Abstract
Light plays an essential role in a variety of physiological processes, including vision, mood, and glucose homeostasis. However, the intricate relationship between light and an animal's feeding behavior has remained elusive. Here, we found that light exposure suppresses food intake, whereas darkness amplifies it in male mice. Interestingly, this phenomenon extends its reach to diurnal male Nile grass rats and healthy humans. We further show that lateral habenula (LHb) neurons in mice respond to light exposure, which in turn activates 5-HT neurons in the dorsal Raphe nucleus (DRN). Activation of the LHb→5-HTDRN circuit in mice blunts darkness-induced hyperphagia, while inhibition of the circuit prevents light-induced anorexia. Together, we discovered a light-responsive neural circuit that relays the environmental light signals to regulate feeding behavior in mice.
Significance Statement
Feeding behavior is influenced by a myriad of sensory inputs, but the impact of light exposure on feeding regulation has remained enigmatic. Here, we showed that light exposure diminishes food intake across both nocturnal and diurnal species. Delving deeper, our findings revealed that the LHb→5-HTDRN neural circuit plays a pivotal role in mediating light-induced anorexia in mice. These discoveries not only enhance our comprehension of the intricate neuronal mechanisms governing feeding in response to light but also offer insights for developing innovative strategies to address obesity and eating disorders.
Introduction
Light information transduced by the retina modulates a wide range of physiological behaviors, such as sleeping, locomotion, and mood (Vandewalle et al., 2009; LeGates et al., 2014). Interestingly, nocturnal rodents typically consume substantially more food at night (Olsen et al., 2017; Challet, 2019). Exposure to excessive light in the dark phase is associated with aberrant food intake and body mass (Fonken et al., 2010; Borniger et al., 2014), suggesting that light may regulate feeding. However, long-term light manipulations often result in a disrupted circadian clock, which is a well-known cause of metabolic disorders (Rudic et al., 2004; Turek et al., 2005; Kettner et al., 2015). The direct effect of light on food intake, therefore, remains elusive.
The lateral habenula (LHb), a part of the epithalamus, is a highly conserved nucleus across species (Aizawa et al., 2011). Hyperactivity of the LHb leads to depressive-like symptoms, which are ameliorated by light therapy (Lam et al., 2016; Cui et al., 2018; Yang et al., 2018a,b; Hu et al., 2020). Recently, it was reported that the firing pattern of LHb neurons in nocturnal mice exhibits a heightened rate during the light cycle compared with the dark cycle (Sakhi et al., 2014a,b). In addition, the neuronal activity of the LHb is affected by light (Huang et al., 2019). It is worth mentioning that stimulation of the glutamatergic terminals from the lateral hypothalamus (LH) to the LHb decreased the consumption of palatable food (Stamatakis et al., 2016), indicating that the LHb might be involved in feeding regulation. More importantly, the LHb plays an essential role in regulating the midbrain monoaminergic centers, including the midbrain serotonin (5-hydroxytryptamine, 5-HT) neurons (Hu et al., 2020).
The central 5-HT systems have been implicated in the control of satiety for several decades (Burke and Heisler, 2015). Both central administration of 5-HT and its precursor have been linked to appetite and body weight suppression, whereas a deficiency of 5-HT in the brain has been associated with hyperphagia and obesity (Breisch et al., 1976; Saller and Stricker, 1976; Blundell and Latham, 1979). Notably, 5-HT-producing neurons reside predominantly in the dorsal Raphe nucleus (DRN; Okaty et al., 2015). Specific removal of tryptophan hydroxylase 2 (TPH2), the enzyme pivotal for 5-HT synthesis, within the DRN, resulted in heightened food intake and subsequent weight gain (Liu et al., 2021). Indeed, the brain 5-HT system has emerged as a promising target in extensive clinical studies. For example, d-fenfluramine and lorcaserin, which function as the selective 5-HT reuptake inhibitor and 5-HT 2C receptor agonist, respectively, were approved as antiobesity therapies in humans, although these drugs were no longer used clinically due to undesirable side effects (McGuirk et al., 1991; Smith et al., 2010; Martin et al., 2011). 5-HTDRN neurons receive direct synaptic inputs from complex networks, including the LHb, to modulate various physiological and pathological processes, such as eating, depression, and anxiety (Simansky, 1996; Lucki, 1998; Pollak Dorocic et al., 2014). However, whether 5-HTDRN neurons can integrate information from the LHb to orchestrate light-associated changes in feeding remains unclear.
In this study, we demonstrated that light exposure dampens food intake not only in nocturnal mice but also in diurnal Nile grass rats and even in humans. Employing an array of transneuronal virus tracing techniques, fiber photometry, and chemogenetic approaches, we identified an LHb→5-HTDRN circuit that plays a key role in orchestrating light-induced anorexia in mice. Activation of the LHb→5-HTDRN circuit abolished the darkness-evoked increase in feeding, whereas inhibition of this circuit impaired light-induced anorexia in mice. Together, we uncovered a novel LHb→5-HTDRN pathway in mediating light-induced alterations in feeding in mice.
Materials and Methods
Animals
Adult male wild-type, TPH2-CreER/Rosa26-LSL-tdTomato and TPH2 flox/flox mice (8–12 weeks old) were used in this study. We crossed TPH2-CreER mice (Jackson Laboratory, #016584) with Rosa26-LSL-tdTomato mice (Jackson Laboratory, #007905) to generate TPH2-CreER/Rosa26-LSL-tdTomato mice. Tamoxifen injection (0.2 g/kg, i.p., MedChemExpress, #42611) in these mice at 8 weeks of age induced Cre activity and therefore led to the expression of tdTomato selectively in 5-HT neurons. The TPH2 flox/flox mouse line was purchased from Jackson Laboratory (#027590) and backcrossed to the C57Bl6j background. We crossed male and female TPH2 flox/flox mice to generate littermate TPH2 flox/flox mice as study cohorts. Nile grass rats were from Dr. Lily Yan at Michigan State University (Soler et al., 2021). Mice and Nile grass rats were housed in a temperature-controlled environment (23°C) using a 12 h light/dark cycle. Mice and Nile grass rats that were used for food intake measurements were singly housed 1 week before the study. The mice were fed a standard chow diet (6.5% fat, Harlan-Teklad, #2920). Water was provided ad libitum. Care of all animals and procedures was approved by the Baylor College of Medicine Institutional Animal Care and Use Committee.
Stereotaxic surgery
Mice were anesthetized (with 2% isoflurane) and placed in a stereotaxic instrument. Artificial eye ointment was applied to prevent corneal drying, and a heat pad was used to hold the body temperature at 37°C. To specifically identify the downstream targets of LHb glutamatergic neurons, we injected AAV9-CaMKIIa-ChR2 (H134R)-GFP into the LHb of wild-type mice (Addgene, #26969; titer, 5 × 1012 GC/ml, 0.2 µl/injection; AP, −1.7 mm; ML, ±0.5 mm; DV, −2.8 mm). To retrogradely label DRN-projecting LHb neurons, retrograde AAV-GFP was injected into the DRN of wild-type mice (Addgene, #50465; titer, 5 × 1012 GC/ml, 0.2 µl/injection; AP, −4.65 mm; ML, 0 mm; DV, −3.6 and −3.3 mm).
To record the response of LHb neurons (which send projections to the DRN) to light through fiber photometry, we stereotaxically injected retrograde AAV-Cre (Addgene, #105553-AAVrg, titer, 5 × 1012 GC/ml, 0.2 µl/injection) into the DRN and delivered AAV8-DIO-GCaMP6 m (Addgene, #100838; titer, 5 × 1012 GC/ml, 0.2 µl/injection) into the LHb. During the same surgery, an optic fiber was placed 0.2 mm above the LHb, and fibers were fixed to the skull using dental acrylic.
To uncover the synaptic properties of the LHb→5-HTDRN circuit, TPH2-CreER/Rosa26-LSL-tdTomato mice received tamoxifen injection, and then the LHb was infected with AAV9-CaMKIIa-ChR2 (H134R)-GFP. Four weeks later, the brain sections containing LHb were subjected to electrophysiological recording.
To specifically infect the DRN-projecting LHb neurons with DREADDs, retrograde AAV-Cre was administrated into the DRN, and AAV8-DIO-hM3Dq-mCherry (Addgene, #44361; titer, 5 × 1012 GC/ml, 0.2 µl/injection) or AAV8-DIO-hM4Di-mCherry (Addgene, #44362; titer, 5 × 1012 GC/ml, 0.2 µl/injection) was delivered into the LHb, while the LHb was infected with AAVDJ-fDIO-hM3Dq-mCherry (Aklan et al., 2020; plasmid was kindly provided by Dr. Huxing Cui and Dr. Deniz Atasoy, and virus was packaged by the Baylor IDDRC Neuroconnectivity Core; titer, 1 × 1012 GC/ml, 0.2 µl/injection).
