A Small Brain Circuit in the Body’s Internal Clock Controls Wakefulness

WAKE‐expressing neurons in Drosophila are arousal‐promoting, as are mWAKE⁺ neurons in the dorsomedial hypothalamus (DMH) in mice.^[15, 17^] We wished to further investigate the role of mWAKE⁺ neurons in regulating arousal and wakefulness. To genetically target mWAKE⁺ neurons, while simultaneously creating a mutant allele, we previously generated transgenic mice (mWake^(Cre)) where exon 5 of mWake was replaced with a tdTomato‐P2A‐Cre cassette, which causes frameshifted nonsense mutations downstream, resulting in a loss‐of‐function allele of mWake (Figure1A). We hypothesized that, on balance, mWAKE broadly labels arousal‐promoting neurons. To test this hypothesis, we performed chemogenetic activation of most or nearly all mWAKE⁺ neurons. To do this, we crossed mWake^(Cre/+) mice with a transgenic mouse line expressing DREADD‐hM3Dq under Cre‐dependent control (LSL^Gq) to generate mWake^(Cre/+); LSL^Gq progeny; these animals were injected with low‐dose clozapine N‐oxide (CNO) or vehicle control at Zeitgeber Time 6 (ZT6). DREADD‐Gq activation of most or all mWAKE⁺ neurons induced intense wakefulness with a dramatic reduction of sleep for ≈6 h after CNO injection (Figure 1B,C; Figure S1A; Table S1, Supporting Information). Next, we asked whether chemogenetically inhibiting most or nearly all mWAKE⁺ neurons would impair arousal. We crossed transgenic mice expressing Cre‐dependent DREADD‐hM4Di (LSL^Gi) with mWake^(Cre/+) mice to generate mWake^(Cre/+); LSL^Gi progeny and injected CNO at ZT10. Electroencephalography (EEG) data demonstrated a marked reduction in signal amplitude, with slow frequency waveforms on EEG following injection of CNO, but not vehicle control (Figure 1D–G; Figure S1B; Table S1, Supporting Information). One CNO‐treated animal died >24 h after CNO injection. These phenotypes were suggestive of a stupor‐like state, rather than sleep, since the animals exhibited minimal to no responsiveness to gentle touch. Moreover, the low‐amplitude EEG waveforms were inconsistent with sleep (Figure S1C, Supporting Information); thus, the elevated delta power observed in the EEG signal likely reflects pathological, rather than sleep‐related, changes in cortical activity. For comparison, we assessed the EEG effects of chemogenetically silencing arousal‐promoting HDC⁺ (histamine decarboxylase) neurons, by repeating these experiments with Hdc^(Cre/+); LSL^Gi mice. CNO treatment of these mice led to no observable differences in EEG spectral power and amplitude (Figure S1D; Figure S1G; Table S1, Supporting Information). Together, these findings suggest that the mWAKE⁺ network plays a crucial role in promoting arousal.

Similar to the expression of WAKE in fly clock neurons, mWAKE is highly enriched in the SCN in mice. mWAKE does not label a specific sub‐cluster of SCN neurons, but instead is expressed across multiple SCN clusters (e.g., VIP‐, GRP‐, or prokineticin2/Prok2‐expressing cells).^[19^] Because our data suggest that mWAKE⁺ neurons generally promote arousal, we hypothesized that mWAKE⁺ SCN neurons play a similar role. As previously described, mWAKE is expressed in the core region of the SCN (Figure2A).^[15, 19^] To address the function of SCN^mWAKE neurons in regulating sleep or wakefulness, we unilaterally injected AAV‐DIO‐ChR2‐EYFP virus or a control virus into the SCN of mWake^(Cre/+) mice (Figure 2B,C; Figure S2A; Table S2, Supporting Information). We then performed optogenetic stimulation of the SCN, while simultaneously recording EEG data. We first examined whether this manipulation would increase wakefulness during the daytime (when mice exhibit increased sleep). No effects on wakefulness were observed with optogenetic stimulation in control animals with injections of AAV‐DIO‐EYFP virus (Figure 2D–F; Table S1, Supporting Information). In contrast, optogenetic activation of SCN^mWAKE neurons led to a significant increase in wakefulness during stimulation, which persisted afterwards for ≈30 min (Figure 2G–I; Figure S2B,C; Table S1, Supporting Information). We next assessed whether a similar arousal‐promoting effect was observed if optogenetic activation of SCN^mWAKE neurons was performed during the nighttime. Stimulating SCN^mWAKE neurons during the active phase of the mice led to an increase in wakefulness (Figure S2D–F; Table S1, Supporting Information). These data suggest that activating SCN^mWAKE neurons enhances arousal.

