Role of the Catecholamine and Limbic Systems in Narcolepsy/Cataplexy

Transcription

1 Role of the Catecholamine and Limbic Systems in Narcolepsy/Cataplexy by Christian Richard Burgess A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Cell and Systems Biology University of Toronto Copyright by Christian Richard Burgess 2012

2 Role of the catecholamine and limbic systems in narcolepsy/cataplexy Doctor of Philosophy, 2012 Christian Richard Burgess Graduate Department of Cell & Systems Biology University of Toronto Abstract In this thesis I investigated the neural circuits that trigger cataplexy in mice. Specifically, I first addressed the theory that cataplexy is a REM sleep disorder. I then investigated a role for the noradrenergic and dopaminergic systems in murine cataplexy. Finally, I addressed the role of the amygdala in triggering cataplexy. From this work several specific conclusions can be drawn: 1. Cataplexy does not share a common executive mechanism with REM sleep, although the two may share a common mechanism that generates muscle atonia. Muscle tone during REM sleep and cataplexy is similar, however increasing REM sleep pressure does not increase cataplexy and positive affective stimuli that can increase cataplexy tend to decrease REM sleep. 2. Systemic manipulation of dopamine receptors can modulate cataplexy without affecting behavioral state. Specifically, manipulation of D2-like dopamine receptors at specific doses can modulate cataplexy while having no effect on sleep-wake state or sleep attacks, and manipulation of D1-like receptors potently affects sleep-wake state and sleep attacks without affecting cataplexy. ii

3 3. Systemic modulation of noradrenergic activity in orexin KO mice is sufficient to modulate cataplexy. Specifically, activation of excitatory α 1 receptors reduces the occurrence of cataplexy while blockade of these receptors exacerbates it. 4. Withdrawal of an endogenous α 1 -mediated noradrenergic drive from motor neurons during wakefulness contributed to the loss of muscle tone during cataplexy. Reestablishing this excitatory drive exogenously alleviated cataplexy-dependant muscle atonia. 5. The amygdala is a critical part of the neural mechanism that triggers cataplexy in orexin KO mice. Ablation of the amygdala resulted in significant decreases in both baseline cataplexy and emotionally-induced cataplexy. The amygdala may trigger cataplexy through direct projections to brainstem areas that regulate muscle atonia. iii

4 Acknowledgments I would first like to thank Dr. John Peever. John gave me an opportunity to pursue research and has been a great supervisor and mentor over the last 7 years. The work in this thesis would not be possible without his guidance. Thanks to all the members of the Peever lab: Zoltan, Jimmy, Peter, Nicole, Lauren, Paul, Arash, Jenn, Sharshi, Angie, Simon, Saba and Gavin. I would like to single out Dr. Patti Brooks who was not only a great colleague but also a wonderful friend and roommate throughout my time in the Peever lab, this thesis would not have been possible without her. Approximately one third of the work in this thesis was completed in the department of Neurology, Harvard Medical School, at Beth Israel Deaconess Medical Center. I would like to thank Dr. Tom Scammell for the opportunity to study in his lab; he provided great mentorship throughout our collaboration. I would also like to thank the other members of the Scammell lab: Takatoshi, Mihoko, Daniel, Chloe, Phillip and LJ for their help and friendship during my time in Boston. I thank my supervisory committee, Dr. Vince Tropepe and Dr. Richard Stephenson, for advice and guidance throughout my PhD. I also thank Dr. Barry Sessle, Dr. Junchul Kim, and Dr. Chris Leonard for serving on my PhD thesis defense committee and Peggy Salmon, Ian Buglass, Tamar Mamourian, Catherine Siu and Nalini Dominique for their assistance throughout my PhD. Thanks also to my parents, Tony and Camilla, my siblings, Erin, Robert, Elna, Shaina, Marc, Andrew and April, my nephews, Luka and Ellington, and my best friend, Egan, for their love and support. iv

5 Table of Contents Acknowledgments... iv Table of Contents...v List of Figures... xi List of Abbreviations... xiv Chapter 1: Introduction Narcolepsy Cataplexy Etiology Treatment Animal models of narcolepsy/cataplexy Sleep states The orexin system Neurobiology of cataplexy The dopaminergic system The noradrenergic system The amygdala Other transmitter systems Cataplexy as a manifestation of REM sleep Current model of cataplexy Thesis overview...38 Chapter 2: Methods Mice Surgical procedures Sleep recording...41 v

6 2.4 Behavioral state analysis Histology...42 Chapter 3: REM sleep and Cataplexy are generated by Independent Mechanisms Abstract Introduction Methods Animals Surgery Electrophysiological recordings REM sleep deprivation protocol Wheel running and chocolate protocol Ultrasonic vocalizations and social reunion paradigm Statistics Results Muscles exhibit atonia during both REM sleep and cataplexy Putative positive emotions trigger cataplexy in mice Stimulating environments increase cataplexy but decrease REM sleep Increasing REM sleep pressure does not affect cataplexy Discussion REM sleep and cataplexy are similar states Muscle tone during REM sleep and cataplexy is similar Positive affective stimuli induce cataplexy and suppress REM sleep in mice REM sleep pressure does not significantly increase cataplexy REM sleep and cataplexy do not share a common executive mechanism...68 Chapter 4: Dopaminergic Regulation of Sleep and Cataplexy Abstract...71 vi

7 4.2 Introduction Methods Animals Surgery Drug preparation Data acquisition Experimental protocols Data analysis Statistical analysis Results Orexin KO mice exhibit cataplexy and sleep attacks Amphetamine reduced cataplexy and sleep attacks in narcoleptic mice D1-like receptors modulate sleep attacks but not cataplexy D2-like receptors modulate cataplexy but not sleep attacks Discussion Amphetamine alleviates cataplexy and sleep attacks A D2-like receptor mechanism modulates cataplexy A D1-like receptor mechanism modulates sleep attacks Physiological significance...90 Chapter 5: Noradrenergic Regulation of Cataplexy Abstract Introduction Methods Animals Surgical preparation Experimental procedures for sleep and microdialysis studies...95 vii

8 5.3.4 Experimental paradigm Verification of probe location Data analysis Statistical analysis Results Focal activation of α 1 receptors on trigeminal motor neurons increased masseter EMG activity in freely behaving mice Masseter muscles experienced atonia during cataplexy Cataplexy is affected by changes in noradrenergic activity Drug manipulations targeted trigeminal motor neurons Loss of noradrenergic drive to motor neurons is not sufficient for triggering cataplexy Restoration of noradrenergic activity increased muscle tone during cataplexy Discussion The noradrenergic system regulates cataplexy Withdrawal of noradrenergic drive promotes cataplexy Noradrenaline could act at REM sleep generating sites to modulate cataplexy Other circuits involved in cataplexy Conclusions Chapter 6: Role for the Amygdala in Triggering Cataplexy Abstract Introduction Methods Animals Surgery Experimental protocol viii

9 6.3.4 Data acquisition and analysis Histology Anterograde tracing Statistical analysis Results The amygdala is anatomically well positioned to regulate cataplexy Amygdala lesions reduced cataplexy under baseline conditions Amygdala lesions decreased cataplexy triggered by a positive stimulus Amygdala lesions decreased cataplexy triggered by a strong positive stimulus Discussion The amygdala is an important part of the cataplexy inducing circuitry Interventions affected sleep-wake behavior Conclusions Chapter 7: Discussion Overview Cataplexy as a REM sleep phenomenon Dopaminergic regulation of cataplexy Noradrenergic regulation of cataplexy The role of the amygdala in triggering cataplexy General model of the mechanisms underlying cataplexy Limitations Future directions Summary References ix

10 List of Tables Table 6-1: Sleep-wake architecture in amygdala-lesioned mice x

11 List of Figures Figure 1.1: Raw traces of EMG activity during an episode of cataplexy Figure 1.2: Sleep-wake regulation in rodents Figure 1.3: Brainstem mechanisms regulating REM sleep atonia Figure 1.4: Orexin projections Figure 1.5: Dopamine projections Figure 1.6: Noradrenaline projections Figure 1.7: A simplified schematic of major inputs and outputs of the amygdala Figure 1.8: Model of how the orexin (hypocretin) system could prevent the loss of muscle tone in response to positive emotions Figure 3.1: REM sleep and cataplexy occur in similar amounts during the dark period Figure 3.2: Muscle tone during REM sleep and cataplexy Figure 3.3: USVs are associated with cataplexy in orexin KO mice Figure 3.4: Wheel running and chocolate increased cataplexy and decreased REM sleep Figure 3.5: Stimulating environments increased waking and decreased NREM sleep during the dark period Figure 3.6: Selective REM sleep deprivation reduced both NREM and REM sleep Figure 3.7: REM sleep deprivation had no effect on behavioral state over the following dark period Figure 3.8: REM sleep accrued a deficit during the deprivation that was in part recovered over the following dark period xi

12 Figure 3.9: Cataplexy was reduced at the beginning of the dark period following deprivation Figure 3.10: Wakefulness during the dark period was not significantly affected by REM sleep deprivation Figure 3.11: Cataplexy and REM sleep are generated by different mechanisms Figure 4.1: Cataplexy, sleep attacks and sleep-wake behavior in narcoleptic mice Figure 4.2: Amphetamine decreased sleep, sleep attacks and cataplexy Figure 4.3: Inactivation of D1-like receptors increased sleep attacks Figure 4.4: Activation of D1-like receptors decreased sleep attacks Figure 4.5: Activation of D2-like receptors increased cataplexy Figure 4.6: Blockade of D2-like receptors decreased cataplexy Figure 5.1: Microdialysis probe insertion into the trigeminal motor nucleus Figure 5.2: Phenylephrine application increased muscle tone in freely behaving mice Figure 5.3: Muscles experienced atonia during cataplexy Figure 5.4: Cataplexy is sensitive to changes in noradrenergic tone Figure 5.5: Drug manipulations targeted trigeminal motor neurons Figure 5.6: Targeted drug manipulations did not affect sleep-wake architecture Figure 5.7: Targeted drug manipulations did not influence cataplexy Figure 5.8: Loss of noradrenergic drive contributes to muscle atonia during cataplexy Figure 5.9: Stimulating α 1 receptors on motor neurons increased muscle tone during sleep in orexin KO mice xii

13 Figure 5.10: Activation of α 1 receptors on motor neurons elevated muscle tone during cataplexy Figure 6.1: The central nucleus of the amygdala projects to brainstem regions that regulate REM sleep Figure 6.2: Excitotoxic lesions of the amygdala Figure 6.3: Amygdala lesions reduce cataplexy in orexin KO mice Figure 6.4: Amygdala lesions reduced cataplexy in orexin KO mice even when accounting for changes in wakefulness Figure 6.5: Hypothesized model of the neural pathways through which positive emotions trigger cataplexy Figure 7.1: Dopaminergic control of cataplexy Figure 7.2: Noradrenergic control of cataplexy Figure 7.3: Amygdaloid control of cataplexy Figure 7.4: General schematic of cataplexy mechanisms xiii

14 List of Abbreviations ºC degree Celsius 5-HT serotonin acsf artificial cerebral spinal fluid AMPA α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid A.U. arbitrary units AW active wake BF basal forebrain BLA basolateral nucleus of the amygdala CeA central nucleus of the amygdala ChAT choline acetyl-transferase CNS central nervous system DA dopamine DAT dopamine transporter DR dorsal raphe DREADDS designer receptors exclusively activated by designer drugs EEG electroencephalogram EMG electromyogram g grams GABA γ-aminobutryic acid GHB γ-hydroxybuterate h hours Hz hertz ICV Intracerebral ventricular k kilo KO knockout L litre LC locus coeruleus LDT lateral dorsal tegmentum LH lateral hypothalamus LM left masseter i.p. intraperitoneal u micro LPT lateral pontine tegmentum m metre M molar MCH melanin concentrating hormone NA noradrenaline xiv

15 NARP neuronal activity regulated pentraxin NET noradrenaline transporter NPY neuropeptide Y NREM non-rapid eye movement OX 1 R orexin receptor 1 OX 2 R orexin receptor 2 PE phenylephrine PFC prefrontal cortex PPT pedunculopontine tegmentum QW quiet wake REM rapid eye movement RM right masseter s seconds SEM standard error of the mean SLD sublateraldorsal nucleus SN substantia nigra TER terazosin TMN tuberomammillary nucleus USV ultrasonic vocalization vlpag ventrolateral periaquaductal gray VLPO ventral lateral preoptic area VMM ventral medial medulla VTA ventral tegmental area WR wheel running xv

16 Chapter 1: Introduction 1

17 2 Chapter 1: Introduction 1.1 Narcolepsy Narcolepsy is a debilitating sleep disorder with symptoms including excessive daytime sleepiness, fragmented nighttime sleep, sleep paralysis, hypnogogic hallucinations, and episodes of muscle atonia during wakefulness (Gelineau, 1880). This combination of hypersomnolence and episodes of muscle weakness was first described by Westphal in 1877 and given its name by Gelineau in 1880 (Gelineau, 1880, Mignot, 2001, Schenck et al., 2007). Loewenfeld, in 1902, termed these periods of muscle atonia cataplexy (Mignot, 2001). The prevalence of narcolepsy is approximately 1 in 2000 individuals, although this can be variable depending on ethnicity and method of estimation (Mignot, 1998, Overeem et al., 2001). Narcolepsy onset is most common during adolescence but can occur throughout life, including in childhood. Excessive daytime sleepiness is generally the first symptom to be recognized, with cataplexy developing later (Overeem et al., 2011). Narcolepsy symptoms can be treated with moderate efficacy but persist throughout and affects patients quality of life Cataplexy The majority of the work presented in this thesis investigates the most unique symptom of narcolepsy: cataplexy. It is the sudden loss of postural muscle tone during waking that can result in full postural collapse and last from a few seconds to several minutes; patients typically remain conscious during cataplexy. Loss of muscle tone can be complete or partial, and most often affects the muscles of the face, neck and legs, although all muscles can be affected even in a single episode (Guilleminault et al., 1974, Overeem et al., 2011). Cataplexy can occur at any time while the patient is awake although it is most often elicited by positive emotions, with laughing excitedly being one of the best triggers in most patients; however, other triggers including anger, fear and stress can also elicit cataplexy in human narcoleptics (Nishino and Kanbayashi, 2005, Overeem et al., 2011). This can be debilitating for human patients with narcolepsy and can impair their ability to perform even relatively simple tasks, such as driving (Guilleminault et al., 1974).

18 3 With the exception of symptomatic cataplexy (i.e. episodes that fit the definition of cataplexy but are associated with other abnormalities) caused by damage to hypothalamic or brainstem regions, cataplexy is only seen in narcolepsy (Nishino and Kanbayashi, 2005). However, cataplexy only occurs in ~70% of narcoleptics (Sasai et al., 2009). It is unclear whether narcolepsy with and without cataplexy are caused by different mechanisms but there is some evidence to suggest this is the case; narcoleptics with cataplexy have almost undetectable levels of orexin in their cerebral spinal fluid, while those without cataplexy often have normal or only slightly suppressed levels (Nishino et al., 2001, Ripley et al., 2001, Bassetti et al., 2010). Because cataplexy onset is sudden and emotionally-triggered it has been hypothesized that changes in autonomic activity may have a role in triggering it (Guilleminault et al., 1986). Human patients with narcolepsy show no autonomic changes at cataplexy onset, although they do show a drop in heart rate and a slight increase in blood pressure during cataplexy (Donadio et al., 2008). During cataplexy patients continue to breathe although the breathing is interrupted by periods of apnea (Guilleminault et al., 1974). Electroencephalographic (EEG) activity is wakelike during an episode of cataplexy, consistent with the patient maintaining consciousness, while electromyographic (EMG) activity is reduced, consistent with the loss of muscle tone (Guilleminault et al., 1974, Rubboli et al., 2000) (Figure 1.1). In addition to reduced muscle tone, the H-reflex is abolished during cataplexy in human patients (Guilleminault et al., 1998). The H-reflex is a reaction of a muscle after electrical stimulation of sensory fibers in the nerves that innervate the muscle. The H-reflex is present and normal during waking in narcoleptics but is abolished during cataplexy and REM sleep (Stahl et al., 1980, Overeem et al., 1999, Overeem et al., 2004). Cataplexy is thought to be the manifestation of REM sleep during waking, as REM sleep is the only state where muscle atonia is seen in healthy humans (Rechtschaffen and Dement, 1967). Many of the observations regarding cataplexy mentioned above support this theory; however, it does not account for certain aspects of cataplexy, including the fact that it is emotionally induced. Alternative theories exist including that cataplexy is an atavistic expression of tonic immobility, a defense mechanism present in some animals that involves loss of muscle tone (Overeem et al., 2002). The mechanisms that trigger cataplexy are unclear.

19 4 Figure 1.1: Raw traces of EMG activity during an episode of cataplexy. Muscle tone is abruptly lost at cataplexy onset in the mouse (top), dog (middle) and human (bottom). Cataplexy episodes affect most postural muscles and at the end of an episode muscle tone is quickly regained. (Neck, cervical portion of the trapezius; Delt., deltoid; W.Ext., wrist extensor; APB, abductor pollicis brevis; Parasp., paraspinal musculature at the level of the insertion of the twelfth rib; Ret. Abd., rectus abdominis; Quadr., quadriceps; Tib.A., tibialis anterior; R, right; L, left.) Modified from (Rubboli et al., 2000) and (Wu et al., 1999)

20 Etiology Narcolepsy was first linked with posterior hypothalamic abnormalities by Constantin von Economo in 1917 (Mignot, 2001). He made this assertion based on an outbreak of encephalitis lethargica that resulted in hypothalamic damage and in many cases induced hypersomnolence and cataplexy. Clinical research in human narcolepsy and basic research using a canine model of narcolepsy began to elucidate the mechanisms responsible for the symptoms of the disorder but the neural mechanism from which narcolepsy results remained unclear (Nishino et al., 1994, Guilleminault et al., 1998). In 1998, two labs independently discovered a pair of hypothalamic neuropeptides termed orexin A and B (or hypocretin 1 and 2), which have since been demonstrated to have a key role in the regulation of wakefulness (de Lecea et al., 1998, Sakurai et al., 1998). The following year another pair of discoveries implicated these peptides in the etiology of narcolepsy with cataplexy. First, Dr. Yanagisawa s group at the University of Texas Southwestern produced an orexin ligand knockout (KO) mouse which expressed a phenotype strikingly similar to narcolepsy, including sudden episodes of complete muscle atonia (Chemelli et al., 1999). Then Dr. Mignot s group at Stanford demonstrated that the canine model of narcolepsy, which had been an important model for research into the neurobiology of narcolepsy for the previous 20 years, resulted from a mutated orexin receptor (Lin et al., 1999). A link between the orexin system and human narcolepsy followed when two publications independently demonstrated the loss of orexin neurons in narcoleptic patients using post-mortem immunostaining (Peyron et al., 2000, Thannickal et al., 2000). Further research has demonstrated that orexin neurons are lost in human narcolepsy with cataplexy, rather than an inability to produce orexin peptides. For example, gliosis has been demonstrated in the region of orexin neurons (Thannickal et al., 2003) and neurotransmitters colocalized with orexin are also lost in narcolepsy, while chemically identified non-orexin neurons remain unaffected (Blouin et al., 2005, Crocker et al., 2005). This evidence strongly suggests that loss of hypothalamic orexin neurons underlies human narcolepsy. The process of orexin neuron loss is thought to be autoimmune in nature. An association with the human leukocyte antigen gene HLA DQB1*0602 suggests an autoimmune origin (Carlander et al., 1993, Mignot et al., 1994, Dolenc-Grosel and Vodusek, 2002). Gene linkage studies have also found associations with other immune related genes, such as T-cell receptor α,

21 6 supporting this assertion (Hallmayer et al., 2009). Recently, an increase in narcolepsy prevalence has been demonstrated in several populations, and has been linked with the H1N1 pandemic (Dauvilliers et al., 2010, Bardage et al., 2011, Han et al., 2011, Marcus, 2011, Partinen et al., 2012). It is unknown whether the flu itself, the vaccine, or an unrelated trigger has caused this increase but further investigation of the phenomenon may shed light on the autoimmune nature of orexin cell loss. The extensive and selective nature of orexin neuron loss in narcolepsy strongly suggests an autoimmune origin, rather than an alternative explanation such as neurodegeneration or neurotoxicity, however there is no definitive evidence for this (Black et al., 2002, Scammell, 2006). It is unknown how the loss of orexin neurons leads to the symptoms of narcolepsy; however, investigations into the normal role for the orexin system, the underlying neurobiology of narcolepsy, and treatments for narcolepsy have all provided useful contributions to our understanding of this disorder Treatment Narcolepsy has traditionally been treated with a combination of stimulants for excessive daytime sleepiness and tricyclic antidepressants for cataplexy (Ahmed and Thorpy, 2010). Stimulants such as methamphetamine and methylphenidate effectively manage sleepiness by increasing release and decreasing reuptake of monoamines (Nishino et al., 1998b, Kanbayashi et al., 2000, Wisor et al., 2001). Modafinil is an effective treatment for sleepiness in narcolepsy but the mechanism by it which increases alertness is unclear. Recent evidence suggests it may act via a dopaminergic mechanism (Scammell and Matheson, 1998, Wisor et al., 2001, Golicki et al., 2010). Clomipramine and imipramine have been used to effectively treat cataplexy, likely by reducing noradrenergic uptake (Black and Guilleminault, 2001). Sodium oxybate, the sodium salt of gamma-hydroxybutyrate (GHB), is the only effective treatment for excessive sleepiness, cataplexy, and fragmented sleep (Owen, 2008, Huang and Guilleminault, 2009, Poryazova et al., 2011). Sodium oxybate has been hypothesized to act via activation of GABA B receptors and perhaps also through activation of GHB receptors, however the precise mechanism by which it treats narcolepsy is not known; recent evidence suggests it may too act via a dopaminergic mechanism (Howard and Feigenbaum, 1997, Schmidt-Mutter et al., 1999, Thorpy, 2005, Owen, 2008, Huang and Guilleminault, 2009).

22 7 Recent novel approaches for narcolepsy treatment include histamine H3 receptor inverseagonists. H3 inverse-agonists presumably act at auto-receptors on histamine neurons to increase histamine release, thereby increasing arousal. H3 inverse agonists have been demonstrated to be effective for the treatment of excessive sleepiness in narcoleptic patients, dogs and mice but further testing is required to evaluate their efficacy in treating cataplexy (Parmentier et al., 2007, Guo et al., 2009, Lin et al., 2011, Inocente et al., 2012). Since narcolepsy results from selective loss of orexin neurons, restoration of orexinergic signaling could be the most effective treatment. Orexin receptor agonists, cell transplant, or gene therapy are all candidate therapies for narcolepsy. Intracerebral ventricular (ICV) application of an orexin receptor agonist in orexin KO mice resulted in increased wakefulness and reduced cataplexy establishing a strong rationale for development of these drugs for human patients (Mieda et al., 2004). Some small molecule agonists have been developed but they cannot cross the blood brain barrier, while orexin peptides themselves have limited therapeutic potential due to their short half-life (Deadwyler et al., 2007). A nasal spray application technique has been tested and has shown promising preliminary results in humans and non-human primates (Deadwyler et al., 2007, Baier et al., 2011). Orexin gene therapy has been tested in orexin KO mice with some promising results. Inducing orexin expression in the hypothalamus and zona incerta, using viral vectors, reduced symptoms of narcolepsy (Liu et al., 2008, Liu et al., 2011a). Cell transplant and gene therapy in humans each produce significant challenges that could complicate their effectiveness as a long-term treatment of narcolepsy (Arias-Carrion and Murillo-Rodriguez, 2009). Much of what is known of the underlying neurobiology of cataplexy is derived from the treatments that effectively alleviate this symptom of narcolepsy Animal models of narcolepsy/cataplexy There are a number of both naturally occurring and engineered animal models of narcolepsy with which to study the neurobiology of cataplexy (Scammell et al., 2009). These models all have good face validity (they demonstrate sleepiness and cataplexy), predictive validity (human narcolepsy treatments reduce symptoms in animal models), and construct validity (lack of a functional orexin system). The first animal model of narcolepsy to be studied was the naturally occurring canine model, with a mutation in an orexin receptor resulting in sleepiness and severe cataplexy elicited by social interaction and palatable food (Mitler, 1975,

23 8 Mitler et al., 1976, Darke and Jessen, 1977, Mitler and Dement, 1977). Since the discovery of the orexin peptides, genetically engineered mouse models of narcolepsy have been developed (Chemelli et al., 1999). Orexin ligand KO, receptor KO, constitutive receptor KO, and orexinneuron ablated models all demonstrate sleepiness and cataplexy (Hara et al., 2001, Willie et al., 2003, Mochizuki et al., 2004, Kantor et al., 2009, Mochizuki et al., 2011). Studies in this thesis utilized an orexin KO mouse model of narcolepsy to study the underlying neurobiology of cataplexy and hypersomnolence. We chose this model as it was well characterized, had clear episodes of cataplexy, and closely mimicked the human narcolepsy phenotype (Mochizuki et al., 2004). Details of the production of these mice can be found in Chemelli et al. (1999). In brief, a specific targeting vector was used to replace exon 1 of the prepro-orexin gene, the precursor gene for both orexin peptides. The founder mice for the colony used in these studies were obtained from the Yanagisawa lab at the University of Texas Southwestern Medical Center, having already been back-crossed to a C57B6/J background. Chemelli et al. (1999) and later Mochizuki et al. (2004) thoroughly characterized the narcolepsy phenotype in these mice, including chronic sleepiness, sleep attacks and cataplexy (Chemelli et al., 1999, Mochizuki et al., 2004). These mice have been used extensively to investigate both narcolepsy and the normal physiological role of orexin peptides, and serve as a useful model to investigate the underlying neurobiology of narcolepsy/cataplexy Sleep states As narcolepsy is a disorder of sleep and wakefulness, it is important to know some of the basic features of sleep in order to understand the symptoms of narcolepsy. Sleep is a period of relative quiescence that has been observed in all animal species yet investigated. Sleep is thought to serve an important function although it remains unclear what function, or functions, sleep is required for. Sleep appears to be homeostatically regulated (a sleep debt is accrued during waking that dissipates during the next sleep phase), supporting the assertion that it is critical to certain biological processes (Leemburg et al., 2010). It is a combination of circadian and homeostatic processes that regulate the sleep-wake cycle (Borbely, 1982). As this thesis focuses on investigations in mice, the remainder of this section will be specific to rodent sleep. Sleep consists of two separate states: non-rapid eye movement (NREM) and rapid eye movement (REM), or paradoxical sleep. Under normal conditions, REM sleep follows NREM

24 9 sleep and the sleep cycle has a duration between 5 and 15 minutes. There are several brain nuclei that promote wakefulness and arousal. During waking, the noradrenergic, serotonergic, and cholinergic brainstem neurons, dopaminergic, orexinergic, and histaminergic midbrain/hypothalamic neurons, and cholinergic, GABAergic basal forebrain neurons actively promote arousal through projections to the thalamus and/or cortex. During sleep, a collection of GABAergic neurons in the ventrolateral preoptic area (VLPO) inhibit these arousal centers (Saper et al., 2010) (Figure 1.2). During REM sleep, melanin concentrating hormone (MCH) neurons in the hypothalamus and cholinergic neurons in the brainstem trigger a REM sleep switch in the pons that permits entrance into this state (Saper et al., 2010). There are still some aspects of these mechanisms that remain unclear but this model of sleep-state control provides the most current understanding of the neurobiology of sleep. A notable difference between NREM and REM sleep is the level of muscle tone. Muscle tone has a stereotypical pattern across the sleep-wake cycle: it is generally high during waking, reduced during NREM sleep, and further reduced during REM sleep (Burgess et al., 2008). During REM sleep, the complete lack of muscle tone is termed REM sleep atonia, and can be punctuated by brief muscle twitches. The underlying neurobiology regulating REM sleep atonia in rodents has been an area of intense investigation (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b) (Figure 1.3). Motor neurons are actively inhibited during REM sleep by GABAergic/glycinergic pre-motor neurons in the spinal cord and medial medulla. These premotor neurons receive excitatory glutamatergic inputs from the sublateral dorsal nucleus (SLD) of the pons, which becomes disinhibited during REM sleep. Lesions of this area can cause REM sleep without atonia, and may contribute to sleep-related movement disorders such as REM sleep behavior disorder (Lu et al., 2006b). When muscle tone is required, the SLD is inhibited by GABAergic neurons in the ventrolateral periaquaductal gray/lateral pontine tegmentum (vlpag/lpt) (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b). During the different sleep/wake states this pontine switch receives modulatory inputs from monoaminergic/orexinergic neurons (that promote muscle tone) or cholinergic/mch neurons (that promote muscle atonia). Cataplexy and REM sleep both involve muscle atonia but it is unclear whether these mechanisms are responsible for both states, or a unique atonia-generating circuitry underlies cataplexy.

