Dual orexin and MCH neuron-ablated mice display severe sleep attacks and cataplexy

Orexin/hypocretin-producing and melanin-concentrating hormone-producing (MCH) neurons are co-extensive in the hypothalamus and project throughout the brain to regulate sleep/wakefulness. Ablation of orexin neurons decreases wakefulness and results in a narcolepsy-like phenotype, whereas ablation of MCH neurons increases wakefulness. Since it is unclear how orexin and MCH neurons interact to regulate sleep/wakefulness, we generated transgenic mice in which both orexin and MCH neurons could be ablated. Double-ablated mice exhibited increased wakefulness and decreased both rapid eye movement (REM) and non-REM (NREM) sleep. Double-ablated mice showed severe cataplexy compared with orexin neuron-ablated mice, suggesting that MCH neurons normally suppress cataplexy. Double-ablated mice also showed frequent sleep attacks with elevated spectral power in the delta and theta range, a unique state that we call ‘delta-theta sleep’. Together, these results indicate a functional interaction between orexin and MCH neurons in vivo that suggests the synergistic involvement of these neuronal populations in the sleep/wakefulness cycle.

Hcrt (Chemelli et al., 1999) or Hcrtr2 (Willie et al., 2003) gene knockout mice and orexin neuron-ablated mice (Hara et al., 2001;Tabuchi et al., 2014) display a narcolepsy-like phenotype. Narcolepsy is a chronic sleep disorder (Mahoney et al., 2019) caused by the specific degeneration of orexin neurons by the immune system (Peyron et al., 2000;Latorre et al., 2018). Narcolepsy patients have characteristic symptoms including excessive daytime sleepiness, hallucinations and cataplexy, a sudden loss of muscle tone triggered by positive emotions such as laughter (American Sleep Disorders Association, 1990;Burgess and Scammell, 2012). Optogenetic activation of orexin neurons induces wakefulness from sleep while optogenetic inhibition induces sleep from wakefulness (Adamantidis et al., 2007;Tsunematsu et al., 2011;Schö ne et al., 2012;Williams et al., 2019). Together, these studies indicate that orexin neurons play an important role in the maintenance of wakefulness and prevent cataplexy induced by positive emotions.
The neuropeptide MCH was originally isolated from fish pituitary as a substance that controls skin pigmentation (Kawauchi et al., 1983). In mammals, MCH neurons are mainly distributed in the tuberal hypothalamus within which the orexin neurons are also located. Optogenetic activation of MCH neurons increases the total time in rapid eye movement (REM) sleep and reduces non-REM (NREM) sleep in mice Konadhode et al., 2013;Tsunematsu et al., 2014). Ablation of MCH neurons promotes wakefulness and decreases time in NREM sleep but has no effect on REM sleep . These observations suggest that MCH neurons are likely involved in the regulation of both NREM and REM sleep. We recently reported that REM sleep-active MCH neurons are involved in memory erasure during REM sleep , further supporting the concept that MCH neurons are involved in multiple physiological functions (Diniz and Bittencourt, 2017;Arrigoni et al., 2019).
While the orexin and MCH neurons have different roles in the regulation of sleep/wakefulness (Konadhode et al., 2014), they have similar projection areas and receptor distributions (Trivedi et al., 1998;Hervieu et al., 2000;Kilduff and de Lecea, 2001;Marcus et al., 2001;Saito et al., 2001). Orexin and MCH neurons have also been reported to interact with each other. For example, application of orexin peptide increases spike frequency in MCH neurons in vitro (van den Pol et al., 2004) and optogenetic activation of orexin neurons inhibits MCH firing through GABA A receptors (Apergis-Schoute et al., 2015), while MCH reverses hypocretin-1-induced enhancement of action potentials in orexin neurons (Rao et al., 2008). Nevertheless, it is still unclear how interactions between orexin and MCH neurons contribute to sleep/wake regulation.
