Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

5-HT recruits distinct neurocircuits to inhibit hunger-driven and non-hunger-driven feeding

Abstract

Obesity is primarily a consequence of consuming calories beyond energetic requirements, but underpinning drivers have not been fully defined. 5-Hydroxytryptamine (5-HT) neurons in the dorsal Raphe nucleus (5-HTDRN) regulate different types of feeding behavior, such as eating to cope with hunger or for pleasure. Here, we observed that activation of 5-HTDRN to hypothalamic arcuate nucleus (5-HTDRN → ARH) projections inhibits food intake driven by hunger via actions at ARH 5-HT2C and 5-HT1B receptors, whereas activation of 5-HTDRN to ventral tegmental area (5-HTDRN → VTA) projections inhibits non-hunger-driven feeding via actions at 5-HT2C receptors. Further, hunger-driven feeding gradually activates ARH-projecting 5-HTDRN neurons via inhibiting their responsiveness to inhibitory GABAergic inputs; non-hunger-driven feeding activates VTA-projecting 5-HTDRN neurons through reducing a potassium outward current. Thus, our results support a model whereby parallel circuits modulate feeding behavior either in response to hunger or to hunger-independent cues.

Introduction

The prevalence of obesity has substantially increased since the 1950s. Strategies to combat obesity and related comorbidities, including type 2 diabetes, cardiovascular disease, and some types of cancer are urgently required. Obesity is primarily a consequence of the consumption of more food than the body requires; calories that are consequently stored as fat. For some individuals, overeating is primarily driven by hunger (a state of nutritional deficit), whereas others overeat in the absence of hunger [1, 2]. A better understanding of the neurobiological mechanisms for these two types of feeding behaviors is essential to develop rational precision medicines to more effectively treat the complex etiology of obesity.

The brain serotonin (5-hydroxytryptamine, 5-HT) system plays critical roles in the regulation of feeding. Brain 5-HT is primarily synthesized by neurons in the dorsal Raphe nucleus (DRN) in the midbrain, which projects to other midbrain regions and the hypothalamus [3]. Hunger decreases 5-HT release from the DRN, while satiation increases it [4], suggesting that dynamic 5-HT bioavailability may participate in the physiological regulation of feeding behavior. Indeed, d-fenfluramine (d-Fen), a pharmacological agent that increases 5-HT content [5], showed a potent anorexigenic activity in rodents and humans [6,7,8]. Conversely, treatments that suppress central 5-HT signals produce hyperphagia and weight gain [9,10,11,12]. The brain 5-HT system has been an attractive target for anti-obesity therapies. For example, d-Fen, used in combination with phentermine (as Fen/Phen), was widely prescribed obesity medication in the 1990s. The therapeutic benefit of Fen/Phen was established to be mediated primarily via 5-HT2C receptors (5-HT2CRs); the 5-HT2CR agonist lorcaserin was clinically used to treat human obesity [13]. Recent evidence indicates that lorcaserin not only inhibits food intake triggered by hunger [14, 15], but also inhibits the consumption of high palatable foods in the absence of hunger [15, 16]. These results suggest that the brain 5-HT system can inhibit both hunger-driven and non-hunger-driven feeding. Although the Fen/Phen regimen [17] and lorcaserin [18] were no longer used in the clinic due to nonspecific adverse effects, better understanding about how the brain 5-HT circuitry regulates various types of feeding behaviors will provide a necessary framework to develop novel strategies to combat the obesity pandemic.

The arcuate nucleus of hypothalamus (ARH) and the ventral tegmental area (VTA) are two brain regions implicated in the control of energy homeostasis [19, 20] and motivated behaviors [21,22,23,24], respectively. Here, we characterized the neurotransmission between 5-HTDRN neurons and distinct types of neurons in the ARH and VTA, and further examined the effects of the 5-HTDRN → ARH and 5-HTDRN → VTA projections on hunger-driven and non-hunger-driven feeding. In addition, we assessed the dynamic changes of 5-HTDRN neuron activities during the course of hunger-driven and non-hunger-driven feeding, and delineated distinct ionic mechanisms by which 5-HTDRN neurons are regulated.

Materials and methods

Mice

TPH2-CreER mice were purchased from Jackson Laboratory (#016584) that express tamoxifen-inducible Cre recombinase selectively in 5-HT neurons, as we validated previously [25]. Dopamine transporter (DAT)-CreER mice were purchased from Jackson Laboratory (#016583) that express tamoxifen-inducible Cre recombinase selectively in dopamine (DA) neurons, as we validated previously [25]. Tamoxifen-inducible POMC-CreER mice were obtained from Dr. Joel Elmquist lab [26] and has been previously validated in the lab [27, 28]. All these Cre lines, as well as AgRP-Cre mice [29], were bred with Rosa26-LSL-tdTOMATO mice [30] (Jackson Laboratory, #007905), to generate reporter mice, and tdTOMATO signals were examined throughout the brain to confirm the expected Cre expression pattern. We also crossed TPH2-CreER onto POMC-CreER/Rosa26-LSL-tdTOMATO, AgRP-Cre/Rosa26-LSL-tdTOMATO, or DAT-CreER/Rosa26-LSL-tdTOMATO mice to generate compound Cre lines for CRACM studies, as described below. Some TPH2-CreER mice were used for optogenetic and fiber photometry studies as outlined below. In addition, some TPH2-CreER/Rosa26-LSL-tdTOMATO mice, as well as C57Bl6j mice (purchased from the mouse facility of Baylor College of Medicine), were used for retrograde tracing and electrophysiology studies. Further, we crossed TPH2-CreER allele onto γ2fl/fl mice [31] or SK3fl/fl mice [32] (Jackson Laboratory, #019083) to generate mice lacking γ2 or SK3 selectively in 5-HT neurons; littermate γ2fl/fl or SK3fl/fl mice were used as controls. All the breeders have been backcrossed to C57Bl6j background for >12 generations. Mice were housed in a temperature-controlled environment in groups of two to five at 22–24 °C, using a 12 h light/12 h dark cycle. All mice were fed standard chow (6.5% fat, #2920, Harlan-Teklad, Madison, WI) ad libitum, unless described otherwise. Water was provided ad libitum.

Neurotracing

To determine whether 5-HTDRN neurons project to the ARH and VTA, 12-week-old TPH2-CreER mice were anesthetized by isoflurane and received stereotaxic injections of Cre-dependent adeno-associated virus (AAV) expressing wheat germ agglutinin (WGA)-zsGreen (AAV9-CBA-DIO-WGA-zsGreen, 7.6 × 1012 GC/ml) into the DRN (200 nl, 4.65 mm posterior, 0 mm lateral, and 3.6 mm ventral to the Bregma). These mice also received an i.p. injection of tamoxifen (0.2 mg/g) to induce Cre activity. One week after injections, mice were perfused with 10% formalin, and brain sections were cut at 25 μm (five series). One series of the sections were blocked (3% normal donkey serum) for 1 h, incubated with goat anti-WGA antibody (1:1000; AS-2024, Vector Laboratories) on shaker at 4 °C for overnight, followed by the donkey anti-goat AlexaFluor 488 (1:200, A11055, Invitrogen) for 2 h. Slides were cover-slipped and analyzed using a Leica DM5500 fluorescence microscope with OptiGrid structured illumination configuration.

To further confirm the projections of 5-HTDRN neurons project to the ARH and VTA, 12-week-old C57Bl6j mice were anesthetized by isoflurane and received stereotaxic injections of Red RetroBeads and Green RetroBeads into the ARH (bilateral 200 nl/site, 1.7 mm posterior, 0.25 mm lateral, and 5.9 mm ventral to the Bregma) and VTA (bilateral 200 nl/site, 3 mm posterior, 0.5 mm lateral, and 4.5 mm ventral to the Bregma), respectively. Four weeks after, the mice were perfused as above. Brain sections were then subjected to immunofluorescent staining for 5-HT. Briefly, sections were incubated with primary rabbit anti-5-HT antibody (1:10,000, 20080, ImmunoStar) overnight, followed by the goat anti-rabbit AlexaFluor 405 (1:500, A31556, Invitrogen) for 1.5 h. The sections were mounted on glass slides and fluorescence images were analyzed using a Leica DM5500 fluorescence microscope with OptiGrid structured illumination configuration. The DRN neurons with 5-HT immunoreactivity co-labeled by Red RetroBeads, by Green RetroBeads, or by both beads were counted in ten mice.

CRACM

In order to examine the neurotransmissions between 5-HTDRN neurons and POMCARH, AgRPARH, or DAVTA neurons, we performed the channelrhodopsin-2 (ChR2)-assisted circuit mapping (CRACM), similarly as we described before [33]. Three different compound mouse strains were used: TPH2-CreER/POMC-CreER/Rosa26-LSL-tdTOMATO, TPH2-CreER/AgRP-Cre/Rosa26-LSL-tdTOMATO, and TPH2-CreER/DAT-CreER/Rosa26-LSL-tdTOMATO. These mice were anesthetized by isoflurane and received stereotaxic injections of Ad-iN/WED (100 nl) and AAV-EF1α-DIO hChR2(H134R)-enhanced yellow fluorescent protein (EYFP; 100 nl) into the DRN. Tamoxifen (0.2 mg/g, i.p.) was injected into these mice to induce Cre activities. After a 4-week recovery, mice were sacrificed and unfixed brain slices (containing both the DRN and ARH, or DRN and VTA, 120 µm in thickness) were prepared from these mice. These brain slices were subjected to fluorescent microscopy to visualize and quantify WGA(+), tdTOMATO(+), and WGA(+)/tdTOMATO(+) neural somas in the ARH or VTA. Notably, EYFP expressed by the AAV-EF1α-DIO hChR2(H134R)-EYFP shares overlapping spectra with GFP. However, EYFP will only label the fibers and terminals in the ARH or VTA, but will not cross the synapse to label the somas of downstream neurons. Thus, the GFP-labeled somas within the ARH or VTA must be filled by WGA-GFP fusion proteins. About 500–600 neural somas were counted from each mouse and three to five mice were included for each strain. These brain slices were also used to perform electrophysiological recordings, as described below.

Under the voltage-clamp mode, light-evoked excitatory postsynaptic currents (eEPSCs) or inhibitory postsynaptic currents (eIPSCs) were recorded in WGA(+)/tdTOMATO(+) neurons in the ARH or VTA in response to 473 nm laser blue light (10 Hz, 10 ms pulse, 0.2 mW; MBL-FN-473, CNI LASER). The eEPSCs were recorded in whole-cell voltage-clamp mode, by holding the membrane potential at Vh = −60 mV. The pipette solution containing 125 mM CsCH3SO3, 10 mM CsCl, 5 mM NaCl, 2 mM MgCl2, 1 mM EGTA, 10 mM HEPES, 5 mM (Mg)ATP, and 0.3 mM (Na)2GTP (pH 7.3 with NaOH). The eIPSCs were recorded in whole-cell voltage-clamp mode, by holding the membrane potential at Vh = −70 mV. The CsCl-based pipette solution containing of 140 mM CsCl, 10 mM HEPES, 5 mM MgCl2, 1 mM BAPTA, 5 mM (Mg)ATP, and 0.3 mM (Na)2GTP (pH 7.30 adjusted with NaOH; 295 mOsm/kg). TTX (1 μM) and 4-AP (400 μM) were added to the artificial cerebrospinal fluid (aCSF), in order to confirm the response was monosynaptic responses. eEPSCs were recorded when 50 μM bicuculline was applied in the bath solution; after an eEPSC was detected, 30 μM D-AP5 and 30 μM CNQX were used to confirm the current was mediated by glutamate receptors. eIPSCs were recorded when 30 μM D-AP5 and 30 μM CNQX were applied in the bath solution; after an eIPSC was detected, 50 μM bicuculline was used to confirm the current was mediated by GABAA receptor. Under the current-clamp mode, 30 μM D-AP5, 30 μM CNQX, and 50 μM bicuculline were applied in the bath solution to block the glutamate and GABA inputs; action potential firing frequency and resting membrane potential were recorded in response to 473 nm blue light (20 Hz, 10 ms pulse, 0.2 mW) in the absence or presence of 100 μM SB242084 (a 5-HT2CR antagonist) [15], or 10 μM SB224289 (a 5-HT1BR antagonist) [34]. Note that some neurons did not spontaneously fire in the presence of D-AP5, CNQX, and bicuculline cocktail, and therefore were not included in the CRACM experiments.