To generate TPH2DRN-KO and their controls, TPH2 flox/flox mice received either AAV8-Cre (UNC GTC Vector Core, #AV5053C; titer, 6.5 × 1012 GC/ml, 0.2 µl/injection) or AAV8-GFP (UNC GTC Vector Core, #AV5075E; titer, 5.6 × 1012 GC/ml, 0.2 µl/injection) injection into the DRN. To specifically infect the DRN-projecting LHb neurons with hM3Dq in TPH2DRN-KO mice, AAV8-Cre and retrograde AAV-Flpo were administrated into the DRN, while the LHb was infected with AAVDJ-fDIO-hM3Dq-mCherry (Aklan et al., 2020; plasmid was kindly provided by Dr. Huxing Cui and Dr. Deniz Atasoy, and virus was packaged by the Baylor IDDRC Neuroconnectivity Core; titer, 1 × 1012 GC/ml, 0.2 µl/injection).
Acute light manipulations and food intake measurements
Mice were kept in a normal light/dark cycle (6 A.M.–6 P.M.) before light exposure, with the light illumination maintained at 300 lux in the housing room. A group of mice were exposed to a constant light cycle (300 lux white light illumination) for 2 weeks in order to disrupt the circadian clock (confirmed by actogram activity) before they were subjected to acute light manipulation and food intake measurements. All other mice were housed in a 12 h light (6 A.M.–6 P.M., 300 lux white light illumination) and 12 h dark (6 P.M.–6 A.M.) cycle and were never exposed to circadian disruptions before the acute light manipulation experiments. During the acute light manipulation experiments, we provided light by turning on the lights in the room (full-length white light, 300 lux; the intensity was determined by averaging the measurements from the top and the four sides of the cage) and darkness by turning off the lights. The timing of these acute light or darkness exposures and the timing of food intake measurements were indicated in each figure describing these experiments. For chemogenetic studies, saline or CNO (3 mg/kg, i.p., Cayman Chemical, #16882) injections were given 0.5 h before the feeding measurements.
Similarly, all Nile grass rats were housed in a 12 h light (6 A.M.–6 P.M., 300 lux white light illumination) and 12 h dark (6 P.M.–6 A.M.) cycle and were never exposed to circadian disruptions before the acute light manipulation experiments. We first verified the diurnal pattern of their feeding behavior by monitoring food intake at the first 3 h of the light (6–9 A.M.) and dark cycle (6–9 P.M.). To determine the effects of light on food intake in Nile grass rats, we manipulated the intensity of light at 50, 300, and 3,000 lux by turning down the light intensity or adding more bulbs (cool LED lights, UV-free) above the cages during 6–7 A.M., and food intake during this period was measured. To test the effects of lorcaserin on food intake in mice, male WT mice were fasted for 12 h (6 A.M.–6 P.M.) and then received intraperitoneal injection of lorcaserin (9 mg/kg, AdooQ Bioscience, A12598) at 5:30 P.M. before food was provided at 6 P.M.; food intake was measured for a period of 2 h (6–8 P.M.). Similarly, male Nile grass rats were fasted for 12 h (9 P.M.–9 A.M.) and then received intraperitoneal injection of lorcaserin (9 mg/kg, AdooQ Bioscience, A12598) at 8:30 A.M. before food was provided at 9 A.M.; food intake was measured for a period of 2 h (9–11 A.M. with 300 lux white light illumination).
HomeCageScan
To evaluate whether light manipulation affects other behaviors in addition to feeding, we put mice into the HomeCageScan (CleverSys) to record their natural behaviors in the absence or presence of light at various periods of the day (2–6 A.M., 6–10 A.M., 2–6 P.M., 6–10 P.M.). Behaviors were analyzed by the HomeCageScan software.
RNAscope
We injected retrograde AAV-Cre (Addgene, #105553-AAVrg; titer, 5 × 1012 GC/ml, 0.2 µl/injection) into the DRN, and a Cre-dependent mCherry virus (AAV8-DIO-mCherry, Addgene, #50459; titer, 5 × 1012 GC/ml, 0.2 µl/injection) was injected into the LHb to label DRN-projecting LHb neurons. Four weeks later, mice were anesthetized and perfused transcardially with 0.9% saline followed by 10% formalin. Brains were removed and postfixed in 10% formalin for 16 h at 4°C and cryoprotected in 30% sucrose for 48 h. Brains were frozen, sectioned at 14 µm using the cryostat, and washed in DEPC-treated phosphate-buffered saline for 10 min. Sections were mounted on DEPC-treated charged slides, dried for 0.5 h at room temperature, and stored at −80°C. On the day of the RNAscope assay, the slides were thawed, rinsed two times in PBS 1×, and placed in an oven for 30 min at 60°C. After that, slides were postfixed in 10% formalin for 15 min at 4°C. Slides were then gradually dehydrated in ethanol (50, 70, and 100%, 5 min each) and underwent target retrieval for 5 min at 100°C. Slides were incubated in protease III (#322337, ACDBio) for 30 min at 40°C. Slides were then rinsed in distilled water and incubated in RNAscope probes for mCherry (#513201, ACDBio), vGLUT2 (#319171-C3, ACDBio), and vGAT (#319191-C2, ACDBio) for 2 h at 40°C. Sections were then processed using the RNAscope Fluorescent Multiplex Detection Reagents (#320851, ACDBio) according to the manufacturer’s instructions. Slides were coverslipped and analyzed using a fluorescence microscope.
Immunohistochemistry
Mice were transcardially perfused with saline, followed by 10% formalin. The brain sections were cut at 30 µm and collected into five consecutive series. For those that have received virus injection, one series of sections was used to check mCherry signals in the LHb under a fluorescence microscope, and only those with sufficient mCherry expression were included in the data analysis. To examine the apposition of LHb ChR2 fibers with 5-HT neurons in the DRN, one series of the sections that had received AAV-ChR2 delivery into the LHb were blocked (3% normal donkey serum) for 1 h, incubated with goat anti–5-HT antibody (1:5,000; 20079, Immunostar) on a shaker at 4° for overnight, followed by the donkey anti-goat Alexa Fluor 594 (1:200, #110382, Jackson ImmunoResearch) for 2 h. Slides were coverslipped and analyzed using a fluorescence microscope.
To examine TPH2 immunoreactivity, the brain sections were pretreated with 0.3% H2O2 for 30 min, blocked (3% normal goat serum) for 1 h, incubated with rabbit anti-TPH2 antibody (1:3,000; Ab111828, Abcam) on a shaker at room temperature for overnight, and biotinylated with goat anti-rabbit secondary antibody (1:1,000; BA-1000, Vector Laboratories) for 2 h. Sections were then incubated in the avidin–biotin complex (1:500, ABC; PK-6101, Vector Laboratories) and developed using the Vector VIP Substrate Kit (SK-4600, Vector Laboratories). After dehydration through graded ethanol, the slides were then immersed in xylene and coverslipped. Images were analyzed using a bright-field microscope. The numbers of TPH2 positive cells in the DRN were counted in 5–6 consecutive brain sections, and the average was used to reflect the data value for each mouse. At least three mice were included in each group for statistical analyses.
For light-induced c-fos immunostaining, WT mice or Nile grass rats were exposed to 300 lux room light from 6 to 6:30 P.M., and then they were kept in the dark until being perfused at 7 P.M. The controls were kept in the dark from 6 to 7 P.M. before perfusion. For CNO-induced c-fos expression, mCherry or hM3Dq mice received intraperitoneal injection of CNO (Cayman Chemical, #16882, 3 mg/kg) 1 h before perfusion (during 9–11 A.M.). The brain sections containing the LHb were pretreated with 0.3% H2O2 for 30 min, blocked (3% normal goat serum) for 1 h, incubated with mouse anti–c-fos antibody (1:2,000; ab208942, Abcam) on a shaker at room temperature for overnight, and biotinylated with goat anti-mouse secondary antibody (1:1,000; BA-9200, Vector Laboratories) for 2 h. Sections were then incubated in the avidin–biotin complex (1:500, ABC; PK-6101, Vector Laboratories) and incubated in 0.04% 3,3′-diaminobenzidine and 0.01% hydrogen peroxide. After dehydration through graded ethanol, the slides were then immersed in xylene and coverslipped. Images were analyzed using a bright-field microscope. The numbers of c-fos–positive cells in the LHb were counted in 6–7 consecutive brain sections and the average was used to reflect the data value for each mouse. Three mice were included in each group for statistical analyses.
For c-fos and TPH2 double immunostaining, the brain sections containing the DRN were pretreated with 0.3% H2O2 for 30 min, blocked (3% normal goat serum) for 1 h, and incubated with rabbit anti-TPH2 antibody (1:3,000; Ab111828, Abcam) and mouse anti–c-fos antibody (1:2,000; ab208942, Abcam) on a shaker at room temperature for overnight. Sections were then treated with the Double Staining Polymer Kit (MP-7724, Vector Laboratories) according to the manufacturer's instructions. Slides were coverslipped, and images were analyzed using a bright-field microscope. The numbers of c-fos–positive cells in 5-HTDRN neurons were counted in 5–6 consecutive brain sections, and the average was used to reflect the data value for each mouse. At least three mice were included in each group for statistical analyses.