To understand the functional communication between orexin and MCH neurons and sleep/wakefulness regulation, we generated transgenic mice in which both orexin and MCH neurons were simultaneously ablated by tetracycline trans-activator (tTA)-induced expression of the diphtheria toxin A fragment (DTA). Analysis of the sleep patterns of orexin and MCH neuron double-ablated mice (OXMC mice) revealed that these mice exhibited very high levels of cataplexy and sleep/wake abnormalities. OXMC mice had increased wakefulness and profound reductions in REM sleep (particularly during the dark phase) and a corresponding increase in cataplexy, suggesting that MCH neurons have a suppressive role on cataplexy. OXMC mice also frequently showed short episodes of behavioral arrest with high d and q power. We defined this unique state as 'delta-theta sleep' (DT sleep) since this state was distinct from other states of sleep/wakefulness or cataplexy. Behavioral and pharmacological assessments showed some similarities as well as characteristic differences between DT sleep and cataplexy.
Orexin neurons and MCH neurons project widely throughout the brain. Dense projections of orexin neurons have been described in the locus coeruleus (LC) and raphe nucleus (Peyron et al., 1998;Date et al., 1999). MCH neurons densely innervate the medial septum (MS) and hippocampus Izawa et al., 2019). In OXMC mice maintained for 4 weeks in the DOX(-) condition, orexin and MCH nerve terminals in these projection sites were completely eliminated ( Figure 2). These results confirmed that, not only the cell bodies, but also the projections of orexin and MCH neurons were eliminated in OXMC mice after DOX(-) for 4 weeks.
Dual-ablated mice frequently displayed short behavioral arrests with high spectral power in the d and bands of the EEG during wakefulness Sleep and wakefulness was assessed in three groups of mice, OXMC DOX(-), OXMC DOX(+) and OX DOX(-) mice by EEG and EMG recording ( Figure 3A). Spectral analyses of the EEG and EMG during wakefulness, NREM and REM sleep were indistinguishable across the three groups in the DOX(+) condition (before DOX removal, Figure 4A and B). After DOX removal, there was no difference in EEG spectrum and EMG integral of OXMC DOX(-) mice during any vigilance state as neuronal ablation proceeded ( Figure 4C, D and E). These results indicate that neither orexin nor MCH neuron ablation affected the EEG power spectrum in any vigilance state.
Ablation of orexin neurons is known to induce narcolepsy-like symptoms, such as fragmentation of sleep and wakefulness, sleep onset REM sleep, and cataplexy-like behavioral arrests (Hara et al., 2001;Tabuchi et al., 2014). Since orexin neurons were ablated in both OX DOX(-) and OXMC DOX (-) mice, these mice showed cataplexy-like behavioral arrests but OXMC DOX(+) mice did not ( Figure 3B and Table 1). In addition, OXMC DOX(-) mice frequently showed behavioral arrest episodes that were similar to cataplexy (Video 1) but which occurred after a sustained period of wakefulness. However, these behavioral arrest episodes were different from cataplexy based on a number of criteria (Table 2), particularly the EEG, which showed high amplitude spectral power in the d and q bandwidths during the behavioral arrests ( Figure 3C). These episodes were defined as a sudden cessation of motor activity characterized by decreased EMG and relatively high-power ratios of d and q in the EEG, preceded by at least 40 s of wakefulness (10 epochs) and followed by a return Figure 1 continued Protocol I is the control (DOX(+)) condition; Protocol II is the experimental (DOX(-) for 4 weeks) condition. EEG and EMG surgeries were performed at age 10 weeks in the sleep-recording group; the mice used for immunostaining did not undergo surgery. Light blue and white bars represent the periods of DOX chow (DOX(+)) and normal chow (DOX(-)) availability, respectively. Gray bars indicate sleep recordings; red arrows indicate when mice were sacrificed for immunostaining. (D) Immunostaining of orexin (brown) and MCH (black) neurons in the LH at DOX(-) 0 week (i and v), 2 weeks (ii and vi), 4 weeks (iii and vii), and DOX(+) 4 weeks (iv and viii). Black and brown arrowheads indicate typical examples of orexin and MCH neurons, respectively. Panels v-viii are magnifications of the areas delineated by the squares in Panels i-iv. Scale bars: i-iv, 500 mm; v-viii, 100 mm. (E) The number of orexin and MCH neurons in OXMC mice from the DOX(+) and (-) conditions (n = 3-6). Values are mean ± SEM. *p<0.05 vs. DOX(+). Data were analyzed by unpaired t test. The online version of this article includes the following source data and figure supplement(s) for figure 1: Source data 1. Source data for Figure 1E.   to wakefulness. The relative power ratio of d and q bands during these arrests was significantly different from NREM sleep, REM sleep or cataplexy ( Figure 3D and E). Consequently, we called the state during these arrests 'delta/theta sleep' (DT sleep). These results suggest that DT sleep is a novel brain state different from other states, as defined by EEG spectral characteristics.

