wakefulness stimulus for breathing. Prog Clin Biol Res
345:23-30; discussion 31.
Orem J, Netick A (1986) Behavioral control of breathing in the cat. Brain Res 366:238-253.
92
Orem J, Netick A, Dement WC (1977) Breathing during sleep and wakefulness in the cat.
Respir Physiol 30:265-289.
Orem JM, Lovering AT, Vidruk EH (2005) Excitation of medullary respiratory neurons in
REM sleep. Sleep 28:801-807.
Pace-Schott EF, Hobson JA (2002) The neurobiology of sleep: genetics, cellular physiology
and subcortical networks. Nat Rev Neurosci 3:591-605.
Pal D, Mallick BN (2004) GABA in pedunculo pontine tegmentum regulates spontaneous
rapid eye movement sleep by acting on GABAA receptors in freely moving rats.
Neurosci Lett 365:200-204.
Pal D, Mallick BN (2009) GABA in pedunculopontine tegmentum increases rapid eye
movement sleep in freely moving rats: possible role of GABA-ergic inputs from
substantia nigra pars reticulata. Neuroscience 164:404-414.
Paxinos G, Watson C (1998) The rat brain in stereotaxic coordinates, 4th Edition. San Diego:
Academic Press.
Phillipson EA, Bowes G (1886) Control of breathing during sleep. In: Handbook of
physiology, section 3, the respiratory system, Vol. II, control of breathing, part 2, 2nd
Edition (S. CN, G WJ, eds), pp 649–689 p. Bethesda, MD: American Physiological
Society.
Plazzi G, Montagna P, Provini F, Bizzi A, Cohen M, Lugaresi E (1996) Pontine lesions in
idiopathic narcolepsy. Neurology 46:1250-1254.
Pollock MS, Mistlberger RE (2003) Rapid eye movement sleep induction by microinjection
of the GABA-A antagonist bicuculline into the dorsal subcoeruleus area of the rat.
Brain Res 962:68-77.
Pollock MS, Mistlberger RE (2005) Microinjection of neostigmine into the pontine reticular
formation of the mouse: further evaluation of a proposed REM sleep enhancement
technique. Brain Res 1031:253-267.
Remmers JE, deGroot WJ, Sauerland EK, Anch AM (1978) Pathogenesis of upper airway
occlusion during sleep. J Appl Physiol 44:931-938.
Rukhadze I, Kubin L (2007) Mesopontine cholinergic projections to the hypoglossal motor
nucleus. Neurosci Lett 413:121-125.
Rye DB, Saper CB, Lee HJ, Wainer BH (1987) Pedunculopontine tegmental nucleus of the
rat: cytoarchitecture, cytochemistry, and some extrapyramidal connections of the
mesopontine tegmentum. J Comp Neurol 259:483-528.
Rye DB, Lee HJ, Saper CB, Wainer BH (1988) Medullary and spinal efferents of the
pedunculopontine tegmental nucleus and adjacent mesopontine tegmentum in the rat. J
Comp Neurol 269:315-341.
Sakai K, Sastre JP, Kanamori N, Jouvet M (1981) State-specific neurons in the ponto-
medullary reticular formation with special reference to the postural muscle atonia
during paradoxical sleep in the cat. In: Brain Mechanisms of Perceptual Awareness
and Purposeful Behaviour., 2nd Edition (Pompeiano O, Aimone Marsan C, eds), pp
405-429. New York: Raven Press.
93
Sallanon M, Denoyer M, Kitahama K, Aubert C, Gay N, Jouvet M (1989) Long-lasting
insomnia induced by preoptic neuron lesions and its transient reversal by muscimol
injection into the posterior hypothalamus in the cat. Neuroscience 32:669-683.
Sanford LD, Tang X, Xiao J, Ross RJ, Morrison AR (2003) GABAergic regulation of REM
sleep in reticularis pontis oralis and caudalis in rats. J Neurophysiol 90:938-945.
Sanford LD, Morrison AR, Mann GL, Harris JS, Yoo L, Ross RJ (1994) Sleep patterning and
behaviour in cats with pontine lesions creating REM without atonia. J Sleep Res
3:233-240.
