Optogenetic induction of hibernation-like state with modified human Opsin4 in mice

Summary We recently determined that the excitatory manipulation of Qrfp-expressing neurons in the preoptic area of the hypothalamus (quiescence-inducing neurons [Q neurons]) induced a hibernation-like hypothermic/hypometabolic state (QIH) in mice. To control the QIH with a higher time resolution, we develop an optogenetic method using modified human opsin4 (OPN4; also known as melanopsin), a G protein-coupled-receptor-type blue-light photoreceptor. C-terminally truncated OPN4 (OPN4dC) stably and reproducibly induces QIH for at least 24 h by illumination with low-power light (3 μW, 473 nm laser) with high temporal resolution. The high sensitivity of OPN4dC allows us to transcranially stimulate Q neurons with blue-light-emitting diodes and non-invasively induce the QIH. OPN4dC-mediated QIH recapitulates the kinetics of the physiological changes observed in natural hibernation, revealing that Q neurons concurrently contribute to thermoregulation and cardiovascular function. This optogenetic method may facilitate identification of the neural mechanisms underlying long-term dormancy states such as sleep, daily torpor, and hibernation.


In brief
To control the induced hibernation-like state with higher time resolution in mice, Takahashi et al. develop an optogenetic method using human opsin4 (OPN4), a G protein-coupled-receptor-type blue-light photoreceptor. Modified OPN4 stably induces hypothermia for 24 h via extremely low-power light illumination.

INTRODUCTION
Recent advances in neuronal manipulation techniques with appropriate operating periods and temporal resolutions have enabled the study of the effects of the excitation or inhibition of particular neuronal circuits on behavior. Optogenetics has helped elucidate the consequences of manipulation of particular neurons in vivo at a high spatiotemporal resolution, providing important insights in neuroscience. 1,2 Channel-rhodopsin2 (ChR2) is the most commonly used opsin in optogenetics and evokes action potentials in neurons using brief pulses of bluelight illumination. 3,4 Since the first published report of applying ChR2 for optogenetics, various modified ChR2s have been developed to improve the sensitivity (e.g., ChR2/H134 5,6 ) and alter the absorbance spectrum (e.g., crimson, 7 ChRmine 8,9 ). ChR2 and its variants are conventionally used for relatively short-term (seconds to minutes) manipulations, and it is challenging to examine their long-term effects (hours to days) on physiological changes of slow processes, such as circadian rhythms, 10 sleep, 11 and torpor 12 with optogenetics. Stabilized step function opsin (SSFO), a bistable excitatory variant of ChR, sustains an open state for 30 min by a single brief light pulse; 13 however, the excitatory manipulation of neurons for over several hours remains challenging. Most neuropeptide MOTIVATION Optogenetics is used to manipulate neurons at a high temporal resolution to understand the neuronal mechanisms of many physiologies. Increasing evidence suggests that current tools are effective for short-term manipulation (seconds to minutes); however, the effects of manipulating neurons are difficult to examine for a longer period (hours to days) on long-term behavioral and physiological responses, such as hibernation. This study aimed to establish a method that allows stable and long-lasting ($24 h) neuronal manipulation with high reproducibility using a modified OPN4 (melanopsin), a highly photosensitive G protein-coupled-receptor-type photosensor. and neuromodulator receptors, such as monoamines and acetylcholine, are G protein-coupled receptors (GPCRs), 14 which evoke slow modulations of neuronal activity, such as changes in membrane potential and input resistance, compared with ion channel receptors. Therefore, a tool for light-inducible G protein signaling would be valuable for examining the functions governed by metabotropic neurotransmission.
Hibernation serves as a biological strategy for thermostatic mammals to adapt to harsh environments, typically described as regulated, hypothermic, and hypometabolic states that last longer than 24 h. 12,[15][16][17] We recently identified a neuronal population that induces a hibernation-like hypometabolic state in mice and rats, 18 which normally do not hibernate. 19 These neurons, called quiescence-inducing neurons (Q neurons), express a gene encoding a neuropeptide, pyroglutamylated RFamide peptide (QRFP), and reside in the anteroventral periventricular nucleus (AVPe) in the preoptic area (POA) of the hypothalamus, 18 known as the thermoregulatory region, which regulates the thermogenic capacity of brown adipose tissue (BAT) via the sympathetic nervous system in mice. 20 The induced long-lasting (for several days) hypothermic/hypometabolic state, which we termed the Q-neuron-induced hypothermic/hypometabolic state (QIH), was similar to hibernation, including resetting of the body temperature set point. 18 To induce QIH, we used chemogenetics, a stimulatory hM3Dq designer receptor exclusively activated by the designer drug (DREADD) system. 21 The mouse core body temperature (T B ) and metabolic rate dropped immediately after clozapine-N-oxide (CNO) administration, while mice recovered from hM3Dq DREADD-mediated QIH (hereafter, QIH M3Dq ) slowly, requiring several days to fully recuperate. 18 From this perspective, the kinetics of T B changes in QIH M3Dq do not recapitulate those observed in natural hibernation, especially in terms of return from the hypometabolic state. 12,22 This slow recovery process hinders the examination of the effects of QIH-related hypometabolism on various physiological processes, during which the effects of QIH on various physiological functions may be obscured. Therefore, we attempted to develop a technique that would allow the excitatory manipulation of specific neurons over a long period ($24 h) with high temporal resolution.
The opsin family of proteins is a GPCR-type photoreceptor found in animals. 23 Among them, opsin4 (OPN4; also known as melanopsin) is expressed only in a small subset of retinal ganglion cells, intrinsically photosensitive retinal ganglion cells (ipRGCs), in primates and rodents. 24,25 OPN4 mainly couples to the Gq subclass of G proteins and responds to blue light, playing an important role in non-visual photoreception in mammals, such as the pupillary light reflex and light entrainment of the circadian clock. 24,26 Exogenously expressed human OPN4 induces a transient current evoked by blue light in Purkinje cells, which is turned off by yellow light, 27 suggesting its usefulness as an optogenetic tool. A previous study showed that orexin-producing neurons were excited by hOPN4 with shortterm (5 s) photostimulation, resulting in the induction of wakefulness in mice. 28 Although OPN4 appears to be an ideal tool for long-term neural manipulation based on its properties, neural manipulation for more than several hours has not been reported.
Herein, we provide evidence that optogenetics with modified OPN4 is useful for inducing and maintaining long-term (up to 24 h) QIH with rapid recovery in mice, which is quite similar to hibernation. C-terminally truncated OPN4 (OPN4dC) was activated by low-power light, allowing us to induce QIH with very low light intensity, which even enables non-invasive transcranial stimulation with blue-light-emitting diodes (LED). This study shows that GPCR-type optogenetic manipulation of neurons is a powerful technique for examining neural function in slow and sustained physiological processes (hours to one day) with a high spatiotemporal resolution.

