Anxiety control by astrocytes in the lateral habenula

The potential role of astrocytes in lateral habenula (LHb) in modulating anxiety was explored in this study. The habenula are a pair of small nuclei located above the thalamus, known for their involvement in punishment avoidance and anxiety. Herein, we observed an increase in theta-band oscillations of local field potentials (LFPs) in the LHb when mice were exposed to anxiety-inducing environments. Electrical stimulation of LHb at theta-band frequency promoted anxiety-like behavior. Calcium (Ca 2+ ) levels and pH in the cytosol of astrocytes and local brain blood volume changes were studied in mice expressing either a Ca 2+ or a pH sensor protein specifically in astrocytes and mScarlet fluorescent protein in the blood plasma using fiber photometry. An acidification response to anxiety was observed. Photoactivation of archaerhopsin-T (ArchT), an optogenetic tool that acts as an outward proton pump, results in intracellular alkalinization. Photostimulation of LHb in astrocyte-specific


Introduction
Anxiety is a primitive emotional state likely generated as a result of a rapid and unconscious assessment of the environment.It is defined as an uncontrollable, diffuse, unpleasant, and persistent state of negative affect, characterized by apprehensive anticipation regarding unpredictable and unavoidable future danger, and accompanied by physiological symptoms of tension and a constant state of heightened vigilance (Barlow, 2002).Anxiety is beneficial for survival as potentially harmful situations are avoided as a consequence of this emotion.However, when anxiety becomes persistent and excessive, even favorable adaptation to the ever-changing environment is suppressed.In humans, anxiety is invariably reflected in behavioral disturbances such as avoidance, non-verbal vocalization, and/or hypervigilance (Celia, 1988;Beck et al., 2005).Recently, several clinical studies have conceptualized human anxiety disorders as defensive disorders, the key feature of which is the inappropriate activation of defensive behaviors as a result of misappraisal of danger (Beck, 1976;Deakin and Graeff, 1991).Therefore, the tone of anxiety is adjusted for optimal behavior in healthy individuals.This study explored the potential role of astrocytes in the lateral habenula (LHb) in regulating anxiety levels in mice (Poskanzer and Yuste, 2016;Nagai et al., 2021;Zhang et al., 2021;Cho et al., 2022).
Animals can experience fear and anxiety when faced with aversive stimuli and threats.While it is difficult to distinguish between fear and anxiety, fear is generally elicited by actual, acute sensory inputs, whereas anxiety is induced by potential, indirect, and anticipated threats, and is associated with arousal and vigilance (Davis et al., 2010;Grupe and Nitschke, 2013;Tovote et al., 2015).Animal behaviors that appear to stem from anxiety have been observed across different species.In rodents, anxiety leads to behaviors such as avoidance of open and bright areas, as these areas have potential threats for harmful consequences.Therefore, the open field test and elevated plus maze test are often used to assess anxiety levels in rodents.
The habenula are a pair of small nuclei located above the thalamus.They are ancient brain structures found in almost all vertebrates.It is one of the few brain regions that controls both dopaminergic and serotonergic systems (Lecourtier and Kelly, 2007).Since these neuromodulators play essential roles in a wide range of motivational and cognitive functions, LHb neuronal circuits have been shown to be potentially relevant in Major Depression Disorder and anxiety (Metzger et al., 2021).
The LHb is known for its involvement in cognitive and emotional behaviors, such as punishment avoidance and anxiety (Hu et al., 2020).Previous studies have shown that increased neuronal activity in the mouse LHb is associated with aversive and anxious behaviors (Tovote et al., 2015;Calhoon and Tye, 2015;Baker et al., 2022).Recent pharmacological inactivation studies have shown that the LHb is involved in regulation of aversively motivated and anxiety-like behaviors (Velazquez-Hernandez and Sotres-Bayon, 2021;Lecourtier et al., 2023).Long-term LHb dysfunction reportedly results in behavioral disorders associated with response flexibility, such as those seen in depression, schizophrenia, and addiction (Hones and Mizumori, 2022).This suggests that the neuroplastic changes that occur within the human LHb play an important role in the pathogenesis of depressive disorders and mechanisms of antidepressant treatment (Loonen and Ivanova, 2019).Increasing evidence has shown that glial cells play a crucial role in maintaining brain microenvironment, regulating neurotransmitter levels, and influencing synaptic plasticity (Onodera et al., 2021;Fang et al., 2022).Astrocytes in the LHb produce a variety of cytokines in response to pro-inflammatory stimuli that affect glutamatergic neurotransmission.Such neurotoxic changes also occur in psychiatric disorders, such as depression (Loonen, 2023).
To understand the mechanisms underlying anxiety-like behavior, electrophysiological and fiber photometry recordings were made from the LHb in freely behaving mice.We observed an increase in theta-band (5 -10 Hz) oscillations in local field potentials (LFPs) recorded from electrodes placed in the LHb when mice were exposed to a highly anxious environment.A previous study has shown that electrical stimulation in the theta-band range (5 Hz) in the LHb produces depressivelike behavior in rats (Jakobs et al., 2019).We hypothesized that astrocytic actions may be responsible for switching and maintaining the mode of neuronal activity related to anxiety.Astrocytic calcium (Ca 2+ ), pH, and local brain blood volume (BBV) changes were analyzed in freely moving mice using fiber photometry method (Qin et al., 2020;Ikoma et al., 2023aIkoma et al., , 2023b;;Asano et al., 2023).Upon exposure to an anxious environment, an acidic shift was observed in LHb astrocytes.The effect of photoactivation of an optogenetic tool, archaerhodopsin-T (ArchT), which is specifically expressed in astrocytes, was also assessed.Since ArchT is an outward proton pump, its photoactivation results in intracellular alkalinization (Beppu et al., 2014).ArchT photoactivation resulted in dissipation of theta-band LFP in an anxious environment and a slight alleviation from anxiety, as assessed by observing animal behavior.We hypothesized that changes in the local brain environment in the LHb would cause a shift in the mode of neuronal activity.Decreased brain pH is a common endophenotype of psychiatric disorders (Hagihara et al., 2018(Hagihara et al., , 2023)).It is possible that the dysregulation of astrocytic pH in the LHb is related to mental illnesses, such as anxiety disorders (Griebel and Holmes, 2013).

Animals
All animal experimental procedures were approved by the Animal Research Committee of the Tohoku University.8 to 16-week-old male mice with the C57BL/6J background were used in this study.Mice were housed and maintained at a controlled temperature of 23-26 • C and 40%− 60% humidity on a 12 h light/dark cycle with food and water provided ad libitum.All behavioral procedures were performed during the light phase between 8 A.M. and 8 P.M.Details of the transgenic mice demonstrating astrocyte-specific expression of a Ca 2+ sensor (Mlc1-tTA:: tetO-YC nano50 ), a pH sensor (Mlc1-tTA::tetO-E 2 GFP), or an optogenetic tool (Mlc1-tTA::tetO-ArchT) used in this study have been described previously (Tanaka et al., 2012;Beppu et al., 2014;Kanemaru et al., 2014;Onodera et al., 2021).Briefly, Mlc1-tTA mice were crossed with tetO mice, and the bigenic mice expressed YC nano50 , E 2 GFP, or ArchT in astrocytes.A total of 14, 4, 4, and 26 C57BL/6J wild-type mice, Mlc1-tTA::tetO-YC nano50 transgenic mice, Mlc1-tTA::tetO-E 2 GFP transgenic mice, and Mlc1-tTA::tetO-ArchT transgenic mice were used, respectively.The numbers of mice used in each experiment are detailed in the figure legends.
To assess local brain blood volume changes using in vivo fiber photometry, blood vessels were fluorescently visualized by expressing a fluorescent protein in the liver, mScarlet, fused to the C-terminus of albumin (Alb) using adeno-associated virus (AAV) serotype 8 [AAV8; detailed method is shown in (Wang et al., 2022;Ozawa et al., 2023)].The expressed Alb-mScarlet was secreted and included in the blood plasma.AAV8/P3-Alb-mScarlet was produced by the ultracentrifugation method (Konno and Hirai, 2020) using the Viral Vector Core at Gunma University.The virus vector was dissolved in phosphate-buffered saline (PBS) at a concentration of 2 × 10 12 vg/ml.The virus-containing solution was systemically administered via intraperitoneal injections (i.p.) for 270 µl per mouse.Surgical implantation of optical fibers and electrodes was performed first (see below), and AAV was injected after the completion of surgery.After waiting for more than three weeks for stable Alb-mScarlet fluorescence in the blood vessels, fiber photometry experiments were performed.A more conventional method is to fluoresce blood vessels by injecting Texas Red dextran into the tail vein; however, in such cases, the fluorescence rapidly decays with time.To understand BBV changes associated with behavior, stable expression of the fluorophore in the blood plasma is needed; thus, the Alb-mScarlet approach was used.