25 10 Figure 1.2: Sleep-wake regulation in rodents During waking brainstem, midbrain, hypothalamic, and basal forebrain nuclei are active (top). These areas provide excitatory input to the thalamus and cortex to promote arousal and wakefulness. When the mouse enters sleep, GABAergic neurons in the ventrolateral preoptic area become active (bottom). These neurons inhibit arousal promoting wake-active neurons to permit the entrance into sleep. (Ach, acetylcholine; OX, orexin; HA, histamine, DA; dopamine; 5-HT, serotonin; NA, noradrenaline)

26 11 Figure 1.3: Brainstem mechanisms regulating REM sleep atonia A. During normal waking, arousal-promoting nuclei directly and indirectly inhibit the sublateral dorsal nucleus (SLD) to suppress muscle atonia. The orexin system has a number of projections through which it promotes muscle tone. B. During REM sleep, monoaminergic and orexin nuclei are inhibited, thus disinhibiting the SLD. In addition, REM sleep-promoting systems such as acetylcholine and MCH may activate the SLD. The SLD then activates inhibitory pre-motor neurons in the medulla and spinal cord which inhibit motor neurons leading to muscle atonia. (NA, noradrenaline; 5-HT, serotonin; vlpag, ventrolateral periaquaductal gray; LPT, lateral pontine tegmentum; SLD, sublateral dorsal nucleus; MM, medial medulla; Ach, acetylcholine; MCH, melanin-concentrating hormone; Green arrows represent excitatory pathways; Red lines represent inhibitory pathways)

27 The orexin system The orexin system is composed of two peptides (orexin A and B) derived from a common precursor peptide (prepro-orexin) and two G-protein coupled receptors (OX 1 R and OX 2 R) (de Lecea et al., 1998, Sakurai et al., 1998). Orexin receptor binding increases intracellular Ca 2+ predominantly by a G q protein coupled receptor mechanism. Orexin A binds equally to both receptors while orexin B shows 10-fold greater affinity for OX 2 R, both peptides contribute to the normal physiological role for the orexin system (de Lecea et al., 1998, Sakurai et al., 1998). The general term orexin will be used to refer to both ligands in this thesis. Orexin neurons also contain other neurotransmitters, including glutamate and dynorphin (Chou et al., 2001). The normal physiological role of these transmitters or whether they have a role in narcolepsy/cataplexy is unclear. Orexin is only synthesized in neurons of the lateral hypothalamus but these neurons have widespread projections throughout the central nervous system (CNS). Orexinergic axons and synapses are present throughout the cortex, forebrain, amygdala, hypothalamus, and brainstem (Peyron et al., 1998) (Figure 1.4). The orexin system densely innervates many brain areas that are important in sleep and arousal, including sending its densest projections to the noradrenergic locus coeruleus (LC) where orexin release has excitatory effects that promote behavioral arousal (Peyron et al., 1998, van den Pol et al., 2002, Espana et al., 2005, Chen et al., 2010). In addition orexin neurons innervate several other regulators of sleep-wake state, including the serotonergic dorsal raphe (DR) and cholinergic basal forebrain/ldt/ppt. Orexin neurons also receive inputs from various brain areas that are important for the control of behavioral arousal. Several of these inputs are from sleep-promoting regions (e.g.: the VLPO) or are reciprocal projections from wake promoting regions (e.g.: BF, DR, and LDT) that could function to promote the maintenance of behavioral states (Sakurai et al., 2005). The orexin system also projects to a number of other sites that regulate different types of arousal. Projections to the ventral tegmental area (VTA) where orexin peptides activate both dopaminergic and non-dopaminergic neurons, could modulate stimuli-specific arousal, reward and motivation as the VTA has a role in motivation and processing emotionally salient stimuli (Nakamura et al., 2000, Narita et al., 2006). Also of note is the demonstration that orexin

28 13 neurons directly innervate both spinal and cranial motor neurons (Yamuy et al., 2004, McGregor et al., 2005). Orexin receptors have a broad expression pattern throughout the brain (Marcus et al., 2001). In situ hybridization for orexin receptor mrna has demonstrated widespread and sometimes non-overlapping expression for both known receptors. OX 1 R is highly expressed in the brainstem noradrenergic and cholinergic neurons and the amygdala, among other areas. OX 2 R is highly expressed in the cortex, paraventricular hypothalamus and ventral hypothalamus, among other areas (Marcus et al., 2001). The widespread distribution of orexin receptors and projections suggests a number of functions for the orexin system. One function of the orexin system is to establish and maintain a waking state. The finding that loss of orexin neurons leads to narcolepsy is a clear indication that the orexin system has a role in sleep-wake regulation and arousal (Peyron et al., 2000, Thannickal et al., 2000). Studies investigating the normal role of the orexin system have demonstrated that orexin is a key regulator of behavioral state. Both gain and loss of function experiments have demonstrated that orexin promotes transitions into and maintenance of wakefulness. Mice lacking a functional orexin system (i.e. orexin ligand knockout mice) demonstrate behavioral state instability and the inability to maintain long waking bouts (Mochizuki et al., 2004). Application of orexin peptides into the brain of mice can increase waking and specifically the length of waking bouts, while orexin receptor antagonists can induce sleepiness (Mieda et al., 2004, Neubauer, 2010).

29 14 Figure 1.4: Orexin projections While orexin neurons are exclusively located in the lateral hypothalamus they project widely throughout the brain and spinal cord. Projections to arousal-related nuclei allow the orexin system to regulate sleep-wake state. In addition, the orexin neurons project to the cortex, thalamus, and may be auto-excitatory. (5-HT, serotonin; DA, dopamine; HA, histamine; NA, noradrenaline; Ach, acetylcholine; GABA, γ-aminobutryic acid)

30 15 Recently, more selective techniques have confirmed a role for the orexin system in wakefulness. Activation of excitatory designer receptors exclusively activated by designer drugs (DREADDs; exogenous receptors selectively expressed on orexin neurons) on orexin neurons results in increased wakefulness; while activation of inhibitory DREADDs on orexin neurons suppressed wakefulness (Sasaki et al., 2011). Optogenetic targeting of orexin neurons, which is both highly selective and acting at short temporal scales, suggests a role in switching from sleep to wake. Optogenetic stimulation of orexin neurons in rats resulted in an increased probability of arousal from both NREM and REM sleep; while optogenetic inhibition of orexin neurons in mice resulted in short latency entrances into NREM sleep (Adamantidis et al., 2007, Tsunematsu et al., 2011). The orexin system has a key role in regulating sleep-wake state; the perturbation of this system leads to narcolepsy, an imbalance between sleep-wake states, and is a useful tool to study the mechanisms and pathways that normally balance behavioral state. The orexin system also has a role in motor control. Orexin neuron activity and peptide release is highest during periods of movement in rats (Kiyashchenko et al., 2002, Lee et al., 2005, Mileykovskiy et al., 2005). Orexin neurons project directly to both cranial and spinal motor neurons and also to subcortical motor-related structures, and administration of orexin can increase locomotion in rodents (Peyron et al., 1998, Thorpe and Kotz, 2005, Samson et al., 2010). Orexin receptor activation directly on motor neurons has an excitatory effect; it has been demonstrated to act both pre and post-synaptically to increase motor neuron excitability (Peever et al., 2003, Yamuy et al., 2004, McGregor et al., 2005). Many symptoms of narcolepsy involve a decoupling of motor control and behavioral state. In healthy humans, motor activity is high during waking, suppressed during sleep, and further suppressed during REM sleep muscle atonia. During cataplexy and sleep paralysis, atonia occurs during wakefulness. Narcoleptics also have an increased prevalence of sleep-related movement disorders such as REM sleep behavior disorder, which is increased movement and muscle tone during REM sleep (Wierzbicka et al., 2009, Franceschini et al., 2011). While the orexin system promotes waking and motor activity, these findings suggest that the orexin system may help to establish a balance between behavioral state and proper motor control. In the absence of this stabilizer narcolepsy develops. While the location of the orexin neuron field (in the lateral hypothalamus) and some early studies suggested a role for orexin in feeding, subsequent studies have elucidated a key role for the orexin system in the promotion and maintenance of wakefulness (Sakurai et al., 1998). It is

31 16 important to note however that the orexin system has been implicated in a number of other behaviors including energy homeostasis (as orexin neurons receive inputs from local NPY and POM C neurons, and are responsive to leptin and ghrelin peptides) and addiction/reward (through interaction with the VTA) (Cason et al., 2010, Moorman and Aston-Jones, 2010, Ponz et al., 2010a, Quarta et al., 2010, Sakurai and Mieda, 2011, Nixon et al., 2012). Research into the natural role for the orexin system has implicated these peptides in a number of behaviors. The afferent and efferent connections formed by the orexin system position it to integrate disparate systems and behaviors, including metabolism, stress, emotion, pain, motion, and arousal (Berridge et al., 2010, Chiou et al., 2010, Kuwaki and Zhang, 2010, Teske et al., 2010, van Dijk et al., 2011). The loss of orexin neurons results in the symptoms of narcolepsy (Peyron et al., 2000, Thannickal et al., 2000). This assertion is supported by the finding that hypothalamic damage, affecting the region of orexin neurons, can result in narcolepsy symptoms (Nishino and Kanbayashi, 2005). The mechanisms that trigger the loss of muscle tone in response to positive emotions (i.e. cataplexy) are present in healthy individuals and may lead to momentary muscle weakness, the feeling that one is weak with laughter (Overeem et al., 1999, Overeem et al., 2004). In healthy individuals it is hypothesized that the orexin system counteracts this momentary weakness and prevents cataplexy, whereas in the absence of the orexin system (i.e. narcolepsy) there is no counter to the muscle weakness and cataplexy occurs. The observation that orexin application to narcoleptic mice (ICV injection) and dogs (ICV/IV application) can reduce severity of cataplexy supports this hypothesis (Fujiki et al., 2003, Mieda et al., 2004). However, why the symptoms of narcolepsy occur as a result of the loss of orexin is unknown. Orexin projections and receptor patterns in the brain indicate some potentially important downstream targets of orexin that could play a role in cataplexy. Willie et al. (2003) investigated behavioral differences in OX 2 R KO mice and orexin KO mice, showing that OX 2 R KO had similar excessive sleepiness to the ligand KO model but significantly less cataplexy (Willie et al., 2003). This suggests that the loss of orexin signaling may trigger sleepiness through an OX 2 R- mediated mechanism while cataplexy is triggered through both OX 2 R and OX 1 R mechanisms. Histamine neurons of the ventral hypothalamus express OX 2 Rs and have a well established role in the promotion of wake and arousal, indicating that orexin loss may promote sleepiness through histaminergic projections (Huang et al., 2001, Marcus et al., 2001, Shigemoto et al.,

32 ). Meanwhile, OX 1 Rs are strongly expressed in the noradrenergic neurons and the amygdala, perhaps implicating these areas more in the expression of cataplexy (Marcus et al., 2001). This is however speculative, as the OX 2 R KO mouse still had some cataplexy and the OX 2 R-mutated canine model reliably shows cataplexy despite intact OX 1 Rs (Reilly, 1999). 1.2 Neurobiology of cataplexy This thesis will focus primarily on the roles of the dopaminergic system, noradrenergic system, and the amygdala in the regulation of cataplexy. Therefore a review of these three systems will be provided, with emphasis on the organization and general function of each system as it relates to cataplexy. While the noradrenergic, dopaminergic and amygdaloid systems all provide a strong rationale for investigation with respect to cataplexy, they are not the only transmitter systems involved; therefore, a brief review of other brain areas and transmitter systems implicated in the regulation of cataplexy is also included The dopaminergic system Dopamine is a catecholamine neurotransmitter synthesized in numerous nuclei throughout the brain. Dopamine is synthesized in neurons from the amino acid L-tyrosine. It is then packaged into vesicles by the vesicular monoamine transporter (Molinoff and Axelrod, 1971, Erickson et al., 1992). After release into the synaptic cleft, uptake of dopamine is mediated by the dopamine transporter, DAT, which allows dopamine to be taken up either by glial or pre-synaptic cells (Geffen et al., 1976, Gainetdinov and Caron, 2003). Pharmacological manipulation of DAT is implicated in the arousing effects of CNS stimulants (Fumagalli et al., 1998, Nishino et al., 1998b, Wisor et al., 2001). Nine dopaminergic cell fields have been demonstrated in the mammalian brain. Designated A8-A16, these cells are located in the mesencephalon, diencephalon, and olfactory bulb (Figure 1.5). Dopamine neurons have widespread projections throughout the CNS (Bjorklund and Dunnett, 2007a). The densest projections are from midbrain dopamine neurons in the A8 (reticular formation), A9 (substantia nigra (SN)) and A10 (ventral tegmental area (VTA)), to the striatum, limbic system and cortex (Eaton et al., 1994, Bjorklund and Dunnett, 2007a). Diencephalic cell fields, including the A11 (dorsal hypothalamus), A12 (arcuate nucleus), A13 (zona incerta), A14 (periventricular nucleus) and A15 (ventral hypothalamus),

33 18 have projections to the pituitary, hypothalamus, and spinal cord (Skagerberg et al., 1988, Bjorklund and Dunnett, 2007b, a). The A16 neurons are located in the olfactory bulb. The projections from diencephalic dopaminergic cell fields are not as well established as those from midbrain cell fields, though it is known that they provide the dopaminergic projections to the spinal cord (Skagerberg et al., 1982). Dopamine acts directly on two different receptor types. D1-like (including receptors D1 and D5) are G-protein coupled receptors (G as ) that are generally expressed on post-synaptic neurons (Missale et al., 1998). Dopamine binding stimulates adenylate cyclase activity which initiates intracellular signaling cascades. For the D1 receptor these cascades include production of camp, followed by activation of protein kinase A (PKA), and then phosphoralation of various proteins and ion channels. The general effect of D1 receptor activation is depolarization of the affected cell (Missale et al., 1998). D2-like receptors (including receptors D2, D3, and D4) are G-protein coupled receptors (G ai ) that are generally inhibitory. D2-like receptor activation inhibits camp production through inhibition of adenylate cyclase (Missale et al., 1998). D2-like receptors can act as auto-receptors on dopamine neurons to decrease neuron excitability (Westerink et al., 1990). Dopamine receptors are expressed in many structures throughout the CNS. D1 and D2 receptors are the most abundant, while D4 and D5 receptors are not expressed in great numbers in the rodent CNS (Meador-Woodruff et al., 1989, Mansour et al., 1990). Dopamine receptors are most abundant in the striatum and limbic system, including the amygdala. They are also expressed in the cortex, forebrain, hypothalamus and brainstem. Both D1-like and D2-like receptors can be expressed either pre or post-synaptically and are rarely co-expressed on the same neurons (Missale et al., 1998). The widespread projections of dopamine neurons and locations of dopamine receptors allow the dopamine system to contribute to a wide range of functions that relate to cataplexy, including emotion, sleep-wake state and motor control.

34 19 Figure 1.5: Dopamine projections The nine distinct dopaminergic cell fields (A8-16; blue triangles) in the mammalian brain project throughout the CNS. Dopamine neurons project to the cortex, limbic system, striatum, and both cranial & spinal motor neurons (designated by blue arrows). These projections position the dopaminergic system to modulate many behaviors including motivation, emotion, and movement. Modified from (Kvetnansky et al., 2009)

35 Functions The dopaminergic system has many functions. Dopamine neurons in the VTA project, through the mesolimbic pathway, to the limbic system. The mesocortical pathway regulates emotion and motivation through projections to the PFC. Another major dopaminergic pathway, the nigrostriatal pathway, controls movement through dopamine release into the basal ganglia (Bjorklund and Dunnett, 2007a). These three pathways from midbrain dopamine nuclei are functionally important connections in the brain and their loss can have devastating effects, for example the loss of substantia nigra (SN) dopamine in Parkinson s disease (Clarenbach, 2000). Dopamine has an established role in reward, motivation and processing of emotions. Areas in the limbic system and striatum, including the amygdala, PFC, and nucleus accumbens, express high concentrations of dopamine receptors and receive dense innervations from midbrain dopamine neurons (Loughlin and Fallon, 1983, Meador-Woodruff et al., 1989, Meador- Woodruff et al., 1991). These midbrain dopamine neurons increase activity in response to certain types of both positive and aversive stimuli (Steinfels et al., 1983b, Strecker et al., 1983). Current evidence suggests that dopaminergic transmission in the limbic system modulates processing of emotionally salient stimuli. The functional connections between the VTA, nucleus accumbens, amygdala and PFC have not been entirely elucidated, but changes in dopaminergic transmission within this circuit could be important in psychological disorders, such as schizophrenia (Laviolette, 2007). As previously mentioned, the orexin system shares reciprocal projections with the VTA that may regulate dopaminergic transmission; changes in the functional VTA-limbic circuit in response to the lack of orexin could be important in emotionally-induced cataplexy (Aston-Jones et al., 2010, Moorman and Aston-Jones, 2010). Early unit recording studies of midbrain dopamine neurons did not show significant differences in firing across the sleep-wake cycle (Miller et al., 1983, Steinfels et al., 1983a). This finding led to the hypothesis that dopamine did not play a role in determining behavioral state. However, this now does not appear to be the case. Most stimulants, which increase behavioral arousal, act via a dopaminergic mechanism (Nishino et al., 1998b, Scammell and Matheson, 1998, Kanbayashi et al., 2000, Wisor et al., 2001). Furthermore, selective pharmacological activation or antagonism of dopamine receptors can modulate sleep-wake behavior in rodents (Monti et al., 1988, Monti et al., 1989, Monti et al., 1990, Monti and Monti,

36 ). In addition, identified dopamine neurons increase Fos expression during wakefulness (A11 and ventral periaquaductal gray (vpag)) and REM sleep recovery (A13) (Leger et al., 2010). Lu and colleagues demonstrated that 50% of vpag dopamine neurons express Fos during waking; these neurons have a functional role in waking as selective lesions of dopamine neurons in the vpag suppressed wakefulness (Lu et al., 2006a). These findings suggest that the dopamine system does have a functional role in determining behavioral state, specifically in promoting waking behavior. The majority of dopaminergic nuclei do not project directly to motor neurons (Bjorklund and Dunnett, 2007b, a). The exception to this is the dopaminergic A11 neurons. It was previously reported that these neurons project to the ventral horn of the spinal cord where they synapse onto motor neurons (Bjorklund and Skagerberg, 1979, Skagerberg et al., 1982, Lindvall et al., 1983, Skagerberg and Lindvall, 1985). Our lab recently showed that A11 neurons innervate the trigeminal motor nucleus as well. These projections have a functional role, as activation of the A11 region leads to D1 receptor-mediated excitation of masseter muscle activity (JJ Fraigne and JH Peever, unpublished data). Previous studies in our lab have implicated the dopamine system in regulating muscle tone during waking. Dopamine provides an endogenous excitatory drive to motor neurons during waking that acts via a D1 receptor mechanism, while a D2 receptor mediated inhibitory drive is present during sleep. Of interest, activation of D1 receptors on trigeminal motor neurons during REM sleep partially reversed REM sleep muscle atonia, suggesting that withdrawal of a dopaminergic drive could contribute to muscle atonia during cataplexy (NA Yee and JH Peever, unpublished data). The dopamine system contributes to the regulation of emotion, arousal and muscle tone, suggesting that it could have a role in triggering cataplexy Established role in cataplexy Early studies in canine narcolepsy demonstrated a clear role for the dopamine system in regulating cataplexy. Systemic application of selective D2-like receptor agonists exacerbated cataplexy while antagonists reduced severity of attacks (Nishino et al., 1991). These results have been shown to be consistent with different compounds and methods of application (Mignot et al., 1993, Okura et al., 2000). As mentioned, D2-like receptors can act as auto-receptors on dopamine neurons to affect dopamine release (Svensson et al., 1987, Westerink et al., 1990),

37 22 therefore to elucidate the mechanism through which systemic drug applications were acting, focal perfusion of D2 receptor drugs was performed into different dopaminergic brain areas (Reid et al., 1996, Honda et al., 1999b, Okura et al., 2004). Perfusion of D2 agonists into the VTA, SN and diencephalic dopamine cell groups all exacerbated cataplexy, presumably by decreasing dopamine release at targets of these neurons (Reid et al., 1996, Honda et al., 1999b, Okura et al., 2004). Dopamine groups have projections directly to motor neurons, downstream to brainstem areas that regulate arousal and muscle tone, and/or upstream to the limbic system and cortex, which provides a number of possible mechanisms through which the dopaminergic system could modulate cataplexy. Studies in human narcoleptics have demonstrated an altered striatal dopaminergic system. Neuroimaging showed that narcoleptic patients have increased D2-like receptor binding in the striatum that correlates with cataplexy (Eisensehr et al., 2003). Similarly, dopamine receptor expression is altered in narcoleptic dogs, with higher receptor density in the ventral striatum and amygdala (Bowersox et al., 1987) In addition, many of the effective treatments for narcolepsy act via a dopaminergic mechanism, including amphetamines, modafinil, and possibly sodium oxybate (Nishino et al., 1998b, Wisor et al., 2001, Wisor and Eriksson, 2005). Therefore, dopamine appears to have a central role in both human and canine narcolepsy, and one that is strongly linked with cataplexy. The canine model of narcolepsy results from a mutation in OX 2 R rather than a more general loss of orexin signaling, but these studies have not been replicated in a murine model of narcolepsy that more closely mimics human narcolepsy. Whether endogenous and/or exogenous dopamine activity mediates these effects on cataplexy at the limbic system, REM sleep circuitry, or directly on motor neurons is unknown The noradrenergic system Noradrenaline is a catecholamine neurotransmitter. It is synthesized from dopamine by the catalyst dopamine beta-hydroxylase (Molinoff and Axelrod, 1971). Noradrenaline is transported into vesicles by the vesicular monoamine transporter (Erickson et al., 1992). After noradrenaline is released into the synaptic cleft, it is taken up by glia or presynaptic neurons by the noradrenergic transporter, NET, where it is repackaged into vesicles or broken down by monoamine oxidase (Gainetdinov and Caron, 2003). As with DAT, NET is a target for CNS stimulants (Xu et al., 2000).

