CIRCADIAN RHYTHMS IN LUNG VENTILATION IN WAKEFULNESS AND SLEEP Kiong Sen Liao A thesis subrnitted in confonnity with the requirements for the degree of Master of Science (M.Sc.), Graduate Department of Physiology, University of Toronto O Copyright by Kiong Sen Liao (2001)
124
Embed
CIRCADIAN RHYTHMS IN LUNG VENTILATION IN WAKEFULNESS AND · PDF fileAbstract Kiong Sen Liao. Department oj* Physiologr.. Universip of Toronto. Sleep exerts important modulating intluences
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
CIRCADIAN RHYTHMS IN LUNG VENTILATION
IN WAKEFULNESS AND SLEEP
Kiong Sen Liao
A thesis subrnitted in confonnity with the requirements for the degree of Master of Science (M.Sc.),
Graduate Department of Physiology, University of Toronto
O Copyright by Kiong Sen Liao (2001)
National Libmry (jrl of C w d a Bibliothèque nationale du Canada
Acquisitions and Acquisitions et Bibliographie Services selvices bibliographiques
395 Wellington Street 395, rue Wellington OttawaON KiAON4 Ottawa ON K I A ON4 Canada canada
The author has granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, distribute or seU copies of this thesis in microform, paper or electronic formats.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts fiom it may be printed or otherwise reproduced without the author's permission.
L'auteur a accordé une Licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/film, de reproduction sur papier ou sur format électronique.
L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation,
To my grandmother, parents and brothers
Abstract
Kiong Sen Liao. Department oj* Physiologr.. Universip of Toronto.
Sleep exerts important modulating intluences on breathing such that a decline in
lung ventilation in sleep predisposes individuals to respiratory impairment. Recent data
from awake humans also suggest that the circadian system intluences breathing. This
circadian intluence may be potentially important because a decline in ventilation due to
circadian effects may exacerbate the decline due to sleep. although this has not been
tested. Therefore. by measuring sleep-wake States (using electroencephalography and
electromyography). lung ventilation (using barometric plethysmo~gaphy) and metabolic
CO? production across 24 hour periods in six freely behaving rats. the present study tested
and confirmed the hypotheses that: (1) there are circadian rhythms in lung ventilation in
wakefulness. non rapid eye movement (NREM) sleep and REM sleep (al1 p<0.001). and (2)
the magnitude of the change in ventilation From wakefulness to NREM sleep is the same
across time of day (p=0.58).
III
Acknowledeements
"Give thanks to the Lord for He is good." (Psalm 1 36: 1 )
1 am grateful to God for His many blessings that He has given me throughout my
life and over these two years in this graduate expenence. He has been faithful and
provided for my needs at every step of the way. He has blessed me with a whole host of
people who have guided me when 1 was lost and perplexed. rncouraged me when I was
discouraged. picked me up when 1 fell. endured with me during tough times. strengthened
me when 1 was weak. challenged me when 1 became complacrnt and corrected and been
patient with me when I erred. The list could go on. but can be summarised in one
sentence ... the good Lord has provided for me in every conceivable way. He has made
the biggest difference in my life as 1 have my walked with Him.
1 would like to thank rny father. Pao Chun Liao and my mother. Chiao Yun Kuo
without whose love. sacrifices. hard work. encouragement and suppon. I would not have
had the many countless opportunities such as being able to attend university or do my
graduate studies. They have modelled canng, giving and dedication. 1 am grateful to my
brothers. Choy Sen and Liung Sen for al1 their love and support. 1 am also mindful of the
immense contribution that my late grandmother made in innumerable ways during my
upbringing. l am also thankful to Yuanni for al1 the help she has given me dunng this
projec t.
Over the past two years. I have worked under the joint supervision of Dr. Richard
Homer and Dr. Richard Stephenson. Both of thesr "chiefs" are very dedicated and
talented scientists. 1 am very grateful for al1 their guidance. advice. kindness. patience
and suppon in providing a very ennching Master's expenence. Through their course and
in the lab. both of then taught me to think critically and to ask questions in addition to a
lot of usefùl knowledge in research. sleep. breathing and chronobiology. 1 will remember
Dr. Horner for his generosity. efficiency. drive and attention to detail. 1 would like to
dedicate Figure 11 to him. 1 will remember Dr. Stephenson for his humility. gentlenrss.
wittiness and ability to put things in perspective.
I am grateful to Dr. James Duffin and the members of the Respiratory Research
group. Their constructive criticism and feedback has been invaluable. 1 would like to
thank Dr. Dufin for his graduate courses in Respiratory Physioloa and Electronics. both
of which were tremendous leaming opportunities. 1 would also like to thank him along
with Dr. Martin Ralph for their time and effort in serving on my M.Sc. supervisory
committee. 1 am gratehl to Dr. Scott Thomas and Dr. Dina Brooks for serving on my
M .Sc. examining committee.
1 want to thank my fhends and fellow graduate students in the lab. Hedieh
Hamrahi and Sandeep Sood. 1 have valued their encouragement. support. Company and
humour deeply. I would like to thank Beverly Chan for helping me with the data analysis
and surpries involved in my project. 1 also thank Hattie Liu and Xia Liu for their
encouragement.
1 b i l l not forget the many ways that my friends and teachers at Hebron School
and Chan and the staff at Shinkows Chinese Restaurant have blessed me richly. 1 am
thankfùl for al1 the prayen and support from my many aunts and uncles and friends at
Fint Alliance Church and St. John Ambulance Scarborough 432. Thcir warmth.
generosity and friendship meant a lot to me during the hard times.
The present study would not have been possible without hnding frorn the
Medical Rcsearc h Council of Canada. the Nat ional Science and Engineering Council of
Canada and the Ontario Thoracic Society. 1 am grateful for the Ontario Graduate
Scholanhip in Science and Technology which helped me finance this degree.
Table of Contents
Contents
Chapter 1: Introduction Ci rcadian Rhyt hms Sleep-wake States Replation of Respiration Effect of Sleep-wake State on Breathing Circadian Modulation of B reathing Hypot heses Rational for Hypotheses
Chapter 3: Results Section 1: Testing Hypothesis 1 and 2 Section 2: Additional Circadian Rhythms Analyses Section 3: Additional Analyses on Effects of Sleep-wake State Section 4: Additional Day-night Cornparisons
Chapter 4: Discussion Technical considerations C ircadian related changes in lung ventilation Potential rnechanisms rnediating the circadian rhythms in lung ventilation The circadian rhythm in ventilation nomalized for COz production in
NREM sleep Sleep-wake related changes in breathing. rnetabolism and body temperature Future research opportunities Conclusions
References
Page
List of Abbreviations Used
SCN
EEG EMG EOG NREM REM Wake
8 1
n 0 a Pi
P z %Bz/%6i
Tb
Mb
BTPS
PB Pb~20 PCH~O
SwvP STPD
Circadian Rhythrns Suprac hiasmatic nucleus
Sleep Electroencep halogram Elec tromyogram Electrooculogram Non-rapid eye movement sleep Rapid eye movement sleep Wakefiilness
EEG freattencies Delta 1 (0.5-2Hz) Delta 2 (24Hz) Theta (4-7.5 Hz) Alpha (7.5- 13 5 H z ) Beta 1 (1 3.5-20Hz) Beta 2 (20-30Hz) Ratio of high (p?) to low (61)
E EG frequencies
Other Core body temperature
(degrees Celsius. OC) Body mass
Conditions Body temperature pressure
(-37°C. 760mrnHg) Barometric pressure Alveoli water vapour pressure Water vapour pressure in
animal chamber Saturated water vapour pressure Standard temperature pressure
(273 Kelvin. 760 rnrnHg)
CV
CO? DRG f
H+ mmHg ml mumin PaCOl
phCo2 PCOl
Pa02
PO2
0 2
t r t~
TOT Y .A
Y COz
Y,
i' ,/i' COz
Y ,Y CO?
VRG VT
ANOVA
Respiration Coefficient of variation
(percent. %) Carbon dioxide Dorsal respiratory group Respiratory frequency (breaths
per minute Hydrogen ion Millimetres of mercury Millilitres Millilitres per minute Artenal partial pressure of
carbon dioxide Alveolar partial pressure Partial pressure of carbon
dioxide Anenal partial pressure of
oxygen (mmHg) Partial pressure of oxygen
(mmHg) Oxygen Inspiratory duration (seconds) Expiratory duration (seconds) Total breat h durat ion (seconds) Alveolar ventilation (ml/ min) Rate of production of CO?
(BTPS, mumin) Lung ventilation (inspired)
(ml /min) Lung ventilation normalized
for COz production Alveolar ventilation
normalized for ventilation Ventral respiratory group Tidal volume (ml. BTPS)
Statistics Analysis of Variance
List of Figures
Figure Number and Figure Contents
Introduction
1 - 1 : Graphical representation of a sine wave. 1 - 1-2: Schematic of the respiratory control system. 1 - 1 -2: Oxygen saturation and the ventilatory response to O?.
1-4: Metabolic hyperbolae at two different metabolic rates. 1 - 1-5: Metabolic hyperbola and the ventilatory response to COz. -
1-6: The chernoreflex control of breathing. 1 - 1-7: Mechanism of decrease in ventilation and increase in PaC02 on going -
from wakefulness to NREM sleep.
1-8: Pictorial representation of the hypotheses tested in the experiment. 1 - 1-9: Mechanisms of sleep-wake and time of day changes in lung ventilation in -
the rat.
1 - 10: Hypnogram of the rat sleep-wake cycle. -
Methods
2-1 : X-ray of EEG and EMG electrodes and the telemetry unit implanted in the - rat.
2-2: Schematic of experimental apparatus. 1 - 2-3 : Graphs of inspired Fractional CO2 concentration vs. time of day. 1 - 2-4: Correlations of inspired CO: (%) vs. deviations from mean lung -
ventilation.
Page
Figure Xumber and Figure Contents (continued)
Results
3-1: Raw EEG. EMG and breathing traces. -
3-2: Graphs of %Bz/%6i, EEG and EMG amplitude in each sleep-wake state - across time of day.
3-3: Graphs showing circadian rhythms in iung ventilation in wakefulness. - NREM and REM sleep.
3 4 : Comparisons of lung ventilation at two tirnes of day. -
3-5: Plots of metabolic COz production. ventilation normalized for CO, - production and body temperature across time of day.
3-6: Graph of phase relationships between lung ventilation. metabolic rate. - ventilation normalized for CO? production. body temperature and respiratory frequency.
3-7: Variability of breathing across time of day. -
3-8: Sleep-wake state etTects on breathing. -
3-9: Sleep-wake state effects on metabolism. ventilation normalized for CO2 - production and body temperature.
3-10: Additional day-night cornparisons in breathing. metabolism and sleep - related variables.
Discussion [llustration of conclusions.
Page
Chapter 1: Introduction
The regulation of respiration is highly complex and lung ventilation may be
modified by many factors such as sleep-wake states and metabolic rate. However, very
few studies have examined whether the circadian timing system plays a role in
modulating ventilation or how sleep-wake states, metabolism and lung ventilation
interact with circadian time. This topic is the focus of this thesis. This research may be
potentially important in revealing the role of the circadian timing system in the seventy
of common respiratory disorders such as hypoventilation syndromes. Speci fically, a
noctumal decline in ventilation may add to and exacerbate, the decline in lung ventilation
due to sleep and produce severe respiratory impairment in susceptible individuals.
With this in mind, the purpose of this study was two fold. First, to determine
whether circadian rhythms in lung ventilation are present in each sleep-wake state. The
second aim of the study was to determine whether the change in magnitude of lung
ventilation from wakefulness to sleep is the same across the day. Pnor to discussing the
methodology and answers to these questions, however, it is necessary to provide a basic
overview of circadian rhythms, sleep-wake states and the regulation of breathing to act as
a hmework with which the present study can be undentood. By no means, however, is
this intended to be a comprehensive review of each of these vast and complicated fields.
Rather, the discussion will be limited to phenornena most relevant to the present study.
CIRCADIAN RHYTHMS
Cyclical variations in behavioural and physiological variables are an inherent
property of living organisms (Aschoff, 198 1). Perhaps the most widely studied of these
rhythms across a large range of species, are those that are approximately 24 hours in
duration. i.e., circadian rhythrns. Circadian rhythms can be observed at many levels of
organization ranging from gene transcription to the physiological and behavioural levels
(Moore, 1999). In this study, the circadian rhythms of the sleep-wake cycle. body
temperature, metabolism and lung ventilation are of interest. Hence, greater emphasis
will be paid to these circadian rhythms rather than the multitude of others that have been
studied. However, in this section, these fonner three will be mentioned whilst over
viewing sorne of the basic concepts in circadian rhythms. The founh mentioned rhythm,
lung ventilation, is linked to the other three, but since it is the focus of this thesis, it will
receive greater attention towards the end of this introductory section.
Circadian rhythms, by definition, are cyclical and are conveniently represented
graphically as a sinusoidal waveform. There is an associated set of terminology relevant
in their discussion - period, mesor, minima and maximum value, amplitude and
acrophase. These terms are illustrated in the Figure 1-1.
The phase of a rhythm refers to the time of an instantaneous point in the cycle
relative to some other point. Phase relationships are often important in determining
possible causal relationships between two oscillating variables. If for example, two
oscillating variables have the same phase they could be dependent on each other. The
period is the time interval between recurrences of a defined phase of a rhythm. Circadian
rhythms have a penod of approximately 24 hours. Hence, the origin of the term circadian
(fiom circa meaning about, and, dies meaning day or about 24 hours). If the period is
less than 24 hours, it is termed an ultradion rhythm. A rhythm that is greater than 24
hours is tenned an infradian rhythm. When an organisrn is entrained or synchronized to
an exogenous (extemal) 12- hour light 12-hour dark cycle. the period of a measured
physiological or behavioural rhythm will also be 2 1 houn. However. if placed in constant
light or dark. the period of a circadian rhythm will begin to free run with a period that is
close to, but not exactly. 24 hours (Pittendrigh. 1960).
The mesor is the anthmetic mean of al1 instantaneous values of an oscillatory
variable within one cycle. The minima and maximum are as their names suggest (see
Figure 1 - I ). They are usually represented as the trough and the peak of a waveform of the
dependent variable plotted against time. The amplitude of a rhythm is the magnitude of
the difference between the rnesor and the peak or the trough value. The acrophase of a
rhythm refers to the time value corresponding to the peak of a sine funcrion.
Fipure 1-1: Craphical representation of a sine wave and the associated terminology.
I Period ,-4
Minimum (Troughi
A cornmon method of assessing whether a physiological or behavioural variable
possesses a circadian rhythm is cosinor analysis (Nelson, Tong, Lee, & Halberg, 1979).
In this method, in which a cosine or sine wave is fitted to the raw data by least squares,
the amplitude, mesor and the acrophase can easily be detennined. However, information
on the actual shape of the waveform of the vanable plotted across time is assumed to be
sine or cosine.
The circadian rhythms in body temperature are one of the most widely measured
of physiological rhythms in mammals and birds. The rhythm in body temperature is a
fairly robust rhythm that is present in many species, and under many expenmental
conditions. Consequently, it is ofien used as a reliable marker of the circadian phase of an
organism (Refinetti & Menaker, 1991). The amplitude of the body ternperature rhythms
in most homeotherms are fairly low, given that there is only a deviation of a one or two
degrees from a mesor value (in the range of 37-39°C). Under entrained conditions, body
temperature maximum occurs dunng the active portion and the minimum occurs during
the rest portion of the rest-activity cycle. Thus, for noctumal animals such as rats, which
are more active dunng the night, the maximum occurs during the dark phase of the light-
dark cycle. For these organisms, the body temperature minimum will occur during the
light phase when they are mainly resting (Refinetti & Menaker, 1991). For diurnal
organisms such as humans, the reverse is true - body temperature is higher dunnp the day
and low the body temperature minimum occurs a few houn pnor to waking (Czeisler,
Weitzman, Moore-Ede, Zimmeman, & Knauer, 1980).
Two fundamental propenies of circadian rhythms are their generation by
endogenous pacemakers and entrainment by environmental stimuli (Moore, 1997). One
possible explanation for the presence of a circadian rhythm is that it represents a response
to an exogenous cycle such as the extemal light-dark cycle, which also, has a period of 24
houn. However, this is not the case, given that circadian rhythms such as the circadian
sleep-wake cycle persist in studies where humans, for example, have been placed in
environments devoid of any time cues (e.g., Dijk & Czeisler, 1995). Under these constant
conditions, circadian rhythms become free mnning with their intrinsic period that is close
to, but not exactly 24 hours. Together, these data suggest that circadian rhythms are
endogenously generated. In order for the endogenously generated rhythm to be
synchronized to an exogenous cycle, the pacemakers mnning these rhythms must be reset
or entrained by exogenous stimuli known as zeitgebers (meaning time giver) (Moore,
1997). Thus, in the presence of a 12-hour light: 12- hour dark, light-dark cycle, as was the
case in Our experiment, al1 circadian rhythms would be expected to have a penod of 24
hours.
Zeitgebers (clock resetting stimuli) may be broadly classified into two categories
- photic (light based) stimuli or non-photic (non light) based stimuli. As judged by its
ability to cause large phase shifis in circadian rhythms, light is the most potent zeitgeber
(Pittendrigh, 1960). Examples of non-photic stimuli include social interaction
(Mrosovsky, 1988) and exercise (Edgar, Martin. & Dement, 199 1).
It is widely accepted that circadian rhythms are generated endogenously within
cells, tissues and organs (Moore, 1999). Pacemaker clocks, of which there are several, are
capable of synchronizing these inherent rhythms with different penods. in mammals. the
suprachiasmatic nucleus (SCN) located in the anterior hypothalamus has been established
as the pnmary pacemaker or time keeping mechanism (Ralph, Foster, Davis, & Menaker,
1990). When lesioned in rats for example. it results in the loss of many circadian rhythrns
such as locomotor and drinking behaviour (Stephan and Zucker. 1972). adrenal
corticosterone (Moore and Eichler. 1972). sleep-wake cycles (Eastman. Mistleberger. &
Rechtshaffen. 1983: Ibuka & Kawamura. 1975). metabolic CO2 production (Nagai.
Nishio. & Nakagawa. 1985) and body temperature (Eastman et al.. 1983). Common to
al1 these experiments. is the disruption of the temporal organization of these
physiological and behavioural variables when the SCN is lesioned. thus illustrating the
role of the SCN as a timekeeper. What is not changed. however. is the quantity of each
variable. For example. when SCN lesions destroy the circadian sleep-wake rhythm in
rats. the total arnount of time spent awake and asleep and the proportion of NREM and
REM sleep remains (Eastman et al.. 1983).
Perhaps the most convincing evidence for the importance of the SCN as the
primary time keeping mechanism in mammals are Iiom SCN transplant experiments
çonducted by Ralph et al.. (1990). When fetal SCN were transpianted into arrhythrniç
SCN lesioned animals. circadian rhythms were restored. Moreover. the restored rhyihms
had a period charactenstic of the SCN donor. These experiments drmonstrated that in
mammals. the SCN is the site of the circadian pacemaker. and that the intrinsic period of
a çircadian rhythm is penetically determined.
In addition to studies related to the SCN. the neural pathways by which external
eues such as light and social interaction are able to influence the SCN and its output to
various physiological systems have been investigated. There are several afferent inputs
into the SCN which mediate entrainment (Moore. 1997) such as the retinohypothalmic
tract. the interpniculate leaflet and the ventral lateral geniculate nucleus. Photic input
from the visual system is sent directly via the retinohypothalmic tract to the SCN
(Johnston, Moore. & Morin. 1988). Photic information is also sent indirectly via the
intergeniculate leaflet (Morin. Blanchard. & Moore. 1992) and the LGN (Card & Moore.
1982). The lateral pniculate nucleus. which has been shown to be important in mediating
non-photic entrainment. is also thought to be involved in integrating photiç and non
photic entraining information to the SCN (Moore. 1997).
The SCN. which shows a circadian rhythm in neuronal firing (Inouye &
Kawamura. 1979). is able to çontrol circadian rhythms such as sleep-wake and drinking
rhythms by efferent projections to regions involved in the control of these functions.
Although the number of projections from the SCN is relatively few. there are many areas
that reçeive seçondary projections from the SCN. The primary projections from the SCN
include the subparaventricular zone and the paraventriçular nuclei of the hypothalamus.
with lesser ones to the basal forebrain and the midline thalamus (Watts. 199 1 ). Primary
projections to the hypothalamus result in circadian information being relayed to the
anterior pituitary. the hypothalamiç. and the brainstem reticular formation regions
associated with autonomie regulation. the control of merabolism. body temperature and a
temporal organization of sleep-wake cycles (Moore. 1997)
It has often been proposed thai by usine an interna! time keeping mechanism. an
organism is better adapted to its environment. not only in space. but also in time (Moore-
Ede. 1986). This would enable orpanisms to predict and anticipate various occurrences in
the extemal environment such as availability of food. and thus. be able to synchronize
both behavioural and physiolopical changes to these events.