Effect of dual ablation of orexin and MCH neurons on sleep and wakefulness
The sleep/wakefulness patterns of OXMC mice were analyzed during the simultaneous ablation of both the orexin and MCH neurons. Figure 5A presents typical 24 hr hypnograms observed during the first 4 weeks of orexin and MCH neuron ablation. Sleep/wakefulness fragmentation was observed in both OXMC DOX(-) and OX DOX(-) mice by 1 week in the DOX(-) condition, particularly at the beginning of the dark phase ( Figure 5A). Cataplexy was detected from 2 weeks DOX(-) onwards in both OXMC DOX(-) and OX DOX(-) mice ( Figure 5A, C and D). DT sleep was only observed in OXMC mice; the first episode was detected at 1 week DOX(-), a week earlier than the initial observation of cataplexy at 2 weeks DOX(-). Since DT sleep often occurred during the early dark phase, Figure 5B presents an expanded hypnogram for 1 hr in the early dark phase at 4 weeks DOX(-) (the period enclosed by the magenta rectangle in Figure 5A). The mean DT sleep duration and number of DT bouts during the dark phase after 4 weeks of DOX(-) were 14.4 ± 0.7 s and 98.3 ± 6.2, respectively ( Figure 5C). Although the total time and the number of DT sleep bouts observed during the light phase progressively increased by 4 weeks after DOX removal, neither the total time, mean bout duration nor the number of DT sleep bouts progressively increased during the dark phase as ablation proceeded ( Figure 5C and D). Interestingly, the mean DT sleep bout duration appeared to plateau at~15 s in both the light and dark phases by 2 weeks post-ablation ( Figure 5C and D).
To clarify the functional role of MCH neurons, the total time, mean bout duration and the number of bouts of each vigilance state were compared at each stage of ablation. The total wakefulness time significantly increased in OXMC DOX(-) mice at 4 weeks after DOX removal in the both light and dark phases as compared with OX DOX(-) mice ( Figure 5C and D and Table 1). Conversely, except for NREM sleep during the light period, OXMC DOX(-) mice exhibited a significant reduction in total REM and NREM sleep and the number of bouts in both light and dark phases at 4 weeks DOX(-) when compared with OX DOX(-) ( Figure 5C and D and Table 1). The total time in cataplexy and mean cataplexy bout duration was greater in OXMC DOX(-) mice as compared with OX DOX(-) mice ( Figure 5C and D and Table 1), indicating that the loss of MCH neurons exacerbated cataplexy symptomatology. Table 1 summarizes the vigilance state and cataplexy characteristics in OXMC DOX(-) mice compared to OX DOX(-) mice. The differences in sleep/wakefulness phenotype between these two strains suggest that MCH neurons are part of a circuit that normally suppresses cataplexy.

