Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
PNAS 2023 Vol. 120 No. 41 e2301951120 https://doi.org/10.1073/pnas.2301951120 1 of 12
RESEARCH ARTICLE
|
Significance
Narcolepsy is a sleep disorder
caused by decient orexin
signaling due to orexin neuron
degeneration. However, the
neural mechanisms underlying
the rapid eye movement (REM)
sleep- related symptomatology of
narcolepsy remain unclear. We
determined previously uncertain
details of orexin neuron activity
dynamics during NREM sleep,
REM sleep, and cataplexy at the
population- level by ber
photometry and at the single-
neuron level by microendoscopy.
Using optogenetic approaches,
we found that orexin neuron
activity during NREM sleep
regulates NREM–REM sleep
transitions whereas orexin
neuron activity during REM sleep
regulates REM sleep structure
and cataplexy. Collectively, this
study advances understanding of
how orexin neurons normally
regulate REM sleep architecture,
and how loss of this regulation
contributes to the REM sleep–
related symptomatology of
narcolepsy.
Author contributions: H.I. and A.Y. designed research;
H.I. and S.M.R. performed experiments; S.I. and D.O. set
up state- dependent photoillumination system and the
ber photometry system, respectively; H.I., N.F., S.M.R.,
Y.M., T.S.K., and A.Y. contributed to the data analysis and
review; and H.I., T.S.K., and A.Y. wrote the paper.
The authors declare no competing interest.
This article is a PNAS Direct Submission.
Copyright © 2023 the Author(s). Published by PNAS.
This open access article is distributed under Creative
Commons Attribution- NonCommercial- NoDerivatives
License 4.0 (CC BY- NC- ND).
1To whom correspondence may be addressed. Email:
yamank@cibr.ac.cn.
This article contains supporting information online at
https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.
2301951120/- /DCSupplemental.
Published October 5, 2023.
NEUROSCIENCE
Deficiency of orexin signaling during sleep is involved
in abnormal REM sleep architecture in narcolepsy
HirotoItoa,b,c , NoriakiFukatsua,b , SheikhMizanurRahamana,b , YasutakaMukaia,b , ShuntaroIzawaa,b , DaisukeOnoa,b, ThomasS.Kildud,
and AkihiroYamanakaa,b,e,f,g,h,1
Edited by Joseph Takahashi, The University of Texas Southwestern Medical Center, Dallas, TX; received February 3, 2023; accepted July 10, 2023
Narcolepsy is a sleep disorder caused by deficiency of orexin signaling. However, the
neural mechanisms by which deficient orexin signaling causes the abnormal rapid eye
movement (REM) sleep characteristics of narcolepsy, such as cataplexy and frequent
transitions to REM states, are not fully understood. Here, we determined the activity
dynamics of orexin neurons during sleep that suppress the abnormal REM sleep archi-
tecture of narcolepsy. Orexin neurons were highly active during wakefulness, showed
intermittent synchronous activity during non- REM (NREM) sleep, were quiescent prior
to the transition from NREM to REM sleep, and a small subpopulation of these cells
was active during REM sleep. Orexin neurons that lacked orexin peptides were less active
during REM sleep and were mostly silent during cataplexy. Optogenetic inhibition of
orexin neurons established that the activity dynamics of these cells during NREM sleep
regulate NREM–REM sleep transitions. Inhibition of orexin neurons during REM sleep
increased subsequent REM sleep in “orexin intact” mice and subsequent cataplexy in
mice lacking orexin peptides, indicating that the activity of a subpopulation of orexin
neurons during the preceding REM sleep suppresses subsequent REM sleep and cata-
plexy. us, these results identify how deficient orexin signaling during sleep results in
the abnormal REM sleep architecture characteristic of narcolepsy.
hypothalamus | orexin | calcium imaging | cataplexy | sleep
Impairment of adequate sleep duration or quality is well known to increase sleepiness and
to negatively impact the quality of wakefulness. Similar homeostatic regulation occurs for
rapid eye movement (REM) sleep, a state characterized by muscle atonia and desynchro-
nized cortical activity in which theta waves are dominant in the electroencephalogram
(EEG). For example, selective deprivation of REM sleep is well known to cause subsequent
REM sleep rebound (1–5). Since the discovery of REM sleep in humans (6), various
neural populations and mechanisms in the hypothalamus, midbrain, and basolateral amyg-
dala have been shown to regulate the initiation, maintenance, or inhibition of REM sleep
(3, 7–12). However, the precise neural mechanisms that regulate appropriate sleep archi-
tecture, particularly REM sleep, based on preceding vigilance states remain mostly
unknown (8).
Narcolepsy is a sleep disorder thought to be caused by selective loss of neurons in the
perifornical and lateral hypothalamic area (LHA) that produce the neuropeptides orexin
A and B (also known as hypocretin- 1 and - 2) (13–16). e loss of orexin peptides due to
orexin neuron degeneration plays a critical role in the etiology of narcolepsy (17–19). e
pathophysiology of narcolepsy has two main features: diculty in maintenance of wake-
fulness, manifested as excessive daytime sleepiness (20–24), and abnormal REM sleep
architecture (21, 24), exemplied by frequent transitions to REM sleep and by intrusion
of REM atonia into wakefulness (cataplexy and sleep paralysis). e most dramatic symp
-
tom of narcolepsy is cataplexy, the sudden loss of muscle tone during wakefulness (20–24).
Abnormal REM sleep architecture in narcolepsy might be explained by an accelerated
buildup of REM sleep pressure (25) and/or a lower threshold to enter REM sleep (4).
Interestingly, in mice lacking the orexin peptides, REM sleep deprivation increased sub-
sequent cataplexy (4).
e neural pathways involved in the etiology of narcolepsy have yet to be fully eluci-
dated. Although it has been frequently proposed that orexin neurons should receive excit-
atory input and be activated to prevent the triggering of cataplexy (20–24), the activity
of orexin neurons during cataplexy dened by EEG and electromyographic (EMG) record-
ings has not been directly measured in mouse models of narcolepsy in which the orexin
peptides are absent. Previous reports have proposed that the loss of orexin signaling in
wakefulness- promoting nuclei causes the sleepiness and sleep fragmentation characteristic
of narcolepsy (20–24). Recent studies reported that orexin signaling during REM sleep
also functions in the stabilization of REM sleep and induces REM sleep–like muscle atonia
OPEN ACCESS
2 of 12 https://doi.org/10.1073/pnas.2301951120 pnas.org
(3, 26). However, the neural mechanisms through which the loss
of orexin signaling causes abnormal REM sleep architecture in
narcolepsy have not been dened. Here, we determined the activ-
ity dynamics of orexin neurons during sleep and their involvement
in the abnormal REM sleep architecture characteristic of
narcolepsy.
Results
Activity Dynamics of Orexin Neurons during Sleep. It has been
reported that in rats, 2 to 9% of orexin neurons were cFos active
after NREM and REM sleep or REM sleep rebound (27, 28).
Particularly, 28% of orexin neurons that sent terminals to the
sublaterodorsal tegmental nucleus (SLD) were cFos active after
REM sleep rebound (3). Although these studies suggest the
existence of REM- active orexin neurons, more precise methods
to detect the activity invivo are warranted. Previous studies using
extracellular recordings described that orexin neuron activity
was the highest during wakefulness, but intermediate, low, or
virtually silent in NREM and REM sleep (29–32). Although ber
photometry recordings of orexin neurons described intermittent
low activity in the cell bodies during NREM sleep (33), orexin
signaling occurs in the SLD projection area during REM sleep
(3). ese inconsistent ndings may be due to the limited number
of orexin neurons recorded, a limited recording period, and/or a
limited resolution of activity imaging. us, the activity dynamics
of orexin neurons during sleep warrant a more comprehensive
study at both the single- cell and population levels, and across
many sleep/wakefulness state changes by extensive observations
for prolonged periods.
To reveal activity dynamics of the orexin neurons across phys-
iological sleep/wakefulness states in detail, we conducted ber
photometry recordings with EEG and EMG in freely moving mice
for 24 h. To express the Ca2+ indicator GCaMP6s specically in
orexin neurons, AAV9- TetO- GCaMP6s was injected into the
LHA of orexin- tTA mice (Fig. 1A). Immunohistochemical analysis
(n = 3) conrmed that, of the neurons labeled for GCaMP6s, 93.8
± 1.2% of them also expressed orexin- A (ratio of orexin
neurons/GCaMP6s- expressing neurons) and, of the neurons labe-
led for orexin- A, 91.1 ± 1.2% of them also expressed GCaMP6s
(ratio of GCaMP6- expressing/orexin neurons). Only 0.7 ± 0.7%
of MCH neurons expressed GCaMP6s, and 1.1 ± 1.1% of
GCaMP6s positive neurons were MCH- immunoreactive (n = 3).