Surgery
Surgery was performed under anesthesia using a mixture of 0.75 mg/ kg medetomidine hydrochloride (Domitor; Nippon Zenyaku Kogyo Co., Ltd., Fukushima, Japan), 4 mg/kg midazolam (Midazolam, Sandoz Inc., Japan), and 5 mg/kg butorphanol tartrate (Vetorphale; Meiji Seika Pharma Co., Ltd., Tokyo, Japan).Atipamezole hydrochloride (0.75 mg/ kg; Antisedan; Zenyaku Kogyo Co., Ltd., Tokyo, Japan) was administered for postoperative medetomidine reversal.For surgery, the anesthetized mice were placed in a stereotaxic frame.The Paxinos and Franklin (2001) mouse brain atlas was used as the basis for coordinates.Behavioral assays or recording experiments were performed after a postoperative recovery period of at least one week.After surgery, the mice were kept isolated in a home cage with 17 × 10.5 cm or 18 × 18 cm floor space and 23.2 cm height.

In vivo electrophysiology
For in vivo electrophysiology recording experiments, a hand-made optrode consisting of a 225 µm clad diameter optical fiber (core diameter 200 µm, 0.39 NA, FT200UMT; Thorlabs, USA) and two 0.1 mm diameter epoxy-coated tungsten wire electrodes (UNIQUE MEDICAL, epoxy coating removed for ~0.3 mm from the cut end) was implanted into the left LHb (AP, − 1.72 mm; ML, − 0.46 mm; DV, − 2.35 mm from the brain surface) (Fig. 1A).The optical fiber served as a support for the electrodes and maintained the two electrodes at a constant distance.Even in experiments in which no optical signal was used, an optical fiber was placed, which served as a spacer to create a short constant distance between the electrodes.The tip of the electrodes protruded below the tip of the optical fiber by 0.3 mm.The optrode was placed in such a manner that only the tips of the electrodes entered the LHb to minimize damage to the LHb.The optical fiber and electrode assembly were fixed to the skull using dental cement.
LFPs were recorded from these electrodes.These LFPs reflected neuronal activity mainly within the LHb; however, a contribution from W. Tan et al. (caption on next page) W. Tan et al. neuronal activity in the hippocampus and in the vicinity of the recorded electrical signals was also expected (Aizawa et al., 2013).Two cranial holes above the cerebral cortex and the cerebellum were drilled before the stainless-steel screw (M1) implantation.These holes were slightly less than 1 mm in diameter.The 1 mm diameter screw was screwed into the cranial hole using a fine Phillips head screwdriver until the screw's far end just touched the brain surface.The screw in the cerebral cortex was used to record intracranial electroencephalography (EEG) (or electrocorticogram; EcoG), and the other screw electrode in the cerebellum served as the ground (Fig. 1A).The plain fit between the hole and the screw allowed physical grasp that could hold the screw in place.However, to ensure the fixation of the screw to the skull over days of experiments, dental cement was applied on top of both the screw and the skull to adhere these two components stably.Lead wires from the four electrodes (two tungsten electrodes in the LHb, one screw electrode in the cortex, and one screw electrode in the cerebellum) were individually soldered to a 4-pin (male) connector.The 4-pin connector was held in place slightly above the skull with an instant adhesive (Loctite 454J; Henkel).The gap between the lead wires and the connector was also filled with the adhesive.Finally, to firmly hold all implantations, lead wires, and the connector in place, the whole structure, except for the 4-pin heads, was covered with dental cement and adhered to the skull.The resultant implant can be seen in Fig. 2A, connected to the recording cable.
A 60 cm multicore cable was prepared with a 4-socket (female) connector at one end.Immediately prior to the behavioral experiments, this 4-socket connector was plugged in to the matching 4-pin connector, which had been surgically attached to the animal's skull, as described in the previous paragraph.The ends of the four wires at the other end of the multicore cable were individually soldered to single-pin connectors.The two wires from the two LHb electrodes were connected to inputs A and B of a DAM50 extracellular amplifier (World Precision Instruments, USA) for differential bipolar recording.The wire from the screw electrode in the cerebellum was connected to amplifier ground.The voltage difference between the two LHb electrodes was amplified and recorded as the local field potential (LFP) of the LHb.The difference between the voltage signal recorded from the screw electrodes in the cerebral cortex and cerebellar cortex (i.e.unipolar recording reference to ground) was amplified and recorded as the cortical EEG.Voltage fluctuations in the cerebellum will also have been included in the recorded signal.LFP and EEG signals were amplified and initially bandpass filtered between 1 Hz and 1 kHz with two DAM50 amplifiers; and then digitized using an A/D converter (TUSB-1612ADSM-S2Z, Turtle Industry, Japan) at a sampling rate of 2 kHz.The acquisition was controlled with the recording software LaBDAQ5 (Matsuyama Advance, Japan).The same pair of tungsten LHb electrodes could also be used for electrical stimulation.The pulse sequence of the electrical train was generated using a pulse generator (AWG-50, ELMOS Co., Ltd., Japan).A constant current isolated stimulator (DS3, Digitimer, UK) was used to send an electrical stimulus to the LHb.
All experiments were performed in an animal cage placed within a standard steel stationary cabinet with an off-white surface.The steel structure served as a Faraday cage and was effective in cutting out the electrical hum noise from the surroundings.Air was circulated using a direct-current (DC)-powered fan, the LED lighting was powered by DC, and the amplifiers and stimulus isolator used (see above) were batterypowered.The two DAM50 amplifiers and the stimulus isolator were placed inside the cabinet.The chassis ground of the amplifiers was connected to the Faraday cage, and the Faraday cage was connected to the isolated ground provided by the facility.The fiber photometry apparatus was also placed inside the cabinet, but the fluorescence excitation light source was placed outside.The optical fibers, BNC cables, and USB cables for the web camera were connected to the equipment inside the cabinet through a hole drilled into the side or roof of the cabinet.All alternating current (AC)-operated equipment, including a laptop with a battery charged with AC, was placed outside the cabinet.Since the optical and electrical wires tethered to the mouse were loosely hung from a hole drilled through the shelf above, the mouse could move relatively freely inside the cage.After the initial setup, the animals were kept inside the cabinet with its door closed; thus, the influence of the experimenter and outside environment was minimal.
In all experiments, animal behavior was monitored and recorded using a web camera at 30.28 fps.Timing of stimulation was marked in one of the multiple channels recorded using the A/D converter, and an electrical pulse was sent at an appropriate time to drive an infrared (IR) LED that was captured within the field of view of the web camera.This allowed synchronization of the recorded electrophysiological data and videos.Similar IR-LED markings of the video for synchronization of multimodal data were performed in all experiments.
Following the recordings, mice were transcardially perfused with 4% paraformaldehyde in PBS.The brains were post-fixed overnight and cut into 50 µm-thick coronal sections using a vibratome (Vibratome VT1000S Leica Microsystems, RRID: SCR_016495, Nussloch, Germany).The placement of the optical fiber and electrodes was confirmed by examining these sections.For the verification of electrode positions for in vivo electrophysiology recording experiments, see Fig. S1.These sections were attached to slide glasses (PLATNUM PRO PRO-03 micro slide glass, MATSUNAMI Glass IND., LTD, Osaka, Japan) and sealed with coverslip glass (NEO cover glass, MATSUNAMI Glass IND., Osaka, Japan) preserved for later confirmation.

