Cannabinoids promote NREM sleep via inhibition of prefrontal cortico-hypothalamic projections

Cannabinoids promote non-rapid eye movement (NREM) sleep, but its underlying mechanism is not known. Here we find that cannabinoid promotion of NREM sleep is inhibited by cannabinoid antagonist injection systemically or into the dorsomedial hypothalamus (DMH), where cannabinoids selectively inhibit glutamatergic inputs synapsing with glutamatergic but not GABAergic neurons. Photoactivation of DMH-projecting ventromedial prefrontal cortical (vmPFC) neurons, their terminals, or their postsynaptic DMH neurons rapidly switches NREM sleep to wakefulness, which is blocked by photoinhibition of DMH outputs. Chemoactivation of DMH glutamatergic but not GABAergic neurons innervated by vmPFCs promotes wakefulness and suppresses NREM sleep, whereas chemoinhibition of vmPFC projections in DMHs produces opposite effects by mimicking cannabinoid effects. DMH-projecting vmPFC neurons are inhibited during NREM sleep and activated during wakefulness. Chemoactivation of DMH-projecting vmPFC neurons blocks cannabinoid promotion of NREM sleep and suppression of wakefulness. Thus, vmPFC neurons innervating DMHs represent the first identified set of cerebral cortical neurons for promotion of physiological wakefulness and suppression of NREM sleep, while cannabinoid inhibition of vmPFC projections in DMHs promotes NREM sleep and suppresses wakefulness. identification and recording of the fluorescent positive DMH neurons. To examine electrophysiology of DMH glutamatergic neurons, we Results were reported as mean ± SEM. Statistical analysis of the data was performed using a Student t test, one-way ANOVA, two-way ANOVA, two-way ANOVA for repeated measures or RM two-way ANOVA, followed by Bonferroni post-hoc test. Statistical significance was set at p <0.05 (*), p <0.01(**).

Long-term sleep problems, which affect 25-30% of human population 1 , is associated with profound health problems, including anxiety, depression, cardiovascular disorders, and obesity 1,2 .
Experimental activation of sleep-promoting cells in the hypothalamus, basal forebrain and brainstem switches wakefulness to sleep 2,[11][12][13] . In 1920s, von Economo indicated the midbrain and posterior hypothalamus to regulate wakefulness 14 . It is now known that main wakepromoting cells are located in the brainstem, hypothalamus, and basal forebrain [11][12][13]15,16 and their activation promotes wakefulness through ascending projections to cerebral cortices 2,12,13 . After nearly a century of intensive and extensive sleep study, however, it remains unknown whether cerebral cortices, as the center of the central nervous system to process high-order information, contain wake-promoting cells.
In this study, we employed in vivo sleep recording, neurophysiology, the cutting-edge neuroanatomical sparse labeling, conditional mouse mutagenesis, opto/chemogenetic strategy, and in vivo calcium imaging to investigate the mechanism underlying cannabinoid modulation of sleep-wake behavior. Surprisingly, we found an unanticipated cerebrohypothalamic pathway linking not only activation of ventromedial prefrontal cortical (vmPFC) projections in the dorsomedial hypothalamus (DMH), promotion of wakefulness and suppression of NREM sleep, but also cannabinoid inhibition of vmPFC projections in the DMH, promotion of NREM sleep and suppression of wakefulness.