SLEEP-WAKE STATES
Like circadian rhythm, sleep-wake states are an integral component of
mammalian behaviour. Sleep is an essential behavioural phenomenon known to exist in
al1 marnmalian species studied so far (Zepelin, 1994). indeed, the consequences of sleep
deprivation can be debilitating and even cause death if extended for a sufticient duration
of tirne (Rechtschaffen. Bergmann, & Winter, 1983). Yet, currently the function of sleep
is both unknown and highly controversial (Rechtschaffen, 1998), but there is general
agreement that it is necessary, restorative, and beneficial.
Sleep is not a hornogeneous phenomenon, but rather, consists of two
fundamentally distinct neurophysiological states - rapid-eye-movement (REM) sleep and
Cun-Dossi, 1990). Also involved in wakefulness and EEG desynchronisation are areas
in the basal forebrain. The newotransrnitters involved in wakefulness, and their sites of
release (in brackets), include acetylcholine (basal forebrain and pons), histamine
(tuberomammillary nucleus), norepinephnne (locus coeruleus) and serotonin (dorsal
raphe) (Shiromani, Scarnmell, Shenn, & Saper. 1999). Tonic activity from these regions
is relayed via ascending pathways to the cerebral cortex to produce cortical activation
that occurs during wakefulness (Szymusiak, 1995).
While wakefulness is the result of the excitation of the reticular activating system,
NREM sleep is charactenzed by inhibition of this region by the inhibitory
neurotransmitter, G A B k from the ventrolateral preaptic area (Shiromani et al., 1999).
There are also regions in brain that have been implicated for triggenng NREM sleep such
as the preoptic anterior hypothalamus and basal forebrain pre-optic areas (Shi romani et
al., 1999). Lesioning these areas results in long lasting insomnia (Nauta. 1 946), electncal
stimulation leads to a NREM sleep like pattern on the EEG (Sterman & Clemente. 1962),
and recordings of cells in this area show cells which change their firing rate in NREM
sleep (Szymusiak & McGinty, 1986).
Cholinergic and cholinoreceptive pontine reticular formation neurons are thought to
play an important role in regulating REM sleep. Lesions of the dorsolateral pontine
tegmentum (including the lateral dorsal tegrnental nuclei and pedunculo-pontine nuclei)
can disrupt and eliminate REM sleep (Webster & Jones. 1988). The muscle atonia that is
characteristic of REM sleep is thought to occur by postsynaptic inhibition and
disfacilitation (i.e., a reduction in the tonic discharge of tonically active presynaptic
excitatory neurons) of spinal motoneurons via brain stem mechanisms (Chase & Morales,
1 994).
In addition to understanding the generation of sleep-wake states. the profound impact
of sleep-wake states on many aspects of regulatory physiology have also been studied
especially when the sleep-wake state have clinical implications. However, addressing the
many changes that occur for example, in the cardiovascular system or autonornic nervous
system with sleep, is beyond the scope of this thesis (see Coote, 1982; Horner, 2000 for
reviews). The impact of sleep-wake state on the respiratory system will be discussed, as
it is important in understanding the impact of sleep-wake States on lung ventilation. Pnor
to doing so, it will be necessary to overview important concepts in the regulation of
breathing.
REGULAT~ON OF RESPIRATION
One of the most fundamental functions of breathing is to facilitate gas exchange
across the lungs, and in doing so, supplying oxygen (O2) and removing carbon dioxide
(CO2) to and h m the tissues. Oz is an essential substrate and COz is the byproduct of
cellular aembic catabolism, the process by which many organisms consente vital
chemical energy in the form of adenosine triphosphate (ATP). When the lungs are
ventilated, the diffision of O2 and COr along their partial pressure gradients across the
alveoli (the gas exchange portions of the lungs) is facilitated. Using the respiratory
control system to regulate lung ventilation, the body can maintain the blood gas and
hydrogen ion concentrations in arterial blood within a narrow range. Breathing can be
altered by many factors such as emotion, speech, exercise etc. However. the present
discussion will be limited to factors goveming the control of resting ventilation, as this is
the focus of this thesis. In this regard. in this section, an oveMew of the respiratory
system will be given, and mechanisms such as the central and petipheral chemoreflexes
and the effect of wakefulness involved in determining resting lung ventilation will be
discussed.
The respiratow control system
Essentially, breathing is controlled by two anatomically separate. but Functionally
integrated elements referred to as the metabolic (or automatic) control system and the
behavioural (or voluntary control system) (see Figure 1-2) (Berger. Mitchell. &
Severinghaus. 1977: Phillipson. 1978). The metabolic control system onpinates in
brainstem (pons and the medulla) structures and is primarily concemed with blood gas
Fipure 1-2: Schematic of the respiratory control system and its cornponents (modified from P hillipson, 1978)
Controllers
Cerebral Cortex
(Behavioural control)
Retfcular Activatlnn - System Brain Stem
v Wakefulness +f (~etabol ic controi 1.e. 1-. I
Stimulus) Wood Ga. hom.ostasis~+ 1
Respiratory I
Motonmurons I 7
Oiaphragm &
Intercostal Muscle
Upper Almays : 8 tungs
I
Upper Airway and Lung Receptors
Propnoreccpton
Penpheral and Central Chemorecepton 1
homeostasis. The behavioural control systern anses in cortical and forebrain stmctures
and is involved in activities such as speech, which utilize the respiratory system for non-
respiratory functions (Phillipson, 1978).
The metabolic control system can be conceptualized as a feedback control system
that is composed of a central controller, effectors and sensors (see Figure 1-2). Each
component of the metabolic control system has been studied separately using a vanety of
neurophysiological and neurohistological techniques. In addition, the integrated
responses of this metabolic control system have been studied using a variety of stimuli
such as hypercapnia (high COr) and hypoxia (low Oz) in animals and humans in much
detail (see Cunningham, Robbins, & Wolff (1986) for review). The present study,
involving the measurement of lung ventilation in each sleep-wake state under normal
room air conditions, represents a study of the integrated response of the respiratory
control system across time of day.
The central controller, located in the brainstem, is the site of the respiratory
rhythm generator needed for automatic breathing (Berger et al., 1977; Feldman & Smith,
1995). Evidence for this cornes From intact brainstem and spinal cord preparations, which
can generate respiratory motor nerve output in the absence of rostral or sensory afferents.
in neonatal and fetal rats, respiratory related patterns continue following removal of the
brainstem (and spinal cord) to an in vitro chamber, and, an isolated section of a particular
slice of the medulla will even generate respiratory related rhythms (Smith, Ellenberger,
Ballanyi, Richter. & Feldman, 1991). These experiments support the existence of a
pacemaker mode1 of respiratory rhythm generation and the region hypothesized to be
involved is the pre-Botzinger region (Smith, Ellenberger, Ballanyi, Richter, & Feldman,
199 1; Feldman & Smith, 1995).
Located in the dorsal respiratory group (DRG) and ventral respiratory group
(VRG) of the medulla oblongata are respiratory neurons whic h display sync hronized
activity wit h both inspiration and expiration. DRG neurons are mainly inspiratory, whilst
neurons in VRG, contain both inspiratory (located more rostrally) and expiratory (located
more caudally) neurons. initially, the DRG and the VRG were thought to contain the
respiratory rhythm generator, but it is now believed that respiratory rhythms are
generated in the rostral ventrolateral medulla. (Duffin, Emre, & Lipski. 1995). Evidence
for this cornes fiom the ceasing of respiratory rhythmogenesis following local cooling of
this region for example (Budzinska, Euler, Kao, Pantaleo, & Yamamoto. 1985).
Aithough the locus of respiratory rhythms has been detemined, the mechanisms by
which it is generated remains to be elucidated. Currently, there is evidence for pacemaker
models in neonatal and fetal rats as alluded to earlier. However in adults, the absence of
cells with pacemaker like properties make network rnodels of rhythm generation appear
to be more likely (Duffin et al., 1995).
The rhythmic impulses fiorn the respiratory rhythm generator are relayed to spinal
and cranial motoneurons, which in tum, innervate pnmary and secondary respiratory
muscles involved in generating airflow and maintaining an effective air passage fcr this
airflow. The pnmary respiratory muscles, such as the diaphragm and the intercostals,
when contracted during inspiration, create a negative pressure by increasing the volume
of the c hest cavity, thereby generating airtlow into the lungs. When these muscles relax
on expiration, air is expelled from the lungs. The electrical impulses leading to
contraction of the diaphragm and the intercostal muscles, originate From motoneurons
located in the ventral horn of the spinal cord and, are relayed to these muscles via the
phrenic and the extemal intercostal nerves, respectively. The secondary respiratory
muscles such as the laryngeal, pharyngeal and hypoglossal (skeletal) muscles modulate
airway resistance and are innervated by the cranial motoneurons (Feldman & Smith.
1995).
Respiratory motoneurons receive both phasic and tonic afferent inputs that affect
their excitability and output (Orem, 1994). Phasic input is received from the respiratory
rhythm generator neurons in the brain stem. Tonic inputs, on the other hand, are of many
different forms and types. Examples of tonic afferents include inputs from the sensors of
the respiratory control system such as lung stretch recepton, central and penpheral
chemoreceptors and non-specific activity of the reticular formation (Phillipson & Bowes.
1986). The lung stretch receptors are located in the muscular portions of the walls of the
Iower airways (bronchi and bronchioles) throughout the lungs, and are innervated by the
vagus nerve (Guyton & Hall, 1996). When the lungs become inflated, they increase their
rate of firing, and consequently, terminate further inspiration, prolong expiration and
decrease respiratory rate, in what has been termed the Hering-Breuer inflation reflex. h
this way, excess inflation of the lungs is prevented (Berger et al., 1977). The afferents
fiom the central and peripheral chemorecepton, which are vital in acid base and blood
gas homeostasis, and the reticular formation, will be discussed in greater depth below.
When these afferents are abolished, one by one, in intact unanaesthetized dogs, there
is a step wise reduction in respiratory Frequency and lung ventilation corresponding to the
progressive suppression of the major respiratory drives related to wakefùlness, vagal,
penpheral and central chemoreceptor stimuli (Sullivan, Kozar, Murphy, & Phillipson,
1978). Moreover, this result also suggests that individual afferent inputs may be summed
to determine the overall level of tonic drive needed for ventilation.
Chemical control of breathing
The central and peripheral chemoreceptors, which constitute an essential part of
the chemical regulatory mechanisms of the respiratory control system, adjust ventilation
in such a way that alveolar and artenal pariial pressure of CO2 (PC02) is tightly
regulated. These sensors form one of the most important afferent inputs involved in the
regulation of breathing under normal conditions. indeed in models of the regulation of
breathing in the resting state, breathing is often assumed to be determined to a large
extent by chemical stimuli and wakefulness (Duffin, 1990). Therefore, understanding the
fûnction and behaviour of central and peripheral chemorecepton, and the role of
wakefulness, is usefd in providing insights into factors and mechanisms which may
possibly govern circadian rhythms in lung ventilation in any given sleep-wake state.
Stimuli involved in control of breathiog
Both the central and penpheral chemoreceptors are sensitive to H+ and CO2, but
only the peripheral chemorecepton are sensitive to Oz (Gorualel Almaraz, Obeso, &
Rigual, 1992). Of these three chemical stimuli, CO2 and / or H+ have been shown to be
more important in controlling lung ventilation under normoxic (normal room air)
conditions in which the fiactional concentrations of Oz and COz in ambient air are 2 1.9%
(16OmmHg) and 0.03%, respectively (Ganong, 1999). The role of O2 in maintaining
normal ventilation requirements under normoxic conditions is relatively minor. hdeed,
the partial pressure of O: (PO?) may Vary quite considenbly without çhanping the total
O2 saturation of arterial blood (see
Figure 1-3 panel A). Hemoglobin.
the oxygen carrying protein in the
blood is fully saturated at normal
arterial PO? (PaO:) of 100 mmHg
(Nattie. 1999). It is also almost
fully saturated with even
considerably lower levels of
ventilation. The presence of a
sigmoid relationship between PaO?
and hemoglobin Oz saturation
means that levels of ventilation
which result in PaOl between 70
mmHg and 100mmHg have little
Figure 1-3: Graphs of the ( A ) oxygen saturation and PO2 and (B) the ventilatory response to O2 at a constant PCOz(see text for details) (modifieci from Yattie, 1999).
effect on the total amount of 0: carried in artenal blood (see Figure 1-31. At normal
levels of ventilation. the baseline firing of carotid and aortic bodies constitute a tonic
drive to the respiratory control system. Removal of this tonic drive for example by
denervation of these structures. results in Iesser resting ventilation (Olson. Vidruk. &
Dempsey. 1988). This tonic drive differs from the changes in afferent input from the
carotid body to the brain that results from minute to minute changes in PaCO? or PaO?.
The ventilatory sensitivity to PaO?. defined as the change in ventilation for a unit change
in Pa02, is low at levels just above or below the normal value of 100mmHg. Therefore.
the control of alveolar ventilation just above or below the normal value is unlikely to be
determined by the Pa02.
in contrast to the ventilatory responses of PaOz, the sensitivity of the ventilatory
response to PC02 is much higher, with only an increase of a few rnrnHg in PC02 fiom the
normal value being sufficient to stimulate ventilation significantly (see Figure 1-5). Thus.
the stimulus that provides input usetùl in the maintenance of normal ventilatory
requirements is CO2 (Nattie, 1 999).
Funher evidence for the importance of COr in normal ventilation comes fiom
experiments on awake sheep (Phillipson, Duffin, & Cooper, 1981). Using a carbon
dioxide membrane lung to add and remove COr to and ftom the lungs of awake sheep,
the level of metabolic COz production was linked to the level of lung ventilation in a
linear manner. When the rate of removal of COz equaled the rate of metabolic CO2
production by the animal, PaC02 remained normal. but ventilation ceased. Thus, these
investigators were able to show that one of the most critical afferent inputs required for
the generation of respiratory rhythms are stimuli related to metabolic CO2 production.
The linear relations hip between lung ventilation and metabolic rate expressed as
either, the rate of O2 consumption (Y 02) or the rate of COz production ( Y CO2), has also
been seen in many examples that increase rnetabolic rate, such as, exercise (Wasserman.
the exact nature of this link between lung ventilation and CO2 production is unclear
(Mortola and Gautier, 1995).
Given that the stimulus that is involved in the maintenance of normal ventilatory
requirements is likely to be CO2 (Nattie, 1999), attention can now be focused on factors
that govern the concentrations of this gas (quantified as partial pressure of CO?, PC02) in
arterial blood.
Determination of resting ventilation and PaC02
The normal artenal PCOr (PaCO?) in mammals of approximately 40 mmHg is
linked to acid base balance and the maintenance of a normal extracellular pH of 7.4
(Nattie, 1999). PaCOr values represent a balance between the metabolic production of
COr by the body tissues, and the amount of ventilation of the alveolar space. For
example, a decrease in alveolar ventilation ( Y a) at a constant rate of COz production
increases anerial and alveolar PC02. This increase in PCOz is detected by the
chemoreceptors leading to the stimulation of ventilation, which eliminates COz and
decreases PaC02 to its equilibrium value. From this illustration, two observations on the
relationship between PaCOz and ventilation can be made. Fintly, it is evident that
ventilation alters PaC02 at a given metabolic rate (also see Figure 1-4). Second,
ventilation is itself affected by PaC02 levels (also see Figure 1-5). The first relationship
is ofien referred to as the ventilation equation, ventilation hyperbola, or the metabolic
hyperbola. The second is called the ventilatory response to COz or the chemoreflex, and,
is the feedback component in the control of ventilation. Together, these relationships are
usefbl in conceptualking the control of ventilation as consisting of a control loop
consisting of two pans - the fonvard loop and the feedback loop (Cunningham et al.,
1986; Dufin, 1990).
Integrating the forward and feedback components involved in the chernical
control of breathing is valuable in deterrnining the PaC02 and ventilation under difTerent
conditions. in steady state, the resting PaCOz level (which will equal alveolar PC02 since
equilibration has occurred) can be calculated. Since the metabolic hyperbola and the
ventilatory response to COz share the same axes. they can be superimposed on each other
after reversing the axes of the metabolic hyperbola (see Figure 1-5). The intersection of
these graphs represents an equilibnum between the rate of production of COz and the rate
at which the lungs elirninate the CO2. This equilibrium point, which gives an indication
of the prevailing ventilation or PaCO? levels, may be altered either by shifts in the
metabolic hyperbola (caused by changes in metabolic rate). or. by changing the
charactenstics of the chemoreflex cuwe. A decrease in metabolic rate and altered
chemoreflexes dunng sleep, for example. will reduce ventilation and increase the PaC02
(see Figure 1-7). Therefore, undentanding the metabolic hyperbola and the
c hernore flexes are important in understanding and predict ing resting lung ventilation
under various conditions.
The metabolic hyperbola
The forward loop, s h o m by the metabolic hyperbola, refers to the effects that
changing ventilation has on PaCOr at a given metabolic rate. It expresses al1 possible
pain of the independent variable, alveolar ventilation (Y A) and the dependent variable.
PACOz (see Figure 14). PaCOz is proportional to the ratio of the rate of production of
CO? from metabolism (Y CO2) and the rate at which it is eliminated from the lung via
ventilation of the alveoli. It can be expressed as an equation:
PaCOz = k Y COz
Y A
where PaCOz is the partial pressure of CO? in artenal blood. Y COz is the rate of
production of COz. Y .+ is alveolar ventilation and k is a constant relating the solubility of
COz gas in blood.
Higher metabolic COz production during exercise. or during the active portion of
the rest-activity cycle. for example. has the etTect of shifting the curve to the right (Le..
for a given ventilation. PaCO? is higher). Fipure 1-4 showing rnetabolic hyperbolae for two different rates of COr production in
Converçely* a drap in metaboli' 'Oz hurnans (derived from White et al. 1985).
production dunng sleep or during the rest
phase of the rest-activity cycle. causes a
leftward shift in the metabolic hyperbola
(i.e.. for a given level of ventilation PaCOz
is lower). Since the metabolic rates oscillate
across a 24-hour penod (e.g.. in Aschoff and
Pohl. 1970). we could expect a whole
continuum of metabolic hyperbolae to Alveolar Ventilation (Umin)
represent these changes in metabolism across times of day.
The Chemoreflexes
As mentioned earlier. C02/H+ at the chemoreceptors represents a major tonic
afferent input to the respiratory rhythm generator and is consequently able to influence
respiratory output considerably. This tonic input from the central and penpheral
chemoreceptors to drive ventilation on
stimulation by COdH+ is called the
chemoreflexes (which may be
graphically illustrated by a graph of
the effects of PaCO- on ventilation -
see Figure 1-5). Also known as the
ventilatory response to CO?, the
chemoreflexes have been studied by
measunng lung ventilation to varying
PaCO? (which itself is changed by
altenng the inhaled CO2
concentrations) under steady state
conditions. or. by using techniques
Fipure 1-5 showing the ventilatory response to CO2, the metabolic hyperbola and the equiiibrium point of these two relationships to give resting alveolar ventilation and PaCOI in the awake adult human (derived from data in White (1 985) and P hitliason (1 978)).
L,
'Ë . , Ventilatory 3 ' , % Response to CO2 Y /
hyperbola
such as the rebreathing method (e-g.. Read ( 1967) and moditied rebreathing technique
Mohan & Duffin (1997)). Using these techniques. the ventilatory response to CO? has
shom to be approximately linear above a certain threshold (see Figure 1-6 panel D). The
dope of this line represents the sensitivity of the response of the chemoreceptors to C02.
The intersection of this line with the PaCO2 axis (at which point the corresponding level
of ventilation is zero) has cornrnonly been extrapolated to represent the threshold or the
PCO? value at which PaCO? no longer drives breathing. However. since there is a basal
level of ventilation independent of PaCOz in normal awake individuals. the actual value
of the threshold is the PaCOt at which ventilation increases above basal ventilation (see
Figure 1-6 panel D). This ventilatory threshold is represented by the inflection point of
the ventilatory response to CO?. To get the ventilatory response to CO1 for PaC02 values
below the ventilatory threshold, the body stores of CO2 need to be reduced by voluntary
hyperventilation (e.g in Duffin & McAvoy, 1988) or mechanical ventilation. The
resulting chemoreflex curve in humans has the shape of a hockey stick or dogleg (see
Figure 1-6). However, obtaining the COr response From resting ventilation yelds a COr
response curve similar to the one depicted in Figure 1-5.