Because mWAKE is a clock‐output molecule, we next investigated whether mWAKE functions in SCN^mWAKE cells to regulate arousal in a time‐dependent manner. To test this possibility, we performed conditional knockout of mWAKE by bilateral injection of AAV‐Cre‐tdTomato virus into the SCN of mWake^(flox/flox) mice (Figure3A,B; Figure S3A; Table S2, Supporting Information), as we did previously for the DMH and lateral amygdala.^[17, 18^] To confirm the efficacy of mWake knockdown following viral Cre injection, we performed RNAscope in situ hybridization, which revealed a substantial reduction in mWake transcript in the SCN (Figure S3B,C; Table S1, Supporting Information), likely due to nonsense‐mediated decay. We then assayed vigilance states by EEG recordings. Knockdown of mWake in the SCN significantly increased wakefulness and reduced NREM sleep during the subjective nighttime, but not the subjective daytime, compared to sham‐injected controls, in constant darkness (DD) (Figure 3C–G; Table S1, Supporting Information). Interestingly, REM sleep was slightly increased during the daytime in the conditional knockout animals (Figure 3E,H; Table S1, Supporting Information). A similar increase in wakefulness and reduction in NREM sleep during the nighttime was observed under light‐dark conditions (LD) (Figure S4A–F; Table S1, Supporting Information). These findings suggest that mWAKE acts in SCN neurons to inhibit arousal selectively at night.

Next, we asked whether mWAKE expression in the SCN was required for circadian rhythms. A previous study using a different constitutive loss‐of‐function mWake allele (Nmf9) reported a mild but significant decrease in period length.^[20^] First, we assessed period length and rhythm strength of locomotor rhythms in constitutive mWake knockout mice, using a wheel‐running assay. No change in period length was observed, while rhythm strength showed a mild, but not significant, reduction in mWake^(‐/‐) mice, compared to littermate controls (Figure S5A–D; Table S1, Supporting Information). The slight difference between mWake^(‐/‐) and Nmf9^(‐/‐) animal phenotypes could be attributed to sex differences between males and females.^[20^] To specifically assess the role of mWAKE in the SCN on circadian rhythms, we assessed period length and rhythm strength in mice with selective knockout of mWAKE in the SCN (Figure 3A,B; Figure S3A; Table S2, Supporting Information). This manipulation had no significant impact on circadian period length or rhythm strength (Figure4A–D; Table S1, Supporting Information). We next asked whether molecular clock rhythms were affected in these animals. To assess this, we examined expression of PER2 (a core clock molecule) at CT1 (Circadian Time 1) and CT13. mWAKE knockout in the SCN did not alter PER2 day/night cycling, compared to controls (Figure S5E,F; Table S1, Supporting Information), suggesting that molecular clock rhythms remain intact following this manipulation. Thus, in contrast to its role in regulating arousal, these findings suggest that mWAKE expression in the SCN is not essential for maintaining robust circadian rhythms.

We have previously shown that rhythms of WAKE/mWAKE expression are dependent on the core circadian oscillator.^[15, 16, 17, 18^] Thus, we investigated whether CLOCK function in SCN^mWAKE neurons is also required for the rhythmic regulation of arousal. To test this hypothesis, we expressed a dominant‐negative CLOCK construct in SCN^mWAKE neurons, by bilaterally injecting AAV‐DIO‐Clock‐DN‐EYFP virus^[18^] into the SCN of mWake^(Cre/+) animals (Figure5A,B; Figure S6A; Table S2, Supporting Information). As predicted, impairing CLOCK function in SCN^mWAKE neurons produced a phenotype similar to that seen with mWAKE conditional knockout in the SCN; wakefulness was enhanced and NREM sleep was reduced during the night, compared to control animals in both DD and LD (Figure 5C–H; Figure S6B–G; Table S1, Supporting Information). We next asked whether molecular and behavioral rhythms were affected in these animals. To investigate this, we first examined PER2 expression during subjective day versus subjective night. In control mice injected with AAV‐DIO‐EYFP, PER2 levels in the SCN exhibited robust day/night differences, with higher expression at CT13 compared to CT1 (Figure S7A,B; Table S1, Supporting Information) in both mWAKE⁺ and mWAKE⁻ cells. In contrast, mice expressing Clock‐DN in SCN^mWAKE cells showed a selective loss of PER2 day/night cycling in mWAKE⁺ cells, while mWAKE⁻ cells maintained significant day/night differences (Figure S7A,C; Table S1, Supporting Information). These results demonstrate both the efficiency and cell‐type specificity of the Clock‐DN virus in impairing core clock function. In terms of behavioral rhythms in these animals, neither period length nor locomotor activity rhythm strength was affected with Clock‐DN expression in SCN^mWAKE neurons (Figure S7D–G; Table S1, Supporting Information). Taken together, these findings suggest that both mWAKE and CLOCK in SCN are required in SCN^mWAKE neurons to regulate arousal in a time‐dependent manner.