Behavioral assessment of DT sleep
To determine whether DT sleep is different from cataplexy or sleep, we performed tactile stimulation on OXMC DOX(-) mice during cataplexy, NREM or DT sleep. Although mice rarely responded to tactile stimulation during cataplexy (Video 2), they always responded to tactile stimulation during NREM or DT sleep by escaping the source of stimulation. Figure 6A illustrates that the probability of wakefulness was 33% after tactile stimulation during cataplexy compared to 100% in NREM and DT sleep. Mice could sense the touch of the brush and escaped during DT sleep and NREM sleep (Video 3). These results suggest that DT sleep differs from cataplexy and may be more similar to NREM sleep.
To further characterize DT sleep, we analyzed the position of mice in the home cage when DT sleep was initiated. The cage was divided into 4 areas, with a nest located in one of them ( Figure 6B). NREM and REM sleep occurred with higher probability when mice were in the nest area week 0 (i and v), 2 weeks (ii and vi), 4 weeks (iii and vii) and the DOX(+) condition at 4 weeks (iv and viii). Panels v-viii are magnifications of the areas delineated by the squares in Panels i-iv. Scale bars: i-iv, 500 mm; v-viii, 100 mm.  Figure 6B bar graph). However, both DT sleep and cataplexy occurred in the nest area at almost a chance level (25%, Figure 6B). These results suggest that DT sleep occurs with a similar timing to cataplexy and, like cataplexy, may be a spontaneous, uncontrolled state.
We next categorized behaviors before cataplexy or DT sleep initiation by video analysis. The probability of running before cataplexy was significantly higher than running before DT sleep ( Figure 6C). In comparison, the probability of grooming behavior before DT sleep was significantly higher than grooming before cataplexy ( Figure 6C). Although the probabilities differed with some behaviors such as running and grooming, both DT sleep and cataplexy often occurred after running or digging. These results suggest that DT sleep can occur under situations similar to cataplexy.
Day-and night-time differences in cataplexy or DT sleep were also analyzed. As described previously , orexin-ablated mice (OX DOX(-)) experienced more cataplexy during the dark phase than during the light phase ( Figure 6D). In contrast, there was no substantial difference in cataplexy occurrence between the light and dark phases in OXMC DOX(-) mice, although DT sleep occurred more frequently in the dark phase compared to the light phase ( Figure 6D). These results suggest that, even though MCH neuron ablation alone does not result in cataplexy, MCH neuron activity might be involved in the temporal regulation of cataplexy during both the light and dark phases.

Pharmacological assessments of DT sleep
To further understand whether DT sleep is related to sleep or cataplexy, OXMC DOX(-) mice were exposed to chocolate to increase cataplexy, administered clomipramine to suppress cataplexy, or modafinil to promote wakefulness. After removing DOX chow for 4 weeks to generate dual orexin and MCH neuron ablation, DOX chow was replaced to stop further ablation and arrest further progress of sleep/wakefulness abnormalities . The experimental protocols are illustrated in Figures 7A, 8A and 9A. Briefly, chocolate (1.7-1.9 g, milk chocolate, Meiji) was provided for 15 min just prior to dark onset (ZT12) for 3 days and sleep/wakefulness was analyzed on the third day. After a one-day interval, the vehicle, clomipramine or modafinil was administered followed by counterbalanced treatments on subsequent days. All administrations were performed just before dark onset.
Chocolate availability is known to increase the time in cataplexy and the number of cataplexy bouts in Hcrt knockout mice (Burgess et al., 2013;Oishi et al., 2013). Chocolate administration to OXMC DOX(-) mice significantly increased the total time in wakefulness and decreased REM and NREM sleep but did not affect total time or the number of bouts of DT sleep ( Figure 7B and C). Clomipramine administration significantly decreased time in wakefulness and REM sleep and increased total time of NREM sleep. As expected, clomipramine also significantly inhibited cataplexy but did not affect DT sleep time, the number of bouts or their duration ( Figure 8B and C). On the other hand, modafinil administration increased the total time of wakefulness, inhibited NREM and REM sleep and, as expected, did not affect cataplexy. Although the total time in DT sleep was not significantly affected by modafinil administration, the mean DT sleep bout duration was reduced ( Figure 9B and C). These results suggest that the neural mechanisms underlying DT sleep might be distinct from those underlying cataplexy.