Fiber photometry
The optrode assembly used for electrophysiological recordings and stimulation was the same as that used for fiber photometry recordings.The optrode was implanted in the left LHb.The center 200 µm diameter Fig. 1.Theta-band increase in LHb upon exposure to anxious environment of the elevated alley (EA).(A) The location of optrode implantation targeted towards the LHb is shown.Two tungsten wire electrodes were attached to the center optical fiber.The tips of the electrodes were inserted into the LHb.Two screw electrodes, one on the cortex and the other on the cerebellum were used for recording EEG.(B) The EA (right) was used as an anxious environment for the mice.The electrical wire and optic fiber were connected to the mouse but the tethered mouse could relatively move freely in the apparatus.Pink dental cement was used for fixing the electrodes and the connector to the skull.A black heat shrink wrap was used to protect the soldered connection between the wire and the connector.Preamplification adjacent to the animal's head was not necessary as the steel cabinet enclosure served as a Faraday cage for adequate noise reduction.(C) Upon being placed on the EA, 5 -10 Hz theta-band activity increased in the LFP recorded using the electrodes in the LHb (bottom).Similar change was not observed in the EEG recorded from the cortex (top).(D) Traces show 2 s of LFP signals in the LHb before, during, and after exposure to the EA.In the LFP trace while the mouse was on the EA, a clean oscillation with mostly single frequency was apparent even in the raw trace (middle).LFPs were low-pass filtered at 20 Hz. (E) Power spectrum of the LFPs in the LHb before (thin gray trace) and during (thick black trace, upper panel), and during (thick black trace) and after (thin gray trace, lower panel) exposure to the EA.The time window of each power spectrum is 10 min.A clear increase in the 5 -10 Hz theta-band power during EA exposure was observed.(F) As shown in Fig. S2, the raw power spectrum was log 10 transformed and the ratio of the average log 10 power of theta (5 -10 Hz) over the delta (1 -4 Hz) band of the LFP in the LHb was calculated for periods before, during, and after exposure to the EA.Data points from individual animals (n = 6) and mean ± SEM are presented as bar graphs and error bars, respectively.One-way ANOVA test revealed that there was a statistically significant difference in theta-delta ratio between the three groups (F(2, 15) = [24.12],P = 0.0004).Bonferroni's multiple comparisons found that the mean value of theta-delta ratio was significantly different between Before and EA (P = 0.0063, 95% C.I. = [− 0.05306, − 0.01300]); EA and After (P = 0.0046, 95% C.I. = [0.02236,0.08501]).There was no statistically significant difference in mean value of theta-delta ratio between Before and After (P = 0.2335).* * P < 0. 01, NS, not significant.C57BL/6J wild-type mice were used in these experiments.
(caption on next page) W. Tan et al. core optical fiber was inserted into a 1.25 mm diameter ceramic ferrule (CFLC230-10, Thorlabs, USA) and the implanted optical fiber was connected to another optical fiber with direct ferrule to ferrule connection being established via a split sleeve.The connected optical fiber led to the fiber photometry apparatus.The fiber photometry apparatus was custom-built using an optical block series (Hamamatsu Photonics, Shizuoka, Japan) mounted with two photomultiplier tubes (PMTs; H10722-210, Hamamatsu Photonics).
The basic strategy for recording and analyzing fluorescence signals was the same as that used in a previous study (Ikoma et al., 2023a).In the current study, a maximum of four fluorescence traces were recorded in the astrocyte YC nano50 and Alb-mScarlet expression experiments (quad recordings).Three excitation wavelengths were used (M415F3, Thorlabs light source with ET430/24x Chroma filter; M505F3 Thorlabs light source with ET505/20x Chroma filter; and M595F2 Thorlabs light source with ET580/25x filter) and fluorescence emission was detected using two PMTs (PMT1, with a custom-made Chroma filter that passed CFP and mScarlet fluorescence emission wavelengths; PMT2, with ET540/30 m Chroma filter that passed YFP fluorescence emission wavelengths).The excitation wavelengths were combined using T470lpxr and T550lpxr (Chroma) dichroic mirrors.The excitation light was sent to the animal using a multiband wavelength mirror (69008bs, Chroma), which allowed the fluorescent light to return from the animal and pass through.Fluorescence emissions were divided using a multiband wavelength mirror, ZT532dcrb-UF1 (Chroma).For the E 2 GFP expression experiments, Prizmatix 450 light source with ET473/24 m Chroma filter was used for excitation, fluorescence was divided using ZT502rdc (Chroma) filter, and the emission was detected using two PMTs with ET540/40 m and ET510/20 m Chroma filters.The timing of light delivery was controlled using pulse generators (AWG-50 and/or Master-8, A.M.P.I., Israel).

Optogenetics
In contrast to the fiber photometry recordings, optogenetic illumination was delivered bilaterally to both LHb structures.Two optical fibers were bilaterally implanted in the LHb at an angle of 10 • (AP, − 1.72 mm; ML, ± 0.83 mm; DV, − 2.4 mm from the brain surface).For photoactivation of ArchT, green (505 nm; M505F3, Thorlabs, USA) light was delivered (power intensity at the output tip, ~1.8 mW).In the 2-Way Bright-Dark apparatus experiments (see Section 2.6.4.), ArchT expressed in astrocytes was stimulated using 505 nm light for 10 s while the mouse was still in the home cage.Subsequently, the mouse was intermittently stimulated with 2 s on and 3 s off during the time it stayed in the 2-Way Bright-Dark apparatus for 20 min.The appropriate positioning of the optical fibers was confirmed using paraformaldehydeperfused tissue at the end of all experiments.

Experimental paradigms 2.6.1. Elevated alley (EA)
An EA was used to induce anxiety in mice.Mice have an innate fear of heights.However, the threat created by EA is indirect, as the mice will not immediately fall over and hurt themselves.Therefore, the emotions created by the EA can be categorized as anxiety.The width of the alley was 5.5 cm, length was 55.5 cm straight, and height was 50 cm from the floor.At the center of the alley, a small wing protruded 6.5 cm on both sides (Fig. 1B).The EA was uniformly illuminated with 100 lux light.Unlike the elevated plus maze, elevated T-maze (Graeff et al., 1998) or light-dark box, the EA apparatus provided a uniform anxiogenic environment with no anxiolytic compartments (e.g., closed arm or dark box) that the mice could resort to.EA is a non-standard single-compartment anxiogenic environment that was newly developed in this study.As with all other apparatuses for animal behavior experiments, the EA was placed in the standard steel stationary cabinet.

All marble cage (AMC)
The marble-burying test is often used to assess anxiety-like behavior in rodents (de Brouwer et al., 2019).Due to their novelty, marbles presumably induce innate fear in mice.As marbles themselves do not cause harm to the mice, the emotions created by their presence can be categorized as anxiety.Burying marbles removes the aversive stimuli from view; thus, marble-burying behavior is considered an appropriate response to relieve anxiety created by aversive stimuli (Gray, 1982;Njung'e and Handley, 1991).Therefore, marble-burying tests are often used to assess anxiolytic drug action.
We noticed that the mice avoided the cage with marbles; thus, we assumed that the mice became anxious in the presence of marbles.Therefore, we filled the entire floor of the cage with marbles (All Marble Cage; AMC) so that the mice would not be able to escape the marble threat, creating a highly anxiogenic environment (Fig. 2A).The AMC is also a non-standard, single-compartment, anxiogenic environment that was newly developed in this study.Smooth glass marbles of various colors and patterns measuring 1.6 cm in diameter were used to fill the entire floor of the 18 cm × 18 cm cage.The surrounding walls were transparent and measured 23.2 cm in height.The cage was illuminated with a maximum luminescence of 100 lux.

Open field test
An open field is also considered an anxiogenic environment as mice are exposed to a novel open space that under normal ecological circumstances would signal potential threat from a predator.The open field test is a commonly used standard test to assess anxiety levels in mice.The mice usually avoid the center of the open field and mostly stay at the edges.Since thigmotaxis (i.e., to move along the edge of an open space) is an index of anxiety in mice, the open field could be divided into more (center) or less (edge) anxiety-creating zones, respectively (Simon et al., 1994;Bitran et al., 1998;Walz et al., 2016).In our study, we assumed Fig. 2. Theta-band increase in LHb upon exposure to anxious environment of the all marble cage (AMC).(A) AMC apparatus was used as an anxious environment for the mice.(B) Upon being placed in AMC, 5 -10 Hz theta-band activity in the LFP was recorded using the electrodes in the LHb (bottom).Similar changes were not observed in the EEG recorded from the cortex (top).(C) Representative traces of LFP in the LHb before, during, and after exposure to the AMC.Theta-band oscillation was apparent in the raw LFP trace recorded during AMC exposure.LFPs were low-pass filtered at 20 Hz. (D) Power spectrum of the LFPs in the LHb before and during (upper panel), and during and after (lower panel) exposure to the AMC.The time window of each power spectrum is 10 min.5 -10 Hz theta-band increased in the AMC.(E) The power of theta (5 -10 Hz) and delta (1 -4 Hz) bands were calculated within a time window of 1 s and divided to calculate the theta-delta ratio.This time window was shifted with 0.05-s steps to calculate the transition of theta-delta ratio over time.In the particular sample shown above, exposure to AMC created an immediate increase in theta-delta ratio, which gradually relaxed during the exposure (top trace).Similar tendencies were observed across multiple episodes collected from six animals, and the average theta-delta ratio transition trace was calculated and shown below (average).Both traces were low-pass filtered at 5 mHz.(F) Average theta-delta ratios of the LFPs in the LHb were calculated for periods before, during, and after exposure to the AMC.Data points from individual animals (n = 6) and mean ± SEM are presented as bar graphs and error bars, respectively.One-way ANOVA test revealed that there was a statistically significant difference in theta-delta ratio between the three groups (F(2, 15) = [11.87],P = 0.0034).Bonferroni's multiple comparisons found that the mean value of theta-delta ratio was significantly different between Before and AMC (P = 0.0264, 95% C.I. = [− 0.1220, − 0.009966]); AMC and After (P = 0.0418, 95% C.I. = [0.002938,0.1241]).There was no statistically significant difference in mean value of theta-delta ratio between Before and After (P > 0.9999).* P < 0. 05, NS, not significant.C57BL/6J wild-type mice were used in these experiments.
W. Tan et al. that the mice would be more anxious when crossing the center of the open field than when it was close to the edges.Therefore, the characteristics of the LFP were compared when the mouse was in the center versus in the surrounding area.An open field cage with a floor space of 36.5 × 21.5 cm and a smooth surface floor with no bedding was used.The surrounding transparent wall (height, 26 cm) was back-covered with black paper.The top was fully open.The cage was illuminated with a maximum luminescence of 100 lux.

