The integrative role of orexin/hypocretin neurons in nociceptive perception and analgesic regulation

The level of wakefulness is one of the major factors affecting nociception and pain. Stress-induced analgesia supports an animal’s survival via prompt defensive responses against predators or competitors. Previous studies have shown the pharmacological effects of orexin peptides on analgesia. However, orexin neurons contain not only orexin but also other co-transmitters such as dynorphin, neurotensin and glutamate. Thus, the physiological importance of orexin neuronal activity in nociception is unknown. Here we show that adult-stage selective ablation of orexin neurons enhances pain-related behaviors, while pharmacogenetic activation of orexin neurons induces analgesia. Additionally, we found correlative activation of orexin neurons during nociception using fiber photometry recordings of orexin neurons in conscious animals. These findings suggest an integrative role for orexin neurons in nociceptive perception and pain regulation.


Attenuated pain-related behaviors by selective activation of orexin neurons.
Our findings that a lack of orexin neurons augments pain-related behaviors prompted us to test whether selective activation of orexin neurons would affect nociceptive processing and pain-related behaviors. To this end, we employed a pharmacogenetic tool, designer receptors exclusively activated by designer drugs (DREADD) 32 , as shown in Fig. 3A. In order to express the hM3Dq receptor exclusively in orexin neurons, a Cre-dependent adeno-associated virus (AAV) vector, AAV-hSyn-FLEX-hM3Dq-mCherry, was injected into the hypothalamus of orexin-Cre transgenic mice in which orexin neurons exclusively expressed Cre recombinase under control of human prepro-orexin promoter 13 . The specific expression and activation of orexin neurons was confirmed by immunohistochemistry. More than 3 weeks after injection of the AAV vector, saline or clozapine-N-oxide (CNO) (1 mg/kg) was intraperitoneally administered. One hour after application, cFos expression in orexin neurons was tested using immunohistochemical analysis. hM3Dq expression was detected as red fluorescence from the mCherry protein fused with hM3Dq. Although hM3Dq was expressed in orexin neurons in both groups, strong cFos expression was only observed in orexin neurons in the CNO-injected group (Fig. 3B, also refer to our previous report 13 ). Behavioral experiments to test nociception were performed one hour after CNO injection (1 mg/kg, intraperitoneal (i.p.)) when orexin neurons were robustly and specifically activated (Fig. 3C). Locomotor assessment confirmed that there was an increase in locomotion which lasted for 4 hours (Fig. 3D) as in our previous study 13 .
Scientific RepoRts | 6:29480 | DOI: 10.1038/srep29480 This hyper-locomotion made it difficult to perform the von Frey test or Hargreaves test. Therefore, we employed a hot plate test and a cold plate test. The temperature of the hot plate was set at 50 °C or 55 °C. The latency to exhibit pain-related behaviors such as hind-paw shaking or jumping was measured. The cut-off time was set at 1 min to avoid burn injuries. The cold plate test was performed as described in the methods. Compared with saline-injected mice, CNO-injected mice showed a significantly longer latency during the hot plate test at both 50 °C (Saline: 18.5 ± 0.7 sec, CNO: 24.3 ± 2.3 sec, n = 12) and 55 °C (Saline: 7.5 ± 0.4 sec, CNO: 9.5 ± 1.0 sec, n = 12) (Fig. 3E). Similar anti-nociceptive effects were observed in the cold plate test. Pain related behaviors were significantly decreased in CNO-injected mice (Saline: 24.4 ± 1.7 counts, CNO: 18.1 ± 2.6 counts, n = 12) (Fig. 3F). We also performed a formalin test to analyze the time course for chemical nociception. CNO-injected mice showed significantly shorter licking and biting time in phase II (Saline: 297 ± 23 sec, CNO: 175 ± 29 sec, . (E,F) Coronal sections of the LC (E) and PAG (F). These areas include dense nerve endings of orexin neurons in DOX(+ ) mice. However, these orexin nerve endings were not observed in DOX(− ) mice. 4V: fourth ventricle, Aq: Aqueduct, TH: tyrosine hydroxylase. n = 6, 7), while the initial response was not affected in phase I (Saline: 63.8 ± 8.2 sec, CNO: 57.4 ± 6.3 sec, n = 6, 7) (Fig. 3G,H). These results demonstrated that activation of orexin neurons using DREADD had an analgesic effect.