Fiber photometry
Mice that received GCaMP6m injection into the LHb were allowed 3 weeks for recovery before acclimatization and investigator handling for 1 week before experiments. Mice were kept in a normal light/dark cycle (6 A.M.–6 P.M.) before light exposure, with the light illumination maintained at 300 lux in the housing room. All recordings were done in the home cage of singly housed experimental animals during the dark cycle (6–7 P.M.), and 300 lux room light was given for 2 s after recording the baseline signals for 5 min. Mice were allowed to adapt to the tethered patchcord for 2 d prior to experiments and given 5 min to acclimate to the tethered patchcord prior to any recording. Continuous <20 μW blue LED at 465 nm and UV LED at 405 nm served as excitation light sources, driven by a multichannel hub (Doric Lenses), modulated at 211 and 330 Hz, respectively. The light was delivered to a filtered minicube (FMC5, Doric Lenses) before connecting through optic fibers to a rotary joint (FRJ 1 × 1, Doric Lenses) to allow for movement. GCaMP6m calcium GFP signals and UV autofluorescent signals were collected through the same fibers back to the dichroic ports of the minicube into a femtowatt silicon photoreceiver (2151, Newport). The digital signals were then amplified, demodulated, and collected through a lock-in amplifier (RZ5P, Tucker-Davis Technologies). The fiber photometry data was collected using Synapse 2.0 (Tucker-Davis Technologies) and downsampled to 8 Hz. We derived the values of GCaMP6 fluorescence change (ΔF / F) by calculating (F465 − F0) / F0, where F0 is the baseline fluorescence of the F465 channel signal 2 s prior to the onset of light exposure. The F405 channel was used as an isosbestic fluorescence channel; we derived the values of isosbestic fluorescence change (ΔF / F0) by calculating (F405 − F0) / F0, where F0 is the baseline fluorescence of the F405 channel signal 2 s prior to the onset of light exposure.
Slice electrophysiology
Electrophysiology recordings were performed as previously described (He et al., 2021b). Mice were deeply anesthetized with isoflurane and transcardially perfused with a modified ice-cold sucrose-based cutting solution (pH 7.3) containing 10 mM NaCl, 25 mM NaHCO3, 195 mM sucrose, 5 mM glucose, 2.5 mM KCl, 1.25 mM NaH2PO4, 2 mM Na-pyruvate, 0.5 mM CaCl2, and 7 mM MgCl2, bubbled continuously with 95% O2 and 5% CO2. The mice were then decapitated, and the entire brain was removed and immediately submerged in the cutting solution. Coronal brain slices (220 mm) containing the DRN or LHb were cut with a Microm HM 650V vibratome (Thermo Fisher Scientific) in an oxygenated cutting solution. Slices were then incubated in oxygenated artificial CSF (aCSF; 126 mM NaCl, 2.5 mM KCl, 2.4 mM CaCl2, 1.2 mM NaH2PO4, 1.2 mM MgCl2, 11.1 mM glucose, and 21.4 mM NaHCO3, balanced with 95% O2/5% CO2, pH7.4) to recover ∼25 min at 32°C and subsequently for 1 h at room temperature before recording. Slices were transferred to a recording chamber and allowed to equilibrate for at least 10 min before recording. The slices were superfused at 32°C in oxygenated aCSF at a flow rate of 1.8–2 ml/min. tdTomato- or mCherry-labeled neurons were visualized using epifluorescence and IR-DIC imaging on an upright microscope (Eclipse FN1, Nikon) equipped with a movable stage (MP-285, Sutter Instrument). Patch pipettes with resistances of 3–5 MΩ were filled with intracellular solution (pH 7.3) containing 128 mM K-gluconate, 10 mM KCl, 10 mM HEPES, 0.1 mM EGTA, 2 mM MgCl2, 0.05 mM Na-GTP, and 4 mM Mg-ATP. Recordings were made using a MultiClamp 700B amplifier (Axon Instruments), sampled using Digidata 1440A, and analyzed offline with pCLAMP 10.3 software (Axon Instruments). Series resistance was monitored during the recording, and the values were generally <10 MΩ and were not compensated. The liquid junction potential was +12.5 mV and was corrected after the experiment. Data were excluded if the series resistance increased dramatically during the experiment or without overshoot for the action potential. Currents were amplified, filtered at 1 kHz, and digitized at 20 kHz. The current clamp was engaged to test neural firing frequency and resting membrane potential at the baseline or in response to CNO (Cayman Chemical, #16882, 10 µM). A neuron was considered depolarized or hyperpolarized if a change in membrane potential was at least 2 mV in amplitude.
Serum corticosteroid measurement
Mice were kept in the dark or exposed to white light (300 lux) for 4 h during 6–10 P.M., and then blood was collected and processed to measure serum corticosteroid levels using the mouse corticosterone ELISA kit (ADI-900-097, Enzo Life Sciences).
Conditioned place preference test
The conditioned place preference test was performed in a dual-zone chamber divided by a removable barrier, with white walls in one zone and black and white striped walls in the other zone. During the 5 d of conditioning sessions, mice received saline injection 30 min before they were confined to the white wall zone for 30 min in the morning and CNO injection 30 min before they were confined to the striped wall zone for 30 min in the afternoon. On the final testing day, mice were given 10 min of free access across the chamber to determine their conditioned place preference.
Elevated plus maze test
The apparatus for the EPM consisted of two opposing open arms (24 cm length × 12 cm width) and two closed arms (24 cm length × 12 cm width), which were connected by a central zone (6 cm length × 6 cm width). The whole apparatus was elevated 50 cm above the floor. Mice received CNO injection 0.5 h before they were subjected to the test. The mouse was placed in the center zone facing toward one open arm and was allowed to freely explore the arena for 8 min. Time spent in the open arm during the last 6 min was measured (EthoVision XT software). The box was wiped clean with a paper towel soaked in 70% ethanol and dried thoroughly after each test session.
Open field test
Mice received CNO injection 0.5 h before they were subjected to the test. Mice were placed in the center of an open field box (42 cm length × 42 cm width × 30 cm height) in a room with dim light and were allowed to explore the arena for 10 min. All animal activity was recorded with a camera placed above the box. Time spent in the center during the last 8 min was measured (EthoVision XT software). The box was wiped clean with a paper towel soaked in 70% ethanol and dried thoroughly after each test session.
Forced swimming test
Mice received CNO injection 0.5 h before they were subjected to the test. Mice were placed in a cylinder of water (temperature of 23–25°C; 20 cm in diameter, 27 cm in height for mice) for 6 min. The depth of water was set to prevent animals from touching the bottom with their hind limbs. Animal behavior was recorded from the side. The time each animal spent immobile during the last 4 min of the test was counted. Immobility was defined as no active movement except that needed to keep the animal from drowning.
Tail suspension test
Mice received CNO injection 0.5 h before they were subjected to the test. Each mouse was suspended above the ground at approximately one-third from the end of the tail. Tape is used to secure the mouse to the wall. Animal behavior was recorded with a camera for 6 min. The time each animal spent immobile during the last 4 min of the test was counted. Immobility was defined as no active movement.
Human studies
Fifty-three (male, 42; female, 11) healthy Han Chinese were recruited and interviewed by trained investigators using a detailed questionnaire including general information, sociodemographic characteristics, and psychological conditions. None of the participants had any acute or chronic physical diseases or a family history of psychiatric disorders. Other exclusion criteria included pregnancy or breast diseases, cancer, infections, allergies, memory or any other cognitive impairment, visual impairment, anti-inflammatory and immunosuppressive therapy, and history of systemic, endocrine, immune, or cerebrovascular diseases. The participants were asked to fast overnight before the day of the experiment. On the experimental day, the participants came to Wuhan Mental Health Center (Wuhan, China), and their height, weight, waist circumference, and vital signs were recorded upon arrival. Then they were asked to have a free breakfast alone in the experiment room, which consisted of two windows, three lights, an entrance, and a dining area. The same room was used for all participants individually when the room was randomly set with a strong light environment (windows open and lights turned on; light intensity, 1,677 lux) or dim light environment (windows closed and light turned off; light intensity, 100 lux). Twenty-six individuals (21 men and 5 women; age range, 18–48 years; mean ± SEM, 30.85 ± 1.57) ate in the dim environment, while 27 individuals (21 men and 6 women; age range, 18–48 years; mean ± SEM, 30.19 ± 1.61) ate in the strong light environment. The breakfast included bread and milk, which were purchased from a local grocery store, with known nutrient ingredients and calories per unit weight. The participants could ask for more bread or milk until they felt full. The amounts of bread and milk consumed by each participant were measured to calculate their nutrient intake.