Optogenetics and hunger-driven or non-hunger-driven feeding

TPH2-CreER mice (12 weeks of age) were anesthetized with isoflurane and received stereotaxic injections of Cre-dependent AAV expressing ChR2-YFP (AAV-EF1α-DIO hChR2(H134R)-YFP, 6.2 × 1012 VP/ml) into the DRN (200 nl; 4.65 mm posterior, 0 mm lateral, and 3.6 mm ventral to the Bregma). Simultaneously, an optic fiber (0.2 mm in diameter with a numerical aperture of 0.22) was implanted to target the ARH (1.7 mm posterior, 0.25 mm lateral, and 5.8 mm ventral to the Bregma) or the VTA (3 mm posterior, 0.5 mm lateral, and 4.3 mm ventral to the Bregma). Four weeks after, mice were subjected to hunger-driven feeding paradigm. Briefly, mice were singly housed with chow diet ad libitum. Mice were then fasted overnight, and chow diet was provided back to the cage at 10 a.m. next morning. Blue (473 nm, 10 ms per pulse, 20 pulses per second) or yellow light (589 nm, 10 ms per pulse, 20 pulses per second) pulses were delivered to the ARH or VTA through the optic fiber for one hour (10–11 a.m.). To avoid damages associated with optic stimulations, we provide 10 ms per pulse at 20 pulses per second (200 ms per second) leaving the majority of the period unstimulated. Light power was applied at 10 mW for both photostimulation and photoinhibition to reach appropriate light power exiting the fiber tip in the brain for activation and inhibition, respectively. Food (chow) intake was measured during the stimulation period and also at 2-h time point. The same mice received blue or yellow light pulses in different trials in a crossover fashion, with a 1-week interval. The order of blue and yellow light stimulation was randomized to avoid potential sequence effects.

After the hunger-driven feeding, the same mice were then subjected to the non-hunger-driven feeding paradigm. One week before the experiment, chow-fed mice were presented with a high-fat diet (HFD) pellets (40% fat; TD.95217, Harlan) for 1 day to avoid neophobia, and then they were maintained on chow diet ad libitum. On the experimental day, at 10 a.m., the HFD was provided to the cage, together with chow diet. Photostimulations (blue or yellow light pulses) were delivered through the optic fiber for 1 h (10–11 a.m.), similarly as described above. Food (chow and HFD) intake was monitored during the stimulation period and also at 2.5-h time point.

Optogenetics and real-time place preference test

Mice were subjected to the procedure room 1 h before the start of each test and remained in the same room throughout the test. The experimental box contained two chambers that are connected by a doorway. The two chambers were 50 × 50 × 25 cm for each with black pexiglass wall and white pexiglass floor, and connected by an opening (12.5 cm) in the center. In each trial, mice were allowed to explore the two chambers for half an hour (during the light cycle). After acclimation, the time spent in each chamber was recorded and analyzed for chamber preference. To validate accurate and sufficient infection of ChR2-EYFP in 5-HTDRN neurons, all mice were perfused with 10% formalin. Brain sections were cut at 25 μm (five series) and subjected to histological validation. Only those mice with EYFP in the DRN and the fiber tract in the ARH or VTA were included in analyses.

Electrophysiology

Twelve-week-old TPH2-CreER/Rosa26-LSL-tdTOMATO mice were anesthetized by isoflurane and received stereotaxic injections of Green RetroBeads and Purple RetroBeads into the ARH and VTA (bilateral, 200 nl/site), respectively. Four weeks after, the mice were subjected to hunger-driven or non-hunger-driven feeding paradigm as described above. For mice undergoing hunger-driven feeding, mice were sacrificed when fed ad libitum, after a 24-h fasting, or after a 24-h fasting followed by a 2-h refeeding. For non-hunger-driven feeding, mice were sacrificed immediately before or 1 h after they started the non-hunger-driven feeding. Briefly, mice were deeply anesthetized with isoflurane at aforementioned time points 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 (Ren et al., 2012). The mice were then decapitated, and the entire brain was removed and immediately submerged in the cutting solution. Slices (250 µm) were cut with a Microm HM 650 V vibratome (Thermo Scientific). Brain slices containing the DRN were obtained for each animal (Bregma −2.06 mm to −1.46 mm; interaural 1.74–2.34 mm). The slices were recovered for 1 h at 34 °C and then maintained at room temperature in aCSF (pH 7.3) containing 126 mM NaCl, 2.5 mM KCl, 2.4 mM CaCl2, 1.2 mM NaH2PO4, 1.2 mM MgCl2, 5.0 mM glucose, and 21.4 mM NaHCO3) saturated with 95% O2 and 5% CO2 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 34 °C in oxygenated aCSF at a flow rate of 1.8–2 ml/min. tdTOMATO(+) neurons, co-labeled with Green or Purple RetroBeads, in the DRN were visualized using epifluorescence and IR-DIC imaging on an upright microscope (Eclipse FN-1, 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 0.05 mM Mg-ATP. Recordings were made using a MultiClamp 700B amplifier (Axon Instrument), sampled using Digidata 1440 A 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 action potential. Currents were amplified, filtered at 1 kHz, and digitized at 20 kHz. The current-clamp mode was engaged to measure neural firing frequency and resting membrane potential. The values for resting membrane potential and firing frequency were averaged within 2-min bin.

A voltage-clamp protocol was used to record the small conductance Ca2+-activated K+ (SK) current as we did before [35, 36]. Briefly, 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 0.05 mM Mg-ATP. The SK tail current was evoked by a 100 ms depolarizing pulse from a holding potential −60 mV to 0 mV and then back to −60 mV. To measure miniature EPSC (mEPSC), the internal recording solution contained 125 mM CsCH3SO3, 10 mM CsCl, 5 mM NaCl, 2 mM MgCl2, 1 mM EGTA, 10 mM HEPES, 5 mM (Mg)ATP, and 0.3 mM (Na)2GTP (pH 7.3 with NaOH) [37]. The mEPSCs were recorded in whole-cell voltage-clamp mode, by holding the membrane potential at Vh = −60 mV in the presence of 1 μM TTX and 50 μM bicuculline. The miniature IPSCs (mIPSCs) were recorded in whole-cell voltage-clamp mode by holding the membrane potential at Vh = −70 mV. The CsCl-based pipette solution contains 140 mM CsCl, 10 mM HEPES, 5 mM MgCl2, 1 mM BAPTA, 5 mM (Mg)ATP, and 0.3 mM (Na)2GTP (pH 7.30 adjusted with NaOH; 295 mOsm/kg). The mIPSCs were recorded in the presence of 1 μM TTX, 30 μM D-AP5, and 30 μM CNQX [38]. Frequency and peak amplitude were measured using the Mini Analysis program (Synaptosoft Inc.).

Validation of γ2 or SK3 deletion

To validate selective deletion of γ2 from 5-HT neurons, γ2-TPH2-KO mice and control littermates were perfused with 10% formalin, and brain sections were cut at 25 μm (five series). These brain sections were subjected to dual immunofluorescent staining for γ2 and 5-HT. Briefly, sections were incubated with primary rabbit anti-γ2 antibody (1:100, NB300-190, Novus Biologicals) and primary goat anti-5-HT antibody (1:5000, 20079, ImmunoStar) overnight, followed by the donkey anti-rabbit AlexaFluor 594 (1:200, A21207, Invitrogen) and donkey anti-goat AlexaFluor 488 (1:500, A11055, Invitrogen) for 1.5 h. γ2 Immunoreactivity (red) and 5-HT immunoreactivity (green) fluorescence signals were analyzed using a Leica DM5500 fluorescence microscope with OptiGrid structured illumination configuration.

We used real-time RT-qPCR to examine levels of SK3 mRNAs in the DRN from control vs. SK3-TPH2-KO mice. To this end, the mouse brains were removed, the DRN was quickly dissected, and subjected to RNA extraction and reverse transcription using the qSript XLT cDNA SuperMix (Quanta Biosciences, 95161-025), according to the manufacturer’s instruction. Briefly, 2 μl of qSript XLT cDNA SuperMix was added to 8 μl of each sample, and 10 μl of this RT mixture were used for cDNA synthesis (25 °C for 10 min, 42 °C for 60 min, and 85 °C for 5 min). The cDNA samples were amplified on a CFX384 Real-Time System (Bio-Rad) using SsoADV SYBR Green Supermix (Bio-Rad). Results were normalized against the expression of house-keeping gene. Primer sequences were for cyclophilin: forward-TGGAGAGCACCAAGACAGACA and reverse-TGCCGGAGTCGACAATGAT; for SK3: forward-TGTTGCACTCTTCTCCCACG and reverse-GGTCATTGAGATTTAGCTGGCT.

Fiber photometry

For the fiber photometry experiments, TPH2-CreER mice (12 weeks of age) were anesthetized by isoflurane and received stereotaxic injections of pAAV9.Syn.Flex.GCaMP6m WPRE.SV40 (200 nl/site, 3 × 109 VP/ml) into the DRN. During the same surgery, an optical fiber (fiber: core = 400 μm; 0.48 NA; M3 thread titanium receptacle; Doric Lenses) was implanted over the DRN (4.65 mm posterior, 0 mm lateral, and 3.30 mm ventral to the Bregma, based on Franklin and Paxinos Mouse Brain Atlas). Fibers were fixed to the skull using dental acrylic and mice were allowed 3 weeks for recovery before acclimatization investigator handling for 1 week before experiments.

The fiber photometry recordings started 4–6 weeks after surgeries to allow for adequate recovery and GCaMP6m expression to stabilize. All recordings were done in the home cage of the singly housed mice that were subjected to either hunger-driven or non-hunger-driven feeding paradigm as described above. Mice were allowed to adapt to the tethered patchcord for 2 days prior to experiments and given 5 min to acclimate to the tethered patchcord prior to any recording. Recordings were performed during the first 5 min of each 30-min bin, to avoid photobleaching. 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. A 465 nm excites calcium-dependent fluorescence from GCaMP6m protein, providing a measure of neuron activity. A 405 nm excites calcium-independent fluorescence from GCaMP6m protein and serves as a control for movement and bleaching artifacts. 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) [39]. The fiber photometry data was down sampled to 8 Hz. To align acute neural responses (in seconds) to each eating bout, animals’ behavior were also simultaneously recorded with a camera. We derived the values of GCaMP6 fluorescence change (ΔF/F0) by calculating (F465 − F0)/F0, where F0 is the baseline fluorescence signal 2 s prior to the onset of each eating bout [40]. The F405 channel is 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 each eating bout. To evaluate the slow changes in neural activity (every 30 min), we derived the values of fluorescence change (ΔF/Fn) by calculating (F465 − F405)/F405, in order to minimize the interference of bleaching artifacts [41].