Selective recording of orexin neuronal activity in vitro.
To elucidate the causality between neuronal activity and analgesic behaviors, it is necessary to perform real-time recordings of orexin neuronal activity during nociceptive processing. To achieve this, we first developed a recording system to detect selective responses of orexin neurons in vitro using a genetically encoded calcium indicator, G-CaMP6 33 . To express G-CaMP6 exclusively in orexin neurons, an AAV vector, AAV-TetO(3G) G-CaMP6, was injected into the LHA of orexin-tTA mice (Fig. 4A). The TetO(3G) sequence is a modified version of the TetO sequence that provides for lower basal expression leakage and higher expression in the presence of tTA 34 . Three weeks after AAV vector injection, we confirmed a dense overlap of the G-CaMP6 expression in orexin neurons (71.6 ± 7.4% of orexin neurons expressed G-CaMP6, n = 3, Fig. 4B). G-CaMP6 fluorescence was restricted to orexin neurons, and there was no ectopic expression of G-CaMP6 other than in orexin neurons. We prepared brain slices which included the LHA and performed simultaneous patch-clamp recording and calcium imaging to clarify the relationship between the electrophysiological properties and G-CaMP6 fluorescence (Fig. 4C). Representative traces indicated that orexin neurons expressing G-CaMP6 showed increased fluorescence intensity with firing in a frequency-dependent manner. Firing frequency was generated by injecting current pulse through a recording pipette at 5, 10, 20 and 50 Hz (Fig. 4D-F). We confirmed that orexin neurons responded to the input current with high fidelity even at a high 50 Hz frequency (Fig. 4D). The fluorescence of G-CaMP6 was enhanced as the frequency of firing increased (∆ F/F% was 46.1 ± 7.4% when the applied firing was 50 Hz, n = 9, Fig. 4F). We also confirmed the calcium response of orexin neurons caused by glutamate bath application. By increasing the concentration of glutamate from 10 μ M to 1,000 μ M, we observed a correlative increase in fluorescence from 2.4 ± 0.8% to 62.4 ± 10.6% (n = 10-12, Fig. 4G,H). These results confirmed the robust correlation between G-CaMP6 fluorescence and activity of orexin neurons.

Fiber photometry system for selective recording of orexin neuronal activity in vivo.
To achieve real-time recordings of orexin neuronal activity in vivo, we developed a novel fiber photometry system. Figure 5A-C show schematic drawings of our fiber photometry system (also refer to the Methods). In short, the system is equipped with an LED as a light source and a photomultiplier tube (PMT) as a detector. Blue light emitted from the LED is reflected by a dichroic mirror and illuminates the G-CaMP6-expressing neurons in the brain. The green fluorescence from G-CaMP6 is then gathered by the same optic fiber and the green light passes the first dichroic mirror and reflects off the second dichroic mirror in front of the PMT. Power intensity at the tip was set to 0.5 mW in the following experiments. To record the activity of orexin neurons, we generated mice which selectively express G-CaMP6 in orexin neurons by injecting AAV-TetO(3G) G-CaMP6 into the orexin-tTA mice as described in Fig. 4A. Three weeks after AAV vector injection, fiber photometric activity recordings of orexin  (B) Immunohistochemical confirmation of orexin neuron activation via hM3Dq. Orexin-Cre mice expressing hM3Dq in orexin neurons were intraperitoneally injected with saline or CNO (1.0 mg/kg). The arrowheads indicate triple co-localization of orexin, hM3Dq-mCherry and cFos. Scale bar is 50 μ m. (C) Diagram illustrating the behavioral test protocol. AAV vectors were injected 3 weeks before the behavioral experiment day. On the day of experimentation, saline or CNO were intraperitoneally injected at 12:00 noon. One hour after injection, nociception assays were performed. (D) Locomotive activity was monitored to confirm the long lasting effects of orexin neuronal activation using pharmacogenetics. Total motion was counted every 10 min (saline: n = 12, CNO: n = 12). (E) Hot plate tests (saline: n = 12, CNO: n = 12). Latency to jump or licking was measured. (F) Cold plate tests (saline: n = 12, CNO: n = 12). The number of pain-rerated behaviors was counted. Data represent the mean ± SEM. Statistical analyses were performed by a Student's t-test (* p < 0.05, * * p < 0.01). (G) Formalin tests for chemical nociception (Saline: n = 6, CNO: n = 7). The licking and biting time after formalin injection was measured. (H) Quantitative analysis of the formalin test; Phase I: 0-5 min, Phase II: 10-45 min. Data represent the mean ± SEM. Statistical analyses were performed by a Student's t-test (n.s. not significant, * p < 0.05, * * p < 0.01).