We had previously established the relationship between GCaMP
signals and ring frequency in orexin- tTA mice (34). For ber
photometry recordings, an optic ber was inserted into the LHA.
To quantify the activity levels between states from the recordings
of six mice, we extracted every 10- min episode, which includes
all three vigilance states (146 episodes, each stage (wakefulness,
NREM, and REM sleep) > 1 min; Fig. 1B) and all transitions
from NREM to REM sleep (171 transitions, each stage > 1 min;
Fig. 1B). Consistent with previous reports (29–31), we found that
orexin neurons were highly active during wakefulness compared
to NREM and REM sleep (Fig. 1 C and D and SI Appendix,
Fig. S1 A, E, and F). However, orexin neurons also exhibited inter-
mittent activity during NREM sleep and continuous low- level
activity (relative to Wakefulness) during REM sleep (Fig. 1 C–F).
During NREM sleep, the activity of orexin neurons was correlated
with a decrease in spectral power in the theta and alpha EEG
bandwidths (SI Appendix, Fig. S1 A–D). Elevated activity of orexin
neurons also occurred during microarousals (SI Appendix, Fig. S1
G and H). Analysis of all 171 transitions recorded from NREM
to REM sleep revealed a reduction of activity prior to almost every
transition to REM sleep (166 out of 171; Fig. 1E and SI Appendix,
Fig. S2A). Specically, the mean activity of orexin neurons in each
mouse was signicantly reduced in the 30 sec prior to the transi-
tion from NREM to REM sleep (tNR: dened as the last 30 sec
just prior to REM sleep, See also Materials and Methods) compared
to NREM sleep and subsequent REM sleep (Fig. 1F). ese results
suggest that orexin neurons become quiescent prior to the transi-
tion to REM sleep.
For direct observation of the activity dynamics of orexin neu-
rons at single- cell resolution, we undertook microendoscopy
recordings using nVista (Fig. 2 A and B). Of the 137 cells
recorded from ve mice, approximately half of the orexin neurons
(47.4%, 65 cells) were essentially quiescent but showed occa-
sional intermittent activity during NREM sleep, and a smaller
subpopulation was also active during REM sleep (Fig. 2 C–F and
Movie S1). Of the recorded cells, almost one- third (32.1%, 44
cells) were classied as belonging to the subpopulation active
during REM sleep (Fig. 2 E and F). To further evaluate the inter-
mittent activity of orexin neurons during NREM sleep, we con-
ducted cluster analysis by nonnegative matrix factorization
(NMF) (35) in the ve mice. e synchronous activity (intensity
in Fig. 2G; representative sample) in each NREM cluster
(Materials and Methods) and its contribution to each NREM
cluster (Fig. 2D) was calculated using NMF. To quantify the
change in the neural network, we computed pairwise correlation
coecients for all pairs of NREM- active cluster cells (Materials
and Methods) across vigilance states. We found the largest pro-
portion of correlated (>0.2) pairs during NREM sleep compared
to other states (Fig. 2 H and I).
To evaluate whether the activity patterns of orexin neurons
described above were specic to orexin- tTA mice, we determined
the activity of orexin neurons during sleep in a dierent mouse
strain, heterozygous orexin- Flp (KI/- ) mice (36), which, like
orexin- tTA mice, have orexin neurons that express orexin peptides.
Synchronous cluster activity during NREM sleep and subpopu-
lations active during REM sleep were also observed in orexin- Flp
(KI/- ) mice (SI Appendix, Fig. S3 and Movie S2). us, identical
results in two strains of genetically engineered (knockin and trans-
genic) mice indicate that it is very unlikely that the observed
phenomena could be due to transgene introduction. us, the
occurrence of orexin neuron activity during sleep suggests a pos-
sible role for these cells in the regulation of vigilance states.
Loss of Orexin Peptides Disrupts the Activity Dynamics of Orexin
Neurons during Sleep. Since the orexin neuropeptides have a
critical role in the etiology of narcolepsy (17, 19, 36, 37), we
hypothesized that these peptides play an important role in orexin
neuron activity during sleep in addition to their well- recognized
role in maintenance of wakefulness. To address this hypothesis,
we used orexin- Flp mice that were generated by knocking in the
Flp transgene in- frame at the start codon of the prepro- orexin
gene. Consequently, homozygous orexin- Flp (KI/KI) mice are
unable to synthesize orexin peptides and thus serve as a model of
narcolepsy, while heterozygous orexin- Flp (KI/- ) mice express the
orexin peptides (36). We, therefore, examined changes in activity
dynamics of orexin neurons lacking orexin peptides during sleep
in narcoleptic orexin- Flp (KI/KI) mice.
Orexin- Flp (KI/KI) mice injected with AAV9- CMV- dFRT-
GCaMP6s were subjected to ber photometry (ve mice; Fig. 3A)
and microendoscopy (ve mice; Fig. 4A). e specicity of expres-
sion of GCaMP6s to orexin neurons was conrmed in heterozygous
orexin- Flp (KI/- ) mice (SI Appendix, Fig. S3C). Only 0.2 ± 0.1% of
MCH neurons expressed GCaMP6s, and 0.6 ± 0.3% of GCaMP6s
positive neurons were MCH- immunoreactive (n = 3). A relation-
ship between the GCaMP6s signal and ring frequency was
PNAS 2023 Vol. 120 No. 41 e2301951120 https://doi.org/10.1073/pnas.2301951120 3 of 12
conrmed by simultaneous patch- clamp recordings and calcium
imaging in brain slices. Orexin neurons increased uorescence
intensity in an induced- action potential frequency- dependent man-
ner (SI Appendix, Fig. S4).
As in nonnarcoleptic orexin- tTA mice, orexin neurons in nar-
coleptic orexin- Flp (KI/KI) mice that lacked orexin peptides exhib-
ited higher activity during wakefulness and intermittent activity
during NREM sleep that was closely correlated with the decreased
power of the theta and alpha EEG bandwidths (Fig. 3 B–D and
SI Appendix, Fig. S5 A–F). Also, as in orexin- tTA mice (SI Appendix,
Fig. S1 G and H), orexin neuron activity was correlated with
microarousals in orexin- Flp (KI/KI) mice (SI Appendix, Fig. S5 G
and H). However, in contrast to orexin- tTA mice, orexin neuron
activity in orexin- Flp (KI/KI) mice was lower during REM sleep
than during NREM sleep or the tNR (Fig. 3 C, E, and F). Analysis
of every transition from NREM to REM sleep (121 in total, each
stage > 1 min) revealed that orexin neurons in orexin- Flp (KI/KI)
mice showed the lowest activity during REM sleep in most of the
transitions (103 out of 121; SI Appendix, Fig. S2B). However, the
highest orexin neuron activity during the tNR occurred in 29
transitions (SI Appendix, Fig. S2B). ese results suggest that the
activity during REM sleep and the quiescence prior to the tran-
sition to REM sleep evident in orexin- tTA mice are impaired in
narcoleptic orexin- Flp (KI/KI) mice in which the orexin neurons
lack the orexin peptides.