The chemosensitivity can be altered by many conditions such as hypoxia (Mohan
& Duffin, 1997), time of day (Raschke & MIller, 1989; Spengler, Czeisler, & Shea.
2000) and sleep-wake state (Douglas, White, & Weil et al., 1982). Of direct relevance to
this study, are the impacts of sleep-wake state and time of day on sensitivity. There is a
reduction in chemosensitivity in NREM sleep and REM sleep compared to wakefulness
(Douglas et al., 1982). in addition to time of day changes in chemosensitivity. there are
also reports of circadian rhythms in threshold (Stephenson, Mohan, Dufin, & Janky,
2000). These changes in the chemoreflex, whether by altering the threshold or
c hemosensitivity, can have a marked influence on ventilation.
Ventilation fiorn any set of chemoreceptor stimuli, including those present under
resting conditions, can be predicted by adding together the contributions to ventilation of
the central and perip heral chemoreflexes, in addit ion to the c hemoreceptor independent
effect stimulatory effect of wakefulness (Duffin, 1990; Duffin et al, 2000) (see Figure 1-
6). The ventilatory responses to central and peripheral chemoreceptors simulation are
both linear, but the central chemoreflex response is more sensitive to COz / H+ than the
penpheral chemoreflex in normoxia or hyperoxia (Dufin, 1990) (see Figure 1-6 panels A
and B). However, the sensitivity of the penpheral chemoreceptors to CO2, is modulated
by the Pa02, such that sensitivity is increased at low Pa02 (Dufin, 1990) (see Figure 1-6
panel B). The nature of this interaction between low O2 and H+ is multiplicative rather
than additive (Cunningham et al., 1986; Duffin, 1990).
The central chernoreflex is mediated by the central chemoreceptors, which are
located in the brainstem and are perfused by artenal blood from the ventral surface by the
basilar artery (Nattie. 1999). Central chemoreceptors were traditionally thought to be
located on, or near to. the surface of the ventrolateral medulla (Loeschcke, 1982;
Mitchell, Loeschcke, Massion, & Severinghaus, 1963), but experiments that have
employed techniques such as focal acidification (using the carbonic anhydrase inhibitor
acetazolamide) suggest that they are present in many locations in the brainstem (e.g.,
Coates, Li, & Nattie, 1993). These areas include the locus coeruleus and the medullary
raphe (Bernard, Li. & Nattie, 1996). which are involved in arousal state regulation.
Central chemoreceptors respond to changes in hydrogen ion concentrations in their
imrnediate environment. However, this hydrogen ion concentration is more closely
related to PaCOz than arterial hydrogen ion concentration (Duffin, 1990).
The mechanism by which PaC02, rather than artenal hydrogen ion concentration,
is able to influence the hydrogen ion concentration in the vicinity of the central
chemoreceptors, involves the relatively rapid diffusion of CO2 corn arterial blood across
the blood brain banier, and, the carbonic anhydrase catalyzed reaction of CO2 with water
to form hydrogen and bicarbonate ions. in contrat to CO2, H+ in arterial blood does not
cross the blood brain bamier easily because it is a polar ion. The relationship of H+ and
CO2 in the environment of the central chemoreceptors cm be expressed using the linear
form of the Henderson Hasselbach equation: [H+] = 24 PC02/ [HC03'] (where [H+] is
the hydrogen ion concentration (in nanomoleçllitre) and [HC03'] is the bicarbonate ion
concentration (in millimoles/litre) which at normal resting ventilation is 24
millimoles/litre). It must be noted, however, that the PC02 at the central c hemorecepton
is not the same as that in arterial blood, even when fully equilibrated. It is in fact higher.
at values close to mixed venous cerebral blood because the low blood flow rates in brain
tissue cause a lag between changes in PaCOz and PCOz at the central chemoreceptors
(Dufin, 2 990).
Unlike the central chemorecepton, which respond to hydrogen ion concentration
only, the peripheral chemoreceptors, the mediators of the penp heral c hemoreflexes,
respond to both hydrogen ion concentration and Pa02. Hence, they are the primary 0-
sensors in the respiratory control system and serve a protective function against low PO2
in arterial blood (Gonzalez, Almaraz, Obeso, & Rigual, 1992). The peripheral
chemorecepton consist of the carotid and aortic bodies and are located at the bifurcation
of the common carotid arteries. The carotid bodies send afferent inputs to the brain via
the carotid sinus branch of the glossopharyngeal nerve, and fibres from the aortic bodies
ascend via the vagus nerve (Ganong, 1990). In marnmals, it is estimated that the carotid
bodies are responsible for approxirnately 90% of the response to hypoxia (Gonzalez,
Almamz, Obeso, & Rigual, 1992).
The penpheral chemoreceptors are estimated to also be responsible for 2040% of
the response to arterial hypercapnia and pH, with the remaining 50-80% of the response
to hypercapnia being mediated by the central chemoreceptors (Gonzalez, Aimaraz,
Obeso, & Rigual, 1992). This result suggests that the central chemoreceptors may be
relatively more important in seîting the level of ventilation in response to changes in
PaC02. Comrnon methods used to elucidate the relative importance of the peripheral and
the central c hemoreceptors in the vent ilatory responses to increased or decreased PaC02,
include removal of the peripheral contribution either by surgical means (Le., carotid body
denervation (Olson et al., 1988) or by the administration of hyperoxic gas mixtures (e.g
Heeringa, Berkenbosch, De Goede. & Olievier, 1979). n ie role of the penpheral
chemorecepton in contributing to the ventilatory response to CO2 under normoxic
conditions has also been estimated using these techniques. but the relative importance of
these chemorecepton in this role may Vary with species (Tenney & Boggs, 1986).
Wakefulness stimulus
in addition to the central and penpheral contributions to breathing, there is also a
neural drive to respiratory neurons that is varied between States of alertness, but is
independent of chemoreceptive stimuli. The site of origin of this neural drive is the
midbrain reticular formation, a region implicated in the generation of wakefulness. When
this region is stimulated, like wakefulness, it also increases rate and depth of breathing
(Orem, 1 994).
The influence of sleep-wake state on breathing was first demonstrated by Fink
( 196 1), who subsequently temed the stimulatory effect of wakehilness on breathing
independent of CO2, the "'wakefulness stimulus". When normal subjects hyperventilate to
lower their PaC02 below threshold leveis, they continue to breath if they are awake.
However, if asleep, drowsy or anest hetized, breathing completely ceases (apnea) after
hyperventilation (Fink, 196 1 ; Datta, Shea, Homer, G u , 199 1). Under these conditions,
the tonic stimulatory effect of wakefulness on respiratory output is no longer present and
breathing is almost entirely dependent on metabolic stimuli CO2. However. if PaC02 is
below threshold levels after hyperventilation. there is an absence of stimuli to drive
breathing. and thus. breathing ceases.
Fipure 1-6: The chernoreflex control of breathing consists of the central (panel A) and two peripheral (panels B and C) chemoreflexes. When added together with the basal ventilation due to wakefulness they can be used to predict the ventilation under a vanety of conditions. One erample is shown in panel D which shows the prediction of ventilation at different PO1 levels, including those a t normoxia (POZ=100mmHg), in awake adult humans (modified from Duffrn, 1990 and Mahamed, 2000).
Changes in ventilation fiom wakefulness to sleep have been documented across
many species including humans (White. Weil. & Zwillich, 1985). dogs (Phillipson.
Murphy, & Kozar. 1976) and rats (Pappenheirner, 1977). There is an overall decline in
alveolar ventilation and breathing becomes more regular on going from quiet waking to
NREM sleep. In general. these
reductions in alveolar ventilation are
either mediated by a decrease in tidal
volume in humans or a decrease in
respiratory frequency in animais
(Krieger. 1989). These changes in
ventilation are accompanied by a 10-
30% reduction in metabolic rate as
indicated by a decrease in O? uptake
and CO? production (Brebbia &
Aitushuler. 1965: White et al.. 1985).
The ventilatoiy responses to COz are
also altered as a function of sleep-wake
state. in NREM sleep. there is a
decreased sensitivity to CO? but no
change in threshold of the COz
Figure 1-7 shows the mechanisrns involved in the decrease in ventilation on going from wakefulness to NREIM sleep. The intersection of the metabolic hyperbola (line a) and the ventilatory response to CO2 during wakefulness yields the equilibrium point 1. O n falling into NREM sleep, metabolic rate declines, shifting the metabolic hyperbola Ieft (line b). The loss of the "wakefulness stimulus" during NREM sleep rnakes the ventilatory response to COz tess sensitive (shown by line d). The net result is a new equilibrium point during NREM sleep (point 2 ) a t which ventilation is lower and P a C 0 2 is higher compared to wakefulness (derived from Phillipson. 197R\
response (Douglas et al.. 1982). Together these changes result in an incrrase of 2-7 rnrnHg
in PaCOz (see Figure 1-7 for mechanism) (Phillipson. 1978).
Ventilation that occurs during REM sleep is variable. Some studies in human infants
and adults have found increases, whereas others in cats have found decreases. During REM
sleep, breathing is irregular and there are large variations in tidal volume and respiratory
frequency (Phillipson et al., 1976; Remmers. Bartlett, & Putman, 1976). Periods of
hyperventilation, regular breathing as well as apneas of varying length have been noted in
REM sleep. Consequently, there are large fluctuations of instantaneous measurements of
ventilation on a breath-to-breath and minute-to-minute basis. This large vanability in
breathing indicates that steady state ventilation may not exist in REM sleep (Phillipson &
Bowes, 1986).
The mechanisms involved in the decrease in lung ventilation during NREM sleep
are related to the withdrawal of the stimulatory effects of the ''wakefulness stimulus" on
breathing. Wakefulness exens a tonic stimulating effect on the rnedullary respiratory
neurons. Falling asleep is associated with the withdrawal of this tonic drive to the
respiratory motor neurons (reviewed in Orem, 1994; Phillipson & Bowes, 1986), which
ultimately affects respiratory motor output. Respiratory related neurons in the ventral
medulla, for example, have been observed to show either a decrease in rate of discharge
or cessation of firing dunng sleep compared with that dunng wakefulness (Orem,
Montplaisir, & Dement, 1974). The l o s of wakefulness has rnany different efkcts on the
muscles of the respiratory system which. in addition to decreases in breathing rates and
alveolar ventilation, are manifested as decreases in peak inspimtotory aimow rate and
increases in upper airway resistance (Orem Netick, & Dement, 1977), for example. It has
been proposed that the degree to which a respiratory muscle will be affected in sleep
depends on the ratio of respiratory and non-respiratory inputs to the motoneurons
innervating the respiratory related muscle (Orem & Dick, 1983). Thus, the activity of the
diaphragrn, which receives more respiratory input than non-respiratoiy input, will be less
affected by sleep than muscles of the upper ainvay such as the genioglossus (tongue
muscle), whic h receives a greater degree of non-respiratory input (Orem, 1994).
The reductions in activity of these important respiratory muscles dunng sleep are
of clinical relevance. For example, in a pathological condition known as obstructive sleep
apnea. sleep related relaxation of pharyngeal muscle activity leads to upper ainvay
narrowing. snoring and upper airway obstruction, which, disrupt sleep, produce anerial
oxygen desaturation and significant hemodynamic changes (Homer, 2000). in another
condition known as congenitaf hypoventilation syndrome, there is a cessation of breathing
during sleep because of the absence of any ventilatory responses to CO2 which are vital in
driving ventilation during sleep in the absence of a wakefulness drive (Shea 1996). Lefi
untreated these conditions are potentially debilitating and life threatening.
Unlike NREM sleep, the mechanisms by which ventilation changes in REM sleep
are cornplex and cannot be attributed to a single event such as withdrawal of the
wakefulness stimuli (Phiilipson & Bowes. 1986). Like wake fulness, there are many
different factors and physiological mechanisms which control ventilation in REM sleep.
Regulation of breathing in REM sleep. unhke in NREM. is largely independent of the
metabolic control system. Therefore, spontaneous changes and gas exchange are probably
not attributable to the metabolic control w e m (Phillipson & Bowes, 1986).
Circadian Modulation of Breathing
Unlike the impact of sleep-wake state on breathing, knowledge on the modulation
of the breathing via the circadian system is limited. Studies that have focused on the
changes in breathing across time of day have been relatively few in number, both in ternis
of describing the changes that occur across the day, and, in the mechanisms mediating
these changes. The reasons for the spane literature in this field probably stems from the
fact that this is a relatively new subject in the field of the control of breathing. Ln
addition, t here are also difficulties of measunng breathing across long penods of time.
whilst controlling for the potential masking effects of sleep-wake state. However.
evidence is accumulating that suggests that the circadian system is involved in the
modulation of breathing, either directly through the respiratory control system. or
indirectly through circadian rhythms in other variables such as metabolic CO2 production
or sleep-wake state, which as previously discussed, are known to influence breathing.
Several studies in awake humans have investigated changes in the respiratory
control system across time of day. Raschke & Moller (1989) found that
chemoresponsiveness to CO2 possessed a circadian rhythm, with the trough of the rhythm
(Le., minimum chemosensitivity, maximum threshold) occumng at 5 a.m. Stephenson et
al. (2000) also found that the chernoreflex threshold (maximum at 6 a.m.) possessed a 24-
hour rhythm, but they found no rhythm in chemosensitivity. The reasons for these
differences are unclear, but it is possible that they may have been caused by di fferences
in protocol. For example, in Raschke & M6ller (1 989)'s study, subjects were allowed to
sleep dunng the study, but were woken up to make respiratory measurernents. In contrast,
in Stephenson et al. (2000)'s study, subjects were constantly awake throughout the length
of the forty hour experimental period.
Times of day differences in chemoreflexes have also been tested in rats and
ducks. The ventilatory responses of awake rats exposed to 3.5% COz were higher during
the night (active) phase compared to the day (rest) phase (Peever & Stephenson. 1997).
suggesting that c hernosensitivity varies across time of day. Similarly. the vent ilatory
response of diving ducks to progressive asphyxia (progressive hypercapnia and hypoxia)
has been found to be lower at night, a time of day when these anirnals remain submerged
longer (Woodin & Stephenson, 1 998).
Like the studies on the circadian influences on the respiratory control system, few
studies have measured tirne of day changes in lung ventilation. Two studies. one in awake
humans, and the other in rats, have show that there are 24 hour rhythms in ventilation.
Spengler, Oliver, Czeisler, & Shea (1997) have shown that a circadian rhythm in
ventilation is present in hurnans kept awake for 40 hour periods under constant conditions
(using the constant routine protocol). This result suggests that there may be circadian
influences on breathing independent of sleep-wake state. Siefen et al. (2000) also showed
that that there was a circadian rhythm in lung ventilation in freely behaving rats.
However, in this study, the sleep-wake States of the rats were not considered as the data
were pooled into 20-minute bins regardless of sleep-wake state. Knowing that rats are
asleep for greater periods of time during the day and are predominantly awake dunng the
night, it is not clear fiom this study, if the circadian rhythm in lung ventilation is
independent of the circadian related changes in sleep-wake state.
Although Peever and Stephenson (1997) considered the effect of wakefulness in
their analysis of time of day differences in breathing in seven rats, they found no
statistically significant differences between lung ventilation dunng the day and lung
ventilation at night. This result suggests that lung ventilation in awake rats does not
possess a circadian rhythm. However. there was a trend towards higher lung ventilation at
night, suggesting that day night differences in Iung ventilation could have been found had
the vanability in the data been less with the use of a larger sample size. It is possible that
this lack of difference may have been a result of lung ventilation being relatively higher
than the expected value dunng the light phase. and. relatively lower than the expected
value dunng the dark phase, or a combination of both. These hypothetical scenarios may
be possibly explained by two expenmental shortcomings. Fintly, arousing the animal to
make measurernents dunng the light phase (when the animal is normally asleep) would
likely cause lung ventilation to be higher during this phase. Secondly. since no EEG or
EMG recordings were used to discriminate sleep-wake States. it is possible that somr of
the measurements were inadvertently made in drowsy, rather than fûlly awake rats. If
such were the case dunng the dark phase, for instance, this would depress lune
ventilation during this phase.
Although the studies discussed above suggest that there are circadian rhyt hms in
lung ventilation independent of sleep-wake state, they are by no means conclusive. They
have either neglected the effect of sleep-wake state on lung ventilation or measured lung
ventilation across 24 hours in wakefulness, but not in sleep. To date, we are not aware of
any studies that have determined if there are circadian rhythrns in lung ventilation in
wakefulness g& NREM sleep or REM sleep. With a deficiency of studies in this area, it
remains to be known if circadian rhythms in lung ventilation are independent of sleep-
wake state, or if the effect of sleep-wake states on lung ventilation can be added to the
circadian effect of lung ventilation.
Hv~otheses
The aim of the present study was to determine the effect of time of day on lung
ventilation taking into account the effect of sleep-wake state. Two hypotheses (see
Figure 1-8) were proposed and tested:
1. Lung ventilation has a circadian rhythm in wakefülness. NREM and REM sleep.
2. The magnitude of change in lung ventilation from wakehlness to NREM sleep is
the sarne across the day (in Figure 1-8: 1-2 = 3-4).
Fipure 1-8 shows a pictorial representation of the two hypotfieses tested in this experiment.
Lig ht Phase Dark Phase
~y pot-hesis -m.*i(
12 24 Time of Day (hours))
Rationale for the hv~otheses
There are many physiological and behavioural variables that oscillate with a 24-
hour rhythm such as sleep-wake state (Eastman et al., 1983), metabolic rate (Aschoff &
Pohl, 1970; Nagai, Nishio, & Nakagawa, 1985; Spengler et al.. 2000) and body
temperature (Eastman et al.. 1983). These specific variables are also known to influence
breathing (Mortola & Gautier, 1995). Therefore. provided the mechanisms mediating the
relationships between these variables and breathing are similar across t ime of day, lung
ventilation will also oscillate in phase with these variables to give circadian rhythm in
lung ventilation.
in this study, by rneasunng and comparing lung ventilation in a given sleep-wake
state across time of day, the masking effects of sleep-wake state on breathing can be
minimized. in this way, any time of day changes that occur in lung ventilation can be
attnbuted to factors other than sleep-wake state. Of these factors, circadian oscillations in
metabolic COz production and the chemoreflexes are likely to play a key role in
mediating circadian rhythms in lung ventilation in each sleep-wake state. As discussed
previously, there are circadian rhythms in CO2 production (e.g., Aschoff & Pohl, 1970).
time of day changes in the chemoreflexes (Peever & Stephenson, 1997; Raschke &
Mdler, 1989; Spengler et al., 2000; Stephenson et al., 2000) in addition to a close
relationship between CO2 production and ventilation (Phillipson et al.. 1981).
Furthemore, there is evidence that ventilation matches COz production at two times of
day in awake adult rats (Peever and Stephenson. 1997) and in rat pups (Saiki & Mortola,
1995). This suggests that PaCOz is similar at these two times or varies relatively little
across time of &y in a given sleep-wake state in comparison to sleep-wake changes in
PaCQ (Spengler et al., 2000). When combined al1 together, these observations (as
illustrated in Figure 1-9 below), suggest that circadian rhythms in lung ventilation are
mediated indirectly by the mechanisms underlymg the chernical control of breathing.
There may also be more direct mechanisms which involve the possibility of
neural projections fiom the SCN, the pnnciple mammalian pacemaker. to parts of the
respiratory control system such as the respiratory rhythms generator or the central
chemorecepton in the brain stem. This is a plausible hypothesis given that there are
secondary projections to the brain stem from the hypothalamus (Moore. 1 997). However.
currently. although SCN efferents have been widely mapped. their functions and the
mechanisms by which they affect other target systems are poorly understood (Piggins &
Rusak, 1999).
Figure 1-9: Hypothetical mechanisms involved in mediating sleep-wake state and time of day changes in lung ventilation in the rat (based on data from Peever and Stephenson, 1997: PhilIipson and Bowes; Lai et al. 1979) (See text below for explanation).
A Sleeii-Wake State Effects /durina the dayl
B Tirne of Dav Effects lin NREM sleepl
Fieure 1-9: Circadian variations in lung ventilation in wakefulness. NREM and REM sleep can be predicted based upon the principles underlying the chernoreflex control of breathing. PaCOz is the balance of the metabolic COz production rate and the rate at which COz is eliminated by alveolar ventilation. A PaCO: value of 30 mmHg is representative of mammals during waketùlness. On falling slrep. tonic inputs from the reticular activating system to the respiratory motoneurons are withdrawn result ing in a decline in lung ventilation. This decline in lung ventilation exceeds the decline in metabolic rate from wakefulness to sleep. Consequently. there is an increase in PaC02 during NREM sleep.
