Narcolepsy, orexins and respiratory regulation


Prof. Dr Fang Han, Department of Pulmonary Medicine, The People's Hospital, Beijing University, Beijing 100044, China. Email:


Narcolepsy is a debilitating sleep disorder characterized by excessive daytime sleepiness, cataplexy and intrusive rapid–eye movement sleep. Deficits in endogenous orexins are a major pathogenic component of the disease. This disorder is also associated with the gene marker HLADQB1*0602. Orexins as hypothalamic neuropeptides have multiple physiological functions, and their primary functions are regulation of the sleep–wake cycle and feeding. Evidence from animal studies using orexin knockout mice and focal microdialysis of an orexin receptor antagonist at the retrotrapezoid nucleus and medullary raphe in rats demonstrated that orexins also contribute to respiratory regulation in a vigilance state–dependent manner, as animals with orexin dysregulation have attenuated hypercapnic ventilatory responses predominantly in wakefulness. These findings are consistent with the notion that the activity of orexinergic neurons is higher during wake than sleep periods. Orexin neurons seem to be a pivotal link between conscious and unconscious brain functions in animals. The human model of hypocretin deficiency is patients with narcolepsy–cataplexy. In contrast to the findings suggested by animal studies, we found significant decreases in hypoxic responsiveness, but not in hypercapnic responsiveness, in narcoleptics, and further analysis indicated that decreased ventilatory responses to hypoxia in human narcolepsy–cataplexy is in relation to HLA-DQB1*0602 status, not hypocretin deficiency. This is confirmed by the fact that the hypoxic responsiveness was lower in HLA positive versus negative controls. Unlike in mice, hypocretin-1 is not a major factor contributing to depressed hypoxic responses in humans. Species differences may exist.


The regulation of breathing relies upon chemical feedback concerning the levels of CO2 and O2. Such chemosensory reflexes influence sleep apnea, chronic obstructive pulmonary disease (COPD), heart failure, and acute adaptation to high altitude.1 A variation in chemosensitivity across various inbred rodent strains and familial clustering of ventilatory traits in humans provide a strong rationale for gene and protein isolation efforts to unravel molecular mechanisms for the healthy and diseased operation of these traits.2 Orexin neurons seem to be a pivotal link between sleep and wake brain functions in animals.3 Loss of hypothalamic-neurotransmitter orexins characterizes narcolepsy with cataplexy, a sleep–wake disorder also associated with the gene marker HLADQB1*0602.4–8 Depressed hypercapnic responses in hypocretin-deficient animals and an increased prevalence of sleep apnea in narcolepsy suggested interactions among ventilatory chemosensitivity, narcolepsy–cataplexy, and sleep apnea.9–15

The present review will summarize evidence from animal studies using orexin knockout mice and focal microdialysis of orexin-receptor antagonist at the critical areas for respiratory activities in the central nervous system (CNS), which indicates that orexins may contribute to respiratory regulation in a vigilance state–dependent manner.16–18 We will also report chemoresponsiveness in patients with narcolepsy–cataplexy, a human model of orexin deficiency, which indicates an unexpected finding that the mechanism for differences between patients and controls in hypoxic responsiveness could relate to HLA marker status rather than orexin deficiency.19


The neurotransmitters orexin-A and B (Hypocretin-1 and 2) are derived from a common 130–131 amino-acid precursor, prepro-orexin, encoded by a gene localized to human chromosome 17q2.20 Orexin A contains 33 amino acids and orexin B contains 28. These neurotransmitters act on two subtypes of G protein–coupled receptors, orexin receptors 1 and 2 (OX1R and OX2R). OX2R is a nonselective, high-affinity receptor for both orexin neuropeptides, whereas OX1R is selective for orexin A alone.21 The orexin-containing neurons are exclusively located within the perifornical and dorsomedial–lateral hypothalamus. These cells diffusely project to many areas of the brain and spinal cord including the forebrain, limbic and brainstem nuclei,3,22 and have a critical role in multiple physiological functions, from their primary roles in the sleep–wake cycle and feeding to the control of the cardiovascular system, pain, locomotion, stress, and addiction as well as their involvement in psychiatric disorders such as panic, anxiety, and depression.3