As in orexin- tTA and orexin- Flp (KI/- ) mice, single- cell microen-
doscopic imaging using nVista in ve orexin- Flp (KI/KI) mice
(Fig. 4 A and B) consistently revealed synchronous activity among
orexin neurons during NREM sleep (Fig. 4 C–E and G–I and
Movie S3). e largest proportion of correlated orexin neuron
pairs in NREM- active cluster cells was observed during NREM
sleep compared to other states (Fig. 4 H and I). However, the
subpopulation of orexin neurons that were active during REM
sleep (Fig. 4 C–F) was signicantly smaller (10.1%, 7 cells) than
Fig.1. Activity recordings of orexin neurons across vigilance states in orexin- tTA mice as determined by ber photometry. (A) Schematics illustrating the injection
of AAV- TetO- GCaMP6s into the LH of orexin- tTA mice to induce expression of GCaMP6s in orexin neurons (Left) and combined ber photometry with EEG/EMG
recordings (Middle). Immunohistochemical conrmation of GCaMP6s expression in orexin neurons (Right). The dashed white lines at the top of each panel
indicate the edge of the optic ber. (B) Schematic of experimental procedures and data analysis. (C) Representative orexin neuronal activity across vigilance
states measured by ber photometry. (D) Summary of orexin neuron activity across vigilance states, expressed as Z- scores (all stages from N = 6 mice). (E) Orexin
neuron activity for all 171 NREM to REM transitions (each stage > 1 min) determined from 24- h recordings in N = 6 mice. (F, Top) Mean Z- score of all NREM to
REM transitions for each individual mouse (gray); mean Z- score of all the N = 6 mice (red). (F, Bottom) Summary of orexin neuron activity during NREM–REM state
transitions for all six mice shown as Z- scores. Values are the mean ± SEM. *P < 0.05, **P < 0.01. Statistical analyses are shown in SIAppendix, TableS1. Vigilance
state parameters of orexin- tTA mice (n = 6) in the ber photometry experiments (24 h) and the 10- min episodes for Z- score (all stages) are described in SIAppendix,
TablesS2 and S3, respectively. tNR, last 30 s during the transition from NREM to REM sleep; PMT, photomultiplier tube; W, wakefulness; NR, NREM; R, REM.
4 of 12 https://doi.org/10.1073/pnas.2301951120 pnas.org
in orexin- tTA mice (Chi- squared test, P = 0.00056) and heterozy-
gous orexin- Flp (KI/- ) mice (Chi- squared test, P = 0.00025; cf.,
Fig. 2 E and F and SI Appendix, Fig. S3 F and G), which express
orexin peptides. e reduced activity of orexin neurons during
REM sleep in the absence of orexin peptides suggests that the
orexin peptides normally contribute to orexin neuron activity
during REM sleep and, furthermore, that this activity may sup-
press the symptoms of narcolepsy such as cataplexy. us, these
results are consistent with the hypothesis that the reduction of
orexin signaling during REM sleep evident in orexin- Flp (KI/KI)
mice is an important contributor to the symptomatology of
narcolepsy.
Orexin Neurons Lacking Orexin Peptides Are Mostly Silent
during Cataplexy. Several neural circuitry models of narcolepsy
have proposed that orexin neurons receive excitatory inputs and
should be activated to prevent the triggering of cataplexy (20–
24, 38). e activity of orexin neurons during “cataplexy- like
attacks” has been described by video recording (39). However,
based on the consensus of the International Working Group on
Fig.2. Activity recordings of orexin neurons across vigilance states in orexin- tTA mice as determined by microendoscopy. (A) Schematics of GCaMP6s expression
(Left) and nVista microendoscopy with EEG and EMG recordings (Right). (B) Representative identication of orexin neurons using nVista. Dashed white line indicates
region of interest (ROI). (C) Representative traces of Ca2+ activity during each vigilance state. (D) Activity of orexin neurons aligned by contribution to the three
NREM clusters in G. (E) Activity during each vigilance state of orexin neurons determined to be active (n = 44 cells) or inactive (n = 93 cells) during REM sleep in
N = 5 mice. (F) Venn diagram showing the number of orexin neurons (133 cells) exhibiting each activity pattern. A total of 137 cells were detected; four neurons
were not classied. (G) Representation of the activity of orexin neurons within NREM clusters 1, 2, and 3. (H) Representative correlations of activity among orexin
neuron pairs from NREM- active clusters 1 and 2 in G. Each dot reects the location of the cells determined by the (x, y) position of the center of the neuron in
the eld of view. (I) Percentage of cells exhibiting activity synchronization, dened by the proportion of correlated (>0.2) pairs during each vigilance state. Values
are the mean ± SEM. *P < 0.05, **P < 0.01. Statistical analyses are shown in SIAppendix, TableS1. W, wakefulness; NR, NREM; R, REM.
PNAS 2023 Vol. 120 No. 41 e2301951120 https://doi.org/10.1073/pnas.2301951120 5 of 12
Rodent Models of Narcolepsy (40), EEG and EMG recordings
in addition to video are necessary to dene cataplexy; otherwise,
the term “abrupt behavioral arrest” is more appropriate when
only video is used. us, the activity of orexin neurons during
cataplexy as measured with EEG and EMG has not previously
been determined in any mouse model of narcolepsy. To address
this knowledge gap, we utilized orexin- Flp (KI/KI) mice (36)
combined with ber photometry (n = 5) and microendoscopy
(n = 3) and found that most orexin neurons in orexin- Flp (KI/
KI) mice were silent during cataplexy but quickly recovered their
activity upon termination of atonia and resumption of the waking
EEG (Fig.5 A–E, SI Appendix, Fig.S6 A–D and Movie S4).
Interestingly, although the number of the recorded neurons may
be small, further analysis through recording both REM sleep and
cataplexy in the same mice revealed that the subpopulation of
orexin neurons that is active during REM sleep and cataplexy
are almost the same [88.9% (eight out of nine neurons) for each
state, SI Appendix, Fig. S6 A–D]. ese results suggest either
increased inhibition, a lack of excitatory inputs or both during
cataplexy. Given the high frequency of spontaneous ring of
orexin neurons (36, 41), inhibitory input is likely necessary to
suppress spontaneous ring of orexin neurons during cataplexy.
Fig.3. Activity recordings of orexin neurons across vigilance states in prepro- orexin knockout [orexin- Flp (KI/KI)] mice as determined by ber photometry.
(A) Schematics illustrating the injection of AAV- CMV- dFRT- GCaMP6s into homozygous orexin- Flp (KI/KI) mice to induce expression of GCaMP6s expression in
orexin neurons (Left) and ber photometry with EEG/EMG recordings (Middle). Histochemical conrmation of GCaMP6 expression in orexin neurons (Right).
The dashed line at the top of the panel indicates the edge of the optic ber. (B) Schematic of the experimental procedures and data analysis. (C) Representative
orexin neural activity across vigilance states measured by ber photometry. (D) Summary of orexin neuron activity across vigilance states, expressed as Z- scores
(all stages from N = 5 mice). (E) Activity of orexin neurons from all 121 NREM to REM transitions recorded (each stage > 1 min) during 24- h recordings in N = 5
mice. (F, Top) Mean Z- score of all the transitions in each individual mouse (gray); mean Z- score of all the N = 5 mice (red). (F, Bottom) Summary of orexin neuron
activity of all the mice during NREM–REM state transitions shown as Z- scores. Values are the mean ± SEM. *P < 0.05, **P < 0.01. Statistical analyses are shown in
SIAppendix, TableS1. Vigilance state parameters of orexin- Flp (KI/KI) mice (n = 5) in the ber photometry experiments (24 h) and the 10- min episodes for Z- score
(all stages) are described in SIAppendix, TablesS2 and S3, respectively. tNR, the last 30 s during the transition from NREM to REM sleep; PMT, photomultiplier
tube; W, wakefulness; NR, NREM; R, REM.
6 of 12 https://doi.org/10.1073/pnas.2301951120 pnas.org
Optogenetic Silencing of Orexin Neurons Increases NREM–
REM Sleep Transitions. Frequent transitions into REM sleep
are characteristic of both the clinical presentation of narcolepsy
and animal models of this disorder (19, 24, 25). From repeated
and detailed single- cell recordings by both ber photometry and
microendoscopy, we found that orexin neurons became essentially
quiescent prior to the transition from NREM sleep to REM sleep
(Figs.1 and 2 and SI Appendix, Fig.S3). While these ndings
suggest that the activity dynamics of orexin neurons during
NREM sleep play a regulatory role in the transition from NREM
sleep to REM sleep (NREM–REM sleep transition), optogenetic
inhibition of orexin neurons was previously shown to only increase
NREM sleep time (42–44). us, we tested whether optogenetic
silencing of orexin neurons increased NREM–REM sleep
Fig.4. Activity recordings of orexin neurons across vigilance states in prepro- orexin knockout [orexin- Flp (KI/KI)] as determined by microendoscopy. (A) Schematics
of GCaMP6s expression (Left) and nVista microendoscopy with EEG and EMG recordings (Right). (B) Representative identication of orexin neurons using nVista.