Panel A illustrates the effects of sleep-wake on lung ventilation. metabolic rate and PaCO. dunng the trouah of the metabolic rate rhythm in the a. The equilibrium PaCO. and alveolar ventilation in wakefulness (point 1) and NREM sleep (point 2) are the intersection points of the metabolic hyperbolae (at the lowest COz production rates in a 24 hour penod) (line a and line b) and the ventilatory response to COz (line c and line d) in each of these sleep-wake States. Metabolic hyperbolae show the changes in PaCO: that result from spontaneous changes in alveolar ventilation at a given metabolic rate whilst the ventilatory responses show the changes in alveolar ventilation that result from induced changes in PaC02.
Panel B shows the time of day effects on lung ventilation. metabolic rate and PaC02 in the a. For clarity. only the metabolic hyperbolae and the CO2 responses at times corresponding to the trough (during the day) and the peak of the circadian CO?
production (during the night) in NREM sleep are illustrated. However, the concepts for wakefùlness are similar. Line b and line e represent the expected metabolic hyperbolae at the 24-hour minimum and maximum metabolic rates in NREM, respectively. Line d and line f show the expected corresponding ventilatory response to COz at these times. Since metabolic rates oscillate with a circadian rhythm, a continuum of metabolic hyperbolae between line b and line e would be expected. Point 2 and point 4 represent the equilibrium values of PaC02 and alveolar ventilation at the peak and trough of the CO2 production rhythm in NREM. In order to maintain PaC02 constant in a given sleep-wake state (at an estimated PaC02 of 45 mmHg in NREM sleep), alveolar ventilation would also be expected to oscillate in phase with the circadian rhythm in CO2 production. The predicted values of alveolar ventilation across time of day would be expected to lie on a vertical line between point 2 and point 4. Assuming ventilation of dead space in the lungs is kept constant, lung ventilation would be expected to be proportional to a given level of alveolar ventilation. Point 2 and point 4 in this figure would be expected to correspond to point 2 and point 4 in Figure 1-8.
Having discussed the theoretical basis of the present experiment, attention will
now be tumed towards testing the hypotheses. The primary objective of the experiment
was to measure breathing, metabolism and body temperature in all the three sleep-wake
states (wakefûlness, NREM and REM sleep) across a 24-hour penod in freely behaving
rats. in order to generate sufficient data in al1 three states across a 24-hour period, rats
were studied because of their naturally occumng polyphasic sleep-wake cycles (see
Figure 1-10) (unlike adult humans whose sleep is consolidated to a period of
approximately 8 hours during the night). That is, even though they are classified as being
noctumal, their sleep is not confined solely to the day but also occurs during the night
when they are predominantly awake and more active.
Lung ventilation and metabolism were measured using non-invasive closed
system whole body plethysmography. Sleep-wake state was measured electrographically
using surgically implanted cortical EEG and neck EMG electrodes attached to a telemetry
unit. Body temperature was used as a measure of the circadian phase of the rats. The
generated data when plotted as a hinction of tirne of day was then tested for the presence
or absence of circadian rhythms.
Fipure 1-10 illustrates the polyphasic nature of rat sleep on a hypnogram of typical rat sleep- wake behaviour across a 24-hour period. Hypograrns show the occurrence and duration of a particular sleep-wake state in a span of time.
(Source: Trachsel et al., 1988)
Tirne after light onset (hours)
O 12 24
REM 1
Chapter 2: Methods Experimental arotocol
Expenments were performed on six male Sprague Dawley rats (mean * SEM
body mass: 3591 19%). Surgery was performed and the rats were allowed to recover on a
12 hour light - 12 hour dark cycle. Durinp this recovery phase. the rats were allowed to
habituate to the experimental apparatus for shon (4 hour) periods of time. About 24
hours before the start of the experiment. parameters of an on-line computerized sleep
detection system were set to aid the expenmenter in making decisions of the sleep-wake
state across the 24-hour day (see :kfeasrtrernent of' sleep-wake stares atzd bodv
temperature). At least 10 hours before breathing and metabolism measurements were
made. the rats were placed in the animal chamber (see Figure 2-2) which contained corn
based rnaterial mixed with bedding from the home cage to reduce the novelty of the
situation. Food and water were freely availablr. Recordings lastrd 24 hours foollowing the
10-hour farniliarization interval. Each rat was studied twice with at least 24 hours
between experiments.
Animal preparation
The rats were implanted with a three-channel radio transmitter (TL 10 M-3 F-50
EET - Data Sciences International) connected to EEG and EMG elcctrodes (see Figure 2-
1) as follows:
hzesthesia und oreparution: Surgery was performed under gencral anesthesia and using
sterile conditions. Anesthesia was induced by an intra peritoneal injection of ketamine
35
(8 5mgkg): xylazine ( 1 5 mgkg). As the sugery progressed. the inhalation anesthetic.
halothane (typically between 0.1 - 2.0%) was
Prior to rnaking any incisions. the rats
buprenorphine (0.0 1 -O.O5mg/kg) as an anal
administered in O? enriched air as required.
were also intra-peritoneally administered
gesic. atropine ( I mgkg) to prevent upper
airway secretions ftom blocking the aimay and 3ml of 0.9% sterile saline for hydration.
Subsequently. the head and abdominal areas were shaved and stenlized with 70% alcohol
and betadine. The telernetry unit was soaked in 0.9% glutaraldehyde for 12 hours
followed by at least one hour sterile saline for sterilization.
Irn~lantation of' ihe telemetv mit: After the pedal withdrawal reflex was no longer
present. a 3-cm incision was made in the skin on top of the skull. The rat was then placrd
in the supine position. an midline incision was made in the skin overlying the abdomen
and another incision was made along the linea alba to expose the peritoneal cavity. The
transmitter was loosely sutured to the rectus abdominus muscle using 3-0 non-absorbable
silk. The leads from the telemetry unit were pushed through a puncture made in the
abdominal muscle and subcutaneously tunneied to the incision in the head region. The
abdominal incisions of the muscle layer and the epidrrmai layer were then sutured using
3-0 vicryl absorbable suture.
Imolantation ot'EhfG and EEG electrodes: After the rat was placed in the prone position.
its head was stabilized using a Kopf stereotaxic frame. A 3-cm midline incision was
made in the muscle layer on the top of the skull. The surface of the skull was cleared of
connective tissue and then swabbed with hydrogen peroxide to establish a clean surface
on which dental acrylic could set.
The incision in the skin on top of the skull was extended caudally, the underlyng
layer of the dorsal neck muscle was exposed and the two EMG electrodes were sutured
into the splenius muscle on either side of the midline using 3-0 non-absorbable silk. Each
EMG electrode was anchored to the splenius muscles with a caudally placed suture and
covered with a silastic sheath to prevent irritation or injury to the muscle. This sheath was
held in place by another suture to the splenius muscles. Next, three holes were drilled into
the skull to house 0-80 x 118" stainless steel screws. Two were used for EEG electrodes
and one served as the ground (reference) electrode. The first EEG electrode was located 2
mm anterior and to the lefi of bregma. and the second. was placed 2 mm posterior and to
the right of the first EEG electrode. The ground rlectmde was situated 2 mm anterior and
to the right of bregma. The tips of the EEG and ground leads From the telemetry unit were
placed in their respective holes and screwed in with 0-80 x li8" stainless steel screws.
Fipure 2-1: X-ray of a rat implanted with EEG and EMG electrodes, temperature sensor and biotelemetry unit barometric plethgsmograph.
Dental acrylic was poured over the screws to firmly anchor the electrodes to the skull and
to provide electrical insulation. The incision on the head was closed with 3-0 absorbable
vicryl and the rats were allowed to recover. During recovery, the incisions were checked
for signs of infection and the general state of health of the animal was monitored.
Experimental Apparatus
Barometric plethysmography cornbined with biotelemetry was used to measure
breathing. metabolisrn and sleep in freely behaving rats (see Figure 2-1 and Figure 2-2).
Figure 2-2: Schematic of the experimental apparatus used to measure lung ventilation, metabolism. body temperature and sleepwake states.
1 1 1 1 Acquisition of Breathing
1mYI and Metabolism signals
CO2 and 0 2
Air In
Chartes' Law Pressure is proportional to Temperature (when volume is constant).
23' C - During inhahtion, . . air is warmed and
humidified $\ F in lungs * L' LA pressure in
?-$L animal chamber i (constant volume)
rises. -$ Reverse occurs in N Y exhalation.
EEG, EMG, Body Temperature Signai Receiver and Processor 1
Measurement o f /un p ventilation
In whole body plethysmography. the rat is placed in a constant volume container
attached to a very sensitive differential pressure transducer capable of detecting small
changes in pressure associated with breathing (Drorbaugh & Fenn. 1955: hlortola &
Frappell. 1998). When air is inhaled. it is warmed and humidified in the lunp. Since this
heating of the gas occurs in a container of constant volume. by Charles law (pressure a
temperature). there is an increase in pressure in the chamber. When air is exhaled from
the lungs the opposite occurs. That is. the warmer. more humid air of the lungs is cooled
and condenses. leading to a decrease in pressure. These pressure changes when
transduced into an electrical voltage form the basis of the calculation of tidal volume (VT)
by the following equation (Drorbaugh & Fenn. 1955):
VT ~BTPSI = Pm * Vca~ * Ga
Pal
whrre. Pm is the voltage change caused by respiratory related pressure changes in the
chamber. Pd is the voltage change in the pressure transducer caused by injection of a
known volume of gas (Val) and Ga is a dimensionless constant that represents the ratio of
the tidal volume at alveolar conditions (VT.\) to the small increase in volume that occurs
when tidal gas expands from its volume at chamber conditions (VTC) . i.r.. Ga = VTA
It was calculated using the equation:
Ga = T b (PB'PcH~o)
Tb( PB -PCH?O)- TJ'B-P~HIO)
where. Tb and Tc are the alveolar and chamber temperatures (in Kelvins). respectively. PB
is the barometnc pressure. and PbHZo and PCH~O are the water vapour pressures of the gas
in the alveoli and the chamber, respectively. PCWO was calculated as follows:
Pc~ ro = S WVP, xRH
1 O0
where, SWVPc is the saturated water vapour pressure at the chamber temperature. Tc. and
RH is the relative humidity (%) in the chamber.
Respiratory tiequency (0 in breaths min" was calculated as:
f = 6 0
TOT
where. t ~ o ~ is the total breath duration calculated by summing inspiratory time (ti) and
expiratory time (tE). That is. TOT = t~ + (E.
Lung ventilation ( ? 1, ml min". BTPS) was calculated by:
i',=t-IV*
The use of barometric plrthysmograp hy to measure lung ventilation non-
invasively in small animals and infants has been widespread (Monola and Frappell.
1998). Good correlations between baromctric plethysmography and other rstablished
techniques to measure breathing such as pneumotachography. have lent validity to the
measurement of breathing using this technique (e.g. Drorbaugh and Fenn. 1955: Stahel
and Nichol. 1988; Siefert et al. 2000).
The barometnc plethysmograph (see Figure 2-2) used in our experiment consisted
of a 9 litre animal chamber and an identical reference chamber placed in a water bath.
The purpose of the reference chamber and water bath. respectively. was to shield the
apparatus from potentially disruptive changes in ambient pressure and temperature.
Pressure tluctuations associated with the rat breathing in the animal chamber were
measured using a differential pressure transducer (Validyne Engineering: Mode1 DP-45-
14) attached to outlets fiom the animal and reference chambers (see Figure 3-1 for
representative traces of breathing during wakefulness, NREM and REM sleep). The
relative humidity and temperature of the animal chamber was measured by a
thermohygrometer (Cole Parmer: Model 37950- 10. resolutions of 0.1 OC and 0.1 % for the
t hermometer and hygrometer, respect ively).
The analog signals from the differential pressure transducer and the
thermohygrometer were fed into and recorded on an eight-channel data acquisition
system ( ADinstmments: Mac Lab 8/s) driven by a Macintosh computer (Performa
5200CD). The same data acquisition system was used to process and record the signals
from the CO? and O? analyzers and the body temperature sensor described below (sec
Measuremenr of metabolic rate and Me~srirentenr of' sleep-wake srare and bo&
temperarure below).
To make measurements of breathing. three steps were followed. First. the chan
record (Grass Lnstmments: Model 78D) was switched on at a speed of 5 mmis to get a
paper record of the sleep-wake state. Second. the animal chamber was completely sealed
using a solenoid valve (Cole Panner: Model 01367-97. 118" x P4". 12 VDC) that
simultaneously closed both the inlet and outlcts. Third. the rate of data sampiing on the
data acquisition system was increased to 100 samples/sec from 4 samples/second to
prevent any distortion of the respiratory signal by alaising duc to inadequate sarnpling
frequencies. When the concentration of CO2 concentration in the animal chamber had
built up to approximately 0.5% (approximately 2-5 minutes since chamber closure
depending on sleep-wake state of the animal). the solenoid values were opcned to resume
airtlow to the animal chamber. When the animal chamber was open. the sampling rate
was reduced and the chart record switchrd off. Over the course of a 24-hour period. the
chamber was closed to make the required measurements between 50 to 80 times.
Fipure 2-3 shows the mean concentration of COr (O/') in ambient air in each of the analyzed portions of data in each sIeep-wake state.
A
0.8 WAKE - - NREM -
REM - - - - - - -- - --
O - - . - . . . 0.4 - - .-. - - -. r- -
I -. - -. .
m 2. g2< L=$$g i g s C .. 5. , -. - - Cr.
0.2 - - - rr
Time since light onset (hours)
However. not al1 these data could be used due to reasons such as activity leading to
movement arti fact and insufficient durat ion in a given sleep-wake state. Throughout the
expenment. the COz concentration in the ambient air in the chamber was maintained
below 0.65%. However. the mean CO2 concentrations of the analyzed data were
considerably Iower and random across the 24-bour experiment as show in Figure 2-3
below.
The 24-hour mean k standard error of the mean (SEM) of inhaled CO2 dunng
measurements were 0.3 1 f 0.0 1% in wakefulness. 0.29 k 0.01% in NREM sleep and 0.28
f 0.01% in REM sleep. These ambient COz concentrations during each sleep-wake state
were not significantly different (p>0.05) fiom each other and the variance in ventilation
explained by inspired CO, concentration were negligible (see Figure 2 4 below: i = 0.08
in wakefulness. ? = 0.03 in NREM sleep and ? = 0.00 in REM sleep). Therefore. there
was no ventilatory response to CO2 at the low CO? concentrations present in the animal
chamber. Furthemore, the random ambient CO? concentrations across the day prompted
us to assume that the effects of CO2 concentrations on lung ventilation across the day
were negligible.
During the times when ventilation was not recorded, air was constantly flushed
through the animal chamber at a flow rate of approximately 3 litres per minute to supply
k s h air and prevent build up of CO? in the chamber.
Fipure 2-4: Graph of the deviation from the 24hour mean lung ventilation plotted versus inhaled CO2 concentration ( O h ) in wakefulness. NREM and REM sleep. The R' between these two variables in each sleep-wake state is also shown.
- WAKE
REM
r = 0.08 (Wake) r * = 0.03 (NREM)
r = 0.00 (REM)
lnspired CO2 (%)
Measurement o f metabolic rate
Metabolic rate. as indexed by the rate of CO, production (mumin). was estimated
by measuring the rate of accumulation of CO? in the animal chamber during the brief
intervals when the inlet and outlets of the animal chamber were closed. Air in the animal
chamber was continuously sampled for analysis of fiactional CO? and O? concentration
(%) and retumed to the animal chamber in a closed circuit. The circuit consisted of a gas
dner (active ingredient anhydrous calcium sulphate). a roller pump (Masterflex: Model
US Drive with US Easy load II pump head). pressure dampers and high resistance (26G
needles) to reduce pressure fluctuations associated with the roller pump. an
electrochemically based O? analyzer (Amtek: Model S-3Nl) and an inka red based CO2
analyzer (Amtek: Model 3D-3A) al1 connected in senes. The roller purnp circulated air
around the circuit at a tlow rate of approximately 50 mljmin. Before the start and
approximately 12 hours into the rxpenment. the COI and 0: analyzers were calibrated
with 2 1.45% 0: and 5.09% CO? gas. By injecting a 60ml bolus of pure CO: gas into the
empty chamber. the lag tirne for a change in the CO2 concentration in the charnber to be
detected by the CO? analyzer was approximately I O seconds.
The formula used for calculating metabolic rate (Y CO2. ml. min". STPD) was:
Y CO2 = d[COr]. Vc - dt
where. d[C02] is the rate of change of CO2 concentration and V, is chamber volume. - dt
The change in CO, concentration per minute was denved from the dope of a linear
regession line fitted through fractional concentration of COz plotted over a time penod
ranging From 1-2 minutes. The volume of the expenmental chamber was determined by
injecting a known volume of pure CO? gas into the chamber (without an animal present).
recording the changes in fractional concentration of COz (%) and substituting into the
following equat ion:
Volume of chamber = Volume of pure CO2 injected into c hamber
Change in fractional CO2 concentration
In the presence of the rat. body density was assumed to be 1 .O3 @ml (Stephenson. 1993)
and v,' was calculated as:
v,' = V, - Mb - 1 .O3
where Mb is the mass of the rat (g) and v,' is the volume of gas in the animal chamber
with the rat present inside it.
Although the ambient concentration of O? was measured. an estimate of the
metabolic rate using the rate of uptake of 0: was not calculated because of the low
resolution and highly noisy. inadequately filtered signal. Throughout the experiment. the
ambient inhalrd Oz concentration was maintained at normoxic levels (between 20.0% and
Measurernertt of'S1ee~-wake stares arid B o d ~ ternRerarrrre
Sleep-wake states were scored visually using the EEG and neck EMG signais
recorded on chan. The signals from the cortical EEG electrodes and neck EMG
electrodes were transduced by the three-channel telemetry unit (Data Sciences: Mode1
TL I OM3-FSO-EET) and emitted as radio waves. A thermal sensor (resolution O. 1 O C . 90%
(mean f SEM) response time of 129 t 9 seconds) inside the telemetry unit measured core
body temperature (Tb). Althouph the thermal sensor allowed circadian changes in body
temperature to be measured. the response times were too slow to permit shorter-term
changes in body temperature associated with changes in sleep-wake state to be
determined. The radio signals from the telemetry unit were picked up by a receiver (Data
Sciences: Model RPC- 1 ) under the plethysmopph. and then fed into a digital to analog
converter (Data Sciences: Model UA-10 Universal Adapter) which also amplified the
analog signal 1000 times. The analog signals were Funher amplified six and a half times
using a buffer amplifier (CWE Inc: Model DC-936 Buffer) (for a total of 6250 times) and
filtered (EEG signal: O. 1 - 100Hz. EMG signal: 3-1OOHz) before being fed into a chan
recorder to give a paprr record (speed: j r nds ) of the EEG and EMG. The analog signals.
afier conversion to digital signals (at a sampling rate of 300 Hz). were also sent to a
separate computer system (IBM Compatible 386. 16 MHz) that auiomatically determined
sleep-wake state by analyzing the fiequencies of the EEG and the amplitudes of the EMG
signals (Hamrahi. Chan. & Horner. in press).
The analysis of the EEG. in 6-second rpochs. used the interval histogram rnethod
(Kuwahara et al.. 1988) to determine the percent of the signal in each of the following six
bandwidths: &(OS-2Hz). &(24Hz). O(4-7.5Hz). a(7.5- l3.5Hz). P i ( 1 XWOHz) and
p2(20-3OHz). In this method which is software dnven. the amplitude of the EEG signal
was divided into 32 equally spaced horizontal slice lines. A period (reciprocal to
frequency) was rneasured as the time interval between two points at which the same slice
line crossed consecutive positive-going slopes of the EEG signal (Kuwahara et al.. 1988).
A histogram was then constnicted for these intervals. and from this histogram. the
percent distribution of the frequencies was calculated.
Additionally. the %P2/%8, and EEG and EMG amplitudes were determined. From
these signals. the computer made a judgment of the sleep-wake state using an algorithm
based on the ratio of %Pz/%& and EMG amplitudes (Homer et al., 1998; Hamrahi et al..
in press). in this algorithm, wakefulness and REM sleep was differentiated from NREM
sleep by a higher %P2/%& in wakefulness and REM sleep compared to NREM sleep.
That is. ftequencies of the EEG are higher in wakefulness and REM sleep compared to
NREM sleep (also see Figure 3-2 in RESULTS). Having similar %PL/%& magnitudes.
wakefulness and REM sleep were differentiated by the absence of muscle tone and low
EMG values in REM sieep. This algorithm has previously been validated in rats by
Hamrahi. et al.. (in press).
