Animal models of narcolepsy and the hypocretin/orexin system: Past, present, and future

Abstract

Animal models have advanced not only our understanding of the etiology and phenotype of the sleep disorder narcolepsy but have also informed sleep/wake regulation more generally. The identification of an inheritable narcolepsy phenotype in dogs in the 1970s allowed the establishment of a breeding colony at Stanford University, resulting in studies that provided the first insights into the genetics and neurotransmitter systems that underlie cataplexy and rapid-eye movement sleep atonia. Although the discovery of the hypocretin/orexin neuropeptides in 1998 initially seemed unrelated to sleep/wake control, the description of the phenotype of the prepro-orexin knockout (KO) mouse as strongly resembling cataplexy, the pathognomonic symptom of narcolepsy, along with identification of a mutation in hypocretin receptor-2 gene as the source of canine narcolepsy, unequivocally established the relationship between this system and narcolepsy. The subsequent discovery of hypocretin neuron degeneration in human narcolepsy demystified a disorder whose etiology had been unknown since its initial description 120 years earlier. These breakthroughs prompted the development of numerous other animal models that have allowed manipulation of the hypocretin/orexin system, thereby advancing our understanding of sleep/wake circuitry. While animal models have greatly informed understanding of this fascinating disorder and the role of the hypocretin/orexin system in sleep/wake control, the question of why these neurons degenerate in human narcolepsy is only beginning to be understood. The development of new immune-mediated narcolepsy models are likely to further inform the etiology of this sleep disorder and animal models will undoubtedly play a critical role in the development of novel narcolepsy therapeutics.

Statement of Significance.

Our understanding of the sleep disorder narcolepsy is intimately intertwined with the identification and development of animal models with dysfunctional hypocretin/orexin systems. This review describes how studies of animal models of narcolepsy have facilitated our understanding of hypocretin/orexin dysfunction that results in the symptomatology of human narcolepsy and how this system is involved in arousal state regulation. We discuss both spontaneously-occurring and genetically engineered narcolepsy models and the advances they have enabled, and speculate on future models that could inform still-enigmatic aspects of this disorder which could, in turn, facilitate the identification of novel therapeutic targets for treating narcolepsy and other disorders of hypersomnolence.

Narcolepsy: An Introduction

Animal models are important tools in the investigation of the etiology, pathology, and phenotyping of neurological diseases, as well as for the development and testing of potential therapeutics. Our understanding of the neural basis of the sleep disorder narcolepsy and arousal state regulation more generally are intimately intertwined with the development and study of animal models of this sleep disorder. The first unambiguous descriptions of narcolepsy symptoms were by Westphal in 1877 and Gélineau in 1880 [1–3]. These authors described similar cases characterized by hypersomnia and episodes of muscle weakness that could be triggered by the patients’ emotional state, but which were distinguished from epilepsy by the maintenance of consciousness during the attacks. The first attempts to establish systematic criteria for the diagnosis of narcolepsy relied on the assessment of a tetrad of narcolepsy symptoms consisting of (1) hypersomnia or, more specifically, excessive daytime sleepiness (EDS), which is characterized an abrupt or persistent, often overwhelming need to sleep during the daytime, irrespective of previous sleep time or quality; (2) cataplexy, an unexpected reduction in, or loss of, muscle tone without the loss of consciousness, which can be brought on by positive emotional stimuli; (3) hypnogogic hallucinations, characterized by vivid visual and/or auditory hallucinatory fabrications occurring when falling asleep; and (4) sleep paralysis, the brief, reversible inability to move occurring during sleep onset or awakening [4]. Many of these symptoms are also associated with other diseases of somnolence, the exception being cataplexy which is pathognomonic of type 1 narcolepsy (NT1). In addition, the tetrad of symptoms frequently presents in an incongruent manner with varying degrees of severity in different patients with narcolepsy (e.g. in type 2 narcolepsy (NT2); see below). Both of these factors contribute to the difficulty in diagnosing patients with narcolepsy, particularly in children where the symptoms can be even more complex [5].

The understanding of narcolepsy symptoms and the diagnostic criteria used to identify individuals suffering from narcolepsy became further refined with the identification of the distinct electrophysiological substates of sleep [6–8] and later with the development of the two-process model of sleep regulation [9]. The identification of sleep substates led to the recognition that, not only were people with narcolepsy exhibiting excessive daytime sleepiness (EDS) and fragmented nocturnal sleep, they were also entering rapid-eye movement (REM) sleep much more rapidly than healthy individuals [10–12]. These premature REM events, referred to as sleep-onset REM periods (SOREMs), led to the concept of narcolepsy as a sleep disorder with symptoms of dysregulated REM sleep and other narcolepsy symptoms were subsequently viewed in this light. Hypnogogic hallucinations and sleep paralysis were thus viewed as intrusions of REM-related dream imagery and atonia during sleep onset or awakening, and cataplexy was viewed as the occurrence of REM-associated atonia during wakefulness.

Narcolepsy is characterized both by difficulty maintaining wakefulness during the daytime and fragmented sleep during the night [13, 14], but EDS cannot be fully explained as a compensatory homeostatic response to nocturnal sleep fragmentation [15, 16]. In the two-process model of sleep regulation, the timing of sleep is viewed as the interaction between a circadian timekeeping and homeostatic sleep mechanism [9]. Individuals with narcolepsy show a compensatory rebound in NREM slow-wave activity (SWA) following sleep deprivation but exhibit an attenuated circadian distribution of sleep, suggesting that the homeostatic system is intact but that the circadian effector mechanisms responsible for promoting alertness may be compromised [15, 17, 18]. These observations that neither homeostatic nor circadian deficits explain the narcolepsy sleep phenotype, but rather that the sleep-related symptoms of narcolepsy are closely correlated with cataplexy, led to the current view of narcolepsy as a disorder of arousal state instability caused by a loss of arousal state boundary control rather than simply as a REM sleep disorder [19].

This refined understanding of the narcolepsy phenotype led to a fine-tuning of the criteria used to diagnose narcolepsy based largely on the Multiple Sleep Latency test to systematically identify and quantify the presence of SOREMs and EDS [20, 21]. With the development of more precise diagnostic criteria, narcolepsy was recognized as having a spectrum of severity, eventually leading to the identification of two types, with both subtypes being characterized by a combination of the sleep and behavioral symptoms mentioned above but with NT1 being distinguished from NT2 by the presence of cataplexy [22, 23].

In contrast to studies in humans, non-human models of neurological conditions such as narcolepsy allow the application of invasive techniques in structured cohorts, thereby allowing for more precise and systematic experimental analyses. In this review, we first provide a broad description of some of the early animal models of narcolepsy and their contribution to our understanding of the narcolepsy phenotype, as well as to our understanding of arousal state control more generally. We then provide a description of recently developed models of narcolepsy (Table 1) and other transgenic models that have enabled investigation of the neural circuitry of the hypocretin/orexin system, which is now known to be dysfunctional in narcolepsy (Table 2). Finally, we discuss the strengths, weaknesses, and overall utility of these animal models and speculate on the future direction of new animal models of narcolepsy that could facilitate advances in our understanding of this fascinating disorder and aid in teasing apart the contribution of various neural circuits involved in controlling sleep states, thereby furthering our understanding of the regulatory mechanisms and underlying functions of these states.