Dashed white line indicates region of interest (ROI). (C) Representative traces of Ca
2+
activity during each vigilance state. (D) Activity of orexin neurons during NREM
sleep aligned by activity cluster. (E) Activity during each vigilance state of orexin neurons determined to be active (7 cells) or inactive (62 cells) in REM sleep in
N = 5 mice. (F) Venn diagram showing the number of orexin neurons (47 cells) exhibiting each activity pattern. A total of 69 cells were detected; 22 neurons were
not classied. (G) Representation of orexin neuron activity within NREM clusters 1, 2, and 3. (H) Representative correlations of activity among orexin neuronal
pairs in the NREM- active clusters 1 and 2 in G. Each dot reects the location of the cells determined by the (x, y) position of the center of the neuron within the
eld of view. (I) Percentage of cells exhibiting activity synchronization, dened as the proportion of correlated (>0.2) pairs, during each vigilance state. Values
are the mean ± SEM. *P < 0.05, **P < 0.01. Statistical analyses are shown in SIAppendix, Table S1. tNR, the last 30 s during the transition from NREM to REM
sleep; W, wakefulness; NR, NREM; R, REM.
PNAS 2023 Vol. 120 No. 41 e2301951120 https://doi.org/10.1073/pnas.2301951120 7 of 12
transitions. e neural silencer Archaerhodopsin- T (ArchT) fused
with enhanced green uorescent protein (EGFP) was exclusively
expressed in orexin neurons following bilateral injection of AAV9-
TetO- ArchT- EGFP into the LHA of orexin- tTA mice (Fig.6 A and
B). Expression of ArchT and optogenetic inhibition was validated
by immunostaining and patch- clamp recordings, respectively
(SIAppendix, Fig.S7).
One- h continuous inhibition of orexin neurons during
Zeitgeber time (ZT) seven to eight signicantly decreased the time
in wakefulness and increased the time in NREM and REM sleep
compared to baseline (Fig. 6 C–E). Importantly, we found that
the transition ratio (REM bouts/NREM bouts) and the cumula-
tive probability of transitioning from NREM sleep to REM sleep
signicantly increased (Fig. 6H) whereas these parameters from
NREM sleep to wakefulness decreased (Fig. 6G), suggesting a
disproportionate increase in the transition from NREM sleep to
REM sleep when orexin neurons are inhibited. EGFP- expressing
control mice did not show any signicant change in these
parameters (SI Appendix, Fig. S8 A–E). Optogenetic inhibition of
orexin neurons also signicantly increased the hypersynchronous
paroxysmal theta (HSPT) bursts that are characteristic of REM
sleep and cataplexy in other mouse models of narcolepsy (45, 46),
indicating that optogenetic inhibition of orexin neurons was suc-
cessful (Fig. 6 D and F). Spectral analysis showed that EEG power
was aected, but the dierence was not signicant (SI Appendix,
Fig. S9A). Moreover, to test whether intermittent orexin neuron
activation during NREM sleep prevented the NREM–REM sleep
transition, we also conducted an intermittent stimulation of orexin
neurons (four times/min) for 1 h at ZT6- 7 using bigenic
orexin- tTA; TetO- ChR2 (47) mice. is intermittent stimulation
protocol reduced the transition ratio and cumulative probability
of the NREM–REM sleep transition, without reducing NREM
and REM sleep duration (SI Appendix, Fig. S10 A–H). Given that
only the activity during NREM sleep is thought to contribute to
the subsequent transition from NREM sleep, these results also
support the idea that the intermittent activity of orexin neurons
Fig.5. Activity of orexin neurons during cataplexy in prepro- orexin knockout [orexin- Flp (KI/KI)] mice. (A) Representative trace of orexin neuron activity during
cataplexy as measured by ber photometry. (B) Activity of orexin neurons of all the state transitions (wake to cataplexy and cataplexy to wake, each stage > 40 s),
measured from 24- h recordings in N = 5 mice. (C, Top) Mean Z- score of all the transitions in each individual mouse (gray); mean Z- score of all the N = 5 mice (red).
(C, Bottom) Summary of orexin neuron activity of all the mice (wake to cataplexy and cataplexy to wake) shown as Z- scores. (D) Representative traces of the activity
of 20 orexin neurons during cataplexy measured using nVista. (E) Summary of orexin neuron activity during transitions between wake and cataplexy for N = 50
orexin neurons from N = 3 mice. Values are the mean ± SEM. **P < 0.01. Statistical analyses are shown in SIAppendix, TableS1. C, cataplexy; W, wakefulness.
8 of 12 https://doi.org/10.1073/pnas.2301951120 pnas.org
during NREM sleep prevents the NREM–REM sleep transition,
whereas a decrease in orexin neuron activity during NREM sleep
promotes the NREM–REM sleep transition (Fig. 6I).
Orexin neurons release not only orexin peptides but also other
neurotransmitters such as glutamate and dynorphin. us, the role
of orexin peptides in the regulation of NREM–REM sleep transi-
tions by orexin neurons remains unknown. To address this question,
AAV9- CMV- dFRT- ArchT- EGFP was bilaterally injected into the
LHA of orexin- Flp (KI/KI) mice and orexin neurons that lacked
orexin peptides were subjected to optogenetic inhibition (Fig. 6 J
and K). e exclusive expression of ArchT- EGFP in orexin neurons
was conrmed using heterozygous orexin- Flp (KI/- ) mice and func-
tional expression of ArchT in orexin neurons lacking orexin peptides
was conrmed by patch- clamp recordings (SI Appendix, Fig. S11
A–G). In contrast to inhibiting orexin neurons that express orexin
peptides in orexin- tTA mice, inhibition of orexin neurons lacking
orexin peptides in orexin- Flp (KI/KI) mice did not aect total time
spent in each state, the transition ratio, HSPT bouts (Fig. 6 L–Q)
or the EEG spectrum (SI Appendix, Fig. S9B). GFP- expressing con-
trol mice did not show any signicant change in these parameters
(SI Appendix, Fig. S8 F–J). ese results suggest that the orexin
peptides are a critical component in the regulation of NREM–REM
sleep transitions by orexin neuron activity during NREM sleep.
Orexin Neuron Activity during REM Sleep Aects Subsequent
REM Sleep Architecture and Cataplexy. e results from ber
photometry and continuous optogenetic inhibition suggest an
inhibitory role for orexin neuron activity in NREM–REM sleep
transitions. However, the signicance of orexin neuron activity
during REM sleep, which is lower in narcolepsy model mice
(Figs.1F, 2F, and SIAppendix, Fig.S3G vs. 3F and 4F), remains
unclear. It has long been thought that orexin neurons have REM
sleep–suppressing activity (8, 17, 25). us, we hypothesized
that orexin neuron activity during REM sleep is involved in
the increased REM sleep characteristic of narcolepsy. To test
this hypothesis, we took advantage of previous studies in which
published gures implied that REM sleep is increased not only
during the dark period, but also during the latter portion (last 4
h) of the light period in narcolepsy model mice compared to wild-
type mice (4, 19). We conrmed this observation in narcoleptic
orexin- Flp (KI/KI) mice compared to both orexin- tTA mice and
wild- type mice. In orexin- Flp (KI/KI) mice, REM sleep during
Fig.6. Eects of optogenetic inhibition of orexin neurons on NREM–REM transitions in mice with or without orexin peptide expression. (A) Schematics of ArchT-
EGFP expression in orexin neurons in orexin- tTA mice (Left) and implantation of bilateral optic bers with EEG/EMG recordings (Right). (B) Immunohistochemical
conrmation of ArchT- EGFP expression in orexin neurons in orexin- tTA mice (N = 6 mice). Dashed lines at the top of each image indicate the edge of the optic
ber. (C) Hypnogram (ZT7- 8) during baseline (Upper) and optogenetic inhibition (Lower). (D) Representative EEG during REM sleep at baseline (Upper) and during
optogenetic inhibition (Lower). Insets are EEG spectra of the indicated epochs. Note episodes of hypersynchronous paroxysmal theta (HSPT) in the EEG. (E) Total
time in each vigilance state during baseline (gray) and optogenetic inhibition (green). (F) The number of HSPT bouts during baseline (gray) and optogenetic
inhibition (green) in orexin neurons from orexin- tTA mice. (G) Optogenetic inhibition of orexin neurons decreases the transition ratio (Left) and cumulative
probability (Right) for the NREM sleep to Wake transition in orexin- tTA mice. (H) Optogenetic inhibition of orexin neurons increases the transition ratio (Left)
and cumulative probability (Right) for the NREM to REM sleep transition in mice expressing the orexin peptides (orexin- tTA mice). (I) Schematic summarizing
the eect of optogenetic inhibition of orexin neurons in orexin- tTA mice. (J) Schematics of ArchT- EGFP expression in orexin neurons from orexin- Flp (KI/KI) mice
which lack orexin peptides (N = 6 mice). (K) Histochemical conrmation of ArchT- EGFP expression in orexin- Flp (KI/KI) mice; dashed lines indicate the location of
the optic ber. (L) Hypnogram (ZT7- 8) during baseline (Upper) and optogenetic inhibition (Lower). (M) Total time in each vigilance state during baseline (gray) and
optogenetic inhibition (green) of orexin neurons in orexin- Flp (KI/KI) mice. (N) The number of HSPT bouts during baseline (gray) and optogenetic inhibition (green).