Cannabinoid modulation of sleep-wake behavior
Because rodents sleep in the day, we injected synthetic cannabinoids 10-30 min before lights-on (8:00 AM) and analyzed electroencephalogram (EEG) and electromyogram (EMG) recordings.
We first examined detailed changes of sleep-wake states after administration of nabilone, a synthetic cannabinoid currently licensed as Cesamet in Canada for the treatment of chemotherapy-associated nausea and vomiting. Nabilone (0.2, 1, 5 mg/kg, i.p.) dose-dependently promoted NREM sleep and suppressed both wakefulness and REM sleep (Fig. 1a-c) for over 12 h (Fig. 1d-f). Specifically, the moderate dose of nabilone (1 mg/kg, i.p.) promoted NREM sleep from approximately 58% of the baseline sleep-wake state to 85% and suppressed wakefulness and REM sleep from approximately 33% and 9% to 13% and 2%, respectively (Fig. 1a-c). EEG hypnograms also showed significant changes of sleep and wake architecture after nabilone administration (Fig. 1g). Nabilone significantly increased bout duration but not number of NREM sleep, decreased bout duration but not number of wakefulness, and decreased both bout number and duration of REM sleep (Fig. 1h-m). HU210 (5, 10, 20, 50 μg/kg, i.p.), the most potent synthetic cannabinoid, also dose-dependently promoted NREM sleep and suppressed both wakefulness and REM sleep (Extended Data Fig. 1a-c).
To examine whether cannabinoids modulate sleep via the CB 1 cannabinoid receptor (CB 1 R), we conducted both systemic and intra-cerebral injection of the CB 1 R antagonist AM281 (2 mg/kg, i.p.; 50 ng/0.2 l, intra-cerebral injection) 17,18 . Systemic injection of AM281 but not 5 vehicle, which did not significantly affect sleep-wake states on its own, significantly inhibited the effects of nabilone on NREM sleep and wakefulness without significant effects on cannabinoid suppression of REM sleep (Extended Data Fig. 1d-f). The DMH is a central hub for modulation of sleep-wake states 12,13,19,20 , ingestion, reproduction, endocrine, cardiovascular function, respiration, and thermogenesis [21][22][23] . We therefore examined the possible DMH location of systemic cannabinoid modulation of sleep-wake states. Bilateral intra-DMH injection of AM281 (Extended Data Fig. 1g,h), which did not significantly affect sleep-wake behavior on its own, significantly inhibited the effects of nabilone on NREM sleep and wakefulness without significant effects on cannabinoid suppression of REM sleep (Extended Data Fig. 1i-k). These results suggest that cannabinoids activate DMH CB 1 R to promote NREM sleep and suppress wakefulness.

Cannabinoids deactivate DMH glutamatergic but not GABAergic neurons
We next performed patch clamp experiments on hypothalamic slices to explore whether cannabinoids act on DMH presynaptic CB 1 R of GABAergi, glutamatergic or both synapses.
HU210 did not significantly change DSE of DMH GABAergic neurons (Fig. 2e,i), but nabilone significantly inhibited DSE of DMH glutamatergic neurons (Fig. 2f,i). Because activation of glutamatergic presynaptic CB 1 R inhibits presynaptic release of glutamate, producing deactivatory (i.e., inhibitory) effects on postsynaptic neurons 24,25 , it was not surprised to find that bath application of nabilone significantly reduced firing rates of DMH LTS(-) but not LTS(+) neurons ( Fig. 2g-i). Our results together suggest that cannabinoids selectively activate CB 1 R at DMH glutamatergic but not GABAergic inputs synapsing with DMH glutamatergic but not GABAergic neurons, leading to sequential inhibition of DMH glutamatergic inputs and glutamatergic neurons.

Mapping brain regions innervated by DMH-projecting vmPFC neurons
Then, which brain region sends glutamatergic inputs to the DMH for modulation of sleep-wake behavior? The vmPFC, i.e., the prelimbic, infralimbic and dorsal preduncular subregions in rodents, is a central hub that controls multiple brain regions through its internal circuitry and external connectivity 27-31 . The vmPFC plays a key role in decision-making, value-coding, thinking and emotional arousal/recognition 30,32 and sends glutamatergic projections into the DMH in rats 33 . Therefore, the vmPFC may actively participate in modulation of sleep-wake behavior through its glutamatergic projections in the DMH. We then conducted neuroanatomical tract-tracing experiments to investigate glutamatergic vmPFC-DMH projections in detail. In mice injected with the retrograde-tracing virus pAAV-SYN-MCS-mCherry-3FLAG into the 7 DMH (Fig. 3a,b), we observed mCherry-labeled neurons specifically in the 5th layer of the vmPFC (Fig. 3c), which is in agreement with a recent study on rats 33 .
It is known that the vmPFC projects to multiple brain regions 27-31 , but it is not clear if the same vmPFC neuron innervating the DMH also project to major brain regions of sleep-and wake-promoting. Employing the recently developed sparse labelling approach 34 , we injected the rAAV2-Camk2-mCherry-2A-Cre into the DMH, which was transported back to vmPFC cell bodies and turned on the Cre-dependent virus mixture of both controller (pAAV-TRE-DIO-FLPo) and amplifier vectors (pAAV-TRE-fDIO-GFP-IRES-Tta) injected into the ipsilateral vmPFC ( Fig. 3d). After vectors containing the bright EGFP were transported and labelled the whole vmPFC neuron, including its axons and terminals, mice were killed for clearing of the whole brain 35 , followed by imaging of the brain with a light sheet microscope. DMH-projecting vmPFC neurons showed their entirety with cell bodies located in the vmPFC and their axons projecting all the way downward and caudalward into the rAAV2-Camk2-mCherry-2A-Cre injection site in the DMH but not its surrounding regions ( Fig. 3e-g), confirming the validity of our targetspecific labeling strategy. Along their way toward the DMH, axons sent collaterals heavily innervating the nucleus accumbens (Fig. 3h), which is the only brain region receiving as highest terminal innervation as the DMH. Moderate density of axonal terminals was found in the motor cortex, anterior cingulate cortex, bed nuclei of the stria terminalis (Fig. 3i), preoptic hypothalamus and anterior hypothalamus (Fig. 3i), with low density in the bottom part of the thalamus. While the preoptic hypothalamus contains sleep-promoting neurons 11-13 , we did not detect axons/terminals in the wake-promoting paraventricular thalamus (PVT), lateral hypothalamus (LH), tuberomammillary nucleus, basal forebrain, ventral tegmental area (VTA), raphe nuclei, locus coeruleus (LC), and parabrachial nucleus 11 .