Statistics
The minimal sample size was predetermined by the nature of the experiments. For food intake measurements, 6–9 different mice or rats per group were included. For histology studies, the same experiment was repeated in at least three different mice. For electrophysiological studies, at least 10 different neurons from three different mice were included. The data are presented as mean ± SEM or as individual data points. Statistical analyses were performed using GraphPad Prism to evaluate normal distribution and variations within and among groups. Methods of statistical analyses were chosen based on the design of each experiment and are indicated in figure legends. p < 0.05 was considered to be statistically significant.
Study approval
Care of all animals and procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee. The human studies were approved by the Ethics Committee of Wuhan Mental Health Center.
Results
Feeding behavior is regulated by light/darkness independent of the circadian clock
To determine whether the existence of light or darkness affects feeding behavior in nocturnal mice, we exposed wild-type (WT) C57Bl6j mice to “light extension” or “darkness extension” paradigms at various times of the day and compared their food intake during these periods to the same time periods under a normal light/dark cycle. Interestingly, a 4 h light extension after the end of the light cycle (6–10 P.M.) significantly suppressed food intake (Fig. 1A, t = 2.577, dF = 14, p = 0.0366). In addition, a 4 h light extension before the onset of the light cycle (2–6 A.M.) induced a similar anorexigenic response (Fig. 1B, t = 5.044, dF = 8, p = 0.001). In contrast, darkness extension after the end (6–10 A.M.) or before the start (2–6 P.M.) of the dark cycle evoked hyperphagia (Fig. 1C, t = 3.02, dF = 11, p = 0.0116; Fig. 1D, t = 3.52, dF = 7, p = 0.0097). To test the effects of light on feeding in food-deprived mice, we combined the fasting-induced refeeding and the light/darkness extension paradigms. Notably, the hunger-driven feeding was similarly suppressed by light extension but potentiated by darkness extension (Fig. 1E, t = 2.295, dF = 16, p = 0.0356; Fig. 1F, t = 2.54, dF = 16, p = 0.0219). We also adopted the HomeCageScan to monitor other behaviors, including locomotor activity, grooming, and sleeping, in response to light or darkness extension. The 4 h light extension decreased eating times and traveling distance without evidently changing the duration of grooming, sleeping, and awakening (Fig. 2A, t = 3.852, dF = 7, p = 0.0063; Fig. 2B, t = 2.678, dF = 7, p = 0.0316; Fig. 2C, t = 0.7721, dF = 7, p = 0.4653; Fig. 2D, t = 0.1328, dF = 7, p = 0.8981; Fig. 2E, t = 0.1735, dF = 7, p = 0.8762; Fig. 2F, t = 1.948, dF = 14, p = 0.0718; Fig. 2G, t = 1.464, dF = 7, p = 0.1866; Fig. 2H, dF = 7, t = 0.7239, p = 0.4926; Fig. 2I, t = 0.5581, dF = 7, p = 0.5942; Fig. 2J, t = 0.6493, dF = 7, p = 0.5369), while the 4 h darkness extension produced the opposite effects (Fig. 2K, t = 2.676, dF = 7, p = 0.0317; Fig. 2L, t = 2.963, dF = 7, p = 0.021; Fig. 2M, t = 1.699, dF = 7, p = 0.1331; Fig. 2N, t = 2.37, dF = 7, p = 0.0496; Fig. 2O, t = 0.1472, dF = 7, p = 0.8871; Fig. 2P, t = 2.15, dF = 7, p = 0.0686; Fig. 2Q, t = 2.235, dF = 7, p = 0.0605; Fig. 2R, t = 2.168, dF = 7, p = 0.0668; Fig. 2S, t = 1.405, dF = 7, p = 0.2029; Fig. 2T, t = 1.22, dF = 7, p = 0.262). To exclude the possibility that light-induced anorexia is caused by stress instead of the light illumination itself, we checked the serum corticosterone level and found that a 4 h light exposure at the dark cycle (6–10 P.M.) had no effect on corticosterone expression (Fig. 2U, t = 0.04579, dF = 9, p = 0.9645).
Light decreases feeding in mice. A, B, Food intake during 6–10 P.M. (A, n = 8) or 2–6 A.M. (B, n = 8) in the absence or presence of light (300 lux). C, D, Food intake during 6–10 A.M. (C, n = 12) or 2–6 P.M. (D, n = 8) in the presence or absence of light (300 lux). E, Fasting-induced refeeding during 6–8 P.M. with or without light exposure (300 lux, n = 9 per group). F, Fasting-induced refeeding during 6–8 A.M. with or without light exposure (300 lux, n = 9 per group). Data are expressed as mean ± SEM and individual data points. Paired (A–D) or unpaired (E, F) Student's t tests. *p < 0.05, **p < 0.01, ***p < 0.001.
Light manipulation affects eating and locomotion without significantly altering grooming and sleeping. A–E, HomeCageScan revealing the number of eating times (A), travel distance (B), duration for grooming (C), sleeping (D), and awakening (E) in mice during 6–10 P.M. in the absence or presence of light (300 lux, n = 8). F–J, HomeCageScan revealing the number of eating times (F), travel distance (G), duration for grooming (H), sleeping (I), and awakening (J) in mice during 2–6 A.M. in the absence or presence of light (300 lux, n = 8). K–O, HomeCageScan revealing the number of eating times (K), travel distance (L), duration for grooming (M), sleeping (N), and awakening (O) in mice during 2–6 P.M. with overhead lights on or off (300 lux, n = 8). P–T, HomeCageScan revealing the number of eating times (P), travel distance (Q), duration for grooming (R), sleeping (S), and awakening (T) in mice during 6–10 A.M. with overhead lights on or off (300 lux, n = 8). U, Serum corticosterone levels after light exposure (300 lux) from 6 to 10 P.M. V, Double-plotted actogram showing wheel-running activities of WT mice that were constantly exposed to light (300 lux) for 2 weeks. Data are expressed as mean ± SEM and individual data points. Paired Student's t tests (A–T). Unpaired Student's t tests (U) *p < 0.05, **p < 0.01.
To exclude the possibility that the alteration in food intake might be influenced by the endogenous clock instead of light/darkness per se, we subjected mice to a 2-week constant light exposure to disrupt their circadian rhythm (Fig. 2V) and then exposed these arrhythmic mice to the darkness. Again, the darkness exposure in either the early morning or the late afternoon increased food intake in these mice (Fig. 3A, t = 3.62, dF = 11, p = 0.004; Fig. 3B, t = 2.848, dF = 11, p = 0.0158). Additionally, the amount of refeeding after an overnight fast was increased during the darkness compared with refeeding during the light exposure (Fig. 3C, t = 2.281, dF = 10, p = 0.0457). Together, these results indicate that darkness induces hyperphagia in a circadian-independent manner.
Light decreases feeding independent of the circadian clock in mice. A, B, Food intake during 2–6 P.M. (A) or 6–10 A.M. (B) with or without light exposure (300 lux) in mice after a 2-week constant light exposure (300 lux, n = 12). C, Fasting-induced refeeding during 6–8 A.M. with or without light exposure (300 lux) in mice after a 2-week constant light exposure (300 lux, n = 6 per group). Data are expressed as mean ± SEM and individual data points. Paired (A, B) or unpaired (C) Student's t tests. *p < 0.05, **p < 0.01.
An LHb→DRN circuit mediates light-induced anorexia
Next, we sought to unravel the central mechanisms that may account for light-induced suppression in feeding. Previous research has established that light treatment plays a pivotal role in modulating the activation of the lateral habenula (LHb; T. Qu et al., 1996; Hattar et al., 2006; Huang et al., 2019). Therefore, we checked the LHb neuronal activity after light exposure at the dark cycle through c-fos immunostaining. Light exposure significantly increased c-fos expression in the LHb in WT mice (Fig. 4A, t = 5.753, dF = 8, p = 0.0004), suggesting that LHb neuronal activity was elevated by light stimulation. Since the LHb is predominated by glutamatergic neurons (Hashikawa et al., 2020), we delivered a CaMKII-ChR2 (H134R)-GFP virus into the LHb of WT mice to specifically induce ChR2 expression in LHb glutamatergic neurons, which allowed us to identify potential downstream targets (Fig. 4B). We observed dense GFP-labeled cell bodies in the LHb and fibers in the DRN, the nucleus accumbens shell (AcbSh), the preoptic area (POA), the zona incerta (ZI), the peduncular part of the lateral hypothalamus (PLH), the ventral tegmental area (VTA), and the dorsomedial hypothalamic nucleus (DMH; Fig. 4B,C, Extended Data Fig. 4-1A–F). Given the well-established role of the DRN in feeding regulation, we focused on the LHb→DRN circuit. By delivering a retrograde GFP virus into the DRN of WT mice, we further confirmed the connections between the LHb and the DRN, as revealed by abundant GFP+ neurons in the LHb (Extended Data Fig. 4-1G–I). Consistent with earlier reports (Ren et al., 2019), we also observed GFP+ cells in the lateral septal (LS), the bed nucleus of the stria terminalis (BNST), the POA, the lateral hypothalamus (LH), the medial amygdala (MeA), and the VTA (Extended Data Fig. 4-1J–L).