Saper CB, Chou TC, Scammell TE (2001) The sleep switch: hypothalamic control of sleep
and wakefulness. Trends Neurosci 24:726-731.
Saper CB, Cano G, Scammell TE (2005) Homeostatic, circadian, and emotional regulation of
sleep. J Comp Neurol 493:92-98.
Sapin E, Lapray D, Berod A, Goutagny R, Leger L, Ravassard P, Clement O, Hanriot L, Fort
P, Luppi PH (2009) Localization of the brainstem GABAergic neurons controlling
paradoxical (REM) sleep. PLoS One 4:e4272.
Sastre JP, Jouvet M (1979) [Oneiric behavior in cats]. Physiol Behav 22:979-989.
Sastre JP, Sakai K, Jouvet M (1981) Are the gigantocellular tegmental field neurons
responsible for paradoxical sleep? Brain Res 229:147-161.
Semba K, Fibiger HC (1992) Afferent connections of the laterodorsal and the
pedunculopontine tegmental nuclei in the rat: a retro- and antero-grade transport and
immunohistochemical study. J Comp Neurol 323:387-410.
Semba K, Reiner PB, Fibiger HC (1990) Single cholinergic mesopontine tegmental neurons
project to both the pontine reticular formation and the thalamus in the rat.
Neuroscience 38:643-654.
Shammah-Lagnado SJ, Negrao N, Silva BA, Ricardo JA (1987) Afferent connections of the
nuclei reticularis pontis oralis and caudalis: a horseradish peroxidase study in the rat.
Neuroscience 20:961-989.
Sherin JE, Shiromani PJ, McCarley RW, Saper CB (1996) Activation of ventrolateral preoptic
neurons during sleep. Science 271:216-219.
Sherin JE, Elmquist JK, Torrealba F, Saper CB (1998) Innervation of histaminergic
tuberomammillary neurons by GABAergic and galaninergic neurons in the
ventrolateral preoptic nucleus of the rat. J Neurosci 18:4705-4721.
Sherrington CS (1961) The integrative action of the nervous system, [2d Edition. New
Haven,: Yale University Press.
Shiromani PJ, Armstrong DM, Berkowitz A, Jeste DV, Gillin JC (1988) Distribution of
choline acetyltransferase immunoreactive somata in the feline brainstem: implications
for REM sleep generation. Sleep 11:1-16.
Shouse MN, Siegel JM (1992) Pontine regulation of REM sleep components in cats: integrity
of the pedunculopontine tegmentum (PPT) is important for phasic events but
unnecessary for atonia during REM sleep. Brain Res 571:50-63.
94
Sood S, Morrison JL, Liu H, Horner RL (2005) Role of endogenous serotonin in modulating
genioglossus muscle activity in awake and sleeping rats. Am J Respir Crit Care Med
172:1338-1347.
Steenland HW, Liu H, Horner RL (2008) Endogenous glutamatergic control of rhythmically
active mammalian respiratory motoneurons in vivo. J Neurosci 28:6826-6835.
Steininger TL, Gong H, McGinty D, Szymusiak R (2001) Subregional organization of
preoptic area/anterior hypothalamic projections to arousal-related monoaminergic cell
groups. J Comp Neurol 429:638-653.
Steriade M, McCarley RW (2005a) Neuronal Activities in Brainstem and Basal Forebrain
Structures Controlling Waking and Sleep States. In: Brainstem control of wakefulness
and sleep, 2nd Edition, pp 381-416 p. New York: Plenum Press.
Steriade M, McCarley RW (2005b) Brainstem control of wakefulness and sleep, 2nd Edition.
New York: Plenum Press.
Steriade M, McCarley RW, ebrary Inc. (2005a) Neuronal Control of REM Sleep In:
Brainstem control of wakefulness and sleep, 2nd Edition, pp 461-511 p. New York:
Plenum Press.
Steriade M, McCarley RW, ebrary Inc. (2005b) Brainstem control of wakefulness and sleep.