The intial evidence for orexin being critical for the regulation of the sleep–wakefulness cycle lies in the fact that orexin neurons have widespread excitatory projections throughout the brain with particularly dense innervations in regions related to sleep–wake regulation, such as the hypothalamus, the locus coeruleus (LC), the dorsal raphe nucleus (DR), the cerebral cortex,22,23 and the brainstem reticular formation, including the rapid–eye movement (REM) sleep–inducing region located in the ventral portion of the oral pontine reticular nucleus (vRPO).24

Orexin-producing neurons discharge in tight synchrony with electroencephalographic (EEG) transitions to states of increased alertness, and cease firing during sleep.25 Selective stimulation of the orexin neurons is sufficient to trigger awakening from non-REM (NREM) and REM sleep. Alternation of short bouts of wakefulness and sleep can be attributed to the orexin-independent mutual inhibition between sleep-promoting neurons and wake-promoting neurons. To maintain a long, consolidated waking period, excitatory inputs from orexin neurons to waking-active monoaminergic and cholinergic neurons in the hypothalamus and brainstem regions are also necessary. Orexin neurons also provide indirect inhibition of sleep-promoting neurons by enhancing the inhibition arising from arousal neurons and/or local inhibition within the lateral hypothalamus.26 Recent findings indicate that orexin neurons are involved in the circadian control of REM sleep.27 Thus, orexin activity promotes and maintains wakefulness by exciting nuclei that govern arousal, whereas its inhibition leads to sleep initiation.28

Orexin neuropeptides have wake-promoting and sleep-suppressing actions. Orexin tends to be released within the brain at high levels during active waking and during REM sleep, and at minimal levels during NREM sleep. Accordingly, intra-cerebroventricular administration of orexin A and B has a robust wake-promoting effect and leads to a remarkable decrease in NREM sleep and REM sleep,29,30 effects similar to those seen with optogenetic stimulation of orexin neurons.31 In rats, dogs and humans, somnolence is induced by pharmacological blockade of both orexin-A and -B receptors.28 In relation to the control of REM sleep generation, orexin projections and receptors have been identified in cholinoceptive areas of the pontine reticular formation involved in REM generation and control of REM-polygraphic signs. Furthermore, orexin enhances acetylcholine and GABA release in these areas.32 The importance of orexin in the regulation of sleep–wakefulness is highlighted by the identification of a link between narcolepsy and orexin deficiency.4–6


Narcolepsy is a disabling sleep disorder characterized by sleepiness, cataplexy, hypnagogic hallucinations, sleep paralysis and disturbed nocturnal sleep, including fragmentation of sleep and direct transitions from wakefulness to REM sleep. It affects about 0.02%–0.05% of the general population in different ethnic groups.33–35 The disorder most commonly starts in the teens or twenties, but it can be present as early as one year of age or begin as late as middle age.36 The diagnosis of narcolepsy rests on clinical grounds. Excessive daytime sleepiness, as the most common presenting complaint, and cataplexy, as the most specific, are the cardinal symptoms. When narcolepsy is suspected, polysomnography (PSG) followed by a multiple sleep latency test (MSLT) is the most useful diagnostic test. A mean sleep latency ≤8 min and two or more sleep-onset REM periods indicate the existence of daytime sleepiness and an abnormal onset of REM sleep.37 In addition to optimizing nocturnal sleep duration and planned daytime naps, medications such as methylphenidate and clomipramine are used for the treatment of daytime sleepiness and/or REM sleep intrusion phenomena.