An LHb→DRN circuit mediates light-induced anorexia. A, Representative c-fos immunoreactivity and quantification in the LHb from mice in darkness or after 1 h light exposure (300 lux, n = 5 per group). B, Scheme for AAV-CaMKII-ChR2-GFP virus injection into the LHb of WT mice and ChR2-GFP-labelled cell bodies in the LHb. C, ChR2-GFP–labeled fibers/terminals in the DRN. D, Scheme for GCaMP6m expression in DRN-projecting LHb neurons. E, GCaMP6m signals in DRN-projecting LHb neurons in response to light stimulation (300 lux, n = 6) and changes in calcium fluorescence of DRN-projecting LHb neurons from six individual mice. Time 0 marks the start of light exposure (300 lux). F, Scheme for specific infection of DRN-projecting LHb neurons with hM3Dq or mCherry and a representative image showing hM3Dq expression in the LHb. G–I, Typical action potential traces (G), firing frequency (H), and resting membrane potential (I) of DRN-projecting LHb neurons expressing hM3Dq in response to 10 µM CNO. J, Food intake during 2–6 P.M. in saline-treated mCherry or hM3Dq mice with overhead lights on or off (300 lux, n = 7 per group). K, Food intake during 2–6 P.M. in CNO-treated mCherry or hM3Dq mice with overhead lights on or off (300 lux, n = 7 per group). L, Food intake during 6–8 P.M. in saline- or CNO-treated mCherry and hM3Dq mice (n = 7 per group). M, N, Traveling distance (M) and time spent in the open arm (N) during EPM test in mCherry and hM3Dq mice after CNO treatment (n = 7 per group). O, P, Traveling distance (O) and time spent in the center (P) during the open field test in mCherry and hM3Dq mice after CNO treatment (n = 7 per group). Q, The immobility time during forced swimming test in mCherry and hM3Dq mice after CNO treatment (n = 7 per group). R, The immobility time during tail suspension test in mCherry and hM3Dq mice after CNO treatment (n = 7 per group). S, Time spent in CNO-paired side in mCherry and hM3Dq mice after conditioning (n = 7 per group). T, Scheme for specific infection of DRN-projecting LHb neurons with hM4Di or mCherry and a representative image showing hM4Di expression in the LHb. U, Food intake during 6–10 P.M. in saline-treated mCherry or hM4Di mice with overhead lights on or off (300 lux, n = 7 per group). V, Food intake during 6–10 P.M. in CNO-treated mCherry or hM4Di mice with overhead lights on or off (300 lux, n = 7 per group). W, X, Traveling distance (W) and time spent in the open arm (X) during the EPM test in mCherry and hM4Di mice after CNO treatment (n = 7 per group). Y, Z, Traveling distance (Y) and time spent in the center (Z) during the open field test in mCherry and hM4Di mice after CNO treatment (n = 7 per group). Data are expressed as mean ± SEM and individual data points. Two-way ANOVA with Tukey's test (J–L, U, V) or unpaired Student's t test (A, M–S, W–Z) or paired Student's t tests (H, I). *p < 0.05, **p < 0.01, ***p < 0.001 (see Extended Data Figs. 4-1 and 4-2 for more details).
Figure 4-1
LHb neurons project to the DRN. A. CamKII-ChR2 expression in the LHb. B-F. LHb glutamatergic neurons-originated fibers in the AcbSh (B), POA (C), ZI and PLH (D), VTA (E) and DMH (F). G. Scheme for retrograde tracing from the DRN. H. Retrograde virus infection in the DRN. I-L. Retrograde GFP expression in the LHb (I), LS/BNST/POA (J), LH/MeA (K), and VTA (L). M. GCaMP6 m expression in DRN-projecting LHb neurons. N. hM3Dq-mCherry expression in DRN-projecting LHb neurons. O. hM4Di-mCherry expression in DRN-projecting LHb neurons. Download Figure 4-1, TIF file.
Figure 4-2
Upstream inputs of DRN-projecting LHb neurons. A. Scheme for rabies tracing from DRN-projecting LHb neurons. B-G. Rabies-labelled GFP in the LHb (B-C), MPA (D), LPO (E), PLH (F) and EP (G). Download Figure 4-2, TIF file.
To test whether the DRN-projecting LHb neurons are responsive to light, we delivered a retrograde Cre virus into the DRN, and a Cre-dependent GCaMP6m virus into the LHb (Fig. 4D). Through this approach, we successfully introduced GcaMP6m expression into DRN-projecting LHb neurons (Extended Data Fig. 4-1M), thereby facilitating the real-time tracking of calcium activity within this specific neuronal population. Fiber photometry recordings showed that the Ca2+ signals were increased in response to a 2 s 300 lux light exposure (Fig. 4E), indicating that DRN-projecting LHb neurons are excited by light.
To uncover the functional contributions of the LHb→DRN circuit to light-induced alterations in feeding, we injected a retrograde Cre virus into the DRN and infected the LHb with a Cre-dependent virus encoding the neuronal activator DREADD hM3Dq (AAV-DIO-hM3Dq-mCherry) or mCherry (AAV-DIO-mCherry; Fig. 4F, Extended Data Fig. 4-1N), which enabled us to use CNO to specifically activate the LHb→DRN circuit in hM3Dq but not mCherry mice. CNO treatment depolarized LHb neurons that were infected with hM3Dq, as demonstrated by elevated resting membrane potential and increased firing rate (Fig. 4G–I, t = 3.66, dF = 16, p = 0.0045; Extended Data Fig. 4-2C, t = 3.62, dF = 16, p = 0.0051). As expected, extended darkness can reliably increase food intake in saline-treated mCherry and hM3Dq mice (Fig. 4J; main effect of darkness, F(1,24) = 7.719, p = 0.0104). Similarly, CNO-injected mCherry mice showed expected hyperphagia when exposed to extended darkness (Fig. 4K; main effect of darkness, F(1,24) = 5.699, p = 0.0252). However, in CNO-injected hM3Dq mice, extended darkness failed to increase food intake (Fig. 4K; main effect of darkness, F(1,24) = 0.1895, p = 0.6672), indicating that activation of the LHb→DRN circuit prevents darkness-induced hyperphagia. Notably, in mice under the normal 12 h light/dark cycle, CNO administration also decreased food intake during the first 2 h of the dark cycle (6–8 P.M.) in hM3Dq mice but not in mCherry mice (Fig. 4L; main effect of CNO, F(1,20) = 5.391, p = 0.0309), suggesting that inhibition of the LHb→DRN circuit is required to permit a normal feeding behavior during the early phase of the dark cycle.
The LHb was reported to play an important role in regulating emotional status, and we therefore employed the elevated plus maze (EPM) and open field test (OFT) to examine the anxiety-like behavior. Activation of the LHb→DRN circuit failed to significantly change locomotor activity and the time spent in the open arms in EPM test (Fig. 4M, t = 1.446, dF = 12, p = 0.1738; Fig. 4N, t = 0.8443, dF = 11, p = 0.4165) and the time spent in the center in the open field (Fig. 4O, t = 1.74, dF = 12, p = 0.1074; Fig. 4P, t = 1.167, dF = 12, p = 0.266), suggesting that activation of this circuit did not induce anxiety-like behavior. Moreover, we performed the forced swimming test (FST) and tail suspension test (TST) to evaluate the depressive-like behaviors. The immobility time during FST and TST was not significantly different between CNO-treated mCherry and hM3Dq mice (Fig. 4Q, t = 0.8952, dF = 12, p = 0.3883; Fig. 4R, t = 0.572, dF = 12, p = 0.5779), demonstrating that acute activation of the LHb→DRN circuit yields no discernible effects on depressive symptoms. We further subjected these mice to the conditioned place preference test and found that activation of the LHb→DRN circuit had no effects on animals’ place preference (Fig. 4S, t = 0.6817, dF = 12, p = 0.5084). Collectively, these findings indicate that the short-term manipulation of the LHb→DRN circuit did not induce any noticeable changes in the affective aspects.