8
In order to examine whether the results obtained with sparse labeling approach are reliable, we conducted another experiment for double labeling of vmPFC neurons innervating the DMH and PVT. Mice received an injection of retrograde-tracing virus pAAV-SYN-MCS-mCherry-3FLAG into the DMH and pAAV-SYN-MCS-EGFP-3FLAG into the PVT (Fig. 3j-l). Almost all DMH-projecting neurons are located in the layer 5 of the vmPFC, whereas PVT-projecting neurons are mainly located in the layer 6 with a few cells in the lay 5, which did not show double labeling with DMH-projecting neurons (Fig. 3m). These results clearly reveal that although vmPFC neurons directly innervate both the DMH and PVT, DMH-projecting vmPFC neurons do not project to the PVT, thus confirming our sparse labeling results.

Optoactivation of vmPFC-DMH projections switches NREM sleep to wakefulness
While we observed that intra-DMH injection of the CB 1 R antagonist AM281 inhibited cannabinoid promotion of NREM sleep and suppression of wakefulness, we also found that cannabinoids selectively inhibited glutamatergic inputs in the DMH. We reasoned that if inhibition of DMH glutamatergic inputs participated in cannabinoid promotion of NREM sleep and suppression of wakefulness, activation of DMH inputs would promote wakefulness and suppress NREM sleep. As neurons in the 5 th layer of the vmPFC directly innervated the DMH ( Fig. 3), we hypothesized that in contrast to inhibitory effects of cannabinoids on vmPFC-DMH projections, activation of vmPFC-DMH glutamatergic projections may promote wakefulness and suppress NREM sleep. To examine the hypothesis, we employed optogenetic strategy to activate DMH-projecting vmPFC neurons with adeno-associated virus (AAV) expressing channelrhodopsin 2 (ChR2). We injected the pseudorabies virus vector expressing Cre recombinase (PRV-Cre) 36 into bilateral DMHs for transport of Cre back to cell bodies in the vmPFC, followed 5 weeks later by an injection of AAV-Camk2-DIO-ChR2-mCherry (or AAV-Camk2-DIO-mCherry as control) into bilateral vmPFCs (Fig. 4a). Mice receiving both PRV-Cre and AAV-Camk2-DIO-ChR2-mCherry were killed for patch clamp experiments on brain slices.
We applied blue laser light stimulation of 1, 10, 20 or 50 Hz (5 mW, 10 ms each pulse, 10 sec total) to red fluorescence mCherry-labelled cells in the 5th layer of the vmPFC. Photostimulation