38 23 There are seven noradrenergic nuclei in the mammalian brain. Designated A1-A7, these nuclei are located in the pons and medulla (Dahlstroem and Fuxe, 1964, Dahlstrom and Fuxe, 1964, Swanson and Hartman, 1975) (Figure 1.6). The A6, generally referred to as the locus coeruleus, contains ~45% of the noradrenergic neurons in the rodent brain and has projections throughout the brain, including to the cortex, basal forebrain, hypothalamus, midbrain and brainstem (Dahlstroem et al., 1964, Swanson and Hartman, 1975). The other nuclei provide the major noradrenergic projections to the brainstem and spinal cord (Grzanna and Fritschy, 1991). These projections allow the noradrenergic system to regulate levels of arousal, attention, and stress. Projections to both cranial and spinal motor neurons allow noradrenaline to modulate motor neuron excitability directly. The noradrenergic system receives some of the densest efferents from hypothalamic orexin neurons, with the A4, A5, A6 (LC) and A7 also expressing high levels of OX 1 R (Peyron et al., 1998, Marcus et al., 2001), suggesting that this system could contribute to the symptoms of narcolepsy. Noradrenaline acts at three main families of receptors: α 1 receptors are expressed postsynaptically and act to increase neuron excitability through a G q protein coupled receptor that acts via a phospholipase C-IP 3 mechanism to ultimately increase intracellular Ca 2+. The α 1 receptors are the most abundant adrenergic receptors in the mammalian brain. These receptors are present in the cortex, amygdala, and motor nuclei, among other areas. α 1 receptors are not present on noradrenergic neurons themselves (McCune et al., 1993, Garcia-Sainz et al., 1999). Due to a previously established role in both cataplexy and motor control, this thesis will focus on the α 1 receptor mediated role for noradrenaline (Mignot et al., 1988b, a, Chan et al., 2006). α 2 receptors are expressed both pre- and post-synaptically and activation of these receptors leads to inhibition through a G i protein coupled receptor; activation of α 2 receptors inhibits both intracellular Ca 2+ and camp production. β receptors are generally post-synaptic and increase neuron excitability through a G s protein coupled receptor; activation of β receptors results in increased adenylate cyclase, which increases production of camp, which ultimately causes phosphorylation of calcium channels (Piascik and Perez, 2001). α 2 and β receptors are present on noradrenergic cells where they act to modulate NA release (Buscher et al., 1999).

39 24 Figure 1.6: Noradrenaline projections The seven distinct noradrenergic nuclei (A1-7; black triangles with red projections) in the mammalian brain project throughout the CNS. The A6 (locus coeruleus (LC)) is the largest population of noradrenergic neurons in the rodent brain. Noradrenaline neurons project to the cortex, hypothalamus, and brainstem, as well as to both cranial & spinal motor neurons. These projections position the noradrenergic system to affect many behaviors including cognition, sleep and motor control. Modified from (Kvetnansky et al., 2009)

40 Functions One function of the noradrenergic system is to regulate behavioral arousal (Berridge and Waterhouse, 2003). Neurons in the LC fire in a state-dependent manner in dogs and mice; neuron activity is highest during periods of active waking, decreased in quiet waking, further decreased in NREM sleep and almost silent during REM sleep (Wu et al., 1999, Takahashi et al., 2010). This suggests that parts of the noradrenergic system may be important in regulating sleep-wake states. In support of this, many drugs that act via a noradrenergic mechanism promote wake and suppress sleep (Kuczenski and Segal, 1997, Kanbayashi et al., 2000, Rothman et al., 2001, Willie et al., 2003). However, studies disagree on whether lesions of noradrenergic neurons or complete noradrenergic depletion affect sleep-wake architecture (Hunsley and Palmiter, 2003, Hunsley et al., 2006, Li and Nattie, 2006, Blanco-Centurion et al., 2007). Recently it has been suggested that the noradrenergic system and the LC specifically may play a role in maintenance of sustained attention rather than waking per se (Gompf et al., 2010). Other noradrenergic nuclei express Fos activity after spontaneous waking and REM sleep deprivation, while the LC expresses Fos in response to novel environments but not spontaneous waking (Leger et al., 2009). These other noradrenergic regions are perhaps more likely to have a role in the regulation of sleep-wake state and/or cataplexy. Another way the noradrenergic system contributes to general arousal is through the stress response. Stressful stimuli induce noradrenergic release from the LC which acts at several upstream brain areas, including the amygdala (Tanaka et al., 1991). Noradrenaline can act at several receptor types in multiple sub-nuclei within the amygdala, modulating activity and responses to incoming sensory information (Ferry et al., 1999). The central nucleus of the amygdala in turn activates the LC through projections containing corticotrophin-releasing factor (Wallace et al., 1989, Van Bockstaele et al., 1998). This circuit could provide the necessary arousal and other behavioral and autonomic changes to respond to stressful conditions. As both the amygdala and the LC have connections with the orexin system, and cataplexy can be triggered by emotional stimuli, this circuit could contribute to cataplexy. Some noradrenergic nuclei project to both cranial and spinal motor neurons, which express noradrenergic receptors (Shao and Sutin, 1991). The A5 and A7 nuclei most densely innervate cranial motor pools and the spinal cord ventral horn (Commissiong et al., 1978,

41 26 Guyenet, 1980, Westlund et al., 1983, Card et al., 1986, Grzanna et al., 1987, Bruinstroop et al., 2011). Numerous studies have demonstrated that noradrenaline has an α 1 receptor-mediated excitatory affect on facial, hypoglossal, trigeminal and lumbar motor neurons (Fung and Barnes, 1981, Larkman and Kelly, 1992, Chan et al., 2006, Lu et al., 2007). Our lab has previously demonstrated the importance of the noradrenergic system in regulating masseter muscle tone across the sleep-wake cycle in rats. Perfusion of noradrenergic receptor agonists and antagonists by reverse-microdialysis onto trigeminal motor neurons elucidated an α 1 receptor mediated excitatory waking drive that is withdrawn during sleep (S Mir and JH Peever, unpublished data). As the noradrenergic system has a role in both motor control and arousal it is an interesting target for investigations into the underlying circuits regulating cataplexy Established role in cataplexy As with the dopaminergic system, pharmacological evidence strongly suggests a role for the noradrenergic system in mediating cataplexy. Tricyclic antidepressants effectively treat cataplexy in humans, dogs and mice, with those compounds that are more selective for noradrenaline being more effective, at least in canine narcolepsy (Babcock et al., 1976, Schachter and Parkes, 1980, Mignot et al., 1993, Willie et al., 2003). Experiments with more selective noradrenergic receptor agonists and antagonists have demonstrated that α 1 receptor activation reduces cataplexy in dogs, while antagonism exacerbates it (Mignot et al., 1988a, Nishino et al., 1990, Renaud et al., 1991, Mignot et al., 1993). In addition, altered expression of adrenergic receptors has been identified in canine narcolepsy, with elevated α 1 receptor binding in the amygdala and α 2 receptor binding in the LC (Mignot et al., 1988b, Mignot et al., 1989, Nishino et al., 1990). More evidence for the involvement of the noradrenergic system in cataplexy comes from studies utilizing single unit recordings in freely behaving narcoleptic dogs: presumptive noradrenergic neurons of the LC ceased firing during cataplexy (Wu et al., 1999). More recently, it was shown that high-frequency optogenetic stimulation of the LC in wildtype mice caused behavioral arrests similar to cataplexy (Carter et al., 2010). While these two results initially seem paradoxical, it has been proposed that this high frequency stimulation of the LC would lead to a depletion of noradrenaline from synapses, and thus a situation similar to the cessation of cell firing (McGregor and Siegel, 2010). This has led to the hypothesis that

42 27 withdrawal of excitatory noradrenergic input to motor neurons is responsible for the loss of muscle tone during cataplexy, although this has never been directly tested The amygdala The amygdala, or amygdaloid complex, is a collection of nuclei in the medial temporal lobe. It is comprised of a number of different nuclei with extensive intra and internuclear connections (Sah et al., 2003). Different classification systems have been used and the amygdala can be subdivided into between three and thirteen different nuclei. Five commonly separated subnuclei include the lateral nucleus, the basolateral nucleus (BLA), the central nucleus (CeA), the medial nucleus and the cortical nucleus. There are also a group of GABAergic interneurons within the region of the amygdala, termed the intercalated cells that contribute to intraamygdaloid signaling (Sah et al., 2003). The nuclei in the amygdaloid complex have many connections between them (Pitkanen et al., 1997, Pitkanen et al., 2003). These connections are important for the processing of incoming sensory information. In general terms sensory information enters the amygdala through the lateral and BLA, then follows a lateral to medial progression, to the CeA, which is the primary output nucleus of the amygdala (Sah et al., 2003) (Figure 1.7). Afferents to the amygdala can be divided into two different groups based on the information encoded by the inputs. Cortical and thalamic inputs carry sensory information to most nuclei within the amygdala. Hypothalamic and brainstem inputs carry behavioral and autonomic information to the amygdala, though predominantly only to the CeA. The amygdala processes these primary sensory inputs, as well as polymodal inputs from the PFC, and behavioral inputs, and then communicates with the brainstem, hypothalamus and ventral striatum to induce appropriate responses to the incoming stimuli (Romanski and LeDoux, 1993, Updyke, 1993, Brinley-Reed et al., 1995, Sah et al., 2003). The remainder of this overview will focus on the morphology and physiology of CeA, as it is the only nucleus to receive projections from and innervate the brainstem, where muscle atonia is generated (Hopkins and Holstege, 1978, Wallace et al., 1989, 1992, Fung et al., 2011, Xi et al., 2011). In addition to being the primary output nucleus of the amygdala, the CeA has also been proposed as an integrator of amygdala activity. The CeA both sends inputs to and receives innervations from the other major amygdaloid nuclei (Pitkanen et al., 1997, Sah et al.,

43 ). Organization even within just the CeA is complex, as it can be subdivided into at least three sub-nuclei and contains numerous cell types and transmitters. Projections from the CeA to the midbrain and brainstem include neurons that contain enkephalin, dynorphin, neurotensin, substance P, somatostatin, corticotrophin releasing factor, glutamate and GABA (Sah et al., 2003). The organization of the CeA and heterogeneity of the neuronal population within it suggest the ability to process many different types of stimuli and affect a wide range of behavioral and autonomic responses.

44 29 Figure 1.7: A simplified schematic of major inputs and outputs of the amygdala The amygdala receives many inputs including from all sensory systems. These inputs are processed through projections within and between different amygdala nuclei. The major outputs of the amygdala establish appropriate reactions to the incoming stimuli. (LA, lateral nucleus; BLA, basolateral nucleus; ITC, intercalated cells, CeA, central nucleus; DA, dopamine; NA, noradrenaline; 5-HT, serotonin; Ach, acetylcholine)

45 Functions The amygdala is part of a larger collection of brain areas that process emotion: the limbic system. The limbic system contains the amygdala, prefrontal cortex, hypothalamus, hippocampus, and areas of the thalamus. These areas share connections and are important for processing sensory information, assigning valence, motivation, emotional memory, and generally ensure appropriate arousal, endocrine and autonomic responses to emotional stimuli (LeDoux, 2000, Siegel and Boehmer, 2006, Boissy et al., 2007, Murray, 2007). Lesions in the region of the amygdala, as in Kluver-Bucy syndrome, result in flat affect or placidity (Horel et al., 1975, Baron-Cohen et al., 2000, Murray, 2007). The majority of amygdala research has focused on its role in fear learning and response. For example, the amygdala is critical for fear conditioning, with lesions of the BLA resulting in an inability to learn conditioned fear and lesions in the CeA resulting in a lack of fear response (Prather et al., 2001, Sah et al., 2003, Phelps and LeDoux, 2005). However, the amygdala has also been linked with positive emotion. Neuroimaging shows increased amygdala activity in response to positive emotional stimuli in humans, including photographs, music, and smiling faces (Garavan et al., 2001, Boissy et al., 2007). The amygdala also encodes positive affective associations in non-human primates and rodents; amygdala firing is correlated with anticipation of rewards and contributes to learning the valence of specific positive stimuli (Nishijo et al., 1988, Schoenbaum et al., 1998, Baxter et al., 2000, Baxter and Murray, 2002, Paton et al., 2006). Many studies have established the amygdala as a key part of the circuit that processes both positive and aversive emotionally salient stimuli, which have both been demonstrated to trigger cataplexy in some narcolepsy patients (Overeem et al., 2011). The amygdala has also been implicated in the regulation of REM sleep, particularly changes in REM sleep in response to emotional stimuli. Pharmacological inhibition of the amygdala after inescapable shock suppressed the normal increase in REM sleep in rats, indicating that at least after stressful stimuli the amygdala functions to promote REM sleep (Tang et al., 2005, Sanford et al., 2006, Liu et al., 2009). However, lesions of the amygdala in non-human primates resulted in no significant changes to REM sleep, bringing into question whether the amygdala has a role in normal sleep-wake behavior (Benca et al., 2000). The amygdala is anatomically positioned to regulate REM sleep, as the CeA of the amygdala sends

46 31 both excitatory and inhibitory projections to the brainstem. Of particular interest is the finding that there are glutamatergic projections from the CeA to the sublateraldorsal nucleus (SLD), a key regulator of REM sleep and muscle atonia (Boissard et al., 2003, Fung et al., 2011, Xi et al., 2011). Whether the amygdala has a functional role in the normal regulation of REM sleep is unclear; however, as the amygdala is implicated in emotion and has projections to REM sleepgenerating brain areas, it is an interesting target for cataplexy research Established role in cataplexy In the context of cataplexy, the limbic system is important as positive emotions, and particularly laughter, are the most reliable trigger of cataplexy in human narcolepsy (Siegel and Boehmer, 2006, Overeem et al., 2011). Even in animal models of narcolepsy, positive emotions may trigger cataplexy; the play or food-elicited cataplexy tests have been used extensively in narcoleptic dogs, where social play or palatable food are given to dogs in order to increase cataplexy (Siegel et al., 1986, Mignot et al., 1988a). In murine models of narcolepsy access to palatable food, social play, or a running wheel have been demonstrated to increase cataplexy (Espana et al., 2007, Clark et al., 2009, Scammell et al., 2009). These data suggest a role for the limbic system in mediating cataplexy but few studies have directly addressed this. Orexin strongly projects to the limbic system and recent evidence from clinical studies suggests that human narcoleptics may have abnormal limbic activity in response to certain stimuli. An fmri study demonstrated decreased hypothalamic activity and increased amygdala activity in response to humorous stimuli in narcoleptics when compared to controls (Schwartz et al., 2008). A separate experiment used a game to assess activity in reward related brain regions in narcolepsy patients; reduced activity in the VTA and PFC were seen in narcolepsy patients versus controls, perhaps due to the lack of orexinergic drive to these areas (Ponz et al., 2010a). Evidence also suggests morphological differences in the amygdala in narcolepsy, with a 17% decrease in amygdala volume compared to controls (Brabec et al., 2011). While these studies tell us little about a functional role for the limbic system in cataplexy, they do highlight abnormalities in areas that are important in emotion and reward processing, thus providing a rationale for the investigation of the amygdala and other limbic areas in regulating cataplexy. A more convincing link between cataplexy and limbic system activity was demonstrated in canine narcolepsy: unit recording in the amygdala showed a subpopulation of neurons that

47 32 increased their firing at cataplexy onset and decreased their firing when normal activity was resumed (Gulyani et al., 2002). Unit recording studies in narcoleptic dogs have provided some of the clearest data to date regarding which brain areas are involved in generating cataplexy; however, these studies are correlative and do not infer a functional role for the neurons recorded. Each of the studies outlined here has correlated amygdala activity with aspects of narcolepsy and cataplexy, but no studies have investigated a functional role for the limbic system in the triggering of cataplexy. A possible, though highly speculative, link between the amygdala and cataplexy is the aforementioned hypothesis that cataplexy is an atavistic expression of tonic immobility (Overeem et al., 2002). Tonic immobility is a defense mechanism, like freezing in rodents, and is present in some animals, most notably sharks, rabbits, chickens and guinea pigs. The amygdala mediates many fear responses including freezing responses in rodents. Freezing is mediated by projections from the CeA to the ventral periaquaductal gray (vpag) and is very different from cataplexy in that muscle tone generally remains high (Reese et al., 1984, Oliveira et al., 2004). Tonic immobility is mediated by projections from the CeA to the ventral lateral periaquaductal gray (vlpag) and can be similar to cataplexy (Leite-Panissi et al., 2003). The vlpag has also recently been implicated in the control of REM sleep-related muscle atonia in rodents (Boissard et al., 2002). While humans, mice and most dogs do not naturally express tonic immobility, it is possible that the loss of the orexin system uncovers this dormant behavior. It is unclear whether cataplexy is related to tonic immobility but if so the amygdala would play a central role in the initiation of cataplexy. In general, this also functionally links the amygdala with circuits that are able to generate the loss of muscle tone normally seen only during REM sleep, at least in a small group of animals, though it is unknown if this plays a functional role in cataplexy Other transmitter systems While this thesis focuses predominantly on the noradrenergic, dopaminergic and amygdaloid regulation of cataplexy a number of other transmitter systems and brain regions have been implicated in this behavior. In this section I will provide a brief overview of some of the other systems involved in cataplexy, including the serotonergic, cholinergic, and histaminergic systems.

48 33 Unlike the dopaminergic and noradrenergic systems, serotonin does not appear to play an important role in cataplexy. While serotonin receptor agonists suppress canine cataplexy, antagonism did not exacerbate cataplexy (Nishino et al., 1995a). It was originally hypothesized that because firing of serotonin neurons in the dorsal raphe nucleus is suppressed during REM sleep, just as LC neurons are, that serotonin would have a role in cataplexy (Trulson and Jacobs, 1979, Jacobs et al., 1981, Trulson and Trulson, 1982). However, subsequent studies have shown that firing of dorsal raphe neurons is not suppressed to REM sleep levels during cataplexy (Wu et al., 2004). These data do not support a strong role for the serotonergic system in cataplexy; however, due to its established role in mood/emotion and motor control a role in cataplexy cannot be ruled out without further investigation. The role of the serotonergic system has not been investigated in murine narcolepsy. Histamine neurons are important in the maintenance of arousal and receive strong innervations from orexin neurons (Monnier et al., 1967, Lin et al., 1988, Monti et al., 1991, Monti, 1993, Peyron et al., 1998). It has been hypothesized that the loss of orexinergic excitation of histamine neurons may underlie the excessive sleepiness in narcolepsy; studies demonstrating decreased CSF histamine in human narcolepsy support this claim (Kanbayashi et al., 2009, Nishino et al., 2009, Scammell and Mochizuki, 2009). More recently, selective genetic tools have been used to address the role for histamine in sleepiness; mice with a transcriptionally disrupted OX 2 R were created, which lack a functional OX 2 R, but in the presence of crerecombinase, transcription and receptor expression is rescued (Mochizuki et al., 2011). When orexin receptors were selectively expressed in the posterior hypothalamus (where histamine neurons are located), the excessive sleepiness normally seen in these mice was rescued (Mochizuki et al., 2011). However, unit recording of presumptive histamine neurons in narcoleptic dogs has shown that histamine neurons do not decrease firing during cataplexy (when compared to waking), suggesting that histamine is not involved in cataplexy (John et al., 2004). Thus it is evident that histamine has a key role in the maintenance of arousal and perhaps consciousness but the histaminergic system likely does not contribute to cataplexy. Cholinergic neurons in both the pons (PPT/LDT) and basal forebrain are involved in the regulation of sleep and wakefulness, particularly REM sleep (McCarley, 2004, Lu et al., 2006b). Both canine and murine models of narcolepsy have altered cholinergic systems, with increased PPT/LDT neurons in canine narcolepsy and increased ChAT staining intensity in LDT neurons

49 34 in murine narcolepsy (Nitz et al., 1995, Kalogiannis et al., 2010). Systemic injection of drugs that increase cholinergic drive increased cataplexy in narcoleptic dogs; this effect may be mediated by pontine regions that generate REM sleep, as activation of muscarinic receptors specifically in the pontine reticular formation increased cataplexy in dogs (Reid et al., 1994a, Reid et al., 1994b, Reid et al., 1994c). Importantly, these findings have recently been replicated in a murine model of narcolepsy (Kalogiannis et al., 2011). Cataplexy was also affected by manipulation of basal forebrain activity although few cataplexy active neurons were detected there (Nishino et al., 1995b, Nishino et al., 1998a, Reid et al., 1998) Cataplexy as a manifestation of REM sleep Similarities between cataplexy and REM sleep have led researchers to hypothesize that cataplexy is the intrusion of REM sleep into wakefulness. Loss of muscle tone, theta rich EEG activity, and reduced H-reflex are all common features of cataplexy and REM sleep. Recently, a number of studies have shown key differences between cataplexy and REM sleep suggesting that they do not share the same executive mechanism. Nishino et al. (2000) demonstrated that REM sleep has an ultradian rhythm while cataplexy does not; REM sleep is homeostatically regulated while cataplexy can be triggered at any time with positive affective stimuli in narcoleptic dogs (Nishino et al., 2000). Lesions of pontine regions that suppress REM sleep lead to increased REM sleep but no change in the occurrence of cataplexy (Kaur et al., 2009). In addition, a number of pharmacological interventions can suppress cataplexy while having no affect on REM sleep (Okura et al., 2000). These studies demonstrate key differences between the regulation of REM sleep and cataplexy that suggest they have different executive mechanisms; however, these two states may share a common mechanism at or near the level of the motor neuron that ultimately triggers muscle atonia. As mentioned previously, much of the circuitry that regulates muscle tone during REM sleep in rodents has been elucidated. The monoaminergic-cholinergic hypothesis of REM sleep has been updated to include a glutamatergic-gabaergic switch that is modulated by these brainstem arousal systems. There is a REM sleep-off area in the pons (vlpag/lpt)) that actively inhibits a group of glutamatergic REM sleep-on neurons (SLD), this forms the basis for the flip-flop switch model of REM sleep control (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b). Neurons in the SLD likely project directly to both inhibitory interneurons in

50 35 the spinal cord as well as GABAergic/glycinergic neurons in the medulla; both the glycinergic interneurons and medullary neurons inhibit motor neurons to induce muscle atonia (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b). If the loss of muscle tone during cataplexy is generated by the same mechanism as REM sleep muscle atonia, one would expect neuron activity to be the same in these regions during both states. In narcoleptic mice, lesions in the region of the vlpag lead to an increase in REM sleep but no observed change in cataplexy, while unit recording in the pons revealed REM-active neurons but no cataplexy-active neurons (Kaur et al., 2009, Thankachan et al., 2009). These studies seem to indicate that REMgenerating circuits are not involved in triggering cataplexy. However, unit recording of medullary neurons (in the nucleus magnocellularis) in freely behaving narcoleptic dogs demonstrated a population of neurons that were highly active during both REM sleep and cataplexy, these neurons could be responsible for sending inhibitory projections that suppress muscle tone (Siegel et al., 1991). In addition, as mentioned previously, direct manipulation of pontine regions that regulate REM sleep atonia with cholinergic agonists can exacerbate cataplexy (Kalogiannis et al., 2011). Whether the circuits responsible for REM sleep atonia also trigger loss of muscle tone in cataplexy is unknown, and warrants further research. The similarities between REM sleep and cataplexy have led to the hypothesis that cataplexy is the intrusion of REM sleep phenomena into wakefulness (Rechtschaffen and Dement, 1967). While some recent data challenge this, it has not been investigated whether increasing physiological need or pressure for one of these states can promote the occurrence of the other in animal models of narcolepsy. Orexin KO mice represent an excellent model to investigate the relationship between REM sleep and cataplexy Current model of cataplexy The current understanding of the neural circuitry that triggers cataplexy has been summarized into a working model (Siegel and Boehmer, 2006). The authors propose that in response to positive emotions, the limbic system inhibits noradrenaline and serotonin systems that normally activate motor neurons. Brainstem inhibitory networks are simultaneously recruited that actively inhibit motor neurons leading to muscle atonia (Figure 1.8). Under normal conditions (i.e. orexin system intact), orexin counters this influence and prevents the loss of muscle tone. This model provides a good framework for understanding cataplexy; however, it is

51 36 based predominantly on pharmacological and unit recording data, and few of the elements of the model have been functionally demonstrated. This model treats the limbic system as a black box and marginalizes the role of the dopaminergic system. Whether the limbic system has a functional role in triggering cataplexy has not been investigated. Whether the monoaminergic-excitatory and brainstem-inhibitory systems have a functional role in mediating the loss of muscle tone during cataplexy is unknown. The goal of this thesis is to address some of these fundamental questions in narcolepsy research in a way that improves the current model of the mechanisms that trigger cataplexy.

52 37 Figure 1.8: Model of how the orexin (hypocretin) system could prevent the loss of muscle tone in response to positive emotions This model demonstrates how the loss of muscle tone (i.e. cataplexy) could be triggered by positive emotions when a functional orexin/hypocretin system is absent. In this model positive emotions trigger limbic system activity that inhibits monoaminergic brainstem nuclei. At the same time inhibitory circuits are recruited by acetylcholine that further suppress motor neuron excitability. In a healthy individual, the intact orexin/hypocretin system counteracts this loss of muscle tone, but one could imagine in the absence of orexinergic excitation motor neurons would be inhibited and disfacilitated leading to loss of muscle tone. From (Siegel and Boehmer, 2006)

53 Thesis overview As detailed above, the mechanisms that regulate cataplexy remain unclear. Cataplexy is thought to be the intrusion of REM sleep atonia into wakefulness and is triggered by strong positive emotions. What is responsible for the loss of muscle tone, how this relates to REM sleep atonia, and why positive emotions can act as a trigger for cataplexy are all questions that remain unanswered. These experiments will help elucidate the neural mechanisms responsible for cataplexy using freely behaving orexin KO mice. Behavioral studies, whole animal pharmacology, local perfusion of drugs via reverse-microdialysis, neuronal tracing and cell-specific lesions will be used to identify key components of the underlying neurobiology of cataplexy. Specific research objectives of this thesis are as follows: 1. Further characterize the narcoleptic phenotype in orexin KO mice and investigate the relationship between cataplexy and REM sleep (Chapter 3). 2. Characterize the role for the noradrenergic and dopaminergic systems in the regulation of murine cataplexy using systemic pharmacology (Chapters 4 and 5). 3. Determine whether a change in noradrenergic drive to motor neurons contributes to the loss of muscle tone during cataplexy (Chapter 5). 4. Characterize the role of the amygdala in triggering cataplexy, particularly cataplexy elicited by positive emotions (Chapter 6).