Table 1.

Narcoleptic animal models and the associated phenotype of each model

Strain	Mutation	Phenotype	References
Canarc-1 mutant canines	Mutation in Hcrtr2 gene	Sleep/wake fragmentation; SOREMs; low daytime activity; increased time spent in drowsy state; cataplexy	[29, 31, 36, 67]
Orexin -/- (Hcrt -/-) mice	Constitutively absent Hcrt peptide precursor gene	Mild sleep/wake fragmentation; SOREMs; decreased wakefulness during dark phase; increased REM; frequent cataplexy; obesity	[64]
Orexin/ataxin-3 mice	Progressive postnatal ablation of Hcrt neurons	Sleep/wake fragmentation; SOREMs; increase in total REM time; decreased wakefulness during dark phase; cataplexy; obesity	[82]
Orexin/ataxin-3 rat	Progressive postnatal ablation of Hcrt neurons	Sleep/wake fragmentation; SOREMs; increase in total REM time; decreased wakefulness during dark phase; cataplexy	[84]
Hcrtr1 -/- (OX1R-/-) mice	Constitutively absent HcrtR1 (OX1R)	Mild sleep/wake fragmentation	[89]
Hcrtr2 -/- (OX2R-/-) mice	Constitutively absent HcrtR2 (OX2R)	Mild sleep/wake fragmentation; infrequent cataplexy	[90]
Hcrtr1 -/- ; Hcrtr2 -/- (OX1R-/-; OX2R-/--) mice	Constitutively absent HcrtR1 and HcrtR2	Sleep/wake fragmentation; SOREMs; increase REM; decreased amount of wakefulness during dark phase; cataplexy	[88, 91]
OX 2 R/Hcrtr2 TD mice	LoxP-mediated disruption of OX2R/ Hcrtr2 expression	Sleep/wake fragmentation; decreased amount and duration of wakefulness during dark phase; infrequent cataplexy	[92]
OX-tTA mice	Expresses tetracycline transactivator within Hcrt neurons	None; useful with KENGE-tet expression system	[109]
OX-tTA;TetO-DTA mice	Allows conditional ablation of Hcrt neurons	Severe sleep/wake fragmentation; SOREMS; increased REM; decreased wakefulness during dark phase; pronounced cataplexy; obesity	[95]
OX-tTA;TetO Htr1a (and OX-tTA;OX-EGFP;TetO Htr1a) mice	Conditionally and reversibly overexpresses 5HT1A in Hcrt neurons	Severe sleep/wake fragmentation during dark period	[164]
O/E3 -/- mice	Constitutively absent O/E3 transcription factor; decreased number of Hcrt neurons	Sleep/wake fragmentation; SOREMs; increased REM; decrease in wakefulness during dark phase; cataplexy	[113]
H1N1 infection in Rag1-/- mice	Mice lacking T and B cells; H1N1 infection causes Hcrt neuronal ablation	Sleep/wake fragmentation; SOREMs; cataplexy	[122]
OX-HA mice	Expresses hemagglutinin (HA) as a neo-self-antigen in Hcrt neurons	Orexin-HA mice transfected with CD8 T cells exhibit a narcolepsy-like phenotype; Cataplexy and sleep attacks	[124]
Pmch-tTA;TetO-DTA mice	Allows conditional ablation of MCH neurons	Decreased NREM and increased wake	[97]
OX-tTA;Pmch-tTA;TetO-DTA mice	Allows conditional ablation of MCH and Hcrt neurons	Increased wake; decreased NREM and REM; severe cataplexy; delta-theta (DT) sleep	[143]

Table 2.

Mouse models allowing optogenetic manipulation of their Hcrt neurons

Strain	Mutation	Phenotype	References
Lentivirus-mediated expression of Channelrhodopsin (ChR2) in mouse Hcrt neurons	Expresses ChR2 in Hcrt Neurons	Photoillumination (PI) activates Hcrt neurons, which results in increased sleep/wake transitions	[101–103]
OX-tTA;TetO ChR2 (C128S) mice	Conditionally expresses ChR2 in Hcrt neurons	PI activates Hcrt neurons	[93, 104]
OX/Halo mice	Constitutively express Halorhodopsin in Hcrt neurons	PI inhibits Hcrt neurons; silencing induces synchronized NREM-like EEG when applied during light phase	[107]
OX/Arch mice	Constitutively express Archaerhodopsin-3 (Arch) in Hcrt neurons	Mice with high Arch expression (aHE) exhibit reduced Hcrt neuron excitability; increased sleep/wake transitions; increased REM; reduced sleep latency; altered EEG spectral distribution; obesity	[108]
OX-tTA;TetO-ArchT mice	Conditionally express ArchT in Hcrt neurons	PI inhibits Hcrt neurons; inhibition during dark phase leads to increase in NREM and increase in sleep/wake transitions	[109]

Narcolepsy in Non-Humans

Naturally occurring narcolepsy phenotypes in animals

Narcolepsy-like symptomatology has been identified in a number of mammals including horses [24–26], a Brahman bull [27], sheep [28], and a cat [29]. However, the initial report of narcolepsy in a Dachshund [29] led William Dement to establish a colony of narcoleptic dogs at Stanford University that, at different times, included Doberman pinschers, Labrador retrievers, miniature poodles, dachshunds, beagles, and Saint Bernards [30, 31]. While these species exhibited similar symptoms, the severity and developmental sequence of symptom occurrence varied significantly among dog breeds [30].

Canine narcolepsy

Early descriptions of the narcolepsy phenotype in canines documented the occurrence of normal sleep stages and distribution, the presence of SOREMs, and both partial and full cataplectic attacks that could be induced by the presentation of positive emotional stimuli such as food or toys [29–31]. While cataplexy was unequivocally present in canine narcoleptics, the initial reports of excessive sleepiness in narcoleptic dogs were somewhat controversial given baseline canine sleep patterns. However, polysomnographic studies determined that, in addition to exhibiting SOREMs, narcoleptic canines spend more time in a drowsy state [32, 33], exhibit lower daytime activity and sleep fragmentation [32–34], and have a shorter sleep latency and a greater number of SOREMs when presented with a homeostatic challenge [35].

Breeding of canine narcoleptics revealed that cataplexy occurred in both sporadic and familial forms [36]. The familial form present in large breed dogs was transmitted as an autosomal recessive gene [36] that was later called canarc-1 [37]. While the transmission of narcolepsy in several other breeds proved to be more complex, the genetic transmission of canine narcolepsy in large breed dogs enabled Dement et al. to establish a narcoleptic canine colony at Stanford University. The development of this colony enabled systematic pharmacological, neurochemical, electrophysiological, and genetic studies of this disorder as well as in vivo testing of potential narcolepsy therapeutics [38]. The overall conclusion from these studies was that canine narcolepsy/cataplexy was the result of a neurochemical imbalance in which monoaminergic tone was reduced (monoaminergic hypoactivity) and brainstem cholinergic tone was increased (cholinergic hypersensitivity) [39].