In, 2nd Edition, p 728 p. New York: Plenum Press.
Steriade M, Datta S, Pare D, Oakson G, Curro Dossi RC (1990) Neuronal activities in brain-
stem cholinergic nuclei related to tonic activation processes in thalamocortical
systems. J Neurosci 10:2541-2559.
Sterman MB, Clemente CD (1962a) Forebrain inhibitory mechanisms: sleep patterns induced
by basal forebrain stimulation in the behaving cat. Exp Neurol 6:103-117.
Sterman MB, Clemente CD (1962b) Forebrain inhibitory mechanisms: cortical
synchronization induced by basal forebrain stimulation. Exp Neurol 6:91-102.
Thakkar M, Portas C, McCarley RW (1996) Chronic low-amplitude electrical stimulation of
the laterodorsal tegmental nucleus of freely moving cats increases REM sleep. Brain
Res 723:223-227.
Thakkar MM, Strecker RE, McCarley RW (1998) Behavioral state control through
differential serotonergic inhibition in the mesopontine cholinergic nuclei: a
simultaneous unit recording and microdialysis study. J Neurosci 18:5490-5497.
Torterolo P, Morales FR, Chase MH (2002) GABAergic mechanisms in the pedunculopontine
tegmental nucleus of the cat promote active (REM) sleep. Brain Res 944:1-9.
Torterolo P, Sampogna S, Morales FR, Chase MH (2006) MCH-containing neurons in the
hypothalamus of the cat: searching for a role in the control of sleep and wakefulness.
Brain Res 1119:101-114.
Torterolo P, Yamuy J, Sampogna S, Morales FR, Chase MH (2001) GABAergic neurons of
the laterodorsal and pedunculopontine tegmental nuclei of the cat express c-fos during
carbachol-induced active sleep. Brain Res 892:309-319.
Tretiakoff c, Bremer F (1920) Encephalite letargique avec syndrome Parkinsonien et
catatonie. Revue Neurologique 7:e4272.
95
Vanni-Mercier G, Sakai K, Lin JS, Jouvet M (1989) Mapping of cholinoceptive brainstem
structures responsible for the generation of paradoxical sleep in the cat. Arch Ital Biol
127:133-164.
Verret L, Leger L, Fort P, Luppi PH (2005) Cholinergic and noncholinergic brainstem
neurons expressing Fos after paradoxical (REM) sleep deprivation and recovery. Eur J
Neurosci 21:2488-2504.
Vincent SR, Satoh K, Armstrong DM, Fibiger HC (1983) NADPH-diaphorase: a selective
histochemical marker for the cholinergic neurons of the pontine reticular formation.
Neurosci Lett 43:31-36.
Wang HL, Morales M (2009) Pedunculopontine and laterodorsal tegmental nuclei contain
distinct populations of cholinergic, glutamatergic and GABAergic neurons in the rat.
Eur J Neurosci 29:340-358.
Webster HH, Jones BE (1988) Neurotoxic lesions of the dorsolateral pontomesencephalic
tegmentum-cholinergic cell area in the cat. II. Effects upon sleep-waking states. Brain
Res 458:285-302.
Webster HH, Friedman L, Jones BE (1986) Modification of paradoxical sleep following
transections of the reticular formation at the pontomedullary junction. Sleep 9:1-23.
Woolf NJ, Butcher LL (1989) Cholinergic systems in the rat brain: IV. Descending
projections of the pontomesencephalic tegmentum. Brain Res Bull 23:519-540.
Yamamoto K, Mamelak AN, Quattrochi JJ, Hobson JA (1990) A cholinoceptive
desynchronized sleep induction zone in the anterodorsal pontine tegmentum: locus of
the sensitive region. Neuroscience 39:279-293.
Yasui Y, Cechetto DF, Saper CB (1990) Evidence for a cholinergic projection from the
pedunculopontine tegmental nucleus to the rostral ventrolateral medulla in the rat.
Brain Res 517:19-24.
Younes M, Park E, Horner RL (2007) Pentobarbital sedation increases genioglossus
respiratory activity in sleeping rats. Sleep 30:478-488.