Relationship between DT sleep and the transition from NREM to REM sleep
Since both d and q power in the EEG were high during DT sleep ( Figure 3D and E), we evaluated the EEG during sleep/wake and wake/sleep transitions to compare EEG spectral characteristics. We found that the EEG spectrum in the transition from NREM to REM sleep was similar to DT sleep. We analyzed by one-way ANOVA followed by the Bonferroni post hoc test. Despite comparable EMG levels, DT sleep has significantly greater spectral power in the d range and less power in the q range than either REM sleep or cataplexy. The online version of this article includes the following source data for figure 3: Source data 1. Source data for Figure 3D and E.  Figure 10A). The EEG spectra during the transition from NREM sleep to REM sleep was not altered by dual orexin and MCH neuron ablation ( Figure 10B). Spectral power in the d, q, a and b bands of the EEG was indistinguishable between DT sleep and the NREM to REM sleep transition ( Figure 10C). In addition, the 24 hr EMG integral did not differ between DT sleep and the NREM to REM transition ( Figure 10C). These results indicated a similarity between DT sleep and the transition state from NREM to REM sleep.

Discussion
To understand functional interactions between orexin and MCH neurons in the regulation of sleep and wakefulness, we generated dual orexin and MCH neuron-ablated mice and compared the resultant sleep abnormalities to those of orexin neuron-ablated mice. Double-ablated mice exhibited pronounced cataplexy and the total time in cataplexy and mean cataplexy bout duration were significantly increased, suggesting that MCH neurons normally have a suppressive role on cataplexy. Double-ablated mice also had exaggerated sleep abnormalities compared to singly-ablated or intact  The online version of this article includes the following source data for Table 1: Source data 1. Source data for Table 1. mice; specifically, increased time in wakefulness and decreased time in NREM and REM sleep. Double-ablated mice also exhibited a novel state that we called DT sleep, defined as an episode of sudden behavioral arrest of brief (~15 s) duration preceded by at least 40 s of wake and characterized by high d and q power in the EEG. Behavioral, electrophysiological and pharmacological assessments discriminated DT sleep from NREM, REM and cataplexy. Cataplexy is well-known to be triggered by strong positive emotions (Mignot, 1998). Previously we showed that cataplexy is prevented by orexin but not by other co-localized neurotransmitters in the orexin neurons, such as glutamate and dynorphin . Orexin neuron activity is thought to prevent cataplexy through activation of OX2R when positive emotion occurs since Hcrt or Hcrtr2 gene knockout or ablation of orexin neurons induces cataplexy. The amygdala is implicated in the emotional stimulation that facilitates cataplexy (Burgess et al., 2013;Hasegawa et al., 2014;Hasegawa et al., 2017;Mahoney et al., 2017;Snow et al., 2017). Neurons in the amygdala project to the brainstem, which is crucial for the regulation of muscle tone and REM sleep (Wallace et al., 1989). Serotonergic neurons in the raphe nucleus projecting to the amygdala are thought to be the neural pathway responsible for suppression of cataplexy by orexin neurons, since activation of serotonergic neurons or activation of serotonergic nerve terminals in the amygdala prevents cataplexy (Hasegawa et al., 2014;Hasegawa et al., 2017). Here, we show that Video 1. Typical behavior for DT sleep in orexin-and MCH neuron-ablated mice.
https://elifesciences.org/articles/54275#video1  MCH neurons decrease cataplexy duration but not cataplexy bout frequency. This difference might suggest that the mechanism of MCH neurons to prevent cataplexy is different from that of serotonergic neurons. Our in situ hybridization results as well as previous reports (Mickelsen et al., 2017;Mickelsen et al., 2019) suggest that both orexin and MCH neurons are glutamatergic neurons. Among LH glutamatergic neurons, these two types of neurons are a relatively minor population (the proportion of MCH neurons and orexin neurons was 15% and 24%, respectively). These results suggest that ablation of a small number of additional LH glutamatergic neurons (10-15%) that co-express MCH exacerbates the narcolepsy phenotype in orexin neuron-ablated mice, underscoring the important role of MCH neurons to suppress cataplexy.