Canarc-1: an inherited mutation in the hypocretin receptor 2 (OX₂R) gene

A major advance in the understanding of human narcolepsy resulted from the discovery that many narcolepsy patients were positive for the major histocompatibility complex (MHC) class II human leukocyte antigen (HLA) DR2 [40]. Subsequent studies identified a strong association with class II polymorphisms in the DQB1*06:02 and DQA1*01:02 loci of HLA genes: 82%–99% of humans with narcolepsy had the DQB1*06:02 allele compared to just 12%–38% in the general population [40–45]. These results suggested the possibility of an autoimmune mechanism underlying human narcolepsy and influenced the search for the gene responsible for the inherited form of canine narcolepsy. Although initial studies failed to find a link between MHC class II dog leukocyte antigens and canarc-1 [37], through the construction of a bacterial artificial chromosome (BAC) library of a backcrossed Doberman pinscher heterozygous for canarc-1, researchers were ultimately able to identify a mutation in the Hypocretin-2 receptor as the defect underlying canine narcolepsy [46].

Hypocretin and Narcolepsy

Cloning of the hypocretin/orexin gene

In 1998, a cDNA isolated from the rat hypothalamus by subtractive hybridization was predicted to code for two structurally similar excitatory neuropeptides [47]. Because the predicted neuropeptides were expressed in the hypothalamus and had similarities to secretin, these peptides were named hypocretin-1 (Hcrt1) and hypocretin-2 (Hcrt2). Just weeks later, another group identified the same two neuropeptides and determined them to be the endogenous ligands for two orphan G protein-coupled receptors [48]. Since intracerebroventricular (ICV) administration of these neuropeptides into the lateral ventricle caused rats to increase their food intake, the second group named the two neuropeptides orexin-A and orexin-B and their associated receptors orexin receptor-1 (OX₁R/HcrtR1) and orexin receptor-2 (OX₂R/HcrtR2) after the Greek word “orexis” which means appetite [48]. OX₁R/HcrtR1 and OX₂R/HcrtR2 are G protein-coupled receptors and, while OX₁R/HcrtR1 shows a selective binding affinity for orexin-A/Hcrt1, OX₂R/HcrtR2 has an equal affinity for both orexin-A/Hcrt1 and orexin-B/Hcrt2 [48].

Hcrt-positive neurons are restricted to the perifornical hypothalamus (PFH) and lateral hypothalamic area (LHA) but have widespread projections to brain regions that differentially express HcrtR1 and HcrtR2, including dense projections to the noradrenergic areas of the locus coeruleus (LC), the nucleus accumbens, serotonergic areas of the dorsal raphe nuclei (DRN), the amygdala, the suprachiasmatic nucleus, basal forebrain, the cholinergic brainstem, and the spinal cord (Figure 1) [48–53]. The widespread projections of the Hcrt cells to brain regions known to be involved in arousal state control provides the anatomical substrate for their involvement in the regulation of the sleep/wake cycle.

Figure 1.

Distribution of hypocretin/orexin and its receptors. Hypocretin (Hcrt)-containing neurons are confined to the perifornical hypothalamus and lateral hypothalamic area. These neurons project to widespread regions of the brain where the Hcrt receptors-1 and -2 (HcrtR1 and HcrtR2) are differentially expressed. HcrtR1-expressing regions are indicated in yellow, HcrtR2 in magenta, and regions where both receptors are expressed in turquoise. HcrtR1 is expressed in the cingulate cortex, the central area of the amygdala (CeA) and the locus coeruleus (LC). Hcrt2R is expressed in the layer VI of the cerebral cortex, the nucleus accumbens (NAc), substantia nigra (SN), ventral tegmental area (VTA), and the tuberomammillary nucleus (TMN). Both receptors are expressed in most regions of the cerebral cortex, the basal forebrain (BF), the bed nucleus of stria terminalis (BNST), most regions of the hypothalamus (Hy), the ventrolateral periaqueductal gray (vlPAG), dorsal part of the deep mesencephalic nucleus (dDPMe), the dorsal raphe (DR), laterodorsal tegmental nucleus (LDT), and the pedunculopontine tegmental nucleus (PPT).

The Hcrt/orexin system and its role in sleep/wake regulation

The hypothalamus was first recognized as an integral part of the wake-promoting brain circuitry nearly 100 years ago, when von Economo observed that patients afflicted with encephalitis lethargica, a disease-causing extreme sleepiness, frequently had lesions in their posterior hypothalamus [54]. At the time of the cloning of Hcrt, it was understood that arousal state regulation occurs through the interaction of several anatomically distributed neuronal populations. Monoaminergic neurons in the brainstem had been determined to promote wakefulness, contribute to the desynchronization of cortical neuronal activity, and inhibit REM-active pontine neurons [55–58]. Furthermore, cholinergic neurons in the basal forebrain and pons that were active during both wake and REM sleep were thought to mediate forebrain desynchronization [59–61]. Because the Hcrt-containing neurons innervated all of these neuronal populations, it was proposed that essential parts of the wake-promoting brain circuitry may reside in the hypothalamus and that the Hcrt-producing neurons were a critical part of this circuitry [49, 62, 63]. Canine narcoleptics, whose phenotype is due to the absence of a functional HcrtR2, further hinted at the importance of this system in sleep/wake control [46].

Murine Models of Narcolepsy

Early murine models of narcolepsy

Hcrt/orexin peptide knockout mice

The first mouse model of narcolepsy, the prepro-orexin null mutant mouse, was described the year after the Hcrt gene was cloned [64]. Open field studies in these mice revealed that they exhibited an unusually high amount of inactivity during the dark period, the active period for nocturnal rodents. Follow-up investigation revealed this to be the result of “sleep attacks,” which were remarkably similar to cataplectic attacks seen in narcoleptic canines and humans with narcolepsy, but the authors refrained from calling them cataplexy because they were unable to demonstrate the maintenance of consciousness during these episodes as occurs during human and canine cataplectic attacks [64, 65]. Nonetheless, EEG/EMG studies of prepro-orexin knockout (KO) mice [64] documented other characteristics present in both humans and canines with narcolepsy including SOREMs [66, 67], sleep/active cycle fragmentation [66, 68], increased active period sleepiness [66, 68–70], and gait ataxia [64, 65]. As the first rodent model of narcolepsy, the prepro-orexin KO mouse was invaluable as it demonstrated that a complex behavior such as sleep could be affected by the elimination of a single gene and that a heretofore mysterious sleep disorder could be “modelled” in rodents in which extensive neuroanatomical and neurochemical information existed.

This first study of mice in which the Hcrt system was genetically manipulated was published concomitantly with the description of canarc-1 as a mutation in the Hcrtr2 gene in narcoleptic Doberman pinschers and Labrador retrievers [46]. Together, these studies demonstrated that disruption of the Hcrt system, whether pre-synaptically in the mouse or post-synaptically as in the dog, resulted in a narcolepsy-like phenotype, thereby implicating Hcrt deficiency in the etiology of narcolepsy. Since EDS and sleep fragmentation are diagnostic symptoms of narcolepsy, these results also directly implicated the Hcrt neurons as an important part of the endogenous sleep/wake regulatory system. This recognition led to an explosion of research on the role of the Hcrt system in both sleep/wake regulation and narcolepsy, as well as increased interest in the generation of other rodent models with manipulations of the Hcrt system.