neurons were performed. Figure 5E shows how we determined whether the recorded fluorescence originated from G-CaMP6 or not by analyzing the wavelength of the fluorescence. The fiber tip was gradually inserted into the brain surface and the wavelength of the fluorescence was measured at depths of 0, 1, 2, 3, 4 and 5 mm from the surface of brain. The location of the inserted optic fiber was confirmed histochemically after the experiments (Fig. 5D). At a depth of 5 mm, where orexin neurons are densely distributed, the 500-600 nm wavelength photon count greatly increased, and the peak shifted toward shorter wavelengths (Fig. 5E). This wavelength correlated well with the fluorescence from G-CaMP6 33 . The data summarized in Fig. 5F represent the area under the curve (AUC) between wavelengths at 500-550 nm that could pass through the fluorescence filter. These results confirmed that the fluorescent signals originated from G-CaMP6.
Selective recording of orexin neuronal activity during nociception. Using this fiber photometry system, we recorded the activity of orexin neurons in response to noxious stimuli under conscious and anesthetic conditions. Mice were head-fixed with a stereotaxic apparatus and an optic fiber was inserted (from bregma − 1.5 mm, lateral 0.8 mm, ventral 5.0 mm); noxious mechanical stimuli were then applied to the tail using forceps equipped with a force sensor at the tip (Fig. 6A). Pinching force was programmable and was applied as a trapezoidal shape with 100, 200 and 300 g of maximum force. Before we use G-CaMP6-expressing mice, we confirmed that physical movement do not affect fluorescence intensity using orexin-EGFP mice as control (Fig. S2). The head position of a test mouse was stable even when mechanical stimuli induced body movement to avoid forceps. A representative trace indicates the activity of orexin neurons when 100, 200 and 300 g mechanical stimuli were applied (Fig. 6B). Pinching with a 300 g force induced a robust correlative increase in the G-CaMP6 signal of orexin neurons with high fidelity, while this response was not observed when weaker (100 g or 200 g) stimuli were applied or when the mice were under anesthesia with isoflurane (Fig. 6C,D).
We also investigated orexin neuronal activity during nociception of heat stimuli. A ramp-shaped noxious heat stimulus gradually increasing from 30 °C to 50 °C or 60 °C was applied to the left hind-paw using a thermal stimulator with a circular probe (diameter: 1 mm, Fig. 6E). Heat stimuli also induced a clear correlative increase in the fluorescent signal (Fig. 6F). A 50 °C stimulus induced a very weak response, however, at 60 °C a significant increase in activity of orexin neurons was induced. We also observed a temperature-dependent increase in the response that was not detected under anesthetic conditions (Fig. 6G,H). These results showed the correlative activation of orexin neurons during nociception.

Discussion
In this study, we selectively manipulated the activity and fate of orexin neurons to elucidate their cellular contribution to nociceptive perception and pain regulation. Adult-stage ablation of orexin neurons, together with a considerable decrease in their co-transmitters in the LHA, demonstrated their roles in anti-nociception and analgesia to all modalities of painful stimuli such as mechanical, thermal, and chemical stimuli. Accordingly, pharmacogenetic activation of the same neurons attenuated thermally and chemically induced pain. Using our novel fiber photometry system, we found a robust correlation between orexin neuronal activity and nociceptive perception induced by mechanical and heat stimuli. Interestingly, these neuronal responses were not observed under inhalation anesthesia by isoflurane.
We found correlative activation of orexin neurons induced by noxious mechanical and heat stimuli. To our knowledge, this is the first study to demonstrate real-time recording of orexin neuronal activity in conscious mice during nociceptive processing. Previous studies have reported that the concentration of orexin peptides is relatively low when feeling pain based on microdialysis 35 , while c-Fos expression in orexin neurons increases after noxious stimuli such as carrageenan-induced inflammation or foot-shock 36,37 . However, the temporal resolution of these experimental techniques was insufficient to detect real-time changes in the activity of orexin neurons. In the present study, we carefully confirmed the relationship between the fluorescence of G-CaMP6 and electrical activity in orexin neurons even at a high frequency firing rate of 50 Hz, and developed a recording system for selective neural circuits in the deep brain. Our findings demonstrate the high sensitivity of the fiber photometry system for detection of transient fast activity during physiological responses. Recently, other groups have developed similar systems to reveal the precise activity pattern during feeding 38 , reward 39 and social behavior 40 . These technical advances shall aid in dissecting the causality between specific neuronal activity and various behaviors. Our system can be further modified to add an optical path for photostimulation using different wavelengths of light. In the future, we expect that simultaneous recording and manipulation of orexin neurons in free-moving mice will characterize their multiple roles in nociception.