What underlying mechanism accounts for the night‐specific behavioral effects of mWAKE knockout in the SCN? In Drosophila, we previously showed that loss of WAKE leads to an increase in clock neuron spiking, specifically at night, which led to loss of daily rhythms of their activity.^[15^] Moreover, we also demonstrated that DMH neurons in mWake mutants also exhibit a selective increase in firing rate at night.^[17^] We thus asked whether mWake knockout would cause a similar phenotype in SCN^mWAKE neurons. Previous work has shown that, in general, SCN neuron spiking frequency is greater during the day versus the night.^[21, 22, 23, 24^] We performed whole‐cell patch‐clamp slice recordings and first found, as expected, that the firing of SCN^mWAKE neurons in control mWake^(Cre/+) mice followed a similar pattern (Figure6A–C; Table S1, Supporting Information). Next, we repeated these experiments in mWake^(Cre/Cre) mutants and found a loss of cycling of spiking frequency in SCN^mWAKE neurons, due to a selective increase in spiking frequency at night (Figure 6B,C; Table S1, Supporting Information). The effects of mWake knockout on SCN^mWAKE activity are likely cell‐autonomous, as the intrinsic excitability of SCN^mWAKE neurons in these mutants was similarly increased during the night, but not the day (Figure 6D,E; Table S1, Supporting Information). These data suggest that, similar to its function in the DMH and LA, mWAKE suppresses neuronal excitability and activity in the SCN at night.

What are the circuit mechanisms by which an mWAKE‐expressing subset in the SCN promotes wakefulness? One potential downstream target for mediating this phenotype is the subparaventricular zone (SPZ), a key relay region downstream of the SCN that integrates circadian signals to regulate sleep and arousal states.^[13, 25, 26, 27^] We thus hypothesized that SCN^mWAKE neurons signal to the SPZ network to promote wakefulness. To test this possibility, we unilaterally injected AAV‐DIO‐ChR2‐EYFP virus into the SCN and performed optogenetic stimulation of the terminals at SPZ during the daytime (ZT2‐6), while simultaneously recording EEG and video. Similar to the phenotype observed with activation of SCN^mWAKE cells, stimulating the SCN^mWAKE → SPZ pathway significantly promoted wakefulness (Figure7A–E; Figure S8A–C; Table S1, Supporting Information). Next, we compared the effects of activating the SCN^mWAKE → SPZ circuit versus directly stimulating SCN^mWAKE neurons. Interestingly, activating the SCN^mWAKE → SPZ circuit resulted in an overall stronger phenotype, as it reduced the time to transition from sleep to movement and increased movement speed compared to the activation of SCN^mWAKE neurons (Figure 7F–I; Video S1 and S2; Table S1, Supporting Information). Similar results were obtained when these experiments were performed during the nighttime (ZT14–18) (Figure S8D–F; Video S3 and S4; Table S1, Supporting Information). Together, these data suggest SCN^mWAKE neurons enhance arousal by signaling to a downstream SPZ region.