RESULTS
OPN4dC optogenetics evokes hibernation-like state with high time resolution To manipulate Q neurons with a higher time resolution, we established optogenetics with modified human OPN4 to induce long-term QIH in mice. The C-terminal region of mammalian OPN4 contains a cluster of phosphorylation sites required for arrestin binding, leading to OPN4 inactivation. 29,30 Thus, the truncation of this region or the introduction of mutations at the phosphorylation sites in the C-terminal tail of OPN4 may enhance signal transduction. We tested OPN4dCs, OPN4 with mutations at nine phosphorylation sites (OPN4(9A)), and OPN4 with mutations at the phosphorylation sites and a deletion of the subsequent C-terminal region (OPN4(9A)dC) for optogenetic induction of QIH ( Figure 1A). To target Q neurons, Cre-dependent adeno-associated viral (AAV) vectors carrying Opn4 and its variants were injected into the AVPe of Qrfp-iCre knockin mice. 18 After continuous illumination by a 473 nm blue laser with optic fiber (3 mW light power at the fiber tip, 100 mW mm À2 irradiance, 6 h stimulation period) in the AVPe, the number of c-Fos-positive neurons increased in OPN4-expressing neurons, showing that Q neurons were efficiently excited ( Figure 1B). Under the same conditions, we measured the surface temperature in the interscapular area in which BAT is located (hereafter referred to temperature around BAT [T BAT ]) using infrared thermal cameras. The photostimulation of Q neurons with OPN4dC triggered eminent drops of T BAT , and hypothermia was stably sustained for 6 h in mice ( Figure 1C). Notably, any form of OPN4s induced QIH efficiently and continuously while shining light with power of 3 mW (100 mW mm À2 ), which was much lower than that usually used for other optogenetic methods (3-30 mW). 31 The C-terminal deleted form, OPN4dC, lacking all phosphorylation sites, was the most effective, while the amounts of OPN4 expression in cells were compatible with each variant (Figures 1C and S1). Thus, we used OPN4dC fused with mCherry in subsequent experiments and referred to the OPN4dC-mediated QIH as QIH OPN4dC .
Qrfp-iCre mice expressing OPN4dC-mCherry in the AVPe (Q-OPN4dC mice) showed immobility, rapid decline in body temperature, and vasodilation in tails 2 min after the initiation of light illumination ( Figure 1D), which is similar to QIH M3Dq . However, contrary to the slow recovery observed in QIH M3Dq , T BAT and T B increased rapidly immediately after light termination in QIH OPN4dC , taking approximately 30 min to reach normal levels ( Figures 1C and 1D; Video S1). We confirmed that female mice also showed a similar sufficient and persistent hypothermia using the same method ( Figure 1E) because sex differences exist in fasting-induced daily torpor and that the duration of torpor is longer in female mice. 33,34 The optogenetic induction of QIH may help identify the physiological changes in the recovery/arousal phase of hibernation and the mechanisms underlying sex differences. ) AAV. AAV carrying hOPN4-, hOPN4dC-, hOPN4(9A)-or hOPN4(9A)dC-mCherry, respectively, was bilaterally injected into the AVPe in Qrfp-iCre mice. Q-mCherry mice expressing mCherry in Q neurons were used for the control group. Optic fibers were unilaterally implanted above the AVPe. The images representing each OPN4 structure were modified from a figure in a previous report. 32