Statistical analyses

The minimal sample size was predetermined by the nature of experiments. The data are presented as mean ± SEM and/or 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. Note that both male and female animals were included; since we did not observe sex differences, data from both sexes were pooled for statistical analyses.

Study approval

Care of all animals and procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee.

Results

5-HTDRN neurons project to the ARH and VTA

To examine whether 5-HTDRN neurons project to the ARH and VTA, we stereotaxically injected AAV-DIO-WGA-zsGreen into the DRN of TPH2-CreER mice (with tamoxifen induction) to selectively express WGA-zsGreen in 5-HTDRN neurons (Fig. S1A, B). WGA is an anterograde trans-synaptic tracer, which can label the somas of downstream neurons that are immediately innervated by 5-HTDRN terminals [42]. We observed abundant WGA-zsGreen-labeled cell bodies in the ARH and VTA (Fig. S1C, D), confirming that 5-HTDRN neurons project to both of these brain regions. Next, we sought to determine whether the ARH and VTA receive projections from the same or segregated 5-HTDRN neurons. To this end, we stereotaxically injected red and green retrograde tracers (RetroBeads) into the ARH and VTA, respectively, in the same mice (Fig. S1E, F). As expected, we detected both red and green RetroBeads in the DRN, which were colocalized with 5-HT immunoreactivity (Fig. S1G), confirming that 5-HTDRN neurons project to the ARH and/or VTA. Consistent with the broad 5-HTDRN-originated projections throughout the brain regions other than the ARH and VTA [43, 44], we noted that the majority of 5-HTDRN neurons were not labeled by retrograde beads injected into the ARH and VTA (Fig. S1G, H). Thus, only a small subset of 5-HTDRN neurons provide projections to the ARH and/or VTA (as co-labeled by either green beads or red beads). Within this 5-HTDRN subset projecting to the ARH and/or VTA, only 13.27% have collateral projections to both regions (co-labeled by both green and red beads), while the majority project to either ARH only (33.71%) or VTA only (53.02%), indicating that ARH- and VTA-projecting 5-HTDRN neurons are largely segregated (Fig. S1H).

5-HTDRN → ARH circuit inhibits hunger-driven feeding but not non-hunger-driven feeding

We next sought to clarify the target of 5-HTDRN within the ARH. We focussed on anorexigenic pro-opiomelanocortin neurons and orexigenic agouti-related peptide neurons (POMCARH and AgRPARH) because they are essential for feeding control [45,46,47]. We confirmed that TPH2-CreER does not induce Cre activity in the ARH, while POMC-CreER or AgRP-Cre does not induce Cre activity in the DRN (Fig. S2). We next stereotaxically injected Cre-dependent AAV expressing the optogenetic ChR2-EYFP into the DRN of the compound TPH2-CreER/POMC-CreER/Rosa26-LSL-tdTOMATO mice to express ChR2 specifically in 5-HTDRN neurons and their fibers/terminals, which allows photostimulation. The same mice also received stereotaxic injections of Ad-iN/WED, an anterograde trans-synaptic viral tracer [42], into the DRN. Ad-iN/WED expressed GFP-tagged WGA (WGA-GFP) in a Cre-dependent manner and therefore 5-HTDRN neurons and their downstream neurons were labeled by WGA-GFP. WGA-labeled somas in the ARH were identified as the immediate targets of 5-HTDRN neurons. Note that these mice also express tdTOMATO in POMCARH neurons due to the combination of POMC-CreER and Rosa26-LSL-tdTOMATO alleles. Therefore, WGA(+)tdTOMATO(+) neurons within the ARH were identified as 5-HTDRN-innervated POMCARH neurons, which account for ~49% of POMCARH neurons [tdTOMATO(+), Fig. 1A–C].

Fig. 1: 5-HTDRN → ARH circuit inhibits hunger-driven feeding but not non-hunger-driven feeding.
figure1

A Recordings in 5-HTDRN-innervated POMCARH neurons [tdTOMATO(+)WGA(+)] in response to photostimulation of ChR2-labeled 5-HTDRN-originated fibers within the ARH. B Representative images showing ARH neurons labeled by tdTOMATO, WGA, or both. Scale bars = 20 µm. Arrows point to the single-labeled neurons and the arrowhead points to a double-labeled neuron. C The composition of tdTOMATO-labeled POMCARH neurons that are tdTOMATO(+) only and tdTOMATO(+)WGA(+). A total of 1779 POMCARH neurons were counted from three mice. D Representative action potential traces in response to blue light pulses (473 nm, 10 ms/pulse, 20 Hz) in the absence or the presence of 100 µM SB242084 in a bath solution containing 30 µM CNQX, 30 µM D-AP5, and 50 µM bicuculline. E, F Firing frequency (E) and resting membrane potential (F) at the baseline or after blue light stimulation in the absence or the presence of 100 µM SB242084. Results are shown as individual data points. ***P < 0.001 between blue light vs. baseline; #P < 0.05 and ##P < 0.01 between control and SB242084 in two-way ANOVA analyses followed by Sidak post hoc test (N = 10–14 neurons from three mice per group). G Recordings in 5-HTDRN-innervated AgRPARH neurons [tdTOMATO(+)WGA(+)] in response to photostimulation of ChR2-labeled 5-HTDRN-originated fibers within the ARH. H Representative images showing ARH neurons labeled by tdTOMATO, WGA, or both. Scale bars = 20 µm. Arrows point to the single-labeled neurons and arrowheads point to double-labeled neurons. I The composition of tdTOMATO-labeled AgRPARH neurons that are tdTOMATO(+) only and tdTOMATO(+)WGA(+). A total of 1945 AgRPARH neurons were counted from three mice. J Representative action potential traces in response to blue light pulses (473 nm, 10 ms/pulse, 20 Hz) in the absence or the presence of 10 µM SB224289 in a bath solution containing 30 µM CNQX, 30 µM D-AP5, and 50 µM bicuculline. K, L Firing frequency (K) and resting membrane potential (L) at the baseline or after blue light stimulation in the absence or the presence of 10 µM SB224289. Results are shown as individual data points. ***P < 0.001 between blue light vs. baseline; #P < 0.05 between control and SB224289 in two-way ANOVA analyses followed by Sidak post hoc test (N = 11–15 neurons from three mice per group). M Schematic experimental strategy using optogenetics to activate the 5-HTDRN → ARH projections. N The experimental paradigm to measure hunger-driven feeding. Chow intake during photostimulation and during the total 2-h period in mice with the 5-HTDRN → ARH projections unstimulated (yellow) or activated (blue). Results are shown as individual data points. **P < 0.01 between yellow vs. blue light in two-tailed paired t test (N = 6 mice per group). O The experimental paradigm to measure non-hunger-driven feeding. HFD intake during photostimulation and during the total 2.5-h period in mice with the 5-HTDRN → ARH projections unstimulated (yellow) or activated (blue). Results are shown as individual data points (N = 6 mice per group). P Left: the experimental paradigm to measure real-time place preference in mice with the 5-HTDRN → ARH projections unstimulated when in chamber 1 and activated when in chamber 2; right: the % time spent in each chamber in mice at hungry or satiated state with or without blue light stimulation coupled to chamber 2. Results are shown as mean ± SEM with individual data points. **P < 0.01 in two-way ANOVA analyses followed by Sidak post hoc test (N = 6 mice per group).

Since WGA-GFP may also travel retrogradely [48], we adopted another approach to confirm the connectivity between 5-HTDRN neurons and POMCARH neurons. CRACM was employed to examine ex vivo electrophysiological responses in ARH slices containing 5-HTDRN-innervated POMCARH [WGA(+)tdTOMATO(+)] neurons from TPH2-CreER/POMC-CreER/Rosa26-LSL-tdTOMATO mice. Under voltage-clamp, we detected blue light-evoked eEPSC only in 53% (10 out of 19) tested WGA(+)tdTOMATO(+) neurons (Fig. S3A–D). The eEPSCs were mediated by glutamatergic neurotransmission as they were blocked by glutamate receptor inhibitors (CNQX and D-AP5); 4-AP and TTX failed to affect the eEPSCs, confirming a monosynaptic nature (Fig. S3A–D). Interestingly, we also detected light-eIPSCs in a few (4 out of 21, 19%) WGA(+)tdTOMATO(+) neurons (Fig. S3E–H). The eIPSCs were mediated by GABA neurotransmission as they were blocked by GABAA receptor inhibitors (bicuculline); 4-AP and TTX failed to affect the eIPSCs, confirming a monosynaptic nature (Fig. S3E–H). These results indicate that 5-HTDRN neurons provide direct glutamatergic and/or GABAergic inputs to a portion of POMCARH neurons.

We then used a current-clamp mode to examine effects of blue light photostimulation on firing frequency and resting membrane potential in 5-HTDRN-innervated POMCARH [WGA(+)tdTOMATO(+)] neurons. Since AgRPARH neurons are known to provide local inhibitory GABAergic inputs to POMCARH neurons [49], we added 50 μM bicuculline, as well as 30 μM D-AP5 and 30 μM CNQX, in the bath solution to block the GABAergic and glutamatergic inputs in the current-clamp recordings. We found that blue light photostimulation on the 5-HTDRN fibers/terminals in the ARH caused rapid increases in firing frequency and resting membrane potential in all tested WGA(+)tdTOMATO(+) neurons (Fig. 1D–F). We next examined the 5-HT receptor mediating this effect. These activations were completely abolished by a selective 5-HT2CR antagonist, 100 µM SB242084. These results indicate that photostimulation of 5-HTDRN terminals activates POMCARH neurons via a 5-HT2CR-mediated mechanism. Notably, SB242084 also significantly reduced the baseline firing frequency and resting membrane potential (Fig. 1E, F), highlighting an excitatory effect of tonic 5-HT2CR actions on these anorexigenic POMCARH neurons.

We then used a similar approach in compound TPH2-CreER/AgRP-Cre/Rosa26-LSL-tdTOMATO mice to identify 5-HTDRN-innervated AgRPARH neurons, which account for ~32% of AgRPARH [tdTOMATO(+)] neurons (Fig. 1G–I). Using the same CRACM approach under the voltage-clamp protocol, we detected eEPSC in 65% (11 out of 17) tested WGA(+)tdTOMATO(+) neurons evoked by blue light pulses (Fig. S3I–L), and detected eIPSC in 20% (5 out of 25) tested WGA(+)tdTOMATO(+) neurons (Fig. S3M–P). Interestingly, under a current-clamp mode, we found that the blue light photostimulation caused rapid decreases in firing frequency and resting membrane potential in all tested WGA(+)tdTOMATO(+) neurons (Fig. 1J–L). Further, these inhibitions were completely abolished by a selective 5-HT1BR antagonist, 10 µM SB224289. These results indicate that activation of 5-HTDRN neurons can inhibit AgRPARH neurons via a 5-HT1BR-mediated mechanism. Notably, addition of SB224289 also significantly enhanced the baseline resting membrane potential (Fig. 1J–L), highlighting an inhibitory effect of tonic 5-HT1BR actions on AgRPARH neurons.