Human narcolepsy: a different etiology than canine narcolepsy or prepro-orexin KO mice

The implication of Hcrt system dysfunction in the phenotype of narcoleptic canines and prepro-orexin KO mice led to the investigation of the role of this system in human narcolepsy. The first direct indication of a link between the Hcrt system and human narcolepsy was that the Hcrt-1/orexin-A peptide was undetectable in the CSF in the vast majority of patients with narcolepsy [71], an observation that ultimately led to a new criterion to aid in the diagnosis of narcolepsy and the differentiation between NT1 and NT2. Most human NT1 patients were found to exhibit reduced Hcrt1 levels in their cerebrospinal fluid (CSF) [71–73] as also occurs in some sporadic cases of canine narcolepsy [74]. While low Hcrt1 CSF levels are very commonly reported in HLA-positive NT1 patients, individuals diagnosed with NT2 and other sleep or neurological disorders exhibit only mildly reduced or normal CSF levels of Hcrt1 [72, 73]. Although it was unclear at the time whether the absence of the peptide in the CSF was due to a problem with synthesis or release of the peptide by the Hcrt neurons, subsequent postmortem examination revealed a massive reduction in the number of Hcrt neurons in the brains of human narcolepsy patients [75, 76]. A crucial subsequent study [77] demonstrated that the neuropeptide prodynorphin and the pentraxin NARP are colocalized in Hcrt neurons in the human hypothalamus but that, in human narcolepsy, the number of cells expressing these markers is reduced to 5%–10% of the normal level. Examination of other cell types within the tuberal hypothalamus supported the concept of a specific degeneration of the Hcrt neurons, since the coextensive but not colocalized melanin-concentrating hormone (MCH) neurons were spared. Thus, after 120 years as a disorder with unknown etiology, the convergence of information from studies conducted in rodents, canines, and humans strongly implicated hypocretin deficiency as the likely cause of human narcolepsy.

Orexin/ataxin-3 mice: Hcrt neuron ablation produces a narcolepsy phenotype

In prepro-orexin KO mice, the Hcrt neuropeptides are absent but the neurons themselves are intact, leaving the possibility that other neuroactive substances expressed by the Hcrt neurons could still be released by these cells and affect the organismal activity or physiological functions [78–81]. Consequently, the next mouse model in which the Hcrt system was manipulated, the orexin/ataxin-3 mouse [82], represented an improvement over the prepro-orexin KO mouse as a narcolepsy model as it recapitulated the cell loss that occurs in human narcolepsy [75, 76]. In orexin/ataxin-3 (ATAX) mice, Hcrt-containing cells are selectively ablated through the targeted expression of the toxic Ataxin-3 transgene, which causes apoptosis, under the control of the HCRT promoter. The ablation of the Hcrt neurons is progressive during development in ATAX mice, which could mitigate the compensation that can occur in constitutive KO models such as the prepro-orexin KO mice. This progressive degeneration of the Hcrt field is also more representative of narcolepsy pathology in humans, as the onset of narcolepsy symptoms does not occur typically until post-puberty [83]. As the Hcrt neurons degenerate, ATAX mice begin to exhibit narcolepsy symptoms such as sleep fragmentation, SOREMs, obesity despite reduced food intake, and cataplexy [82]. However, cataplectic attacks in orexin/ataxin-3 mice are highly variable between individuals, occurring between 10 and 30 times in a 4-h recording period and lasting anywhere between a few seconds to several minutes [82]. The same transgene was used to develop an ATAX rat model [84], whose larger size facilitates the application of techniques that are not feasible in mice [85].

Hcrt receptor KOs produce different narcolepsy phenotypes

As indicated above, disruption of the Hcrt system, whether pre-synaptically in the prepro-orexin KO mouse or post-synaptically as in the dog, results in a narcolepsy phenotype. Thus, genetic manipulation of the Hcrt receptors has been another approach to generate narcolepsy models. The development of single KOs for each Hcrt receptor and a double KO of both Hcrt receptors allowed researchers to dissect the contribution of each receptor to the narcolepsy phenotype and, more generally, to sleep/wake control. The behavioral and sleep phenotypes of HcrtR1 and HcrtR2 KO mice were characterized by vigilance state disturbance and sleep fragmentation, albeit the disruption was much less severe in HcrtR1 KO mice than that observed in either the prepro-orexin or the HcrtR2 KO strains [86]. Furthermore, the HcrtR1 KO did not exhibit cataplectic events while the HcrtR2 KO did, but less frequently than observed in prepro-orexin KO or ATAX mice. Lastly, neither receptor KO exhibited SOREMs [87–89]. A systematic comparison of the narcolepsy phenotype in prepro-orexin and HcrtR2 KO mice also confirmed a less severe narcolepsy phenotype of HcrtR2 KO mice compared to prepro-orexin KO and ATAX mice and concluded that normal regulation of wake/non-REM sleep transitions depends critically upon activation of HcrtR2 [90]. Dual Hcrt/orexin receptor KO (OX₁R^−/−; OX₂R^−/−) mice exhibited a robust narcolepsy phenotype characterized by the frequent occurrence of cataplectic attacks, SOREMs and pronounced sleep disturbance that was more severe than the OX₂R^−/− KO alone and more similar to that of prepro-orexin KO and ATAX mice, thus suggesting a role for OX₁R signaling [88, 91]. These results imply that, while HcrtR2 may be more involved in the expression of narcolepsy symptoms, both receptors have some overlapping, perhaps reciprocal, function in generating narcolepsy symptoms and that the contribution of HcrtR1 to the narcolepsy phenotype should not be overlooked.

Conditional models of narcolepsy

The first conditional mouse model of narcolepsy utilized a transcription-disrupter strategy in which a loxP-flanked gene cassette was used to disrupt the production of the OX2R/HCRTR2 [92]. The absence of OX2R signaling in this strain was verified by the absence of a cellular electrophysiological response of tuberomammillary nucleus (TMN) neurons to orexin-A application. OX2R loxP-flanked transcription-disrupter (TD) mice had poor maintenance of wakefulness, indicative of sleepiness and fragmented sleep, but infrequent bouts of cataplexy. Both of these defects were completely recovered by crossing them with mice that globally deleted the transcription-disrupter cassette by expression of Cre in the female germline. Focal restoration of OX2R in TMN neurons and the adjacent posterior hypothalamus using a Cre-dependent AAV rescued the sleepiness but not the fragmented sleep of these mice, demonstrating that the TMN plays an essential role in the wake-promoting effects of orexins, but also indicating that these peptides stabilize sleep through other targets.