Activation of OPN4dC leads to membrsane depolarization and calcium mobilization in neurons
We performed in vitro studies for the protein characterization of OPN4dC to indicate that blue-light illumination evoked intracellular signaling that excites neurons. First, the membrane localization of OPN4dC-mCherry and wild-type OPN4-mCherry was confirmed in HEK293T cells ( Figure S2A). Since mammalian OPN4 mainly activates Gq signaling, 35 we measured the concentration of inositol 1-phosphate (IP1), which is a metabolite in the Gq-signaling cascade. 36 The IP1 assay revealed that continuous blue-light illumination (1 mW/mm 2 for 20 or 30 min) increased the amount of IP1 in cells expressing either wild-type OPN4 or OPN4dC and that OPN4dC had a higher effect than that of the wild type ( Figure S2B). Cells expressing hM3Dq treated with CNO for 20 min generated much higher amounts of IP1 than OPN4s ( Figure S2B). We also confirmed the induction of Ca 2+ transients using blue light. In either the assay system using fluorescence-based Ca 2+ indicator (Fluo4-AM) or bioluminescence-based indicator (Nano-lantern), OPN4dC exhibited a higher Ca 2+ response to light than that of wild-type OPN4, as expected ( Figures S2C and S2D). Next, luciferase (Luc) activity driven by the nuclear factor of activated T cell (NFAT) response element was used to examine the longer effects of blue light on OPN4-mediated Gq signaling. 37 LED light illumination (1 mW/mm 2 for 20 min) increased Luc activity, which was sustained for a longer time in cells expressing OPN4dC compared with those expressing wild-type OPN4 ( Figure S2E), while the protein expression levels did not significantly differ between them. OPN4dC was sensitive in this assay (half-maximal effective concentration [EC50]; 9.95 nW/mm 2 ) for blue LED light stimulation, and 475 nm blue light efficiently induced the cellular response, compared with the other wavelengths of light with the same power (405 nm violet and 520 nm green) ( Figures S2F and S2G), which was consistent with previous studies using full-length OPN4. 24,29 Ectopic expression of wild-type hOPN4 demonstrably increases the firing rate of orexin-producing neurons, which is a relatively slow and long-lasting response. 28 We confirmed that OPN4dC induced action potentials in primary cortical neurons in culture by whole-cell patch-clamp recording using a 20 s blue-light stimulus ( Figure S3A). Exposure to blue light (120 mW at the fiber tip, 150 mW/mm 2 , 20 s) depolarized the membrane potential and increased the firing rate ( Figures S3A-S3C). The latency to the first action potential firing from the start of light illumination was 6.09 ± 3.83 s (n = 7 neurons from three independent preparations), which is longer than channel-type optogenetic proteins (e.g., ChR2 11 ). Moreover, the increase in firing rate and membrane potential depolarization was sustained even after the termination of light stimulation under this condition, while the duration varied among neurons ( Figures S3B and S3C). We then performed in vivo electrophysiology recordings in Calb1positive pyramidal neurons of the hippocampal CA1 38 in freely moving mice and showed that their firing frequency was increased during non-invasive stimulation (10 mW at the fiber tip positioned on the brain surface, 10 s, 10 times) ( Figure S3D). In this case, the firing frequency returned to the basal level within 15 s after stimulation. Neurons expressing OPN4dC persistently showed a light response during repeated stimulation ( Figure S3E).
We then confirmed whether Q neurons expressing OPN4dC in the slice preparations responded to light stimulation by Ca 2+ imaging with GCaMP6s. 39 The intracellular Ca 2+ concentration in Q neurons was clearly increased during stimulation (75 mW at the fiber tip positioned above the neurons, 3 s, 10 times) and returned to the baseline immediately after the termination of photostimulation ( Figure S3F). These results indicate that the C-terminal deletion of OPN4 enhances the intracellular response and that OPN4dC has an obvious excitatory effect on neurons.
In vivo characterization of hOPN4dC Next, we optimized the stimulation condition of OPN4dC at the behavioral level ( Figure 2A). To assess the effect of light intensity on QIH, we applied various laser powers (0.1, 1, 3, or 10 mW at the fiber tips) in sequence to induce QIH OPN4dC ( Figure 2B). All conditions we tested induced marked hypothermia, but 0.1 mW (3.2 mW mm À2 ) light had a relatively weak effect. These data indicate that OPN4dC expressed in Q neurons responds to extremely weak light (as low as 0.1 mW) and that the effect is saturated at 3 mW ( Figure 2B). However, the hypothermia did not persist at a higher light intensity (100 mW at the fiber tips), although the laser power was lower than that normally used ( Figure 2C). In a previous study, we succeeded in inducing short-term hypothermia by SSFO with moderate light intensity (10 mW, 1 s width, every 30 min), 18 whereas the light intensity suitable for OPN4dC was insufficient for either SSFO or ChR2 under all tested conditions ( Figure S4). This indicates that OPN4dC and other opsins are efficiently activated under specific light conditions to be optimized according to their properties.
The peak of spectral sensitivity of mammalian OPN4 is approximately 480 nm, which lies in the blue/cyan range of visible light. 24 We next examined the dependence of wavelength on the stimulation of OPN4dC through QIH induction (10 mW at the fiber tip, 1 h; Figure 2D). Activation of OPN4dC using blue (473 nm) or green (532 nm) lasers triggered a remarkable drop in T BAT , whereas the effect of yellow light (589 nm) was weaker. We did not observe any effect of red light (632 nm) on T BAT , suggesting that OPN4dC could be used in combination with other optogenetic tools (e.g., red calcium indicator). This wavelength dependency of OPN4dC activity is consistent with a previous report on full-length OPN4 24 and our in vitro study ( Figure S2G).
To determine whether the decrease in body temperature depended on the frequency and duration of each light pulse, we applied light illumination to Q neurons under various stimulation conditions (10 mW at the fiber tip, 6 h). Photostimulation at 0.1, 1, 2, and 10 Hz decreased T BAT in a frequency-dependent manner in the early phase (0.5-1 h) ( Figure 2E). However, the 10 ms pulse conditions were insufficient to sustain hypothermia during the stimulation ( Figure 2F). Among these conditions, only 1 Hz, 500-ms-wide light pulses and continuous stimulation succeeded the long-lasting QIH, suggesting that appropriate amounts of photons are necessary for OPN4dC activation ( Figure 2E).