54 39 Chapter 2: Methods This thesis includes studies using multiple techniques and approaches, performed at multiple institutions (University of Toronto and Harvard University). This chapter will explain the general methods that were common to all experiments performed. Detailed accounts of the methods used in each experiment, including all equipment and reagents used are presented in the individual chapters.

55 40 Chapter 2: Methods 2.1 Mice All of the research presented in this thesis was performed in freely behaving wildtype or orexin KO mice. All procedures and experimental protocols were approved by the University of Toronto s Animal Care Committee or by the Institutional Animal Care and Use Committees of Beth Israel Deaconess Medical Center and Harvard Medical School, and were in accordance with the Canadian Council on Animal Care and the National Institutes of Health Guide for the Care and Use of Laboratory. Animals were cared for under the supervision of the support staff of the Bioscience Support Facility (University of Toronto, Toronto, Canada) or the CLS Animal Research Facility (Center for Life Sciences Boston, Boston, Massachusetts). All animals were maintained at a room temperature of C on a 12:12 light:dark cycle. Food and water were available ad libitum. Mice were group-housed with up to five same-sex siblings in plastic cages (dimensions 28cm x 16cm x 12cm, Nalgene Labware) on standard cob bedding. These studies utilized orexin KO mice and wildtype littermates. Orexin KO mice were obtained from Dr. Masashi Yanigasawa s lab at the University of Texas Southwestern Medical Center, where the mouse line was generated (Chemelli et al., 1999). These founder mice were backcrossed several generations on a C57Black/6 background. The majority of both orexin KO and wildtype mice used in these studies were obtained from heterozygous-heterozygous mating pairs. Orexin KO mice were genotyped using PCR with genomic primers 5'- GACGACGGCCTCAGACTT CTTGGG, 3'-TCACCCCCTTGGGATAG CCCTTCC, and 5 - CCGCTATCAGGACATAGCG TTGGC (with forward primers being specific for either wildtype or KO mice and the reverse primer being common to both). 2.2 Surgical procedures To record brain and muscle activity in mice, we implanted electroencephalogram (EEG) and electromyogram (EMG) electrodes. Sterile surgery was performed under isoflurane (1-2%; University of Toronto) or ketamine/xylazine (100 and 10 mg/kg i.p.; Harvard University) anesthesia. Effective depth of anesthesia was determined by the abolishment of the pedal withdrawal reflex.

56 41 Mice were placed in a stereotaxic apparatus and the skin was retracted to expose the skull surface. The skull was leveled, so that bregma and lamba were in the same horizontal plane. EEG recordings were obtained using two stainless steel micro-screws (1mm anterior and 1.5mm lateral to bregma; 3mm posterior and 1.5mm lateral to bregma). EMG electrodes consisted of multistranded stainless steel wires that were sutured onto the neck and/or masseter muscles. All electrodes were attached to a micro-strip connector, which was affixed onto the mouse s head with dental cement. After surgery, mice were given 0.5 ml of 0.9% saline, ketoprofen (3mg/kg; s.c.; University of Toronto) or meloxicam (5mg/kg; s.c.; Harvard University). 2.3 Sleep recording In order to record sleep-wake state and behavior, mice were housed in a specifically designed containment system. This caging system was housed inside a sound-attenuated, ventilated, and illuminated chamber that obviated visual, olfactory and auditory disturbances. Sleep-wake state and muscle activity were recorded by attaching a lightweight cable to a plug on the mouse s head. This was then connected to a data amplifier system. The EEG signal was amplified 1000 times and band-pass filtered between 1 and Hz. EMG signals were amplified 1000 times and band-pass filtered between 30 Hz and Hz. All electrophysiological signals were digitized at Hz, monitored and stored on a computer. In experiments with orexin KO mice, infra-red video recordings were also captured and synced to the electrophysiological recordings. 2.4 Behavioral state analysis We used both EEG and EMG signals (neck and/or masseter muscles) as well as video to identify up to six distinct behavioral states: active wake, quiet wake, NREM sleep, REM sleep, sleep attacks, and cataplexy. Active wake was characterized by high frequency, low voltage EEG signals coupled with high levels of EMG activity. Quiet wake was characterized by high frequency, low voltage EEG signals but in the absence of overt motor activity. NREM sleep was characterized by high amplitude, low frequency EEG signals and minimal EMG activity. REM sleep was characterized by low amplitude, high frequency theta EEG activity and very low EMG levels (i.e., REM atonia) interspersed by periodic muscle twitches. Sleep attacks, sudden transitions into sleep that have been observed in these mice, were characterized by gradual loss

57 42 of muscle tone, NREM-like EEG characteristics and automatic behavior (i.e. chewing) in the masseter muscles. Cataplexy was classified as a sudden loss of EMG activity, in neck and/or masseter muscles, following at least 40 seconds of active waking and with a duration of at least 10 seconds. Cataplexy was scored using both electrophysiological and video recordings. This definition is consistent with the consensus definition of murine cataplexy (Scammell et al., 2009). We also scored the transitions between behavioral states; however we did not include these data in our analyses. Sleep states were visually identified, analyzed, and scored in Spike 2 or Sleepsign for Animals. 2.5 Histology In order to determine microdialysis probe location, brain lesion area, or neuronal projections, histology was performed. At the end of each experiment mice were anesthetized via isoflurane or ketamine-xylazine and sacrificed. The brain was removed and placed in paraformaldehyde or formalin for 24 hours and then cryoprotected in 20-30% sucrose (in 0.1M PBS) for 48 hours. The brain was then frozen in finely crushed dry-ice and transversely sectioned in 30-40µm slices using a microtome. Brain sections were stained and mounted on slides. Tissue sections were viewed using a microscope and photographed. The locations of probe tracts, lesions, or projections were then plotted on a standardized stereotaxic map of the mouse brainstem (Paxinos and Franklin, 2001).

58 43 Chapter 3: REM sleep and Cataplexy are generated by Independent Mechanisms Other researchers contributed to this work: Petri Takala, BSc: Assisted with USV experiments and counting USVs. LJ Agostinelli, BSc: Assisted with selective REM sleep deprivation/gentle handling. Takatoshi Mochizuki, PhD: Assisted with setting up counts for wheel running experiments. John Yeomans, PhD: Assisted with experimental design for the USV experiments.

59 44 Chapter 3: REM sleep and Cataplexy are generated by Independent Mechanisms 3.1 Abstract It is hypothesized that cataplexy is the manifestation of REM sleep muscle atonia during wakefulness. Some recent studies have challenged this hypothesis, suggesting instead that REM sleep and cataplexy may be generated by unique neural mechanisms. Here we used a series of experiments to investigate some aspects of REM sleep and cataplexy and determine whether the two states share an underlying neural mechanism. We observed several phenotypic similarities between the two states, including frequency of occurrence and suppression of muscle tone. If they share a common mechanism, stimuli that increase propensity for one state may increase the other as well. We therefore sought to promote either state and observe changes in the other. Increasing the propensity for cataplexy by using specific environmental stimuli caused a reduction in REM sleep. We then increased REM sleep propensity, by selectively depriving mice of REM sleep, and observed a short-term reduction in cataplexy. These results suggest that cataplexy and REM sleep are not generated by the same neural mechanism, although they may share a common muscle atonia-generating mechanism. 3.2 Introduction Narcolepsy is characterized by excessive daytime sleepiness and cataplexy, the sudden loss of postural muscle tone during waking (Gelineau, 1880). Narcolepsy with cataplexy results from the loss of orexin containing neurons in the lateral hypothalamus, but the underlying neural circuitry responsible for cataplexy is unclear (Peyron et al., 2000, Thannickal et al., 2000, Thannickal et al., 2003). Cataplexy shares some characteristic features with REM sleep, leading to the hypothesis that cataplexy is the intrusion of REM sleep atonia into wakefulness (Rechtschaffen and Dement, 1967, Siegel et al., 1991, Chemelli et al., 1999). Numerous lines of evidence suggest that cataplexy and REM sleep share a common mechanism. Cataplexy and REM sleep share a similar phenotype, including theta-rich EEG and loss of muscle tone (Chemelli et al., 1999). Some pharmacological evidence suggests that they are regulated by a similar mechanism, as tricyclic antidepressants that are used to treat cataplexy

60 45 also reduce REM sleep (Babcock et al., 1976, Gervasoni et al., 2002, Willie et al., 2003). In addition, in narcoleptic canines a population of neurons in the medial medulla thought to be important for generating muscle atonia are active only during REM sleep and cataplexy (Siegel et al., 1991). Recently, however, several studies have suggested that cataplexy and REM sleep are distinct behavioral states: neurons in the pons, an area recently shown to be important in the regulation of REM sleep, fire most during waking and REM sleep but not during cataplexy in orexin KO mice, REM sleep follows an ultradian pattern while cataplexy does not, and systemic pharmacological manipulation of the dopamine system can induce changes in cataplexy without affecting REM sleep (Nishino et al., 2000, Okura et al., 2000, Thankachan et al., 2009, Burgess et al., 2010). Using a series of experiments investigating motor aspects, regulation and triggering of REM sleep and cataplexy, we addressed whether these two states share a common mechanism or are two different, unique behavioral states. We observed that episodes of REM sleep and cataplexy have similar frequency of occurrence, duration and levels of muscle tone. We then demonstrated that potentially positive affective stimuli increased cataplexy in mice, while these same stimuli decreased REM sleep. Finally, because REM sleep is homeostatically regulated, we investigated whether increasing REM sleep pressure would promote cataplexy. We found REM sleep deprivation resulted in a short-term decrease in cataplexy. Our data suggest that cataplexy and REM sleep are unique states with different executive mechanisms, though they may share a similar atonia-generating mechanism. 3.3 Methods These studies were approved by the Institutional Animal Care and Use Committees of Beth Israel Deaconess Medical Center and Harvard Medical School and were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the University of Toronto s animal care committee and were in accordance with the Canadian Council on Animal Care Animals We used 43 orexin KO mice (29 male and 14 female), weeks of age and weighing 26-34g. Mice were genotyped using PCR with genomic primers 5 -

61 46 GACGACGGCCTCAGACTTCTTGGG, 3 TCACCCCCTTGGGATAGCCCTTCC, and 5 - CCGCTATCAGGACATAGCGTTGGC (with forward primers being specific for either wildtype or KO mice and the reverse primer being common to both) Surgery We anesthetized mice with isoflurane (1-2%; University of Toronto) or ketamine/xylazine (100 and 10 mg/kg i.p.; Harvard University) and placed them in a stereotaxic alignment system (Model 1900, Kopf). We then implanted mice with electrodes for recording the EEG and EMG. In brief, stainless steel screws were implanted for frontoparietal EEG recordings (1.5 mm lateral and 1 mm anterior to bregma; 1.5 mm lateral and 3 mm posterior to bregma). EMG electrodes were made from fine, multi-stranded stainless steel wire (AS131, Cooner Wire, Chatsworth, CA), which were sutured into the masseter or neck muscles. All electrodes were attached to a micro-strip connector affixed to the animal's head with dental cement. After surgery, mice were given 0.5 ml of 0.9% saline, ketoprofen (3mg/kg; s.c.; University of Toronto) or meloxicam (5mg/kg; s.c.; Harvard University) Electrophysiological recordings Behavioral state and muscle activity were recorded by attaching a lightweight cable to a plug on the mouse s head, which was connected to an amplifier system. The EEG was amplified 1000 times and band-pass filtered between 0.3 and 100 Hz. EMG signals were amplified 1000 times and band-pass filtered between 30 Hz and 100 Hz. All electrophysiological signals were digitized between Hz (Spike 2 Software, 1401 Interface, CED Inc.), monitored and stored on a computer. Infra-red video recordings were also captured and synced to the electrophysiological recordings. Raw EMG signals were full-wave rectified and quantified in arbitrary units (A.U.). Average EMG activity for masseter muscle activity was quantified in 5s epochs for each behavioral state REM sleep deprivation protocol In order to determine if REM sleep pressure affected the occurrence of cataplexy, we selectively deprived orexin KO mice of REM sleep. REM deprivation experiments were performed at Harvard University with the assistance of LJ Agostinelli and under the supervision

62 47 of Dr. Scammell (Department of Neurology, Beth Israel Deaconess Medical Center). Two weeks after surgery, we transferred mice to recording cages in a sound-attenuated chamber with a 12:12 light-dark cycle (30 lux daylight-type fluorescent tubes with lights on at 07:00), constant temperature (23 ±1 C), and with food and water available ad libitum. The recording cable was attached to a low torque electrical swivel, fixed above the cage that allowed free movement. Mice were habituated to the cables for 4 days before the experiments and remained connected throughout the study. We first recorded baseline sleep-wake behavior across 24 hours using EEG/EMG and infrared video recordings. During this control recording, all procedures were the same as during the deprivation day protocol. Following the control recording, we then performed selective REM sleep deprivation on half of the mice during the last 4 hours of the light period (15:00-19:00) by visually identifying REM sleep using EEG/EMG criteria and then waking mice from sleep using gentle handling. A paired mouse was also aroused regardless of what state it was currently in to serve as a sham deprivation (i.e. a similar number of arousals without the same loss of REM sleep). After 3 days of recovery, the mice reversed roles such that mice in the sham group were now deprived of REM sleep, while those previously deprived served as shams Wheel running and chocolate protocol In order to establish whether presumptive positive affective stimuli could increase cataplexy we provided running wheels and chocolate to orexin KO mice. Wheel running experiments were performed at Harvard University under the supervision of Dr. Scammell. These mice served as a control for experiments detailed in chapter 6, therefore they had received bilateral amygdala PBS injections during EEG/EMG implantation surgery. This intervention did not appear to affect behavior. We first examined baseline sleep-wake behavior across 24 hours using EEG, EMG and infrared video recordings. We then studied mice under two conditions that should increase cataplexy: access to a running wheel, and to a running wheel plus chocolate. We placed a low torque, polycarbonate running wheel (Fast-Trac, Bio-Serv, Frenchtown, NJ) in each cage and recorded wheel rotations using a photodetector beneath each wheel (designed and constructed by Dr. Mochizuki). Running wheels increase cataplexy in orexin KO mice (Espana et al., 2007), and this style of wheel was chosen because it does not interfere with the EEG recording cable. After 7 days of habituation to the wheel, we recorded sleep/wake behavior and wheel rotations. The next night, we gave mice 3g of milk chocolate (Hershey s) at dark onset and recorded sleep-wake behavior and wheel running activity over the next 12 hours (19:00-

63 48 7:00). We chose to use chocolate because chocolate has been used as a reward in rodent operant studies (Holahan et al., 2011, King et al., 2011), and cataplexy in mice and dogs is increased by palatable food (Siegel et al., 1986, Clark et al., 2009) Ultrasonic vocalizations and social reunion paradigm Ultrasonic vocalization (USV) experiments were performed at the University of Toronto in collaboration with Dr. Yeomans (Department of Psychology, University of Toronto). USVs have been proposed as a measure of affect in rodents. In order to investigate whether stimuli that evoke USVs would also increase cataplexy, we subjected female orexin KO mice to a social reunion paradigm; female mice were used as males will respond to social reunion with aggressive behaviors, while females respond positively. Social reunion is an effective method to elicit USVs in mice (Irie et al., 2012). At 12:00, seven hours before testing, female mouse pairs were separated, with one mouse remaining in the home cage while the other was placed in an identical cage with fresh bedding and food. At the beginning of the testing session (19:00), the separated mouse was re-introduced into the home cage. Video and audio recordings were taken for 20 minutes after the reunion. Cataplexy was identified by an experienced observer, blinded to the genotype of the mouse pair, using videography. We were not able to use the consensus definition of murine cataplexy to identify arrests because of the absence of EEG/EMG (Scammell et al., 2009). Therefore, all sudden postural collapses of greater than 10 seconds were classified as cataplexy. USVs were quantified by an observer experienced in scoring digital acoustic wave patterns (Petri Takala), using Avisoft SASLab Pro software (Avisoft Bioacoustics, Berlin, Germany) Statistics The statistical tests used for analysis are included in the text of the results section. All statistical analyses were computed using SigmaStat (SPSS Inc., Chicago, IL) and applied a critical α value of p<0.05. Data are presented as mean ± SEM. 3.4 Results Cataplexy in orexin KO mice occurred almost exclusively during the dark period (n=9; Figure 3.1), whereas REM sleep occurred during both the dark and light periods. During the

64 49 time when cataplexy occurs, amounts of REM sleep and cataplexy are similar: during the dark period orexin KO mice spent 2.6% of the time in cataplexy and 3.5% of the time in REM sleep. The average duration of each bout of REM sleep and cataplexy was similar during the dark period (61 ± 3s vs. 66 ± 5s; paired t-test; p=0.394), while the number of bouts was also similar (25 ± 3 vs. 18 ± 4; p=0.230) Muscles exhibit atonia during both REM sleep and cataplexy In orexin KO mice (n=11), we analyzed masseter muscle EMG to quantify muscle tone during cataplexy and REM sleep. Masseter muscle tone during cataplexy was lower than REM sleep muscle tone (t-test, p=0.024; Figure 3.2A and B). During REM sleep, muscles exhibit phasic muscle twitches on a background of muscle atonia; by comparing the frequency of muscle twitches during REM sleep and cataplexy, we observed that these twitches generally did not occur during cataplexy (p=0.001; Figure 3.2A and C) and those that were detected were significantly smaller in amplitude (p=0.010; Figure 3.2D). Removing phasic twitch activity from our measure of muscle tone during REM sleep demonstrated that the suppression of masseter muscle tone during REM sleep atonia and cataplexy was similar (p=0.557; Figure 3.2E).

65 50 Figure 3.1: REM sleep and cataplexy occur in similar amounts during the dark period A. Cataplexy occurred almost exclusively during the dark period, but REM sleep and cataplexy occurred in similar amounts during the dark period. B and C. The frequency and duration of REM sleep and cataplexy episodes was similar during the dark period. *, p<0.05.

66 51 Figure 3.2: Muscle tone during REM sleep and cataplexy. A. Raw traces demonstrating masseter muscle tone during REM sleep and cataplexy. Phasic muscle twitches (see arrows) were largely absent during cataplexy. B. Overall EMG tone during cataplexy was lower than during REM sleep. C and D. Muscle twitches that occurred during REM sleep were largely absent during cataplexy, and those that remained were smaller in amplitude. E. Muscle tone during cataplexy was similar to levels during REM sleep atonia. A.U., arbitrary units; *, p<0.05.

67 Putative positive emotions trigger cataplexy in mice To investigate the relationship between cataplexy and emotion in orexin KO mice, we looked at the association between USVs and cataplexy. It is hypothesized that USVs are a measure of affect in rodents (Knutson et al., 2002, Burgdorf et al., 2007, Wang et al., 2008). Because we were interested in whether USVs are associated with cataplexy, all experiments were performed in orexin KO mice. Using a social reunion paradigm previously demonstrated to induce USVs (data not shown), we observed cataplexy in 56% of testing sessions and a trend towards more vocalizations during sessions where cataplexy was observed (Figure 3.3A). Mice that tended to vocalize more spent increased time in cataplexy; there was a positive correlation between total number of USVs and total time spent in cataplexy across the recording period (R 2 =0.55; P= Figure 3.3B). In order to investigate whether there was a temporal relationship between USVs and cataplexy onset, we separated USVs into time bins based on when in the recording period they occurred: those that occurred during the first 2 minutes post-reunion, those that occurred in the minute preceding an episode of cataplexy, and those that occurred any other time across the recording period. Despite many USVs being emitted at the beginning of the testing session, cataplexy was not associated with these USVs, as no episodes of cataplexy were observed in the 2 minutes post-reunion. There was, however, a trend towards more USVs being emitted in the minute preceding an episode of cataplexy than on average across the rest of the recording period (Figure 3.3C). When we looked at individual episodes of cataplexy, we observed that only 29% of episodes were immediately (<1 min) preceded by USVs, although the average latency between the last USV and the onset of cataplexy was very short (9 ± 4s) in these episodes, suggesting that there may be a temporal relationship between USVs and cataplexy in specific cases. These results suggest that emotionally salient stimuli increase cataplexy in orexin KO mice.

68 53 Figure 3.3: USVs are associated with cataplexy in orexin KO mice A. Orexin KO mice show increased vocalizations in testing sessions in which cataplexy occurred. B. In testing sessions during which cataplexy occurred, total numbers of vocalizations across the recording period positively correlate with total time spent in cataplexy (n=19 testing sessions). C. In the one minute preceding a cataplectic attack, mice had more USVs than they did on average throughout the rest of the recording period. This suggests that there may be a loose temporal relationship between the occurrence of USVs and the onset of cataplexy.

69 Stimulating environments increase cataplexy but decrease REM sleep To determine whether cataplexy inducing conditions could lead to changes in the occurrence of REM sleep we gave orexin KO mice access to a running wheel and chocolate. These stimuli together significantly increased amounts of cataplexy by 92% (n=8; 2.6 ±0.4% vs. 5.0 ±0.6%; RM ANOVA; p<0.001). Wheel running (WR) alone increased frequency of cataplexy by 83% (18 ±4 vs. 33 ±6; p=0.013; Figure 3.4B), while the average duration of cataplexy was reduced (66 ±5s vs. 40 ±6s; p<0.001; Figure 3.4C). Chocolate in addition to WR increased the frequency of cataplexy by 272% when compared to baseline conditions (18 ±4 vs. 67 ±10; p<0.001, Figure 3.4B); the average duration of cataplexy was reduced (66 ±5s vs. 34 ±2s; p=0.006; Figure 3.4C). These stimulating environments increased cataplexy while simultaneously decreasing amounts of REM sleep by 54% (3.5 ± 0.4% vs. 1.6 ±0.3%; p<0.001; Figure 3.4D) and 89% (3.5 ± 0.4% vs. 0.4 ±0.2%; p<0.001) during the WR and WR and chocolate conditions, respectively. WR alone decreased the frequency of REM sleep by 48% (25 ± 3% vs. 13 ±2%; p<0.001; Figure 3.4E) while having no effect on duration of bouts (61 ± 3s vs. 55 ±4s; p=0.731). WR in addition to chocolate decreased the frequency of REM sleep by 84% (25 ± 3% vs. 4 ±2%; p<0.001; Figure 3.4E) while not significantly affecting the duration of REM sleep bouts (61 ± 3s vs. 38 ±8s; p=0.061; Figure 3.4F). These data demonstrate that stimuli that increase the probability of cataplexy decrease REM sleep amounts. Both WR and WR and chocolate conditions also affected NREM and wake amounts. Waking during the dark period was significantly increased by 20% under the WR condition (RM ANOVA; p<0.001; Figure 3.5A) and by 38% under the WR and chocolate condition (p<0.001). WR and chocolate had such a potent wake promoting effect that orexin KO mice maintained wakefulness for 90% of the dark period despite chronic sleepiness. NREM sleep was reduced under stimulating conditions, decreasing by 42% (p<0.001) and 85% (p<0.001) during the WR and WR and chocolate conditions, respectively (Figure 3.5B). Because cataplexy occurs during waking it is possible that the increases in cataplexy observed with environmental stimuli are due to increased waking; however, this is unlikely to be the case as the increase in wake is not proportional to the increase in cataplexy (ex: a 38% increase in waking vs. 92% in cataplexy). We also expressed cataplexy with respect to the amount of waking in each condition (i.e. cataplexy/wake %) and see no changes in the significance of observed trends (data not shown).

70 55 Figure 3.4: Wheel running and chocolate increased cataplexy and decreased REM sleep A, B and C. Wheel running (WR) and WR with chocolate increased cataplexy during the dark period in orexin KO mice. Increases were due to an increase in the number of episodes of cataplexy as the average duration of cataplexy bouts decreased with these stimuli. D, E, and F. Wheel running (WR) and WR with chocolate decreased REM sleep in orexin KO mice. The observed decrease was due to a decrease in the number of bouts of REM sleep. * denotes a significant difference from baseline; ** denotes a significant difference from baseline and WR; p<0.05.

71 56 Figure 3.5: Stimulating environments increased waking and decreased NREM sleep during the dark period A and B. Wheel running (WR) and WR with chocolate increased amounts of wakefulness and decreased amounts of NREM sleep during the dark period in orexin KO mice. * denotes a significant difference from baseline; ** denotes a significant difference from baseline and WR; p<0.05.