Chemoactivation of DMH glutamatergic but not GABAergic neurons innervated by vmPFCs promotes wakefulness and suppresses NREM sleep
As the DMH contains glutamatergic neurons 20 , we examined the effects on sleep-wake states of activation of DMH glutamatergic neurons innervated by the vmPFC. We employed our vGlut2-iCreERT2 mutant mice 36 and transsynaptic tagging strategy 37 for a specific transgene expression in DMH glutamatergic neurons innervated by vmPFCs. Instead of employing optogenetic strategy, here we utilized chemogenetic strategy for a stable activation of neurons for several hours, during which changes of sleep-wake states could be assessed after a clozapine-N-oxide (CNO) injection at 10:00 am, a time approximating peak sleep drive in mice. Mice received pAAV-hSyn-DIO-EGFP-T2A-NLS-Flipo injection into bilateral vmPFCs, followed 5 weeks later by pAAV-EF1a-fDIO-hM3Dq-mCherry (or pAAV-EF1a-fDIO-mCherry as control) injection into bilateral DMHs (Fig. 5a). Patch clamp recording of mCherry-labeled DMH neurons revealed a significant increase in their firing rates after bath application of CNO onto hypothalamic slices (Fig. 5b), suggesting both successful Cre-dependent transgene expression in and CNO-elicited activation of DMH glutamatergic neurons innervated by vmPFCs. In mice receiving pAAV-EF1a-fDIO-hM3Dq-mCherry, but not pAAV-EF1a-fDIO-mCherry, within bilateral DMHs together with pAAV-hSyn-DIO-EGFP-T2A-NLS-Flipo confined in bilateral vmPFCs ( Fig. 5c and Extended Data Fig. 3d,g), CNO (0.5 mg/kg, i.p.) but not vehicle injection significantly promoted wakefulness and suppressed NREM sleep without significant effects on REM sleep (Fig. 5d,e). The CNO-induced promotion in wakefulness was primarily due to longer wake bouts and less number of wake bouts (Fig, 5f), suggesting an efficient promotion of wakefulness after activation of vmPFC-DMH projections. The promotion in wakefulness across 13 12h in light phase was followed by insignificant changes of sleep-wake states across 12h in darker phase (Extended Data Fig. 7).
As the DMH also contains GABAergic neurons 20 , we further employed chemogenetic strategy to examine the effects on sleep-wake states following activation of DMH GABAergic neurons innervated by vmPFCs. We used Dlxa-Cre mice carrying Cre in forebrain GABAergic neurons used in our recent study 39 . Mice received pAAV-hSyn-DIO-EGFP-T2A-NLS-Flipo injection into bilateral vmPFCs, followed 5 weeks later by pAAV-EF1a-fDIO-hM3Dq-mCherry (or pAAV-EF1a-fDIO-mCherry as control) injection into bilateral DMHs (Extended Data Fig.   8a). In mice with virus injection within bilateral vmPFCs and DMHs, a CNO injection (0.5 mg/kg, i.p.) at 10:00 am did not produce significant effects on sleep-wake states (Extended Data Fig. 8b-f). Our results together suggest that activation of glutamatergic but not GABAergic neurons innervated by vmPFCs promotes wakefulness and suppresses NREM but not REM sleep.

NREM switch to wakefulness by optoactivation of DMH-projecting vmPFC neurons is blocked by optoinhibition of DMH outputs in LHs
While DMH glutamatergic neurons mainly innervate the LH 20

Inhibition of vmPFC outputs in DMHs suppresses wakefulness and promotes NREM sleep
To access whether vmPFC neurons innervating DMHs are required for physiological sleep-wake behavior, we performed chemoinhibition of vmPFC projections in DMHs. Mice received AAV-Camk2-hM4Di-mCherry (or AAV-Camk2-mCherry as control) injection into bilateral vmPFCs, followed by insertion of 1 pair of cannulae above bilateral DMHs at 40 O oralward angle (Fig. 6a).
Five weeks later, mice received CNO injection into bilateral DMHs through cannulae at 10:00 pm, a time approximating peak activity in mice. In mice with virus injection confined to bilateral vmPFCs and DMHs (Fig. 6b,c), CNO but not vehicle significantly suppressed wakefulness and promoted NREM sleep without significant effects on REM sleep (Fig. 6d,e). Thus, activation of vmPFC neurons innervating the DMH is required for physiological sleep-wake behavior.