OPN4dC-mediated axonal fiber excitation
We previously reported that the dorsomedial hypothalamus (DMH), which is implicated in thermoregulation, 40 is the main target of Q neurons. 18 Since OPN4dC was expressed in the axons of Q neurons in the DMH, we attempted optogenetic activation of OPN4dC in the axonal fibers to induce behavioral changes. We implanted optic fibers in the DMH unilaterally in Q-OPN4dC mice and performed axonal photostimulation (3 mW at the fiber tip, 6 h; Figure 2G). Although the effect on body temperature was slightly weaker than that on cell bodies, unilateral axonal stimulation in the DMH effectively induced T BAT reduction and c-Fos expression ( Figure 2H). This result indicates that OPN4dC is also useful for examining neuronal functions at the circuit level.
OPN4dC optogenetics maintains hypothermia for 24 h with good reproducibility We next attempted to optogenetically induce a 24-h-long QIH. First, we used the commonly used opsins ChR2 and SSFO. For the use of ChR2, we tested various stimulating frequencies and found that 2 Hz was the most effective for the short-term induction of QIH. Both opsins induced a profound hypothermic phenotype, which persisted for approximately 10 h with ChR2 Article ll OPEN ACCESS and 3 h with SSFO. However, we failed to maintain QIH expression for 24 h using these opsins ( Figure 3A). In contrast, OPN4dC was able to induce a 24-h-long QIH with weak light (3 mW at the fiber tip, 24 h; Figure 3B). Even though the mice maintained a hypothermic state during the stimulation, we observed a small increase in locomotor activity (Act) in the dark period, which correlated with T B (Figure 3B, bottom). This increase was reproducible in each attempt with the same time course, suggesting the possibility that, even during the long robust hypothermia, a circadian influence remained and that mice became active in the dark period. The T B change during QIH OPN4dC was similar to that of QIH hM3Dq , especially during the induction phase ( Figure 3B). However, QIH OPN4dC exhibited rapid recovery from hypothermia after lights off, which is in stark contrast to QIH hM3Dq . We repeated 24 h-long stimulation four times at an interval of 2 days (3 mW, 24 h per stimulation) to examine the reproducibility of QIH OPN4dC ( Figure 3C). Q-OPN4dC mice (male, n = 2; female, n = 2) showed stable QIH every time, with small variations among individuals and trials ( Figures 3D and 3E). The body temperature in all individuals returned to normal within 30 min of the end of the laser stimulation ( Figure 3F). The good reproducibility within the same mouse suggested that continuous laser stimulation for as long as 24 h did not cause significant tissue/cell damage or bleaching of OPN4dC. mW-scale photostimulation, which is suitable for OPN4dC optogenetics, Article ll OPEN ACCESS did not induce the expression of tissue inflammation and injury markers, glial fibrillary acidic protein (GFAP), and ionized calcium-binding adaptor molecule 1 (Iba1) ( Figure S5). In contrast, stimulation with 10 mW at 2 Hz caused a significant increase in Iba1 and GFAP immunoreactivity ( Figures S5C-S5E), suggesting that ChR2 may fail to sustain QIH due to local brain tissue damage. Collectively, these results indicate that OPN4dC optogenetics in Q neurons works without causing tissue damage and with extreme sensitivity, sustainability, and reproducibility.