To further confirm the involvement of the LHb→DRN circuit in light-induced anorexia, we delivered a retrograde AAV carrying Cre into the DRN and a Cre-dependent AAV expressing the neuronal inhibitor hM4Di into the LHb (Fig. 4T, Extended Data Fig. 4-1O), which enabled us to use CNO to inhibit DRN-projecting LHb neurons (hM4Di mice). Mice that received Cre-dependent AAV expressing mCherry into the LHb served as controls (mCherry mice). After an initial injection of saline, both groups exhibited a decrease in food intake following exposure to light during the dark cycle (6–10 P.M.), confirming that light exposure suppresses feeding behavior (Fig. 4U; main effect of light exposure, F(1,22) = 18.39, p = 0.0003). However, when CNO was administered, the hM4Di mice showed a blockade of light-induced anorexia. In contrast, the mCherry mice continued to exhibit decreased food intake upon light exposure (Fig. 4V; main effect of light exposure, F(1,22) = 3.476, p = 0.0757). Thus, these results indicate that the activation of DRN-projecting LHb neurons is required for light-elicited suppression of feeding during the dark period. Notably, when the LHb→DRN circuit was inhibited, there were no significant alterations observed in locomotor activity and the time spent in the open arms during the EPM test (Fig. 4W, t = 0.6959, dF = 12, p = 0.4998; Fig. 4X, t = 0.7735, dF = 12, p = 0.4542). Similarly, there were no notable changes in locomotor activity and the time spent in the center during the open field test (Fig. 4Y, t = 0.5642, dF = 12, p = 0.583; Fig. 4Z, t = 0.4226, dF = 12, p = 0.68), suggesting that inhibition of this circuit did not have a discernible impact on locomotion and anxiety-like behavior.
To identify the upstream inputs of DRN-projecting LHb neurons, we employed a trans-synaptic tracing method based on a modified rabies virus (Beier et al., 2015). Briefly, the DRN was infected with a retrograde Cre virus, while the LHb received the Cre-dependent helper virus (AAV-DIO-GTB) expressing the TVA receptor and rabies glycoprotein, which are required for the replication of the rabies virus. Three weeks later, we injected the EnvA-ΔG-Rabies-GFP virus into the LHb to infect helper+ DRN-projecting LHb neurons (Extended Data Fig. 4-2A–C). One week following rabies injection, we detected GFP+ neurons in the medial preoptic area (MPA), the lateral preoptic area (LPO), the LH, and the entopeduncular nucleus (EP; Extended Data Fig. 4-2D–G).
5-HTDRN neurons receive excitatory inputs from the LHb
To investigate the connection between the LHb and 5-HT neurons in the DRN (5-HTDRN neurons), brain sections from mice infected with the CaMKII-ChR2 (H134R)-GFP virus within the LHb were subjected to 5-HT staining. Our observations revealed that 53.2% of 5-HT neurons within the DRN are in close apposition to fibers/terminals originating from the LHb (Fig. 5A), suggesting that 5-HTDRN neurons are innervated by the LHb. To validate the glutamatergic nature of DRN-projecting LHb neurons, we selectively labeled these neurons with mCherry by administrating a retrograde Cre virus into the DRN and a Cre-dependent mCherry virus into the LHb (Fig. 5B). The triple RNAscope for mCherry, vGLUT2, and vGAT revealed that all the mCherry+ neurons express vGLUT2 but not vGAT (Fig. 5C), confirming that the DRN-projecting LHb neurons are glutamatergic.
5-HTDRN neurons receive excitatory inputs from the LHb. A, Immunofluorescence images showing the close localization of LHb-originated fibers and 5-HT neurons in the DRN. B, Scheme for the labeling of DRN-projecting LHb neurons with mCherry. C, RNAscope showing mCherry, vGLUT2, and vGAT in the LHb. D, Scheme for AAV-CaMKII-ChR2-GFP virus injection into the LHb of TPH2-CreER/Rosa26-LSL-tdTomato mice. E, Light-evoked EPSCs in 5-HT neurons in the presence or absence of DNQX (n = 10 neurons from 3 mice). F, The percentage of 5-HTDRN neurons in response to blue light stimulation (n = 10 neurons from 3 mice). G, H, Representative c-fos immunoreactivity (G) and quantification for the number of c-fos+ cells (H) in the LHb from mCherry (n = 6) and hM3Dq (n = 5) mice after CNO treatment. I, J, Double staining for TPH2 (red) and c-fos (brown) in the DRN of mCherry (n = 6) and hM3Dq (n = 5) mice after CNO treatment (I) and quantification for the number of c-fos+ cells in 5-HT neurons (J). Data are expressed as mean ± SEM and individual data points. Unpaired Student's t test (H, J). ***p < 0.001.
To further explore the functional connections between the LHb and 5-HTDRN neurons, we injected the CaMKII-ChR2 (H134R)-GFP virus into the LHb of TPH2-CreER/Rosa26-LSL-tdTomato mice (Fig. 5D), in which 5-HT neurons were labeled by tdTomato upon tamoxifen injection. Slice electrophysiological recordings showed that blue light stimulation can trigger excitatory postsynaptic currents (EPSCs) in 7 out of 10 (70%) tdTomato+ neurons, which was blocked by the application of the AMPA receptor antagonist DNQX (Fig. 5E,F). The evoked EPSCs persisted in the presence of TTX and 4-AP (Fig. 5E), indicating that LHb glutamatergic neurons provide monosynaptic inputs to 5-HTDRN neurons. To test whether activation of the LHb can elevate the activity of 5-HTDRN neurons, we introduced hM3Dq or mCherry constructs into DRN-projecting LHb neurons (hM3Dq or mCherry mice, as described in Fig. 4I). As anticipated, the administration of CNO led to an augmentation of c-fos expression specifically in hM3Dq mice, with no such effect observed in mCherry controls (Fig. 5G,H, t = 5.683, dF = 9, p = 0.0003). Importantly, a noticeable increase in c-fos expression was observed in 5-HTDRN neurons of hM3Dq mice compared with their mCherry counterparts (Fig. 5I,J, t = 8.166, dF = 9, p < 0.0001). Together, these results signify the reception of excitatory inputs from the LHb by 5-HTDRN neurons.
LHb→5-HTDRN circuit mediates light-induced anorexia
To investigate the possible involvement of 5-HTDRN neurons in light-induced suppression in feeding, we checked c-fos expression in 5-HTDRN neurons after light exposure, which showed an increase (Fig. 6A,B, t = 11.44, dF = 4, p = 0.0003). Consequently, we generated TPH2DRN-KO mice and their controls by delivering AAV-Cre or AAV-GFP virus into the DRN of TPH2 flox/flox mice to explore the role of TPH2, a rate-limiting enzyme in 5-HT synthesis, in mediating the suppressive effect of light on feeding (Fig. 6C–E, t = 22.79, dF = 14, p < 0.0001). Although light extension at the dark phase decreased food intake in both control and TPH2DRN-KO mice, such effects were significantly attenuated in TPH2DRN-KO mice (Fig. 6F; main effect of light exposure, F(1,24) = 10.51, p = 0.0035), demonstrating that 5-HT signals are at least partially required for light-induced anorexia in mice.
An LHb→5-HTDRN circuit mediates light-induced anorexia. A, B, Representative fluorescence images (A) and quantification (B) showing c-fos and TPH2 expression in the DRN of male wild-type mice after 1 h light exposure (300 lux, n = 3 per group). C, Scheme for AAV-Cre-GFP virus injection into the DRN of TPH2 flox/flox mice. D, E, TPH2 staining (D) and quantification (E) in the DRN of control and TPH2DRN-KO mice (n = 8 per group). F, Food intake during 6–10 P.M. in control (n = 7) and TPH2DRN-KO (n = 7) mice with or without light exposure (300 lux). G, H, TPH2 staining (G) and quantification of TPH2+ cells (H) in the DRN of control and TPH2DRN-KO mice (n = 7 per group). I, Scheme for the creation of control and TPH2DRN-KO mice with specific expression of hM3Dq or mCherry in DRN-projecting LHb neurons. J, Flp-dependent hM3Dq expression in the LHb. K–M, Representative c-fos immunoreactivity (K, L) and quantification (M) in the LHb from mCherry and hM3Dq mice after CNO treatment (n = 7 per group). N–P, Typical action potential traces (N), firing frequency (O), and resting membrane potential (P) of DRN-projecting LHb neurons expressing Flp-dependent hM3Dq in response to 10 µM CNO (n = 6 neurons). Q, Food intake during 2–6 P.M. in saline-treated control and TPH2DRN-KO mice that were infected with mCherry or hM3Dq, in the presence or absence of light (300 lux, n = 7 per group). R, Food intake during 2–6 P.M. in CNO-treated control and TPH2DRN-KO mice that were infected with mCherry or hM3Dq, in the presence or absence of light (300 lux, n = 7 per group). S, Food intake during 6–8 P.M. in saline- or CNO-treated control and TPH2DRN-KO mice infected with hM3Dq (n = 7 per group). Data are expressed as mean ± SEM and individual data points. Two-way ANOVA with Tukey's test (F, Q–S) or paired Student's t test (O, P) or unpaired Student's t test (B, E, H, M). #p < 0.05 for the effect of CNO in control versus TPH2DRN-KO mice. *p < 0.05, **p < 0.01, ***p < 0.001.