Prior to the stan of the experiment. the tidelity of transmission of the EEG and
EMG signals were tested by inputting a 0.5V amplitude. lOHz sine wave generated from
a signal generator into the telemetry unit. and. recording and verifiing the output on the
sleep computer and chan record. The resulting output on the sleep computer indicated
that 99.8% of the signal was in the a(7.5-l3.5Hz) range as expected. After implantation
of the transmitter in the rat. the %P2/%tii and €MG thresholds were then set by
repeatedly observing these parameters in al1 three sleep-wake States Wr approximately 5
hours. During the experiment. the computer judgments were used only as an aid to
determining the sleep-wake stata of the rats across the day. In this respect. since
rxperiments were 21 hours long and drowsiness on the pan of the experimenter was
inevitable. the computer assisted sleep scoring proved invaluable. ARer completion of the
experiment. the chan records of the EEG and EMG were used as definitive in assessing
sleep-wake state before breathing and metabolism measures were made.
Data Analysis
Selection of ~or t ions of data suitable for anaiysis and calculations performed
Thiny-second periods of established wakefulness. NREM and REM sleep were
selected From the chan records to give a sampie of at least 30 breaths in each of these
sleep-wake states. These selected episodes occurred at least 20 seconds afier wakefulness
or sleep onset. and. at least 20 seconds afler the chamber was closed. These critena were
set to ensure that the rats were in established sleep-wake states and to avoid measuring
breaths that were associated with a minor arousal that may have occurred due to changes
in the noise level in the chambrr. When the rats were first placed in the chamber. it was
observed From the EEG that the rats were startled by noise of the solenoid valve used to
close the chamber. However. aRer the rats were regularly errposed and habituated to these
noises during the 10-hour familiarization period. arousal from the operation of the
expenmental apparatus was no longer an issue at the start OC and duing the experiment.
For eac h of the selected episodes. the corresponding breat hing. metabolic rate.
body ternprrature and EEG and EMG related data was determined. For each breath in the
inspiratory time (ti). expiratory time (tE), total breath duration (tTor) were calculated.
Pressure signals deemed to be the result of movement artifact. sighs or sniffs were
omitted from the analysis. In each sleep-wake state for each intervention. the mean value
of these parameters from the selected breaths were then calculated and used in
subsequent analyses. In addition. the coefticient of variation (CV) for tidal volume.
frequency and lung ventilation was computed to give an estimate of the variability of
breathing. The equation used to calculate the coefficient of variation for each selected
episode was:
Coefficient of Variation = Standard deviation x 100
Also for each selected episode, the mean % of the EEG frequency in the 6, and pz
bandwidths. the %Br to %&ratio (%P2/%6i) and the mean EEG and EMG amplitudes
were computed.
Analvsis of Circadian Rhvthms in Breathin~. Metabolism and Bodv Temperature
In each sleep-wake state. the presencc or absence of circadian rhythms in lung
ventilation. respiratory frequency. tidal volume (and their associated coefficients of
variation), CO2 production. ventilation normalized for COz production and body
temperature was detemined by tïtting a sinusoidal curve through the normalized data set
for each variable.
Pre~ararion of' data for model fitting: So that the data from al1 animals could be
combined. prior to fitting a sine wave model. the inter-animal variability in the mean
value of eac h parameter was removed throuph normaiizat ion. This process of nonnalizing
the data for each rat in day 1 and day 2 occuned in two steps. First. a 24-hour mran
(rnranrhu) was calculated. Second. each individual data point was normalized and
expressed as a drviation from the rneanldb using the equation: Normalized value =
Value - rneanirm. These normalized data were then replotted as a hnction of time
before fitting a sine wave model to the data.
The normalized values for both days I and 2 of the enperiment were combined
afler no differences in mean lung ventilation were found between day 1 and day 2 of the
expenment (F[1. 231 < 1, p=0.98 using a 2 way Repeated Measures ANOVA, with sleep-
wake state (Wakefulness. NREM sleep. REM sleep) and day (Day 1, Day 2) as factors.
Model: The general equation of the fitted model was:
y = yu + asin(2iidb + c)
where. y is the physiological variable. y0 is the mesor (24 hour fitted mean value level.
equal to O for time adjusted deviations). a is the amplitude. x is the variable time, b is the
period set at 24 hours. and c is the acrophase - Il12 radians. The assumption that the
period of the sinusoidal function was 24 houn was based on the fact that the rats were
rntnined to a 12: 12 light-dark cycle.
A computer program (Jandel Scienti fic: Sigmaplot 4) was used to f i t the model by
regression analysis of variance. In this process. an algorithm that minimized the
deviations between the fitted model and the actual data (by the Least Squares method)
was used. In addition to fitting a sine wave model to the deviations of the data. Sigmaplot
4 also calculated the probabilities that the fitted mode1 possessed a circadian rhythm. A p-
value of <O.Os was accepted as statistically significant.
.-lssessirw differences in acro~lzases
The phase relationships between the acrophases of the circadian rhythms in lung
ventilation. CO? production. ventilation normalized for CO2 production. body
temperature and respiratory frequency in each sleep-wake state were assessed using t-
tests from acrophase values and their standard erron generated fiom the titted model. The
equation (Ott. 1993) used was:
t = meanl - mean?
d (SEM,? + SEM??)
degrees of freedom = ni +n? -2
where mean and SEM (standard error of the mean) are in hours and nl and n? are the
number of samples used in the calculation of mean, and mean? respectively.
Determininp the mamitude of change in lunp ventilation from wakefulness to
'IREM across time of day
To test hypothesis 2. the mean change in lung ventilation from wakefulness to
NREM slerp from two 5-hour time periods in the day (light phase) and night (dark phase)
were compared using a 2 way Repeated Measures ANOVA with sleep-wake state
(wakefulness. NREM sleep) and time of day (day. night) as factors. Thesr windows of
data corresponded approximately to the peak and trough of the temperature rhythm and
began 2 hours afier lights on and lights off. respectively.
Additional analvses performed to determine possible causes of time of dav chanees
in lunp ventilation
In addition to companng the magnitude of change in lung ventilation from
wakefulness to NREM sleep during the day and night. other variables such as CO2
production. ventilation nonnalized for CO: production. body temperature. the ratio of
high to low EEG fiequencies (%Pz/%&) and EEG and EMG amplitudes were compared
ar two times of day in wakefulness and NREM sleep only. REM sleep was not included
in this analysis because of a lack of availability of data in this sleep-wake state in some
rais at some times of the day. These means were then compared using a 2 way Repeated
Measures ANOVA with the factors being sleep-wake state (wakefulness. NREM sleep)
and time of day (day. night). If significant differences were found using the ANOVA.
Tukey's post hoc test was used to identib the source of these differences.
Chapter 3: Results
General observations on behaviour and slee~mwake ~atterns
As the experiment was conducted. the rats did not appear to be disturbed by the
experimental apparatus. Although no forma1 quantitative analysis was conducted on their
sleep-wake cycles. the distribution of measurernents made in eac h sleep-wake state were
retlective of the ultradian rhythms in rats sleep-wake cycles (see Figure 3-2). In this
regard. there was an uneven distribution of each sleep-wake state across a 24-hour period.
The rats generally slept more during the light phase (day) and generally were more awake
and active during the dark phase (night). When asleep. REM sleep was more predominant
towards the end of the light phase. but was not very frequent during the dark phase. These
behaviour patterns across each 24-hour experirnental day in addition to the circadian
body temperature rhythms (see Figure 3-5 row D) are characteristic of freely behaving
rats entrained to a IZ hour light-dark cycle.
The computer based EEG and EMG analysis was consistent with EEG and EMG
traces used in visual scoring methods (see Figure 3-1). Three examples from the
combined data can be used to illusrrate these consistencies. Firstly as shown in Figure 3-2
panel A. the mean of the ratio of high to low EEG frequencies (%P2/%61) was higher in
wakefulness and REM sleep compared with NREM sleep (F[1.23) = 9.0. p=0.03).
Secondly. the EEG amplitudes as illustrated in Figure 3-2 panel D. were greater in
NREM sleep t han in wakrfùlness and REM sleep (F( 1.23) = 60.0. pc0.00 1 ). Thirdly. the
neck EMG amplitudes were highest in wakefulness followed by NREM sleep. and thrn.
REM sleep (F[1.23] = 132.1. p<0.00 1) (see Figure 3-2 panel E). Although there are some
higher EMG values in NREM sleep. these were due to the effects of posture on muscle
tone.
Shortly after the lights came on. the rats exhibited short bouts of "deep" sleep that
showed up on the EEG as low frequency. high amplitude waves. As tirne progressed.
visual observation of the EEG showed that the amplitude was relatively similar. but the
percentage of lowest frequency bandwidth. delta 1 (%& ) decreased exponentiall y
throughout the light phase from its maximum. shortly at the beginning of the day (see
Figure 3-2 panel C).
On average. alter combining the two 24 hour periods of experiments for each rat.
the mean I standard error of mean nvmber of measurements per rai were 32.1 I 3.0 in
wakefulness. 15.0 f 1.7 in NREM sleep and 15.0 + 1.6 in REM sleep. Aççompanying
each of these measurements was a paper record of the EEG. EMG and breathinp traces.
Representative traces retlecting the breathinp and the EEG and EMG in wakelulness.
NREM and REM sleep are shown below in Figure 3-1.
Figure 3-1: Raw EEG. EMG and breathing traces in wakefdness. NREM and REhI slee~.
Fieure 3-2: Ratio of high (%p2) to low (%61) EEG frequencies and the EEG and EMG amplitudes in each of the analyzed portions of data in wakefulness, 'IRE31 and REM steep (n=6).
NREM REM -
Time since light onset (hours)
in this chapter, the results of the analyses conducted on the data in this experiment
are presented in four sections. Section 1 presents data to answer the questions posed by
the hypotheses stated in the Introduciion. Section 2 provides the results of additional
circadian analyses performed on variables such as CO? production. ventilation
norrnalized for COr production. body temperature and variability in breathing. Using
combined data across the 24-hour experimental durations. Section 3 provides data to
study the effects of sleep-wake state on breathing and metabolisrn. Finally in Section 4.
using data from the peak and trough of the body temperature rhythm. additional day-night
comparisons of certain EEG and EMG data and metabolisrn were conducted to make
funher inferences on possible mechanisrns involved in mediating the circadian rhythms
in lung ventilation in each sleep-wakr state.
Section 1 :TESTING HYPOTHESIS 1 and 2
a) Hypothesis 1: Circadian rhythrns in lung ventilation in wakefulness, NRELM and
RE 34 sleep
Figure 3-3 row A shows the circadian rhythms in lung ventilation that were found
in wakefulness (F[3.192] = 1 1.3. p<0.0001). NREM sleep (F[3,269] = 8.9, p<0.0001) and
REM sleep (F13.891 = 9.7. p<0.0001). The prak of these circadian rhythms occurred
during the dark phase as indicated by the mean acrophase (for al1 three sleep-wake states)
of 17:29 hours (1 7 hours 29 minutes) after lights on. The amplitudes of these rhythms
were 7.7%. 6.8% and 10.1 % of their corresponding 24-hour mean values in wakefulness.
NREM and REM sleep. respectively.
The 24 hour rhythms in lung ventilation in waketùlness and NREM sleep were
due to circadian rhythms in respiratory frequency (wake: F[3.192] = 8.6. p<0.000 1.
NREM: F[3.269] = 2 1.3. p<0.000 1). rather than tidal volume (see Figure 3-3 rows B and
C). In contrast. there was no significant circadian rhythm in respiratory frequency during
REM sleep (F[3.89] = 1.8. p=0.16). The acrophases of the 21 hour rhythms in respiratory
frequency in wakefulness and NREM sleep were 18:14 t 1:11 and 17:37 f 0:14 hours
aRer light onset. respectively. and the amplitudes were 7.2% and 7.5% of the mesor
values in wakefulness and NREM sleep, respectively. Tidal volume did not possess a
circadian rhythm in al1 three sleep-wake states (wake: F[3. 1921 < 1. p4.O. NREM:
F[3,269] < 1. p=O.jO and REM: F[3.89] < 1. p=0.51) (see Figure 3-3 row C).
Fipure 3-3: Graphs showhg the circadian rhythms in lung ventilation (v 1) in wakefulness. NREII sleep and RESI sleep (row A) obtained by fitting a sine wave through plots of lung ventilation across 24-hour periods. Row B and C show the components of these 24-hour rhythms in lung ventilation - tidal volume (VT) and respiratory frequency (f) (n=6). The presence of a circadian rhythm is marked with an asterisk
NREM - *
REM -
Time since light onset (hours)
b) Hypothesis 2: The magnitude of change in lung ventilation from wakefulness to
YREM sleep at two times of day
The data that were used to test hypothesis 2 consisted of two five-hour windows.
beginning 2 hours afler lights on and off (frum both experimental days). to give an
estimate of the lung ventilation in wakefulness and NREM sleep dunng the light phase
(referred to as "day") and dark phase (referred to as "night"). REM sleep was not used in
this day night cornpanson because of the lack of REM sleep data for one rat dunng the
dark phase.
A 24.2% (91 ml min") decline in ventilation occurred from wakefulness to
NREM sleep during the day and a similar decrease of 23.6?6 ( 10 1 ml min") was observed
during the night phase (F[ 1.23 j < 1. p=0.58) (see Figure 3-4 below).
Fipure 34 : Lung ventilation during two times of day in wakefulness and NREM sleep and the magnitude of change in lung ventilation from wakefulness to NREM sleep during the day and night (n=6). * indicates a significant tirne of day effect (p<O. 05) and t indicates a significant sleepwake state effect (p<O.OS)
01 5 a
200 Day Night Time of Day
Time of Day
D ~ Y Night
Section 2: AûDlTlONAL CiRCADlAN RHYTHMS ANALYSES
a) Body temperature. COt production and ventilation normalized for CO2
production
in addition to circadian rhythms being obsewed in lung ventilation in
wakefulness. NREM and REM sleep. circadian rhythms in these three sleep-wake States
were also found in CO2 production and body temperature. Ventilation nomalized for
CO? production had a significant 21-hour rhythm in NREM sleep. but not wakefulness or
in REM sleep (see Figure 3-5. Table 1 and below for funher details).
CO: production had a circadian rhythm in wakefulness (F[3.192] = 14.2.
p<0.000 1). NREM sleep (F[3.268] = 41.9. pc0.0001) and REM sleep (F[3.84] = 7.2.
p=0.0003) (see Figure 3-5 row B). The amplitudes of these rhythms were 9.0%. 9.6?/0 and
10.8% of thrir 21-hour mesor values in wakefulness. NREM sleep and REM sleep
respectively. The mean acrophase of these rhythms was 1858 hours after Iight onset (see
Figure 3-6).
Whrn ventilation was normalized foi CO2 production. a circadian rhythm was
detected only during NREM sleep (F[3.268] = 5.7. p=0.0008). but not during
wakehlness (F[3.192] c 1. p= 0.53 or REM slerp (F[3.84] < 1. p = 1.00) (see Figure 3-5
row C). The acrophase of this rhythm was 9 3 2 hous aller light onset (see Figure 3-6)
and the amplitude was 2.5 (5.9% of the mesor). There were no intluential data points in
this data set (al1 data points had Cook's D test values of < 1 ). When the acrophases of the
ventilation and CO? production rhythm in NREM sleep were compared. there were no
signi ficant di fferences (p>0.05). indicating that these two rhythms in NREM sleep were
in phase.
Al1 rats possessed a circadian body temperature rhythm in wakehlness (F[3,192]
= 68.4. pc0.000 l), NREM sleep (F[3.269] = 58.5. p<0.000 1) and REM sleep (F[3,89] =
8.9, p<0.0001) (see Figure 3-5 row D). The acrophase of these rhythms in al1 three sleep-
wake states occurred approxirnately 17 houn alter light onset (see Figure 3-6). There
were signi ficant di fferences between the amplitudes of t hese temperature rhythms. with
the highest amplitude being in wakefulness (0.6'C) and the lowest in REM sleep (0.3OC)
(see Table 3- 1). However. as mentioned before. there were no differences between the
24-hour means of the temperature rhythm in wakeîùlness. NREM and REM sleep. This
was likely due to a combination of the slowly changing nature of the body temperature
signal in comparison to the fairly rapid changes in sleep-wake state and the relatively
slow response time of the temperature sensor. Thus. the acrophase values obtained are
more likely to reflect the overail acrophase of the body temperature rhythm rather than
the acrophase of this variable in each sleep-wake state.
b) Phase relationships between variables
Figure 3-6 shows the acrophases of lung ventilation. COi production. ventilation
normalized for CO? production. body temperature and respiratory frequency in each
sleep-wake state. There were no differences in the acrophases of circadian rhythms of
lung ventilation. COI production. body temperature and respiratory tiequency in each
sleep-wake state (al1 pO.05) That is, these rhythms were al1 in phase. There were also no
differences in phase for each of the above variables across sleep-wake statcs (al1 p>0.10
using t-test). In pneral. the peak of the lung ventilation. CO: production and body
temperature peaked around the middle of the dark phase in al1 three states (in the range of
16:44 - 19:36 hours after lights on).
One variable that did show a significant phase difference between the rest of the
variables was ventilation normalized for COr production (Y [/Y COz) in NREM sleep
(p<0.05). The acrophase of this circadian rhythm occurred approximately 8 hours in
advance of the acrophases of the rhythms in lung ventilation. CO? production. body
temperature and respiratory frequency.
Figure 3-5: Graphs showing lung ventilation (row A), CO2 production (row B), ventilation normalized for COz production (row C ) and body temperature (row D) as a function of timc of day in wakefulness, XREM and REM sleep (n=6). The presence of a circadian rhythm is s h o m using an asterisk (*).
WAKE 500 -
NREM REM -
Time since light onset (hours)
Fipure 3-6: Phase relationships between the acrophases of the circadian rhythm in lung ventilation, COz production, ventilation normalized for COZ production, body temperature and respiratory frequency ( n 4 ) . Each point indicates the mean acrophase and the error bars represent standard error bars. (* indicates that there was a statistically significant ( ~ 4 . 0 5 ) difference between ventilation normalized for CO2 production in NREM sIeep and the other variables listed).
Lung Ventilation 1 (mllmin) 1
1
: WAKE I NREM .$ REM
CO2 production I l
(rnllmin) O
Ventilation nonnalised for '
CO2 production I l
Body I
Temperature (OC) 1 1
I
Respiratory Frequency
(breathslmin) j
O 6 12 18 24
Time since light onset (hours)
c) Variability in lung ventilation, respiratory frequency and tidal volume
Though there were circadian rhythms in lung ventilation in wake fulness. NREM
and REM sleep. no circadian rhythms in the coefficients of variation (CV) for lung
ventilation were found in al1 three sleep-wake States (wake: F13.1921 < 1. p=0.41.
Coefficients of variation for respiratory frequency were arrhythmic in wake (F[3.192] <
1. p=0.98) and REM (F[3.86] 4. p=0.5). However. for NREM sleep. the CV for
respiratory frequency was rhythmic (F[3.269] = 6.9. p=0.0002) (see Fipre 3-7 row B).
No circadian rhythms in CV for tidal volume were observed (wake: F[3.192] c 1. p=0.37.
NREM: Fi3.2691 < 1. p= l .O and REM (F[3.86]4. p=0.84) (see Figure 3-7 row C).
Firure 3-7: Breath to breath variability in lung ventilation (t CV). respiratory frequency (f CV) and tidal volume (VT CV) in wakefulness, NREM and REM sleep across a 24-hour period (n=6). The presence of a circadian rhythm is indicated with an asterisi(*).
Time since lig ht onset (hours)
Section 3: ADDITIONAL ANALYSES on EFFECTS of SLEEP-WAKE STATE
In the analyses conducted in this section. the 24-hour means of each parameter in
wakefulness. NREM and REM sleep were compared using a One Way Repeated
Measures ANOVA.
a) Lung ventilation, respiratory frequency, tidal volume and their assaciated
coefficients of variation
Sleep-wake state (wakefulness, NREM and REM sleep) had significant effrcts on
the 24-hour means in lung ventilation (F[Z. 171 = 40.2) (see Figure 3-8 panel A). The
differences in lung ventilation across sleep-wake states depended on both respiratory
Frequency (F[Z. 171 = 19.0. p<O.OO 1 ) (see Figure 3-8 panel) and tidal volume (F[ 2.171 =
49.7. p < 0.00 1 ) (see Figure 3-8 panel C).