Place aversion test in the 2-Way Bright-Dark apparatus by electrical stimulation of LHb
The 2-Way Bright-Dark apparatus consisted of one bright and one dark cage.The bright cage (16.8 × 16.8 × 17.5 cm) had a floor filled with the same glass marbles used in AMC and the surrounding transparent walls were back-lined with aluminum foil which reflected light.The floor of the other cage (16.8 × 16.8 × 17.5 cm) was filled with soft white bedding and the walls were back-lined with black paper (dark cage).Illumination was set to 100 lux in both cages.The two cages were connected by an opening large enough to allow the mouse wired with electrodes and optical fibers to travel easily between the cages.
This 2-Way Bright-Dark apparatus can be considered fundamentally close to the standard light-dark box used in experiments to evaluate anxiety.The light-dark box consists of an open light compartment and a wholly closed dark compartment, often used to test the effects of antipsychotics (Bourin and Hascoët, 2003).However, the optical fiber and electrical cable connecting the mouse to the recording equipment would prevent the mouse from moving freely in the light-dark box.Therefore, the light-dark box is unsuitable for experiments with the tethered mouse.We used the 2-Way Bright-Dark apparatus and made the two cages different in terms of the level of anxiety that would be induced in mice by using different bedding and wallpaper.Since there is no ceiling on the two cages, the mice can move freely in the 2-Way Bright-Dark apparatus while being tethered with an optical fiber and an electrical cable.The 2-Way Bright-Dark apparatus is also conceptionally similar to the elevated plus maze (EPM), which is commonly used to assess anxiety.In our initial trials with the EPM, we noticed that the narrow corridors and high walls surrounding the closed arm prevented smooth travel of the wired mouse.The bright cage of the 2-Way Bright-Dark apparatus was also similar to that of the AMC; thus, it was similar to the elevated open-arm environment in the EPM, which would create anxiety.The dark cage provided an anxiolytic environment to which the mice could resort to, which was similar to the closed-arm environment in the EPM.
The mouse was first placed in the bright cage, and each time the mouse entered the dark cage, an 8 Hz electrical stimulation (0.2 ms duration, 125 ms interval, 250 µA intensity) was applied between the electrodes placed in the LHb.Commonly, the 5 -10 Hz frequency band is referred to as the theta-band.In this study, we show that theta-band frequency is enhanced in the LFP recorded from the LHb when the mouse is placed in an anxiogenic environment (Figs. 1 and 2).Therefore, 8 Hz stimulation frequency was chosen to evaluate whether such stimulus can evoke anxiety-like behaviors.This 8 Hz stimulus was repeated throughout the 20 min that the mouse was placed in the 2-Way Bright-Dark apparatus.Animal behavior was recorded using a web camera from above and animal location was extracted with a protocol created using FIJI software (distributed under GPL).

Place preference in the 2-Way Bright-Dark apparatus with astrocytic ArchT optogenetic stimulation
The same 2-Way Bright-Dark apparatus with bright and dark cages as described above was used, with illumination set to 100 lux in the dark cage and 2500 -3500 lux in the bright cage.For unknown reasons, basal anxiety levels appeared to be lower in Mlc1-tTA::tetO-ArchT transgenic mice, and they did not mind staying in bright cages.Therefore, the illumination was set at a higher intensity, and the residence time in the bright cage decreased.We used these conditions for conducting experiments on astrocytic ArchT-expressing transgenic mice.In the ArchT photostimulation sessions, the optical fiber implanted above the LHb was illuminated with a 2 s on and 3 s off cycle for the entire 20 min during which the mouse was placed in the 2-Way Bright-Dark apparatus.A previous study from our lab has shown that 10 s of ArchT photostimulation resulted in intracellular alkalinization and the pH would be maintained alkalized, even after the cessation of the light stimulation (Beppu et al., 2014).The pH returned gradually to the baseline level with a half-decay time of about 40 s.In a recent publication, our group has also successfully inhibited online cerebellar motor learning in mice by using 2 s on 3 s off cycle of light activation of ArchT expressed in cerebellar glial cells (Kanaya et al., 2023).Therefore, we assumed that the pH in the LHb astrocytes would remain alkaline with this protocol of light stimulation throughout the 20 min experimental time.Mouse travel in the 2-Way Bright-Dark apparatus was tracked, and travel in each cage was quantified and analyzed.

Data analysis
Behavioral analysis of the recorded video was performed using FIJI software.Animals with incorrect fiber and electrode placement were excluded from the analysis.Raw traces of LFP and EEG were filtered between 1 Hz and 1 kHz and sampled at 2 kHz.Fluorescence data from PMT outputs along with the LFP and EEG data were analyzed offline using AxoGraph (AxoGraph Scientific, Australia).Spectrum analysis of LFP and EEG was performed using a custom code written in Python programming language.The fast Fourier transform (FFT) function equipped in the NumPy library was used.Hanning function to a window of 2000 samples was applied and the window was shifted with 100 sample steps.The short-term Fourier transform (STFT) was applied to the trace and the power spectrogram was created.
Statistical analyses were performed using GraphPad Prism10 (GraphPad Software, Boston, USA).Data were presented as means ± standard error of the mean (SEM) unless otherwise specified.For statistical comparison of the power of the LFP at each frequency range, log 10 transform of the power was first calculated before collating the data (see Fig. S2 for the calculation method).For three-group comparisons, one-way ANOVA followed by Bonferroni's multiple comparisons test was used (Figs.1F and 2F).For two-group comparisons, nonparametric tests (Mann-Whitney U-test) were used and the data were presented as median and range (Figs.4D and 8D, E).