It is relatively easy to trace the efferent projections from orexin neurons. In contrast, the afferent pathways that convey noxious stimulation to the orexin neurons remain less clear. There are several candidate regions, however. The neurons in the spinal cord send axons directly to the hypothalamus 41 . The parabrachial nucleus is innervated by lamina I projection neurons in the spinal cord 42 and also projects to orexin neurons 43,44 . Considering that the amygdala projects to orexin neurons 44 , it is reasonable to suspect their involvement in nociception, especially in the affective component of pain. Loss of orexin neurons in human beings results in the sleep disorder narcolepsy with a characteristic symptom called cataplexy [45][46][47] . Cataplexy is the sudden loss of muscle tone caused by strong emotion with a preservation of consciousness during the attack 48 . Furthermore, prepro-orexin knockout mice showed weaker cardiovascular and locomotor responses to emotional stress in awake and freely moving conditions 49 . The central amygdala receives multiple nociceptive information from the brainstem, as well as highly processed polymodal information from the thalamus and the cerebral cortex 50 . However, the amygdala and the parabrachial nucleus mediate quite varied physiological responses, therefore it is difficult to decipher specific functional pathway conveying nociceptive information in these areas.
The C1 neurons located in the rostral ventrolateral medulla are also interesting candidates. Comprehensive analysis using transgenic mice expressing a retrograde tracer, GFP-TTC, selectively in orexin neurons identified several brain regions including basal forebrain cholinergic neurons, serotonergic neurons in the raphe nucleus, and neurons in the ventrolateral medulla 51 . Moreover, it was confirmed that C1 neurons innervate the orexin neurons by immunohistochemistry and electron microscopy 52 . Considering that C1 neurons are activated by noxious stimuli and involved in autonomic responses 53 , orexin neurons might mediate the transmission of nociceptive information from the C1 neurons during nociception. Although activation of adrenaline receptors in orexin neurons inhibits neuronal activity 54 , the synapses formed by C1 neurons with orexin neurons are densely filled with small clear vesicles likely containing glutamate 52 . Indeed, pharmacogenetic activation of A1/C1 catecholamine neurons results in activation of orexin neurons 55 . To confirm this, retrograde trans-synaptic infection of recombinant rabies viral vectors in selective types of neurons 56 might enrich our understanding of the possible inputs of orexin neurons, and selective manipulation technologies will confirm the physiological importance of those pathways.
Interestingly, orexin neurons could not be activated by noxious stimuli under anesthesia by isoflurane. It was reported that isoflurane and sevoflurane inhibit c-Fos expression in orexin neurons 57 , that selective ablation of orexin neurons delays the emergence from anesthesia 57 , and that orexin A facilitates emergence of the rat from isoflurane anesthesia 58 . These findings suggest the possibility that inhibition of orexin neurons by anesthetics plays a role in maintaining stable anesthesia.
Correlative activation of orexin neurons might affect memory formation by modulating the vigilance level of the entire brain. Noxious stimuli, such as foot-shock, are commonly used as unconditioned stimuli in associative learning. Immediate memory formation to avoid the succeeding experience is very important for survival. Indeed, orexin receptor 1 in the locus coeruleus is involved in fear memory consolidation 59 . It was also reported that orexin A has a beneficial inhibitory effect on orofacial pain-induced deficits in spatial learning ability and memory 60 . Our results show that noxious stimuli immediately activate orexin neuronal activity. The role of orexin neurons in linking nociception and cognition is an interesting topic for future study.