(O) Transition ratio (Left) and cumulative probability (Right) for the transition from NREM sleep to Wake. (P) Transition ratio (Left) and cumulative probability
(Right) for the transition from NREM to REM sleep. (Q) Schematic summarizing the absence of any eect of optogenetic inhibition of orexin neurons in orexin- Flp
(KI/KI) mice. Values are the mean ± SEM. *P < 0.05, **P < 0.01. Statistical analyses are in SIAppendix, TableS1. HSPT, Hypersynchronous paroxysmal theta burst.
PNAS 2023 Vol. 120 No. 41 e2301951120 https://doi.org/10.1073/pnas.2301951120 9 of 12
ZT8- 12 and during ZT12- 18 is signicantly increased compared
to phenotypically normal orexin- tTA mice (Fig.7 A and B) and
wild- type mice (6.7 ± 0.4% vs. 7.9 ± 0.4% (not signicant) during
ZT 3 to 8, 8.4 ± 0.6% vs. 6.2 ± 0.3% (P < 0.01) during ZT8- 12,
and 4.9 ± 0.8% vs. 1.7 ± 0.3% (P < 0.01) during ZT 12 to 18,
SIAppendix, TableS1).
Since sleep disruption at night increases sleepiness during the
daytime in humans, we similarly hypothesized that orexin neuron
activity during REM sleep in the preceding period (ZT3- 8)
aects the increase in subsequent REM sleep from ZT8- 12, as
shown in narcoleptic mice (Fig. 7 A and B). To test this, we
performed state- dependent optogenetic inhibition of orexin neu
-
rons from ZT3 to ZT8 in orexin- tTA and analyzed the eect on
vigilance states during the subsequent 10 h (light period: ZT8- 12;
dark period: ZT12- 18) after inhibition had been terminated.
Sleep/wakefulness states were automatically discriminated using
Fig.7. REM sleep state–dependent inhibition of orexin neurons with and without orexin peptides regulates subsequent REM sleep architecture and cataplexy.
(A and B) Line graphs showing the time spent in each vigilance state (A) and bar graphs summarizing the time spent in REM sleep during each time period (B) in
orexin- tTA mice (black, N = 6 mice) and prepro- orexin knockout mice [orexin- Flp (KI/KI); red, N = 6 mice]. (C) Decision tree algorithm for real- time vigilance- state
determination using EEG, EMG, and locomotion. (D and E) Photoillumination of orexin neurons expressing ArchT- EGFP during REM sleep results in REM sleep
state–dependent inhibition. (F and G) Bar graphs indicating the illumination “cover ratio” for REM sleep in orexin- tTA mice (F) and prepro- orexin knockout mice
[orexin- Flp (KI/KI); G]. (H and I) Line graphs showing the eects of REM sleep state–dependent inhibition on the time spent in each vigilance state during state-
dependent illumination (ZT3- 8), the subsequent light period (ZT8- 12), and the subsequent dark period (ZT12- 18) in orexin- tTA mice (H) and prepro- orexin knockout
[orexin- Flp (KI/KI)) mice (I)]. (J–M) Bar graphs showing the eects of REM sleep state–dependent inhibition on the time spent in REM sleep and cataplexy at each
time period (ZT3- 8, ZT8- 12, and ZT12- 18) in orexin- tTA mice (Left in J–L) and prepro- orexin knockout mice [orexin- Flp (KI/KI); Right in J–L and M). Statistics for other
vigilance states during each time period are described in SIAppendix, TableS1. Data are the mean ± SEM. *P < 0.05, **P < 0.01. Statistical analyses are shown in
SIAppendix, TableS1. FFT, fast Fourier transform; TTL, transistor–transistor logic.
10 of 12 https://doi.org/10.1073/pnas.2301951120 pnas.org
EEG, EMG, and locomotor activity, which enabled immediate
closed loop- triggering of photoillumination during specic states
(48) (Fig. 7 C–E and SI Appendix, Fig. S14 A and B). REM sleep
state–dependent triggering of photoillumination spanned 84.3
± 6.7% of REM sleep (total time) with little illumination during
nontarget states (Fig. 7F). e time spent in vigilance states under
REM sleep state–dependent inhibition (ZT3- 8) did not dier
from baseline (Fig. 7 H and J, Left) or yoked (joined together)
controls in which the subjected mice (yoked controls) received
optogenetic inhibition whenever the matched mice showed REM
sleep (SI Appendix, Fig. S12 A–C, E, and G, Left). However, the
time spent in subsequent REM sleep (ZT8- 12) in orexin- tTA
mice signicantly increased compared to baseline (Fig. 7H and
K, Left) and yoked control (SI Appendix, Fig. S12 E and H, Left),
but not in the EGFP- expressing control mice (SI Appendix,
Fig. S13 A–C and H, Left). After NREM sleep–specic inhibition
(SI Appendix, Fig. S14 A–C), a signicant reduction in NREM
sleep and an increase in wakefulness during the dark period were
also observed (SI Appendix, Fig. S14 E and F). In contrast,
wakefulness- specic inhibition (SI Appendix, Fig. S14 A–C) did
not aect vigilance states (SI Appendix, Fig. S14 G and H). us,
the inhibition of orexin neuron activity during REM sleep at a
time of day when REM sleep is normally high caused a subse-
quent overexpression of REM sleep. erefore, it produced a
REM sleep increase in orexin- tTA mice at ZT8- 12, but not at
ZT3- 8, similar to the REM sleep structure at ZT3- 8 and ZT8- 12
in narcolepsy.
ese results encouraged us to hypothesize that REM sleep
state–dependent inhibition of orexin neurons lacking orexin pep-
tides might exacerbate subsequent cataplexy in narcolepsy model
mice. Orexin neurons in orexin- Flp (KI/KI) mice which lack orexin
peptides were state- dependently inhibited during REM sleep from
ZT3- 8. REM sleep state–dependent triggering of photoillumina-
tion spanned 74.0 ± 5.6% of REM sleep (total time) with little
illumination during nontarget states (Fig. 7G). Time in vigilance
states did not dier among time periods (Fig. 7 I and J–L, Right).
However, time in cataplexy in orexin- Flp (KI/KI) mice signicantly
increased during the dark period after REM sleep state–dependent
inhibition compared to baseline (Fig. 7 I and M) and yoked con-
trols (SI Appendix, Fig. S12 D, F, and J), but not in the
GFP- expressing control mice (SI Appendix, Fig. S13 D–F and J).
us, our results suggest that orexin neuron activity during
preceding REM sleep has an inhibitory role on subsequent cata-
plexy during the dark period. Taken together, these results show
that the activity dynamics of orexin neurons during NREM and
REM sleep are involved in the abnormal REM sleep architecture
in narcolepsy.
Discussion
Here, we revealed the activity dynamics and physiological roles of
orexin neurons during sleep in both orexin- tTA mice and orexin- Flp
(KI/- ) mice and in orexin- Flp (KI/KI) narcolepsy model mice
(SI Appendix, Fig. S15). Orexin neurons in both orexin- tTA mice
and orexin- Flp (KI/- ) mice showed intermittent and synchronous
activity in NREM sleep but were essentially quiescent before tran-
sitions to REM sleep (tNR), suggesting that orexin neuron activity
during NREM sleep is involved in the regulation of the REM sleep
transition. is role was also conrmed by the disproportionate
increase or decrease in REM sleep transitions by continuous inhi-
bition or intermittent activation, respectively, of orexin neurons
in orexin- tTA mice. A subpopulation of orexin neurons showed
weak activity during REM sleep. REM sleep state–dependent inhi-
bition revealed that orexin neuron activity during the preceding
REM sleep has an inhibitory role in subsequent REM sleep archi-
tecture in orexin- tTA mice and cataplexy in orexin- Flp (KI/KI) mice.