Theta-band activity in the LHb in anxiogenic environments
Since the LHb is a subregion of the brain known to code negative emotions, reactions of the LHb to anxiogenic environments were examined in this study.In previous studies, it has been shown that placing rats on an elevated platform can cause acute stress (Xu et al., 1998;Rocher et al., 2004).In our study, we first used the EA apparatus to create anxiety.Upon placing the mouse on the EA, a prominent change in the LFP frequency spectrum was observed in the signals recorded using the LHb electrodes (Fig. 1C).The signal characteristics were directly observable in the raw LFP traces, which often had a clean near-single periodic cycle of sine waves (Fig. 1D) when the mice were placed on the EA.Compared to before the EA, the 5 -10 Hz theta-band increased on the EA, which then decreased after the mouse was removed from the EA apparatus (Fig. 1E).A change in the 1 -4 Hz delta-band was not apparent in the sample recording shown in Fig. 1C.To normalize the error distribution, the power of the spectrum was log 10 transformed.To control for general signal level of each recording, the theta-to-delta ratio was calculated across six animals and is summarized in Fig. 1F.A significant increase in the theta-delta ratio was observed on the EA.
AMC tests were performed for more stable creation of high anxiety.These also revealed characteristic changes in the LFP recorded using the LHb electrodes along with a clear emergence of theta-band activity (Fig. 2B-D).There was a small decrease in the delta-band activity.The transition of the theta-delta ratio was plotted against time (Fig. 2E).An immediate increase in theta-delta ratio was observed upon placement in the AMC.The ratio decayed slowly during the remainder of the stay in AMC.To normalize the error distribution, the power of the spectrum was log 10 transformed.The theta-to-delta ratio for six animals was calculated and is summarized in Fig. 2F.A significant increase in the thetadelta ratio was observed in the AMC.
In both EA and AMC experiments, the mouse was placed in an allanxiogenic environment by the experimenter.We next tested whether the theta-band LFP activity in LHb can spontaneously switch on in mice upon choosing to enter an anxiogenic zone.For this purpose, we used the plain open field test (Fig. 3A).Mice tended to avoid the center of an open area, stayed mostly at the periphery, and crossed the central area infrequently.The open field was arbitrarily divided into center and surroundings (edge), and the time segment that the mouse spent in the center was marked.Upon manually placing the mouse in the open field test cage, the theta-delta ratio instantaneously elevated, and then slowly decayed during the 10 min of the test period (Fig. 3B).However, upon closer examination of the LFP traces with higher time resolution, the emergence of theta-band frequency upon entering the center zone became apparent (Fig. 3C).LFP traces were aligned when the mouse entered the central area, and the power spectra before and after entering the center were calculated and averaged across 21 episodes of center entrance (Fig. 3D).A theta-band increase and delta-band decrease were clearly detected.The average theta-delta ratio increase was summarized for six animals, all of which showed a prominent increase upon entering the center (Fig. 3E).
It is possible that the increase in the theta-band activity could simply be generated by more movement in the center zone.To evaluate this possibility, the speed of the mouse immediately before and after entering the center zone was calculated.In many cases, the mouse slowed down upon center entry, and in other cases, the mouse sped up.In either case, the theta-delta ratio increase was statistically significant (Fig. S3).Therefore, the theta-delta ratio increase is not dependent on movement changes but rather is highly correlated with center zone entry.

Theta-band electrical stimulation of LHb resulted in anxiety-like behavior
Next, we tested the effect of artificial creation of neuronal activity in the LHb by 8 Hz stimulation through the electrodes.It was assumed that such stimulation generates theta-band (5 -10 Hz) neuronal activity in the LHb; however, this could not be confirmed because of the large stimulus artifact created by the stimulation electrode.Therefore, the effects created could be due to the increased neuronal activity in general.The 2-Way Bright-Dark apparatus consisted of a bright cage filled with glass marbles and a dark cage with comfortable bedding (Fig. 4 A, left).The mouse was first placed in the bright cage and, as expected, it moved to the more comfortable dark cage and mostly stayed there for the remainder of the experimental time (Fig. 4 A Control).Subsequently, we performed a protocol similar to that used in the passive avoidance test.Every time the mouse entered the dark cage from the bright cage, an 8 Hz electrical stimulation (E-stim) was delivered to the LHb electrodes (Fig. 4B, left, C).Due to this stimulus, the mouse tended to stay in the bright cage, and the total residence time in the bright cage was significantly longer with E-stim compared to control (Fig. 4 B E-stim, D).This suggests that the 8 Hz LHb E-stim created an aversion to dark cage and that anxiety-like behavior could be artificially evoked.

Development of quad fluorescence recording for estimating the local brain environment fluctuations
To measure the anxiety-associated local brain environmental fluctuations, fiber photometry was performed in mice expressing a fluorescence resonance energy transfer (FRET) type Ca 2+ sensor in astrocytes (Mlc1-tTA::tetO-YC nano50 ; Fig. 5A).AAV, which can produce Alb-mScarlet in blood plasma, was injected approximately four weeks prior to the experiment, which allowed direct visualization of the local BBV.An optical fiber was implanted in the vicinity of the LHb.Since the target was a small area, we used a relatively smaller diameter fiber (200 µm).Triple-excitation light was sent sequentially to the optical fiber to excite the CFP, YFP, and mScarlet (Fig. 5B).A non-conventional dichroic mirror, which only reflects yellow fluorescence, was placed between the two PMTs, and a custom-made dual-band filter was placed in front of PMT1, which allowed the detection of fluorescence from both CFP and mScarlet.PMT2 only detected fluorescence from YFP.This configuration allowed the detection of three fluorescence wavelengths using only two PMTs.Excitation of CFP (at 415 nm) led to the emission of CFP and YFP fluorescence (by FRET), which were detected using PMT1 (fCFP) and PMT2 (fYFP), respectively.With an increase in the cytosolic Ca 2+ , a decrease in fCFP and an increase in fYFP are expected.Direct excitation of YFP (at 505 nm) led to the emission of YFP, which was detected using PMT2 (dYFP).Changes in the Ca 2+ should not affect the dYFP.Excitation of mScarlet (at 595 nm) led to the emission of mScarlet, which was detected using PMT1 (mScarlet).
An example of fluorescence fluctuation in response to electrical stimulation (E-stim) of the LHb is shown in Fig. 5C.In this case, a relatively simple mirrored response of fYFP increase and fCFP decrease was observed, with no relative change in dYFP.This was expected for the Ca 2+ -increased response of the FRET sensor.We have previously shown that dYFP can be heavily influenced by changes in the cytosolic pH and local BBV fluctuations (Ikoma et al., 2023a).A small decrease in dYFP may be due to a small acidic shift in the cytosol.There was also little change observed in mScarlet, suggesting that BBV was not altered by E-stim.Since fYFP would also be affected by a shift in pH (much more so than the effect of pH on fCFP), the difference between fYFP and dYFP would likely provide a more reliable estimate of changes in cytosolic Ca 2+ in comparison to the more conventional ratiometry between fYFP and fCFP.The fluorescence fluctuations in response to typical rapid eye movement (REM) sleep were also examined (Fig. 5D).Typical theta-band increases in EEG were observed during REM sleep with accompanying increases in mScarlet fluorescence, which indicated a BBV increase associated with REM sleep.This increase in mScarlet fluorescence was accompanied by a YC nano50 fluorescence decrease.It has been reported that fluorescence in the brain parenchyma is partially obstructed by blood vessels; thus, an increase in the blood vessel diameter and BBV would result in a decrease in the detected fluorescence from the parenchyma (Ikoma et al., 2023b).Since dYFP only depends on cytosolic pH and BBV, this trace was inverted, overlaid, and scaled using mScarlet.The two traces almost completely matched.Since the changes in the two YFP signals (fYFP and dYFP) were both larger than the fCFP changes, it was also suggested that cytosolic pH acidification accompanied the increase in BBV during REM sleep.These data show that our quad recordings of fluorescence (fYFP, dYFP, fCFP, and mScarlet) allowed the estimation of several key components of local brain environment fluctuations.

Fig. 4. 8 Hz electrical stimulation of LHb induces anxiety-like behavior. (A)
The 2-Way Bright-Dark apparatus consisted of two cages, one covered with soft bedding surrounded with black walls (dark cage) and the other filled with only glass marbles surrounded with reflecting aluminum foil-covered walls (bright cage).The mouse could freely move between the two cages.The mouse movements were tracked using a web camera fixed above.In the control condition, without electrical stimulation, the mouse placed first in the bright cage quickly moved to the dark cage and mostly stayed there during the remainder of the recording.Shown on the right is the heat map from a representative mouse showing that it preferred the dark cage.(B) Every time the mouse entered the dark cage, an electrical stimulus of frequency 8 Hz was delivered to the electrodes inserted in the LHb.The heat map of such a mouse shows that the mouse avoided the dark cage and spent more time in the bright cage when stimulated in such a manner.(C) Schematic of the timing of 8 Hz electrical stimulation is shown.(D) The relative residential time in the bright cage during the 20 min of 2-Way Bright-Dark apparatus test is summarized.Compared to the control (n = 6), 8 Hz electrical stimulated mice spent more time in the bright cage (n = 5).A non-parametric test, Mann-Whitney U-test was used.The results indicate there is a significant difference between control and electrical stimulation, [U = 2, P = 0.0173].* P < 0.05.C57BL/6J wild-type mice were used in these experiments.