Fiberless transcranial manipulation of Q neurons
The high photosensitivity of OPN4dC inspired us to examine whether transcranial light delivery using an LED device could activate OPN4dC expressed in the AVPe to non-invasively induce QIH. A small blue LED light source was positioned directly above the surface of the skull of Q-OPN4dC mice ( Figure 4A). Illuminating blue light (470 nm, 1 mA current, 250 mW power) caused a decrease in both T BAT and T B in Q-OPN4dC mice, although the efficiency was slightly weaker than that of direct stimulation of Q neurons via an optic cannula ( Figures 4B and 4C; Video S2). The control group (Q-mCherry) exhibited no changes in body temperature, indicating that LED illumination itself had no effect on the body temperature under this condition. We confirmed c-Fos expression in OPN4dC-positive Q neurons following transcranial LED exposure ( Figure 4D). This light power was low enough to not cause significant heat production in the LED device under this condition ( Figure 4B). Importantly, fiberless stimulation of OPN4dC can be applied to other types of neurons. Figure S3D demonstrates that transcranial stimulation can excite hippocampal neurons, although it remains to be elucidated whether this stimulation indeed modulates physiology or behavior in animals.
OPN4dC optogenetics reveals that cardiovascular function is actively regulated by Q neurons Hibernators arise every several days, i.e., interbout arousal (IBA), during which the body temperature rapidly returns to approximately 37 C and is maintained within the normal range for several hours. 41 They return to deep hibernation and sustain low body temperatures until subsequent IBA. In terms of drastic physiological transitions from the normal to torpid state (and vice versa), we need to know how the central and peripheral nervous systems behave in each step (induction/entry, torpor, and recovery/IBA). 42 Using QIH OPN4dC , we mimicked these states and interstate transitions in mice. We examined the changes in vital signs, including T B , T BAT , electrocardiogram (ECG), and heart rate (HR), with high temporal resolution. QIH OPN4dC mice showed a rapid decrease in HR immediately after the start of light illumination (3 mW, 6 h, same as Figure 1C), followed by a reduction in T B and T BAT (Figures 5A and 5B). Likewise, during the recovery phase, HR increased first, and then T B and T BAT slowly returned to basal levels. This result suggests that changes in cardiovascular function in the QIH are not secondary to changes in body temperature. Repeated short-term stimulation (3 mW, 1 h laser exposure with an interval of 1 h, three times) resulted in similar fluctuations in HR with high reproducibility ( Figure 5A). We used general anesthesia (isoflurane inhalation) as a control condition with hypothermia for comparison. During QIH OPN4dC , HR was decreased more rapidly and deeply compared to during anesthesia ( Figure 5C). Even when T B was almost identical in the two groups (at 2.5-3.5 h after beginning of stimulation), HR in QIH OPN4dC was lower than that in anesthesia ( Figure 5D). Similarly, when T B in QIH OPN4dC was higher than that in anesthesia (5-6 h after the beginning of stimulation), their HRs were almost the same in both conditions ( Figure 5D). These results demonstrate that changes in HR are directly regulated by Q neurons and are not secondary to decreased body temperature. The HR right after the recovery from QIH OPN4dC was higher than that in the basal condition ( Figures 5B and 5C). HR overshoot is also observed in the natural hibernation of hibernators, such as the Syrian hamster. 22 We then compared the kinetics of T B and HR between QIH OPN4dC , QIH M3Dq , and isoflurane-induced anesthesia ( Figure 5F). In the hibernation of natural hibernators, upon entering torpor, HR initially decreased rapidly, followed by a more gradual and delayed decline in T B , followed by a more gradual decrease in HR as T B decreases rapidly 22 (Figure 5E). Therefore, the scatterplot of these parameters generally takes the form of an open loop 22 ( Figure 5E). QIH M3Dq also induced a rapid decrease in T B and HR; however, these two factors were highly correlated ( Figure 5F). In anesthetized mice, the HR-T B loop was more flattened, indicating that HR variability closely correlates with T B . In QIH OPN4dC , cardiac function was suppressed in advance of the T B decrease, and the offset timing was completely controllable ( Figure 5F). In addition, the In normal condition, Q neurons expressing mCherry were stimulated by the same condition as QIH OPN4dC . In QIH M3Dq , Q neurons expressing hM3Dq were excited by CNO intraperitoneal administration (0.1 mg/kg). In anesthesia, the same mice that were used in QIH OPN4dC , and normal conditions were exposed to 1% isoflurane for 6 h without photostimulation. The start time of light exposure, CNO intraperitoneal administration, and isoflurane inhalation were set to time 0. HR for each condition is plotted as a function of the concurrent T B . The plots are presented by each color for 1 h before the stimulus onset, for 6 h of stimulation, and for 2 h of recovery. In QIH M3Dq , extra plots of the time to complete recovery are also presented (orange, 8-24 h), as no clear recovery is observed even 6 h after the initiation of stimulation. The lines and shading in (C) denote the mean and SD of each group, respectively. All p values (*p < 0.01, **p < 0.001, ***p < 0.0001) are from one-way ANOVA with Tukey's multiple comparisons test.  Figures 5E and 5F).
Overall, these results demonstrated that QIH OPN4dC is similar to natural hibernation in terms of body temperature and cardiovascular regulation.