Calcium imaging of vmPFC neurons innervating DMH and PVT
Next, we examined the real-time activity of vmPFC neurons innervating the DMH and PVT across sleep-wake cycle. Calcium activity of the vmPFC was recorded through a photometry fiber implanted above the vmPFC in freely moving mice after injection of the retrograde-tracing AAV2 (rAAV2) expressing the calcium indicator GCaMP6 into the DMH (Fig. 7e) or PVT (Extended Data Fig. 10a). During each imaging session, sleep-wake states were monitored with EEG and EMG recordings. Calcium activity of vmPFC neurons innervating the DMH varied significantly across sleep-wake states, with significant inhibition during NREM sleep and significant activation during wakefulness and REM sleep (Fig. 7f-h). However, calcium activity 16 of vmPFC neurons innervating the PVT did not show significant changes across sleep-wake states (Extended Data Fig. 10b-d). These results suggest that vmPFC neurons innervating the DMH represent a new class of wake-promoting neurons in the brain.
These results replicated our findings shown above that activation of DMH-projecting vmPFC neurons, their terminals or their postsynaptic glutamatergic neurons promoted wakefulness and suppressed NREM sleep without significant effects on REM sleep. Systemic nabilone significantly promoted NREM sleep and suppressed wakefulness, which was blocked by CNO injection (Fig. 8c-e). Our results suggest that cannabinoids promote NREM sleep and suppress wakefulness through inhibition of vmPFC glutamatergic projections in the DMH.

Discussion
The cerebral cortex processes high-order information, but whether the cerebral cortex has wake- When wake-promoting cells in the basal forebrain, LH and brainstem send wake-promoting signals to the mPFC, the mPFC may in turn modulate arousal and wakefulness through its topdown outputs to major wake-promoting neurons in the basal forebrain, LH and brainstem 12,40 .
Because the vmPFC exclusively targets GABAergic neurons in both the LC 41  In sharp contrast to wake-promoting and NREM-suppressing effects produced by activation of vmPFC-DMH projections, a single dose of synthetic cannabinoids significantly promoted NREM sleep and suppressed wakefulness for over 12 hours, which were inhibited by systemic or intra-DMH injection of CB 1 R antagonist AM281 and blocked by chemoactivation of vmPFC neurons innervating DMHs. Surprisingly, cannabinoids selectively activated CB 1 R at DMH glutamatergic but not GABAergic inputs synapsing with DMH glutamatergic but not GABAergic neurons, thus inhibiting DMH glutamatergic but not GABAergic neurons. These results further support the idea that activation of vmPFC neurons innervating DMH neurons is required for physiological sleep-wake behavior. NREM sleep is important for restoration of body physiology especially brain function 43  In conclusion, vmPFC neurons innervating the DMH represent the first identified set of cerebral cortical cells for promotion of physiological wakefulness and suppression of NREM sleep without significant involvement in REM sleep (Fig. 8f). In contrast, synthetic cannabinoids produce profound promotion of NREM sleep and suppression of wakefulness through activation 20 of CB 1 R at vmPFC glutamatergic inputs synapsing with DMH glutamatergic neurons, thus inhibiting DMH glutamatergic neurons (Fig. 8g).

Animals
We used adult (over 2 months old) male CD1 mice, vGlut2-iCreERT2 mutant mice carrying iCreERT2 in brain glutamatergic neurons containing vGlut2, and Dlxa-Cre mice carrying Cre in forebrain GABAergic neurons, which were used in our recent studies 36,39

Sleep-wake recording and analysis
Under anesthesia with isoflurane (4% for induction, 1.5% for maintenance), an adult mouse was placed in a stereotaxic apparatus. A longitudinal incision of 1 cm in length was made in the midline and through the scalp to expose the skull for implantation of the EEG/EMG headmount (8402, Pinnacle Technology Inc., USA) onto the mouse skull (AP 3 mm) with stainless steel screws serving as EEG electrodes across the cortex. Dental acrylic was applied to secure the headmount in place. EMG probes were inserted into the cervical portion of the trapezoid muscles.
Metacam (0.2 mg/kg, s.c.) was injected once per day for 3 days after surgery. After surgery, mice were allowed to recover undisturbed for at least 1 week before they were briefly anesthetized with isoflurane (4% induction, 1.5% maintenance) and the headstage pre-amplifier (8406, Pinnacle Technology Inc., USA) was attached to a connector of the mouse EEG headmount.
Mice were then returned to their Plexiglas recording cages (25.4  We analyzed EEG traces with custom written MATLAB scripts and functions (MathWorks).
EEG traces were band-pass filtered in the forward and reverse direction to eliminate phase distortions. Spectral analysis (spectrogram plotting) was performed using a multi-taper fast Fourier transform (FFT, padding factor =1) on each 40-50s data segments taken before and after optical stimulation (Chronux package, window = 4 s, step = 0.5 s, tapers [2 3]). Power spectra were calculated in a linear scale to achieve the power distribution in the frequency band of interest (0-20 Hz). EEG power value was extracted from the power spectrum (band power: total power over 0.5-20 Hz). Power spectral density was calculated by the Fourier transform of the autocorrelation. EEG peak frequency was extracted from the power spectral density. EEG was sampled at 400 Hz.