Neurons in the amygdala innervate and suppress the ventrolateral periaqueductal grey (vlPAG/ LPT), DR, LC and tuberomammilary nucleus (TMN), areas that are known to be involved in the regulation of REM sleep or cataplexy (Burgess et al., 2013). It has been reported that MCH neurons also inhibit DR, LC, TMN and vlPAG neurons (Torterolo et al., 2008;Sapin et al., 2010;Del Cid-Pellitero and Jones, 2012;. These areas might be part of the neural circuit contributing to cataplexy prevention since MCH neurons may innervate different types of neurons in these areas. Alternatively, other downstream targets of MCH neurons such as the sublaterodorsal tegmental nucleus might be involved in the regulation of cataplexy and REM sleep Monti et al., 2016). Recently, chemogenetic activation of MCH neurons in Hcrt knockout mice was shown to increase cataplexy bout duration without affecting the number of bouts (Naganuma et al., 2018), a finding that is inconsistent with our results. However, the absence of orexin during development in constitutive Hcrt knockout mice might induce compensatory changes in neural circuitry, since Hcrt knockout mice have many fewer cataplexy bouts per night than DTA mice in which orexin neurons are ablated after maturation . In addition, the chemogenetic activation paradigm used by Naganuma et al. could have induced strong activation of MCH cells above the physiological range. Recently, we reported that MCH neurons can be divided in three types by activation pattern: wake-active, REM-active and wake-and REM-active . Activation of wake-active MCH neurons within the physiological range in orexin neuron-ablated mice would likely be valuable to further understand the role of MCH neurons in cataplexy.
Orexin-and MCH-neurons double-ablated mice frequently displayed brief (~15 s) episodes of behavioral arrest during wakefulness that we called DT sleep, which was characterized by high power d and q waves in the EEG with low EMG amplitude. The first appearance of DT sleep was at 1 week DOX(-), which was sooner after DOX removal than the appearance of cataplexy at 2 weeks and suggests that DT sleep may be due to a smaller reduction in the number of orexin and MCH neurons. When initially observed, we assumed that DT sleep was cataplexy since both behaviors shared the characteristic of sudden behavioral arrest during wakefulness. However, the EEG spectrum was clearly distinct from cataplexy. Another critical difference between DT sleep and cataplexy was that mice responded to tactile stimulation by immediately returning to wakefulness from DT sleep but not from cataplexy. Furthermore, the behavioral arrest of DT sleep was observed without complete muscle atonia; thus, mice could maintain posture during DT sleep. This observation underscores the difference between DT sleep and cataplexy, since muscle atonia is an indispensable component of cataplexy.
Pharmacological assessments further confirmed differences between DT sleep and cataplexy. Chocolate is known to increase cataplexy by activating the prefrontal cortex and amygdala (Burgess et al., 2013). Chocolate increased cataplexy in double-ablated mice, but did not increase DT sleep. Clomipramine is a selective serotonin reuptake inhibitor that can inhibit REM sleep and cataplexy (Willie et al., 2003). Clomipramine significantly decreased cataplexy in double-ablated mice, but did not affect the number of bouts or mean bout duration of DT sleep. Modafinil is known to be a dopamine reuptake inhibitor that can promote wakefulness (Willie et al., 2005). Although modafinil significantly increased wakefulness, decreased NREM sleep and decreased the duration of DT sleep, the number of DT bouts was unaffected. Based on these results, DT sleep is classified as sleep but is distinct from NREM or REM sleep. Modafinil decreased the duration of DT sleep and DT Source data 1. Source data for Figure 7B and C.
sleep was often observed after grooming behavior, suggesting that dopaminergic neurons might be involved in the initiation of DT sleep since dopaminergic neurons are activated during grooming (Fornaguera et al., 1995;Cromwell et al., 1998).