The KENGE-tet system and OX-tTA transgenic mice

The next major advance in narcolepsy models was the development of an inducible model based on the tetracycline-controlled transactivator (tTA) tetracycline-off system [93, 94]. The “Tet-on/Tet-off” or the knockin-mediated enhanced gene expression by improved tetracycline-controlled gene induction (KENGE-tet) system enables reliable gene expression control in targeted cell types (Figure 2A and B) through knock-in (KI) targeting transgene insertion just downstream of the β-actin gene polyadenylation signal of the tetO cassette. This system allows inducible control of transgene expression in cells expressing tTA which, in the case of the orexin-tTA (OX-tTA) mice, are the Hcrt/orexin-containing cells. Expression can be controlled through the addition or removal of doxycycline (DOX) from the animal’s diet (Figure 2A and B). OX-tTA mice have allowed the application of the versatile KENGE-tet system for the development of several inducible bigenic strains including OX-tTA;TetO-ChR2(C128S)-EYFP [95], OX-tTA;TetO-Archaerhodopsin [96] and MCH-tTA;TetO ArchT [97] mice, as well as others as described below.

Figure 2.

(A) Breeding scheme to create bigenic mice. Orexin-tTA (pOX) mice or MCH-tTA (pMCH) are crossed with TetO-DTA mice to create Orexin-tTA;TetO-DTA or MCH-tTA;TetO-DTA mice, respectively. (B) Induction of neuron-specific cell death using the tet-off system. In the presence of doxycycline (Dox+ condition), Dox binds to the tetracycline transactivator (tTA) which prevents tTA from binding to TetO in the promoter region of the diphtheria toxin A (DTA) gene, DTA expression is repressed, and orexin and MCH neurons remain intact. Conversely, in the absence of DOX (Dox− condition), tTA can bind TetO region of the promoter, resulting in DTA expression and apoptosis of orexin neurons in Orexin-tTA;TetO DTA mice or MCH neurons in MCH-tTA;TetO-DTA mice.

Bigenic OX-tTA;TetO-DTA mice

Breeding OX-tTA mice with TetO-DTA mice yields bigenic OX-tTA;TetO-DTA mice in which the TetO-DTA transgene occurs specifically in Hcrt-containing neurons. In this bigenic line, diphtheria toxin A (DTA), a substance known to induce cellular apoptosis through inhibition of protein synthesis, can be conditionally expressed in Hcrt-containing neurons under the control of the KENGE-tet system (Figure 2A and B) [95]. While maintained on DOX chow, the Hcrt cells remain intact but, upon removal of DOX from the diet, ablation of the Hcrt neuronal field progresses swiftly, with ablation of >85% of the neurons occurring within 7 days. As the Hcrt neurons degenerate, narcolepsy symptoms begin to appear, with sleep/wake fragmentation appearing at 1 week and cataplexy at 2 weeks after removal of dietary DOX. After 11 weeks, OX-tTA;TetO-DTA mice exhibited a robust narcolepsy phenotype with 56.2 ± 7.7 cataplectic events occurring during the dark period and 15.0 ± 2.2 bouts during the light period. Sleep/wake transitions during the dark period were greatly increased, while transitions to REM were greatly decreased, and the average number of Hcrt neurons remaining was just 7 ± 1 cells [95]. Compared to ATAX mice, OX-tTA;TetO-DTA mice exhibited more extensive Hcrt field ablation and a more pronounced narcolepsy phenotype. While this stronger phenotype caused by the near total ablation of the Hcrt neuronal field obviously offers a more robust system to test the efficacy of current [98] and novel [99] therapeutics for narcolepsy than other models, the real advantage of the OX-tTA;TetO-DTA model lies in the experimenter’s ability to control the timing and duration of neuronal ablation.

Optogenetic models of the Hcrt/orexin system

Lentivirus-mediated expression of Channelrhodopsin (ChR2) in the Hcrt neurons

Channelrhodopsins are blue light-gated ion channels whose stimulation results in the depolarization and, thus, activation of cells expressing these proteins [100]. Among these, Channelrhodopsin-2 (ChR2) from Chlamydomonas reinhardtii is commonly used in optogenetic studies. Lentiviral injections of Hcrt::ChR2-mCherry into the hypothalamus produced mice in which the Hcrt cells expressed ChR2 labeled with mCherry, allowing the visualization of cells expressing ChR2 and optogenetic activation of those cells [101, 102]. Optogenetic stimulation of Hcrt neurons expressing ChR2 with 5–30 Hz blue light pulses reduced the latency in sleep-to-wake transitions [101, 103]. A subsequent study demonstrated that photoinhibition of downstream Hcrt projections in the LC blocked the Hcrt neuron-mediated increase in sleep-to-wake transitions, and that concomitant stimulation of both Hcrt and LC neurons resulted in a further increase in the probability of sleep-to-wake transitions than that induced solely by photostimulation of the Hcrt neurons [102].

OX-tTA;TetO-ChR2(C128S) transgenic mice

Appropriate expression of light-sensitive proteins such as ChR2 via viral delivery depends upon the accuracy of stereotaxic injections, efficacy of viral penetration into the relevant neuronal population, and subsequent gene transcription and translation of the protein. In contrast, transgenic mice can provide stable expression which may enable more reproducible results. Several inducible models have been produced that express light-sensitive proteins in the Hcrt cells using the OX-tTA transgenic line discussed above [93, 95]. For example, bigenic OX-tTA;TetO-ChR2(C128S) mice allow the expression of ChR2 specifically in Hcrt cells. Optogenetic studies using bilateral blue light excitation of Hcrt neurons (flanked by yellow pulses to close the channel) in Orexin-tTA;TetO-Chr2(C128S) mice was shown to induce awakening within 4 s after stimulation in over 75% of the trials whereas the probability of an NREM-Wake transition was only 20% for control trials in which yellow light pulses were flanked by yellow pulses [104]. In contrast with previous studies [101, 103] in which unilateral stimulation of Hcrt neurons changed the probability of NREM-Wake transitions over the subsequent 60 s, these results demonstrate that direct excitation of large populations of Hcrt neurons can induce wakefulness with a very short latency.

OX/Halorhodopsin transgenic mice

Halorhodopsin (HaloR) is an orange light-gated chloride pump whose stimulation results in an influx of chloride ions, thereby causing hyperpolarization and neuronal inhibition [105, 106]. HaloR expression was induced in the Hcrt neurons of mice through a LacZ-mediated gene substitution under control of the prepro-orexin promoter [107]. Optogenetic silencing of Hcrt neurons resulted in a reduced neuronal discharge rate of these cells as well as in cells in the DRN, a projection site of Hcrt neurons. In vivo optogenetic silencing of Hcrt neurons in OX/Halo transgenic mice during the light period induced a synchronized EEG indicative of NREM sleep but had no discernible effect when applied early during the dark period, the normal waking period in nocturnal mice. Interestingly, serotonergic neurons in the dorsal raphe of ATAX mice exhibited a normal discharge rate, suggesting that compensatory regulation occurs in the chronic absence of input from Hcrt neurons [107].