Since 5-HTDRN neurons activate anorexigenic POMCARH neurons and inhibit orexigenic AgRPARH neurons, we sought to examine the effects of 5-HTDRN → ARH circuit on feeding control. To this end, we stereotaxically injected Cre-dependent AAV expressing ChR2-EYFP into the DRN of TPH2-CreER mice to express ChR2 specifically in 5-HTDRN neurons and their fibers/terminals, and implanted the optic fiber to target the ARH (Fig. S3Q–R). After recovery, mice were first subjected to a hunger-driven feeding paradigm. Briefly, after overnight fasting, mice were provided with chow diet at 10 a.m. next morning, while photostimulation was applied for the first hour of refeeding period (Fig. 1N). Compared to mice receiving yellow light stimulation (as a control, 589 nm, 10 ms per pulse, 10 mW, 20 pulses per second), mice receiving blue light stimulation (473 nm, 10 ms per pulse, 10 mW, 20 pulses per second) displayed significantly reduced chow intake during the photostimulation period and also at the 2-h time point (Fig. 1N). We then examined dark-cycle feeding in chow-fed mice which is primarily driven by hunger, and found that photostimulation of the 5-HTDRN → ARH circuit also significantly inhibited dark-cycle chow intake (Fig. S3S).

We then employed a non-hunger-driven feeding paradigm. Briefly, satiated mice were provided both chow and a HFD at 10 a.m. when mice normally do not eat (Fig. 1O). When provided with both chow and HFD, these satiated mice consumed primarily HFD but minimal chow (Fig. 1O and Fig. S3T), indicating that this HFD feeding is largely driven by the palatability of the HFD but rather by hunger. Importantly, there were no differences in chow or HFD intake between mice receiving yellow light or blue light stimulation (Fig. 1O and Fig. S3T). Next, we used a fasting paradigm that mice were first fasted overnight and then provided with both HFD and chow in the morning. In this setting, we found that photostimulation of the 5-HTDRN → ARH circuit significantly reduced HFD intake in these hungry mice but had no effect on chow intake (Fig. S3U, V). These data suggest that the stimulation of the 5-HTDRN → ARH circuit inhibits hunger-driven feeding, but has little effect on non-hunger-driven feeding behavior.

To examine the valence associated with 5-HTDRN → ARH poststimulation, mice were assessed in a real-time place preference test. Briefly, mice were placed into a rectangular open field, which is comprised of two identical sides connected through an opening in the center (Fig. 1P). Without photostimulation, overnight fasted mice displayed no place preference (Fig. 1P). In the next trial, blue light pulses were paired with entry into the right compartment and fasted mice showed a preference for the blue light-paired area (Fig. 1P). However, in the satiated condition, mice did not show preference for the blue light-paired area (Fig. 1P). These data indicate that stimulation of the 5-HTDRN → ARH circuit transmits a positive valence in hungry mice, but not in satiated mice, suggesting that the reduction in hunger-driven feeding is not due to the mice feeling unwell.

5-HTDRN → VTA circuit inhibits non-hunger-driven feeding but not hunger-driven feeding

We next examined the effect of 5-HTDRN innervation of neurochemically defined DA neurons within the VTA (DAVTA neurons) because they are a key mediator of reward-related behaviors. We confirmed that TPH2-CreER does not induce Cre activity in the VTA, while the DAT-CreER does not induce Cre activity in the DRN (Fig. S2). Compound TPH2-CreER/DAT-CreER/Rosa26-LSL-tdTOMATO mice were used to identify 5-HTDRN-innervated DAVTA neurons, which account for ~49% DAVTA cells [tdTOMATO(+)] (Fig. 2A–C). Using a CRACM approach under the voltage-clamp protocol, we detected eEPSC in 84% (16 out of 19) tested WGA(+)tdTOMATO(+) neurons evoked by blue light pulses (Fig. S4A–D), and detected  eIPSC in 15% (4 out of 26) tested WGA(+)tdTOMATO(+) neurons (Fig. S4E–H).

Fig. 2: 5-HTDRN → VTA circuit inhibits non-hunger-driven feeding but not hunger-driven feeding.
figure2

A Recordings in 5-HTDRN-innervated DAVTA neurons [tdTOMATO(+)WGA(+)] in response to photostimulation of ChR2-labeled 5-HTDRN-originated fibers within the VTA. B Representative images showing VTA neurons labeled by tdTOMATO, WGA, or both. Scale bars = 20 µm. The arrow points to a single-labeled neuron and the arrowhead points to a double-labeled neuron. C The composition of tdTOMATO-labeled DAVTA neurons that are tdTOMATO(+) only and tdTOMATO(+)WGA(+). A total of 2562 DAVTA neurons were counted from five mice. D Representative action potential traces in response to blue light pulses (473 nm, 10 ms/pulse, 20 Hz) in the absence or the presence of 100 µM SB242084 in a bath solution containing 30 µM CNQX, 30 µM D-AP5, and 50 µM bicuculline. E, F Firing frequency (E) and resting membrane potential (F) at the baseline or after blue light stimulation in the absence or the presence of 100 µM SB242084. Results are shown as individual data points. ***P < 0.001 between blue light vs. baseline in two-way ANOVA analyses followed by Sidak post hoc test (N = 9–22 neurons from three mice per group). G Schematic experimental strategy using optogenetics to activate the 5-HTDRN → VTA projections. H The experimental paradigm to measure hunger-driven feeding. Chow intake during photostimulation and during the total 2-h period in mice with the 5-HTDRN → VTA projections unstimulated (yellow) or activated (blue). Results are shown as individual data points (N = 6 mice per group). I The experimental paradigm to measure non-hunger-driven feeding. HFD intake during photostimulation and during the total 2.5-h period in mice with the 5-HTDRN → VTA projections unstimulated (yellow) or activated (blue). Results are shown as individual data points. *P <0.05 between yellow vs. blue light in paired t test (N = 6 mice per group). J Left: the experimental paradigm to measure real-time place preference in mice with the 5-HTDRN → VTA projections unstimulated when in chamber 1 and activated when in chamber 2; right: the % time spent in each chamber in mice at hungry or satiated state with or without blue light stimulation coupled to chamber 2. Results are shown as mean ± SEM with individual data points. *P < 0.01 and ***P < 0.001 in two-way ANOVA analyses followed by Sidak post hoc test (N = 6 mice per group).

Under a current-clamp mode, we found that blue light photostimulation caused rapid increases in firing frequency and resting membrane potential in all tested 5-HTDRN-innervated DAVTA neurons [WGA(+)tdTOMATO(+)] (Fig. 2D–F). Further, these activations were completely abolished by a selective 5-HT2CR antagonist, 100 µM SB242084. These results indicate that stimulation of 5-HTDRN neurons activates DAVTA neurons via a 5-HT2CR-mediated mechanism. Notably, unlike responses seen in 5-HTDRN-innervated POMCARH neurons (Fig. 1E, F), the baseline firing frequency and resting membrane potential of these 5-HTDRN-innervated DAVTA neurons were not significantly altered by SB242084 (Fig. 2E, F), indicating that 5-HT2CRs have minor tonic actions on these DAVTA neurons in a baseline condition.

We then examined the effects of 5-HTDRN → VTA circuit on both hunger-driven and non-hunger-driven feeding behaviors, using TPH2-CreER mice with ChR2-EYFP expressed in 5-HTDRN neurons and the optic fiber implanted to target the VTA (Fig. 2G and Fig. S4I). In the hunger-driven feeding paradigm, photostimulation of the 5-HTDRN → VTA circuit by blue light pulses failed to significantly alter chow intake compared to the same mice receiving control yellow light stimulation (Fig. 2H). However, HFD intake during the non-hunger-driven feeding paradigm was significantly reduced by the photostimulation of the 5-HTDRN → VTA circuit, while chow intake was not affected (Fig. 2I and Fig. S4J). These data indicate that the stimulation of the 5-HTDRN → VTA circuit inhibits non-hunger-driven feeding, but has little effect on hunger-driven feeding. In the real-time place preference test, we observed that blue light photostimulation of the 5-HTDRN → VTA circuit induced a mild negative valance in hungry mice, but strikingly, the same photostimulation transmitted a strong positive valence in satiated mice (Fig. 2J).

5-HTDRN neurons increase activity during hunger-driven feeding and non-hunger-driven feeding

We then examined the dynamic changes of 5-HTDRN neuron activity during hunger-driven feeding and non-hunger-driven feeding in vivo in real time. To this end, we stereotaxically injected AAV-FLEX-GCaMP6m into the DRN of TPH2-CreER mice (with tamoxifen induction) to express the calcium sensor GCaMP6m selectively in 5-HTDRN neurons and implanted the optic fiber to target the DRN, which allowed in vivo monitoring of 5-HTDRN neuron activity in functional mice (Fig. 3A and Fig. S5A). Mice were subjected to the hunger-driven feeding paradigm. Overnight fasted mice ate 0.25–0.35 g of chow every 30 min, which resulted in a steady increase in cumulative chow intake during a 2-h refeeding period (Fig. 3B). Consistent with previous reports [40, 43], we observed rapid increases in 5-HTDRN neuron activity that were tightly associated with onset of each eating bout and only lasted for a few seconds (Fig. S5B). Interestingly, we noted that 5-HTDRN neuron activity also displayed a slow, but constant elevation during the 2-h feeding period (Fig. 3C). Linear regression analysis revealed a significant correlation between cumulative chow intake and 5-HTDRN neuron activity (Fig. 3D).

Fig. 3: Dynamics of 5-HTDRN neuron activity during hunger-driven feeding and non-hunger-driven feeding.
figure3

A Schematic illustration to stereotaxically inject AAV-FLEX-GCaMP6m virus into the DRN of TPH2-CreER mice to express GCaMP6m in 5-HTDRN neurons. B The schematic fiber photometry recordings of 5-HTDRN neuron activity in mice during the course of hunger-driven feeding (left), the amount of chow intake during each 30-min bin within the 2-h refeeding (middle), and the cumulative chow intake (right). Results are shown as mean ± SEM with individual data points (N = 6 mice per group). C 5-HTDRN neuron activity in mice during the course of 2-h hunger-driven feeding or in control hungry mice with no access to food. Results are shown as mean ± SEM (N = 5 or 6 mice per group). ***P < 0.001 vs. time 0; ##P < 0.01 and ###P < 0.001 vs. no food group in two-way ANOVA analyses followed by Sidak post hoc test. D The linear regression curve of cumulative chow intake and 5-HTDRN neuron activity during hunger-driven feeding. Results are shown as individual data points with the linear regression curve plotted. E The schematic fiber photometry recordings of 5-HTDRN neuron activity in mice during the course of non-hunger-driven feeding (left), the amount of HFD intake during each 30-min bin within 2.5-h feeding (middle), and the cumulative HFD intake (right). Results are shown as mean ± SEM with individual data points (N = 6 mice per group). F 5-HTDRN neuron activity in mice during the course of 2.5-h non-hunger-driven feeding. Results are shown as mean ± SEM (N = 4 or 6 mice per group). ***P < 0.001 vs. time 0; #P < 0.05 and ###P < 0.001 vs. no food group in two-way ANOVA analyses followed by Sidak post hoc test. G The linear regression curve of cumulative HFD intake and 5-HTDRN neuron activity during non-hunger-driven feeding. Results are shown as individual data points with the linear regression curve plotted.