DISCUSSION
The present study showed that OPN4dC is useful for the long-term manipulation of neurons. While the development of novel optogenetic tools has provided researchers with optimized methods for neural manipulation, almost all applications of these opsins have been applied for a short period (usually seconds to minutes), and few studies have examined their effects on long-term (sometimes lasting more than 1 day) physiological changes. DREADD is usually used for long-term manipulation rather than optogenetics, while the termination of DREADD-mediated manipulation is dependent on the clearance of CNO in vivo. Thus, the OPN4dC method has two advantages: a long-term effect, such as DREADD, and high temporal resolution. Another characteristic of OPN4dC optogenetics is that it induces Gq signaling. Optogenetics using longwavelength vertebrate cone opsin with C-terminal domains of 5-HT 1A receptor 43 and engineered mosquito OPN3 44 reportedly activates Gi signaling, resulting in inhibitory action on neurons. For Gq signaling, Opto-a 1 AR, a chimeric protein of bovine rhodopsin and the a 1A -adrenergic receptor, was developed. 45 However, as vertebrate rhodopsin is a monostable opsin, bleaching during long-term application is a concern. Indeed, cortical astrocytes expressing Opto-a 1 AR required supplement with 9-cis retinal to work properly. 46 Therefore, we used human OPN4, a bistable opsin, to induce a hibernation-like state. Ectopic expression of OPN4 has been used for trials of optogenetics-based therapy for retinal degeneration 24,47 because of its high photosensitivity to visible blue light and sluggish phototransduction in cells. 48,49 The full-length OPN4 has also been used in a few studies for manipulating Gq signaling in neurons 27,28 and astrocytes. 50,51 For instance, the excitation of orexin neurons using OPN4 with blue light (19 mW, 5 s) promoted wakefulness. 28 However, the authors of that paper failed to manipulate neurons in the hippocampus using the same method, 52 suggesting that there is something missing to make OPN4 work efficiently. We found that the truncation of C-terminal region of OPN4 significantly increased the activity of Gq signaling ( Figure S2), presumably by decreasing internalization and degradation, and, consequently, the effect on body temperature was enhanced ( Figure 1B). Another point that should be considered is that the commonly used light power (3-30 mW) 31 is too high for OPN4 to function properly. Indeed, we observed that bright blue light (100 mW) was not as effective as weak light (10 mW; Figures 2B  and 2C), although this mechanism needs to be elucidated in further studies using biochemical and electrophysiological assays. One possibility is the shortage of retinal in the brain tissue because we continuously exposed blue light to OPN4dC. However, endogenous OPN4 usually responds to strong light with high temporal resolution in ipRGCs of the retina, where the retinal can be efficiently reproduced. The slow kinetics of membrane potentials in OPN4dC-expressing neurons were similar to those of ipRGCs that express endogenous OPN4. 30 The latency to evoke a spike after light stimulation was 6.09 ± 3.83 s, and the increase of firing rate continued after the end of stimulation, while the period seems to depend on the neuronal type (for >3 min in primary neuron culture, approximately 15 s in pyramidal neurons in the hippocampus and approximately 100 s in orexin-producing neurons 28 ). Conversely, ChR2, a channel-type opsin, usually evokes spikes within several milliseconds, allowing precise control of the firing frequency by light pulse. 4 We are able to select appropriate tools for the physiologies that we would like to examine according to the properties of opsins.
OPN5-expressing neurons in the POA were recently identified as inducing hypothermia by violet-light illumination in mice, 53 and OPN5-positive neurons likely overlapped with Q neurons. 53,54 Considering the expression of such endogenous photosensitive proteins, stimulating light with appropriate ranges of intensity and wavelength must be used. Under the conditions used in this study, we did not observe any effect of blue-light stimulation on T BAT in the control mice ( Figure 1C). For long-term manipulation, light-induced tissue injury should also be considered. Exposure to 3-30 mW light could induce glial activation, a marker of inflammation in the brain, 8,55 and we observed the induction of Iba1 and GFAP expression with 10 mW light stimulation (2 Hz, 10 ms width, 24 h; Figure S5). The present study showed that the light power used in most of the experiments for evoking QIH was 3 mW (100 mW mm À2 ), but 0.1 mW (3.2 mW mm À2 ) blue laser was sufficiently strong to induce QIH (Figures 2B and 2C). The extremely low light intensity facilitates more localized light stimulation than has been achieved so far.
We also succeeded in stimulating them by transcranial, fiberless illumination of weak blue LED light (250 mW), even though Q neurons are in the deep brain. Several groups have developed methods for transcranial optogenetics. [56][57][58] For example, highly sensitive ChR such as ChRmine 8 and SOUL 55 have been applied for transcranial optogenetics. Here, we have added OPN4dC optogenetics to the list of non-invasive methods. Considering that LED are quite easy to process in terms of their shape and size, we also emphasize the usefulness of OPN4dC for photostimulation by an LED in peripheral tissues besides the CNS. Indeed, channels and GPCR optogenetics have been used in autonomic neurons of the heart. 59 In summary, the proposed method has several advantages. (1) Its effect was long lasting and stable for at least 24 h with good reproducibility. (2) OPN4dC is sensitive and enables non-invasive manipulation. (3) OPN4dC is a metabotropic receptor that mimics the slow kinetics of G protein signaling in neurons (and likely other types of cells) and provides researchers with a proper option for neural manipulation involved in particular physiological functions. As an example of OPN4 optogenetic application, we demonstrated the fluctuation of cardiovascular function during all steps of QIH in mice. OPN4-based techniques may be an excellent tool to address the mechanisms of hibernation and other physiologies in the future.

Limitations of the study
This study showed the sensitivity of the OPN4dC was sufficiently high for transcranial manipulation. However, because of concerns regarding heat production, we did not manipulate Q Cell Reports Methods 2, 100336, November 21, 2022 9 Article ll OPEN ACCESS neurons by OPN4dC using blue LED over several hours. Further studies are needed to improve the sensitivity and/or expression levels of OPN4dC to establish long-term transcranial manipulation. The possibility that OPN4dC may be activated by room light or other light sources should also be considered. For OPN4dC expressed in Q neurons in the AVPe, we observed no effect on body temperature under room light (270-280 Lux white LED) when the mouse skull was completely covered with black skin. However, further characterization of OPN4dC as well as preparing a proper experimental control (e.g. mCherry-expressing control) will be necessary, especially when this method is applied to neurons in more superficial regions of the brain.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:  . We thank the animal resource center at the University of Tsukuba for housing the mice; Dr. Y. Niwa for setting up the telemetry system; Dr. E Hasegawa for preparing some AAV vectors; Dr. Genshiro Sunagawa for discussion; and Nicholas Grooms for editing the manuscript. We would like to thank Editage (www.editage.com) for English language editing. All the brain diagrams used in figures were made based on the illustrations in Paxinos and Franklin's Atlas of the Mouse Brain. 60

DECLARATION OF INTERESTS
All authors declare they have no competing interests.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Arisa Hirano (hirano.arisa.gt@u.tsukuba.ac.jp).