Although the role of orexin neurons in promoting wakefulness is
well established (20–24), our results revealed that the loss of the
normal roles for orexin neurons during sleep as well as during
wakefulness are involved in the REM sleep–related symptomatol-
ogy of narcolepsy.
Orexin neurons showed dierent activity patterns between
NREM sleep and REM sleep, suggesting that these cells receive
distinct neural regulation during NREM vs. REM sleep. Li et al.
reported that hyperexcitable orexin neurons drive sleep fragmen-
tation during aging through intermittent activity during NREM
sleep (33). e intermittent synchronous activity during NREM
sleep might be generated intrinsically by burst activity from orexin
neurons. In slice patch- clamp recordings, some orexin neurons
showed intermittent burst activity (49). We also reported that
orexin neurons form a positive- feedback loop both directly and
indirectly (50). ese neural mechanisms might form the basis of
synchronous activity and this synchronous activity in NREM sleep
could be resisted by inhibitory inputs from sleep- promoting neu-
rons such as the preoptic area during NREM sleep (51, 52).
is intermittent activity during NREM sleep may push the
brain state toward wakefulness and thus may contribute to the
generation of a “microarousal”- like state in NREM sleep, which is
often observed in rodents. Microarousal has also been associated
with the activity of orexin neurons and might help animals stay
alert during NREM sleep for survival. is activity inhibits
NREM–REM sleep transitions and, importantly, orexin peptides
appear to be indispensable for this inhibitory role. Loss of this
function could be one of neural mechanisms behind the frequent
transitions in narcolepsy (17, 19, 25, 36, 53–55). Weber et al. (7)
reported a gating role for GABAergic neurons in the ventrolateral
periaqueductal gray (vlPAG) in the transition from NREM sleep
to REM sleep. Orexin neurons are known to project to vlPAG (56)
and thus may interact to suppress the transition to REM sleep.
We also observed activity in a subpopulation of orexin neurons
during REM sleep. However, this REM activity was diminished
in orexin neurons from orexin- Flp (KI/KI) mice which lack the
orexin peptides, suggesting that the orexin peptides contributed
to the generation of orexin neuron activity during REM sleep. We
previously reported impaired activity of orexin neurons that lacked
the orexin peptides when recorded by patch- clamp electrophysi-
ology (36). Orexin neuron activity during REM sleep may be
more susceptible to impairment due to loss of orexin peptides
than other states because orexin neurons exhibit a continuous,
low activity during REM sleep by a smaller population of cells.
Moreover, we found that orexin neuron activity was suppressed
in cataplexy and appeared similar to orexin neuron activity during
REM sleep in orexin- Flp (KI/KI) mice. It has previously been
reported that other wake- promoting neurons, such as histamin-
ergic neurons in the tuberomammillary nucleus (TMN), are
highly active in cataplexy as well as wakefulness (57). e inverse
activities between orexin neurons and other wake- promoting neu-
rons during cataplexy could underlie the neuronal basis of cata-
plexy or sleep paralysis with “sleep- like” muscle atonia under
“wakefulness- like” consciousness.
Orexin signaling during REM sleep suppressed an increase in
subsequent REM sleep in phenotypically normal orexin- tTA mice
and cataplexy in orexin- Flp (KI/KI) narcolepsy model mice. is
observation supports the idea that orexin signaling during REM
sleep relieves/reduces the requirement for REM sleep. e previous
ndings that orexin peptides strongly inhibit REM sleep and that
REM sleep pressure accumulates in narcolepsy (25) are consistent
with our present results. Interestingly, 48 h of REM sleep
PNAS 2023 Vol. 120 No. 41 e2301951120 https://doi.org/10.1073/pnas.2301951120 11 of 12
deprivation has been found to increase subsequent cataplexy in
narcoleptic mice (4). Optogenetic inhibition of orexin signaling
during REM sleep in orexin- Flp (KI/KI) mice replicated this phe-
nomenon at the neural circuitry level. us, orexin signaling in
REM sleep could be one of the neural mechanisms regulating
REM sleep requirements or pressure. e exacerbation of cata-
plexy by REM sleep state–dependent inhibition of orexin neurons
that lack orexin peptides also suggests colocalized neurotransmit-
ters, such as glutamate, may play a role in the inhibition of cata-
plexy. In this regard, mice lacking orexin neurons generally exhibit
more severe narcolepsy symptoms than prepro- orexin knockout or
orexin receptor knockout mice (24, 54).
State- dependent inhibition of orexin neurons aected subsequent
vigilance states (ZT8- 12) but not during inhibition (ZT3- 8). A pos-
sible explanation for this could be that the accuracy of state- dependent
inhibition was ~90% and not 100%. e algorithm underlying our
closed- loop photoillumination system requires one 4- s epoch to
determine the vigilance state, which means that orexin peptides can
be released during the initial 4 s. It is also known that orexin peptides
act via volume transmission. Trace levels of orexin peptides in the
cerebrospinal uid released by such “leak activity” or by activity
during other states might weaken the eect of state- dependent inhi-
bition. In fact, trace levels of orexin peptides in the basal forebrain
were reported to remain similarly higher during both NREM sleep
and wakefulness, but reduced during REM sleep (58). Alternatively,
optogenetic inhibition may be followed by rebound neuronal hyper-
activity due to hyperpolarization- induced current (H current) in
orexin neurons (49). is rebound activity associated with intermit-
tent photoinhibition at each state may weaken the eect of
state- dependent inhibition. ese were considered to be methodo-
logical limitations of state- dependent inhibition.
As another possibility, in narcolepsy, REM sleep is greater dur-
ing the last 4 h (ZT8- 12) of the light period compared to WT as
shown in the Results (Fig. 7 A and B), but not during the typical
sleep period (ZT3- 8). us, it is possible that state- dependent
inhibition from ZT3- 8 did not increase REM sleep during this
typical sleep period (ZT3- 8) when relatively more REM sleep
normally occurs but increases REM sleep during the latter part of
the light period (ZT8- 12) when a relatively small amount of REM
sleep normally occurs. It is also reasonable that “the rebound REM
as cataplexy” occurs in the dark period with a 4- h delay because
wakefulness (which is dominant during the dark period) is nec-
essary for the occurrence of cataplexy.
Another limitation is that we could not record cataplexy in all ve
orexin- Flp (KI/KI) mice in which microendoscopy recording was con-
ducted, but recorded cataplexy only in three mice that had REM- active
orexin neurons. is was because, due to the limited recording period
for microendoscopy experiments, it was dicult to observe a spon-
taneously occurring rare event such as cataplexy during such record-
ings, particularly given the preceding 40- s wakefulness criterion
required by the international consensus. Although we observed the
activity of orexin neurons during REM sleep and cataplexy, the num-
ber of neurons recorded during both REM sleep and cataplexy is
relatively small. us, to clarify this point, further studies may be
warranted. Moreover, although we used chocolate to induce cataplexy,
it is unknown whether the circuit underlying chocolate- induced cat-
aplexy is the same as for spontaneous cataplexy.
Here, we identied the roles for orexin neuron activity dynam-
ics in NREM and REM sleep using mice with and without orexin
peptides. Taken together, our ndings provide important clues
not only for the development of treatments for narcolepsy, but
also for understanding the neural regulatory mechanisms under-
lying sleep/wakefulness states.
Materials and Methods
A full description of experimental materials and methods are in SIAppendix,
Supplementary Materials and Methods.
Animals. All experimental protocols involving the use of mice were approved
by the Institutional Animal Care and Use Committees, Research Institute of
Environmental Medicine, Nagoya University, Japan (#19232 and #19268) and
SRI International (#01026). All efforts were made to reduce the number of mice
used and to minimize the pain and suffering of mice.
Adeno- Associated Viral (AAV) Production and Injection. AAV Helper- Free
System (Agilent Technologies, Inc., Santa Clara, CA, USA), AAV vectors were used
to produce and purify. HEK293 cells were transfected with pAAV vector plasmid
that included a gene of interest, pHelper and pAAV- RC using a standard calcium
phosphate method.