OX/Archaerhodopsin-3 transgenic mice

In Orexin/Archaerhodopsin-3 (OX/Arch) transgenic mice, a modified Archaerhodopsin-3 (Arch) gene was inserted downstream of the prepro-orexin promoter, resulting in expression of the yellow light-sensitive Arch proton pump specifically within Hcrt neurons [108]. Illumination of Arch-expressing Hcrt neurons with yellow light results in a hyperpolarized membrane potential and cellular inhibition. Elevated expression of the Arch transgene affected cellular and physiological parameters of Hcrt neurons independent of photoillumination. The excitability of Hcrt neurons was reduced and both circadian and metabolic parameters were perturbed in a subset of OX/Arch mice that exhibited high levels of Arch expression. OX/Arch mice also had increased REM sleep under baseline conditions but did not exhibit cataplexy. These aberrations resembled some aspects of mouse models with Hcrt neuron ablation, yet the number of Hcrt neurons in OX/Arch mice was not reduced. Thus, OX/Arch mice may be useful to investigate Hcrt system dysfunction when these neurons are intact, as may occur in some NT2 cases (see below).

OX-tTA;TetO-Archaerhodopsin mice

Archaerhodopsin-TP009 (ArchT) is a light-activated proton pump whose stimulation results in a larger current than Arch resulting in a hyperpolarized membrane potential and reduced firing rate of ArchT-expressing cells [96]. In mice expressing ArchT in Hcrt-containing neurons using the KENGE-tet system, application of green light to the Hcrt neuronal field resulted in inhibition of these neurons and induced a time-of-day-dependent effect on the sleep/wake phenotype [109]. Silencing of the Hcrt-containing neurons during the daytime, the major sleep period for rodents, had little effect on EEG activity or sleep/wake patterns whereas 1-h silencing of the Hcrt neurons during the dark period led to increased NREM sleep and increased the number of sleep/wake transitions. These results are consistent with the idea that Hcrt neurons are involved in maintaining wakefulness [107–109].

Conclusions from optogenetic manipulation of the Hcrt system

Optogenetics allows the instantaneous, short-term activation or inhibition of targeted brain regions and cell types. Various optogenetic models allowing the manipulation of Hcrt neurons have been generated utilizing several different methods. Not only does optogenetic manipulation of the Hcrt system allow the dissection of its contribution to arousal circuits through reversible, short-term activation and inhibition of these circuits but, importantly in the context of understanding narcolepsy, it also allows the assessment of the effects of early Hcrt neuron loss prior to the network reorganization that likely occurs in animals chronically lacking these cells, thus providing insight into the initial stages of narcolepsy onset. The earliest reported symptoms of narcolepsy in humans are sleep disruption and EDS, which are mirrored by the increase in sleep/wake transitions and increased NREM sleep during the major wake period in mice whose Hcrt-containing neurons have been optogenetically inhibited [109].

Other rodent narcolepsy models

O/E3 transcription factor null mutant mice

O/E3 is a helix-loop-helix transcription factor that is involved in the regulation of neuronal differentiation in the hypothalamus, spinal cord, cerebellum, adipose and bone tissue. O/E3 KO mice exhibit a wide range of anatomical defects and alterations including the defective migration of gonadotropin-releasing hormone synthesizing neurons from the olfactory placode to the hypothalamus, a reduction in bone mass, Purkinje cell death in the cerebellum, as well as infertility, dwarfism, and gait ataxia [110, 111]. O/E3 is expressed in Hcrt neurons and is downregulated in ATAX mice compared to wild type (WT) mice [112].

O/E3 KO mice exhibit a loss of Hcrt neurons and hypocretinergic innervation of brain structures involved in sleep regulation [113]. O/E3 KO mice have disrupted sleep patterns and SOREMs as occurs in people with narcolepsy. The narcolepsy-like phenotype present in the O/E3 KO mice is reversed by ICV injections of Hcrt1, indicating that the loss of the Hcrt neurons and no other neuronal deficiencies cause the narcolepsy-like symptoms observed in this strain. MCH neurons, which are co-extensive but not colocalized with Hcrt neurons, are intact in the O/E3 KO mice. Together, these observations indicate that O/E3 is an important transcription factor in the development and differentiation of the Hcrt neuronal field [113].

Immune-mediated models of narcolepsy

Genetic, environmental, and serological data suggest that an autoimmune mechanism may contribute to the etiology of narcolepsy. As previously mentioned, NT1 is strongly linked to the presence of HLA-DQB1:*06:02 and HLA-DQA1:*01:02 alleles and many autoimmune diseases are associated with HLA genes [114]. HLA genes encode proteins that present antigens to T cells and, through this interaction, can elicit a specific immune response against cells producing those presented antigens. T-cell receptor α chain locus polymorphisms have also been linked to the occurrence of narcolepsy [115, 116]. Serological studies of CSF from narcolepsy patients identified the presence of antigenic antibodies targeting neuronal proteins [117–119]. In addition, an increase in the occurrence of narcolepsy was observed following the 2009–2011 H1N1 infection in China and several Scandinavian countries following widespread vaccination programs during this outbreak [119–121]. The increased incidence of vaccination-related narcolepsy was later linked to the use of the adjuvant Pandemrix in vaccines used in northern European countries, whereas other adjuvanted pH1N1 vaccines did not affect narcolepsy incidence.

While these observations collectively suggest a link between the immune system and narcolepsy, whether the immune system is directly involved in the degeneration of Hcrt-containing neurons resulting in narcolepsy remained unclear until several recent experiments. In transgenic mice lacking B and T cells, influenza A infection targets neuronal populations involved in sleep/wake regulation, including Hcrt and MCH neurons. These mice exhibited SOREMs, sleep fragmentation as indicated by an increase in sleep/wake transitions during the major sleep phase, and the presence of “cataplexy-like” incidences of behavioral arrest [122, 123]. A mouse line in which hemagglutinin (HA) was expressed as a neo-self-antigen in Hcrt neurons (OX-HA mice) was transfected with neo-self-antigen-specific T cells and the behavioral and sleep phenotype was assessed [124]. Transfer of cytotoxic CD8 T cells into OX-HA mice resulted in the targeted ablation of Hcrt-containing neurons as well as cataplexy and sleep attacks [124]. Most importantly, people with narcolepsy exhibited CD4⁺ T cells that showed a substantial immune reaction to circulating Hcrt fragments in their CSF [125] as well as CD8⁺ Hcrt-recognizing T cells in blood [126], strongly suggesting these antigens may be mediating an autoimmune response that targets the Hcrt neurons.

Melanin-concentrating hormone neurons, sleep, and narcolepsy

Background

MCH is a 19 amino acid neuropeptide is cleaved from the same prepro-peptide precursor (ppMCH) as two other neuropeptides, neuropeptide EI (NEI), and neuropeptide GE (NGE) [127]. MCH is the endogenous ligand for two G-protein coupled receptors (MCHR1 and MCHR2) in the brains of humans and other primates, although only MCHR1 is expressed in rodents. Activation of MCHR1 and MCHR2 causes decreases in cyclic adenosine monophosphate (cAMP) levels [128–133]. In the mammalian brain, MCH-producing neurons are co-extensive, but not colocalized, with Hcrt-producing neurons in the LHA, but extend more rostrocaudally into the incertohypothalamic areas than Hcrt-producing neurons [49, 127, 134–136]. In a manner similar to the Hcrt cells, MCH neurons have widespread projections, innervating both cholinergic and monoaminergic arousal centers such as the locus coeruleus [127].