To specifically delineate the contribution of 5-HTDRN neurons to the effects of the LHb→DRN circuit on feeding, we manipulated the LHb→DRN circuit in control and TPH2DRN-KO mice (Fig. 6G,H, t = 17.34, dF = 12, p < 0.0001). To this end, the AAV-GFP or AAV-Cre virus, together with the retrograde AAV-Flpo, was administrated into the DRN. Meanwhile, a Flpo-dependent AAV encoding DREADD hM3Dq (AAV-fDIO-hM3Dq-mCherry) or mCherry (AAV-fDIO-mCherry) was delivered into the LHb (Fig. 6I,J). By employing this method, we gained the capability to modulate the activity of the LHb→DRN circuit in both control and TPH2DRN-KO mice. c-fos immunostaining and slice recordings confirmed the efficiency of the Flpo-dependent hM3Dq virus to activate infected neurons (Fig. 6K–P, dF = 12, t = 6.413, p < 0.0001). Under the saline-treated condition, extended darkness enhanced food intake in the mCherry and hM3Dq groups for both control and TPH2DRN-KO mice (Fig. 6Q; main effect of darkness, F(1,48) = 39.95, p < 0.0001). Consistent with the early result, CNO-induced activation of the LHb→DRN circuit blunted the darkness-induced hyperphagia in control mice (Fig. 6R). However, this inhibition was not observed in TPH2DRN-KO mice (Fig. 6R; main effect of TPH2DRN-KO, F(1,48) = 10.51, p = 0.0042). Thus, serotoninergic signals from DRN neurons are required to mediate the effects of the LHb-originated projections to prevent darkness-induced hyperphagia. Similarly, in mice under the normal 12 h light/dark cycle, activation of the LHb→DRN circuit decreased dark cycle food consumption in both control and TPH2DRN-KO mice, although the reduction was attenuated in TPH2DRN-KO mice (Fig. 6S; main effect of TPH2DRN-KO, F(1,48) = 20.24, p < 0.0001).
Light inhibits feeding in diurnal Nile grass rats and humans
We used the diurnal Nile grass rats to determine whether light also affects feeding in diurnal animals, which consume most of their daily food during the light period. First, we showed that Nile grass rats consumed significantly more food during the light cycle than the dark cycle, confirming the diurnal property of this species (Fig. 7A, t = 3.02, dF = 10, p = 0.0129). We then monitored 1 h food intake (6–7 A.M.) in the Nile grass rats under strong (3,000 lux), normal (300 lux), and dim (50 lux) light housing conditions. Compared to those under normal light, strong light decreased food intake while dim light increased feeding (Fig. 7B; main effect of light, F(1.121,5.607) = 10.85, p = 0.017). In addition, light exposure significantly increased the expression of c-fos in the LHb and 5-HTDRN neurons in Nile grass rats (Fig. 7C–E, t = 4.91, dF = 4, p = 0.008; Fig. 7F, t = 3.145, dF = 4, p = 0.0347). These results indicate that light activates LHb neurons and 5-HTDRN neurons in these diurnal rats similarly as it does in nocturnal mice. Finally, we showed that i.p. injections of a selective 5-HT 2C receptor agonist, lorcaserin, significantly reduced food intake in both Nile grass rats and mice (Fig. 7G, t = 4.842, dF = 16, p = 0.0002; Fig. 7H, t = 2.28, dF = 10, p = 0.0458), indicating that the anorexigenic effects of 5-HT signals are conserved in these two species.
Light suppresses feeding in diurnal Nile grass rats. A, Food intake at the onset of light (6–9 A.M.) or dark (6–9 P.M.) cycle in Nile grass rats (n = 6). B, Food intake during 6–7 A.M. in Nile grass rats under strong (3,000 lux), normal (300 lux), and dim (50 lux) light conditions after an overnight fasting (n = 6). C, Representative c-fos immunoreactivity in the LHb from Nile grass rats perfused under the dark condition or 30 min after light exposure (3,000 lux). D, Representative TPH2 (red) and c-fos (brown) immunoreactivity in the DRN of the same Nile grass rats described in C. E, Quantification for the number of c-fos+ cells in the LHb (n = 3 per group). F, Quantification for the number of 5-HT neurons that are c-fos+ (n = 3 per group). G, H, Food intake after saline or lorcaserin administration in WT mice (n = 9 per group, G) or Nile grass rats (n = 6 per group, H). Data are expressed as mean ± SEM and individual data points. Unpaired Student's t tests (A, E–H) or one-way ANOVA (B). *p < 0.05, **p < 0.01, ***p < 0.001.
Finally, we recruited a group of healthy Han Chinese individuals, with similar height, weight, body mass index, and waist circumference (Fig. 8A, t = 0.8875, dF = 51, p = 0.379; Fig. 8B, t = 0.2676, dF = 51, p = 0.7901; Fig. 8C, t = 0.6911, dF = 51, p = 0.4926; Fig. 8D, t = 1.501, dF = 51, p = 0.1394), and randomly assigned them to have breakfast at the strong light or dim light condition. Compared to those eating with normal light illumination, those in dim environment consumed significantly higher total energy (Fig. 8E, t = 2.423, dF = 51, p = 0.019), associated with increases in protein (Fig. 8F, t = 3.137, dF = 51, p = 0.0028) and fat intake (Fig. 6G, t = 3.061, dF = 51, p = 0.0035) and a trended increase in carbohydrate consumption (Fig. 6H, t = 1.628, dF = 51, p = 0.1096). Accordingly, these data support the notion that light illumination can suppress food consumption in humans.
Light suppresses feeding in humans. A–D, Height (A), weight (B), BMI (C), and waist circumference (D) of the recruited individuals eating in normal or dim light environments. E–H, Total energy (E), Protein (F), fat (G), and carbohydrate (H) consumed by humans under normal (1,677 lux) or dim (100 lux) light illumination states. Data are expressed as mean ± SEM and individual data points. Unpaired Student's t tests (A–H). *p < 0.05, **p < 0.01.
Discussion
Here, we demonstrated that light exposure suppresses feeding, while darkness promotes increased feeding in mice, regardless of the circadian rhythmicity. Through an array of transneuronal virus tracing techniques, fiber photometry, and chemogenetic approaches, we delineated a neural pathway connecting the LHb to 5-HTDRN neurons in mice. Activation of the LHb→5-HTDRN circuit effectively mitigates the hyperphagia induced by darkness, whereas inhibiting this pathway prevents the light-induced reduction in appetite in mice. Together, we identified an LHb→5-HTDRN pathway that transmits light signals to regulate feeding behavior in mice.
Light is a powerful modulator of many biological activities, including mood, locomotion, sleep, arousal, and glucose metabolism (Vandewalle et al., 2009; LeGates et al., 2014; Meng et al., 2023). Dark cycle light exposure is also relevant to eating abnormalities (Fonken et al., 2010). However, chronic light stimulation often leads to a shift in circadian rhythmicity, which is closely associated with metabolic disruptions (Rudic et al., 2004; Turek et al., 2005; Kettner et al., 2015). For instance, deletion of the circadian clock genes, such as Per1, Bmal1, and Cry1, leads to eating irregularity and energy imbalance in mice (Adamovich et al., 2014; Kettner et al., 2015). Therefore, the altered feeding behavior observed in previous studies that adopted long-term light exposure protocols was likely a byproduct of the compromised circadian clock, instead of the direct effect of light. In the current study, we provided evidence that light has a direct effect on feeding in mice that does not require the circadian clock. First, in most of our studies, animals were never exposed to any chronic circadian disruption protocols, and the acute (4 h) light/darkness extension paradigms we used were unlikely to significantly disturb the endogenous circadian clock (Tahkamo et al., 2019; Koritala et al., 2023; Regmi et al., 2024). In addition, acute light exposure always reduced food intake in mice, regardless of whether it occurred at the beginning or at the end of the dark cycle; similarly, acute darkness exposure always increased food intake in mice at different times of the light cycle. Thus, these effects appeared not to depend on the circadian rhythmicity. Further supporting this notion, we found the same anorexigenic or orexigenic effects of light or darkness in mice with disrupted rhythmicity (by a 2-week constant light exposure). Hence, our findings strongly support that light can regulate feeding behavior in a circadian-independent manner.