72 Increasing REM sleep pressure does not affect cataplexy To test whether increasing REM sleep pressure can increase the occurrence of cataplexy, we selectively REM sleep-deprived orexin KO mice (n=10) for 4 hours prior to the onset of the dark period. During the 4-hour deprivation period, REM sleep was significantly decreased compared to baseline (2-way RM ANOVA, p<0.001; Figure 3.6A). This decrease was due to significantly fewer (p<0.001; Figure 3.6B) and shorter bouts of REM sleep (p<0.001; Figure 3.6C). However, the deprivation was not completely selective as there was a significant decrease in NREM sleep (p<0.001; Figure 3.6D) and an increase in waking (p<0.001; Figure 3.6E). To control for this loss of sleep, a sham deprivation was performed that resulted in the same loss of NREM sleep and increase in waking, while losing less REM sleep over the deprivation period (Sham group vs. Deprivation group, p<0.001; Baseline vs. Sham group, p<0.001; Figure 3.6). As previously demonstrated, cataplexy rarely occurs during the light period; we nonetheless analyzed it during the deprivation period and did not observe significant changes in the amount of cataplexy (Figure 3.6A). After selective REM sleep deprivation there was no significant difference observed between groups during any sleep state when averaged across the entire 12 hour dark period (RM ANOVA, p>0.05 for all comparisons; Figure 3.7A and B). Cataplexy was not significantly different across groups either, with similar amounts of cataplexy, bouts and durations of attacks being observed in all groups (Figure 3.7B-D). When viewed over a full 12-hour period, the 4- hour REM sleep deprivation did not have an effect on cataplexy. Rebound after selective deprivation could act at shorter timescales than 12 hours. In order to determine if state-specific rebounds were occurring at shorter timescales, we investigated cumulative loss plots across the entire recording period. Figures 3.8A and 3.9A demonstrate the deficit in REM sleep and cataplexy compared to baseline that is accumulated and regained over the entire recording period (including one hour of pre-deprivation baseline recording and 4 hours of the next light period). A negative slope confers a loss of that particular state compared to baseline recording, a positive slope confers a gain of that particular state, while a slope parallel to the x-axis means no change. Figure 3.8A demonstrates that a REM sleep deficit was accrued in both the sham and deprivation groups, although the magnitude of the deficit was greater in the deprivation group. Interestingly, figure 3.9A demonstrates that cataplexy seems to accrue a deficit during the first part of the dark period, perhaps while some

73 58 sleep is being regained, then a small rebound in cataplexy is observed toward the end of the dark period in the deprivation group. These observations suggest that shorter timescales are required to investigate the effects of REM sleep deprivation on cataplexy. Observing changes in REM sleep and cataplexy in one hour intervals demonstrated some effects of selective REM sleep deprivation. Focusing first on REM sleep, we observed significant loss of REM sleep in the deprived group, and to a lesser extent in the sham group, during the deprivation. There was a significant rebound in REM sleep during the first hour of the dark period in both the deprivation and sham groups (RM ANOVA, p<0.05; Figure 3.8B), indicating that our intervention did cause an expected homeostatic rebound in REM sleep. The deprivation group showed another increase in REM sleep during the third hour of the dark period. The amount of cataplexy observed was decreased in both the sham and deprivation groups during the first and third hour of the dark period when compared to control mice (RM ANOVA, p<0.05; Figure 3.9B). This would indicate that increasing REM sleep pressure inhibits cataplexy. Over the remainder of the dark period we did not observe any changes in the amount of cataplexy, except between 5:00-6:00 when there was a significant increase in cataplexy only in the REM sleep deprived group. As mentioned previously, it is possible that changes in cataplexy could be a product of changes in the amount of waking. This does not appear to account for the effects seen here, as wakefulness was not affected at any point during the dark period following deprivation (Figure 3.10). In agreement with this, calculating cataplexy amounts relative to waking (i.e. cataplexy/wake %) did not change the significance of the results (data not shown). It is possible that the deprivation led to a change in general arousal and perhaps locomotion, which caused a change in the amount of cataplexy; however, we could not measure these parameters.

74 59 Figure 3.6: Selective REM sleep deprivation reduced both NREM and REM sleep A. REM sleep was significantly reduced in the REM deprivation group when compared to baseline and sham deprivation groups. Cataplexy was rarely observed during the 4h deprivation period. B and C. Both the number and the average duration of REM sleep bouts were decreased in the REM sleep deprivation group. D. Deprivation was not selective to just REM sleep as NREM sleep amount was also reduced in both the sham and deprivation groups compared to baseline, with a concomitant increase in wakefulness. *, p<0.05 compared to baseline; **, p<0.05 compared to baseline and sham groups.

75 60 Figure 3.7: REM sleep deprivation had no effect on behavioral state over the following dark period A and B. Over the entire dark period there was no significant difference in wake, NREM sleep, REM sleep or cataplexy in either the sham or deprivation group compared to baseline. C and D. There was no difference in the occurrence or duration of cataplexy periods during the dark period in sham or deprivation groups when compared to baseline.

76 61 Figure 3.8: REM sleep accrued a deficit during the deprivation that was in part recovered over the following dark period A. A deficit in REM sleep is accrued in both the sham and deprivation groups during the deprivation, although the magnitude of the deficit is greater in the deprivation group. This deficit is then partially recovered over the ensuing dark period. B. Hourly amounts of REM sleep demonstrated successful REM sleep deprivation and a brief REM sleep rebound in both the REM sleep deprived group and sham group. *, p<0.05 compared to Sham and REM Dep groups; #, p<0.05 compared to REM Dep group; $, p<0.05 when compared to Control group.

77 62 Figure 3.9: Cataplexy was reduced at the beginning of the dark period following deprivation A. In both the sham and deprivation groups a cataplexy deficit is accrued during the first part of the dark period. B. Hourly amounts of cataplexy demonstrating a short term decrease in cataplexy during the first part of the dark period. Only between 5:00 and 6:00 was cataplexy significantly greater in the REM deprivation group. *, p<0.05 compared to Sham and REM Dep groups; $, p<0.05 when compared to Control and Sham groups.

78 63 Figure 3.10: Wakefulness during the dark period was not significantly affected by REM sleep deprivation The amount of wake was significantly elevated in the sham and REM sleep deprivation group during the first three hours of the deprivation, but was not different at any point during the dark period. *, p<0.05 compared to Sham and REM Dep groups; $, p<0.05 when compared to Control group.

79 Discussion Cataplexy and REM sleep share many common features, including theta-rich EEG, loss of muscle tone and reduction of the H-reflex (Guilleminault et al., 1974, Guilleminault et al., 1998, Chemelli et al., 1999, Overeem et al., 2004). This has led to the hypothesis that cataplexy is the expression of REM sleep atonia during wakefulness. Here we performed a series of experiments to address the relationship between REM sleep and cataplexy in orexin KO mice in order to shed light on whether they share a common mechanism. We observed some phenotypic similarities between REM sleep and cataplexy; however, increasing propensity for either cataplexy or REM sleep did not increase occurrence of the other state, as might be expected if they were generated by the same mechanism REM sleep and cataplexy are similar states We first observed the natural occurrence and characteristics of REM sleep and cataplexy. REM sleep occurs both during the light and dark periods, though generally more during the light period, the normal sleeping period for mice. Cataplexy occurs almost exclusively during the dark period. During the dark period both states occur in similar amounts, with both the number of episodes and the average duration of episodes being similar. These data demonstrate similarities between REM sleep and cataplexy in orexin KO mice that suggest these states may be regulated by similar mechanisms. It is difficult to compare amounts of cataplexy observed between labs as recording conditions (cage, tether, etc.), techniques (whether EMG, EEG, and video were used), and the mice themselves (number of backcrosses, background, etc.) may change the expression of cataplexy; however, these values are within the range of values observed previously in orexin-deficient mice (Chemelli et al., 1999, Mochizuki et al., 2004, Espana et al., 2007, Kalogiannis et al., 2011) Muscle tone during REM sleep and cataplexy is similar We observed that muscle tone during episodes of cataplexy was similar to REM sleep, only with an absence of muscle twitches. Although human narcoleptics sometimes report muscle twitches during cataplexy, this finding is supported by work in narcoleptic canines demonstrating a lack of rapid eye movements and phasic activity, common during REM sleep, during cataplexy

80 65 (Siegel et al., 1991, Siegel et al., 1992, Overeem et al., 2011). Several different brainstem regions have been proposed as generators of muscle atonia (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b). Siegel and colleagues demonstrated that a population of neurons in the medial medulla thought to be involved in generating REM sleep atonia were active only during REM sleep and cataplexy, suggesting that these neurons could ultimately be responsible for atonia during both states (Siegel et al., 1991). In addition, they demonstrated a population of cells in the medial mesopontine region that are active during waking and REM sleep but not during cataplexy, which could be responsible for the phasic activity seen during REM sleep but absent during cataplexy (Siegel et al., 1992). The mechanisms generating REM sleep atonia are still unclear but recent work has suggested a number of brain regions and a combination of inhibition and disfacilitation of motor neurons (Soja et al., 1991, Kohlmeier et al., 1997, Boissard et al., 2002, Boissard et al., 2003, Morrison et al., 2003a, Fenik et al., 2004, Fenik et al., 2005b, a, Sood et al., 2005, Chan et al., 2006, Lu et al., 2006b, Brooks and Peever, 2008, Burgess et al., 2008). Our lab has demonstrated a key role for GABAergic and glycinergic mechanisms in regulating muscle atonia, while glutamatergic inputs to motor neurons are responsible for muscle twitches during REM sleep (Brooks and Peever, 2008, Burgess et al., 2008). Selective activation of the inhibitory mechanisms, without recruitment of the excitatory inputs, may generate atonia during cataplexy. Further investigation of REM sleep atonia-generating brain regions during cataplexy is needed to determine whether the loss of muscle tone during these two states is caused by the same mechanism. A limitation of this study is that only one muscle was investigated and the activity averaged over the entire episode of each state. Recent work has shown that different types of muscles can be differentially suppressed during REM sleep and that muscle tone may change subtly over the course of a single episode of REM sleep or cataplexy (Fraigne and Orem, 2011, Kalogiannis et al., 2011). A more detailed analysis of several different muscles during cataplexy may yield different results.

81 Positive affective stimuli induce cataplexy and suppress REM sleep in mice Laughter is reported to be the best trigger for cataplexy in human patients (Gelineau, 1880, Overeem et al., 2011). Positive affect also causes cataplexy in narcoleptic canines, as social play and palatable food are good triggers of cataplexy (Mitler and Dement, 1977, Baker et al., 1982, Kushida et al., 1985, Siegel et al., 1986). It is more difficult to determine the emotional state of mice; however, using a proposed correlate of affect in rodents (USVs) we show that emotionally salient stimuli increased cataplexy in orexin KO mice (Knutson et al., 2002, Burgdorf et al., 2007). We then used this finding to promote cataplexy with other rewarding stimuli (WR and chocolate). Mice will bar press for access to a running wheel and for palatable food, indicating that they value these stimuli (Holahan et al., 2011, King et al., 2011) and it has been previously demonstrated that access to a running wheel and palatable food each separately can increase cataplexy in orexin KO mice (Espana et al., 2007, Clark et al., 2009). We observed that these stimuli increased the occurrence of cataplexy while they significantly reduced REM sleep. If REM sleep and cataplexy were generated by the same neural mechanisms, one might expect stimuli that promote cataplexy to also promote REM sleep. These stimuli may have actively suppressed REM sleep circuits while simultaneously activating cataplexy circuits. A number of methodological limitations restrict the conclusions that can be drawn from these experiments, including that chocolate contains a number of compounds, including caffeine which powerfully affects sleep-wake regulation that could be responsible for the observed effects on cataplexy rather than positive affect. This concern has been partly addressed by another study, using a different food stimulus that also observed a link between palatable food and cataplexy in orexin KO mice (Clark et al., 2009). In addition, caffeine has been demonstrated to increase wakefulness in orexin KO mice while having no significant effect on cataplexy (Willie et al., 2003). A further limitation of this study is that the observed relationship between USVs and cataplexy is preliminary and correlative, making strong conclusions difficult. In addition, we found both USVs and cataplexy to be widely variable between mice, requiring a large number of testing sessions and a large number of mice to make meaningful conclusions beyond those made here. As mentioned previously, we did not have EEG/EMG data to score episodes of cataplexy; this may have led to incorrect identification, particularly as orexin KO mice exhibit

82 67 rapid transitions into NREM sleep (i.e. sleep attacks) that could be mistaken for cataplexy. We don t think this is the case as the intervention (i.e. social reunion) is alerting and was performed at the onset of the dark period when mice are generally alert. Mochizuki et al (2004) demonstrated that orexin KO mice can maintain wakefulness for ~45 minutes after a cage change, while we only measured for 20 minutes post-reunion (Mochizuki et al., 2004). Despite these limitations, the implications of this simple experiment are interesting, particularly as the circuits responsible for USVs have been characterized, and may overlap with the circuits that ultimately trigger cataplexy (Burgdorf et al., 2007) REM sleep pressure does not significantly increase cataplexy We observed that increasing REM sleep pressure did not increase cataplexy over the following dark period. Cataplexy was initially suppressed during the dark period after REM deprivation; however, there was one time point toward the end of the dark period that showed an increase in cataplexy only in the REM deprivation group. It is unclear if this increase resulted from the REM sleep deprivation, a homeostatic rebound-like response to the decreased cataplexy at the start of the dark period, or an unrelated phenomenon. A recent study in human patients with narcolepsy also investigated the relationship between increasing REM sleep pressure and cataplexy. Vu and colleagues performed two nights of selective REM sleep deprivation in human patients with narcolepsy and observed no change in cataplexy or other REM sleep related narcolepsy symptoms (i.e. hallucinations and sleep paralysis), while they observed a normal REM sleep homeostatic rebound (Vu et al., 2011). If REM sleep and cataplexy were generated by the same neural mechanisms, one might expect interventions that promote a REM sleep rebound to increase cataplexy. REM sleep deprivation did not appear to significantly increase cataplexy in orexin KO mice or human narcolepsy patients. Our investigation of whether REM sleep pressure can increase the occurrence of cataplexy has methodological limitations that make more detailed analysis difficult. While we successfully reduced REM sleep, the intervention was not selective as NREM sleep was also affected. We did observe a REM sleep rebound in the deprived group indicating there was increased REM sleep pressure, and previous studies have demonstrated as little as two hour of REM sleep deprivation is enough to result in a rebound (Shea et al., 2008). However, it is possible that a longer term deprivation may be required to uncover effects on cataplexy. There

83 68 are automated, selective REM deprivation protocols that could be used to address these concerns and may yield different findings. In addition, a less stressful approach to increasing REM sleep, like increasing ambient temperature, could be used to investigate the effects of increased REM pressure on cataplexy (Szymusiak et al., 1980, Amici et al., 1998, Baker et al., 2005) REM sleep and cataplexy do not share a common executive mechanism The experiments and analyses described in this chapter were designed to address the question of whether REM sleep and cataplexy share a common mechanism. Our data suggest that REM sleep and cataplexy are unique behavioral states that do not share an executive mechanism. Given the similar loss of muscle tone during cataplexy and REM sleep it is possible they share the same mechanisms that ultimately generate muscle atonia. In support of this it has been previously demonstrated that REM sleep atonia-promoting neurons in the medulla fire selectively during REM sleep and cataplexy in narcoleptic dogs (Siegel et al., 1991). We propose that the mechanisms that generate REM sleep and cataplexy converge at brain regions demonstrated to generate muscle atonia (Figure 3.11). An alternative proposal to the theory that cataplexy is a disorder of REM sleep is that it is an atavistic expression of tonic immobility (TI) (Overeem et al., 2002). TI is a defense mechanism seen in some animals (ex: sharks, guinea pigs, chickens, and rabbits, though not humans or mice) that can be similar to cataplexy. During periods of TI, animals can show both stereotyped postures and flaccidity, or loss of muscle tone; they also show a wake-like EEG activity, awareness of their external environment, and reduced heart rate (Klemm, 1971a, b). Studies suggest amygdala projections to the brainstem regions that generate muscle atonia during REM sleep are involved in TI (Klemm, 1976, Leite-Panissi et al., 2003). In addition, muscle twitches and phasic eye movements, defining features of REM sleep that are largely absent during cataplexy, are absent during TI (Braun and Pivik, 1983, Overeem et al., 2002). Our data could support the hypothesis that cataplexy is an atavistic expression of TI, that ultimately triggers loss of muscle tone through brainstem REM sleep atonia-generating sites.

84 69 Figure 3.11: Cataplexy and REM sleep are generated by different mechanisms The mechanisms that regulate cataplexy and REM sleep are unique. While REM sleep is homeostatically regulated, cataplexy can be triggered by positive affective stimuli. We propose that both systems ultimately trigger loss of muscle tone by activating muscle atonia-generating regions in the pons, resulting in inhibition of motor neurons and loss of muscle tone. Green arrows indicate excitatory projections and red lines indicate inhibitory projections.

85 70 Chapter 4: Dopaminergic Regulation of Sleep and Cataplexy This chapter has been adapted from a published manuscript in the journal Sleep (Burgess et al. 2010) Other researchers contributed to this work: Gavin Tse, MSc: Assisted with the wild type mouse amphetamine experiments Lauren Gillis, BSc: Genotyped mice and assisted with the organization of the mouse colony

86 71 Chapter 4: Dopaminergic Regulation of Sleep and Cataplexy 4.1 Abstract Narcolepsy is characterized by excessive sleepiness and cataplexy, the sudden loss of postural muscle tone during waking. The mechanisms that underlie sleepiness and cataplexy are unclear; however, there is evidence that dysregulation of the dopaminergic system may have a role. To establish whether the dopaminergic system plays a role in murine narcolepsy, cataplexy, sleep attacks and sleep-wake behavior were monitored after injection of saline, amphetamine or specific dopamine receptor modulators. Amphetamine (2mg/kg) decreased sleep attacks and cataplexy, suggesting that dopamine transmission could modulate such behaviors. Specific dopamine receptor modulation also affected sleep attacks and cataplexy. Activation and blockade of D1-like receptors decreased and increased sleep attacks, respectively, without affecting cataplexy. Pharmacological activation of D2-like receptors increased cataplexy and blockade of these receptors potently suppressed cataplexy. Manipulation of D2- like receptors did not affect sleep attacks. We found that cataplexy is modulated by a D2-like receptor mechanism, whereas dopamine modulates sleep attacks by a D1-like receptor mechanism. These results support a role for the dopamine system in regulating sleepiness and cataplexy in murine narcolepsy and suggest that cataplexy and REM sleep can be differentially regulated. 4.2 Introduction Narcolepsy is characterized by excessive daytime sleepiness, cataplexy, hypnagogic hallucinations and sleep paralysis (Siegel and Boehmer, 2006). Excessive sleepiness and cataplexy, the involuntary loss of postural muscle tone during waking, are the most debilitating symptoms of the disorder (Siegel and Boehmer, 2006). Although loss of orexin neurons underlies narcolepsy, the specific neurochemical mechanisms that trigger sleepiness and cataplexy are still unknown (Peyron et al., 2000, Thannickal et al., 2000, Thannickal et al., 2003, Blouin et al., 2005). Abnormalities in dopaminergic neurotransmission may contribute to both sleepiness and cataplexy. For example, clinical studies demonstrate that human narcoleptics have an altered striatal dopaminergic system. Specifically, brain imaging studies show that narcoleptics have

87 72 increased D2-like receptor binding that is tightly correlated with cataplexy (Eisensehr et al., 2003). Drugs used to treat human narcolepsy and other sleep disorders affect the dopamine system. For example, amphetamine, modafinil and gamma-hydroxybutyrate, which are used to treat sleepiness and cataplexy, have effects on the dopamine system (e.g., dopamine reuptake/release and receptor expression) (Howard and Feigenbaum, 1997, Schmidt-Mutter et al., 1999, Wisor et al., 2001, Wisor and Eriksson, 2005). Animal studies also show that sleep and cataplexy can be manipulated by the dopamine system (Bagetta et al., 1988, Monti et al., 1988, Monti et al., 1989, Monti et al., 1990, Ongini et al., 1993, Isaac and Berridge, 2003). Pharmacological manipulation of D2-like, but not D1-like, receptors in dopaminergic brain areas (e.g. substantia nigra and ventral tegmental area) modulates cataplexy in narcoleptic dogs (Reid et al., 1996, Honda et al., 1999b, Okura et al., 2004). Sleep too is controlled by dopaminergic mechanisms. For example, loss of wake-active dopamine cells in the ventral periaquaductal gray promotes sleep in rats (Lu et al., 2006a). Despite evidence linking the dopamine system and narcolepsy symptoms, it is unknown if manipulation of dopamine receptors affects sleep or cataplexy in murine narcolepsy. Here, we used orexin KO mice, which serve as a model of human narcolepsy (Chemelli et al., 1999, Mochizuki et al., 2004), to determine if modulation of dopamine receptors can affect cataplexy and sleep attacks (Chemelli et al., 1999). We show that cataplexy is predominantly mediated by D2-like receptors, whereas sleep attacks are modulated by a D1-like receptor mechanism. These data demonstrate that dopaminergic mechanisms contribute to narcolepsy symptoms and further establish the orexin KO mouse as a useful model for studying the mechanisms underlying sleepiness and cataplexy. 4.3 Methods All procedures and experimental protocols were approved by the University of Toronto s animal care committee and were in accordance with the Canadian Council on Animal Care Animals Experiments used male, orexin KO mice on a C57BL/6 background (n=17; age: 15.1 ± 1.0 weeks; mass: 29.2 ± 0.8g) and male wildtype littermates (n=21; age: 15.9 ± 0.8 weeks; mass: 28.1 ± 0.9g). Mice were genotyped using PCR with genomic primers 5'-GACGACGGCCTCAG ACTTCTTGGG, 3'-TCACCCCCTTGGG ATAGCCCTTCC, and 5 -CCGCTATCAGGACATA

88 73 GCGTTGGC (with forward primers being specific for either wildtype or KO mice and the reverse primer being common to both) Surgery Mice were anesthetized using isoflurane (1-2%) and implanted with EEG and EMG electrodes. EEG recordings were obtained using four stainless steel micro-screws (1mm anterior ±1.5mm lateral to bregma; 3mm posterior ±1.5mm lateral to bregma). EMG electrodes were made from multistranded stainless steel (AS131, Cooner Wire, Chatsworth, CA) wires, which were sutured onto both neck and left/right masseter muscles. All electrodes were attached to a micro-strip connector (CLP L-D, Electrosonic, Toronto, ON), which was affixed onto the animal s head with dental cement (Ketac-cem, 3M, London, ON). After surgery, mice were given 0.9% saline and ketoprofen (3mg/kg). Mice were individually housed in a soundattenuated and ventilated chamber on a 12:12 light-dark cycle (110 Lux; lights on 7:00, lights off 19:00) for days post surgery. Food and water were available ad libitum Drug preparation The following drugs were used to modify dopaminergic transmission: quinpirole (0.125 and 0.5mg/kg; a D2-like receptor agonist; Tocris, Ellisville, MO), eticlopride (0.25 and 1mg/kg; a D2-like receptor antagonist; Sigma Aldrich, Oakville, ON), SKF (5 and 20mg/kg; a D1- like receptor agonist; Tocris), SCH (0.25 and 1mg/kg; a D1-like receptor antagonist; Tocris) and amphetamine (2mg/kg; Sigma Aldrich). Drugs were made from frozen stock solutions before each i.p. injection. Dose ranges were chosen based on previous studies demonstrating behavioral effects in mice (Gessa et al., 1985, Zarrindast and Tabatabai, 1992, Tirelli and Witkin, 1995, Ralph and Caine, 2005). In this study we refer to D1-like or D2-like dopamine receptors. D1-like receptors include D1 and D5 receptors while D2-like receptors include D2, D3, and D4 receptors. The drugs used are selective for either D1-like or D2-like receptors. Ki values for quinpirole are 4.8, 24, 30 and 1900nM at D2, D3, D4 and D1 receptors, respectively. Ki values for eticlopride are 0.50 and 0.16nM at D2 and D3 receptors, respectively. Ki values SKF are 1, 0.5, 150, 5000 and 1000nM for D1, D5, D2, D3 and D4 receptors, respectively. Ki values for SCH are 0.2, 0.3, 1100, 800 and 3000nM at D1, D5, D2, D3 and D4 receptors, respectively.

89 Data acquisition Sleep-wake state and muscle activity were recorded by attaching a lightweight cable to the plug on the mouse s head, which was connected to a Physiodata Amplifier system (Grass 15LT, Astro Med, Brossard, QC). The EEG activity was amplified 1000 times and band-pass filtered between 1 and 100 Hz. EMG signals were amplified 1000 times and band-pass filtered between 30 Hz and 1 khz. All electrophysiological signals were digitized at 500Hz (Spike 2 Software, 1401 Interface, CED Inc.) and monitored and stored on a computer. Infra-red video recordings were captured and synchronized with the electrophysiological recordings to couple motor behavior with EEG and EMG recordings Experimental protocols Mice were placed in a round plexi-glass cage (diameter: 20cm) and given 24 hours to habituate to this new environment. After this period, mice were connected to the recording apparatus and given another 48 hours to habituate at which point a habituation injection (i.e., saline) was given. In one group of mice, a single dose of amphetamine (n=12) was given and sleep, cataplexy and sleep attacks recorded. In another group of mice (n=26), dopamine drug injections were given, each separated by 48 hours; injections were given in random order. All injections (0.3mL i.p.) were given at the onset of the dark phase (i.e., 19:00) and behavior monitored for the following 4 hours Data analysis Data was collected for 4 hours after injections and was scored, using EEG, EMG (neck and masseter) and video. Each of the 5 second epochs was scored as wake, non-rapid eye movement (NREM) sleep, rapid eye movement (REM) sleep, cataplexy, sleep attack or transition state (e.g., NREM-REM). Sleep attacks were classified as a gradual loss of neck muscle tone associated with NREM-like EEG characteristics and automatic behavior. In narcoleptic mice, automatic behavior is defined as chewing, which we confirmed by both videography and masseter EMG recordings. Cataplexy was classified as a sudden loss of skeletal muscle tone in both neck and masseter following at least 40 seconds of active waking and with a duration of at least 10 seconds (Scammell et al., 2009). Both videography and electrophysiological recordings were used to identify sleep-wake behavior, cataplexy and sleep attacks.