The drop in ventilation from wakefulness to NREM slcep was due to a 15.1%
decline in tidal volume (p<0.05) and a 14.1% decrease in frequency of breathing
(p<0.05). Though tidal volume was 7.1% lower in REM sleep than NREM slrep. thrre
was no statistically significant difference in ventilation beiween REM sleep and NREM
sleep (p>O.Oj). This was probably due to a 10.2% increase in frequency in REM sleep.
which although non-significant was enough to counterbalance the decrease in tidal
volume.
The vanability in ventilation. respiratory frequency and tidal volume in eac h
sleep-wake statc are shown in Figure 3-8 panels E. F and G and the values are also listed
in Table 1. Sleep-wake states had a significant effect on the 21 hour mean coefficient of
variation (CV) for ventilation (F[2.17] = 18.5, p<O.OO 1 ). respiratory frequency (F[Z. 171 =
40.0, p<0.001) and tidal volume (F[2.17] = 47.2. p<0.00 1). CV for ventilation was lower
in NREM sleep (mean: 15.5 f 0.24%) than wakefulness (mean: 23.1 + 0.4 1%) and REM
sleep (mean: 24.3 t 0.24%) (p<0.05).
The higher variability in ventilation during wakefulness and REM sleep were also
reflected in the respiratory frequency and tidal volume. CV for respiratory Frequency was
9.6 f 0.17% in NREM sleep. 26.0 f 0.38% in wakefiiness and 3 1.6 f 0.53% in REM
sleep. Similarly for tidal volume. CV was 15.8 f 0.20% in NREM sleep. 23.7 +- 0.20% in
wakefulness and 26.6 + 0.25% in REM sleep.
Fipure 3-8 illustrates the sleep-wake state related changes in lung ventilation. respiratory frequency and tidal volume and their associated coefficients of variation (n=6).
WAKE NREM REM
WAKE NREM REM
WAKE NREM REM
WAKE NREM REM
WAKE NREM REM *
p<O.OS
WAKE NREM REM
b) CO1 production, ventilation norrnalized for CO2 production and body
temperature
Sleep-wake state (wakefulness. NREM and REM sleep) had significant effects on
the 24-hour means in ventilation (F[Z. 171 = 40.2). CO? production (F[2.17] = 8 1.2.
p<O.OO 1) and ventilation normalized for CO2 production (F[2. 171 = 4.5. p = 0.04) (see
Figure C 3-9 panels A. B. C). However. contrary to expectations. no sleep-wake state
related differences were be found betwern the mean 24 hour body temperatures (F[?. 171
< 1. p = 0.54) likely because of the slow response time of the body temperature sensors
inside the telemetry unit (see Figure 3-10 panel D).
Using data combined across the 24-hour periods. CO: production was 27.096
lower in NREM sleep and 35.0% lower in REM sleep compared with waketùlness.
However. there was no difference in CO? production between NREM sleep and REM
sleep (pO.05). The drcline in rnetabolic rate from wakefulness to NREM sleep and REM
sleep was matched by a 24.2% drop in ventilation from wakefulness to NREM sleep
(p<0.05) and a 26.6% decline in ventilation from wakefulness to REM sleep (p<0.05).
When the 24-hour mean ventilation was normalized for CO2 production. there
were no differences between wakefulness and NREM sleep (p>0.05). However.
ventilation normalized for CO: production was 19.0% and 12.8% higher in REM slrrp
compared to wake fulness and NREM sleep (both p<O.O5). respectively.
Figure 3-9 showing effects of sleepwake state on (A) Iung ventilation, (B) COl production, (C) ventilation normalized for CO2 production and (D) body temperature (n=6).
WAKE NREM REM 15
WAKE NREM REM 60 *
WAKE NREM REM
WAKE NREM REM * p<o.os
Table 1: 24 hour means. acrophases and amplitudes of ventilation (Y ,). CO1 production (Y CO,). ventilation normalized for COt production ( Y 11 Y CO,), body temperature (W. tidal volume (V,) and respiratory frequency (f). Coefficients of variation for ventilation (< , CV). tidal volume (V, ( 3 3 and respiratory frequency (f (3") are also shown (n=6).
Variable Parameter REM" Directionofchange
24-hour mem
5. CO, (mumin) Acmphue I hours:mi~~. )
Amplitude
A n = 6 but no values in REM sleep were recorded in one day in one rat.
10.0 + 0.8 18:36 I 1 : 16
0.9 t 0.1 (9.0';)
U' = NREM c REM
W = REM < NREM
I I 1
t Sigificantly different (pc0.05) using Tukey's multiple cornparison test.
Who ur mean co2 .\cmphase (hours:min. I
.bplitude
24-ho w man T b (OC) Xcrophase ( hours:&. )
.-litude
* Sipifrcantly different brtsed on pair bise cornparisons (Wake vs. NREM. Wake vs. REM. SREM vs. REM) using means and standard enors generated by fitting sine wave models through data.
7.3 2 0.6 1 R:47 2 Or34
0.7 10.1 (9.6'0)
37.4 + O. 1 1723 t 050 0.4 + 0.03
37.4 t O. 1 17:34 .t 0:39 0.6 f 0.04
The trends retlected in the 2 way Rh4 ANOVA and Tukey's post hoc test are shown in the last colurnn. Acrophases are expressed as hours and minutes afler light onset. The absence of amplitude and acrophase values indicates that no circadian rhythm was present for that variable. When percentages (%) are expressed for amplitudes. they refer to the % of the 24-hour mesor value.
7
39.9 f 1.6
6.5 k 0.6 19% + 1 : 19
0.7 2 0.2 ( 10.800)
37.3 i 0.1 : i l : 0.3 1 0.06
W > NREM = EREXtt W = NREM = REM W = W 3 I = REhl
42.1 k 2.4 9:32 I 2: 17
2.5 5 0.6 (5.3'0)
k* = ';REM = REM U ' = N R E M = R E ? J U' ' NREM REM*
47.5 2 4.5
Section 4: ADDITIONAL DAY-NIGHT COMPARISONS
Whilst the main purpose of the previous sections were to test for the presence of
circadian rhythms and the effects of sleep wake States on lung ventilation. the primary
aim of this section is to perform day-night cornparisons in wakefulness and NREM sleep
on the other variables such as CO? production and EEG and EMG related variables that
were measured along with lung ventilation. Like the data involved in testing hypothesis
2. the data for this section also consistrd of two five-hour windows. each beginning 2
hours afler lights on and off.
a) Day-night cornparison of CO2 production
In general. these results confinned the results from the circadian rhythrns analyses
performed on the data across II hours in wakefulness and NREM sleep. In both
wakefulness and NREM sleep. lung ventilation. body temperature. CO2 production and
were higher in the night (dark-active) phase than the day (light-rest) phase. (F[L 231 =
Figure 3- 10 panels A.B and D). In wakefulness. CO? production and ventilation increased
by 16.1% and 13.6% from day to night. respectively. These increases in CO: production
and ventilation from the day to night were similar in NREM sleep (17.5% for CO2
production and 14.3% for ventilation).
Corresponding to the 24.2% (91 ml min-') and 23.6% (101 ml min-') aecline in
lung ventilation that occurred from wakefulness to NREM sleep during the day and the
night respectively (see Section 1 of RESULTS). there were also similar changes in CO?
production. CO? production dropped by 30.5% (3.0 ml min-') on going from wakefulness
to NREM sleep during the day and a comparable decrease of 29.7% (3.4 ml min") during
the night (F[ 1.231 4 1. p=0.42) (see Figure 3- 10 panel B).
b) Day-night cornparison of %P2/%61, EEG and EMG amplitudes
Although no sipificant main effects of time of day were found in %P1/%61
(F[ 1.231 = 1.8. p=0.24), there was an interaction between time of day and sleep-wake
state in this variable (F[1.23) = 7.5. p=0.04) (see Figure 3-10 panel E). This interaction
was due to the different effects of time of day on %B2i%6i depending on whether the rat
was awake or in NREM sleep. More specifically. % P ? I % ~ ~ was lowcr during the day
compared to the night in NREM sleep (p<0.05). but was not different dunng day than the
night in wakefûlness (p>0.05).
In both wakrtiilness and NREM sleep. EEG amplitude (F[1.23] = 20.8. pc0.006)
(see Figure 3-10 panel E) and EMG amplitude (F[1.23] = 7.7. p=0.04) (see Figure 3-1 0
panel G) were hipher during the day than the night. There were also interactions between
time of day and sleep-wake state for EEG amplitude (F[1.23] = 2 1.9. p=0.005) (see
Figure 3-10 panel F). The EEG amplitude was 72.8% larger in NREM sleep than
waketùlness dunng the light phase. and 89.1% larger in NREM sleep than wakehlnrss in
the dark phase.
F i a r c 3-10 illustrates day-night cornparisoos in (A) lung ventiIation, (B) CC& production, (C) ventilation normalized for COI production, (D) body temperature, (E) ratio of high to low EEG frequencies, (F) EEG amplitudes and (G) ELMG amplitudes (n=6). Statistically significant (p4.05) day-night differences independen t of sleep-wake states effects (main effect of sleep-wake sta te are indicuted by an asterisk (k) and sleep-wake state effects independent of time of day (main effects of sleepwake state) are flagged using a cmss (t). The pound jrnbol (W) shows s t a t i s t i c a ~ ~ significant interactions (p4.05) between sleep-wake state and time of day.
Day Night
Time of day -W NREM + WAKE
D& ~ i g h t
Time of day
Chapter 4: Discussion
The present study confirmed both hypotheses that there are circadian rhythms in
lung ventilation in wakefulness. NREM and REM sleep, and. that the magnitude of the
change in lung ventilation from wakefulness to NREM sleep is the same across the day.
Circadian rhythms in lung ventilation have previously been studied in expenments that
have either not considered sleep-wake state (Seifert. Knowles. & Mortola. 2000) or only
looked at the waking state (Spengler, Oliver. Czeisler. & Shea. 1997). However. to Our
knowledge this is the first study that has found circadian rhythms in lung ventilation in
NREM and REM sleep. in addition to wakefulness. Furthemore. these sleep-wake States
were measured using electrographic techniques which are used routinely in sleep studies.
Of note. is the observation that the circadian rhythms in lung ventilation in each
sleep-wake state did not differ in amplitude or phase. but only varied in their mean 24-
hour level. Consequently. the change in magnitude in lung ventilation from wakefulness
to NREM (and REM) sleep was similar across time of day. This fact was confirmrd by
comparing the decrease in lung ventilation from wakefulness to NREM sleep at two
times of the day that corresponded to approximateiy the peak and trough of the body
temperature rhythm.
Taken together. these results imply that both time of day and sleep-wake state
may both modulate lung ventilation such that circadian and sleep-wake state effects on
lung ventilation may be additive. This result is of clinical relevance because if a decline
in lung ventilation due to circadian factors supenmposes a decrease in lung ventilation
due to NREM sleep, a daily minimum in lung ventilation during sleep would be
produced. It is during this daily minimum that respiratory symptoins of patients with an
underlying respiratory disorder characterized by impaired gas exchange rnay be
intensified. This time may pose a period of increased vulnerability to hypoventilation.
Technical considerations
The present experiment could be faulted for several reasons. One relates to the
measurement of breathing in this study. There are some disadvantages of measunng
breathing indirectly using the closed system method of barometnc plethysmography. It is
possible that closing the chamber and allowing CO, to build up to estimate the level of
CO? impacted lung ventilation. However, the correlation of the fractional concentration
of inspired CO2 venus ventilation (see Figure 2 3 ) indicated that there was no ventilatory
response to COz at the concentrations present in the experiment. Funhermore. the fact
that the mran ambient % CO2 concentrations were systematic across time of day during
measurements in al1 three sleep-wake states. aliows us to attribute the time of day
changes in lung ventilation to Factors other than inspired CO, concentration. The
advantage of the closed system method was that it ailowed fairly rapid estimatrs of
metabolic rate to be made during slrep-wake states that are short in duration in the rat.
The open tlow method of barometric plethysrnography (e.g.. Jacky. 1978) is unlikely to
stimulate breathing. but is incapable of measuring the rapid changes in metabolic rate that
occur with sleep-wake state.
Using the closed system method of barometric plethysmography. there is also a
potential concem of dismpting the sleep-wake cycles of the rat and causing phase shifis
because of the constant changes in environmental noise as the chamber was opened and
closed. However. the rats seemed to habituate to these patterns and their body
temperature rhythms and sleep-wake cycles suggests that they behaved like entrained and
undisturbed rats. For example. body temperature peaked during the active phase and was
at its minimum during the light phase. Also, the sleep-wake behaviour and EEG
characteristics across time of day were consistent with other studies of rats entrained to a
12 hour light dark cycle (Harnrahi et al.. in press; Trachsel et al.. 1988)
An additional concem relates to discerning sleep-wake states. Although the
judgment of sleep-wake state using electrographic means such as the EEG and EMG. is
less subjective than using behavioural criteria, it is possible to be somewhat subjective.
given that it is scored by the experimenter. This is most likely in making judgments of
wakefulness. NREM and REM sleep dunng transitions between these sleep-wake states.
For exarnplr. at the onset of NREM sleep. hiph frequency components of the EEG and
the EMG are still present. making it difficult to determine when NREM slrep actually
begins. In the present experiment. attempts were made io bypass these problems
associated with sleep-wake state transitions by choosing sections to measure breathing
and metabolism in which the sleep-wake state was established Tor a period ofat least 20
seconds. This duration was arbitrarily chosen to prevent the selection of sleep-wake state
transitions. but. whether or not the animal was in a physiological steady state
characteristic of each sleep-wake state cannot be determined. but assumrd. These
sections were then subjected to analyses of the EEG frequencies and amplitude and EMG
amplitudes and compared. Statistical tests veri fied the di fferences in %P2/%6i and E EG
and EMG amplitudes between sleep-wake states and were sirnilar to a previous study in
rats where a similar approach was used (Hamrahi et al.. in press).
Circadian related changes in lung ventilation
The presence of circadian rhythms in lung ventilation in wakefulness are
consistent with other expenments that have measured lung ventilation and'or assessed the
characteristics of the respiratory control system across time of day. In awake. but not
sleeping humans. circadian rhythms in lung ventilation (Spengler et al.. 1997).
chemosensitivity (Raschke & Moller. 1989; Spengler et al. 2000) and CO? response
threshold (Stephenson et al.. 2000) have been shown. Circadian rhythrns in lung
ventilation have also been demonstrated in Freely behaving rats where sleep-wake state
was not assessed (Seifert et al.. 2000). However unlike in humans. circadian rhythms in
the ventilatory response to CO2 have not been shown in rats. but there are suggestions
that it too may exist given that day night differences in the response to CO2 have been
found in awake adult rats (Peever and Stephenson. 1997).
These studies in rats are of most interest when it comes to comparing similar
studies using the same species. Unlike Peever and Stephenson (1997) who found no
statistically significant differences in ventilation in awake rats breathing normal room air
at IO a m . and 10 p.m.. we found that lung ventilation was significantly higher during the
night than the day. Possible reasons for this discrepancy rnay lie in the large variability
in Peever and Stephenson's (1997) lung ventilation data and differences in methodolo~.
Given that there was a trend towards higher lung ventilation at night than day in their
study. it is probable that this result in their study would have also been of statistical
significance had they had a larger sample size to reduce variability. in support of this. the
overall day-night difference in lung ventilation during wakefulness (mean * SEM) in this
study (5 1 -t 27 mVmin) was similar to Peever and Stephenson's (1997) study (46 * 30
mumin). However. due to the lower variability in Our data. we were able to find
statistically significant day-night differences in lung ventilation in contrast to Peever and
Stephenson (1 997).
The determination of wakefulness using behavioural criteria rather than using
EEG and EMG rneasures may have also contributed to a lack of difference in lung
ventilation between day and night in Peever and Stephenson's (1997) study. In their
study. it is possible that the rats could have been judged to be awake when actually
drowsy or asleep. If such was the case dunng the night. for example. overall lung
ventilation would be lower. Also. artificially arousing the rats so that measurements could
be made in wakefulness during the day (when the rats asleep more) could elevate lung
ventilation. Both of these scenarios would lead to day night differences in lung
ventilation being less likeiy to be found. By seleciing periods of established wakehlness.
NREM and REM sleep using electrographic criteria. thrse potential problems were
circumvented in Our study.
in contrast to Peever and Stephenson ( 1997). but consistent with Our study. Siefert
et al. (2000) also found day night differences in lung ventilation in air breathing rats
whilst showing a circadian rhythm in lung ventilation. They found that this circadian
rhythm in lung ventilation was coincident with the rest-activity rhythm and body
temperature. However. since sleep-wake States of the animals were not considered and
given that these rats are nocturnal animals which are predominantiy awake during the
night and asleep during the day. it is plausible that the changes that the day night
oscillations in lung ventilation were a result of time of day changes in sleep-wake state.
That is. rffects of sleep-wake state on lung ventilation may have masked the circadian
rhythm in lung ventilation. Usine the EEG and EMG to venQ sleep-wake states and
separate out potential masking effects of sleep-wake state on lung ventilation. the present
study demonstrates that the circadian rhythm in lung ventilation is independent of sleep-
wake state.
One possible manifestation of the masking effect of sleep-wake state in Siefert et
al.3 (2000) study can be found by comparing the different mechanisms by which lung
ventilation was altered as a function of sleep-wake state and time of day. In Siefert et al
(2000) the circadian rhythm in lung ventilation was attributed to time of day changes in
both tidal volume and respiratory frequency. In contrast. in our study. the iime of day
changes in lung ventilation in wakefulness and NREM sleep were due to changes in
respiratory fiequency only. However. sleep-wake state changes in lung ventilation were
due to both tidal volume and frequency. Therefore. the time of day changes in lung
ventilation due to tidal volume in Siefert et al's (2000) study can probably be attributed to
the masking effects of sleep-wake state.
Potential mechanisms mediating the circadian rhythm in lung ventilation
Of interest in this discussion are the potential mechanisms mediating the circadian
rhythms in lung ventilation in wakefùlness. NREM and REM sleep. in this regard. the
phase relationships between lung ventilation. metabolic COI production and body
temperature in each sleep-wake state are illuminating. The fact that al1 of these variables
have similar acrophases in each and across sleep-wake states at approximately 17 hours
afler light onset ( 5 hours after lights offset) suggests that they are inter-related.
There is evidence that the circadian rhythms in lung ventilation in each sleep-
wake state are in part caused indirectly by the circadian changes in CO? production and
body temperature. For example. when ventilation was nonnalized for CO? production the
circadian rhythm in lune ventilation was no longer present in wakehlness and REM
sleep. The constant ventilation to CO2 production ratio (i' &'CO2) across the day in
wakefulness and REM sleep implies that the circadian changes in lung ventilation
matches the circadian changes in metabolic CO: production in these two sleep-wake
states. This is consistent with other studies in which day-night comparisons of this
variable have been made in newbom (Saiki & Mortola, 1995) and adult rats (Peever &
Stephenson. 1997). It is also consistent with studies that have demonstrated the cntical
role of CO2 in determining lung ventilation (Phillipson et al.. 198 1 ) and with models of
respiratory control that emphasize the chemoreflex control of breathing (e.g.. Duffin.
1990) (also see Introduction). Since PaCO? is detemined by a balance of metabolic
production of COz and ventilation. the constant Y ,/Y COI ratio across the day in
wakefulness and REM slerp implies that PaCOz is also constant across the day in thsse
sleep-wake states.
Hints that thennoregulatory mechanisms may also be responsible for the circadian
changes in lung ventilation in each sleep-wake state comes from the observation that the
time of day changes in lung ventilation are mediated by changes in respiratory frequency
rather than tidal volume. It is interesting to note that the acrophase of respiratory
frequency. when fitted with the sine wave model. was exactly the same as that for body
temperature. The increase in respiratory frequency. in panting for example. is a method of
facilitating heat loss in species such as dogs and cats (Tenney & Boggs. 1986). Panting
causes a large increase in ventilation of the respiratory dead space. and therefore,
increases evaporative heat loss of the respiratory tract (Mortola & Gautier. 1995).
Although rats are not known to use panting as a method of heat loss (Gordon. 1990).
Boden, Hams. & Parkes (2000) have demonstrated that there is a respiratory drive in
addition to the increase in CO2 production at raised body temperatures in the rat. It is
possible that this additional drive. waxes and wanes to alter respiratory frequency in
phase with the circadian rhythm in heat production (also measured by COz production)
(Gordon. 1993) and body temperature.