Manipulation of MCH neurons influences sleep and sleep-related memory retention

ICV administration of MCH induces dose-dependent changes in sleep, causing increases in both REM and NREM sleep [137]. The transcription factor c-FOS is greatly elevated in MCH neurons during recovery from REM sleep deprivation [137] and both cellular electrophysiological [138] and calcium imaging [139] studies indicate that MCH neurons have increased activity during REM sleep. Optogenetic activation of MCH neurons mediated through viral injections of ChR2 expressed downstream of the MCH promoter increased sleep even in the presence of strong waking drives [140]. Optogenetic activation of MCH neurons through the Cre-dependent ChR2 variant ChETA led to a prolongation of REM events when photostimulation was applied at REM onset without affecting NREM duration when stimulation occurred at NREM onset [141]. Furthermore, inhibition of MCH neurons through stimulation of Cre-dependent ArchT decreased the frequency and amplitude of the hippocampal theta rhythm normally associated with REM sleep without affecting the duration of REM sleep [141]. In contrast to Hcrt neurons which have a strong wake-promoting effect, these results collectively suggest that MCH neurons are a sleep-promoting neuronal population.

The KENGE-tet system was used to generate bigenic mouse lines that allowed the expression of ChR2 and ArchT specifically in the MCH neurons and conditional ablation of MCH-containing neurons (Figure 2) [97]. In mice expressing ChR2 in their MCH-containing neurons (MCH-tTA;TetO ChR2 mice), optogenetic activation of MCH neurons led to an increase in NREM-to-REM transitions, as well as increased time spent in REM sleep at the expense of time spent in NREM [97]. However, optogenetic inhibition of MCH-containing neurons in mice expressing ArchT (MCH-tTA;TetO ArchT mice) showed no changes in sleep/wake phenotype following acute silencing of these neurons [97]. Lastly, conditional ablation of MCH neurons in MCH-tTA;TetO-DTA mice resulted in partial insomnia, suggesting that MCH neurons indeed play a role in normal sleep/wake control.

Since MCH neurons were found to densely innervate the hippocampus, MCH-tTA;TetO-DTA mice were subjected to the hippocampus-dependent object recognition (NOR) test before and after MCH neuron ablation [142]. Surprisingly, mice in which MCH neurons were ablated showed significant memory improvement in the NOR compared with control mice with intact MCH neurons, but this improvement required to sleep during the retention period. On the other hand, performance on the contextual fear conditioning (CFC) test (which is not hippocampal-dependent) was not improved. Chemogenetic activation of MCH neurons and optogenetic activation of MCH terminals innervating the hippocampus prevented memory retention, whereas chemogenetic inactivation of MCH neurons facilitated memory retention. Since MCH neurons are active during REM sleep and because REM sleep has been implicated in learning and memory, MCH neurons were optogenetically inhibited specifically during either wake, NREM or REM sleep and only MCH neuron inhibition during REM sleep facilitated memory retention [142]. These results suggest that MCH neuron activation during REM sleep impairs memory retention, which may be why it is so difficult to remember dreams.

Dual ablation of Hcrt and MCH neurons exacerbates narcolepsy symptomatology

To determine whether Hcrt neurons and MCH neurons interact to regulate sleep/wake state, trigenic OX-tTA;MCH-tTA;TetO-DTA (OXMC) mice were generated [143]. In these mice, both Hcrt and MCH neurons were ablated when DOX was removed from the diet. Hcrt neuron-ablated mice exhibited a narcolepsy phenotype including sleep/wake fragmentation and cataplexy as described previously [95]. However, OXMC mice showed a much more severe narcolepsy phenotype compared with Hcrt neuron-ablated mice including highly fragmented sleep, much higher cataplexy levels, and also a unique state in which the delta and theta ranges of the EEG were elevated while the mice were behaviorally immobile [143]. These results suggest that MCH neurons interact with the Hcrt system either directly or in their terminal fields and that MCH neurons normally have a role to prevent sleep/wake fragmentation and cataplexy.

Future Directions

Strengths and weaknesses of current narcolepsy models

The narcolepsy phenotype in the animal models discussed above (Table 1) is largely similar, with most exhibiting some combination of sleep disruption, SOREMs, difficulty maintaining wakefulness, cataplexy, and some strains exhibiting obesity in the absence of increased food intake. The strengths of each model largely depend upon what the model was intended to assess. For example, transgenic strains enable optogenetic manipulation of the Hcrt system to allow short-term, reversible manipulation of Hcrt neurons, thereby facilitating the assessment of specific aspects of Hcrt neural circuitry and the consequences of Hcrt activation or inhibition (Table 2). Since human narcolepsy usually occurs as the result of the post-pubertal loss of the Hcrt neurons, models such as the ATAX mouse and rat models, in which Hcrt-containing neurons are progressively ablated during early ontogeny, or the OX-tTA;TetO-DTA mouse model, in which the ablation of Hcrt-containing neurons can be induced later in development, are more representative of the etiology of the narcolepsy in humans than the prepro-orexin null mutant mouse in which the Hcrt neurons remain intact and are constitutively present, or Hcrt receptor KOs, since the Hcrt receptor terminal fields appear to be largely intact in humans with narcolepsy [144]. Furthermore, the OX-tTA;TetO-DTA mouse model offers a distinct advantage over the ATAX models in that this model allows experimenter control over both the timing and extent of neuronal ablation. Both models allow assessment of the network reorganization that occurs as a consequence of Hcrt neuron loss.

While animal models of narcolepsy have greatly advanced our understanding of this disorder, current animal models of narcolepsy are not without their own weaknesses. For example, the ad libitum sleep patterns of narcoleptic dogs complicated assessment of their sleep phenotype and the identification of the genetic defect in large breed narcoleptic dogs as a mutation in HcrtR2 indicated that they have a different disease than humans. While sleep dysregulation is clearly present in murine models of narcolepsy, the sleep patterns of mice are polyphasic and not representative of the consolidated sleep pattern of healthy humans. In addition, the main sleep phase of mice is during the light period rather than at night as in most humans. One of the most prominent disconnects in current animal models of narcolepsy is the emphasis on elevating the amount of cataplexy for experimental purposes (particularly for pharmacological studies) whereas cataplexy is relatively rare in most people with narcolepsy. While the benefits of using animal models for the investigation of narcolepsy and the testing of narcolepsy therapeutics far outweigh the pitfalls, the limitations imposed by these differences are important to bear in mind when analyzing and interpreting results. A substantial amount of interindividual variability in the amount of cataplexy also occurs in most models.

The need for a model of narcolepsy type 2

There remains room for the development of new models to aid in the investigation of unknown aspects of narcolepsy and to shed light on sleep/wake regulatory mechanisms, more generally. One unmet need is an animal model with a phenotype representative of NT2. Human NT2 patients have a less severe sleep phenotype than NT1 patients [23]. The sleep of NT1 patients is characterized by SOREMs, while type 2 patients also have SOREMs but the latency to REM sleep is longer on average than in type 1 patients [23, 145]. In addition, NT1 is also characterized by lower nocturnal sleep efficiency than NT2 [23, 145]. Indeed, even within these narcolepsy subtypes, symptoms appear to occur on a spectrum of severity [146].