90 75 The frequency and duration of cataplexy and sleep attacks, as well as the total time spent in cataplexy/sleep attacks was determined for each drug and compared to the saline treatment. To determine the total time spent in cataplexy, we summed the duration of each cataplectic attack across the 4-hour recording period. For narcoleptic and wildtype mice the time spent in each sleep-wake state for all drug treatments was determined and compared to the saline treatment. NREM and REM latency was defined as the time from drug administration to the first bout of each state Statistical analysis The statistical tests used for each analysis are included in the results section. Comparisons between frequency (i.e., number of bouts), duration and total time spent in cataplexy/sleep attacks were made using one-way repeated measures analysis of variance (RM ANOVA). Drug effects on sleep-wake state were made using a two-way RM ANOVA. Differences in sleep-wake behaviors between narcoleptic and wildtype mice were determined using 2-way ANOVA. All statistical analyses used SigmaStat (SPSS Inc.) and applied a critical 2-tailed α value of p<0.05. Data are presented as mean ± SEM. 4.4 Results Orexin KO mice exhibit cataplexy and sleep attacks During the 4-hour recording period (i.e., 19:00-23:00), orexin KO mice had an average of 1.9 ± 0.7 (range: 1-5) episodes of cataplexy that lasted 50 ± 11s (range: s). Cataplectic attacks occurred during periods of alert wakefulness and were characterized by postural collapse and loss of skeletal muscle tone with a theta-dominant, waking-like EEG pattern (Figure 4.1A and C). Cataplectic episodes were terminated by re-entrance into wakefulness, with mice resuming normal motor behaviors such as grooming or eating. During control conditions, 12% of narcoleptic mice (2 of 17 mice) did not present with cataplexy. Orexin KO mice also exhibited sleep attacks. Even though sleep attacks also occurred during active wakefulness, they differed from cataplexy because they were characterized by gradual loss of posture and muscle tone and because EEG activity patterns were NREM sleeplike in nature (Figure 4.1B and C). Another feature separating sleep attacks and cataplexy was automatic behavior. In orexin KO mice, automatic behavior was defined as chewing, which was visualized by repeated jaw movements and masseter EMG activity. Automatic behavior was

91 76 common during sleep attacks, but never observed during cataplexy. Sleep attacks were more frequent than cataplexy episodes, and on average mice exhibited 4.7 ± 1.0 (range: 2-9) episodes that lasted 30 ± 3s (range: 10-70s). Narcoleptic mice also had abnormal sleep-wake architecture (Figure 4.1D). Compared to wildtype littermates, orexin KO mice had more REM sleep (KO: 5.3 ± 0.8% and wildtype: 1.7 ± 0.4%; 2-way ANOVA, p=0.001) and more transitions into and out of sleep (p=0.009), while they spent the same amount of time in wakefulness (p=0.839) and NREM sleep (p=0.499; Figure 1D).

92 77 Figure 4.1: Cataplexy, sleep attacks and sleep-wake behavior in narcoleptic mice A and B. Raw EEG and EMG traces demonstrating the defining features of cataplexy and sleep attacks. C. Raw EEG and EMG traces showing masseter activity during cataplexy and a sleep attack. Note that muscle tone is absent during cataplexy, but cyclic during the sleep attack; this rhythmicity represents automatic chewing behavior, which is common in sleep attacks. D. Orexin knockout mice have significantly more REM sleep and a greater number of sleep-wake transitions than wildtype (wildtype) mice. * denotes p<0.05 when compared to wildtype mice.

93 Amphetamine reduced cataplexy and sleep attacks in narcoleptic mice We aimed to determine if amphetamine, which increases brain dopamine levels (Sharp et al., 1987), affects sleep and cataplexy in narcoleptic mice. Amphetamine stimulated wakefulness and suppressed sleep in both orexin KO and wildtype mice (n=6, 2-way RM ANOVAs; wildtype: p<0.001, KO: p<0.001; Figure 4.2A and B). It increased wakefulness by 41% above saline levels in orexin KO mice (p<0.001) but decreased NREM and REM sleep by 52% and 69%, respectively (NREM: p<0.001 and REM: p=0.002 Figure 4.2A). Amphetamine administration potently suppressed time spent in sleep attacks by 61% (RM ANOVA; p=0.032; Figure 4.2C). This reduction was due to a decrease in sleep attack frequency (44% below saline; RM ANOVA; p=0.030; Figure 4.2D) as duration of individual attacks was unaffected by amphetamine treatment (RM ANOVA; p=0.289; Figure 4.2E). Amphetamine also suppressed cataplexy, decreasing the number of episodes by 67% of saline levels (RM ANOVA, p=0.042; Figure 4.2G) without affecting the average duration of individual episodes (RM ANOVA, p=0.149; Figure 4.2H).

94 79 Figure 4.2: Amphetamine decreased sleep, sleep attacks and cataplexy A. Amphetamine increased wakefulness and decreased both NREM and REM sleep in narcoleptic mice. B. Amphetamine increased wakefulness, decreased NREM sleep and abolished REM sleep in wildtype mice. C-E. Amphetamine decreased total time spent in sleep attacks (C) and attack frequency (D), but had no significant affect on sleep attack duration (E). F-H. Amphetamine reduced the frequency of cataplectic episodes (G), but had no statistical effect on the total time spent in cataplexy (F) or episode duration (H). * denotes p<0.05 when compared to saline.

95 D1-like receptors modulate sleep attacks but not cataplexy First, we aimed to determine if blockade of excitatory D1-like receptors would reduce wakefulness and promote sleep in narcoleptic mice. Compared to saline treatment, low doses of the D1-like receptor antagonist (SCH 23390; 0.25mg/kg) had no effect on sleep-wake amounts (n=5; 2-way RM ANOVA, p>0.05; Figure 4.3A) or sleep latency (RM ANOVA, p=0.279). However, a higher dose of SCH (1mg/kg) decreased wakefulness by 25% (2-way RM ANOVA, p=0.016) and increased NREM sleep by 122% (p=0.024). Neither high nor low doses of SCH had significant effects on REM sleep amounts (2-way RM ANOVA, p=0.963; Figure 4.3A). Blockade of D1-like receptors also promoted sleepiness because sleep latency decreased from 1735 ± 717s (i.e., saline treatment) to 172 ± 17s following treatment with a high dose of SCH (RM ANOVA, p=0.033; data not shown). Sleep attacks were affected by D1-like receptor blockade. Compared to saline treatment, low doses of SCH (0.25mg/kg) had no effect on sleep attacks (RM ANOVA, p=0.101; Figure 4.3C); however, higher doses (1mg/kg) increased the total time spent in sleep attacks (RM ANOVA, p=0.009; Figure 4.3C) by increasing the number of sleep attacks by 88% (RM ANOVA, p=0.022; Figure 4.3D); this drug dose had no effect on the duration of individual attacks (RM ANOVA; p=0.265; Figure 4.3E). Blockade of D1-like receptors had no significant effect on cataplexy (RM ANOVA, p=0.870; Figure 4.3F-H). Next, we wanted to determine if D1-like receptor activation would increase wakefulness and decrease sleep in narcoleptic mice. Compared to saline treatment, both low (5mg/kg) and high doses (20mg/kg) of SKF increased wakefulness by 25% (n=6; 2-way RM ANOVA, p<0.001) and 23% (p<0.001), while decreasing NREM sleep by 88% (p<0.001) and 76% (p<0.001; Figure 4.4A). Neither high nor low drug doses had significant effects on REM sleep even though REM amounts decreased by 92% (p=0.157) and 98% (p=0.127). Activation of D1- like receptors also promoted arousal because sleep latency increased from 1768 ± 785s (i.e., saline) to ± 932s (RM ANOVA, p<0.001) and ± 1056s (p<0.001) following treatment of 5mg/kg and 20mg/kg of SKF (data not shown). Both high and low doses of SKF potently suppressed sleep attacks in narcoleptic mice (RM ANOVA, p=0.004; Figure 4.4C). Compared to saline treatment, high and low doses of SKF decreased the number of sleep attacks by 77% (p=0.004; Figure 4.4D) and 58%

96 81 (RM ANOVA, p=0.022), but neither dose had an effect on duration of attacks (p=0.560; Figure 4.4E). High and low doses of SKF completely abolished sleep attacks in 83% and 33% of narcoleptic mice. Neither high nor low doses of SKF had significant effects on cataplexy (RM ANOVA, p=0.549; Figure 4.4F-H). Lastly, we aimed to determine if D1-like receptor manipulation similarly affects sleepwake behavior in wildtype and narcoleptic mice. In wildtype mice, SCH administration (1.0mg/kg) decreased wakefulness by 13% (n=4; 2-way RM ANOVA, p=0.003) and increased NREM sleep by 73% (p=0.002; Figure 4.3B); whereas receptor activation by SKF treatment (20mg/kg) increased wakefulness by 23% (n=5; 2-way RM ANOVA, p=0.002) and decreased NREM sleep by 87% (p=0.005; Figure 4.4B). Neither intervention had significant effects on REM sleep amounts (SCH 23390: p=0.853; SKF 38393: p=0.706). The impact of D1- like receptor manipulation on sleep-wake behavior was similar in wildtype and orexin KO mice (2-way ANOVA; SCH 23390: p=0.295; SKF 38393: p=0.341).

97 82 Figure 4.3: Inactivation of D1-like receptors increased sleep attacks A. In narcoleptic mice, SCH (D1-like antagonist; 1mg/kg) increased NREM sleep and decreased wakefulness. B. In wildtype mice, SCH also increased NREM sleep and decreased wakefulness. C-E. SCH (1mg/kg) increased both the total time spent in sleep attacks (C) and sleep attack frequency (D), but had no affect on attack duration (E). F-H. SCH treatment had no affect on time spent in cataplexy (F) or on cataplexy frequency (G) or duration (H). * denotes p<0.05 when compared to saline.

98 83 Figure 4.4: Activation of D1-like receptors decreased sleep attacks A. In narcoleptic mice, SKF (D1-like agonist) decreased NREM sleep and increased wakefulness. B. In wildtype mice, SKF also decreased NREM sleep and increased wakefulness. C-E. SKF treatment decreased both the time spent in sleep attacks (C) and attack frequency (D), but had no affect on sleep attack duration (E). F-H. SKF had no affect on time spent in cataplexy (F) or on cataplexy frequency (G) or duration (H). * denotes p<0.05 when compared to saline.

99 D2-like receptors modulate cataplexy but not sleep attacks We administered quinpirole (0.125 and 0.5mg/kg) to activate D2-like receptors and eticlopride (0.25 and 1mg/kg) to inactivate them, in order to determine if D2-like receptors influence sleep and cataplexy. Neither high nor low doses of quinpirole affected amounts of sleep or wakefulness (n=7; 2-way RM ANOVA, p=0.262; Figure 4.5A); these interventions also had no effect on sleep attacks (RM ANOVA, p=0.357; Figure 4.5C-E). However, quinpirole had potent effects on cataplexy. The highest dose (0.5mg/kg) increased the number of cataplexy episodes by 172% above baseline levels (RM ANOVA, p=0.030; Figure 4.5G), without affecting cataplexy duration (RM ANOVA, p=0.107; Figure 4.5H). Even though this dose increased the total time spent in cataplexy by 174%, this effect was not statistically significant (RM ANOVA, p=0.193; Figure 4.5F). Activation of D2-like receptors with a modest quinpirole dose (0.125mg/kg) had no measurable effect on either the duration (RM ANOVA, p=0.058) or frequency (RM ANOVA, p=0.931) of attacks. Blockade of D2-like receptors with low doses (0.25mg/kg) of eticlopride decreased wakefulness by 18% (n=7; 2-way RM ANOVA, p=0.026; Figure 4.6A), increased NREM sleep by 90% (p=0.039), but had no effect on REM sleep (p=0.922). High doses of eticlopride (1mg/kg) had no detectable effects on sleep-wake behavior in narcoleptic mice (2-way RM ANOVA, p=0.366; Figure 4.6A). Even though high and low doses did not affect sleep attacks (RM ANOVA, p=0.758; Figure 4.6C-E), they had robust suppressive effects on cataplexy. Compared to saline treatment, high doses of eticlopride reduced the total time spent in cataplexy by 97% (RM ANOVA, p=0.024; Figure 4.6D); this decrease was attributable to the 88% reduction in the number of cataplectic attacks (RM ANOVA, p=0.029; Figure 4.6G). The duration of cataplectic episodes was not affected by high doses (RM ANOVA, p=0.240; Figure 4.6H) even though episode duration was suppressed by 78% below baseline levels. Although partial blockade of D2-like receptors by low eticlopride doses tended to suppress the frequency (RM ANOVA, p=0.104; Figure 4.6G) and duration (RM ANOVA, p=0.240; Figure 4.6H) of cataplectic episodes, these effects were not statistically significant. Finally, we aimed to determine if effects of D2-like receptor manipulation on sleep-wake behavior were similar in wildtype and orexin KO mice. In wildtype mice, neither quinpirole (0.5mg/kg) nor eticlopride (1.0mg/kg) affected sleep-wake amounts (2-way RM ANOVA; quinpirole: n=6, p=0.840; eticlopride: n=6, p=0.469; Figure 4.5B and Figure 4.6B). The

100 85 influence of D2-like receptor manipulation on sleep-wake behavior was similar in wildtype and orexin KO mice (2-way ANOVA; quinpirole: p=0.849; eticlopride: p=0.366).

101 86 Figure 4.5: Activation of D2-like receptors increased cataplexy A. Quinpirole (D2-like receptor agonist) had no affect on sleep-wake behavior in narcoleptic mice. B. Quinpirole also had no affect on sleep-wake behavior in wildtype mice. C-E. Quinpirole had no effect on the total time spent in sleep attacks (C) or on the frequency (D) or duration of attacks (E). F-H. Quinpirole did not significantly increase the total time spent in cataplexy (F), but at 0.5mg/kg it increased cataplexy frequency (G) without affecting duration (H). * denotes p<0.05 when compared to saline.

102 87 Figure 4.6: Blockade of D2-like receptors decreased cataplexy A. In narcoleptic mice, 0.25mg/kg of eticlopride (D2-like antagonist) increased NREM sleep and decreased wakefulness, but at 1mg/kg it had no effect on sleep-wake behavior. B. Eticlopride had no effect on sleep-wake behavior in wildtype mice. C-E. Eticlopride had no effect on the total time spent in sleep attacks (C) or on the frequency (D) or duration of attacks (E). F-G. Eticlopride (1mg/kg) decreased both the total time spent in cataplexy (F) and its frequency (G) without affecting the duration of cataplectic episodes (H). * denotes p<0.05 when compared to saline.

103 Discussion We demonstrate that the dopaminergic system modulates cataplexy and sleep attacks in a murine model of narcolepsy. Specifically, we showed that amphetamine suppresses both cataplexy and sleep attacks, suggesting that dopamine transmission could modulate these behaviors. We then showed that pharmacological activation of D2-like receptors triggers cataplexy and blockade of these receptors suppresses it. Manipulation of D2-like receptors did not influence sleep attacks. We also showed that activation and blockade of D1-like receptors decreased and increased sleep attacks, respectively; however, manipulation of D1-like receptors did not affect cataplexy. Our results suggest that dopamine transmission modulates cataplexy and sleep attacks by different receptor mechanisms Amphetamine alleviates cataplexy and sleep attacks We found that amphetamine suppressed cataplexy, sleep attacks and sleep. One mechanism by which amphetamine may exert its effects is by modulating dopamine transmission. Numerous studies show that systemic amphetamine application elevates dopamine levels and it is hypothesized that this elevation underlies amphetamine s arousal-promoting effects (Sharp et al., 1987, Di Chiara and Imperato, 1988, Wisor et al., 2001). Indeed, amphetamine and modafinil, both of which affect dopaminergic transmission, are clinically effective treatments for sleepiness and sleep attacks (Nishino et al., 1998b, Scammell and Matheson, 1998, Wisor et al., 2001). Amphetamine-induced changes in dopamine levels may also contribute to the suppression of cataplexy in narcoleptic mice. This assertion is supported by the fact that both systemic amphetamine administration and direct manipulation of dopaminergic nuclei modulate cataplexy in narcoleptic dogs (Shelton et al., 1995, Reid et al., 1996, Honda et al., 1999b, Okura et al., 2004). However, because amphetamine also increases levels of both serotonin and noradrenaline, it is possible that amphetamine-induced changes in cataplexy are mediated by multiple monoaminergic systems (Rothman and Baumann, 2006). Indeed, clomipramine, a tricyclic antidepressant that affects dopaminergic, noradrenergic and serotonergic transmission, reduces cataplexy without affecting sleep attacks in orexin KO mice (Willie et al., 2003).

104 A D2-like receptor mechanism modulates cataplexy This study showed that D2-like receptors play a significant role in regulating murine cataplexy. Dopamine receptor modulation of cataplexy showed specificity for the D2-like receptor as neither D1-like receptor activation nor blockade affected cataplexy even though D1 drugs have pronounced effects on sleep. A D2-like receptor mechanism has also been linked to canine narcolepsy; Nishino et al. (1991) demonstrated that activation of D2-like receptors increased cataplexy and blockade of these receptors decreased it in narcoleptic dogs (Nishino et al., 1991). It is noteworthy that D2 drugs modulate cataplexy in both canine and murine narcolepsy because canine narcolepsy results from an OX 2 R mutation, whereas the murine model used here results from loss of orexin itself. This is an important observation because it indicates that D2 drugs modulate cataplexy by an orexin-independent mechanism (Lin et al., 1999). We suggest that D2 drugs exert their effects on dopamine cells themselves. This is supported by the fact that D2-like receptor agonists promote cataplexy when applied directly onto dopaminergic neurons (Reid et al., 1996, Honda et al., 1999b, Okura et al., 2004). D2 drugs could act by manipulating auto-receptors on dopamine neurons to affect dopamine release, which has been shown to affect behavioral arousal (Svensson et al., 1987, Westerink et al., 1990, Olive et al., 1998). Changes in dopamine transmission could in turn affect the activity of noradrenergic, serotonergic and cholinergic cells, which have been implicated in the regulation of cataplexy (Mignot et al., 1993, Reid et al., 1994a, Reid et al., 1994b, Reid et al., 1994c, Nishino et al., 1995a, Nishino et al., 1995b, Wu et al., 1999, Kalogiannis et al., 2010, Kalogiannis et al., 2011) A D1-like receptor mechanism modulates sleep attacks This study confirms a role for dopaminergic neurotransmission in regulating sleep-wake behavior. We showed that stimulation of D1-like receptors suppressed sleep attacks and promoted wakefulness in orexin KO mice. Conversely, we demonstrated that blockade of excitatory D1-like receptors not only triggered sleepiness by reducing the latency from wakefulness to NREM sleep, it also increased NREM sleep amounts and sleep attacks. These findings suggest that a dopaminergic drive acting on D1-like receptors stimulates wakefulness. Wake-active dopamine cells in the ventral periaquaductal gray could be one source of this excitatory drive because lesioning these dopamine cells increased sleep in rats (Lu et al., 2006a). It is unlikely that dopamine-mediated suppression of sleep attacks and sleepiness acts on the

105 90 orexin system because 1) dopamine inhibits rather than excites wake-promoting orexin neurons and 2) KO mice do not synthesize orexin (Li and van den Pol, 2005, Yamanaka et al., 2006). This is also supported by the fact that modafinil s wake-promoting effects are stronger in orexin KO than wildtype mice (Willie et al., 2005). There is evidence that D2-like drugs modulate sleep-wake behavior (Monti et al., 1988, Monti et al., 1989, Nishino et al., 1991, Ongini et al., 1993). We confirmed that blockade of D2- like receptors can influence sleep-wake amounts. However, we did not observe sleep-wake effects with injection of a D2-like receptor agonist. Previous studies have shown that quinpirole does not impact sleep-wake regulation in a monotonic dose-dependent fashion; instead, only low (0.015mg/kg) or high (1mg/kg or greater) doses modulate sleep-wake behavior (Monti et al., 1988). We used mid-range doses (0.125 and 0.5mg/kg) and therefore did not expect changes in sleep-wake behaviors. Indeed, we used this approach to determine whether D2 drugs could affect cataplexy independent of sleep-wake regulation. We found that D2 drugs can modulate cataplexy with negligible effects on sleep and sleep attacks, while D1 drugs modulate sleep and sleep attacks without affecting cataplexy. This observation illustrates that sleep attacks and cataplexy are controlled by distinct mechanisms Physiological significance Narcolepsy and REM sleep share some physiological and behavioral similarities, the most salient example being the loss of postural muscle tone. This similarity has led to the hypothesis that narcolepsy is a REM sleep disorder and that a faulty REM sleep mechanism underlies cataplexy. However, we show that dopamine drugs can manipulate REM sleep and cataplexy independently. Okura et al (2000) also show that D2 antagonists reduce cataplexy without affecting REM sleep in narcoleptic dogs (Okura et al., 2000). Although REM sleep atonia and cataplexy may be caused by the same mechanism (i.e., at the motor neuron level), our data suggest that REM sleep and cataplexy are triggered by distinct mechanisms (Siegel et al., 1991). Two other pieces of experimental data support this claim. First, Thankachan et al. (2009) demonstrated that putative REM sleep-generating cells only discharge during REM sleep but never during cataplexy in narcoleptic mice; and second, Nishino et al. (2000) showed in narcoleptic dogs that REM sleep follows an ultradian rhythm whereas cataplexy does not (Nishino et al., 2000, Thankachan et al., 2009).

106 91 As stated previously, it is impossible to discern where D2 drugs are acting to modulate cataplexy. However, previous studies provide strong evidence that these drugs act at autoreceptors on dopamine neurons in the VTA, SN and diencephalic dopamine nuclei (Svensson et al., 1987, Meador-Woodruff et al., 1989, Westerink et al., 1990, Reid et al., 1996, Honda et al., 1999a, Okura et al., 2004). By manipulating dopamine release from the VTA and SN, these drugs could modulate cataplexy via projections (both direct and indirect) to downstream arousalrelated neurons (Luppi et al., 1995). For example: noradrenergic neurons cease firing during cataplexy and, unlike dopamine neurons in the VTA and SN, project directly to spinal motor neurons and REM sleep atonia-generating circuits (Luppi et al., 1995, Wu et al., 1999). The A11, a diencephalic dopamine cell group, projects directly to motor neurons (Bjorklund and Skagerberg, 1979, Lindvall et al., 1983, Skagerberg and Lindvall, 1985). This suggests that D2 drugs could modulate cataplexy by manipulating dopamine release directly on motor neurons (Skagerberg et al., 1982, Okura et al., 2004). Indeed our lab has demonstrated a role for dopamine receptors on motor neurons in the maintenance of REM sleep atonia (JJ Fraigne, NA Yee, and JH Peever unpublished data). A11 dopamine neurons also project to the pontine region that generates muscle atonia during REM sleep (termed the SLD in rodents) (Boissard et al., 2002, Lu et al., 2006b, Leger et al., 2010). Activity of A11 neurons suggests that they have a role in inhibiting the SLD during waking (Leger et al., 2010). Therefore the presumptive reduction in dopamine release from the A11 when a D2 receptor agonist is applied could both disfacilitate motor neurons and disinhibit SLD neurons to promote cataplexy.

107 92 Chapter 5: Noradrenergic Regulation of Cataplexy

108 93 Chapter 5: Noradrenergic Regulation of Cataplexy 5.1 Abstract Narcolepsy is characterized by hypersomnolence and cataplexy, the loss of postural muscle tone during waking. The neural mechanisms that trigger the decoupling of state and appropriate muscle tone during cataplexy are unknown although it is hypothesized that cataplexy is the intrusion of REM sleep atonia into wakefulness. Noradrenergic locus coeruleus neurons cease firing during cataplexy. The correlation between cessation of neuron activity and loss of muscle tone has lead to the hypothesis that withdrawal of noradrenergic excitation from motor neurons (i.e. disfacilitation) underlies cataplexy-dependent muscle atonia. Here we establish a role for the noradrenergic system in regulating murine cataplexy and directly test this hypothesis by blocking and facilitating noradrenergic drive on motor neurons during cataplexy. We demonstrate that there is a withdrawal of noradrenergic tone during cataplexy and this withdrawal is necessary but not sufficient to induce complete muscle atonia. These data demonstrate an important role for the noradrenergic system in cataplexy and suggest that it is directly affecting motor neuron excitability and neural circuits upstream of motor neurons, perhaps REM sleep atonia-generating brainstem regions, to cause cataplexy. 5.2 Introduction Cataplexy, a symptom of the sleep disorder narcolepsy, is a motor pathology that is characterized by the rapid, involuntary loss of postural muscle tone that interrupts normal waking behaviors (Siegel and Boehmer, 2006). Cataplexy can last from seconds to minutes and can be either mild and partially impair movement or severe and cause complete postural collapse. Breakdown in the orexin system is linked to narcolepsy in humans, dogs, and mice; loss of orexin cells, inability to produce orexin or dysfunctional orexin receptors can all result in cataplexy (Chemelli et al., 1999, Peyron et al., 2000, Thannickal et al., 2000, Hara et al., 2001, Hungs et al., 2001, Thannickal et al., 2003, Mochizuki et al., 2004, Blouin et al., 2005). While cataplexy is thought to result from the intrusion of REM sleep atonia into wakefulness, the specific neurochemical cue that silences motor neurons during cataplexy is unknown. The noradrenergic system plays a key role in mediating cataplexy. Drugs that affect CNS noradrenergic tone or noradrenergic receptors impact cataplexy in both humans and dogs with

109 94 narcolepsy (Babcock et al., 1976, Mignot et al., 1988b, a, Mignot et al., 1993, Nishino et al., 1993, Moller and Ostergaard, 2009, Ahmed and Thorpy, 2010). High frequency optogenetic manipulation of LC neurons in normal mice triggers behavioral arrests that mimic cataplexy (Carter et al., 2010). Changes in noradrenergic cell activity are also tightly linked to cataplexy in narcoleptic canines; LC neurons cease firing at cataplexy-onset, when muscle tone is lost (Wu et al., 1999). As motor neurons express noradrenergic receptors and noradrenergic neurons both project and provide an endogenous excitatory drive to motor neurons, it has been hypothesized that withdrawal of noradrenergic drive to motor neurons could underlie the loss of muscle tone during cataplexy (Fung and Barnes, 1987, Grzanna et al., 1987, Lai et al., 1989, Shao and Sutin, 1991, Kwiat and Basbaum, 1992, Larkman and Kelly, 1992, Fung et al., 1994, Wu et al., 1999, Fenik et al., 2005b, Chan et al., 2006). Here we use a mouse model of narcolepsy (Chemelli et al., 1999, Mochizuki et al., 2004) to test directly whether the loss of noradrenergic input to motor neurons is responsible for cataplexy. We first confirm a role for the noradrenergic system in murine cataplexy using systemic pharmacology. By blocking and activating α 1 adrenergic receptors directly on motor neurons we show that loss of noradrenergic drive is not sufficient to induce cataplexy-dependent muscle atonia. These data refute the hypothesis that withdrawal of noradrenergic drive from motor neurons causes the loss of muscle tone during cataplexy; other sites where withdrawal of noradrenaline could exert cataplexy-inducing affects are also discussed. 5.3 Methods All procedures and experimental protocols were approved by the University of Toronto s animal care committee and were in accordance with the Canadian Council on Animal Care Animals Mice were housed individually and maintained on a 12:12 light dark cycle with both food and water available ad libitum. These experiments used 19 orexin KO male mice (age: 14.2 ± 0.8 weeks; mass: 28.2 ± 0.8g) on a C57BL/6 background. Mice were genotyped using PCR with genomic primers 5'-GACGACGGCCTCAGACTTCTTGGG, 3'- TCACCCCCTTGGGATAGCCCTTCC, and 5 -CCGCTATCAGGACATAGCGTTGGC (with forward primers being specific for either wildtype or KO mice and the reverse primer being common to both).