For AAV injection, mice were anesthetized with isoflurane (Wako Pure
Chemical Industries) anesthesia (< 2%) and fixed in a stereotaxic frame (David
Kopf Instruments). Using injectors (BJ- 110; BEX CO, Ltd., Itabashi, Tokyo, Japan
or Nanoject III; Drummond Scientific Company, Broomall, PA, USA), AAV was
injected into the LHA
EEG/EMG Surgery. Two screws (U- 1430- 01, Wilco, Yokohama, Japan) were
implanted on the skull (AP = +0.5 to 1.0 mm; ML = −0.5 to −1.0 mm and AP
= −2.5 to 3.0 mm; ML = −2.5 to −3.0 mm) to record EEG, and two stainless
steel wires (209- 4811, RS PRO, Yokohama, Japan) were inserted on either side
of the nuchal muscle to record EMG.
In Vivo Ca
2+
Imaging Using nVista.
More than 3 wk after AAV injection, a GRIN
lens (length of 8.4 mm, diameter of 1 mm, Inscopix, Palo Alto, CA, USA) was
implanted above the LHA (AP = −1.4 mm; ML = +0.9 mm; DV = −4.6 to 5.0
mm). At the same time, an acrylic bar for head fixation was attached to the skull
using Super- bond (Super- bond C&B, Sun Medical, Moriyama, Japan). More
than 1 wk after implantation, the mouse was attached to a stereotaxic frame
using the acrylic bar. A baseplate (Inscopix, Palo Alto, CA) was attached using
dental cement (REPAIRSIN, GC) to hold a microendoscope (nVista, Inscopix).
Then the microendoscope was attached to the head to monitor GCaMP6s fluo-
rescence through the implanted GRIN lens. Images were acquired at 10 to 20
frames/s with 0.2 to 1.1 mW of LED light (475 nm) using nVista HD Acquisition
Software (version 3; Inscopix). All images were processed using Mosaic Software
(version 2.0; Inscopix) and Inscopix Data Processing Software (IDPS, version
1.6.1, Inscopix).
Optogenetic Inhibition/Stimulation and Vigilance State–Dependent
Illumination. At least 3 wk after AAV injection, LED cannulae (fiber diameter
500 µm, fiber length 5 mm, bilateral, 525 nm, 3.2 mW; Bio Research Center)
were implanted (AP = −1.4 mm; ML = ±0.9 mm; DV = −4.5 mm) during EEG/
EMG surgery. LED cannulae (fiber diameter 400 µm, fiber length 5 mm, bilateral;
Kyocera Corporation, Japan) were implanted (AP = −1.4 mm; ML = ±0.9 mm;
DV = −4.5 mm). The vigilance state for each epoch was automatically defined
in real time using the following decision tree algorithm by SleepSignRecorder
(Kissei Comtec). Photo illumination was triggered by Transistor- Transistor Logic
(TTL) output from SleepSignRecorder. After the experiments, the cover ratio of
illumination for each vigilance state was calculated by comparison with offline
determination.
Electrophysiology.
Brains were rapidly isolated and chilled in ice- cold bubbled
(95% O2 and 5% CO2) cutting solution. Coronal brain sections of 300- µm thick-
ness were made using a vibratome (VT- 1200S, Leica). The slices were incubated in
a bubbled (95% O2 and 5% CO2) bath solution. electrophysiological properties of
the cells were continuously monitored using the Axopatch 200B amplifier (Axon
Instruments, Molecular Devices, Sunnyvale, CA). Patch- clamp data were recorded
using an analog- to- digital (AD) converter (Digidata 1550A, Molecular Devices)
and pClamp 10.7 software (Molecular Devices).
Statistical Analysis. Statistical analyses were performed using OriginPro 2020
software (LightStone, Tokyo, Japan), easy R (1.37), or Python (3.7). All data are
presented as the mean ± SEM. Details of the statistical tests are described in
SIAppendix, TableS1. Significant differences were set at P < 0.05.
12 of 12 https://doi.org/10.1073/pnas.2301951120 pnas.org
Data, Materials, and Software Availability.
All study data are included in the
article and/or supporting information.
ACKNOWLEDGMENTS. We thank Dr. Nomoto and Dr. Inokuchi at the University
of Toyama for introducing the NMF clustering analysis. We thank S. Tsukamoto,
S. Nasu, E. Imoto, and T. Miyazaki for technical assistance and helpful support.
We thank the Center for Animal Research and Education at Nagoya University
for breeding animals. This work was supported by JST CREST (JPMJCR1656),
AMED- CREST (JP20GM1310007), Mitsubishi foundation, Senshin- iyaku founda-
tion and Kao Kenkokagaku foundation to A.Y., KAKENHI grants (19J22270) to H.I.,
(21K20688 and 22K15225) to Y.M., and (21H02526, 20KK0177 and 18H02477)
to D.O., (26293046, 26640041, 16H01271, 17H05563, 18H05124, 18KK0223
and 18H02523) to A.Y. and NIH R01 NS098813 and R01 NS103529 to T.S.K.
Author aliations: aDepartment of Neuroscience II, Research Institute of Environmental
Medicine, Nagoya University, Nagoya 464- 8601, Japan; bDepartment of Neural Regulation,
Nagoya University Graduate School of Medicine, Nagoya 466- 8550, Japan; cJapan Society
for the Promotion of Science Research Fellowship for Young Scientists, Tokyo 102- 0083,
Japan; dCenter for Neuroscience, Biosciences Division, SRI International, Menlo Park, CA
94025; eChinese Institute for Brain Research, Beijing 102206, China; fNational Institute
for Physiological Sciences, Aichi 444- 8585, Japan; gNational Institutes of Natural Sciences,
Aichi 444- 8585, Japan; and hDivision of Brain Sciences Institute for Advanced Medical
Research, Keio University School of Medicine, Tokyo 160- 8582, Japan
1. A. Rechtschaffen, B. M. Bergmann, M. A. Gilliland, K. Bauer, Effects of method, duration, and sleep
stage on rebounds from sleep deprivation in the rat. Sleep 22, 11–31 (1999).
2. J. A. Horne, REM sleep - by default? Neurosci. Biobehav. Rev. 24, 777–797 (2000).
3. H. Feng et al., Orexin signaling modulates synchronized excitation in the sublaterodorsal tegmental
nucleus to stabilize REM sleep. Nat. Commun. 11, 3661 (2020).
4. A. Roman, S. Meftah, S. Arthaud, P. H. Luppi, C. Peyron, The inappropriate occurrence of rapid
eye movement sleep in narcolepsy is not due to a defect in homeostatic regulation of rapid eye
movement sleep. Sleep 41, zsy046 (2018).
5. S. Arthaud, P. A. Libourel, P. H. Luppi, C. Peyron, Insights into paradoxical (REM) sleep homeostatic
regulation in mice using an innovative automated sleep deprivation method. Sleep 43, zsaa003 (2020).
6. E. Aserinsky, N. Kleitman, Regularly occurring periods of eye motility, and concomitant phenomena,
during sleep. Science 118, 273–274 (1953).
7. F. Weber et al., Regulation of REM and Non- REM sleep by periaqueductal GABAergic neurons. Nat.
Commun. 9, 354 (2018).
8. S. H. Park, F. Weber, Neural and homeostatic regulation of REM sleep. Front. Psychol. 11, 1662 (2020).
9. E. Hasegawa et al., Rapid eye movement sleep is initiated by basolateral amygdala dopamine
signaling in mice. Science 375, 994–1000 (2022).
10. A. K. Barnes, R. Koul- Tiwari, J. M. Garner, P. A. Geist, S. Datta, Activation of brain- derived
neurotrophic factor- tropomyosin receptor kinase B signaling in the pedunculopontine tegmental
nucleus: a novel mechanism for the homeostatic regulation of rapid eye movement sleep. J.
Neurochem. 141, 111–123 (2017).
11. S. Datta, C. M. Knapp, R. Koul- Tiwari, A. Barnes, The homeostatic regulation of REM sleep: A role for
localized expression of brain- derived neurotrophic factor in the brainstem. Behav. Brain Res. 292,
381–392 (2015).
12. Y. Mukai, A. Yamanaka, Functional roles of REM sleep. Neurosci. Res., 10.1016/j.
neures.2022.12.009 (2022).
13. L. de Lecea et al., The hypocretins: Hypothalamus- specific peptides with neuroexcitatory activity.
Proc. Natl. Acad. Sci. U.S.A. 95, 322–327 (1998).