Cannabinoid treatment
Under anesthesia with isoflurane, the EEG/EMG headmount was placed onto the mouse skull and secured, after which mice for future intra-DMH injection received placement of double

Electrophysiology analysis
After decapitation, the mouse brain was removed and coronal brain vibrosections (300 μm) were To evoke PSCs, a stable whole-cell recording configuration was established on the DMH cell.

Neuroanatomical tract-tracing experiments
Adult male CD1 mice were anesthetized with isoflurane (4% induction, 1.5% maintenance) and placed on a stereotaxic apparatus. Ophthalmic ointment was smeared to the eyes to prevent drying. After shaving the hair and cleaning the incision site with medical alcohol and iodine, the
At the end of the last surgery, the EEG/EMG headmount was placed onto the mouse skull and secured, after which an optical fiber housing unit or double cannulae (Plastics One, Canada) was placed above the skull with the fiber or cannula insertion vertically above bilateral vmPFCs (AP  DV -2.7 mm). Five weeks later, mice received an injection of pAAV-EF1a-fDIO-hM3Dq-mCherry or control virus (pAAV-EF1a-fDIO-mCherry) into bilateral DMHs. Next, the EEG/EMG headmount was placed onto the mouse skull and secured. For chemogenetic study on Dlxa-Cre mice, pAAV-hSyn-DIO-EGFP-T2A-NLS-Flipo was injected into bilateral vmPFCs, followed 4 weeks later by pAAV-EF1a-fDIO-hM3Dq-mCherry or control virus (pAAV-EF1a-fDIO-mCherry) injection into bilateral DMHs. Next, the EEG/EMG headmount was placed onto the mouse skull and secured. After surgery, the vGlut2-icreERT2 mice also received tamoxifen injection (1 mg/mouse/day, dissolved in corn oil) for 8 days. Four to six weeks after surgery, all mice received baseline EEG/EMG recording for 24 hours before an i.p. injection of CNO (0.5 mg/kg, i.p.) or vehicle at 10:00 AM or a bilateral intra-DMH injection of CNO (10 M/0.2 l/side) or vehicle. We conducted intra-DMH microinjection of CNO by extending 1 mm out of the implanted cannulae at a rate of 0.1 μl/min for 2 min. Localization of CNO injection sites was done by measuring the distance of 1 mm below the cannulae tract tip.

Confirmation of virus injection sites
After experiments, mice were anesthetized with 5% chloral hydrate (8 mg/kg, i.p.) and perfused transcardially with normal saline, followed by 100 ml ice-cold 4% paraformaldehyde in PBS.
Brains were carefully removed from the skull and postfixed in 4% paraformaldehyde for 12 hours, then dehydrated with 30% sucrose at 4°C overnight. With a freezing microtome (CM 3050S, Leica), brains were cut into 40-μm frontal sections, which were then mounted onto glass slides. Histological images were acquired with an upright fluorescent microscope (Nikon).
Finally, we identified and recorded locations of central points of virus injection sites within (black circles) bilateral and outside (white circles) unilateral or bilateral or within unilateral vmPFCs or DMHs. For intra-DMH injection of AM281 or CNO, the injection sites were 1 mm below the tips of bilateral cannulae. For each experiment, we routinely conducted numerous mice until 5 mice receiving target virus and 5 mice receiving control virus or 6 mice receiving intra-DMH injection of AM281 showed injection sites within bilateral vmPFCs and/or DMHs.

Fiber photometry
Under anesthesia with isoflurane (4% for induction, 1.5% for maintenance), adult C57/BL6 mice For each test, the photometry signal F was converted to ∆ F/F as following formula by Matlab.
At the same time, we recorded EEG and EMG for 1-2 sessions, each with 1-4h duration. To analyze sleep-wake states, we selected all the sessions in which mice had all three states, i.e.
NREM, REM and wakefulness, followed by calculation of ∆ F/F photometry ratios during contiguous three sleep-wake states. After collecting data, mice were anesthetized with 5% chloral hydrate (8 mg/kg, i.p.) and perfused transcardially with normal saline and 4% paraformaldehyde in PBS. Brains were dehydrated with 30% sucrose at 4°C overnight and cut with a freezing microtome for localization of both the DMH or PVT injection site and vmPFC fiber site.