110 Surgical preparation To implant EEG electrodes, EMG electrodes and a microdialysis guide cannula, sterile surgery was performed under isoflurane anesthesia (1-2%). Effective depth of anesthesia was determined by the abolition of the pedal withdrawal reflex. Body temperature was monitored with a probe (TC-1000, CWE Inc., Ardmor, PA) and maintained at 37 ± 1ºC. For microdialysis studies, mice were placed in a stereotaxic apparatus (Model 962, Kopf, Los Angeles, CA). The skin was retracted to expose the skull surface. The skull was positioned so that bregma and lamba were in the same horizontal plane. To implant a microdialysis probe into the left trigeminal motor nucleus, a stereotaxic drill (Model 1471, Kopf) was used to drill a ~1mm hole 5.10 mm caudal and 1.37 mm lateral to bregma. A microdialysis guide cannula (MD2255, BASi, West Lafayette, IN) was slowly lowered 4.0 mm below the skull surface by stereotaxic manipulation and secured in place with dental cement (Ketac-cem, 3M, London, ON). EEG recordings were obtained using two stainless steel micro-screws (1mm anterior and 1.5mm lateral to bregma; 3mm posterior and 1.5mm lateral to bregma). EMG electrodes consisted of multistranded stainless steel (AS131, Cooner Wire, Chatsworth, CA) wires that were sutured onto neck and masseter muscles. All electrodes were attached to a micro-strip connector (CLP L-D, Electrosonic, Toronto, ON), which was affixed onto the animal s head with dental cement (Ketac-cem, 3M). Following surgery, mice were given 0.9% saline and ketoprofen (3mg/kg; s.c.). Mice were individually housed in a sound-attenuated and ventilated chamber for 9-12 days post surgery Experimental procedures for sleep and microdialysis studies During experiments, animals were housed in a movement-responsive caging system that eliminates the need for a commutator or liquid swivel (25cm height, 20cm diameter; Raturn; BASi). This caging system was housed inside a sound-attenuated, ventilated, and illuminated (lights on: 110 lux) chamber. Sleep-wake state and muscle activity were recorded by attaching a lightweight cable to the microstrip connector on the mouse s head, which was connected to a Physiodata Amplifier system (Grass 15LT, Astro Med, Brossard, QC). The EEG was amplified 1000 times and bandpass filtered between 0.3 and 100 Hz. EMG signals were amplified 1000 times and band-pass

111 96 filtered between 30 Hz and 100 Hz. All electrophysiological signals were digitized at 1000Hz (Spike 2 Software, 1401 Interface, CED Inc.), monitored and stored on a computer. Infra-red video recordings were also captured and synchronized to the electrophysiological recordings. A microdialysis probe was used to exogenously perfuse α 1 noradrenergic receptor modulators directly onto motor neurons. Under isoflurane anesthesia, the microdialysis stylet was removed from the guide cannula, and a microdialysis probe (MD-2211, BASi) was lowered into the left trigeminal motor nucleus. The microdialysis probe was connected to FEP teflon tubing (inside diameter=0.12mm; Eicom, Japan), which was connected to a 1mL gastight syringe (MDN-0100, BASi) via a zero dead-space liquid switch (UniSwitch Liquid Switch Syringe Selector, BASi). The probe was continually perfused with filtered (0.2µm Nylon, Fisher Scientific, Ottawa, ON) artificial cerebral spinal fluid (acsf: 125mM NaCl, 5mM KCl, 24mM NaHCO 3, 2.5mM CaCl 2, 1.25mM MgSO 2 ) at a flow rate of 1ul/min with a syringe pump and controller (MD-1001 and MD-1020, BAS). Phenylephrine (Sigma Aldrich, Oakville, ON), an α 1 receptor agonist, was prepared fresh in acsf immediately before each experiment. Terazosin (Sigma Aldrich), an α 1 noradrenergic receptor antagonist was prepared from a stock solution. For injection studies, drugs were diluted with saline to the desired concentration (phenylephrine: 5 and 10mg/kg; terazosin: 5 and 10mg/kg). For microdialysis studies, drugs were diluted with acsf to the desired concentration (phenylephrine: 0.2mM, 1mM, 5mM; terazosin: 1mM) and filtered immediately before use. We used these concentrations of phenylephrine and terazosin because previous data showed that they can successfully affect motor neuron excitability during natural motor behaviors in rats (Chan et al., 2006) (Mir and Peever, unpublished data).

112 97 Figure 5.1: Microdialysis probe insertion into the trigeminal motor nucleus Probe insertion into the trigeminal motor nucleus allowed for focal perfusion of pharmacological agents onto trigeminal motor neurons. Some of these motor neurons innervate the masseter muscles. We then measured masseter EMG activity as our index of motor neuron excitability. EEG screws were implanted in the skull to record neural activity. A headplug was affixed onto the mouse s head to allow recording of EEG and EMG.

113 Experimental paradigm Mice were placed in a round plexi-glass cage and given 24 hours to habituate to this new environment. After this period, mice were connected to the recording tether and given at least another 48 hours to habituate to this condition before the injection procedures. A saline injection was given on the first night after this period to habituate to systemic injections. All injections (0.3mL i.p.) were given at the onset of the dark phase (i.e., 19:00) and drug doses were randomized. Each mouse only received one injection per day and no more than five treatments in total. As terazosin has a longer half life than phenylephrine, data were collected and analyzed for 3 hours after phenylephrine administration and 6 hours after terazosin administration (Hengstmann and Goronzy, 1982, Sonders, 1986). To investigate noradrenaline s role in regulating muscle tone during cataplexy, we microdialyzed an α 1 receptor agonist or antagonist into the left trigeminal motor pool. Before this, mice were placed into the sleep recording chamber and tethered for EEG and EMG recordings. Mice were given 48 hours to habituate to the recording environment before the microdialysis probe was inserted. Under isoflurane anesthesia, the microdialysis probe was inserted between 11:00-12:00 and acsf was perfused. Perfusion of acsf (for baseline recordings) or candidate drugs (phenylephrine or terazosin) began at 19:00. Each drug was applied continuously for 2-4 hours and an acsf washout period followed every drug period. Drugs were perfused during the dark phase (19:00-7:00) to maximize the amount of cataplexy recorded Verification of probe location We used three criteria to demonstrate that microdialysis probes were both functional and located in the motor nucleus. First, we demonstrated that microdialysis probe insertion into the motor nucleus induced a robust increase in only left masseter muscle EMG activity, without affecting the EMG activity of the right masseter muscle. Second, at the end of experiments, 0.01mM α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA, a non-nmda receptor agonist; Tocris, Ellisville, MO) was perfused into the motor nucleus, which induced a rapid and potent increase in basal levels of left masseter muscle tone without affecting right masseter EMG activity. This result verified that: 1) motor neurons were viable and able to

114 99 respond to excitation; 2) microdialysis probes were functional at the end of each experiment; and, 3) probes were located within the motor nucleus. Third, we used post-mortem histological analysis to demonstrate that microdialysis probes were physically located within the motor nucleus. At the end of each experiment, mice were anesthetized via isoflurane and sacrificed. The brain was removed and placed in 4% paraformaldehyde (in 0.1M PBS) for 24 hours and then 30% sucrose (in 0.1M PBS) for 48 hours. The brain was then frozen and sectioned in 40µm slices using a microtome (SM2000R, Leica, Depew, NY). Brain sections were then mounted on slides and dried before being stained with Neutral Red. Tissue sections were viewed using a microscope (BX50Wi, Olympus, Center Valley, PA) and photographed (Q-color 3, Olympus). The location of microdialysis probe lesion tracts were then plotted on a standardized stereotaxic map of the mouse brainstem (Paxinos and Franklin, 2001) Data analysis We used both EEG and EMG signals (right masseter and neck muscles) as well as video to identify five distinct behavioral states: active wake, quiet wake, NREM sleep, REM sleep, and cataplexy. Active wake was characterized by high-frequency, low-voltage EEG signals coupled with high levels of EMG activity. Quiet wake was characterized by high frequency, low voltage EEG signals and the absence of overt motor activity. NREM sleep was characterized by high amplitude, low frequency EEG signals and minimal EMG activity. REM sleep was characterized by low amplitude, high frequency theta EEG activity and very low EMG levels (i.e., REM sleep atonia) interspersed by periodic muscle twitches. Cataplexy was classified as a sudden loss of muscle tone, in neck and right masseter muscles, following at least 40 seconds of active waking and with a duration of at least 10 seconds. Sleep states were visually identified in 5 second epochs and scored in Spike 2 (CED) with the Sleepscore v1.01 script. Raw EMG signals were full-wave rectified and quantified in arbitrary units (A.U.). Average EMG activity for left and right masseter and neck muscle activity was quantified in 5 second epochs for each behavioral state. When noradrenergic agents were applied onto the left trigeminal motor nucleus, EMG data were not analyzed for the first 15 minutes of perfusion because the flow latency from the syringe to the microdialysis probe was 8-10 minutes. Each mouse served as its own control.

115 100 Spectral analysis was performed using EEG Band Detect v1.06 in Spike 2. The EEG was windowed using a Hamming function and subjected to a fast Fourier transform to yield the power spectrum. The power within four frequency bands was recorded as absolute power and as a percentage of the total power of the signal that was calculated over each 5-sec epoch. The band limits used were (delta) Hz, (theta) Hz, Hz and Hz Statistical analysis The statistical tests used for analysis are included in the text of the results section. All statistical analyses were computed using SigmaStat (SPSS Inc., Chicago, IL) and applied a critical α value of p<0.05. Data are presented as mean ± SEM. 5.4 Results Focal activation of α 1 receptors on trigeminal motor neurons increased masseter EMG activity in freely behaving mice To determine whether reverse-microdialysis of specific drugs is a useful technique for investigating motor neuron excitability in freely behaving mice, as it has proven to be in rats, we administered phenylephrine (PE) at concentrations of 0.2mM, 1.0mM and 5.0mM to wildtype mice. PE increased muscle tone during quiet waking, NREM sleep and REM sleep. There was no effect on muscle tone during active waking with any dose of the drug, likely because endogenous noradrenergic drive is high during this state. During quiet waking, PE increased left masseter muscle tone (p=0.011, one-way RM ANOVA; n=9; Figure 5.2A) by 4.4 ± 5.5% (p=0.789), 9.1 ± 4.3 (p=0.507), and 23.6 ± 5.8% (p=0.008) compared to baseline during perfusion of 0.2mM, 1.0mM and 5.0mM respectively; however, this increase was only significant under the 5mM treatment. During NREM sleep, PE increased left masseter muscle tone (p<0.001, one-way RM ANOVA; n=9) by 9.6 ± 3.1% (p=0.102), 12.6 ± 3.8% (p=0.054), and 25.7 ± 6.0% (p<0.001) compared to baseline with perfusion of 0.2mM, 1.0mM and 5.0mM, respectively. During REM sleep, PE increased left masseter muscle tone (P<0.001, one-way RM ANOVA; n=9) by 7.3 ± 2.2% (p=0.157), 12.0 ± 2.6% (p=0.006), and 26.1 ± 2.6% (p<0.001) compared to baseline with perfusion of 0.2mM, 1.0mM and 5.0mM respectively. We did not see a significant increase in right masseter or neck

116 101 muscle tone during any behavioral state with application of PE at any concentration (Figure 5.2B and C). Activation of α 1 receptors on trigeminal motor neurons with PE increased masseter muscle tone in freely behaving mice during quiet waking, NREM sleep and REM sleep. This demonstrates that we can focally manipulate motor neuron excitability in freely behaving mice, making this a useful model to study the effects of neurotransmitters on motor neurons during specific motor behaviors, such as cataplexy.

117 102 Figure 5.2: Phenylephrine application increased muscle tone in freely behaving mice A. Phenylephrine dose-dependently increased left masseter muscle tone during sleep and quiet waking. B. Right masseter muscle tone was not affected by this intervention. C. Raw traces demonstrating EEG and EMG (left and right masseter) activity with phenylephrine application. * denotes p<0.05 when compared to saline.

118 Masseter muscles experienced atonia during cataplexy Narcoleptic mice (n=8) had episodes of cataplexy throughout the recording period. Cataplexy was characterized by abrupt behavioral arrest and postural collapse. During each episode, EMG tone decreased while EEG activity remained wake-like. Immediately after cataplexy, muscle tone returned to waking levels and mice resumed normal behaviors (Figure 5.3A). During a baseline 6-hour recording period, narcoleptic mice had an average of 3 ± 1 cataplectic episodes that lasted 66 ± 15s. It is hypothesized that cataplexy results from the intrusion of REM sleep atonia into waking. However, it is unknown if postural muscles experience atonia or if they simply have reduced muscle tone during cataplexy. Therefore, we quantified levels of masseter and neck muscle tone during cataplexy and REM sleep muscle atonia. We found that EMG tone rapidly decreased at cataplexy onset and returned at cataplexy offset (n=8; RM ANOVA, p<0.01 for both neck and masseter muscles; Figure 5.3B). Levels of both masseter and neck muscle tone remained at levels comparable to REM sleep atonia during cataplexy (REM vs. cataplexy; paired t-test, masseter: p=0.663, neck: p=0.202; Figure 5.3C and D). These results illustrate that muscles experience atonia during cataplexy and that the trigeminal motor system is a good index for determining how motor neurons are controlled during cataplexy.

119 104 Figure 5.3: Muscles experienced atonia during cataplexy A. Example EEG and EMG (neck and masseter) traces showing loss of skeletal muscle tone during an episode of cataplexy in a orexin knockout mouse. B. Group data showing that neck and masseter muscle tone are lost during cataplectic attacks. EMG tone was maximal during waking periods before and after each attack, but minimal during cataplexy. C. EEG and EMG traces demonstrating that muscle tone reached similar levels during REM sleep and cataplexy. D. Group data (n=8) showing that levels of muscle tone were comparable during REM sleep and cataplexy. This observation demonstrates that masseter and neck muscles experience muscle atonia during cataplexy. * indicates p<0.001; A.U., arbitrary units; values are plotted as means + SEM.

120 Cataplexy is affected by changes in noradrenergic activity Changes in noradrenergic tone influence cataplexy in narcoleptic humans and dogs (Babcock et al., 1976, Mignot et al., 1988b, a, Mignot et al., 1993, Zaharna et al., 2010); however, it is unknown if noradrenergic receptor manipulation also modulates cataplexy in narcoleptic mice. To determine if a noradrenergic mechanism influences cataplexy, we injected different doses of either an α 1 receptor agonist (5 and 10mg/kg phenylephrine) or antagonist (5 and 10mg/kg terazosin). We found that stimulation of α 1 receptors (a single 10mg/kg dose of phenylephrine) reduced amounts of cataplexy by 90% (n=6; RM ANOVA, p=0.04; Figure 5.4A). This decrease was caused by a 92% reduction in the number of cataplectic episodes (p=0.01, Figure 5.4B); the duration of cataplexy episodes was unaffected by this intervention (p=0.33; data not shown). Although 5mg/kg of phenylephrine reduced overall cataplexy amounts by 73% and reduced the number of episodes by 67%, these reductions were not statistically significant (n=6; RM ANOVA, p>0.05 for both variables; Figure 5.4A and B). In contrast, we found that α 1 receptor antagonism increased cataplexy. A single injection of terazosin (10mg/kg) robustly increased the number of cataplectic episodes by 92% (n=6; RM ANOVA, p=0.04; Figure 5.4D) despite having no significant effect on the total amount of time spent in cataplexy (p=0.208; Figure 5.4C) or episode duration (p=0.13; data not shown). A 5mg/kg dose of terazosin had no effect on either the number or duration of cataplectic episodes (n=6; RM ANOVA, p>0.05 for both variables; Figure 5.4D). These data indicate that systemic changes in noradrenergic receptor activation affect cataplexy in mice.

121 106 Figure 5.4: Cataplexy is sensitive to changes in noradrenergic tone A and B. Systemic administration of a α 1 receptor agonist (5 and 10 mg/kg phenylephrine) decreased cataplexy by reducing the number of cataplexy attacks. C and D. Systemic administration of a α 1 antagonist (5 and 10mg/kg terazosin) increased the number of cataplexy attacks. * indicates p<0.05 ; values are plotted as means + SEM.

122 Drug manipulations targeted trigeminal motor neurons Postmortem histology showed that all microdialysis probes (1 probe per mouse) terminated in the left trigeminal motor pool (Figure 5.5A and B). We also showed that probe insertion caused an immediate, but transient activation of only left masseter muscle tone (n=11; RM ANOVA, p=0.003; Figure 5.5C). Right masseter muscle activity was not affected by this intervention (RM ANOVA, p=0.144), indicating that probe insertion selectively influences trigeminal motor neurons in the targeted motor pool. In addition, we perfused 0.01mM AMPA at the end of experiments. AMPA application caused marked motor neuron activation that increased left masseter EMG tone (n=7; paired t-test, p=0.03) without affecting right masseter muscle activity (paired t-test, p=0.271; Figure 5.5D). We also wanted to determine if brain regions surrounding the trigeminal motor pool were affected by drug interventions. The sublaterodorsal nucleus (SLD) is located immediately adjacent (~0.2mm dorsomedial) to the trigeminal motor pool and it plays a role in regulating REM sleep (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b). Increased noradrenergic transmission at the SLD region has been shown to suppress REM sleep (Crochet and Sakai, 1999b). Therefore we hypothesized that drug application would not affect sleep and cataplexy if SLD function was unaffected. Sleep-wake architecture was unaffected by manipulation of α 1 receptors at the trigeminal motor pool. Phenylephrine perfusion (1mM) did not influence amounts of REM sleep or NREM in narcoleptic mice (acsf vs. phenylephrine: 2-way RM ANOVA, p=0.124; Figure 5.6A). However, this intervention did reduce EEG theta power during REM sleep in these mice (2-way RM ANOVA, p=0.023; Figure 5.6B). Antagonism of α 1 receptors by terazosin (1mM) perfusion had no effect on either sleep-wake amounts (n=4; acsf vs. terazosin: 2-way RM ANOVA, p=0.069) or EEG spectral power during REM sleep (2-way RM ANOVA, p=0.611; Figure 5.6C and D). We suggest that local REM-generating circuits are largely unaffected by noradrenergic receptor manipulation at the trigeminal motor nucleus. Because the SLD region is hypothesized to control the loss of muscle tone during cataplexy, we also wanted to verify that drug manipulations did not affect amounts of cataplexy. Phenylephrine perfusion had no effect on the total amount of time spent in cataplexy (paired t- test, p=0.991; Figure 5.7A), nor did it affect the number (p=0.800) or duration (p=0.280) of attacks (data not shown). This intervention also had no affect on EEG spectral power during cataplexy (2-way RM ANOVA, p=0.939; Figure 5.7B). Similarly, terazosin application had no

123 108 effect on the total amount of cataplexy (paired t-test, p=0.471; Figure 5.7C) or the number (p=0.270) or duration (p=0.440) of attacks (data not shown). It also did not influence EEG spectral power during individual cataplectic attacks (2-way RM ANOVA, p=0.369; Figure 5.7D). Together, these findings indicate that drug manipulations at the trigeminal motor nucleus had negligible impact on the cell systems triggering cataplexy.

124 109 Figure 5.5: Drug manipulations targeted trigeminal motor neurons A. A histological example demonstrating the location of a microdialysis probe tract in the trigeminal motor pool. B. Microdialysis probe locations in the left trigeminal motor pool in the 11 mice used in these studies. C. EMG traces from left (LM) and right masseter (RM) muscles showing dialysis probe insertion into the left trigeminal nucleus caused a brief, but transient increase in left masseter EMG tone. Group data showing that probe placement at the trigeminal motor pool increased left masseter muscle tone. D. Left and right masseter EMG traces showing that AMPA perfusion into the left motor pool potently increased left (but not right) masseter EMG tone. Group data showing that this drug intervention significantly increases only left masseter tone. Right masseter tone is unaffected by this manipulation, which indicates that drug intervention preferentially targeted motor neurons in the left trigeminal motor pool. * indicates p<0.05; A.U., arbitrary units; values are plotted as means + SEM.

125 110 Figure 5.6: Targeted drug manipulations did not affect sleep-wake architecture A. Stimulation of adrenergic receptors (by phenylephrine perfusion) at the trigeminal motor pool did not affect amounts of REM sleep, suggesting that adjacent REM generating circuits are unaffected by this intervention. B. Group data showing the EEG theta power (plotted as % total power) decreased during REM sleep during phenylephrine application. C. Terazosin perfusion did not affect amounts of wakefulness, NREM or REM sleep. D. Group data showing that terazosin perfusion at the trigeminal pool did not influence EEG power spectra during REM sleep. * indicates p<0.05; values are plotted as means + SEM.

126 111 Figure 5.7: Targeted drug manipulations did not influence cataplexy A. Compared to baseline, activation of α 1 receptors (by phenylephrine perfusion) at the trigeminal motor pool had no effect on cataplexy amounts. B. This intervention also had no effect on EEG spectral power (plotted as % total power) during cataplectic attacks. C. Antagonism of α 1 receptors by terazosin perfusion at the trigeminal motor pool also had no effect on amounts of cataplexy. D. This drug manipulation also had no affect on EEG spectral power (plotted as % total power) during cataplectic attacks. These results indicate that noradrenergic receptor manipulation at the trigeminal nucleus has no measureable influence on the neural circuits that trigger cataplectic episodes. * indicates p<0.05; values are plotted as means + SEM.

127 Loss of noradrenergic drive to motor neurons is not sufficient for triggering cataplexy Since LC neurons stop firing during cataplexy in narcoleptic dogs, it has been hypothesized that loss of noradrenergic drive to motor neurons underlies cataplexy. However, it is unknown if changes in noradrenergic excitation of motor neurons actually contributes to muscle atonia during cataplexy. Therefore, we aimed to determine: 1) if there is an endogenous noradrenergic drive at the trigeminal motor pool during normal waking and 2) if loss of this excitatory drive triggers cataplexy in narcoleptic mice. First, we identified the presence of an endogenous noradrenergic drive onto trigeminal motor neurons during normal periods of wakefulness. We found that antagonism of α 1 - adrenergic receptors by terazosin perfusion (1mM) at the left trigeminal motor pool markedly suppressed left masseter muscle tone during waking (n=4; p=0.03). Specifically, we showed that receptor antagonism reduced waking masseter tone by 33% in the 30s period before cataplexy and by 31% in the 30s period after cataplexy (paired t-test, before cataplexy: p=0.004; after cataplexy: p=0.053; Figure 5.8A). However, this intervention had no effect on right masseter activity (Figure 5.8B). Second, we showed that noradrenergic drive onto trigeminal motor neurons was lost during cataplexy episodes. Although α 1 receptor antagonism significantly reduced left masseter tone during periods of waking immediately preceding and following cataplectic attacks, this same intervention had no effect on masseter muscle tone during cataplexy (p=0.210; Figure 5.8A). This observation suggests that noradrenergic excitation of motor neurons is negligible during cataplexy. Finally, we showed that loss of noradrenergic drive is not entirely responsible for triggering muscle atonia during cataplexy. Although we found that α 1 receptor blockade at the trigeminal motor pool reduced masseter muscle tone during normal waking, it did not reduce tone to the same levels that occurred during cataplexy (p=0.020; Figure 5.8C). Together, these findings indicate that an endogenous noradrenergic drive maintains waking levels of masseter tone; however, loss of this drive is not the only mechanism responsible for muscle atonia during cataplexy.

128 113 Figure 5.8: Loss of noradrenergic drive contributes to muscle atonia during cataplexy A. Waking levels of left masseter muscle tone are significantly reduced by α 1 receptor antagonism (terazosin) at the left trigeminal motor pool (n=4). However, this same intervention had no effect on masseter muscle tone during cataplexy, demonstrating that the waking noradrenergic drive is withdrawn during cataplexy. B. α 1 receptor manipulation at the left trigeminal motor pool affected left masseter muscle tone, but has no impact on right masseter muscle tone. C. Antagonism of α 1 receptors on trigeminal motor neurons (by terazosin perfusion) reduced waking masseter muscle tone, but it did not lower it to cataplectic levels, indicating that loss of noradrenergic drive is not the only mechanism triggering muscle atonia during cataplexy. * indicates p<0.05; A.U., arbitrary units; values are plotted as means + SEM.