All experimental protocols involving animals were reviewed and authorized by the Johns Hopkins Institutional Animal Care and Use Committee (IACUC). Mice were typically housed in groups, except under specific experimental conditions described below. Housing occurred in a shared facility with a 14‐hr light/10‐hr dark photoperiod (14:10 LD), followed by a minimum two‐week acclimation period to a 12‐hr light/12‐hr dark (12:12 LD) cycle in a separate satellite room prior to testing. Genotyping assays employed Taq‐Man rtPCR probes (Transnetyx). Wild‐type C57BL/6J mice (Jackson Laboratory, stock #000664) served as the genetic background, with all experimental strains backcrossed to this line at least seven generations. The study incorporated 5 genetically modified lines, which have been previously described: an mWAKE conditional allele (mWake^flox),^[17^] an mWAKE allele with a tdTomato‐P2A‐Cre cassette integrated into exon 5 (mWake^Cre), ^[17, 18^]mWake knockout (Wake^−/−),^[17^] (LSL^Gq),^[38^] (LSL^Gi)^[38^] and (Hdc^Cre).^[39^]

Ale mice aged 8–10 weeks were anesthetized with isoflurane to induce surgical‐depth anesthesia. Following cranial fur removal, a midline incision was made along the anterior‐posterior axis of the skull. After exposing and cleaning the skull, a 3‐channel EEG headmount (Pinnacle Technology) was positioned 3 mm anterior to the bregma and secured using four manually drilled anchor screws. Bilateral EMG electrodes were implanted into the nuchal muscles, and the incision was sutured before finalizing headmount fixation with dental acrylic. All animals recovered for ≥ 2 weeks post‐surgery prior to experimental setups.

Sleep‐wake behavior was monitored using a tethered EEG/EMG system (Pinnacle Technology). Post‐recovery, mice were individually housed in cylindrical acrylic chambers (8‐inch diameter) with ad libitum access to food and water. Animals underwent a 5‐day acclimation period to tethering while maintained under a 12:12 LD cycle. Signals were amplified (100× gain), sampled at 400 Hz, and filtered (EEG: 0.5 Hz high‐pass; EMG: 10 Hz high‐pass) before digitization via Sirena Acquisition Software (Pinnacle Technology).

Sleep staging was conducted in 10 s epochs using Sirenia Sleep Software (Pinnacle Technology) by trained scorers blinded to experimental conditions. Epochs were classified as WAKE, NREM, or REM, with artifact‐contaminated intervals excluded from analysis. Animals exhibiting persistent signal noise or poor waveform quality were removed from datasets. Spectral analysis utilized custom MATLAB scripts (MathWorks) to compute Fast Fourier Transforms (FFT; 512–1024 points) with Welch's power spectral density estimation. Spectrograms were generated using short‐time Fourier transforms (30‐second Hann windows, 60% overlap).

Male mice aged 8 to 16 weeks were anesthetized to surgical depth using isoflurane, and the fur on the top of the head was completely removed. Each mouse was then secured in a Stoelting stereotaxic frame, and the microinjector tip was positioned at Bregma, with all coordinates set to zero. Small craniotomies (≈0.5 mm) were made to facilitate virus injections (50–100 nL) at a controlled rate of ≈50 nL min⁻¹. Detailed information on injection coordinates, volumes, and virus sources can be found in Table S2 (Supporting Information). Following the injection, animals were single‐housed, and given at least two weeks to recover and allow for viral gene expression. EEG head mount and/or optogenetic fiber were implanted immediately after the virus injection. The accuracy of injection sites was confirmed through post‐hoc imaging.

DREADD receptors coupled to Gi or Gq were expressed in a Cre‐dependent fashion in mWAKE⁺ neurons of mWake^(Cre/+) mice by crossing with a transgenic effector mouse B6N;129‐CAG‐LSL‐HA‐hM4Di‐mCitrine (a CAG‐promoter‐driven HA‐tagged hM4Di transgene and mCitrine fluorescent protein, both downstream of a Lox‐Stop‐Lox (LSL) sequence for Cre‐dependent expression; “LSL^Gi”), or a B6N;129‐Tg(CAG‐CHRM3*‐mCitrine)1Ute/J line (a CAG‐promoter‐driven HA‐tagged hM3Dq transgene and mCitrine fluorescent protein, both downstream of a LSL sequence for Cre‐dependent expression, “LSL^Gq”). Clozapine N‐oxide (CNO) (Sigma‐Aldrich) was prepared as a stock solution of 50 mg mL⁻¹ in DMSO, and then freshly diluted to 0.1 mg mL⁻¹ for LSL^Gi and 0.05 mg mL⁻¹ for LSL^Gq in sterile PBS before intraperitoneal (IP) injection. Solution clarity was monitored throughout dosing, and the solution was warmed to 37 °C if precipitates were observed. Vehicle control was prepared as sterile saline + 0.01% DMSO. All injections occurred at the same ZT time within each experiment, and all animals (both males and females) were treated with vehicle or CNO each day in a cross‐over design, with ≥2 days recovery between experimental recording days. Each DREADD experiment included genetic controls (CNO injections in mWake^(Cre/+); LSL^Gi or mWake^(Cre/+); LSL^Gq animals).