We found that temporally-controlled ablation of orexin neurons in adult mice results in enhanced pain-related behaviors against noxious mechanical, thermal and chemical stimuli. Previous studies have also shown the involvement of the orexin system in nociception. Intravenous injection of orexin A produced analgesic effects in a hot plate test and a carrageenan test 21 . Intra-PAG microinjection of orexin A decreased formalin-induced nociceptive behaviors 61 , and this analgesic effect is mediated by the orexin 1 receptor and endocannabinoid signaling 17 . Orexin neurons innervate the spinal cord 12 (Fig. S1), and the spinal cord also expresses the orexin 1 receptor 62 . It has been reported that the spinal orexin 1 receptors mediate the anti-hyperalgesic effects Scientific RepoRts | 6:29480 | DOI: 10.1038/srep29480 of intrathecally-administered orexins in a rat model of diabetic neuropathic pain 63 . Our results enhance these findings and confirm the cellular role of orexin neurons in nociception. While prepro-orexin null mice show stress-induced analgesia, their baseline pain thresholds are the same as in wild type mice 36 . This behavioral difference between prepro-orexin null and orexin-tTA; TetO DTA mice might imply the concerted effect of co-transmitters within the orexin neurons. It was also reported that orexin neuron-ablation in transgenic mice using a neurotoxic protein, polyglutamine repeat of ataxin-3, resulted in stress-induced analgesia 64 . However, it should be noted that in the hot plate test, a significant difference appeared only after application of strain stress between wild type and orexin-ataxin-3 mice. Our previous report showed that the sleep fragmentation and cataplexy phenotype in orexin-tTA; TetO DTA mice was much stronger than in orexin-ataxin-3 mice 31 . Compensation during development or higher ablation efficiency might explain the different experimental results between orexin-tTA; TetO DTA mice and orexin-ataxin-3 mice. Our DREADD experiments revealed that activation of orexin neurons exerts analgesic effects for several hours. However, the initial time course in the formalin test was similar. This might suggest that i.p. injection itself activates endogenous analgesic mechanisms including orexin neurons. During the hot and cold plate tests, hyperactivity after CNO can be a confounding factor as the mice have less contact with the floor so there may be less heat transfer to the paw. The consistency between the results of neuronal activation and neurotransmitter administration is important because it is possible that the effects of excessive peptide administration do not reflect the physiological role of the peptide-containing neurons in vivo.
In addition, we observed a marked reduction of dynorphin-expressing neurons in the LHA by selective ablation of orexin neurons. Dynorphin plays critical roles in nociceptive processing 65 , and possibly acts with orexin to induce complex effects on downstream neurons. Although orexin and dynorphin exert opposing actions on motivated behavior 66 , the combined effect of these neuropeptides in nociception may be synergetic. One example of the discordant effect of these two peptides is that an opioid receptor antagonist did not inhibit the anti-nociceptive effects of orexin peptides 18 , while cocaine self-administration in OX1R KO mice was restored by kappa opioid receptor blockade 66 . Neurotensin is also expressed in orexin neurons and is known to regulate nociception 67 . Therefore, the analgesic effects caused by orexin neuronal activation might result from the combination of orexin peptides and other co-transmitters such as dynorphin or neurotensin. Considering that placebo analgesia is involved in the opioidergic descending pain inhibitory system 68,69 , the concerted action of orexin and dynorphin might be important for acute pain perception. Note that, similar to our previous report 13 , we did not employ colchicine treatment in our experiments, although many studies use colchicine to block axoplasmic transport and enhance the immunostaining signal of dynorphin in the soma.
Taken together, noxious stimuli induced correlated activation of orexin neurons, and pharmacogenetic activation of orexin neurons attenuated pain perception by heat and cold stimuli. Although it is difficult to determine from our experiments specific neuronal projection involving nociceptive processing, the use of sophisticated genetic tools that have recently been developed 39 might help to dissect the complex physiological role of orexin neurons in the future. Our new findings suggest an integrative role for orexin neurons in nociceptive perception and analgesia, and highlight the linkage between nociception and wakefulness.

Behavioral tests. von Frey test.
Sensitivity to mechanical stimuli was tested using self-made von Frey filaments 71 . Mice were individually placed into plastic containers with perforated metal floors (Model 37450, Ugo Basile, Varese, Italy), and acclimated for at least 30 min before testing. A series of 10 filaments (bending forces 0.2~6.8 g in quasi-logarithmic order, 0.5 mm in diameter) was applied to the midplantar surface of the left hind-paw. We applied the weakest filament first, then each filament was applied twice at intervals of 1 min. The threshold was determined as the minimum force eliciting at least one clear withdrawal response of the paw within the two trials.