Materials availability
Plasmids generated in this study will be shared by the lead contact upon request.
Data and code availability d All data reported in this paper will be shared by the lead contact upon request. d This paper does not report original code. d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Stereotaxic surgery
Qrfp-iCre heterozygous mice (12-20 weeks old) were anesthetized with ventilated 1-2% isoflurane (Pfizer, cat# 4987114133,403) and affixed to a stereotaxic frame (David Kopf Instruments) while being kept warm by a heat pad. Qrfp-iCre mice were injected with 0.15 mL AAV bilaterally into the AVPe (coordinates relative to the bregma: +0.38 mm AP, ±0.20 mm ML, À5.10 mm DV) at a controlled rate (0.1 mL/min) using a needle syringe (Neuros Syringe 1701 RN, 33 gauge, HAMILTON) and a stereotaxic injector (Legato 130, KD Scientific). The needle's position was maintained for 5 min after the injection. All AAV-injected mice were given a recovery period of at least three weeks before subsequent experiments.
For tetrode recording, Calb1-Cre mice were injected with 0.5 mL AAV-OPN4dC unilaterally into the hippocampal CA1 (À2.0 mm AP, +1.5 mm ML, À1.5 mm DV) and implanted with the microdrive. The tetrodes (14-nm diameter, nichrome wire, gold plated to 200-300 kU) in the microdrive were independently adjustable and arranged to surround the optic fiber (200 mm diameter, Thorlabs, FT200EMT) located in the center. The tip of the optic fiber was placed 2.0 mm posterior/1.5 mm lateral from the bregma and at the surface of the brain. One of the stainless steel screws attached to the skull was used as the electrical ground for recording. Postoperatively, animals were carefully monitored and i.p. injected with 0.5 mL of 10% sucrose solution (in saline) when a loss of body weight (>10%) was observed. In to 2-3 weeks postoperatively, the depth of the tetrodes was slowly and manually adjusted until sharp wave ripples and associated spikes could be stably observed. All adjustments were made in a familiar small plastic bucket (20-cm diameter) located in the recording room.

Behavioral tests
All behavioral experiments were performed in a temperature-controlled chamber (HC-100, SHINFACTORY) at stable ambient temperature set at 22 C, under a 12:12 h light-dark cycle (light ON from 8 a.m. to 8 p.m., 270-280 Lux white LED light). The mice were housed in clear synthetic resin cages (25 3 20 3 19 cm) with aspen chips. The mice were allowed ad libitum access to food and water. Mouse body temperature was monitored using an implantable telemetry system (DSI) and/or an infrared camera, as described below. As needed, mouse movements were tracked using a video camera (FDR-AX60, SONY) installed above the cages.

Infrared thermography
Thermographic analyses were performed as previously described. 18 Briefly, hair around the interscapular region of the subjects was shaved to measure the T BAT . Infrared thermographic images were obtained every 2 s (0.5 Hz) using an infrared camera (InfReC R500EX or R550S, Nippon Avionics) positioned 1 m above the bottom of the cage. The obtained images were analyzed using InfReC software (NS9500Pro, Nippon Avionics). The highest values every 10 s were used for data representation.
Telemetric vital signs monitoring Core body temperature (T B ) and HR were assessed using a telemetry system (PhysioTel and PONEMAH, DSI), as previously described. 18 In brief, to record the T B and ECG continuously, a telemetry temperature transmitter (TA11TA-F10 or TA11ETA-F10, DSI) was implanted in the abdominal cavity of mice under isoflurane anesthesia for at least seven days before the recording. Locomotor activity (Act) was monitored concurrently with the T B ( Figure 3B). For ECG measurements, two wires from the transmitter (TA11ETA-F10) were placed on the surface of the thoracic cavity. The ECG was analyzed with the PONEMAH Physiology Platform v.6.30 software (DSI) to calculate HR. Telemetry data were collected every 10 s (0.1 Hz) and used for data representation.