Given the unequivocal role of the Hcrt system in narcolepsy, there are at least two approaches that are immediately evident: partial ablation of the Hcrt neuron population or engineering dysfunction into these cells. The rationale for partial ablation is supported, in part, from a postmortem study of an NT2 human brain that revealed a 33% Hcrt cell loss [147] compared to the 70%–100% Hcrt cell loss reported in patients with confirmed cataplexy (i.e. NT1) [76]. Partial hypocretin deficiency has been studied in the ATAX rat with the intent of relating the degree of cell loss to the level of Hcrt-1/Orexin-A in the CSF [85]. Although there were no differences in the amounts of wake, NREM, or REM sleep across the 24-h period relative to WT Sprague-Dawley rats, the diurnal distribution of REM sleep was strikingly different: ATAX rats had less REM sleep during the light phase and more REM sleep during the dark phase due to longer REM bouts than in WT rats. Direct transitions from wake to REM sleep were detected during both phases in ATAX rats but not in WT rats (cataplexy was not evaluated due to the absence of video recordings). However, Hcrt neurons were present in ATAX rats even at 60–124 weeks of age, although “staining was generally weaker and there were fewer HCRT axonal and dendritic processes” [85].

Partial ablation of Hcrt neurons was also implemented in OX-tTA;TetO-DTA mice by replacement of Dox chow with normal chow for 1.5, 3.5, or 14 days followed by restoration of Dox chow [95]. Restoration of Dox chow after 14 days arrested the increase in cataplexy and further degeneration of Hcrt neurons when assessed over the subsequent 11 weeks, while cataplexy continued to increase and Hcrt neuron degeneration proceeded in the mice maintained on normal chow for the same period. In another study, OX-tTA;TetO-DTA mice were subjected to varying feeding regimens to induce different degrees of Hcrt field ablation and the sleep/wake phenotype of these different groups were evaluated using a non-invasive system that detects activity and respiration [148]. The partial ablation groups exhibited an intermediate “narcoleptic sleep phenotype” compared to the intact or fully-ablated mice, suggesting partial ablation of the Hcrt neuron population as a potential model of NT2. While this study has interesting implications for the differentiation of the two narcolepsy subtypes, NT2 is primarily distinguished from NT1 by the absence of cataplexy and, unfortunately, the methodology used in this study was unable to assess cataplexy. Nonetheless, this report of an intermediate narcolepsy phenotype as the result of the partial ablation of the Hcrt field provides further evidence in support of the idea that NT1 and NT2 lie on a severity spectrum.

Another potential approach to creating a model of NT2 is to engineer dysfunction into the Hcrt neurons, which may have inadvertently occurred in the generation of OX/Arch mice as discussed above. In OX/Arch mice that highly express archaerhodopsin-3 (aHE mice) in their Hcrt neurons, these cells were hyperpolarized even in the absence of photoillumination [108]. These altered cellular properties of Hcrt neurons from aHE mice were reflected at the organismal level as perturbations in sleep, circadian, and metabolic parameters. However, aHE mice did not exhibit either cataplexy or sleep fragmentation. Thus, OX/Arch aHE mice share some, but not all, of the characteristics of NT2, and may be useful to investigate Hcrt system dysfunction when these neurons are intact, as we speculate may occur in some NT2 cases.

The future of narcolepsy therapeutics and animal models of narcolepsy

Animal models are fundamental to the development and testing of novel narcolepsy therapeutics. Obviously, the ultimate target in the development of narcolepsy treatments would be to eventually develop a cure for the etiology of the disorder: the loss of Hcrt-producing neurons in the LHA and PFH. To date, narcolepsy therapeutics have been symptomatic therapies [149]. Several experimental therapeutic approaches have aimed at more fully addressing the etiology of narcolepsy; however, each is not without its own unique drawbacks and hurdles. For example, ICV administration of Hcrt1 decreases the severity of the narcolepsy phenotype in several animal models [150, 151] but administration of the Hcrt neuropeptide in humans with narcolepsy is problematic due to difficulty in crossing the blood-brain barrier (BBB) and ICV administration is not feasible in humans [150–152]. Follow up studies have focused on alternative methods to circumvent the BBB and engineering of small molecule Hcrt receptor agonists capable of crossing the BBB [151, 153–156]. While Hcrt neuropeptide replacement therapies have shown promise, to better address the etiology of narcolepsy, researchers have also focused on different ways to replace the Hcrt-producing neurons themselves. Grafts of Hcrt-containing neurons have been shown to reduce symptom severity in Hcrt2/Saporin-lesioned rats and ATAX mice [157, 158]. Beyond the risk of the invasive surgical procedure to place the graft itself, grafted Hcrt cells showed a relatively low rate of survivability and, like other tissue transplants, grafts carry the risk of rejection [159]. Ideally, Hcrt-containing cells could be generated or repurposed from existing brain structures through less invasive methods [160].

Recently, several studies have demonstrated that astrocytes can be converted to fully functioning, mature neurons through genetic manipulation [161–163]. This co-opting of glial cells has been demonstrated to improve survival and reverse disease phenotypes in animal models of other neurodegenerative diseases such as Parkinson’s and Huntington’s disease whose etiology lies in the loss of specific neuronal populations [161, 162]. In many neurodegenerative diseases, other cells including astrocytes are also damaged, thereby limiting the availability of cells to repurpose for these types of therapies. By contrast, patients with narcolepsy exhibit a very specific loss of the Hcrt-containing neurons whereas other neuronal populations appear to remain intact. Furthermore, humans with narcolepsy exhibit an increase in glial density that occurs concomitantly with the loss of Hcrt-containing neurons [76], making narcolepsy an interesting candidate for the application of this type of therapy.

Future animal models of narcolepsy should build upon the now solid evidence for autoimmune-mediated degeneration of the Hcrt neurons to enable immunotherapeutic approaches. As the etiology of narcolepsy is revealed in more detail, other immune-mediated models such as the Orexin-HA mice should be developed. Understanding more precisely how the immune system causes the degeneration of the Hcrt neurons in humans with narcolepsy has the potential to inform many different aspects of the disorder and promises to provide further therapeutic targets for preventing, treating, or halting the onset of narcolepsy. A more precise understanding of the etiology of Hcrt neuron degeneration will inevitably provide us with a better understanding of the early stages of the disease and to identify individuals who might be more predisposed to developing narcolepsy. Understanding the earliest stages of the disease and factors leading to individual predisposition to developing narcolepsy will facilitate earlier detection, allowing the application of preventative therapeutics. While we can only speculate upon the future directions of narcolepsy research and narcolepsy animal models that might further inform on the disease, it is abundantly clear that animal models will continue to play an important role in the study of narcolepsy and sleep/wake regulation and the development and testing of narcolepsy therapeutics [149].

Disclosure Statement

Non-financial disclosure: none.

Financial disclosure: none.