AAV‐DIO‐ChR2‐EYFP or a control virus (AAV‐DIO‐EYFP) was unilaterally injected into the SCN of mWake^(Cre/+) male mice, as detailed above and in Table S2 (Supporting Information). Following the injection, an optical fiber (200 µm O.D., 0.39 NA, RWD Life Science Inc.) housed within a ceramic ferrule was implanted toward the SCN or SPZ. An EEG head mount (Pinnacle Technology) was then secured to the skull using four screws, and dental acrylic was applied to affix both the ferrule and head mount in place. Mice were given at least two weeks to recover post‐surgery before being connected to the optogenetic stimulation system (Thorlabs, OGKL2) and EEG recording setup (Pinnacle Technology). After a minimum of four days of habituation, EEG recordings were conducted while optogenetic stimulation (470 nm, 4–6 mW, 10 Hz, 20% duty cycle) was applied for 10 min between ZT2 and ZT6. Animals were recorded for at least 1 h after stimulation. The duration to “Return to baseline” was defined as the time required for sleep to fall within the baseline (pre‐stimulation period) ± SEM range determined by EEG, within a 5 min window. A top‐mounted camera (DMK 22AUC03, LORE+ Technology) recorded video at 15 Hz. Following this, recordings were conducted using the same protocol between ZT14 and ZT18 under IR light.

To track animal position during optogenetic experiments, we used a computer vision‐based approach leveraging the YOLOv11 object detection model as previously described.^[40^] The model was trained on a dataset of 160 manually annotated images, each containing a bounding box around the animal. These images were randomly sampled from video recordings of 8 different animals to enhance the model's generalizability. Training was conducted on two NVIDIA RTX 3090 GPUs for 200 epochs, with a batch size of 64 and an input resolution of 640 × 640 pixels. The final model achieved a mean average precision (mAP) of 0.94 across intersection over union (IoU) thresholds from 0.5 to 0.95 (mAP50–95). This trained model was then applied to behavioral video recordings to extract the animal's position, from which average speed was calculated in 1 min time bins.

Mice were deeply anesthetized and perfused with ice‐cold 1× PBS (1.1 mM KH₂PO₄, 155.2 mM NaCl, 3.0 mM Na₂HPO₄, pH 7.4), followed by 4% paraformaldehyde (PFA) in 1× PBS. Brains were dissected and post‐fixed overnight in 4% PFA at 4 °C. Coronal brain sections (40 µm) were cut using a vibratome (VT‐1200s, Leica), then washed twice in PBS containing 0.3% Triton X‐100 (PBS‐T). Sections were blocked in PBS‐T supplemented with 5% goat serum for 30 min at room temperature (RT), followed by incubation with primary antibodies diluted in the same blocking solution for 24 h at 4 °C. The following primary antibodies were used: rabbit anti‐PER2 (Millipore Sigma, AB2202, 1:500) and chicken anti‐RFP (Rockland, 600‐901‐379, 1:1000). After four washes in PBS‐T, sections were incubated overnight at 4 °C with fluorescent secondary antibodies diluted in PBS‐T with 5% goat serum. The following secondary antibodies were used: Alexa Fluor 647 goat anti‐rabbit (Invitrogen, A21244, 1:1000) for PER2 detection and Alexa Fluor 568 goat anti‐chicken (Invitrogen, A11041, 1:1000) for tdTomato detection. After four final washes in PBS‐T, sections were mounted with mounting medium with DAPI (Vector Labs, H‐1500‐10) and coverslips were added. Confocal imaging was performed using a Zeiss LSM700 microscope with 10× and 40× objectives.

PER2 immunoreactivity in the SCN was quantified using Fiji (ImageJ). Maximum intensity projections were generated from confocal z‐stacks of 40 µm slices, 3–5 slices per animal were used for the analysis. In animals with viral injections targeting the SCN, mWAKE+ cells were identified by co‐expression of tdTomato and EYFP. Cells lacking both tdTomato and EYFP signals were classified as mWAKE⁻.