Local brain environment reactions of LHb in anxiogenic environments
Local brain environment reactions in the LHb in response to mice being placed in a highly anxiogenic AMC environment were examined.As mentioned earlier, an increase in the theta-delta ratio in the LFP recorded using the LHb electrodes was observed while the mice were placed in the AMC (Fig. 6A, top two traces).The theta-delta ratio increase was not constant, and some fluctuations were observed (Fig. 6A, bottom).All YC nano50 fluorescence deflected negatively, suggesting an increase in BBV.This was confirmed using the mScarlet recordings, which showed a prominent increase during AMC exposure (Fig. 6B).The mScarlet and dYFP traces nearly mirrored each other.Complete mirroring would suggest no change in the cytosolic pH.However, the time course of pH and BBV changes could be similar; thus, from these traces alone, the presence or absence of pH fluctuations could not be concluded.A small increase in fYFP relative to dYFP was observed upon AMC entry, which was accompanied by a transient decrease in fCFP, suggesting a transient Ca 2+ increase.However, basal Ca 2+ , as estimated by fYFP minus dYFP, was mostly negative, albeit the presence of spontaneous transient Ca 2+ increases throughout the remainder of the time the mouse spent in the AMC (Fig. 6C).The average Ca 2+ instantaneous increase, sustained decrease, and mScarlet increase for four animals are shown in Fig. 6D.

Cytosolic pH reduction in LHb astrocytes in anxiogenic environment
To verify the pH fluctuations in the anxiogenic environment, transgenic mice with astrocyte-specific expression of a pH sensor, E 2 GFP (Bizzarri et al., 2006) were used.E 2 GFP was conjugated with a membrane-tethered structural domain, Lck; thus, the detected signals mainly reflected pH changes in the submembrane region of the cytosol (Shigetomi et al., 2010;Onodera et al., 2021).E 2 GFP was excited using 450 nm light, and fluorescence emission signals were detected at 560 nm (EmR) and 502 nm (EmG).Similar to YC nano50 , the astrocytic E 2 GFP response to REM sleep was first examined.Both fluorescence signals showed a negative deflection, suggesting that the BBV increase obstructed the detection of fluorescence in the brain parenchyma (Ikoma et al., 2023b).
In the context of E 2 GFP fiber photometry recording performed in the lateral hypothalamus, we assumed that the negative deflection of the fluorescence during the early phase of REM sleep primarily reflected the effects of increase in BBV.Therefore, the baseline fluorescence level, which was unaffected by changes in BBV and pH, was estimated and subtracted such that the two fluorescence traces overlapped completely during the early phase.After adjusting the baseline, a small difference in the time courses of the two fluorescence traces remained, with the negative deflection of the EmR trace being greater than that of the EmG trace (Fig. 7A).E 2 GFP fluorescence excited at 450 nm and detected using EmR was expected to largely decrease with acidification, while the EmG signal was expected to be relatively stable (Bizzarri et al., 2006) (Fig. 7B).Therefore, the EmR -EmG calculation showed a pH estimate trace with a downward deflection, reflecting cytosolic acidification (Fig. 7A).
The cytosolic pH changes during REM sleep were estimated to be mostly acidic, as was the case in the lateral hypothalamus (Ikoma et al., 2023b) (Fig. 7D, left).Due to the above baseline adjustment method, if cytosolic acidification started during the early phase, the calculated acidification estimate would be an underestimation of the true acidification.The fluorescence signal changes from the E 2 GFP recordings compared to background noise was quite small, probably because of the reduced E 2 GFP expression in the habenula compared to the lateral hypothalamus and also due to the use of a smaller diameter (200 µm) optical fiber than that used previously (400 µm diameter).
The mouse was placed in the AMC to measure the pH changes associated with anxiety.A typical increase in theta-band activity was detected in the LFP recorded from the LHb.For individual animals, the same baseline adjustment of E 2 GFP fluorescence signals using REM sleep episode was used to analyze the fluorescence signal reactions in the LHb to the AMC environment.Similar to the YC nano50 recordings, both EmR and EmG traces showed a negative deflection, suggesting an increase in BBV upon AMC exposure.In addition, EmR was found to deflect more negatively than EmG.Therefore, we examined the acidic reaction of astrocytes in the LHb in response to AMC exposure (Fig. 7C).All tested mice showed an acidic pH shift in the anxiogenic environment (Fig. 7D, right).

Astrocytic ArchT optogenetic manipulation induced anxiolytic effects
Cytosolic acidification of astrocytes in the LHb was induced when mice were placed in an anxiogenic environment.Acidic reactions in astrocytes may evoke astrocyte-to-neuron interactions, such as release of gliotransmitters, and may result in the creation of theta-band oscillations in neuronal signals and amplify anxiety reactions in mice.If such was expressed selectively in astrocytes, but the expression was relatively low in the cortex and especially in the hippocampus (Gavrilov et al., 2018).Higher expression was observed in the LHb and thalamus.(B) Configuration of the fiber photometry system.415 nm and 505 nm lights were used for excitation of YC nano50 , and 595 nm light was used to excite mScarlet.These three excitation lights were filtered, combined, and delivered with a sequential time course to the mouse brain, as shown below.The emitted light came back through the optical fiber and split into two wavelengths.fCFP and mScarlet were detected using PMT1, and fYFP and dYFP were detected using PMT2.(Ikoma et al., 2023b), this suggests that an increase in the brain blood volume (BBV) occurred.Increase in blood vessel diameter likely occurs upon REM sleep which obscures any fluorescence signal coming from the brain parenchyma.When the fluorescence from mScarlet secreted in the blood plasma was observed, a clear increase in this signal was detected.When the dYFP trace was inverted and scaled, it overlapped almost completely with the mScarlet trace (bottom).Mlc1-tTA::tetO-YC nano50 transgenic mice were used in these experiments.mechanisms operate in the LHb, countering the acidification by optogenetic alkalization of the cytosol may result in an anxiolytic effect.To explore this possibility, transgenic mice with astrocyte-specific expression of ArchT were used (Fig. 8A; Mlc1-tTA::tetO-ArchT).ArchT is an outward proton pump, and its photoactivation results in cytosolic alkalinization (Beppu et al., 2014).Two optical fibers were bilaterally implanted immediately above the LHb to deliver 505 nm light for ArchT photoactivation (Fig. 8A).One of the fibers was accompanied by a pair of tungsten wire electrodes to measure the LFP activity in the LHb.When the mouse was placed in an anxiogenic environment, AMC, an increase  in the 5 -10 Hz theta-band LFP was observed (Fig. 8B).However, 10 s of bilateral ArchT photoactivation of the LHb resulted in attenuation of the theta-band activity of LFP in the LHb.
Next, we tested whether astrocytic ArchT photoactivation would result in attenuation of anxiety using the 2-Way Bright-Dark apparatus (Fig. 8C).When the mouse was first placed in the dark cage, it tended to stay in the dark cage in the absence of ArchT photoactivation (Fig. 8C, control).In a separate group of mice, ArchT was photoactivated while they stayed in the home cage, and photoactivation continued after they were transferred to the dark cage of the 2-Way Bright-Dark apparatus with a 2 s on and 3 s off cycle.The overall residence time in the dark cage was still higher than that in the bright cage; however, the mouse tended to venture more into the bright cage (Fig. 8C, ArchT stim).
The ratio of the cumulative distance traveled by the mouse in the bright cage was divided by the total distance traveled and plotted against time after the mice were introduced into the 2-Way Bright-Dark apparatus.In both control and ArchT photoactivated mice, the travel distance ratio in the bright cage increased gradually to a saturating level The average profile of the cumulative travel distance in the bright cage when the mouse was first placed in it (n = 5 from the control and ArchT group, respectively).When the mice were placed in the bright cage of the 2-Way Bright-Dark apparatus, the mice tended to leave the bright cage and enter the dark cage swiftly in control.However, with ArchT photoactivation, the mouse tended to travel more in the bright cage for a while before leaving it; however, no statistical significance was found between the two groups.The cumulative travel distance ratio at 0.5 min was compared between the two groups.A non-parametric test, Mann-Whitney U-test was used.The results indicate there is no significant difference between control and ArchT photostimulation, [U = 8, P = 0.3651].NS, not significant.Mlc1-tTA::tetO-ArchT mice were used in these experiments. of ~0.32 by the end of the 20-min trial (Fig. 8D).This shows that, even in the control group, the curiosity to explore a new environment overwhelmed the anxiety created by the bright cage and the mice travelled approximately one-third of the total distance traveled in the bright cage.However, the increase in the travel distance ratio in the bright cage was faster in ArchT photoactivated mice.When the ratio was compared at 8.5 min after the dark cage transfer moment, a significant increase in the travel distance ratio was observed in ArchT-photoactivated mice (Fig. 8D, right).
In a previous study on anxiety in rodents using the EPM, anxiety reactions differed when the animal was placed in the EPM with its face towards the open or closed arm (Pellow et al., 1985).Here, we tested animal behavior when the mouse was first placed in the bright cage of the 2-Way Bright-Dark apparatus.As expected, the mice often moved quickly to the dark cage to avoid the anxiogenic bright cage.However, in ArchT-photoactivated mice, this reaction was not always instant and a certain travel distance in the bright cage often accumulated early during the trial (Fig. 8E).Therefore, the average travel distance ratios in the bright cage at 0.5 min from the start of the trial were compared; however, statistical significance was not attained (Fig. 8E, right).Taken together, a slight anxiolytic effect was observed following astrocytic ArchT photoactivation in the LHb.