129 Restoration of noradrenergic activity increased muscle tone during cataplexy Our final goal was to determine if muscle atonia during cataplexy could be prevented by restoring noradrenergic drive onto motor neurons. However, we first wanted to determine if activating α 1 receptors on motor neurons could prevent the natural loss of masseter tone during sleep in orexin KO mice. Phenylephrine perfusion increased masseter tone by 30 ± 5% (n=9; paired t-test, p=0.001) during NREM sleep and by 35 ± 9% during REM sleep in narcoleptic mice (p=0.008; Figure 5.9A and B). These results demonstrate that loss of muscle tone during sleep can be rescued by restoring noradrenergic drive to α 1 receptors on motor neurons. To determine if loss of noradrenergic drive underlies muscle atonia during cataplexy, we pharmacologically restored this excitatory drive to motor neurons during cataplexy. We found that activating α 1 receptors at the left trigeminal motor pool increased left masseter tone and prevented complete muscle atonia during cataplexy (n=9; acsf vs. phenylephrine: paired t-test, p=0.001; Figure 5.10A and B). Specifically, we found that receptor activation increased masseter tone by 93 ± 33% above baseline cataplexy levels (p=0.001; Figure 5.10). In fact, phenylephrine perfusion increased left masseter muscle tone to levels not significantly different from normal waking (cataplexy baseline vs. cataplexy phenylephrine; 1-way ANOVA on Ranks, p>0.05; Figure 5.10C), However, this targeted intervention had no effect on either right masseter (Figure 5.10B) or neck muscle tone (data not shown). Muscle atonia in both muscle groups remained completely intact during cataplexy, illustrating that drug manipulation only influenced the trigeminal motor nucleus, and not the circuitry regulating muscle atonia. Together, these results show that restoration of noradrenergic drive to motor neurons significantly elevates muscle tone during cataplexy.

130 115 Figure 5.9: Stimulating α 1 receptors on motor neurons increased muscle tone during sleep in orexin KO mice A. EEG and EMG traces showing that stimulating α 1 receptors at the trigeminal motor pool increased masseter tone during both NREM (top trace) and REM (bottom trace) sleep in narcoleptic mice. B. Group data (n=9) showing that phenylephrine perfusion is capable of increasing masseter muscle tone during NREM and REM sleep in narcoleptic mice. * indicates p<0.05; A.U., arbitrary units; values are plotted as means + SEM.

131 116 Figure 5.10: Activation of α 1 receptors on motor neurons elevated muscle tone during cataplexy A. EEG and EMG traces showing that muscle tone is increased by restoring noradrenergic activity at the left trigeminal motor pool. Compared to control (i.e., acsf), phenylephrine perfusion onto motor neurons in the left trigeminal nucleus prevented atonia in the left masseter muscle during cataplexy, but atonia persists in right masseter and neck muscles. B. Group data (n=9) showing that left masseter tone is significantly increased during cataplexy when phenylephrine is perfused at the left trigeminal motor pool. C. Group data showing that phenylephrine perfusion restored masseter muscle tone to near waking levels during cataplexy. * indicates p<0.05; A.U., arbitrary units; values are plotted as means + SEM.

132 Discussion The neural mechanisms and transmitter systems that underlie cataplexy remain unclear. Withdrawal of activity of the noradrenergic system has been correlated with, but has not been demonstrated to trigger, the loss of muscle tone during cataplexy (Wu et al., 1999). This study is important because it establishes a direct link between the noradrenergic system and the loss of muscle tone in cataplexy. We identified an endogenous noradrenergic drive during waking that is withdrawn during cataplexy. We then demonstrated that muscle tone during cataplexy could be increased by applying an exogenous excitatory noradrenergic drive to motor neurons. This study is the first to directly test the hypothesis that withdrawal of noradrenergic drive to motor neurons underlies the loss of muscle tone during cataplexy and establishes a functional role for the noradrenergic system in the regulation of cataplexy in orexin KO mice The noradrenergic system regulates cataplexy Using specific α 1 receptor activation and blockade we established that the noradrenergic system modulates cataplexy in narcoleptic mice. Systemic activation of α 1 receptors significantly reduced the severity of cataplexy. This finding is in agreement with previous studies demonstrating that tricyclic antidepressants, which act in part via a noradrenergic mechanism, are one of the most effective treatments for cataplexy in narcoleptic humans, dogs and mice (Foutz et al., 1981, Mignot et al., 1993, Willie et al., 2003, Moller and Ostergaard, 2009, Ristanovic et al., 2009, Zaharna et al., 2010). Systemically blocking α 1 receptors in orexin KO mice increased the occurrence of cataplexy. This finding suggests blockade or loss of noradrenergic drive promotes cataplexy. Other studies support this concept, including the finding that rare cases of pontine lesions in regions that include noradrenergic cells resulted in narcolepsy with cataplexy (D'Cruz et al., 1994, Mathis et al., 2007). One particular noradrenergic area of interest is the LC because its neurons cease firing during cataplexy in narcoleptic dogs (Wu et al., 1999). In addition, highfrequency optogenetic stimulation of the LC, at frequencies thought to induce depolarization block or depletion of noradrenaline, induces behavioral collapse similar to cataplexy (Carter et al., 2010). Our data, along with these studies, indicate that loss of the noradrenergic system or normal noradrenergic activity can lead to cataplexy or cataplexy-like states.

133 118 Recently, gene-replacement inducing orexin expression in the zona incerta was shown to reduce narcolepsy symptoms in orexin-deficient mice (Liu et al., 2011a). The zona incerta densely innervates the LC (Liu et al., 2011a). Orexinergic excitation of noradrenergic LC neurons could be responsible for reducing the severity of cataplexy in these mice, suggesting that increasing excitatory noradrenergic drive can reduce cataplexy. We show that restoring the drive directly at the level of motor neurons can alleviate cataplexy in a single muscle group (i.e. restore muscle tone during cataplexy); however this does not indicate that withdrawal of noradrenergic drive from motor neurons is the primary cause of cataplexy-dependent muscle atonia Withdrawal of noradrenergic drive promotes cataplexy The most direct evidence of a role for the noradrenergic system in cataplexy is the finding that LC neurons cease firing during cataplexy in dogs (Wu et al., 1999). Parts of the noradrenergic system project to both cranial and spinal motor pools (Grzanna et al., 1987, Bruinstroop et al., 2011). Noradrenaline has been demonstrated to have an α 1 receptor mediated excitatory effect on motor neurons that ultimately increases muscle tone (Fung and Barnes, 1987, Lai et al., 1989, Fung et al., 1991, Fenik et al., 2005b, Chan et al., 2006). The confluence of these findings has led to the hypothesis that withdrawal of an excitatory noradrenergic drive to motor neurons underlies the loss of muscle tone in cataplexy (Wu et al., 1999, Siegel and Boehmer, 2006). Our data confirm that a noradrenergic drive is not present during cataplexy and are the first to demonstrate that the withdrawal of this drive accounts for some loss of muscle tone. We observed an endogenous α 1 receptor mediated noradrenergic drive during waking periods preceding and following cataplexy. Blocking this drive on motor neurons during waking resulted in a significant reduction in muscle tone, but did not induce full muscle atonia seen during cataplexy. Therefore, the withdrawal of the endogenous noradrenergic drive from motor neurons is only partially responsible for the loss of muscle tone during cataplexy. It has been demonstrated that the LC and subcoeruleus noradrenergic neurons do not project to motor neurons in great numbers. Recent data using genetically targeted tracing demonstrated few projections from the LC directly to the spinal cord ventral horn in rats (Bruinstroop et al., 2011). Importantly for our study, it was previously demonstrated that the LC and subcoeruleus have few projections to the trigeminal motor nucleus in rats (Grzanna et al.,

134 ). These data suggest that cessation of LC neuron activity would not trigger muscle atonia, at least in rodents. In contrast, the A7 noradrenergic neurons send dense projections to both the spinal cord ventral horn and trigeminal motor nucleus (Grzanna et al., 1987, Bruinstroop et al., 2011). In addition, the A5 noradrenergic cells have been demonstrated to project to the trigeminal motor nucleus (Grzanna et al., 1987). The activity of A5 and A7 neurons during cataplexy is not known but cessation of these neurons may underlie the partial, noradrenergic withdrawal-dependent muscle tone suppression we observed. Further investigation into the activity of the different noradrenergic cell groups during sleep and cataplexy could help elucidate the source of the waking noradrenergic drive to motor neurons Noradrenaline could act at REM sleep generating sites to modulate cataplexy The noradrenergic system could regulate cataplexy through direct projections to brainstem areas that control REM sleep muscle atonia (Leger et al., 2009). Cataplexy and REM sleep share many characteristic features, including the loss of muscle tone. This has led to the hypothesis that cataplexy may be the intrusion of REM sleep atonia into wakefulness (Rechtschaffen and Dement, 1967). Indeed, brainstem neurons that mediate the suppression of muscle tone fire selectively during REM sleep and cataplexy in narcoleptic dogs (Siegel et al., 1991). The neural circuitry that controls muscle atonia during REM sleep in rodents has been recently elucidated (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b). There are REM sleep-suppressing regions in the pons that actively inhibit downstream REM sleeppromoting systems that ultimately trigger muscle atonia (Boissard et al., 2002, Boissard et al., 2003, Lu et al., 2006b). We propose that noradrenaline could act at excitatory α 1 receptors on REM sleep-suppressing regions and/or at inhibitory α 2 receptors on REM sleep-promoting systems to suppress muscle atonia during waking (Herbert and Saper, 1992, Crochet and Sakai, 1999a). In support of this, noradrenaline suppresses REM sleep when injected into REM sleeppromoting regions in the cat (Crochet and Sakai, 1999a). Noradrenergic neurons in the brainstem express Fos in a pattern consistent with the hypothesis that they act to inhibit REM sleep-generating mechanisms; A1, A2, A5 and A7 neurons are Fos positive during REM sleep

135 120 deprivation, while only 2% of LC cells were active under the same conditions (Leger et al., 2009). It is clear that LC neurons are inhibited during both REM sleep and cataplexy (Wu et al., 1999, Takahashi et al., 2010), but unclear if these neurons play any role in the loss of muscle tone during these states. LC neurons are likely actively inhibited during REM sleep by GABAergic projections from the dorsal paragigantocellular reticular nucleus (Nitz and Siegel, 1997, Verret et al., 2006), but could be inhibited by a different mechanism during cataplexy. A recent study in humans demonstrated increased sympathetic nerve activity and blood pressure at cataplexy-onset (Donadio et al., 2008). Increases in blood pressure reduce LC firing (Elam et al., 1984, Murase et al., 1994), perhaps explaining the reduction of LC activity during cataplexy. However, narcoleptic canines do not show elevated blood pressure during cataplexy (Siegel et al., 1986) so further investigation of why LC neurons cease firing and whether this has a functional role in cataplexy is required Other circuits involved in cataplexy Numerous transmitter systems have been implicated in the regulation of cataplexy. Many effective treatments for narcolepsy affect serotonin and dopamine systems in addition to noradrenaline (Rothman et al., 2001, Leonard et al., 2004, Alexander et al., 2005, Wisor and Eriksson, 2005, Moller and Ostergaard, 2009, Delucchi et al., 2010). Serotonin has been demonstrated to have a negligible role, as systemic receptor modulators have little effect and unit recording studies do not show remarkably different firing patterns during cataplexy (Nishino et al., 1995a, Wu et al., 2004). Conversely, we have demonstrated that systemic modulation of the dopaminergic system can affect cataplexy in orexin KO mice (Burgess et al., 2010); these results are supported by previous findings in dogs, where both systemic and focal manipulation of dopaminergic drive affects cataplexy most likely via a D2 auto-receptor mechanism (Nishino et al., 1991, Reid et al., 1996, Honda et al., 1999b, Okura et al., 2000, Okura et al., 2004). Acetylcholine, which has a well established role in the control of REM sleep, can modulate cataplexy: work in narcoleptic dogs showed that increasing cholinergic drive systemically and focally into pontine regions regulating REM sleep atonia can increase cataplexy (Reid et al., 1994a, Reid et al., 1994b, Reid et al., 1994c); these findings have been recently reproduced and built upon in narcoleptic mice (Kalogiannis et al., 2010, Kalogiannis et al., 2011). While our

136 121 data establish a role for noradrenergic drive in modulating cataplexy both systemically and at the level of motor neurons, there are other transmitter systems that contribute to this phenomenon Conclusions These data establish a functional role for the noradrenergic system in the regulation of murine cataplexy. We demonstrated that withdrawal of noradrenergic excitation from motor neurons contributes to, but is not solely responsible for the loss of muscle tone during cataplexy. Evidence suggests that the loss of noradrenergic drive may also directly impact REM sleep generating regions to induce muscle atonia. We propose that the suppression of monoaminergic cell firing during cataplexy disfacilitates both motor neurons and REM sleep suppressing regions, permitting muscle atonia during waking.

137 122 Chapter 6: Role for the Amygdala in Triggering Cataplexy Other researchers contributed to this work: Takatoshi Mochizuki, PhD: Assisted with setting up counts for wheel running experiments. Yo Oishi, PhD: Assisted with the anterograde tracing experiments. Tom Scammell, MD: Assisted with experimental design.

138 123 Chapter 6: 6.1 Abstract Role for the Amygdala in Triggering Cataplexy One of the most striking aspects of cataplexy is that it is usually triggered by strong, positive emotions, but almost nothing is known about the neural pathways through which positive emotions trigger muscle atonia. We hypothesized that the amygdala is necessary for cataplexy because it contains neurons that are active during cataplexy and is thought to mediate positive emotions. Using anterograde tracing in mice, we found that neurons in the amygdala heavily innervate neurons in the LPT and vlpag, brain areas that suppress muscle atonia. We then found that bilateral, excitotoxic lesions of the amygdala (central and basolateral nuclei) markedly reduced cataplexy in orexin KO mice, a mouse model of narcolepsy. These lesions did not alter basic sleep-wake behavior, but substantially reduced cataplexy under baseline conditions and when mice had access to running wheels and chocolate, conditions of high arousal that should elicit positive emotions. These observations demonstrate that the amygdala is a part of the underlying cataplexy circuitry and help generate a new model that explains how positive emotions trigger cataplexy. 6.2 Introduction Narcolepsy is caused by loss of the hypothalamic neurons that produce orexin neuropeptides (Peyron et al., 2000, Thannickal et al., 2000). Loss of these cells or loss of just the orexin peptides results in severe sleepiness and cataplexy, the sudden loss of postural muscle tone during waking. In people with narcolepsy, cataplexy is most often triggered by positive emotions such as those associated with laughter, joking, or delight (Overeem et al., 1999). Similarly, in narcoleptic dogs, cataplexy is usually triggered by palatable food or play (Baker et al., 1982, Siegel et al., 1989, Nishino et al., 1991), and in mouse models of narcolepsy, cataplexy is increased by rewarding stimuli such as wheel running and palatable food (Espana et al., 2007, Clark et al., 2009). This connection to positive emotions has been recognized as a key aspect of cataplexy since its first description in 1880 (Gelineau, 1880), but the neural mechanisms through which positive emotions trigger cataplexy remain unknown.

139 124 Several lines of evidence suggest that the amygdala could be a key site through which emotions trigger cataplexy. There is clear evidence that the amygdala is important for responses to positive stimuli in humans, non-human primates, and rodents. Activity increases in human amygdala in response to positive affective stimuli, and amygdala neurons encode positive value of conditioned images in non-human primates (Nishijo et al., 1988, Garavan et al., 2001). In rats, amygdala neurons encode positive stimulus associations and amygdala lesioned rats failed to approach stimuli of positive affective valence (Schoenbaum et al., 1998, Paton et al., 2006). Anatomically, the amygdala is well-positioned to influence muscle tone and REM sleep phenomena as the central nucleus of the amygdala in rats innervates regions in the pons that regulate muscle atonia during REM sleep (Wallace et al., 1989, 1992). In addition, Gulyani and colleagues recorded from the amygdala of freely behaving narcoleptic dogs and found a large number of neurons with increased activity during cataplexy (Gulyani et al., 2002). These cells often showed a sharp increase in firing at the onset of cataplexy and then a quick return to baseline just as muscle tone recovered, suggesting they may be part of a cataplexy effector mechanism. Building on these observations, we hypothesized that the amygdala is necessary for cataplexy. We first examined if the amygdala innervates regions of the pons that regulate muscle atonia in mice. Then, to test whether the amygdala is functionally necessary for cataplexy, we produced bilateral excitotoxic lesions of the amygdala in orexin KO mice and examined their effect on cataplexy under conditions likely to elicit positive emotions. 6.3 Methods These studies were approved by the Institutional Animal Care and Use Committees of Beth Israel Deaconess Medical Center and Harvard Medical School and were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals Animals We used 36 male, orexin KO mice, weeks old and weighing 26-34g. Founder mice were a kind gift from M. Yanagisawa (University of Texas Southwestern) and were then backcrossed to C57BL/6J mice for over 10 generations. Mice were genotyped using PCR with genomic primers 5 -GACGACGGCCTCAGACTTCTTGGG, 3 TCACCCCCTTGGGATAG

140 125 CCCTTCC, and 5 -CCGCTATCAGGACATAGCGTTGGC (with forward primers being specific for either wildtype or KO mice and the reverse primer being common to both) Surgery We anesthetized mice with ketamine-xylazine (100 and 10 mg/kg i.p.) and placed them in a stereotaxic alignment system (Model 1900, Kopf). Using an air pressure injection system and glass micropipette (tip diameter ~10 um), we bilaterally injected ibotenic acid (5% in PBS; nl injected over 3-5 min) into the area of the amygdala (1.35 mm posterior to bregma, ±2.75 mm lateral, 4.5 mm ventral). To induce sham brain lesions, we microinjected the amygdalae of control mice with an equal volume of sterile PBS. We then implanted mice with electrodes for recording the EEG and EMG. In brief, stainless steel screws were implanted for frontoparietal EEG recordings (1.5 mm lateral and 1 mm anterior to bregma; 1.5 mm lateral and 3 mm posterior to bregma). EMG electrodes were made from fine, multi-stranded stainless steel wire (AS131, Cooner Wire, Chatsworth, CA), which were sutured into the neck extensor muscles. All electrodes were attached to a micro-strip connector affixed to the animal's head with dental cement. After surgery, mice were given 0.5 ml of 0.9% saline and meloxicam (5mg/kg; i.p.) Experimental protocol Two weeks after surgery, we transferred mice to recording cages in a sound-attenuated chamber with a 12:12 light-dark cycle (30 lux daylight-type fluorescent tubes with lights on at 07:00), constant temperature (23 ±1 C), and with food and water available ad libitum. The recording cable was attached to a low torque electrical swivel, fixed above the cages that allowed free movement. Mice habituated to the cables for 4 days before the experiments began and remained connected throughout the study. We first examined baseline sleep-wake behavior across 24 hours with EEG, EMG and infrared video recordings. We then studied mice under two conditions that should increase cataplexy: access to a running wheel, and access to a running wheel and chocolate. We placed a low-torque, polycarbonate running wheel (Fast-Trac, Bio-Serv, Frenchtown, NJ) in each cage and recorded wheel rotations with a photodetector beneath each wheel. Running wheels increase cataplexy in orexin KO mice (Espana et al., 2007), and this style of wheel was chosen because it does not interfere with the EEG recording cable. After 7 days habituation to the wheel, we

141 126 recorded sleep-wake behavior and wheel rotations. The next night, we gave mice 3g of milk chocolate (Hershey s) at dark onset and recorded sleep-wake behavior and wheel running over the next 12 hours (19:00-7:00). We chose to use chocolate because it is used as a reward in rodent operant studies (Holahan et al., 2011, King et al., 2011), and cataplexy in mice and dogs is increased by palatable foods (Siegel et al., 1986, Clark et al., 2009) Data acquisition and analysis EEG/EMG signals were acquired using Grass Model 12 amplifiers (West Warwick, RI) and digitized at 256 Hz. Signals were digitally filtered (EEG: Hz, EMG: Hz) using SleepSign (Kissei Comtec, Matsumoto, Japan). We manually scored behavior as wake, NREM sleep, REM sleep, or cataplexy in 10s epochs. Behavior was scored as cataplexy based on the consensus definition of murine cataplexy (Scammell et al., 2009). Specifically, if the mouse had one or more epochs of muscle atonia accompanied by EEG theta that was preceded by at least 40s of active wakefulness and was also followed by wakefulness the period of atonia was scored as cataplexy. Cataplexy was scored using both EEG/EMG as well as infrared video recordings Histology After recordings, we anesthetized mice with ketamine-xylazine (100 and 10 mg/kg i.p.) and transcardially perfused them with 0.1M PBS followed by 10% formalin. We postfixed brains in formalin for 24 hours, and then cryoprotected them in 20% sucrose for ~48 hours. We coronally sectioned brains at 40 microns in a 1:3 series using a microtome. We stained one series with thionin and mapped the lesions on standard brain atlas maps (Paxinos and Franklin, 2001). Criteria for inclusion in the lesion group were symmetrical, bilateral lesions of the amygdala that encompassed most of the central nucleus and basolateral nucleus of the amygdala without much injury to adjacent structures. The largest lesions encompassed the entirety of the amygdala while occasionally lesioning cells in the piriform cortex and ventral regions of the caudate-putamen. There were no obvious observable behavioral differences between mice with smaller vs. larger lesions. We excluded mice that received off-target ibotenic acid injections resulting in asymmetrical or unilateral lesions. In 6 of the excluded mice that had unilateral hit/unilateral miss lesions, such that the total volume of lesioned area was similar to bilateral amygdala lesioned mice, we still scored sleep-wake behavior and cataplexy. In these mice there was no

142 127 change in sleep-wake behavior or cataplexy when compared to control mice (p>0.05 for all states) Anterograde tracing Under ketamine-xylazine anesthesia, we microinjected an adeno-associated viral vector coding for green fluorescent protein (AAV-GFP; 20-50nl; serotype 8; 7x10 12 p/ml; Harvard Gene Therapy Initiative) into the amygdala of wild type mice to anterogradely label projections. Two to three weeks later, we perfused the mice and sliced brain sections as above. We immunostained for GFP by washing sections in PBS with triton x (0.25%) and then incubating in primary antiserum (Invitrogen at 1:20,000) for two days at room temperature. Sections were then washed in PBS with triton x (0.25%) and incubated in biotinylated secondary antiserum (against rabbit IgG, 1:1000, Vector) for two hours, washed and incubated in ABC reagents for two hours. Sections were then washed again and incubated in solution of 0.06% 3,3-diaminobenzidine tetrahydrochloride (DAB, Sigma) Statistical analysis Paired t-tests were used for comparisons within each group, and unpaired t-tests were used for comparisons between groups. Comparisons of frequency, duration and total time spent in each behavioral state between treatments were made using unpaired t-tests. All statistical analyses were performed using SigmaStat (SPSS Inc.) and applied a critical 2-tailed α value of p<0.05. Data are presented as mean ± standard error of the mean. 6.4 Results The amygdala is anatomically well positioned to regulate cataplexy The vlpag and adjacent LPT are thought to be key sites for the suppression of muscle atonia and REM sleep, as lesions of this region increase REM sleep and may permit muscle atonia to occur outside of REM sleep (Lu et al., 2006b, Kaur et al., 2009). We found that microinjection of the anterograde tracer AAV-GFP into the central nucleus of the amygdala densely labeled projections to the vlpag and LPT (n=3; Figure 6.1). In addition, we saw innervations of the lateral hypothalamus, including putative orexin neurons (data not shown). These tracing studies demonstrate that the CeA of mice has strong direct projections to the vlpag/lpt. In addition, the CeA neurons innervate the orexin neurons which send a

143 128 presumably excitatory projection to the vlpag/lpt. These pathways are similar to those previously described in rats and guinea pigs (Boissard et al., 2002, Fung et al., 2011), and confirm that the amygdala is well positioned to integrate neuronal signals related to emotions and relay them on to neurons regulating atonia Amygdala lesions reduced cataplexy under baseline conditions Most bilateral lesions of the amygdala encompassed both the central and basolateral nuclei, but lesions varied, with a few affecting only the CeA and several involving much of the amygdala (Figure 6.2). Nine mice met criteria for acceptable bilateral lesions. Six other mice had essentially unilateral lesions (unilateral hit/unilateral miss, such that the overall lesioned area was approximately equal to bilateral lesion mice) and had normal amounts of cataplexy (data not shown). Sham-lesioned control mice (n=8) showed no evidence of injury. Bilateral amygdala lesions significantly reduced the amount of time spent in cataplexy during the dark period (p=0.02, Figure 6.3A), with 28% fewer bouts of cataplexy (p<0.001) and a 30% decrease in the average duration of cataplexy bouts (p<0.001). The lesions did not affect wake, NREM or REM sleep parameters (Table 6.1) Amygdala lesions decreased cataplexy triggered by a positive stimulus To determine whether the amygdala is necessary for cataplexy triggered by a positive stimulus, we examined cataplexy and sleep-wake behavior in amygdala-lesioned orexin KO mice with access to a running wheel (n=9). Lesioned mice spent 58% less time in cataplexy than control mice (p=0.023; Figure 6.3B). The lesioned mice also had 52% fewer cataplexy bouts and a 20% decrease in the average duration of cataplexy bouts, but these changes were not statistically significant. Amygdala lesions had no effect on sleep-wake architecture in the presence of a running wheel (Table 6.1). The lesioned mice had a slight decrease in total wheel rotations (7375 ±2603 vs ±2058 rotations, lesion vs. control, p=0.161).

144 129 Figure 6.1: The central nucleus of the amygdala projects to brainstem regions that regulate REM sleep Anterograde tracing from the central nucleus of the amygdala (CeA) showing observable projections to the ventrolateral periaquaductal gray (vlpag) and lateral pontine tegmentum (LPT) in the mouse. These areas have been implicated in the regulation of muscle atonia during REM sleep.