The LHb is a well-recognized aversion center, and its contributions to depressive-like mood disorders have been proved by a growing body of evidence (Benarroch, 2015; Hu et al., 2020). In both rodents and humans, major depression is characterized by LHb hyperactivity, and inactivation of the LHb can improve depressive-like symptoms (Aizawa et al., 2013; Cui et al., 2018; Huang et al., 2019). Meanwhile, numerous studies have identified the LHb as an upstream regulator of the brainstem 5-HT neurons (Pollak Dorocic et al., 2014; Huang et al., 2019). While 5-HT insufficiency is critically involved in the etiology of LHb dysfunction-associated depression, 5-HT supplementation can significantly ameliorate affective disorders (Coplan et al., 2014; Teissier et al., 2015; Yang et al., 2018a). Notably, 5-HT plays an essential role in controlling feeding (Simansky, 1996; He et al., 2021a). Despite the fact that mood dysregulation is often accompanied by eating disturbance (Jeong et al., 2013; N. Qu et al., 2020), we consider that the acute alterations in feeding in mice are directly derived from the manipulation of the LHb→5-HTDRN circuit, since previous research pointed out that it takes at least 2–3 weeks for light manipulations to ameliorate or generate depressive-like symptoms (An et al., 2020). Of course, the altered feeding patterns, if sustained in cases of chronic exposure to repeated light/dark manipulations, may contribute, at least in part, to shifts in mood.
Apart from projecting to the brainstem 5-HT neurons, the LHb also innervates DA neurons in the ventral tegmental area (VTA; Geisler and Trimble, 2008; Levinstein et al., 2020). In addition, LHb neurons send inputs to the rostromedial tegmental nucleus (RMTg), which in turn transmits GABAergic tones to inhibit 5-HT and DA neurons (Brinschwitz et al., 2010). Notably, earlier studies showed that DRN-projecting LHb neurons are mainly located in the medial part, whereas those projecting to the VTA are centralized in the lateral part, of the LHb (Baker et al., 2016; Metzger et al., 2021). Thus, the VTA-projecting and DRN-projecting LHb neurons are largely separate. Consistently, we failed to detect collateral terminals in the VTA from DRN-innervating LHb neurons. Despite the complex relationship between the direct activation of 5-HT neurons by glutamatergic LHb neurons and the indirect inhibition of 5-HT neurons by GABAergic RMTg neurons, our study revealed a monosynaptic glutamatergic input from LHb neurons to 5-HTDRN neurons. Confirming this excitatory input, we show that chemogenetic activation of DRN-projecting LHb neurons in mice triggered increased c-fos expression in 5-HTDRN neurons. These results indicate that activation of the DRN-projecting LHb neurons produces a net excitatory effect on 5-HTDRN neurons in mice.
In addition to 5-HT neurons, the DRN contains glutamate (vGLUT3DRN) and GABA (vGATDRN) neurons (Bhave and Nectow, 2021). Activation of Vglut3DRN neurons suppresses food intake while activation of vGATDRN neurons promotes eating (Nectow et al., 2017). Of note, ∼60% of 5-HT neurons in the DRN coexpress vGLUT3 (Ren et al., 2018). Although LHb neurons form synaptic connections with both 5-HT and vGLUT3 neurons, the role of the LHb→vGLUT3DRN circuit remains unclear (Ren et al., 2018). Given the fact that knocking down of TPH2 in the DRN blocks the anorexigenic effects elicited by activation of the LHb→DRN circuit, 5-HT signals play an indispensable role in mediating the appetite-suppressing action of this circuit, although we could not fully exclude the contribution of glutamate. Notably, we specifically targeted the ventral region of the DRN for TPH2 ablation, which experiences a substantial influx of glutamatergic inputs from the LHb. The remaining TPH2 in the other areas of the DRN might be adequately capable of averting the onset of anxiety or depressive-like emotional states as well as alterations in locomotor activity.
Activation of the LHb→5-HTDRN circuit was sufficient to decrease dark cycle food intake in mice, suggesting that this pathway plays a physiologically important role in orchestrating feeding. Interestingly, activating the same circuit during the light cycle failed to result in observable changes in food intake in mice. This phenomenon is likely due to the minimal amount of food consumed during the regular light phase, making the difference hard to detect. Additionally, neuronal activity of the LHb is higher during the light period than the dark cycle (Sakhi et al., 2014a,b), so further activation of the LHb→5-HTDRN pathway at the light phase would produce a less obvious effect on feeding in mice. 5-HT neurons control feeding largely through projections to the hypothalamus (Simansky, 1996; Y. Xu et al., 2008; He et al., 2021b). We recently reported that activation of the 5-HTDRN projections to the arcuate nucleus of the hypothalamus attenuates hunger-driven feeding (N. Qu et al., 2020). Interestingly, the expression levels of agouti-related protein (AgRP) and activities of AgRP neurons exhibit fluctuations between the light and dark periods (Lu et al., 2002; Yang et al., 2011). Whether these alterations are partly due to the change in the LHb→5-HTDRN tone warrants further exploration.
The mechanism through which light information is conveyed to the LHb remains to be elucidated. Our rabies virus-based tracing results indicate that the MPA, LPO, and lateral hypothalamus send monosynaptic projections to the DRN-projecting LHb neurons. Previous studies have demonstrated that these nuclei receive innervation from the retina (Zhang et al., 2021; Kerschensteiner and Feller, 2024), suggesting potential pathways for the mediation of light-induced activation of LHb neurons. Additionally, earlier reports have documented synaptic connections between the retina and the LHb (Hattar et al., 2006), and it remains to be determined whether LHb activation suppresses feeding through direct inputs from the retina.
It is worth mentioning that while nocturnal animals eat primarily during the dark phase, diurnal species and humans prefer to consume food during the light phase (Rathod and Di Fulvio, 2021; Shankar and Williams, 2021). Therefore, an important question is whether the role of light exposure in feeding regulation is conserved between nocturnal and diurnal species. Interestingly, in diurnal Nile grass rats, we found that light exposure produced similar anorexigenic effects and elevated LHb neuron and 5-HTDRN neuron activity. We further confirmed that activation of 5-HT2CR decreased food intake in these rats as in mice. Thus, our results support that the activation effects of light exposure on LHb neurons and 5-HTDRN neurons are similar between nocturnal mice and diurnal Nile grass rats, and the anorexigenic function of lorcaserin is conserved. This notion is also consistent with the observation that this acute light-induced anorexia is independent of the circadian clock that differs in nocturnal versus diurnal species. Studies using genetic mouse models indicate that multiple 5-HT2CR–expressing neural populations in the brain contribute to the anorexigenic effects of 5-HT, including those in the arcuate nucleus, the ventral tegmental area, and the nucleus of the solitary tract (Heisler et al., 2002; Y. Xu et al., 2008, 2010, P. Xu et al., 2017; Berglund et al., 2013; D’Agostino et al., 2018; Leon et al., 2019; He et al., 2022; Wagner et al., 2023). It remains to be further tested whether the same or similar 5-HT2CR–expressing neural populations in the Nile grass rats mediate the same anorexigenic effects as in mouse brains.
Notably, it has been reported that ambient light luminance affects the amount of food consumed by humans (Wansink and van Ittersum, 2012). For example, the consumption volume is higher when participants eat in the absence of light than in the presence of light (Scheibehenne et al., 2010). Another human study reported that restaurant customers in dim ambient light settings purchase 38.85% more calories than those in bright ambient light settings (Biswas et al., 2017). Consistent with these findings, our human results confirmed that reducing environmental light intensity can increase energy consumption in healthy individuals. In addition, the selective 5-HT reuptake inhibitor and 5-HT2CR agonist, fluoxetine and lorcaserin, cause anorexia in humans (Ferguson, 1986; O’Kane et al., 1994; Afkhami-Ardekani and Sedghi, 2005; Bohula et al., 2018), similarly as they do in rodents (He et al., 2021b). Considering the similar actions of 5-HT signals to suppress feeding in various species, and the similar responses of LHb neurons to light exposure, we suggest that manipulating the LHb→5-HTDRN circuit may provide avenues to influence feeding behavior in humans.
In summary, we discovered that light regulates feeding in mice, an effect mediated through the LHb→5-HTDRN circuit. These findings expand our understanding of the neuronal pathways that coordinate environmental cues with feeding behavior and may provide insights into developing novel approaches to treat obesity or eating disorders. One limitation of this study is that the animal experiments were performed solely in males. Further investigation is warranted to determine whether the light-induced activation of the LHb→5-HTDRN circuit and subsequent feeding suppression extends to females. In addition, light exposure is likely to increase locomotor activity in Nile grass rats, as reported previously (Kim et al., 2023). It remains to be explored whether these alterations in locomotion will confound the effects of light on feeding behavior in Nile grass rats.
Footnotes
This work was supported by grants from the USDA/CRIS (51000-064-01S to Y.X.), American Diabetes Association (1-15-BS-184 to Q.T.), and Wuhan Municipal Health Youth Talent Training Program (to N.Q.). We thank Dr. Lily Yan from Michigan State University for the Nile grass rats and all the human participants whose contributions made this work possible.
↵*H.L. and N.Q. contributed equally to this work.
The authors declare no competing financial interests.
- Correspondence should be addressed to Yong Xu at yongx{at}bcm.edu. Hailan Liu at hailan.liu{at}bcm.edu, or Na Qu at questina{at}163.com.