Genetic predisposition and environmental factors are both considered important for the development of narcolepsy. Most cases of human narcolepsy are sporadic: 1–2% of first-degree relatives may have a clear-cut family history of the disorder.38 Susceptibility to narcolepsy is tightly associated with a specific HLA allele, DQB1*0602, and this marker is seen in 90% to 100% of patients, and also occurs in as many as 50% of people without narcolepsy in different populations.8 The major pathophysiology of narcolepsy has been recently elucidated based on the discovery of its link to orexin dysregulation. The clues suggesting the possible involvement of orexin in narcolepsy initially came from animal models, mice lacking either the orexin gene (prepro-orexin knockout mice) or orexin neurons (orexin/ataxin-3 transgenic mice), as well as mice and dogs with null mutations in the OX2R gene, share a narcolepsy-like phenotype, including altered sleep–wake regulation and the sudden onset of muscle atonia.4,5,39 In contrast, most human cases of narcolepsy are not caused by gene mutations. Systematic screening of mutations in the orexin system in patients with cataplexy (including familial, early-onset and HLA-negative cases) has so far identified only one patient with a mutation in orexin-related genes, and this patient was atypical with very early disease onset (six months old).7 Narcoleptic patients were found to have low levels of orexin-A in their cerebro-spinal fluid (CSF). Nishino et al.6 first reported lumbar CSF orexin-A levels below the detection limit (40 pg/mL) in narcoleptics, while control subjects had an average of 280 pg/mL. These findings have been extended and replicated by other groups.40–44 Based on DQB1*0602 positivity and the presence of cataplexy, more than 95% of narcolepsy patients have orexin deficiency with low orexin levels of <110 pg/mL.41 Further pathologic evidence demonstrates a degenerative loss of orexin neurons in human narcolepsy. Experiments using in situ hybridization, immunochemistry and radioimmunological assays of peptides in the post-mortem brain tissue of narcoleptic patients found undetectable levels of pre-orexin RNA, loss of orexin peptides, and an 80–100% reduction in the number of orexin-containing neurons in the hypothalamus.39,45 Together, these results indicate that orexin production or orexin neurons are selectively damaged in narcoleptic patients, and narcolepsy is the direct consequence of orexin-neuron degeneration.


A diverse group of processes are involved in the control of ventilation. Several brainstem nuclei form a central neuronal network; peripheral and central respiratory chemoreceptors sense hypoxic and hypercapnic challenges; these are processed through neuromuscular circuits resulting in an increase (or decrease) of tidal volume, respiratory frequency and minute ventilation to maintain constant levels of arterial PO2, PCO2 and [H+]. Basal respiration and respiratory reflex regulations are considerably different during the awake and sleep states. Sleep and wakefulness regulate breathing in a state-dependent activity pattern. Neurotransmitters are also involved in the central respiratory drive.1,46

The hypothalamus has long been known to be involved in the modulation of ventilation.47 For example, Horn and Waldrop48 demonstrated that the posterior hypothalamus exerted an excitatory effect on both hypoxic and hypercapnic ventilatory responses. Electrical impulses from the lateral hypothalamic area elicit increases in respiratory activity; however, this effect is diminished in mice genetically engineered to lack orexin,49 indicating that it is mediated partially by hypothalamic orexin-containing neurons. Immunohistochemical evidence has also documented orexin A immunoreactive nerve fibres projecting to the major brain stem–spinal cord respiratory-related network, such as the dorsolateral pons including the Kolliker-Fuse nucleus, dorsal and rostral ventrolateral medulla, pre-Bötzinger complex, the nucleus tractus solitarius, raphe nuclei, hypoglossal nuclei and phrenic nuclei. All these areas are critical for respiratory rhythmogenesis, respiratory activity regulation and upper-airway and diaphragm-movement control.50–53

Orexin neurons are connected to both the central respiratory nuclei and arousal regions, and may serve to match ventilation to changes in states of consciousness. Activation of orexin receptors at various sites of the brain stem and spinal cord influences breathing parameters that determine breathing rate and depth, and coordination between upper-airway and thoracic-pump muscles. Hypothalamic orexin neurons are CO2/pH sensitive in vitro and in vivo.54,55 Using microinjection technique to microperfuse orexin in either of the pre-Bötzinger region and the phrenic nucleus produced a significant dose-dependent increase in diaphragm electromyographic activity.50 Exogenous orexin injected into the hypoglossal motor nucleus increases genioglossus muscle activity.52 Microinjection of orexin B into the Kolliker-Fuse nucleus significantly increases breathing frequency.56 In addition, orexin-B in the pontine respiratory group prolongs the preinspiratory activity of the hypoglossal nerve, a component essential for the maintenance of airway patency.54