RNAscope in situ hybridization was performed using the RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics, 323100) according to the manufacturer's instructions. Mice were anesthetized and perfused with ice‐cold 1× PBS followed by 4% PFA in 1× PBS. Brains were post‐fixed in 4% PFA for 24 h at 4 °C, then cryoprotected sequentially in 10%, 20%, and 30% sucrose solutions in PBS at 4 °C (1 day for 10% and 20%, 2 days for 30%) until they sank. Brains were embedded in Tissue‐Tek O.C.T. compound (VWR, 4583), frozen using dry ice, and stored at −80 °C. Coronal sections (13 µm) were collected using a cryostat, mounted on Superfrost Plus slides, and stored at −80 °C until use. RNAscope was then performed following the ACD protocol (320293). The following probe was used: mWake‐C1 (1057631‐C1). Signal amplification was achieved using Opal 690 dye (AKOYA Biosciences, NEL810001KT; 1:2000 dilution). Following RNAscope, immunostaining for tdTomato was performed on the same sections as described above. Sections were counterstained with DAPI (incubated for 30 seconds at RT), and mounted using ProLong Gold Antifade Mountant (Invitrogen, P36930↗) with coverslips. Confocal imaging was performed on a Zeiss LSM700 microscope using 40× objectives.

mWake RNAscope signal was quantified using Fiji. Maximum intensity projections were generated from 13 µm sections containing the SCN, 3 slices per animal were used for the analysis. The region of interest (ROI) within the SCN was manually defined based on tdTomato fluorescence in mice injected with AAV‐tdTomato or AAV‐Cre‐P2A‐tdTomato.

Adult male mice (3–5 months old) were individually housed in cages equipped with a vertical running wheel (ActiMetrics) and had continuous access to food and water. Following a one‐week acclimatization period under a 12:12 h LD cycle, their wheel‐running activity was recorded for one week in LD conditions. This was followed by another week in constant darkness (DD) to assess free‐running activity. Wheel‐running data was collected in 1 min intervals and analyzed using ClockLab software (ActiMetrics). Period length and rhythm strength estimates were derived from 7 consecutive days of DD recordings. Two mutant animals were excluded due to markedly reduced wheel running, precluding reliable circadian rhythm analysis.

Male mice aged 6 to 9 weeks were deeply anesthetized with isoflurane, after which their brains were rapidly extracted and dissected in an ice‐cold, oxygenated slicing solution (95% O₂, 5% CO₂) containing 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 2.5 mM CaCl₂, 248 mM sucrose, 26 mM NaHCO₃, and 10 mM glucose. Coronal brain slices (250 µm thick) were generated using a vibratome (VT‐1200s, Leica) and subsequently incubated in oxygenated artificial cerebrospinal fluid (ACSF: 124 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 2.5 mM CaCl₂, 26 mM NaHCO₃, and 10 mM glucose; 290–300 mOsm). The slices were maintained at 28 °C for 30 min before being transferred to room temperature for an additional hour. For electrophysiological recordings, the slices were placed in a recording chamber with continuous perfusion of oxygenated ACSF at room temperature. Cells of interest were identified using an upright microscope (BX51WI, Olympus) with infrared differential interference contrast (IR‐DIC) and native fluorescence imaging. Patch‐clamp recordings were performed using glass electrodes (5–8 MΩ) filled with an internal solution consisting of 130 mM K‐gluconate, 5 mM NaCl, 10 mM C₄H₈N₃O₅PNa₂, 1 mM MgCl₂, 0.2 mM EGTA, 10 mM HEPES, 2 mM MgATP, and 0.5 mM Na₂GTP (pH 7.2–7.3, 300 mOsm). Whole‐cell patch‐clamp recordings were acquired using a Multiclamp 700B amplifier (Molecular Devices), with data sampled at 20 kHz, low‐pass filtered at 2 kHz, and digitized using a Digidata 1440A (Molecular Devices).

Statistical analyses were conducted using Prism 8 (GraphPad). For comparisons between two groups that are normally distributed, unpaired Student's t‐tests, two‐tailed were used. For analyses involving multiple factors, two‐way ANOVAs were conducted for normally distributed data, with repeated measures applied when appropriate and post hoc Sidak tests for pairwise comparisons following ANOVA. Data are presented as mean ± SEM for all figures; sample size (n), p value and all statistical details are indicated in each figure legend and Table S1 (Supporting Information).