We then subjected these mice to the non-hunger-driven feeding paradigm. Similarly, we observed transient elevations in 5-HTDRN neuron activity associated with each eating bout (Fig. S5C). In addition, the 5-HTDRN neuron activity displayed a slow elevation, which was significantly correlated with the cumulative HFD consumption (Fig. 3E–G). Together, these results indicate that during both hunger-driven and non-hunger-driven feeding, 5-HTDRN neuron activity gradually escalates in the scale of minutes and hours, which is correlated with the amount of food consumed.

GABAA currents mediate dynamic changes in ARH-projecting 5-HTDRN neuron activity during hunger-driven feeding

We next explored whether the activity of 5-HTDRN neurons projecting to the ARH and/or VTA changes in association with hunger-driven feeding. To this end, TPH2-CreER/Rosa26-LSL-tdTOMATO mice were stereotaxically injected with green and purple RetroBeads into the ARH and VTA, respectively (Fig. S6A). Brain slices containing the DRN were freshly prepared from mice fed ad libitum, overnight fasted, or overnight fasted followed by 2-h refeeding. Electrophysiological recordings were performed in tdTOMATO-labeled 5-HTDRN neurons projecting to the ARH (co-labeled by green beads) or in tdTOMATO-labeled 5-HTDRN neurons projecting to the VTA (co-labeled by purple beads, Fig. S6A). The firing frequency and resting membrane potential of ARH-projecting 5-HTDRN neurons in fasted mice were significantly reduced compared to fed mice, and 2-h refeeding increased firing frequency and resting membrane potential to the level of fed mice (Fig. 4A, B). On the other hand, firing frequency and resting membrane potential of VTA-projecting 5-HTDRN neurons were not affected by these conditions (Fig. 4A, B).

Fig. 4: GABAA currents mediate dynamic changes in ARH-projecting 5-HTDRN neuron activity during hunger-driven feeding.
figure4

A, B Firing frequency (A) and resting membrane potential (B) in ARH-projecting or VTA-projecting 5-HTDRN neurons in mice fed ad libitum, fasted for overnight, or fasted for overnight followed by 2-h refeeding. Results are shown as mean ± SEM with individual data points. **P < 0.01 and ***P <0.001 in one-way ANOVA analyses followed by Sidak post hoc test (N = 15–27 neurons from three mice per group). C Amplitude of mIPSC in ARH-projecting or VTA-projecting 5-HTDRN neurons in mice fed ad libitum, fasted for overnight, or fasted for overnight followed by 2-h refeeding. Results are shown as mean ± SEM with individual data points. ***P < 0.001 in one-way ANOVA analyses followed by Sidak post hoc test (N = 16–17 neurons from three mice per group). DF Amplitude of mIPSC (D), firing frequency (E), and resting membrane potential (F) in ARH-projecting 5-HTDRN neurons in control or γ2-TPH2-KO mice fed ad libitum, fasted for overnight, or fasted for overnight followed by 2-h refeeding. Results are shown as mean ± SEM with individual data points. ***P < 0.001 in two-way ANOVA analyses followed by Sidak post hoc test (N = 14–25 neurons from three mice per group). G Averaged 5-HTDRN neuron activity in control or γ2-TPH2-KO mice during the course of 2-h hunger-driven feeding. Note that the data from control mice are the duplicate of data described in Fig. 3C. Results are shown as mean ± SEM. *P < 0.05 and **P < 0.01 vs. time 0 in control mice in two-way ANOVA analyses followed by Sidak post hoc test (N = 5 or 6 mice per group). H The amount of chow intake during 2-h hunger-driven feeding in control or γ2-TPH2-KO mice. Results are shown as mean ± SEM with individual data points. *P < 0.05 in t test (N = 5 or 10 mice per group). I Averaged 5-HTDRN neuron activity in control or γ2-TPH2-KO mice during the course of 2.5-h non-hunger-driven feeding. Note that the data from control mice are the duplicate of data described in Fig. 3F. Results are shown as mean ± SEM. *P < 0.05 and **P < 0.01 vs. time 0 in control mice; #P < 0.05, ##P < 0.01, and ###P < 0.001 vs. time 0 in γ2-TPH2-KO mice in two-way ANOVA analyses followed by Sidak post hoc test (N = 5 or 6 mice per group). J The amount of HFD intake during 2.5-h non-hunger-driven feeding in control or γ2-TPH2-KO mice. Results are shown as mean ± SEM with individual data points (N = 7 or 16 mice per group).

To further explore the mechanisms for the dynamic 5-HTDRN neuron activity during the hunger-driven feeding, we examined the mEPSC and the SK currents in ARH- and VTA-projecting 5-HTDRN neurons, and did not observe any significant changes in these currents in the different feeding conditions (Fig. S6B–D). However, the amplitude of miniature IPSC (mIPSC) in ARH-projecting 5-HTDRN neurons was significantly enhanced by fasting compared to fed mice, and 2-h refeeding largely restored mIPSC amplitude to the level of fed mice. mIPSC amplitude in VTA-projecting 5-HTDRN neurons was not affected (Fig. 4C). No differences were observed in the frequency of mIPSC by feeding conditions (Fig. S6E). Given the inhibitory nature of mIPSC, we hypothesized that GABA-mediated mIPSC contributes to the dynamic ARH-projecting 5-HTDRN neuron activity during the course of hunger-driven feeding.

To examine this, we generated γ2-TPH2-KO mice which lack the pore-forming γ2 subunit of GABAA receptor [31, 50, 51] specifically in 5-HT neurons and their control littermates (Fig. S6F). We found that mIPSC amplitude, but not frequency, in ARH-projecting 5-HTDRN neurons from γ2-TPH2-KO mice was significantly smaller than control mice, and the dynamic pattern of mIPSC we observed in control mice during the hunger-driven feeding was abolished in γ2-TPH2-KO mice (Fig. 4D and Fig. S6G). Importantly, the firing frequency and resting membrane potential of ARH-projecting 5-HTDRN neurons from γ2-TPH2-KO mice did not display the dynamic alterations during the course of hunger-driven feeding, as seen in those from control mice (Fig. 4E, F). Consistently, fiber photometry demonstrated that 5-HTDRN neuron activity in γ2-TPH2-KO mice was not gradually regulated during the course of hunger-driven feeding (Fig. 4G), although the transient elevations associated with onset of eating bouts were still observed (Fig. S6H) similarly as those in control mice (Fig. S5B). Importantly, in the hunger-driven feeding paradigm, the amount of chow intake during the 2-h refeeding period was significantly reduced compared to control mice (Fig. 4H). However, during non-hunger-driven feeding, 5-HTDRN neuron activity in γ2-TPH2-KO mice displayed similar elevations and HFD and chow intake were comparable to control mice (Fig. 4I, J, S6I). Together, these results indicate that GABAA currents mediate dynamic changes in ARH-projecting 5-HTDRN neuron activity during hunger-driven feeding.

SK currents mediate dynamic changes in VTA-projecting 5-HTDRN neuron activity during non-hunger-driven feeding

We also examined the ex vivo dynamic firing activities of ARH- and/or VTA-projecting 5-HTDRN neurons during the course of non-hunger-driven feeding (Fig. S6A). Firing frequency and resting membrane potential of VTA-projecting 5-HTDRN neurons were significantly increased 1 h after non-hunger-driven feeding compared to the baseline levels, whereas the activity of ARH-projecting 5-HTDRN neurons did not change (Fig. 5A, B). We did not detect any significant changes in mIPSC or mEPSC in these neurons (Fig. S7A–D). Interestingly, SK currents in VTA-projecting 5-HTDRN neurons, but not in ARH-projecting neurons, were significantly reduced 1 h after non-hunger-driven feeding (Fig. 5C).

Fig. 5: SK currents mediate dynamic changes in VTA-projecting 5-HTDRN neuron activity during non-hunger-driven feeding.
figure5

A, B Firing frequency (A) and resting membrane potential (B) in ARH-projecting or VTA-projecting 5-HTDRN neurons in mice before and 1-h after non-hunger-driven feeding. Results are shown as mean ± SEM with individual data points. *P < 0.05 in t test (N = 13–27 neurons from three mice per group). C Amplitude of SK-like currents in ARH-projecting or VTA-projecting 5-HTDRN neurons in mice before and 1-h after non-hunger-driven feeding. Results are shown as mean ± SEM with individual data points. ***P < 0.001 in t test (N = 14–27 neurons from three mice per group). DF Amplitude of SK-like currents (D), firing frequency (E), and resting membrane potential (F) in VTA-projecting 5-HTDRN neurons in control or SK3-TPH2-KO mice before and 1-h after non-hunger-driven feeding. Results are shown as mean ± SEM with individual data points. **P < 0.01 and ***P < 0.001 in two-way ANOVA analyses followed by Sidak post hoc test (N = 21–32 neurons from three mice per group). G Averaged 5-HTDRN neuron activity in control or SK3-TPH2-KO mice during the course of 2.5-h non-hunger-driven feeding. Note that the data from control mice are the duplicate of data described in Fig. 3F. Results are shown as mean ± SEM. *P < 0.05 and **P < 0.01 vs. time 0 in control mice in two-way ANOVA analyses followed by Sidak post hoc test (N = 5 or 7 mice per group). H The amount of HFD intake during 2.5-h non-hunger-driven feeding in control or SK3-TPH2-KO mice. Results are shown as mean ± SEM with individual data points. ***P < 0.001 in t test (N = 11 or 13 mice per group). I Averaged 5-HTDRN neuron activity in control or SK3-TPH2-KO mice during the course of 2-h hunger-driven feeding. Note that the data from control mice are the duplicate of data described in Fig. 3C. Results are shown as mean ± SEM. *P < 0.05 and **P < 0.01 vs. time 0 in control mice; #P < 0.05 and ##P < 0.01 vs. time 0 in SK3-TPH2-KO mice in two-way ANOVA analyses followed by Sidak post hoc test (N = 5 or 6 mice per group). J The amount of chow intake during 2-h hunger-driven feeding in control or SK3-TPH2-KO mice. Results are shown as mean ± SEM with individual data points (N = 11 or 15 mice per group).

To examine the physiological role of SK currents in 5-HTDRN neurons in feeding control, we generated control littermates and SK3-TPH2-KO mice with 5-HT-specific deletion of SK3 (Fig. S7E), the most abundant SK isoform in the DRN [52]. SK currents in VTA-projecting 5-HTDRN neurons from SK3-TPH2-KO mice were significantly smaller than control mice, and the dynamic pattern of SK currents we observed in control mice during non-hunger-driven feeding was abolished in SK3-TPH2-KO mice (Fig. 5D). Importantly, the firing frequency and resting membrane potential of VTA-projecting 5-HTDRN neurons from SK3-TPH2-KO mice were significantly elevated, and they did not display the dynamic alterations during the course of non-hunger-driven feeding as seen in those from control mice (Fig. 5E, F). Consistently, we used fiber photometry to demonstrate that unlike control mice, the 5-HTDRN neuron activity in SK3-TPH2-KO mice did not alter during the course of non-hunger-driven feeding (Fig. 5G), although the transient elevations associated with onset of eating bouts were still observed (Fig. S7F). Importantly, the amount of HFD intake, but not chow intake, during the non-hunger-driven feeding paradigm was significantly reduced compared to control mice (Fig. 5H and Fig. S7G). Interestingly, during hunger-driven feeding, the 5-HTDRN neuron activity in SK3-TPH2-KO mice, although starting from a higher baseline level, displayed similar elevations as in control mice, and their refeeding behavior was comparable (Fig. 5I, J). Together, these results indicate that SK currents mediate dynamic changes in VTA-projecting 5-HTDRN neuron activity during non-hunger-driven feeding. We then subjected control and SK3-TPH2-KO mice to the open-field test, the light–dark test, the elevated plus maze test, and forced swim test, and found comparable locomotor activity, anxiety-like, and depression-like behaviors in these mice (Fig. S8). Thus, we suggest that the reduced non-hunger-driven feeding in these mice was not secondary to changes in animals’ locomotion and mood. Finally, we compared SK3 protein levels in the DRN before and after mice conducted non-hunger-driven feeding, and found that 2.5-h non-hunger-driven feeding significantly reduced SK3 levels in the DRN (Fig. S9).