Correlation between T B and HR with anesthesia
The T B and HR were simultaneously recorded ( Figure 4C) using the method described above.
Animals in the ''QIH OPN4 '' and ''Normal'' groups were Qrfp-iCre mice that received AAV-hOPN4dC and AAV-mCherry into the AVPe, respectively. For anesthesia, the mice were administered 1% isoflurane for 6 h by using the inhalation anesthesia machine (MK-AT210, MUROMACHI KIKAI) at an ambient temperature of 22 C. After exposure to anesthesia for 6 h, the mice were immediately returned to the chamber in which they had been kept during the normothermic phase. Two mice were used for each group (''QIH OPN4 '' and ''Normal''). The order of the experiments was randomized and unbiased. T B -HR loop observation The data used for the ''QIH OPN4 ,'' ''Normal,'' and ''Anesthesia'' were the same as the data presented in Figure 4C. Animals in the ''QIH M3Dq '' were Qrfp-iCre mice that received AAV-M3Dq into the AVPe. The time of each phase ( Figure 4F) was designated as e4 Cell Reports Methods 2, 100336, November 21, 2022 Article ll OPEN ACCESS follows: -1-0 h: normothermic phase, 0-1 h: entry phase, 1-6 h: torpor phase, 6-7 h: arousal entry phase, and 7-8 h: arousal recovery phase. The ''0 h'' timepoint denoted the onset time of the photostimulation, isoflurane inhalation and CNO (ab141704, Abcam) intraperitoneal administration (ZT4). For QIH M3Dq , recording was continued until the mice had fully recovered (8 h-).
Optogenetic manipulation AAV-injected mice were implanted with an optic cannula (200-mm diameter, NA:0.50, 6.0 mm long, KYOCERA) unilaterally above the AVPe (+0.38 mm AP, 0.00 mm ML, À4.80 mm DV). To stimulate the axon terminal ( Figures 2G and 2H), the optic cannula was unilaterally implanted above the DMH (À1.58 mm AP, +0.25 mm ML, À4.75 mm DV). The fiberoptic cannulas implanted in mice were connected to a fiberoptic patch cable (200-mm diameter, NA:0.22, 1.0-m long, Doric Lenses) using ceramic sleeves (Thorlab). We used DPSS lasers (473 nm blue, BL473T8-100FC, Shanghai Laser) for the optogenetic manipulation. The intensity of the output laser at the optical fiber tip was measured using a laser checker (Coherent) and was adjusted to 3 mW (100 mW mm À2 ), unless otherwise indicated in the figure legend. Irradiation (mW mm À2 ) was calculated based on the light power at the fiber tip using ''predicted irradiance values'' (https://web.stanford.edu/group/dlab/cgi-bin/graph/chart.php). Laser stimulation was applied continuously for 6 h, controlled by a TTL pulse generator (Amuza), except when noted. To verify the light intensity at which hOPN4dC responded sufficiently to induce deep hypothermia, the laser power was set to 0.1, 1, 3, or 10 mW ( Figure 2B). To stimulate ChR2 and SSFO ( Figure S4), a laser was applied at 10 Hz for 10 ms-width for 6 h and at 1 Hz for 1 s width every 30 min for 6 h. To verify the spectral sensitivity of OPN4dC in vivo ( Figure 2D), the laser powers (473, 532, 589, and 632 nm, Shanghai Laser) were set to 10 mW. To test the responses of OPN4dC to various stimulus frequencies ( Figure 2E), 473 nm laser power was set to 10 mW and photostimulation was applied at each frequency (0.1, 1, 2, and 10 Hz) for 10-ms-width for 6 h.
Mice were habituated to the cages overnight prior to the optogenetic experiments. Laser stimulation was initiated at ZT4 (0 p.m.), except for Figure 3, in which the laser stimulation started at ZT6 (2 p.m.) and was applied for 24 h. After the experiment, the positions of the implanted cannula and gene expression were confirmed using immunohistochemistry. The data were included only if proteins were expressed in the AVPe by AAV infection and optic cannulas were implanted in the precise position.
Transcranial fiberless optogenetics A laboratory-made small LED light source device (8-mm diameter, 4-mm height, 0.12 g weight) was attached directly to the surface of the skull (around the bregma) of Q-OPN4dC and Q-mCherry control mice using dental cement and covered with the scalp. The device was equipped with two commercially available blue LED chips (center wavelength: 470 nm, ISO 0603 size). The LEDs were molded with transparent epoxy resin. The LEDs on the device were driven via current-controlled operation by a lab-made battery-powered LED driver equipped with a microcontroller and a Bluetooth chip. It could control the period and intensity of LED light using lab-made control software on a PC. The current was set to 1.0 mA and the output power of the LED was measured as 250 mW. After two days of recovery from the surgery, the LED device on the mouse head was connected to the LED driver with a patch cable. The mice were allowed to move freely in and habituate to the experimental cages for 1 h before light exposure.
Tetrode recording and analysis procedures Two to three weeks postoperatively, recording sessions were conducted in a familiar bucket. All the data were acquired using a 32-channel Digital Lynx 4SX acquisition system (Neuralynx). The local field potential (LFP) was filtered at 2-9,000 Hz. Spike waveforms were filtered between 0.6 and 6 kHz and those above a peak threshold of 50 V were time-stamped and digitized at 32,556 Hz. The optical fiber in the microdrive was coupled to a 473 nm DPSS laser (Shanghai Laser & Optics Century, BL473T8-300FC). Ten minutes after the baseline recording, blue light (10 mW, 10-s duration, 10-min inter-stimulus interval, 10 trials) was delivered while recording the spikes and LFP.
The multi-unit activity was analyzed using scripts written in MATLAB (MathWorks). Firing rates during the 10 s pre-or post-light onsets were calculated for the OFF or ON epochs. For each electrode, the mean firing rates during the OFF epochs were used to normalize the firing rates of both the OFF and ON epochs for pairwise comparison across the electrodes.

Membrane localization of hOPN4dC
HEK293T cells seeded on a cover glass were transfected with the expression construct, pEF1a-hOPN4-mCherry or pEF1a-hOPN4dC-mCherry. Twenty-four hours after transfection, the cellular membranes were stained with PlasMem Bright Green (P504, DOJINDO). Images were obtained using a laser confocal microscope (TCS SP8, Leica Biosystems).