Hargreaves test. Thermal sensitivity to noxious heat was tested using a Hargreaves apparatus (Model 7371, Ugo Basile, Varese, Italy) 71 . Mice were placed into individual plastic cages with glass floors 30 min before the Scientific RepoRts | 6:29480 | DOI: 10.1038/srep29480 experiments. The withdrawal latency was measured in response to an infrared beam (intensity: IR40) applied to the plantar surface of the left hind-paw. The cut-off latency was set at 20 s to avoid damaging the tissue by heat.
Cold plate test. The nociceptive threshold to cold stimuli was measured by the cold plate test. Mice were placed on a cold stainless steel plate surrounded by clear Plexiglas (12 × 20 × 10 cm). The temperature of the cold plate was continuously monitored and kept at − 5 °C by circulating water containing 20% ethylene glycol beneath the plate. Pain-related behaviors such as hind-paw lifting, licking and jumping were observed and videotaped. The total duration of the pain-related behaviors was measured during a 3 min observation period.
Formalin test. Following an acclimation period of at least 30 min, 10 μ l of 5% formalin dissolved in 0.01 M PBS was injected subcutaneously into the plantar surface of the left hind-paw with a 30-gauge needle. Mice were returned to the plastic box immediately after the injection. The total duration of pain-related behaviors (i.e. licking and biting of the injected side hind-paw) were measured every 5 min up to 50 min after the injection. We identified and analyzed two phases in the pain-related behaviors: an initial acute phase (phase I, 0-5 min after the injection) followed by a prolonged tonic phase (phase II, 10-45 min after the injection). Stereotaxic AAV injection. Surgeries for AAV injections were conducted under pentobarbital anesthesia (50 mg/kg, i.p.) and isoflurane (2%, inhalation) using a stereotaxic instrument (David Kopf Instruments, Tujunga, CA, USA). Recombinant AAV-hSyn-FLEX-hM3Dq-mCherry (serotype: DJ; 600 nl/injection, 3 × 10 12 copies/ml) was stereotaxically and bilaterally injected into the LHA of orexin-Cre mice. A glass micropipette pulled with a puller (Sutter Instrument Novato, CA, USA) with a tip diameter of 100 μ m was filled with AAV. An air pressure injector system (Pneumatic PicoPump; World Precision Instruments, Inc., Sarasota, FL, USA) was connected to the glass micropipette with a polyethylene tube. Air pressure (10-20 psi) was applied to inject the AAV. Injection sites were as follows: from bregma − 1.4 mm, lateral ± 0.7 mm, ventral − 5.0 mm for AAV-hSyn-FLEX-hM3Dq-mCherry. Three weeks after the AAV injection, mice were subjected to behavioral experiments.

Adeno-associated virus (AAV) production and purification.
Immunohistochemistry. Mice were deeply anesthetized with isoflurane and transcardially perfused with 20 ml of chilled saline, followed by 20 ml of chilled 10% formalin solution (Wako Pure Chemical Industries, Ltd., Osaka, Japan). The brain was removed, post-fixed in 10% formalin solution at 4 °C overnight, and immersed in 30% sucrose in PBS at 4 °C for at least 2 days. A series of 40 μ m sections were obtained with a cryostat (Leica CM3050 S; Leica Microsystems, Wetzlar, Germany).
Brain slices were transferred into a recording chamber (RC-27L; Warner Instruments, USA) on a fluorescence microscope stage (BX51WI; Olympus, Tokyo, Japan) and were superfused with bubbled (95% O 2 and 5% CO 2 ) bath solution at rate of 0.8 ml/min using a peristaltic pump (Dynamax, Rainin, USA). An infrared camera (C2741-79; Hamamatsu Photonics, Hamamatsu, Japan) was installed in the fluorescence microscope along with a charge-couple device camera (Evolve 512 delta; Photometrics, USA) and both images were separately displayed on a monitor. A micropipette puller (P-1000; Sutter Instruments) was used to prepare the patch pipette (GD150-10, Harvard Apparatus, USA) of 5-8 MΩ resistance. Patch pipettes were filled with a KCl-based internal solution (in mM: 145 KCl, 1 MgCl 2 , 10 HEPES, 1.1 EGTA, 2 MgATP, 0.5 Na 2 GTP; pH 7.3 with KOH) and the osmolarity of the solution was confirmed to be 280-290 mOsm. Orexin neurons were identified by green fluorescence of G-CaMP6. Positive pressure was introduced within the patch pipette and it was moved toward the cell. Upon contacting the cell, the pressure was released and a giga-seal was made. The patch membrane was ruptured by suction to form a whole-cell configuration. The membrane potential was monitored with an Axopatch 200B amplifier (Axon Instrument, USA). Orexin neurons were hyperpolarized by negative current injection around − 70 to − 80 mV through an amplifier to stop spontaneous firing. To generate faithful firing, a depolarizing rectangular current of ~50 pA, width of 10 msec, frequency of 5, 10, 20 and 50 Hz was applied using an electric stimulator (SEN-3301; Nihon Kohden, Japan) and was applied to the cell through the recording pipette. The output signals were low-pass filtered at 5 kHz and digitized at a 10 kHz sampling rate. Patch clamp data were recorded through an analog-to-digital converter (Digidata 1322A; Axon Instruments, Molecular devices, USA) using pClamp 10.2 software (Molecular Devices, USA).