Discussions

The primary cause of the global obesity epidemic is the overconsumption of food, which is driven by hunger-dependent and -independent mechanisms. Here, we aimed to define specific brain circuits modulating hunger-driven and non-hunger-driven feeding behavior with the objective of informing the rational design of precision medicines for the treatment of obesity. We employed a commonly used paradigm [37, 53, 54] to measure hunger-driven feeding as a rapid chow refeeding behavior induced by overnight fasting. Non-hunger-driven feeding was assessed as HFD intake in satiated mice presented with a free choice of HFD and chow diet; importantly, we only measured HFD intake for 2.5 h during the morning, instead of a longer period (e.g., 12 and 24 h) or during the late afternoon/evening, which should minimize food intake triggered by hunger or circadian cues. Of course, one limitation is that we did not measure food intake in longer periods and may have missed potential rebound in feeding. Nevertheless, using these feeding paradigms, we provide the insight into the in vivo activity of 5-HTDRN neurons in response to different types of foods and reveal that 5-HTDRN neurons progressively increase their activity during food consumption. Further, we report that activation of the 5-HTDRN → ARH projections efficiently decreased hunger-driven feeding, but had no effect on non-hunger-driven feeding, whereas activation of the 5-HTDRN → VTA projections robustly inhibited non-hunger-driven feeding, but had no effect on hunger-driven feeding.

The ARH was first demonstrated to be a necessary homeostatic feeding center through lesion studies in last century, and the past few decades have revealed the importance of POMCARH and AgRPARH neurons in mediating this effect [55, 56]. We and others have reported that 5-HT, like other hormones and neurotransmitters, regulates the activity of POMCARH and AgRPARH neurons [57,58,59]. However, the source of 5-HT sufficient to modulate the POMCARH and AgRPARH activity, and feeding behavior has not been fully clarified. Here, we report that 5-HTDRN neurons innervate around half of anorexigenic POMCARH cells via 5-HT2CRs and a subset of orexigenic AgRPARH neurons via 5-HT1BRs. We report that all 5-HTDRN-innervated POMCARH neurons were activated by 5-HT2CR-mediated 5-HTergic inputs. Notably, since these current-clamp recordings were performed in the presence of GABA and glutamate receptor inhibitors to block the majority of fast synaptic inputs, we suggest that the observed rapid activations in tested neurons likely reflect responses to direct synaptic inputs from 5-HTDRN neurons. It is striking that in the presence of GABA and glutamate receptor inhibitors, the 5-HT2CR antagonist (SB242084) completely abolished the spontaneous firing of 5-HTDRN-innervated POMCARH neurons. Thus, we suggest that the 5-HT inputs (acting through 5-HT2CRs) and glutamatergic inputs provide the majority of excitatory drive to this subset of POMCARH neurons. These findings are consistent with earlier reports that 5-HT2CR agonists activate a portion of POMCARH neurons [14, 58]. Further, deletion of 5-HT2CRs from POMC neurons leads to hyperphagia and obesity in mice [26] and 5-HT2CRs expressed by POMC neurons partly mediates anorexigenic effects of 5-HT analogs during hunger-driven feeding [60]. We also observed that all 5-HTDRN-innervated AgRPARH neurons were inhibited by 5-HT1BR-mediated 5-HTergic inputs. These results are consistent with previous reports that 5-HT1BRs inhibit AgRPARH neurons and 5-HT1BR agonists reduce hunger-driven feeding [57]. Together, these observations indicate that a subset of 5-HTDRN neurons project to the ARH, where they activate POMCARH neurons via 5-HT2CRs and inhibit AgRPARH neurons via 5-HT1BRs to reduce hunger-driven feeding. Such findings suggest that in addition to the 5-HT2CR, the 5-HT1BR may be another suitable target for the treatment of obesity. Indeed, co-administration of 5-HT2CR and 5-HT1BR agonists in animals produces a greater reduction in food intake compared to single agonist treatment [61].

A separate population of 5-HTDRN neurons innervate a subset of DAVTA neurons. DAVTA neurons are instrumental in modulating motivation and reward-related behaviors. We report that 5-HTDRN influences the activity of DAVTA neurons, and this effect is mediated via the 5-HT2CRs. In line with this finding, we previously reported that lorcaserin activates DAVTA neurons, effects that are blocked by deletion of 5-HT2CRs in DA neurons [15]. Further, deletion of 5-HT2CRs in DA neurons blocks effects of lorcaserin to inhibit binge-like eating (a feeding behavior largely driven by hedonic value of high palatable food), but not hunger-driven feeding [15]. Together, these results indicate that a subset of 5-HTDRN neurons project to and activate DAVTA neurons via 5-HT2CR-mediated mechanisms, which inhibits non-hunger-driven feeding. In addition to the VTA and ARH, 5-HT2CRs are also expressed in other brain regions, e.g., the lateral hypothalamic area and the ventral hippocampus [62], both of which are involved in feeding control [63,64,65]. The potential role of 5-HT2CRs in these brain regions warrant future investigations.

Notably, a small portion of 5-HTDRN neurons do have collateral projections to both the ARH and VTA, which may confound the circuit-specific stimulation. Nevertheless, given the drastic differences in the phenotypic outcome associated with the photostimulation of these two projections, we suggest that these collaterals likely play a minor role in our experimental settings. Thus, our findings are in line with the notion that neural circuits regulating hunger-driven feeding vs. non-hunger-driven feeding are segregated. Consistently, while animals without AgRPARH neurons can appropriately consume high palatable food [66], these neurons are dispensable for normal feeding responses induced by hunger signals, e.g., ghrelin [66] and asprosin [67]. However, it has to be pointed out that these two feeding-regulatory systems can interplay with each other. For instance, ghrelin can enhance the responses of DAVTA neurons to hedonic cues [68], whereas chronic HFD feeding reduces the responses of AgRPARH neurons to ghrelin [35]. Indeed, DAVTA neurons are reported to receive indirect inputs from AgRPARH neurons [35] and POMCARH neurons [33], which may provide anatomical basis for the crosstalk between the homeostatic and the hedonic circuitry.

In addition to 5-HT-mediated neurotransmissions, we also noted that 5-HTDRN neurons provide glutamatergic or GABAergic inputs to a portion (but not all) of innervated neurons. This is consistent with recent findings that 5-HTDRN neurons are partially overlapping with glutamatergic or GABAergic neurons [43, 44]. Notably, hunger-driven feeding can be inhibited by selective stimulation of glutamatergic neurons in the DRN or by inhibition of GABAergic neurons in the DRN [69]. We found that in the presence of glutamate and GABA receptor inhibitors, 5-HTDRN neurons can provide 5-HTergic inputs to all tested ARH neurons in the in vitro preparation. However, we cannot fully rule out the in vivo contribution of glutamatergic and/or GABAergic outputs from 5-HTDRN neurons to the regulation of hunger-driven feeding, as well as non-hunger-driven feeding. Future studies are warranted to further dissect out the contributions of 5-HT signaling, vs. glutamate or GABA co-released by 5-HTDRN neurons, to the regulation of feeding behavior.

It has been reported that 5-HTDRN neurons can be transiently (on a time scale of seconds) activated by various stimuli, e.g., sex, social interaction, sucrose, or food pellet intake [40, 43]. Consistently, we observed transient activations of 5-HTDRN neurons associated with onset of eating bouts (either chow or HFD). However, we found that these eating bout-associated transient elevations in 5-HTDRN neurons persisted in γ2-TPH2-KO and SK3-TPH2-KO mice. Given these mutant mouse models displayed altered hunger-driven or non-hunger-driven feeding, we suggest that the transient elevations in 5-HTDRN neurons do not have regulatory effects on feeding behavior per se, although they may encode certain aspects of eating, e.g., initiation of an eating bout or detection of food. On the other hand, we found that 5-HTDRN neurons also chronically increased their activities during the course of feeding on a time scale of minutes to hours. Considering that activation of 5-HTDRN neurons inhibits feeding, we suggest that 5-HTDRN neurons function as a key component of a negative feedback loop. Thus, feeding, either driven by hunger or by hunger-independent cues, gradually activates 5-HTDRN neurons, and this slowly elevated 5-HTDRN neuron activity then terminates the meal to prevent overeating. Notably, meal termination can be regulated by other neural populations, including calcitonin gene-related peptide-expressing neurons in the external lateral parabrachial nucleus [70] and catecholamine neurons in the nucleus of solitary tract [71]. It remains unclear whether these neural nodes, including 5-HTDRN neurons, function in a parallel fashion, or within an interconnected network to regulate meal termination.

The slow dynamic changes in 5-HTDRN neurons are not associated with presynaptic alternations (e.g., changes in frequencies of mEPSC and mIPSC), which could happen in seconds. Rather, distinct postsynaptic mechanisms are involved in the regulations of 5-HTDRN neuron activity during hunger-driven vs. non-hunger-driven feeding. On one hand, hunger-driven feeding increased ARH-projecting 5-HTDRN neuron activity by reducing the amplitude of mIPSC. On the other hand, non-hunger-driven feeding activated VTA-projecting 5-HTDRN neurons by reducing a SK outward potassium current. It is worth noting that comparable mIPSCs were detected in ARH- and VTA-projecting 5-HTDRN neurons, although only the mIPSC in the former subset was regulated by hunger-driven feeding. Similarly, comparable SK currents were detected in ARH- and VTA-projecting 5-HTDRN neurons, although only the SK current in the latter subset was regulated by non-hunger-driven feeding. Thus, while both these currents exist in the two subsets of 5-HTDRN neurons, their responses to hunger-driven feeding and non-hunger-driven feeding may involve different regulatory mechanisms which remain unclear. Since 5-HTDRN neuron activity was correlated with the amount of chow or HFD intake, we speculate that the consumed nutrients, calories, and/or the nutritional state of the animal may contribute to the slow dynamic changes in 5-HTDRN neurons, although such possibilities need to be further validated. In addition, our observations that non-hunger-driven feeding can reduce SK3 protein levels in the DRN suggest that carbohydrates and/or lipids in the diets may regulate levels of ion channels, and therefore influence neural activity.

It is worth mentioning that the role of 5-HTDRN neurons in the reward-related behaviors are complex. Correia et al. reported no place preference or aversion associated with repetitive place-contingent pairing of optogenetic activation of 5-HTDRN neurons [72]. However, multiple groups demonstrated that optogenetic activation of 5-HTDRN neurons or the 5-HTDRN → VTA projections increase place preference and other reward-related behaviors [73, 74]. Consistent with the latter, here we found that activation of the 5-HTDRN → VTA projections increased place preference in satiated mice, which was, however, totally reversed in hungry mice. On the other hand, activation of the 5-HTDRN → ARH projections increased place preference only in hungry mice, but not in satiated mice. Thus, our results add to this complexity and suggest that 5-HTDRN neurons play different roles in regulating place preference depending on the different downstream neural circuits and on the nutritional states of animals. Similarly, a circuit from the paraventricular hypothalamus to the lateral parabrachial nucleus has been reported to increase place preference only in hungry mice, an effect that is completely blocked in satiated mice [75].