Discussion
LHb is one of the few brain regions that regulates both dopaminergic and serotoninergic systems.LHb modulates emotional and cognitive behaviors through these downstream systems.LHb neurons can be activated by various negative emotional stimuli such as inescapable foot or tail shock, maternal deprivation, and social defeat stress (Wirtshafter et al., 1994).Therefore, LHb essentially acts as a negative reinforcement center in the brain (Wang and Aghajanian, 1977).Increases in spontaneous firing of LHb neurons in response to aversive stimuli have been demonstrated using in vivo electrophysiological recordings (Lecca et al., 2017).Recently, whole-brain Ca 2+ imaging in zebrafish revealed that the ventral habenula, which is the equivalent of LHb in mammals, was the only region that exhibited hyperactivity in response to repetitive shocks (Andalman et al., 2019).Our results show an increase in theta-band oscillations in the LFP recorded from electrodes placed in the LHb when the mouse was exposed to the highly anxious environment of the EA.We also established a new anxiogenic environment, the AMC, and confirmed robust theta-band oscillations in the LHb when placed in the AMC.Importantly, similar elevation in theta-band activity was observed when the mouse spontaneously entered the central area from the periphery in a classical open-field paradigm.This theta-band increase did not depend on whether the mouse slowed down or sped up upon center zone entry.These data suggest that the unique theta-band feature in the LHb correlates with anxiety.
Astrocyte-neuron interaction is increasingly being proposed as a mechanism responsible for various psychiatric disorders.Drugs, such as fluoxetine and clozapine, have been used to treat mental disorders, such as depression and schizophrenia.Studies have shown that these drugs affect astrocyte function (Allaman et al., 2011;Tanahashi et al., 2012).It has also been reported that intracellular Ca 2+ signals increase in hippocampal astrocytes in response to anxiogenic environments (Cho et al., 2022).The astrocytic potassium channel, Kir4.1, has also been shown to be upregulated in the LHb in a rat model of congenitally learned helplessness (cLH) depression (Yang et al., 2018).Such upregulation has been shown to result in increased burst firing in the LHb and triggering of depressive-like behaviors (Cui et al., 2018).It has also been reported that habenula-specific inhibition of glial glutamate transporter, GLT1, results in an increased firing rate of the LHb neurons, which also results in depressive-like behavior in mice (Cui et al., 2014).
Our study shows that functional alterations in neurons and astrocytes in the LHb occur when mice are exposed to an anxiogenic environment.With the quad fluorescence recording system in freely moving mice, LHb astrocytes responded to the anxiogenic environment by a reduction in intracellular Ca 2+ and an increase in the local BBV.This reduction in basal Ca 2+ level was unexpected.Such decrease has been observed in various situations, including REM sleep in the lateral hypothalamus (Ikoma et al., 2023b); however, the mechanism of triggered Ca 2+ decrease has not yet been determined.Spontaneous transient Ca 2+ increases were observed during the time the mice spent in the anxiogenic environment.
Anxiolytic effects on behavior were observed when the LHb was photostimulated in mice expressing ArchT specifically in astrocytes, which was accompanied by reduction of theta-band power in the LFP.These results suggest that astrocytic reactions to the anxiogenic environment may result in the activation of astrocytes, leading to theta-band oscillations in the LHb (Matsui and Jahr, 2003;Hamilton and Attwell, 2010;Letellier et al., 2016;Shen et al., 2017).This reciprocal neuron-astrocyte interaction could exacerbate anxiety.In fact, an acidic shift in LHb astrocytes was observed in an anxiogenic environment.Astrocytes can become acidified via activation of glutamate transporters, which bring in H + which is coupled during electrogenic transport (Wadiche et al., 1995;Rose and Ransom, 1996;Bergles et al., 1997;Matsui et al., 1999).Glutamate transporters can also send out HCO 3 -via anion channel conductance which is activated along with glutamate transport, which would also result in cytosolic acidification (Wadiche et al., 2006).Acidification triggers astrocytic release of glutamate via anion channel composed of LRRC8 subunits in cerebellar astrocytes (Nilius et al., 1998;Lee et al., 2010;Gourine et al., 2010;Hyzinski--García et al., 2014;Park et al., 2015;Beppu et al., 2021;Kanaya et al., 2023).Astrocyte glutamate release by acidification can send neurons to a state where they become prone to theta-band oscillations.Astrocytic ArchT photoactivation would lead to countering of acidification via the H + outward pump and may be the reason for theta-band dissipation and anxiolytic behavior shift.
In conclusion, our study demonstrated that reciprocal interactions between LHb neurons and astrocytes control the local brain environment, which may regulate anxiety-like behaviors in mice.Therefore, a new therapeutic strategy may be directed towards LHb astrocytes to control anxiety disorders.

Fig. 3 .
Fig. 3. Increased theta-band activity with spontaneous mouse relocation from the edge to the center of the open field.(A) When a mouse is placed in an open field, the mouse tends to stay close to the edge of the cage.A heat map was created based on the time of residence in each location.The heat map shows a higher residential time of the mouse on the edges and, especially, in the corner of the open field.An example of the mouse's track during the 15-min test in the open field is shown.The grey area indicates an arbitrary set region defined as the edge zone of the open field.(B) The raw trace (third trace from top) and the power spectrum (second trace from top) of the LFPs in the LHb are shown.The time when the mouse was placed in the open field and the time that the mouse was taken out of the open field are shown.The mouse's transition between the center and edge of the open field is shown on top.The transition of theta-delta ratio of the LFP was calculated with relatively large time window (1 s, bottom trace) and low-pass filtered at 400 mHz.Theta/delta increased upon introduction to the open field, which gradually decreased during the exposure.(C) Shown are the sample LFP recordings during three episodes when the mouse entered the central area from the periphery.Even in these individual raw LFP traces after low-pass filtration at 15 Hz, an increase in theta-band oscillation is observed.(D) The power spectrum of the LFPs in the LHb when the mouse was still in the edge area but just prior to entering the central area was calculated.The power spectrum after the mouse entered the central area was also calculated.A clear decrease in the 1 -4 Hz delta-band power and an increase in the 5 -10 Hz theta-band power was observed.The time window of each power spectrum is 2 s.The average of power spectrums across 21 episodes was calculated and shown here.(E) The average increase in theta/delta ratio of the log 10 transformed power spectrum upon entering the central area was calculated for multiple episodes in individual animals.Data from individual animal is plotted and the mean ± SEM from multiple animals (n = 6 animals) are shown with a bar graph and error bars.C57BL/6J wild-type mice were used in these experiments.