14. T. Sakurai et al., Orexins and orexin receptors: A family of hypothalamic neuropeptides and G
protein- coupled receptors that regulate feeding behavior. Cell 92, 573–585 (1998).
15. C. Peyron et al., A mutation in a case of early onset narcolepsy and a generalized absence of
hypocretin peptides in human narcoleptic brains. Nat. Med. 6, 991–997 (2000).
16. T. C. Thannickal et al., Reduced number of hypocretin neurons in human narcolepsy. Neuron 27,
469–474 (2000).
17. R. M. Chemelli et al., Narcolepsy in orexin knockout mice: molecular genetics of sleep regulation.
Cell 98, 437–451 (1999).
18. L. Lin et al., The sleep disorder canine narcolepsy is caused by a mutation in the hypocretin (orexin)
receptor 2 gene. Cell 98, 365–376 (1999).
19. T. Mochizuki et al., Behavioral state instability in orexin knock- out mice. J. Neurosci. 24, 6291–6300
(2004).
20. C. R. Burgess, T. E. Scammell, Narcolepsy: Neural mechanisms of sleepiness and cataplexy. J.
Neurosci. 32, 12305–12311 (2012).
21. T. E. Scammell, Narcolepsy. N Engl. J. Med. 373, 2654–2662 (2015).
22. S. Pintwala, J. Peever, Circuit mechanisms of sleepiness and cataplexy in narcolepsy. Curr. Opin.
Neurobiol. 44, 50–58 (2017).
23. C. L. A. Bassetti et al., Narcolepsy - clinical spectrum, aetiopathophysiology, diagnosis and
treatment. Nat. Rev. Neurol. 15, 519–539 (2019).
24. C. E. Mahoney, A. Cogswell, I. J. Koralnik, T. E. Scammell, The neurobiological basis of narcolepsy.
Nat. Rev. Neurosci. 20, 83–93 (2019).
25. M. H. Hansen, B. R. Kornum, P. Jennum, Sleep- wake stability in narcolepsy patients with normal,
low and unmeasurable hypocretin levels. Sleep Med. 34, 1–6 (2017).
26. L. I. Kiyashchenko, B. Y. Mileykovskiy, Y. Y. Lai, J. M. Siegel, Increased and decreased muscle tone
with orexin (hypocretin) microinjections in the locus coeruleus and pontine inhibitory area. J.
Neurophysiol. 85, 2008–2016 (2001).
27. L. Verret et al., A role of melanin- concentrating hormone producing neurons in the central
regulation of paradoxical sleep. BMC Neurosci. 4, 19 (2003).
28. M. Modirrousta, L. Mainville, B. E. Jones, Orexin and MCH neurons express c- Fos differently
after sleep deprivation vs. recovery and bear different adrenergic receptors. Eur. J. Neurosci. 21,
2807–2816 (2005).
29. M. G. Lee, O. K. Hassani, B. E. Jones, Discharge of identified orexin/hypocretin neurons across the
sleep- waking cycle. J. Neurosci. 25, 6716–6720 (2005).
30. B. Y. Mileykovskiy, L. I. Kiyashchenko, J. M. Siegel, Behavioral correlates of activity in identified
hypocretin/orexin neurons. Neuron 46, 787–798 (2005).
31. K. Takahashi, J. S. Lin, K. Sakai, Neuronal activity of orexin and non- orexin waking- active neurons
during wake- sleep states in the mouse. Neuroscience 153, 860–870 (2008).
32. I. A. Azeez, F. Del Gallo, L. Cristino, M. Bentivoglio, Daily fluctuation of orexin neuron activity and
wiring: The challenge of “Chronoconnectivity”. Front. Pharmacol. 9, 1061 (2018).
33. S. B. Li et al., Hyperexcitable arousal circuits drive sleep instability during aging. Science 375,
eabh3021 (2022).
34. A. Inutsuka et al., The integrative role of orexin/hypocretin neurons in nociceptive perception and
analgesic regulation. Sci. Rep. 6, 29480 (2016).
35. K. Ghandour et al., Orchestrated ensemble activities constitute a hippocampal memory engram.
Nat. Commun. 10, 2637 (2019).
36. S. Chowdhury et al., Dissociating orexin- dependent and - independent functions of orexin neurons
using novel Orexin- Flp knock- in mice. Elife 8, e44927 (2019).
37. M. Mieda et al., Orexin peptides prevent cataplexy and improve wakefulness in an orexin neuron-
ablated model of narcolepsy in mice. Proc. Natl. Acad. Sci. U.S.A. 101, 4649–4654 (2004).
38. A. Seifinejad, A. Vassalli, M. Tafti, Neurobiology of cataplexy. Sleep Med. Rev. 60, 101546 (2021).
39. S. Zhou et al., Activity of putative orexin neurons during cataplexy. Mol. Brain. 15, 21 (2022).
40. T. E. Scammell, J. T. Willie, C. Guilleminault, J. M. Siegel, A consensus definition of cataplexy in
mouse models of narcolepsy. Sleep 32, 111–116 (2009).
41. A. Yamanaka et al., Hypothalamic orexin neurons regulate arousal according to energy balance in
mice. Neuron 38, 701–713 (2003).
42. T. Tsunematsu et al., Acute optogenetic silencing of orexin/hypocretin neurons induces slow- wave
sleep in mice. J. Neurosci. 31, 10529–10539 (2011).
43. T. Tsunematsu et al., Long- lasting silencing of orexin/hypocretin neurons using archaerhodopsin
induces slow- wave sleep in mice. Behav. Brain Res. 255, 64–74 (2013).
44. R. H. Williams et al., Transgenic archaerhodopsin- 3 expression in hypocretin/orexin neurons
engenders cellular dysfunction and features of type 2 narcolepsy. J. Neurosci. 39, 9435–9452
(2019).
45. A. Vassalli et al., Electroencephalogram paroxysmal theta characterizes cataplexy in mice and
children. Brain 136, 1592–1608 (2013).
46. S. Bastianini, A. Silvani, C. Berteotti, V. Lo Martire, G. Zoccoli, High- amplitude theta wave bursts
during REM sleep and cataplexy in hypocretin- deficient narcoleptic mice. J. Sleep Res. 21, 185–188
(2012).
47. K. F. Tanaka et al., Expanding the repertoire of optogenetically targeted cells with an enhanced gene
expression system. Cell Rep. 2, 397–406 (2012).
48. S. Izawa et al., REM sleep- active MCH neurons are involved in forgetting hippocampus- dependent
memories. Science 365, 1308–1313 (2019).
49. A. Yamanaka, Y. Muraki, N. Tsujino, K. Goto, T. Sakurai, Regulation of orexin neurons by the
monoaminergic and cholinergic systems. Biochem. Biophys. Res. Commun. 303, 120–129 (2003).
50. A. Yamanaka, S. Tabuchi, T. Tsunematsu, Y. Fukazawa, M. Tominaga, Orexin directly excites orexin
neurons through orexin 2 receptor. J. Neurosci. 30, 12642–12652 (2010).
51. T. Sakurai et al., Input of orexin/hypocretin neurons revealed by a genetically encoded tracer in
mice. Neuron 46, 297–308 (2005).
52. C. B. Saper, P. M. Fuller, N. P. Pedersen, J. Lu, T. E. Scammell, Sleep state switching. Neuron 68,
1023–1042 (2010).
53. J. Hara et al., Genetic ablation of orexin neurons in mice results in narcolepsy, hypophagia, and
obesity. Neuron 30, 345–354 (2001).
54. S. Tabuchi et al., Conditional ablation of orexin/hypocretin neurons: A new mouse model for the
study of narcolepsy and orexin system function. J. Neurosci. 34, 6495–6509 (2014).
55. F. Pizza et al., Nocturnal sleep dynamics identify narcolepsy type 1. Sleep 38, 1277–1284 (2015).
56. T. Nambu et al., Distribution of orexin neurons in the adult rat brain. Brain Res 827, 243–260
(1999).
57. J. John, M. F. Wu, L. N. Boehmer, J. M. Siegel, Cataplexy- active neurons in the hypothalamus:
Implications for the role of histamine in sleep and waking behavior. Neuron 42, 619–634 (2004).
58. L. Duffet et al., A genetically encoded sensor for invivo imaging of orexin neuropeptides. Nat.
Methods 19, 231–241 (2022).