Owing to the diverse etiologies of obesity, it has been difficult to treat with medication and is therefore ideally suited to a precision medication approach. Here, we investigated specific brain regions underpinning different types of feeding behaviors. We report that hunger-driven and non-hunger-driven feeding can gradually activate 5-HTDRN neurons, via either inhibiting responsiveness to inhibitory GABAergic inputs or by reducing a potassium outward current. Activated 5-HTDRN neurons in turn inhibit hunger-driven and non-hunger-driven feeding through their projections to the ARH and VTA, respectively. These 5-HTDRN-originated neural circuits constitute a negative feedback loop to trigger the end of a meal. These findings suggest that individuals with increased hunger could benefit from a medication targeting the 5-HTDRN → ARH circuit, individuals who overeat in the absence of hunger could benefit from a medication targeting the 5-HTDRN → VTA circuit.

References

  1. 1.

    Alonso-Alonso M, Woods SC, Pelchat M, Grigson PS, Stice E, Farooqi S, et al. Food reward system: current perspectives and future research needs. Nutr Rev. 2015;73:296–307.

    PubMed  PubMed Central  Article  Google Scholar 

  2. 2.

    Kenny PJ. Reward mechanisms in obesity: new insights and future directions. Neuron. 2011;69:664–79.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  3. 3.

    Lechin F, van der Dijs B, Hernandez-Adrian G. Dorsal raphe vs. median raphe serotonergic antagonism. Anatomical, physiological, behavioral, neuroendocrinological, neuropharmacological and clinical evidences: relevance for neuropharmacological therapy. Prog Neuropsychopharmacol Biol Psychiatry. 2006;30:565–85.

    CAS  PubMed  Article  Google Scholar 

  4. 4.

    De Fanti BA, Hamilton JS, Horwitz BA. Meal-induced changes in extracellular 5-HT in medial hypothalamus of lean (Fa/Fa) and obese (fa/fa) Zucker rats. Brain Res. 2001;902:164–70.

    PubMed  Article  Google Scholar 

  5. 5.

    Rowland NE, Carlton J. Neurobiology of an anorectic drug: fenfluramine. Prog Neurobiol. 1986;27:13–62.

    CAS  PubMed  Article  Google Scholar 

  6. 6.

    Foltin RW, Moran TH. Food intake in baboons: effects of a long-acting cholecystokinin analog. Appetite. 1989;12:145–52.

    CAS  PubMed  Article  Google Scholar 

  7. 7.

    McGuirk J, Goodall E, Silverstone T, Willner P. Differential effects of d-fenfluramine, l-fenfluramine and d-amphetamine on the microstructure of human eating behaviour. Behav Pharm. 1991;2:113–9.

    Article  Google Scholar 

  8. 8.

    Rogers PJ, Blundell JE. Effect of anorexic drugs on food intake and the micro-structure of eating in human subjects. Psychopharmacology. 1979;66:159–65.

    CAS  PubMed  Article  Google Scholar 

  9. 9.

    Blundell JE, Leshem MB. Central action of anorexic agents: effects of amphetamine and fenfluramine in rats with lateral hypothalamic lesions. Eur J Pharm. 1974;28:81–8.

    CAS  Article  Google Scholar 

  10. 10.

    Geyer MA, Puerto A, Menkes DB, Segal DS, Mandell AJ. Behavioral studies following lesions of the mesolimbic and mesostriatal serotonergic pathways. Brain Res. 1976;106:257–69.

    CAS  PubMed  Article  Google Scholar 

  11. 11.

    Ghosh MN, Parvathy S. The effect of cyproheptadine on water and food intake and on body weight in the fasted adult and weanling rats. Br J Pharm. 1973;48:328P–9P.

    CAS  Google Scholar 

  12. 12.

    Saller CF, Stricker EM. Hyperphagia and increased growth in rats after intraventricular injection of 5,7-dihydroxytryptamine. Science. 1976;192:385–7.

    CAS  PubMed  Article  Google Scholar 

  13. 13.

    Smith SR, Weissman NJ, Anderson CM, Sanchez M, Chuang E, Stubbe S, et al. Multicenter, placebo-controlled trial of lorcaserin for weight management. N. Engl J Med. 2010;363:245–56.

    CAS  PubMed  Article  Google Scholar 

  14. 14.

    D'agostino G, Lyons D, Cristiano C, Lettieri M, Olarte-Sanchez C, Burke LK, et al. Nucleus of the solitary tract serotonin 5-HT2C receptors modulate food intake. Cell Metab. 2018;28:619–30.e615.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  15. 15.

    Xu P, He Y, Cao X, Valencia-Torres L, Yan X, Saito K, et al. Activation of serotonin 2C receptors in dopamine neurons inhibits binge-like eating in mice. Biol Psychiatry. 2017;81:737–47.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  16. 16.

    Blumenthal SA, Pratt WE. d-Fenfluramine and lorcaserin inhibit the binge-like feeding induced by mu-opioid receptor stimulation of the nucleus accumbens in the rat. Neurosci Lett. 2018;687:43–8.

    CAS  PubMed  Article  Google Scholar 

  17. 17.

    Connolly HM, Crary JL, McGoon MD, Hensrud DD, Edwards BS, Edwards WD, et al. Valvular heart disease associated with fenfluramine-phentermine. N. Engl J Med. 1997;337:581–8.

    CAS  PubMed  Article  Google Scholar 

  18. 18.

    FDA. Safety clinical trial shows possible increased risk of cancer with weight-loss medicine Belviq, Belviq XR (lorcaserin). 2020. https://www.fda.gov/drugs/drug-safety-and-availability/safety-clinical-trial-shows-possible-increased-risk-cancer-weight-loss-medicine-belviq-belviq-xr. Acessed 13 Feb 2020.

  19. 19.

    Elmquist JK, Coppari R, Balthasar N, Ichinose M, Lowell BB. Identifying hypothalamic pathways controlling food intake, body weight, and glucose homeostasis. J Comp Neurol. 2005;493:63–71.

    CAS  PubMed  Article  Google Scholar 

  20. 20.

    Morton GJ, Cummings DE, Baskin DG, Barsh GS, Schwartz MW. Central nervous system control of food intake and body weight. Nature. 2006;443:289–95.

    CAS  PubMed  Article  Google Scholar 

  21. 21.

    Stuber GD, Hopf FW, Tye KM, Chen BT, Bonci A. Neuroplastic alterations in the limbic system following cocaine or alcohol exposure. Curr Top Behav Neurosci. 2010;3:3–27.

    PubMed  Article  Google Scholar 

  22. 22.

    Epstein DH, Preston KL, Stewart J, Shaham Y. Toward a model of drug relapse: an assessment of the validity of the reinstatement procedure. Psychopharmacology. 2006;189:1–16.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  23. 23.

    Pignatelli M, Bonci A. Role of dopamine neurons in reward and aversion: a synaptic plasticity perspective. Neuron. 2015;86:1145–57.

    CAS  PubMed  Article  Google Scholar 

  24. 24.

    Szczypka MS, Kwok K, Brot MD, Marck BT, Matsumoto AM, Donahue BA, et al. Dopamine production in the caudate putamen restores feeding in dopamine-deficient mice. Neuron. 2001;30:819–28.

    CAS  PubMed  Article  Google Scholar 

  25. 25.

    Denis RGP, Joly-Amado A, Webber E, Langlet F, Schaeffer M, Padilla SL, et al. Palatability can drive feeding independent of AgRP neurons. Cell Metab. 2017;25:975.

    PubMed  Article  Google Scholar 

  26. 26.

    Berglund ED, Liu C, Sohn JW, Liu T, Kim MH, Lee CE, et al. Serotonin 2C receptors in pro-opiomelanocortin neurons regulate energy and glucose homeostasis. J Clin Investig. 2013;123:5061–70.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  27. 27.

    Kocaman N, Kuloglu T. Expression of asprosin in rat hepatic, renal, heart, gastric, testicular and brain tissues and its changes in a streptozotocin-induced diabetes mellitus model. Tissue Cell. 2020;66:101397.

    CAS  PubMed  Article  Google Scholar 

  28. 28.

    Unger EK, Keller JP, Altermatt M, Liang R, Matsui A, Dong C, et al. Directed evolution of a selective and sensitive serotonin sensor via machine learning. Cell. 2020;183:1986–2002.e1926.

    CAS  PubMed  Article  Google Scholar 

  29. 29.

    Tong Q, Ye CP, Jones JE, Elmquist JK, Lowell BB. Synaptic release of GABA by AgRP neurons is required for normal regulation of energy balance. Nat Neurosci. 2008;11:998–1000.

  30. 30.

    Madisen L, Zwingman TA, Sunkin SM, Oh SW, Zariwala HA, Gu H, et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat Neurosci. 2010;13:133–40.

    CAS  PubMed  Article  Google Scholar 

  31. 31.

    Schweizer C, Balsiger S, Bluethmann H, Mansuy IM, Fritschy JM, Mohler H, et al. The gamma 2 subunit of GABA(A) receptors is required for maintenance of receptors at mature synapses. Mol Cell Neurosci. 2003;24:442–50.

    CAS  PubMed  Article  Google Scholar 

  32. 32.

    Deignan J, Luján R, Bond C, Riegel A, Watanabe M, Williams JT, et al. SK2 and SK3 expression differentially affect firing frequency and precision in dopamine neurons. Neuroscience. 2012;217:67–76.

    CAS  PubMed  Article  Google Scholar 

  33. 33.

    Qu N, He Y, Wang C, Xu P, Yang Y, Cai X, et al. A POMC-originated circuit regulates stress-induced hypophagia, depression, and anhedonia. Mol Psychiatry. 2020;25:1006–21.

  34. 34.

    Morecroft I, White K, Caruso P, Nilsen M, Loughlin L, Alba R, et al. Gene therapy by targeted adenovirus-mediated knockdown of pulmonary endothelial Tph1 attenuates hypoxia-induced pulmonary hypertension. Mol Ther. 2012;20:1516–28.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  35. 35.

    Mazzone CM, Liang-Guallpa J, Li C, Wolcott NS, Boone MH, Southern M, et al. High-fat food biases hypothalamic and mesolimbic expression of consummatory drives. Nat Neurosci. 2020;23:1253–66.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  36. 36.

    Cao X, Xu P, Oyola MG, Xia Y, Yan X, Saito K, et al. Estrogens stimulate serotonin neurons to inhibit binge-like eating in mice. J Clin Invest. 2014;124:4351–62.

    PubMed  PubMed Central  Article  Google Scholar 

  37. 37.

    Liu T, Kong D, Shah BP, Ye C, Koda S, Saunders A, et al. Fasting activation of AgRP neurons requires NMDA receptors and involves spinogenesis and increased excitatory tone. Neuron. 2012;73:511–22.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  38. 38.

    Fenselau H, Campbell JN, Verstegen AM, Madara JC, Xu J, Shah BP, et al. A rapidly acting glutamatergic ARC—>PVH satiety circuit postsynaptically regulated by alpha-MSH. Nat Neurosci. 2017;20:42–51.