Kuwaki and colleagues used prepro-orexin knockout mice as an animal model and studied the possible contribution of orexin to chemoresponsiveness.16 A decreased attenuation of respiratory excitation was noticed during fight-or-flight responses.57 Later they confirmed that orexin knockout mice have attenuated hypercapnic ventilatory responses during the awake but not during the sleep period, which is consistent with the notion that the activity of orexinergic neurons is higher during wake than sleep periods.9 Further studies from the same group found that supplementation of orexin-A or -B partially restores an attenuated response to hypercapnia, while blockage of orexin receptor 1 by SB-334867 weakened the hypercapnic chemoreflex of the wild-type mice,10 indicating that orexin receptor 1 is necessary for chemoreflex control. However, orexin knockout mice did not respond differently from wild-type mice in regard to the response to hypoxia. Therefore, orexins may not be involved in ventilatory hypoxic response in animals. These findings are also confirmed by recent studies using OX1R and OX2R antagonist. Focal antagonism of the OX1R in two putative central chemoreceptor sites, the retrotrapezoid nucleus (RTN) or the medullary raphe, results in a reduction of the CO2 response predominantly in wakefulness (−30% and −16%, respectively).13,14 Selective inhibition of both orexin receptors (OX1R and OX2R) at all central locations by an orally administered dual–orexin receptor antagonist, almorexant, induced an attenuated CO2 response by 26% only in wakefulness during the dark (active) period of the diurnal cycle to a level observed during NREM sleep in the light period in controls. Almorexant also decreased wakefulness and increased NREM and REM sleep during the dark period, as previously reported, and unexpectedly decreased the number of sighs and post-sigh apneas during wakefulness in both the light and the dark period and during both wakefulness and NREM sleep in the dark period.15 All these studies indicated that the orexin system participates importantly in central chemoreception in a vigilance-state– and diurnal-cycle–dependent manner and the sleep–wake difference in the CO2 response can in large part be attributed to orexin.15–17 Orexins may also be involved in the mechanisms of stabilization of breathing. Post-hypoxic long-term facilitation, a physiologic feature that is presumed to stabilize the respiratory control system and reduce sleep apnea, is absent in orexin knockout mice,58 which is consistent with the finding that orexin knockour mice had more frequent apnea during sleep.


The human model of orexin deficiency is patients with narcolepsy and cataplexy6,7,45 who are uniformly HLADQB1*0602 positive and have orexin levels <110 pg/mL. In addition to the possible influence of orexin per se, there is evidence leading us to consider an impaired chemoresponsiveness in patients with narcolepsy–cataplexy. Narcolepsy is associated with a higher expression of sleep apnea,11,12 which has been considered as a disorder of respiratory control.59 Sleep in narcolepsy patients is characterized by overwhelming excessive sleepiness occurring during the daytime, and fragmented nocturnal sleep that is disrupted by numerous awakenings. A number of studies have demonstrated that one night of sleep deprivation significantly reduced both ventilatory chemosensitivity and arousal responses in healthy adults.60 The long-term nocturnal sleep disturbance in narcoleptics may alter their chemoresponsiveness, mimicking the changes caused by sleep deprivation. The fragmented sleep and hypoxia induced by a high incidence of sleep apnea may further impair the ventilatory responsiveness of these patients.61

To test the hypothesis that narcolepsy–cataplexy patients with orexin deficiency may have depressed chemoresponses to hypoxia and/or hypercapnia, we measured hyperoxia, hypercapnic and hypoxic responsiveness in 130 patients with narcolepsy–cataplexy (age 20 ± 10 years; 69% male) and 117 age- and sex-matched controls (age 22 ± 6.9 years; 62% male). In contrast to findings suggested by animal gene-knockout models, we found no group differences with regard to hypercapnic responsiveness, but rather found significant differences in hypoxic responsiveness. Reduced hypoxic responsiveness in narcolepsy–cataplexy patients was unrelated to BMI, age and sex.19

Chronic and acute sleep deprivation significantly reduced ventilatory chemosensitivity in normal people, and vigilance state was found to influence chemoresponsiveness.60 All narcolepsy patients in this study had severe daytime sleepiness, as indicated by the short sleep latency and high mean ESS score. It is reasonable to speculate about the possible contribution of daytime sleepiness in reduced ventilatory chemosensitivity. However, the hypoxic response did not change following the improvement of daytime sleepiness after methylphenidate treatment in cataplectic patients (80 patients, Han et al., unpublished data, 2009). As cataplexy attacks in this group of cataplectic patients were relatively severe and some of them occasionally had “status cataplecticus”, the low ventilatory response to hypoxia in cataplectics may be related to cataplexy itself, the REM sleep–like atonia of the respiratory striated muscles (but not the diaphragm). However, hypoxic responses did not change with the improvement of cataplexy after oral clomipramine administration (20 patients, Han et al., unpublished data, 2009). Therefore, neither sleepiness nor cataplexy may be the major factor involved in the depressed respiratory hypoxic chemoreflex in patients with cataplexy.