The transition from NREM to REM sleep is well known to be characterized by relatively high power in d and q waves (Benington et al., 1994). Thus, we compared DT sleep to the transition from NREM to REM sleep. We found that d, q, a and b wave power were not significantly different between DT sleep and NREM to REM sleep transitions. The activity of orexin neurons is high in wakefulness and low in NREM and REM sleep (Lee et al., 2005). In comparison, the activity of MCH neurons is high in REM sleep and low in NREM sleep and wakefulness (Hassani et al., 2009; Blanco- Source data 1. Source data for Figure 9B and C. Centurion et al., 2019;Izawa et al., 2019). Therefore, the minimum activity of these two types of neurons should occur just before initiation of REM sleep. It is possible that this transition might be manifest as DT sleep, since the brains of dual orexin-and MCH neuron-ablated mice are similar to conditions in which both orexin neurons and MCH neurons are inactive. Furthermore, NREM and REM sleep were attenuated in double-ablated mice, suggesting that DT sleep enables a compensatory role of both NREM and REM sleep. Whereas the loss of orexin neurons in narcolepsy has been proposed to disrupt a 'flip/flop switch' that facilitates the normal transition from wakefulness to sleep , the dual loss of orexin and MCH neurons appears to prolong the normal transition from NREM to REM sleep through the DT sleep state.
To date, specific degeneration of MCH neurons has not been reported in humans. Ablation of MCH neurons in mice induces a relatively weaker effect on sleep and wakefulness compared with ablation of orexin neurons Tsunematsu et al., 2014). MCH concentration in cerebrospinal fluid is not typically measured in sleep disorders patients. However, a patient with hypersomnia had low concentrations of both orexin (95 pg/ml) and MCH peptide (6 pg/ml) in the cerebrospinal fluid (Wada et al., Japan sleep annual meeting 2018, P-128). Detailed study of this and similar cases will help to understand the role of MCH neurons on sleep/wakefulness in humans.
Here, we revealed a new role of MCH neurons to prevent cataplexy and regulate sleep and wakefulness. Orexin neurons and MCH neurons interact in the hypothalamus and potentially in terminal projection sites to influence sleep/wake control. Since human studies have already suggested interaction between these two systems (Blouin et al., 2013), further investigations are warranted to reveal the regulatory mechanisms underlying the roles of these two systems in the control of sleep/ wakefulness and cataplexy.

Experimental procedures Animal usage
All experimental procedures were performed in accordance with the guide of the Institutional Animal Care and Use Committes at the Research Institute of Environmental Medicine at Nagoya University and SRI International. All efforts were made to decrease animal suffering and to minimize the number of animals used.

EEG and EMG surgical procedure
Male mice were anesthetized with 2% isoflurane (095-06573, FUJIFILM Wako Pure Chemical Corporation, Japan) and implanted with three EEG electrodes (U-1430-01, Wilco, Japan) on the skull and two EMG electrodes (AS633, Cooner wire, Mexico) in the rhomboid muscle at 10 weeks of age as previously described Tabuchi et al. (2014); Figure 1C). Carprofen (Zoetis Inc, Japan), 20 mg/kg (subcutaneous injection), was administered the day of, and the day after, surgery for its anti-inflammatory and analgesic properties. After surgery, mice were housed separately for about 7 days for recovery. A cable with a slip ring (Kissei Comtec Co., Ltd, Japan) was connected to mice in the cage for 7 days before the initiation of EEG and EMG recordings.