To complicate matters Further. there are also the effects of arousal levels to
consider within each sleep-wake state which may have a direct or indirect impact on
vent ilation. We have arbitrarily divided sleep-wake state into the commonly accepted
stages of wakefulness. NREM and REM sleep. However. even within each of these sleep-
wake States there is a lack of unifonnity. Take for example. a cornpanson of the ratio of
the high to iow frequencies (%Pd%6i) of the EEG in NREM sleep using a five-hour
sample of data from the day and from the night (see Figure 3-10 panel E). The lower
%P$%& in NREM sleep during the day implies that the NREM sleep was "deeper"
during the day than during the night. In addition. there were time of day di fferences in the
EEG amplitude when considering both wakefulness and NREM sleep. It is known that in
humans. where NREM sleep is scored in more stages. that the stage of NREM slerp
influences the dope of CO2 response line (Douglas et al.. 1982). a key deteminant of
resting ventilation.
Though metabolic COz production. body temperature and arousal level in each
sleep-wake state may be involved in the circadian rhythm in lune ventilation. the present
study is unable to make any hrther conclusions on which of these variables may be more
quantatively or qualitatively important. or how they are related to each other given the
complicated nature of the interactions between body temperature, metabolic rate and lung
ventilation and sleep-wake state.
The circadian rhythm in ventilation normalized for CO2 production in YRE-M sleep
So far. the focus of the discussion on possible mechanisms mediating the
circadian changes in lung ventilation has been on indirect mechanisms such as CO?
production. body temperature and possible differences in arousal level within each slerp-
wake state. There is also the possibility that more direct mechanisms such as connections
between the primary mammalian circadian pacemaker. the SCN. and the components of
the respiratory system may exist. This study. which involves the integrated responses of
lung ventilation in each sleep-wake state across time of day. does not address this
question or which specific part of the respiratory control system may be involved.
However. there are suggestions that in addition to indirect mechanisms. there are also
more direct rnechanisms which may be involved. In this regard. the rather surptising
circadian rhythm in ventilation normalized for CO2 production (i' ,/i' COz) during NREM
sleep. but not in wakelùlness or REM sleep. is insighttùl.
The presence of a circadian rhythm in i' CO2 during NREM sleep implies
that across a 23-hour period. lung ventilation does not change proportionally to metabolic
COz production. and consequently. this suggests that PaCO: too may not be constant
across time of day in NREM sleep. The reason for the presence of a circadian rhythm in
i' CO? in NREM sleep. but a lack of one in wakehlness and REM sleep. is unclear.
When the acrophases of the ventilation and metabolism rhythm were tested. they
showed no differences. indicating that these two variables were in phase. Perhaps, NREM
sleep is the most stable sleep-wake state to observe such a rhythm given that the data for
the Y ,/Y COI in wakefulness and REM sleep is noisier. Before proceeding, it is
imporiant to insert a note of caution with regard to this result. Given that a large nurnber
of statistical tests were performed on the data. this data may represent a statistical arti fact.
More specifically. this result may possibly be a Type I statistical error. That is. falsely
rejecting the nul1 hypothesis that there is no 24-hour rhythm in Y ~li ' COz when it is in
fact tme.
Assuming that the circadian rhythm in d< CO? in NREM sleep is real. in
mechanistic terms. it suggests that across a 24-hour period in NREM sleep. the
chernoretlexes are influenced directly by the circadian timing system such that PaCO?
would also oscillate. This is not an unreasonable hypothesis given that direct modulation
of the respiratory control system by the circadian timing system has been suggested in
expenments in humans (Rascke and Moller. 1997: Stephenson et al. 2000; Spengler et al.
2000) and in ducks (Woodin and Stephenson. 1997). Intriguing and interesting to
speculate on. is the question as to why such a hypothetical mechanisrn might only be
observed in NREM sleep and not in wakehilness or REM sleep.
A possible functional role of the circadian rhythm in Y dY CO? in NREM sleep is
in REM sleep replation. Examination of the circadian rhythm shows that l i t COz in
NREM sleep nses throughout the light phase when the rats are ncrmally asleep to peak
approxirnately nine and a half houn before slowly beginning to decrease again. Likewlse
from persona1 observation in our study. the frequency and duration of REM sleep during
the rest (light) phase increased progressively from its minimum at the start of rest phase
(also see Figure 1-1 0 and Trachsel. Tobler. & Borbelly, 1988). Since PaCO? and H+ are
invenely related to Y $Y CO?, this pattern suggests that PaC02 and H+ are at their
minimum at the acrophase of the Y Ili' CO2 rhythm in NREM sleep. At this time. the
higher Y CO? value implies that ventilation is relatively higher for a given lrvel of
CO? production.
Dev & Loeschcke (1979) have proposed that hydrogen ions alter respiratory
discharge by inhibiting the metabolism of acetylcholine at the synapse. Like many other
enzymes. the activity of acetylcholinesterase. the enzyme which breaks down
acetylcholine. is pH sensitive. Cholinergie agents stimulate breathing when applied to the
ventral medullary surface. as do antagonists of acetylcholine esterase (e.g physostigmine)
(Chemiack. 1993). Dev & Loeschcke (1979) found that atropine. a muscarinic receptor
antagonist blocked the response to CO:. This has been confirmrd by Nattie. Mills. & Ou
(1988) who used more specific antagonists to show that the M2 muscannic binding site
was the pertinent receptor. Knowing that cholinergie and cholinoreceptive mechanisms
play a key role in the neurobioloa of REM sleep generation (Lydic & Baghodoyan.
1994). that acetylcholine. the neurotransmitter involved in these systems is modulatrd by
pH. and. that REM sleep always follows NREM sleep. it is possible that the Y ,/Y CO2
rhythm in NREM sleep represents a feed forward mechanism by which the respiratory
system regulates REM sleep by changing hydrogen ion concentration.
These ideas are consistent with two concepts. Firstly. it is compatible with the
proposa1 that REM slerp is hinctionally related to NREM sleep rather than waking (e.g.
Benington & Heller. 1994). Secondly. it is also consistent with the role of humoral inputs
affecting multiple aspects of neuronal circuits including their synaptic connectivity.
responsiveness to stimuli and neuronal composition in addition to the intrinsic properties
of neurons within the affected circuits (Knieger & Fang. 1999). Given that REM is a
distinct neurophysiological state (Phillipson & Bowes. 1986). H+ may be one such
humoral input modulating the neuronal circuitry involved in REM sleep. However. the
hypothetical role of circadian control of breathing in NREM sleep to Vary H+ across tirne
of day (which rnay possibly modulate neuronal circuitry involved in REM sleep
generation across time of day) rernains to be tested.
Sleep-wake related changes in breathing, metaboiism and body temperature
In general. our tindings of the effect of sleep-wake statç on lung ventilation are
consistent with other studies in humans (White et al.. 1985). dogs (Phillipson et al., 1976)
and rats (Pappenheimer. 1977) that have found lung ventilation to decline from
wakehlness to sleep. In rats. Pappenheimer (1977) found a 10-20% decline in minute
ventilation on going from wakefulness to NREM sleep. In cornparison. we found a mean
decline of 24.2% in lung ventilation from wakefulness to NREM sleep using data
generated over a 24-hour period. Though there was a decline in ventilation from
wakefulness to NREM sleep. we found no di fference in lung ventilation between NREM
sleep and REM sleep. There were also profound state related changes in the variability of
breathing as indicated by coefficients of variation in tidal volume. frequency and
ventilation consistent with what has been reported in the literature (Phillipson & Bowes.
1986). As expected. breathing during NREM sleep was regular and less variable than in
wakefulness and NREM sleep.
There were no changes in the coefficient of variation in lung ventilation and tidal
volume in al1 three sleep-wake states across time of day. However. there was a circadian
rhythm in the coefficient of variation in respiratory frequency during NREM sleep. but
not during wakefulness or REM sleep. The reasons for these novel observations are
unclear. but they may reflect time of day changes in the activity of the respiratory rhythm
generator.
The changes in ventilation that occurred from wakefulness to NREM sleep were
due to a decreased tidal volume and respiratory frequency. The decrease in respiratory
frequency in NREM is consistent with Megirian. Ryan. & Sherry's (1980) and
Pappenhiemer's (1977) study in rats. However. the decrease in tidal volume was different
to Pappenheimer (1977) who found either an increase or no change in tidal volume
dunng NREM sleep compared to wakefulness. The values on lung ventilation. tidal
volume and respiratory frequency in the rat literature are quite variable possibly due to
the large differences in environmental conditions. experirnental techniques. protocols and
strains that were used. indeed. even with similar experimental protocols Strohl et al..
(1997) found differences in respiratory and metabolic variables amongst strains of rat.
Our values for tidal volume in rats breathing room air agree with the values in the
literature ( c g . Holloway & Heath. 1984: Mortola. 199 1: Strohl et al.. 1997: Aaron &
Powell. 1993). The respiratory rate. however. was generally higher in our study in al1
sleep-wake states. Consequently. our values for lung ventilation were also generally
higher. Perhaps the primary reason that our vaiues are higher is because Our results
include data collected during the night when the metabolic rate. and hence. the lung
ventilation is higher. The majority of the experiments in the rat are likely to be collected
during the day. at a time when the rats are mostly resting and asleep. and therefore.
metabolic rate and respiration are likely to be lower. The absence of a correlation
between inspired COz level and lung ventilation (see Figure 2 4 ) indicates that this was
not due to the CO? level in the chamber stimulating breathing.
It must be noted that metabolic rate values in the awake rat are unlikely to be the
basal metabolic rate. which. is commonly and more easily measured in humans. Basal
metabolic rate. a reflection of the minimal metabolic expenditure required for the
maintenance of homeothermy. is defined as the metabolic rate of an individual that is
resting in a thermoneutral state. but not sleeping, 14 to 18 hours after eating (Gordon.
1993). Our values are unlikely to represent basal metabolic rate becausr it was very
difficult to achieve a state in which the rats are at absolute rest. but not sleeping. When
the rats were awake. they continually engaged in some form of motor activity such as
grooming. feeding. exploring etc. Bramante (1 958). for example. found that in a 5-hour
period of measuring activity and oxygen consumption simultaneously. the rat exhibited
no activity only 4.9% of the tirne. We measured breathing and metabolism during periods
in which there were no gross body movements (which cause large fluctuations in the
pressure traces of the plethysmogaph as well as markedly elevating metabolic rates) such
as when the rat was exploring. but undoubtedly engaged in some form of microactivities.
Thus. this needs to be considered when interpreting Our breathing and metabolic rate
values during wakefulness. in contrast to wakefulness. breathing and metabolism in
NREM and REM sleep are less afkcted by such concems.
in companng sleep-wake state related differences in metabolic COz production. as
expected we found that CO2 production was higher in wakehlness than both NREM and
REM sleep. but did not find any difference between NREM and REM sleep. This trend is
consistent with other studies in humans (White et al.. 1985), but another study in rats
found that in this species. metabolic rate was türther reduced in REM sleep (Schmidek.
Zachariassen. & Hammel. 1983). The reason for this discrepancy is probably duc to the
smaller sample sire and the fewer measurements of REM sleep metabolism in our study
that may have contributed to the trend towards lower COz production in REM sleep.
which was not statistically significant.
Of interest in discussing metabolism and ventilation is how these two variables
are related to each to other (as indexed by the ventilation normalized for metabolic CO2
production. lit CO.) to regulate PaCOz in cach of the sleep-wake States. PaCO? has
been reponed and calculated to increase dunng NREM sleep compared to wakehlness
(Pappenheimer. 1977: Phillipson & Bowes. 1986). This has been explainrd by a greatrr
decrease in ventilation relative to the decrease in metabolic rate that occurs with slrep.
which results in a lower CO. ratio. In contrast. but consistent with White et al.'s
(1985) study in humans. we observed no differences in ,/Y CO. between wakefulness
and NREM sleep. However. chanps in lung ventilation (i' 1) have a different effect on
PaCO? depending on whether the changes are mediated by changes in tidal volume or
respiratory frequency. Assuming that dead space volume is 0.75 ml (tiom an rstimatrd
value of 0.7 1 ml / 100g. Pappenheimer. 1977). and. that it is constant across sleep-wake
state. the alveolar ventilation (i' is equal to respiratory fiequency x (tidal volume -
dead space volume)) in wakefulness and NREM sleep can be recalculated from Table 1
to be -3 19ml /min. and -2 19 mumin respectively. Thus. the recalculated i- . , < CO:
values are 3 1.9 wakefulness and 30.0 in NREM sleep. Comparing i' CO2 instead of
i' d Y CO?. the lower Y CO2 value in NREM sleep compared to the equivalent value
in wakefulness. does indeed suggest that PaCOz is higher dunng NREM sleep than in
wakefulness.
in contrast to a lack of difference in Y [/Y CO? between wakefulness and NREM
sleep. i' [/Y COz was higher in REM sleep compared to wakefulness and NREM sleep.
This was due to a decline in metabolism. but not lung ventilation on p i n g fiom NREM
to REM. This irnplies that lung ventilation was higher relative to the metabolic rate
during this state. To date. we are not aware of any studies that have looked at the impact
of REM sleep on Y ,/Y CO2. so cornparison of this variable in rats is not possible.
However. the higher l i t COz is consistent with the finding that ventilation in REM
sleep is oAen independent of metabolic control and is intluenced by the behavioural
inputs (Orern. 1994: Phillipson & Bowes. 1986).
Whereas. lung ventilation and metabolism were dependent on sleep-wake state.
our data did not reveal any effect of sleep-wake state on body temperature. This is in
contrast to body temperature decreases on falling aslerp (Alfoldi. Rubicsek. Csemai. &
Obal. 1990: Li. Randall. & Nattie. 1999). This finding was likely due to the short
duration of sleep-wake cycles in the rat and the slow response time of the temperature
sensor. Although, it is likely that rats change their thermoregulatory set point uith sleep-
wake state (Glotzbach & Heller. 1976). core body temperature is unlikely to change very
quickly because of the thermal inertia present in the intemal ogans of the body. Since the
temperature sensor. which had a 90% response time of approximately 2-3 minutes. was
implanted in the peritoneal cavity. it is unlikely that the temperature changed rapidly
enough to be detected across sleep-wake states that oflen changed quicker than this 2-3
minute period.
Future researc h opportunities
There are many questions raised by this experiment that should provide the
stimulus for further research. Firstly, it is not known from the present experiment whet her
PaCO? is constant across time of day in a given sleep-wake state. There are suggestions
that PaCOz may be constant across the day in wakefulness and REM sleep. but whether
this is also true in NREM sleep needs to be verified given that there is a circadian rhythm
in Y ,/Y CO2 in this sleep-wake state. If so, experiments need to address its possible
functional significance including the proposai that it may be involved in REM sleep
regulation.
Secondly. questions need to be directed at the mec hanisms behind the circadian
rhythm in lung ventilation in rach sleep-wake state. The results of this experiment
suggest that the circadian rhythms in lung ventilation may in pan be an indirect product
of metabolism. body temperature or arousal level within a panicular sleep-wake state.
One possible approach to answenng this question would be to allow the rats to fiee run in
constant conditions. By examining the periods and phase relationships of rhythms in lung
ventilation. metabolism and body temperature. it may be possible to see which of these
variables is more closely coupled to lung ventilation.
One possible mechanism that is likely to be involved along with the circadian
rhythm in COz production. are time of day changes in the chernoreflexes which we
speculated on. However. it is yet to be determined if there are time of day dependent
changes in the respiratory control system in each sleep-wake state manifested as changes
in the thresholds or chemosensitivity or both. The circadian rhythm in NREM sleep
suggests that the circadian timing system may be directly involved in the modulation of
lung ventilation in this sleep-wake state. Given that there is a vast amount of literature on
the nervous system of the rat. ir may be feasible to perhaps study the neural basis of the
observations in this expçrirnent. Of potential benefit in this regard. would be studies in
which lung ventilation. body temperature. metabolism and sleep-wake state were
measured across the day in SCN lesioned rats.
Conclusions
Overall. the present experiment demonstrates that there are circadian rhythrns in
lune ventilation in wakefulness. NREM sleep and REM sleep. Furthemore, we have also
shown that the magnitude of the change in lung ventilation From wakefulness to NREM
sleep is the same across time of day. Togeth;;. these data show that the lung ventilation
may not only be modulated by sleep-wake state. but also by circadian factors. This is of
potential clinical sipificance for patients who already hypoventilate due to an underlying
respiratory abnonnality. For such individuals a time of day decline in lung ventilation
supenmposed on a decline in lung ventilation due to sleep. may exacerbate respiratory
symptoms.
Hvpothesis l
ë- 3
2 0 0 ~ O 12 24 Tirne tlnce llght onset (houn)
Time of Day
Day Night
These circadian rhythms in lung ventilation in each sleep-wake state are likely to
be mediated panly by time of day changes in metabolic COz production. body
temperature and arousal state within each sleep-wake state. When lung ventilation was
normalized for COz production (i' [li' CO?) there was no circadian rhythm in wakefulness
or REM sleep. However, v I/Y COz did exhibit a circadian rhythm in NREM sleep
suggesting that direct modulation of the respiratory control system may be involved in
this sleep-wake state.
References
Aaron. E. A.. & Powell. F. L. (1993). Effect of chronic hypoxia on hypoxic ventilatory response in awake rats. Journal oj'Applied Physiologv. 74(4), 1635- 1640.
Aschoff. J. ( 198 1). A survey of biological rhythms. in J. Aschoff (Ed.), Handbook of' Behaïioural Neurobiologv (Vol. 2. pp. 3-1 0). New York: Plenum Press.
AschotT. J.. & Pohl. H. (1970). Rhythmic variations in energy mmetabolism. Federation Proceedings. 2 9(4). 1 54 1 -5 2.
Aserinsky. E.. & Kleitman. N. ( 1953). Regularly occumng penods of eye motility. and concomitant phenornena. dunng sleep. Scie~ice. 1 18. 273-274.
Benington. J. H.. & Heller. H.. C. (1994). Does the function of REM sleep concem non- REM sleep or waking. Progress in iVeurobiology. 44.443449.
Berger. A. J.. Mitchell. R. A.. & Severinghaus. J. W. ( 1 977). Regulation of respiration. Part I . :Vew England Journal oj'Medicine. 2 97. 92-96.
Bernard. D. G.. Li. A.. & Nattie. E. E. (1996). Evidence of central chemoreception in the medullary rap he. Journal oj'Applied Physiologv. 80. 1 08- 1 1 5.
Bliwise. D. L. (1994). Normal aging. In M. H. Kryger. T. Roth. & W. C. Dement (Eds.). Principles and practice oj'sleep medicine (2nd ed.. pp. 26-39). Philadelphia: WB Saunders.
Boden. A. G.. Hams. M. C.. & Parkes. M. J. (2000). A respiratory drive in addition to the increase in CO2 production at raised body temperature. Experimental Physioloy*. 85(3). 309-3 19.
Borbely. A. A.. & Achennann. P. (1999). Sleep homeostasis and models of sleep regulation. Journal of Biological Rhythms. 14(6). 557-68.
Bramante. P. 0. (1958). Energy rnetabolism of the albino rat in minimal levels of spontaneous muscular activity. Journal of dpplied Physioloyv. 2 4. 1 1 - 16.
Brebbia. D. R.. & Altushuler. K. Z. (1965). Oxygen consumption rate and electroencephalographic stage of sleep. Science. 150. 162 1 - 1623.
Budzinska. K.. Euler. C. v.. Kao. F. F.. Pantaleo. T.. & Yamamoto. Y . (1985). Effect of eraded focal cold block in the rostral areas of the medulla. Acta Physiologica kandinai*ia. 12 4. 329-340.
Card, J. P., & Moore. R. Y. (1 982). Ventral lateral geniculate nucleus efferents to the rat suprachiasmat ic nucleus exhibit avian pancreatic polypeptide like immunoreactivity. Journal of Comparative Neuro[ogi>. 206. 390-396.
Cankadon, M. A., & Dement, W. C. (1994). Normal human sleep: an overview. in M. H. Kryger, T. Roth. & W. C. Dernent (Eds.), Principles and practice of' sleep medicine (2nd ed.. pp. 16-25), Philadelp hia: WB Saunders.
Chase, M. H.. & Morales. F. R. (1994). The control of motoneurons during sleep. In M. H. Kryger, T. Roth. & W. C. Dement (Eds.). Principles and practice oj'sleep medicine (pp. 1 63- 1 75). Philadelp hia: WB Saunders.
Coates. E. L.. Li. A.. & Nattie. E. E. (1993). Widespread sites of brainstem ventilatory chemoreceptors. Journal oj"4pplied Physiologv. 75. 5 - 14.
Coote. J. H. (1982). Respiratory and circulatory control dunng sleep. Journal of' Experimental Bioloyv. 100. 223 -244.