In the group of 130 young cataplectic patients with a mean age of 20 years old, we found19 a higher apnea–hypopnea index (AHI) (2.8 ± 5.4 vs 0.8 ± 1.6/hr, P= 0.03) and lower minimal oxygen saturation during sleep (87% ± 7 vs 91% ± 4, P= 0.0002) than that of controls. Five percent of them, but none of the control subjects, had an AHI >15; the prevalence of an AHI >5 was higher (16%) than that in the control group (5%), which is comparable to the reported value of 4%–5% in the older (age >30) general population.62 We did not exclude the patients with sleep apnea from the present study, and the decision that sleep apnea is not the major cause of daytime sleepiness was decided clinically (severity, nature, and onset of sleepiness; presence of cataplexy) and on the basis of the PSG results. No significant improvement of daytime sleepiness following two weeks of continuous positive airway pressure (CPAP) therapy was also used to confirm the diagnosis of narcolepsy for those with AHI >5 and whose sleepiness was felt to be not solely explained by narcolepsy. Ventilatory responses during wakefulness can be blunted by the consequences of sleep-disordered breathing, such as sleep fragmentation, hypoxia, and hypercapnia.61 A reversible blunted hypoxic response may exist in patients with both narcolepsy and sleep apnea. We did not look at the treatment effect of the CPAP on the chemoresponsiveness of the patients in the present study. However, analysis after the exclusion of those with AHI >5 and AHI >15 did not reveal any difference from that of inclusion of these patients in regard to the statistical significance. In summary, the co-existence of sleep apnea did not play an important role in the depressed hypoxic reflex in this group of cataplectic patients, probably due to the small numbers of patients with sleep apnea, the young age and short disease course, and mild severity of sleep apnea with a normal whole-night mean value of arterial oxygen saturation (SaO2).

All subjects with narcolepsy–cataplexy and 22% of the control population were also HLA-DQB1*0602 positive. Further comparison using multivariate analysis indicated that depressed ventilatory responses to hypoxia in human narcolepsy–cataplexy is related to HLA-DQB1*0602 status, not orexin deficiency. This is confirmed by the fact that the hypoxic responsiveness was lower in HLA positive versus negative controls (0.13 ± 0.08 vs 0.20 ± 0.14 L/min/%SpO2, P= 0.00004).19 The finding in humans indicates that, unlike in mice, orexin-A is not a major factor contributing to depressed hypoxic responses. This difference cannot be fully explained by the fact that orexin knockout mice do not provide the most accurate pathophysiological model of human narcolepsy, as we also did not find depressed hypoxic responses by using orexin/ataxin-3 transgenic mice in which the etiology and course of the disease are similar to human narcolepsy (Han et al., unpublished data, 2009). Species differences may exist. The animal study showed that orexin-B had a stronger effect than orexin-A on increasing spontaneous ventilation.10 This is a question worth investigating further when a specific method of measuring CSF orexin-B in humans becomes available.


Human narcolepsy–cataplexy with genetic marker of HLADQB1*0602 and deficits in the endogenous orexin system also occurs with exaggeration of sleep apnea. The underlying mechanisms remain to be elucidated. Animal studies indicate that the orexin system is one of the essential modulators required for coordinating the circuits controlling respiration and behavior. A blunted response to ventilatory stimuli, impairment in long-term facilitation and lack of orexin excitatory drive to hypoglossal premotoneurons all contribute to the development of sleep apnea in animals. Moreover, respiration can be effectively altered by drugs targeting the orexin system, potentially suggesting new strategies for treating respiratory disorders. Although human data seem not to support a contribution of orexin to the hypercapnic chemoreflex, they do indicate that narcolepsy with orexin deficiency seems to have another type of ventilatory regulation effect, resulting in the occurrence of sleep apnea. A close relationship between reduced hypoxic responsiveness and DQB1*0602 occurred not only in patients, but also in healthy, unaffected subjects. We speculate a novel effect of HLA or of a gene polymorphism located nearby on the well-known interindividual variation in hypoxic responsiveness. As recent developments in narcolepsy research support the hypothesis of narcolepsy being an immune-mediated disease,63 a close association of hypoxic response with HLA-DQ0602 implies the possibility of immune-mediated destruction of type-I glomus cells in the carotid bodies, the peripheral chemoreceptor detectors of hypoxia.


The study was supported by research grants from NSFC (C30770938, C30300120) and Ministry of Education China (985-2-084-113).


The author indicated no potential conflict of interests.