Sleep/wake recordings EEG and EMG signals were filtered at EEG 1.5-30 Hz and at EMG 15-300 Hz and amplified by an amplifier (AB-610J, Nihon Koden, Japan). The digital sampling rate was 128 Hz. Animal behavior was monitored through a Charged Coupled Device (CCD) video camera (SPK-E700CHP1, Keiyo Techno, Japan) and an infrared activity sensor (Kissei Comtec). All EEG and EMG data were recorded by VitalRecorder (Kissei Comtec) and analyzed by SleepSign software (Kissei Comtec).

Vigilance state analysis
EEG analysis was performed by Fast Fourier Transform (FFT). Power spectra profiles over a 0-10 Hz window with 1 Hz resolution for d (1 d <6 Hz) and q (6 q <11 Hz) bandwidths were calculated. EEG and EMG were automatically screened in 4 s epochs by SleepSign and classified as previously reported (Tsunematsu et al., 2011;Tabuchi et al., 2014;Tsunematsu et al., 2014). Wake was characterized by high EMG amplitude or locomotion score with low EEG amplitude, NREM sleep was characterized by low EMG amplitude and high EEG d power, and REM sleep was characterized by low EMG and low EEG amplitude with over 50% of q activity. Cataplexy was defined by muscle atonia lasting more than 10 s after more than 40 s of wakefulness, and low amplitude with over 50% of q activity in the EEG (Scammell et al., 2009). DT sleep was defined by the following criteria: (1) a period of behavioral arrest, (2) that was preceded by at least 40 s of wakefulness, (3) characterized by high d and q power (at least 60-80% of the d power found in NREM sleep) in the EEG, (4) low EMG activity, and (5) was terminated by a return to wakefulnes. Vigilance state classifications were automatically assigned by SleepSign software and then visually corrected. The transition from NREM to REM sleep was identified as the three epochs (12 s) before a REM sleep bout. EEG power spectral density were analyzed and viewed using the Signal Processing Toolbox of Matlab (Mathworks, USA).

Behavioral assessment
Behaviors were manually identified from video recordings. Behaviors occurring during the 4 s epoch immediately preceding DT sleep or cataplexy bouts were analyzed and calculated as percentages of the total recording time.

Tactile stimulation
Tactile stimulation was performed by applying a gentle touch around the face using a brush during the light phase (11:00-18:00) (Videos 2 and 3). Six OXMC DOX(-) mice (20-26 weeks old) were studied. Stimulations were conducted when immobility had lasted at least 10 s during NREM sleep, cataplexy or DT sleep. The interval between individual stimulations was at least 1 min. Vigilance states that occurred after stimulation were identified.
Identification of the location of each sleep/wake stage within the home cage We divided each home cage into four quadrants, one of which contained a nest ( Figure 6B). The home cage quadrant in which each mouse was located at the beginning of each vigilance state was identified by video recording.

Immunohistochemistry
Mice were deeply anesthetized with an i.p. injection of somnopentyl (6.48 mg/kg) (Kyoritsu Seiyaku Corporation, Tokyo, Japan) and 3% isoflurane, and then sequentially perfused with saline and 10% formalin (066-03847, FUJIFILM Wako). The brains were removed and immersed in 10% formalin overnight at 4˚C and then immersed in a 30% sucrose solution in PBS for at least 2 d. The brains were frozen in embedding solution (4583, Sakura Finetek Japan, Japan) and stored in a À80˚C freezer. Brains were subsequently sectioned at 40 mm thickness in a cryostat (CM3050-S, Leica Microsystems K.K., Japan). Coronal sections of mouse brains were placed in phosphate buffer (PBS) with 0.3% H 2 O 2 (H1009-500ML, MilliporeSigma) to inactivate endogenous peroxidase for 40-45 min at RT. After washing three times for 10 min in PBS containing 0.25% Triton X-100 (35501-15, Nacalai Tesque, Japan) and