Cunningham. D. J. C.. Robbins. P. A.. & Wolff, C. B. (1986). Integration of respiratory responses to changes in alveolar partial pressures of CO2 and 0: in arterial pH. In W. Chemiack N.S.. J.G (Ed.). Handbook O/* Physiologr. Section 3: m e Respirarory %rem: Control of' Breathing. part ? (Vol. 2 . pp. 175-528). Baltimore: American Physiological Society.
Czeisler. C. A., Weitman. E. D.. Moore-Ede, M. C.. Zimmerman. J. C.. & Knauer. R. S. (1980). Human sleep: its duration and organization depend on its circadian phase. Science. 210, 1264-1247.
Datta. A.K.. Shea. S.A.. Homer. R.L.. Guz, A. (199 1). The influence of induced hypocapnia and sleep on endogenous respiratory rhythm in humans. Journal of' Physiologv (London). 440. 1 7-3 3.
Dement. W. C.. & Kleitman. N. (1957). Cyclic variations in EEG during sleep and their relation to eye movements. body motility and breathing. Electroencephalogrqhy and Clinical !Vewophysiologv, 9. 673-690.
Dev. N. B.. & Loeschke. H. H. ( 1 979). Topography of the respiratory and circulatory responses to acetylcholine and nicotine of the ventral surface of the meduila oblongata. Pflugers Archiv. 3 79. 19-27.
Dijk. D. J.. & Czeisler. C. A. (1995). Contribution of the circadian pacemaker and the sleep homeostat to sleep propensity. sleep stnicture. electroencep halographie slow waves. and sleep spindle activity in humans. Journal of- ~Veuroscience. I j ( 5 ) . 3526-3538.
Douglas, N. J.. White. D. P.. & Weil J V et al. (1982). Hypercapnic ventilatory response in sleeping adults. American Review of Respiratory Disease. 126. 758-762.
Drorbaugh. J. E., & Fenn. W. 0. (1955). A barometnc method for measuring ventilation in new born infants. Pedianics, 16, 8 1-87.
Duffin. J. (1990). The chemoreflex control of breathing and its measurement. Canadian Journal of'dnaesthesia ( 3 7). 93312.
Duffin. J.. Emre. K.. & Lipski. (1 995). Breathing rhythm generation: focus on the rostral ventrolateral medulla. News in Plysiologi, and Science. 10. 133- 144.
Du ffin. J.. Mohan R.M.. Vasiliou. P. Stephenson. R.. Mahamed S. (2000) A model of the chemoreflex control of breathing in humans: model parameten measurement. Respiration Physiolog-. 120. 1 3 -26.
Duffin. I.. & McAvoy. G. V. (1988). The peripheral chemoreceptor threshold in man. Journal of Physiologv (London), 406. 1 5-26.
Eastman. C. 1.. Mistleberger. R. E.. & Rechtshaffen. A. ( 1983). Suprachiasmatic nuclei lesions eliminate circadian temperature and sleep rhythms in the rat. Phpiologv and Behariour-. 32, 357-368.
Edgar. D. M.. Dement, W. C.. & Fuller. C. A. ( 1993). Effect o f SCN Içsions on sleep in squirrel monkeys: evidence for opponent processes in sleep-wake regulation. Journal oj'lVeuroscience, 13(3), 1065- 1079.
Edgar. D. M.. Manin. C. E.. & Dement. W. C. (1991). Activity feedback to the rnammalian circadian pacemaker: Influence on observed measures of rhythm period length. Journal oj'Biologica1 Rhwhms. 6(3). 1 85- 199.
Feldman. J. L., & Smith. J. C. (1995). Neural conirol of respirato~ pattern in mamals : an overview. in S. A. Dempsey & A. I. Pack (Eds.). Regulation oj'dreathing (2nd ed.. pp. 39-69). New York: Marcel Dekker.
Fink. B. R. (1961). Lnfluence of cerebral activity in wakefuiness on regulation of breathing. Journal of Applied Physiologv. 16. 15-20.
Ganong. W. F. (1 99). Regulation of respiration. Reiiew of medical p&siology ( 19th ed.. pp. 640-649). Nonvalk: Appleton & Lange.
Gonzalez. C.. Almaraz. L.. Obeso. A.. 81 Rigual. R. (1992). Oxygen and acid chemoreception in the carotid body chemoreceptors. Trends i!l Netiroscience. I j . 146-153.
Gordon, C . J. (1990). Thermal biology of the laboratory rat. Phvsiologv and Behavior, 47. 963-99 1.
Gordon, C. J. (1993). Metabolism, Temperottire regulation in [aboratop rodents (pp. 47- 71). New York: Cambridge University Press.
Hamrahi. H.. Chan. B., & Horner. R (in press). Online detection of sleep-wake state and application to produce intermittent stimuli exclusively in rats. Journal oj'Applied Physiologv.
Heeringa, J.. Berkenbosch, A.. De Goede, J., & Olievier. C. N. (1979). Relative contributions of central and peripheral chemoreceptors to the ventilatory responsr to CO? during hyperoxia. Respiratiori Physioiogr~. 3 7. 365-3 79.
Holloway, D. A.. & Heath. A. G. (1984). Ventilatory changes in the golden hamster. Mesocricetus auratus. compared with the laboratory rats. Rattus norvegicus. during hypercapnia andior hypoxia. Comparative Bioclzemist~ and Physioloyi>. 77A(2), 267-73.
Horner. R. L. (2000). Physiological effects of sleep on the cardiovascular systrm. In T. D. Bradley & J. S . Floras (Eds.). Sleep disorders and cardiovasctilar and cerebrovascular disease (in press ed.. ). New York: Marcel Dekker.
ibuka. N.. & Kawamura. H. (1975). Loss of circadian rhythm in sleep-wakehiness cycle in the rat by suprachiasmatic nucleus lesions. Brait1 Research. 96. 76-8 1.
[nouye. S. 1. T.. & Kawamura. H. (1979). Persistence of circadian rhythmicity in a mammalian hypothalamic "island" containing the suprachiasmatic nucleus. Proceedings of the National Academy ofSciences. 76. 5962-66.
Johnston. R. F.. Moore. R. Y.. & Morin. L. P. (1988). Loss of entrainment and anatomical plasticity afler lesions of the hamster retinohypothalamic tract. Brain Research, 460.297-3 1 3 .
Jones. B. E. (1994). Basic mechanisms of slerp-wake States. In M. H. Kryger. T. Roth. & W. C. Dement. (Eds.). Principles and practice of'sleep medicine (2nd ed.. pp. 135- 162). Philadelphia: WB Saunders.
Kmeger. J. M.. & Fang. J. (1999). Cytokines and sleep regulation. In R. Lydic & H. A. Baghodyan (Eds.). Handbook of' Behavioral State Control (pp. 609-621). Boca Raton: CRC Press.
Krieger. J. ( 1 989). Breathing during sleep in normal subjects. In M. H. Kryger. T. Roth. & W. C. Dement. (Eds.). Principles and practice of'sleep medicine ( 1 st ed.. pp. 257-268). Philadelphia: WB Saunders.
Loeschcke. H. H. (1 982). Central chemosensitivity and the reaction theory. Journal of Ph-vsiologv (Lo~tdon). 332. 1-24.
McNew, J. J.. Burson. R. C.. Hoshizaki. T.. & Adey. W. R. (1971). Sleep wake cycle of an unrestrained isolated chimpanzee under entrained and fi-ee nmning conditions. Aviation. Space and Environmental Medicine (43). i 555-6 1.
Megirian. D.. Ryan. A. T.. & Shemy, J. H. (1980). An electrophysiological analysis of sleep and respiration of rats breathing different gas mixtures: diaphragmatic muscle hnction. Electroe~zcephalography and Clinicai Neurophpiologv, 50.303- 3 13.
Mitchell. R. A.. Loeschcke. H. H.. Massion. W. H.. & Severinghaus. 1. W. (1963). Respiratory responses mediated through superficial chemosensitive areas on the medulla. Jownal of Applied PIyiologr: 18. 523-533.
Mahammed. S. (2000). C hemoreflex adaptations to hypoxia. (M.Sc. Thesis). University of Toronto. Toronto.
Mohan. R.. & Dufin. J. (1997). The effect of hypoxia on the ventilatory response to carbon dioxide in man. Respiration Physioiogv. 108. 10 1 - 1 15.
Moore. R. Y. (1997). Circadian rhythms: basic neurobiology and clinical applications. .4nnual Reviews in Medicine, 48. 253-66.
Moore. R. Y. ( 1999). A clock for the ages. Science. 284.2 102-3.
Moore-Ede. M. C. (1986). Physiology of the circadian timing system: predictive vs. reactive homeostasis. .4merican Jownal of Physiologv: Regulato~~ Integrative Comp. Physiol.. 250. R735-R752.
Morin. L. P.. Blanchard. J.. & Moore. R. Y. (1992). intergeniculate leaflet and suprachiasmatic nucleus organization and connections in the golden hamster. Vistral netrroscience, 8. 2 1 9-230.
M o m i . G.. & Magoun. H. ( 1949). Brainstem reticular formation and activation of the E EG. Electroencephalography and Clinical iVeurophysiologv, I . 455-473.
Monola. J. P. (1991). Hamsters vs rats. ventilatory responses in adults and newboms. Respiration Physiologv. 85. 3 05-3 1 7.
Mortola. J. P.. & Gautier. H. (1995). interaction between metabohsm and ventilation: effects of respiratory gases and temperature. in P. A. 1. Dempsey J.A. (Ed.). Regulation of breathing (Vol. 79. pp. 10 1 1 - 1064). New York: Marcel Dekker.
Mortola, J. P. and Frappell, P. B. (1998). On the barometnc rnethod for measurements of ventilation, and its use in small animals. Canadian Journal of'Physiologv and Phamacology 76,937-944.
Mrosovsky. N. ( 1988). Phase response curves for social entrainment. Journal of' Comparative Physiologv and Anaton~y, 162, 35-46.
Nagai. K.. Nagai. N., Sugahara. K.. Niijima, A., & Nakagawa. H. (1994). Circadian rhythms and energy metabolism with special reference to the suprachiasmatic nucleus. Nerrroscience and Biobehavioral Reviews, I8(4). 5 79-84.
Nagai. K.. Nishio. T., & Nakagawa. H. (1985). Bilateral lesions of suprachiasmatic nucleus eliminate rhythms of oxygen consumption and the respiratory quotient in rats. Es-perienria. 4 / ( 9 ) . 1 1 36-8.
Nattie. E. E. ( 1999). CO2, brainstem chemoreceptors and breathing. Progress in neurobiofogi: 59. 299-33 1 .
Nattie. E. E.. Mills, J. W.. & Ou. L. C. (1988). Pirenzine prevenis diethylpyrocarbonate inhibition of cerebral CO? sensitivity. Joirrnal of Appfied P/ivsiolog. 65. 1962- 66.
Nauta. W. J. H. (1946). Hypothalamic regulation of sleep in rats: expenmental study. Journal of'iVeurophysiology, 9. 285-3 16.
Nelson. W.. Tong. L. Y.. Lee. J.. & Halberg. F. (1979). Methods of cosinor- rhythmometry. Chronobiologia. 6. 305-323.
Olson. E. B. J.. Vidnik. E. H.. & Dempsey. J. A. (1988). Carotid body excision significantly changes ventilatory control in awake rats. Joirrnal 01' Applied Physiologv 44(2). 666-67 1.
Orem. J. (1994). The wakefulness stimulus for breathing. In S. C. E. Saunden N.A. (Ed.). Sleep and Breathiny (2nd ed.. pp. 133-155). New York: Marcel Dekker.
Orem. J.. & Dick. T. (1983). Consistency and signal strength of respiratory neuronal activity. Journal ojNeirrophysiolog-. 50. 1089-1 107.
Orem. J.. Montplaisir. J.. & Dement. W. C. (1 974). Changes in activity of respiratory neurons during sleep. Brain Reseurch. 82. 309-3 15.
Orem, J.. Netick. A.. & Dement. W. C. (1977). increased upper aiway resistance to breat hing during sleep in the cat . Electroencephalography and Clinical ~Veurophysiologv. 43. 1 4-22.
Ott. L (1993) An introduction to statistical methods and data analysis. Belmont: Wadsworth.
Peever, J. H., & Stephenson. R. (1 997). Dav night differences in the respiratory response to hypercapnia in awake adult rats. Respiration Physiolog?.: 109. 24 1-248.
Phillipson. E. A. (1978). Control of breathing dunng sleep. Americari Revim of' Respiratory Disease. 1 18. 909-939.
Phillipson. E. A.. & Bowes, G. (Eds.). (1986). Control oj'breatliing during sleep. in N . S. Chemiack & I. G . Widdicombe (Eds.). Handbook of Physiolog- Section 3: 7he Respiratory Svstem (Vol. 2. pp. 649-690). Baltimore: American Physiological Society.
Phillipson, E. A.. Duffin. J.. & Cooper. J. D. ( 198 1). Critical dependence of respiratory rhythmicity on metabolic COz load. Journal oj* .4pplied Physiologv: Respirat,. Environ. Exercise Physiol.. .fO( 1 ). 15-54.
Phillipson, E. A.. Murphy. E.. & Kozar. L. F. (1976). Repulation of respiration in sleeping dogs. Journal ofApplied Physiolop. JO. 688-693.
Pittendrigh. C. S. ( 1960). Circadian rhythrns and the circadian organization of living systems. Cold Spring Harbor &mopsia on Quantitative Biologv. 25. 159- 1 84.
Ralph. M. R.. Foster. R. G.. Davis. F. C.. & Menaker. M. (1990). Transplanted suprachiasrnatic nucleus determines circadian penod. Science. 24 7.4 19-21.
Raschke. F.. & Moller. K. H. (1989). Untersuchungen m r Tagesrhythmik der Chemosensitivat und deren Beitrag ni nachtlichen Atmungsregulationsstorungen. Pneumologie. 43. 568-57 1.
Rechtsc haffen. A. ( 1998). Current perspectives on the hnction of sleep. Perspectii*es in Biolop and hfedicine. 41(3). 359-39 1.
Rec htsc haffen. A.. Berpann. B. M.. & Winter. J. B. ( 1983). Physiological correlates of prolonged sleep deprivation in rats. Scierice. 221. 182-1 84.
Rechtschaffen. A.. & Kales. A. (1968). .4 manual for standardised te~minologv. techniques. and scoring systeni for sleep stages oj' human subjects. Washington DC: Public Health Service. U S Govemment Printing Office.
Refinetti. R.. & Menaker. M. (199 1). The circadian rhythm of body temperature. Physiology and Behavioiir. 5 I(3). 6 1 3-63 7.
Remmen. .J. E.. Bartlett. D. J.. & Putman M.D. (1976). Changes in the respiratory cycle associated with sleep. Respiration Physiology 28. 227-238.
Saiki. C., & Monola, J. P. (1995). Hypoxia abolishes the moming - night differences of metabolism and ventilation in 6 day - old rats. Canadian Journal of Ph-vsiologr and Phonnocology. 73, 1 59- 1 64.
Seifert, E. L., Knowles, J.. & Mortola, J. P. (2000). Continuous circadian measurements of ventilation in behaving adult rais. Respiration Physiologv. 120(2), 179- 183.
Shea. S. A. (1996). Behavioural and arousal related influences in breathing in humans. E-perimental Phwiologv. 81, 1 -26.
Shiromani. P. J., Scammell, T.. Shenn. J. E.. & Saper. C. B. (1999). Hypothalamic regulation of sleep. In R. Lydic & H. Baghdoyan (Eds.). Handbook of- Behavioirral State Confrol: Cellular and :blolecular Mechanisms (pp. 3 1 1-325). Boca Raton: CRC Press LLC.
Smith, J. C.. Ellenberger. H., Ballanyi. K.. Richter. D. W.. & Feldman. J. L. (1991). Pre- Botzinger complex: a brain stem region that may generate respiratory rhythm in mammals. Science, 254. 726-729.
Spengler. C. M.. Czeisler. C. A.. & Shra. S. A. (2000). An endogenous circadian rhythm of respiratory control in humans. Journal o/'Phyiology, 5215(3). 683-694.
Spengler. C. M.. Oliver. H.. Czeisler. C. A., & Shea. S. A. ( 1997). Effects of circadian rhythms and sleep deprivation and respiratory control in humans. .herican Journal of*Respirato~y and Critical Care Medicine. 155. A777.
Stahel . C.D.. & Nicol. S.C. (1988). Cornpanson of barometnc and pneumotachographic measurements of resting ventilation in the litt le penguin (Eudyptula minor). Comparative Biochemistry and Physiolog~.. 89A(3). 387-390.
Stephenson. R. ( 1993). The contributions of body tissues. respiratory system. and plumage to buoyancy in waterfowl. Canadian Journal of Zooloy. 7 1 . 1 52 1 - 1 529.
Stephenson. R.. Mohan. R. M.. Duffin. J.. & Jarsky. T. M. (2000). Circadian rhythms in the c hemoreflex control of breathing. American Journal of Ph~~siolog-: Regdatory Integrative Comp. Physiol.. ? 78. R282-R286.
Stenade. M.. Datta. S.. Pare. D.. Oakson. G.. & Curr-Dossi. R. (1990). Neuronal activities in brain-stem cholinergie nuclei relaied to tonic activation processes in thalamoconical systems. Journal ofiVeuroscience. 10. 254 1-2559.
Steman. M. B.. & Clemente. C . D. (1962). Forebrain inhibitory rnechanisms: cortical desynchronisation induced by basal forebrain stimulation. Experirnental Neurologv. 6. 9 1 - 1 02.
Strohl, K. P., Thomas. A. J.. St Jean. P., Schlenker. E. H.. Koletsky, R. J.. & Schork. N. J. (1997). Ventilation and metabolism among rat strains. Journal of. Applied Physiolog 82(1). 3 17-323.
Szymusiak, R. (1995). Magnocellular nuclei of the basal forebrain: substrates of sleep and arousal regulation. Sleep. 18.478-500.
Szymusiak, R.. & McGinty. D. (1986). Sleep related neuronal discharge in the basal forebrain of cats. Brain Research. 3 70. 82-92.
Tenney. S. M.. & Boggs. D. F. ( 1986). Comparative rnammalian respiratory control. In W. J. G. Cherniack N.S. (Ed.). Handbook of' Physiologv. Section 3: 73e Respiratory $vsrem: Control of Breathitzg. part 2 (Vol. 2 . pp. 833-855). Baltimore: American Physiological Society.
Tobler. 1.. Borbely. A. A.. & Groos. G. ( 1983). The effect of sleep deprivation on sleep in rats with suprac hiasmatic lesions. lVeuroscience Lerlers. 42. 39-54.
Trachsel. L.. Tobler. 1.. & Borbelly. A. A. (1988). Electroencephalogram analysis of non rapid eye movement sleep in rats. American Journal of'Physiologv: Regulato~ Integrative Comparati~~e Physiologv. 255(24). R27-R3 7 .
Trachsel. L.. Tobler. 1.. & Borbely. A. A. ( 1986). Slerp regulation in rats: cffrcts of sleep deprivation. light and circadian phase. American Journal oj' Physiologv: Regulafoy hegrotive Comp. Physiol.. 2 j I , R 103 7-R 1044.
Wasserman. K.. Whipp. B. J.. & Casaburi. R. ( 1986). Respiratory control during exercise. in N. S. Chemiack & J. G . Widdicombe (Eds.). Handbook oj'Physiofogv Section 3: 7?ze Respiratory Svstem (Vol. 2. pp. 595-619). Baltimore: American Physiological Society.
Watts. A. (1991). The efferent projections of the suprachiasmatic nucleus: anatomical insights into the control of circadian rhythms. In D. C. Klein. R. Y. Moore. & S. T. Reppert (Eds.). In the Suprachiasmatic ~Vuc1etr.s - nie illind's clock (pp. 77- 106). New York: Oxford.
Webster. H. H.. & Jones. B. (1988). Neurotoxic lesions of the dorsolatera1 pontomesencephalic tegmentum-cholinergic ce11 area in the cat: III Effects upon sleep-waking States. Bruin Research. 458. 285-302.
White. D. P.. Weil. J. V.. & Zwillich. C. W. (1985). ~Metabolic rate and breathing during sleep. Journal ofrlpplied Physiologv, 59. 3 84-39 1.
Woodin. M.. & Stephenson. R. (1998). Circadian rhythms in diving behaviour and ventilatory response to asphyxia in canvasback ducks. Americon Journal oj' Physiologv. 2 74(43). R686-R693.
Zepelin, H. (1994). Marnmalian sleep. in M. H. Kryger. T. Roth. & W. C. Dement (Eds.), Principles and practice of sleep nzedicine (2nd ed.. pp. 69-80). Philadelphia: WB Saunders.