Fig. 5 .
Fig. 5. Measurements of local brain environmental changes in the habenula by observation of FRET-based Ca2+ sensor expressed in astrocytes and mScarlet secreted in the blood plasma.(A) Brain map of the LHb (left) and actual brain slice preparation from the Mlc1-tTA::tetO-YC nano50 mouse.The implantation location of the optical fiber is shown with vertical dashed white lines.YC nano50 was expressed selectively in astrocytes, but the expression was relatively low in the cortex and especially in the hippocampus(Gavrilov et al., 2018).Higher expression was observed in the LHb and thalamus.(B) Configuration of the fiber photometry system.415 nm and 505 nm lights were used for excitation of YC nano50 , and 595 nm light was used to excite mScarlet.These three excitation lights were filtered, combined, and delivered with a sequential time course to the mouse brain, as shown below.The emitted light came back through the optical fiber and split into two wavelengths.fCFP and mScarlet were detected using PMT1, and fYFP and dYFP were detected using PMT2.(C) Optical fluorescence response to electrical stimulus delivered to the LHb.fYFP signal increased and fCFP decreased, and the two traces relatively mirrored each other.dYFP and mScarlet signals were relatively unchanged.All fluorescence traces are shown in percentage scale relative to their baseline fluorescence level.(D) A characteristic theta-band increase in the recorded EEG was observed during REM sleep in mice (top two traces).Negative deflections of all three optical signals (fYFP, dYFP, fCFP) from the YC nano50 expressed in astrocytes in the LHb were observed.As reported previously(Ikoma et al., 2023b), this suggests that an increase in the brain blood volume (BBV) occurred.Increase in blood vessel diameter likely occurs upon REM sleep which obscures any fluorescence signal coming from the brain parenchyma.When the fluorescence from mScarlet secreted in the blood plasma was observed, a clear increase in this signal was detected.When the dYFP trace was inverted and scaled, it overlapped almost completely with the mScarlet trace (bottom).Mlc1-tTA::tetO-YC nano50 transgenic mice were used in these experiments.
Fig. 5. Measurements of local brain environmental changes in the habenula by observation of FRET-based Ca2+ sensor expressed in astrocytes and mScarlet secreted in the blood plasma.(A) Brain map of the LHb (left) and actual brain slice preparation from the Mlc1-tTA::tetO-YC nano50 mouse.The implantation location of the optical fiber is shown with vertical dashed white lines.YC nano50 was expressed selectively in astrocytes, but the expression was relatively low in the cortex and especially in the hippocampus(Gavrilov et al., 2018).Higher expression was observed in the LHb and thalamus.(B) Configuration of the fiber photometry system.415 nm and 505 nm lights were used for excitation of YC nano50 , and 595 nm light was used to excite mScarlet.These three excitation lights were filtered, combined, and delivered with a sequential time course to the mouse brain, as shown below.The emitted light came back through the optical fiber and split into two wavelengths.fCFP and mScarlet were detected using PMT1, and fYFP and dYFP were detected using PMT2.(C) Optical fluorescence response to electrical stimulus delivered to the LHb.fYFP signal increased and fCFP decreased, and the two traces relatively mirrored each other.dYFP and mScarlet signals were relatively unchanged.All fluorescence traces are shown in percentage scale relative to their baseline fluorescence level.(D) A characteristic theta-band increase in the recorded EEG was observed during REM sleep in mice (top two traces).Negative deflections of all three optical signals (fYFP, dYFP, fCFP) from the YC nano50 expressed in astrocytes in the LHb were observed.As reported previously(Ikoma et al., 2023b), this suggests that an increase in the brain blood volume (BBV) occurred.Increase in blood vessel diameter likely occurs upon REM sleep which obscures any fluorescence signal coming from the brain parenchyma.When the fluorescence from mScarlet secreted in the blood plasma was observed, a clear increase in this signal was detected.When the dYFP trace was inverted and scaled, it overlapped almost completely with the mScarlet trace (bottom).Mlc1-tTA::tetO-YC nano50 transgenic mice were used in these experiments.

Fig. 6 .
Fig. 6.Habenular local brain environmental changes are associated with anxiety.(A) As shown previously, a characteristic theta-band activity was observed in the LFP recorded using the electrodes in the LHb upon all marble cage (AMC) exposure.The transition of theta-delta ratio of the LFP was calculated with relatively a large time window (1 s, bottom trace) and low-pass filtered at 5 mHz.(B) Four fluorescence traces recorded using the optical fiber inserted in the LHb are shown.Upon exposure to the AMC, fYFP, dYFP, and fCFP all decreased but mScarlet increased, suggesting an increase in the local brain blood volume (BBV).(C) fYFP minus dYFP yields an estimate of the Ca 2+ transitions.Transiently upon AMC introduction, a clear increase in Ca 2+ was observed.However, while the mouse was still in the AMC, the relationship between fYFP and dYFP reversed, and a sustained Ca 2+ decrease was estimated.Later, an almost mirror image of the mScarlet and dYFP trace was observed, which suggests a prominent increase in BBV while the animal stayed in the AMC.(D) Immediate Ca 2+ increase and sustained Ca 2+ decrease response to AMC exposure across four animals are summarized.A sustained increase in BBV is also shown by increase in mScarlet.Data from each animal are shown along with the bar graph and the error bar representing mean ± SEM.Mlc1-tTA::tetO-YC nano50 transgenic mice were used in these experiments.

Fig. 7 .
Fig. 7. Astrocytic pH alteration with REM sleep and anxiety.(A) A pH sensor, Lck-E 2 GFP, was specifically expressed in astrocytes (Mlc1-tTA::tetO-E 2 GFP).A characteristic theta-band increase in the recorded EEG was observed upon REM sleep in mice (top two traces).With the 450 nm excitation, the fluorescent signal taken at ~560 nm with PMT1 (shown in a red line) is expected to decrease with acidification, while the signal taken at ~502 nm with PMT2 (shown in a red line) is expected to remain relatively constant.Thus, the (Ex450 nm EmR -EmG) (the dark line) should result in a pH estimate with negative deflection reflecting acidification.(B) Optical signal detection strategy for E 2 GFP.(C) The characteristic theta-band activity in the LFP recorded using the electrodes in the LHb was observed upon being placed in the all marble cage (AMC), as previously described.The optical signals obtained by 450 nm excitation were recorded by PMT1 (EmR) and PMT2 (EmG), respectively.Both signals showed a negative deflection during AMC; however, the deflection was more significant for EmR.Thus, the estimate of pH dynamics was negatively deflected reflecting acidification.(D) The amplitudes of estimated pH deflection to REM sleep (n = 10 REM sleep episodes from four animals) (left) and AMC (n = 7 episodes from four animals) (right) are shown.The pH change estimate was mostly acidic in REM sleep, whereas all pH changes were estimated to be acidic in AMC.Data from each animal are shown along with the bar graph and error bar representing mean ± SEM.Mlc1-tTA::tetO-E 2 GFP transgenic mice were used in these experiments.

Fig. 8 .
Fig. 8. Astrocytic ArchT optogenetic manipulation induces anxiolytic behavior.(A) The location of bilateral implantation of optical fiber targeted towards the LHb is shown (left).Actual brain slice preparation from the Mlc1-tTA::tetO-ArchT mouse is shown on the right.White dashed lines symmetrically tilted at 10 degrees indicate the location of optical fibers, and the red dashed lines indicate the location of the habenula.(B) Upon being placed in the AMC, 5 -10 Hz theta-band activity in the LFP was recorded using the electrodes in the LHb.The theta-band oscillations became weak after photoactivating the ArchT expressed in astrocytes (timing shown below the spectrogram).The power spectrum of the LFPs in the LHb before (left) and after (right) astrocytic ArchT photoactivation.5 -10 Hz theta-band decreased after astrocytic ArchT photoactivation.Time windows of the power spectrum before and after ArchT photoactivation were both 4 min (C) Astrocytic ArchT photoactivation in the 2-Way Bright-Dark apparatus experiment.The 2-Way Bright-Dark apparatus, consisting of the dark cage on the left and the bright cage on the right is shown on top.Bottom: Astrocytic ArchT photoactivation schedule before and during the 2-Way Bright-Dark apparatus experiment.ArchT was photoactivated for 10 s initially while the mouse stayed in its home cage and subsequently, intermittently, stimulated with 2 s on and 3 s off during the 20 min it stayed in the 2-Way Bright-Dark apparatus.Heat map of the mouse shows that the mouse mostly avoided entering the anxiety-inducing glass marbles in the bright cage.In contrast, with photoactivation of the ArchT expressed in astrocytes, the mouse ventured more into the bright cage and spent more time in it.The mouse was first placed in the dark cage of the 2-Way Bright-Dark apparatus.(D) The cumulative travel distance in the bright cage was divided by the total distance traveled and plotted against time.The mouse was first placed in the dark cage and stayed in the 2-Way Bright-Dark apparatus for 20 min.This travel distance profile was averaged across n = 8 mice in the control and ArchT photostimulated groups, respectively.The timing of the photoactivation of the ArchT is shown in C. The mice tended to stay in the dark cage but gradually ventured to the bright cage.The cumulative travel distance ratio at 8.5 min was compared between the two groups and a significant increase was observed in the ArchT photoactivated group.A non-parametric test, Mann-Whitney U-test was used.The results indicate there is a significant difference between control and ArchT photostimulation, [U = 13, P = 0.0482].* P < 0.05.(E) The average profile of the cumulative travel distance in the bright cage when the mouse was first placed in it (n = 5 from the control and ArchT group, respectively).When the mice were placed in the bright cage of the 2-Way Bright-Dark apparatus, the mice tended to leave the bright cage and enter the dark cage swiftly in control.However, with ArchT photoactivation, the mouse tended to travel more in the bright cage for a while before leaving it; however, no statistical significance was found between the two groups.The cumulative travel distance ratio at 0.5 min was compared between the two groups.A non-parametric test, Mann-Whitney U-test was used.The results indicate there is no significant difference between control and ArchT photostimulation, [U = 8, P = 0.3651].NS, not significant.Mlc1-tTA::tetO-ArchT mice were used in these experiments.