    CAS  PubMed  Article  Google Scholar 

  39. 39.

    Li C, Navarrete J, Liang-Guallpa J, Lu C, Funderburk SC, Chang RB, et al. Defined paraventricular hypothalamic populations exhibit differential responses to food contingent on caloric state. Cell Metab. 2019;29:681–94 e685.

  40. 40.

    Li Y, Zhong W, Wang D, Feng Q, Liu Z, Zhou J, et al. Serotonin neurons in the dorsal raphe nucleus encode reward signals. Nat Commun. 2016;7:10503.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  41. 41.

    Lerner TN, Shilyansky C, Davidson TJ, Evans KE, Beier KT, Zalocusky KA, et al. Intact-brain analyses reveal distinct information carried by SNc dopamine subcircuits. Cell. 2015;162:635–47.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  42. 42.

    Leinninger GM, Opland DM, Jo YH, Faouzi M, Christensen L, Cappellucci LA, et al. Leptin action via neurotensin neurons controls orexin, the mesolimbic dopamine system and energy balance. Cell Metab. 2011;14:313–23.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  43. 43.

    Ren J, Friedmann D, Xiong J, Liu CD, Ferguson BR, Weerakkody T, et al. Anatomically defined and functionally distinct dorsal raphe serotonin sub-systems. Cell. 2018;175:472–87.e420.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  44. 44.

    Ren J, Isakova A, Friedmann D, Zeng J, Grutzner SM, Pun A, et al. Single-cell transcriptomes and whole-brain projections of serotonin neurons in the mouse dorsal and median raphe nuclei. Elife. 2019;8:8.

    Article  Google Scholar 

  45. 45.

    Zhan C, Zhou J, Feng Q, Zhang JE, Lin S, Bao J, et al. Acute and long-term suppression of feeding behavior by POMC neurons in the brainstem and hypothalamus, respectively. J Neurosci. 2013;33:3624–32.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  46. 46.

    Aponte Y, Atasoy D, Sternson SM. AGRP neurons are sufficient to orchestrate feeding behavior rapidly and without training. Nat Neurosci. 2011;14:351–5.

    CAS  PubMed  Article  Google Scholar 

  47. 47.

    Krashes MJ, Koda S, Ye C, Rogan SC, Adams AC, Cusher DS, et al. Rapid, reversible activation of AgRP neurons drives feeding behavior in mice. J Clin Investig. 2011;121:1424–8.

  48. 48.

    Fink KB, Gothert M. 5-HT receptor regulation of neurotransmitter release. Pharm Rev. 2007;59:360–417.

    CAS  PubMed  Article  Google Scholar 

  49. 49.

    Atasoy D, Betley JN, Su HH, Sternson SM. Deconstruction of a neural circuit for hunger. Nature. 2012;488:172–7.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  50. 50.

    Lorez M, Benke D, Luscher B, Mohler H, Benson JA. Single-channel properties of neuronal GABAA receptors from mice lacking the 2 subunit. J Physiol. 2000;527:11–31.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  51. 51.

    Wulff P, Goetz T, Leppä E, Linden AM, Renzi M, Swinny JD, et al. From synapse to behavior: rapid modulation of defined neuronal types with engineered GABAA receptors. Nat Neurosci. 2007;10:923–9.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  52. 52.

    Stocker M, Pedarzani P. Differential distribution of three Ca(2+)-activated K(+) channel subunits, SK1, SK2, and SK3, in the adult rat central nervous system. Mol Cell Neurosci. 2000;15:476–93.

  53. 53.

    He Y, Shu G, Yang Y, Xu P, Xia Y, Wang C, et al. A small potassium current in AgRP/NPY neurons regulates feeding behavior and energy metabolism. Cell Rep. 2016;17:1807–18.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  54. 54.

    Kong D, Dagon Y, Campbell JN, Guo Y, Yang Z, Yi X, et al. A postsynaptic AMPK—>p21-activated kinase pathway drives fasting-induced synaptic plasticity in AgRP. Neurons Neuron. 2016;91:25–33.

    CAS  PubMed  Article  Google Scholar 

  55. 55.

    Elmquist JK, Elias CF, Saper CB. From lesions to leptin: hypothalamic control of food intake and body weight. Neuron. 1999;22:221–32.

    CAS  PubMed  Article  Google Scholar 

  56. 56.

    Heisler LK, Lam DD. An appetite for life: brain regulation of hunger and satiety. Curr Opin Pharm. 2017;37:100–6.

    CAS  Article  Google Scholar 

  57. 57.

    Heisler LK, Jobst EE, Sutton GM, Zhou L, Borok E, Thornton-Jones Z, et al. Serotonin reciprocally regulates melanocortin neurons to modulate food intake. Neuron. 2006;51:239–49.

    CAS  PubMed  Article  Google Scholar 

  58. 58.

    Heisler LK, Cowley MA, Tecott LH, Fan W, Low MJ, Smart JL, et al. Activation of central melanocortin pathways by fenfluramine. Science. 2002;297:609–11.

    CAS  PubMed  Article  Google Scholar 

  59. 59.

    Xu Y, Berglund ED, Sohn JW, Holland WL, Chuang JC, Fukuda M, et al. 5-HT2CRs expressed by pro-opiomelanocortin neurons regulate insulin sensitivity in liver. Nat Neurosci. 2010;13:1457–9.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  60. 60.

    Xu Y, Jones JE, Kohno D, Williams KW, Lee CE, Choi MJ, et al. 5-HT2CRs expressed by pro-opiomelanocortin neurons regulate energy homeostasis. Neuron. 2008;60:582–9.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  61. 61.

    Doslikova B, Garfield AS, Shaw J, Evans ML, Burdakov D, Billups B, et al. 5-HT2C receptor agonist anorectic efficacy potentiated by 5-HT1B receptor agonist coapplication: an effect mediated via increased proportion of pro-opiomelanocortin neurons activated. J Neurosci. 2013;33:9800–4.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  62. 62.

    Molineaux SM, Jessell TM, Axel R, Julius D. 5-HT1c receptor is a prominent serotonin receptor subtype in the central nervous system. Proc Natl Acad Sci USA. 1989;86:6793–7.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  63. 63.

    Jennings JH, Rizzi G, Stamatakis AM, Ung RL, Stuber GD. The inhibitory circuit architecture of the lateral hypothalamus orchestrates feeding. Science. 2013;341:1517–21.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  64. 64.

    Wu Z, Kim ER, Sun H, Xu Y, Mangieri LR, Li DP, et al. GABAergic projections from lateral hypothalamus to paraventricular hypothalamic nucleus promote feeding. J Neurosci. 2015;35:3312–8.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  65. 65.

    Sweeney P, Yang Y. An excitatory ventral hippocampus to lateral septum circuit that suppresses feeding. Nat Commun. 2015;6:10188.

    CAS  PubMed  Article  Google Scholar 

  66. 66.

    Denis RG, Joly-Amado A, Webber E, Langlet F, Schaeffer M, Padilla SL, et al. Palatability can drive feeding independent of AgRP neurons. Cell Metab. 2015;22:646–57.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  67. 67.

    Duerrschmid C, He Y, Wang C, Li C, Bournat JC, Romere C, et al. Asprosin is a centrally acting orexigenic hormone. Nat Med. 2017;23:1444–53.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  68. 68.

    Cone JJ, McCutcheon JE, Roitman MF. Ghrelin acts as an interface between physiological state and phasic dopamine signaling. J Neurosci. 2014;34:4905–13.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  69. 69.

    Nectow AR, Schneeberger M, Zhang H, Field BC, Renier N, Azevedo E, et al. Identification of a brainstem circuit controlling feeding. Cell. 2017;170:429–42.e411.

    CAS  PubMed  Article  Google Scholar 

  70. 70.

    Campos CA, Bowen AJ, Schwartz MW, Palmiter RD. Parabrachial CGRP neurons control meal termination. Cell Metab. 2016;23:811–20.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  71. 71.

    Aklan I, Sayar Atasoy N, Yavuz Y, Ates T, Coban I, Koksalar F, et al. NTS catecholamine neurons mediate hypoglycemic hunger via medial hypothalamic feeding pathways. Cell Metab. 2020;31:313–26.e315.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  72. 72.

    Correia PA, Lottem E, Banerjee D, Machado AS, Carey MR, Mainen ZF. Transient inhibition and long-term facilitation of locomotion by phasic optogenetic activation of serotonin neurons. Elife. 2017;6:6.

    Article  Google Scholar 

  73. 73.

    Nagai Y, Takayama K, Nishitani N, Andoh C, Koda M, Shirakawa H, et al. The role of dorsal raphe serotonin neurons in the balance between reward and aversion. Int J Mol Sci. 2020;21:6.

    Google Scholar 

  74. 74.

    Liu Z, Zhou J, Li Y, Hu F, Lu Y, Ma M, et al. Dorsal raphe neurons signal reward through 5-HT and glutamate. Neuron. 2014;81:1360–74.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  75. 75.

    Garfield AS, Li C, Madara JC, Shah BP, Webber E, Steger JS, et al. A neural basis for melanocortin-4 receptor-regulated appetite. Nat Neurosci. 2015;18:863–71.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

Download references

Acknowledgements

The investigators were supported by grants from the NIH (R01DK114279, R01DK109934, and R21NS108091 to QT; R01ES027544 and R01DK111436 to ZS; R00DK107008 to PX; R01DK109194 and R56DK109194 to QW; P01DK113954, R01DK115761, R01DK117281, and R01DK125480 to YX; R01DK120858 to QT and YX; K01DK119471 to CW; and P20GM135002 to YH), USDA/CRIS (51000-064-01 S to YX and QW), American Diabetes Association (1-17-PDF-138 to YH, 7-13-JF-61 to QW, and 1-15-BS-184 to QT), American Heart Association awards (16POST27260254 to CW), the Pew Charitable Trust awards to QW (0026188), Baylor Collaborative Faculty Research Investment Program grants to QW, the Faculty Start-up grants from USDA/ARS to QW, the Biotechnology and Biological Sciences Research Council (BB/K001418/1 and BB/NO17838/1 to LKH), and the Medical Research Council (MC/PC/15077 to LKH). QW is the Pew Scholar of Biomedical Sciences and the Kavli Scholar. The anxiety tests (e.g., open-field test, light–dark test, and elevated plus maze test) were performed in the Mouse Neurobehavior Core, Baylor College of Medicine, which was supported by National Institutes of Health Grant No. P30HD024064. The Ad-iN/WED virus was kindly provided by Dr. Martin Myers (University of Michigan). The AAV9-CBA-DIO-WGA-zsGreen virus was kindly provided by Dr. Richard Palmiter (University of Washington).

Author information

Affiliations

Authors

Contributions

YH and XC were involved in experimental design and most of procedures, data acquisition and analyses, and writing the manuscript. HL, KMC, PX, YL, CW, MY, YH, HL, CL, TY, YY, KY, JW, RZ, and CZ assisted in surgical procedures and production of study mice. FL, ZS, LH, QW, and QT were involved in study design and writing the manuscript. GS and YX the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Corresponding authors

Correspondence to Gang Shu or Yong Xu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

He, Y., Cai, X., Liu, H. et al. 5-HT recruits distinct neurocircuits to inhibit hunger-driven and non-hunger-driven feeding. Mol Psychiatry (2021). https://doi.org/10.1038/s41380-021-01220-z

Download citation

Search

Quick links