Calcium imaging. Brain slices were transferred into a recording chamber and orexin neurons were identified by green fluorescence of G-CaMP6 33 . Slices were superfused with bubbled (95% O 2 and 5% CO 2 ) bath solution at the rate of 0.8 ml/min. Excitation light for G-CaMP6 was emitted from a light source (Spectra light engine; Lumencor, Beaverton, OR, USA) controlled by Metamorph software (Molecular Devices, Sunnyvale, CA, USA). Light was guided to the microscope with a liquid light fiber with a diameter of 1 cm. Brain slices were illuminated with blue light of 475 ± 17.5 nm wavelength and 9.7 mW power through the objective lens of a fluorescence microscope. G-CaMP6 fluorescence intensity was recorded continuously using Metamorph software at a rate of 2 Hz with 100 msec of exposure time. To synchronize the calcium imaging and patch clamp recording, pClamp software was triggered by the TTL output from Metamorph software. Metamorph data were analyzed by setting the region of interest (ROI) on G-CaMP6 expressing orexin neurons and the Δ F/F was calculated from the average intensity of the ROI.
In vivo recordings of neuronal activity using fiber photometry. A fiber photometry system (COME2-FTR/OPT, Lucir, Tsukuba, Japan) was used to record the activity of orexin neurons in conscious mice. This system utilizes a single silica fiber to deliver excitation light and detect fluorescence from G-CaMP6 simultaneously. Excitation blue light (465 nm, 0.5 mW at the tip of the silica fiber) was produced by a high-power LED system (PlexonBright OPT/LED Blue_TT_FC, Plexon, Dallas, TX, USA). The excitation blue light emitted from the LED was reflected by a dichroic mirror and coupled to a 400 μ m silica fiber through an excitation bandpass filter (path 472 ± 35 nm). G-CaMP6 fluorescence was collected by the same silica fiber and guided to a photomultiplier (PMTH-S1M1-CR131, Zolix instruments, Beijing, China) passed through a bandpass emission filter (path 525 ± 25 nm). The signal was digitized using an A/D converter (Micro1401), and recorded by Spike2 software (Cambridge Electronic Design Limited, Cambridge, UK). Signals were collected at a sampling frequency of 100 Hz and the software averaged every 10 samples to minimize fluctuations and noise. In order to deliver blue light to orexin-neurons, the silica fiber was implanted in the LHA.
At least 3 weeks after viral injection, mice were surgically implanted with a silica fiber for recording of orexin neurons. The fiber was placed just above the LHA (from bregma − 1.5 mm, lateral ± 0.8 mm, ventral − 5.0 mm). Recording was started once the mice had recovered from anesthetization. Forceps with a force sensor (PIS-2001, manufactured by Aizawa S., Goto College of Medical Arts and Science, Tokyo, Japan) were used for mechanical stimulation. Pinch stimulation with the forceps was applied to the tail at forces of 100 g, 200 g and 300 g in either conscious mice or anesthetized mice. Additionally, a heating probe with a heat sensor was used for heat stimulation. A ramp-shaped noxious heat stimulus gradually increasing from 30 °C to 50 °C or 60 °C at 2 °C/sec was applied to the left hind-paw using a thermal stimulator with a circular probe (diameter: 1 mm, intercross-2000 N, Intercross, Tokyo, Japan). The probe was then cooled by circulating water, and this gradient of cooling was approximately 1.5 °C/sec.