Leptin-activated hypothalamic BNC2 neurons acutely suppress food intake

The study addresses how leptin regulates appetite beyond the canonical arcuate nucleus (ARC) neuron populations of orexigenic AGRP/NPY and anorexigenic POMC neurons. Although these two populations are thought to function in a yin–yang manner to control feeding, several discrepancies exist: acute POMC activation minimally suppresses feeding, AGRP/NPY activation produces rapid feeding and negative valence, and leptin receptor (LepR) deletion in AGRP/NPY neurons causes severe obesity whereas LepR deletion in adult POMC neurons has minimal effects. AGRP/NPY activation also acutely affects glucose metabolism, unlike POMC. These observations suggest an unidentified leptin-responsive population that rapidly suppresses appetite. The authors hypothesize that a distinct LepR-expressing GABAergic neuron population in the ARC exists that acutely suppresses food intake and mediates aspects of leptin’s action.

Prior work established: (1) AGRP/NPY neuron activation rapidly induces feeding and conveys negative valence, while POMC activation produces limited acute anorexia and can also be associated with negative valence; (2) Genetic defects in POMC processing or MC4R cause obesity, but AGRP/NPY mutations have not consistently altered body weight, though adult ablation of AGRP/NPY neurons causes anorexia; (3) LepR deletion in AGRP/NPY neurons results in extreme obesity, whereas deletion in adult POMC neurons has modest effects; (4) AGRP/NPY neurons acutely alter glucose metabolism, POMC neurons do not; (5) Prior analyses suggested additional GABAergic LepR neuron populations critical for energy balance. Together, these studies imply the existence of a missing, fast-acting leptin-regulated satiety neuron population distinct from AGRP/NPY and POMC.

• Single-nucleus RNA sequencing (snRNA-seq): Microdissected mouse ARC from adult males; neuronal nuclei were FACS-enriched using anti-NeuN and sequenced with the 10x platform. 3,557 cells profiled, 3,481 neurons identified by neuronal markers. Leiden clustering identified 21 clusters; ARC identity validated using known markers. LepR expression examined across clusters.
• In situ hybridization (ISH): Multiplex RNA ISH in adult mouse ARC for Lepr, Bnc2, Agrp, Pomc, Slc32a1 to assess co-localization; analysis of human ARC single-cell data for Lepr/Bnc2 co-localization.
• Mouse genetics: Generated BNC2-P2A-iCre knock-in (BNC2-Cre) mice. Verified Cre expression by ARC AAV-DIO-mCherry and ISH.
• Leptin signaling assays: pSTAT3 immunostaining in BNC2 neurons after leptin or refeeding; FOS immunostaining following refeeding or leptin.
• Electrophysiology: Whole-cell patch clamp from ARC BNC2 neurons expressing AAV-DIO-GFP; applied leptin (100 nM) to slices from fasted or ad libitum-fed mice; measured depolarization and AP firing; reversibility with washout.
• In vivo calcium imaging: Fiber photometry in BNC2-Cre mice expressing AAV-DIO-GCaMP6s in ARC; recorded responses to inedible object, chow, or peanut butter (PB) under fasted vs fed conditions; assessed sensory vs consumption phases including caged-food (sensory-only) paradigms and naïve exposure to new foods; tested activity decay upon food removal.
• Chemogenetics: AAV-DIO-hM3Dq (stimulatory) or AAV-DIO-hM4Di (inhibitory) DREADDs targeted to ARC BNC2 neurons; CNO vs PBS injections at dark onset or light cycle; measured food intake and body weight.
• Optogenetics: AAV-DIO-ChR2-GFP (activation) or soma-targeted stGtACR2-FusionRed (inhibition) in ARC with optic fiber implants; 20 Hz, 20 min stimulation/inhibition; assessed food intake dynamics and locomotion; real-time place preference in hungry vs sated states.
• Circuit mapping (CRACM): In BNC2-Cre::NPY-FlpO mice, co-injected AAV-DIO-ChR2-GFP (BNC2) + AAV-fDIO-mCherry (NPY) to test BNC2→AGRP/NPY connections; recorded optogenetically evoked IPSCs in AGRP/NPY neurons and pharmacologically dissected with TTX, 4-AP, and picrotoxin. Tested reciprocal AGRP/NPY→BNC2 connectivity using AAV-fDIO-ChR2-GFP (NPY) + AAV-DIO-mCherry (BNC2).
• Functional interaction in vivo: Simultaneous BNC2 inhibition (hM4Di) and AGRP/NPY GCaMP6s fiber photometry during chow presentation to test if BNC2 activity mediates food-cue suppression of AGRP/NPY activity.
• LepR loss-of-function: BNC2-Cre x LSL-Cas9-GFP mice received bilateral ARC AAV carrying sgRNAs targeting Lepr (sgLepr) or control (sgCtrl); validated by loss of leptin-induced pSTAT3 in BNC2 neurons. Monitored body weight, composition, intake, energy expenditure (EE), respiratory exchange ratio (RER), locomotion, and responses on chow and high-fat diet (HFD).
• BNC2 neuron ablation: AAV-mCherry-flex-dtA vs AAV-DIO-mCherry controls in ARC; tested acute anorectic responses to leptin and semaglutide (GLP-1R agonist).
• Metabolic assays: Fasting glucose; glucose tolerance tests (GTT) and insulin tolerance tests (ITT) following chemogenetic activation or inhibition of BNC2 neurons to assess acute effects on glucose homeostasis independent of body weight.

• Identification of BNC2/LepR neurons: snRNA-seq revealed a distinct LepR-expressing ARC neuron cluster (Cluster 17) marked by Bnc2, separate from AGRP/NPY (Cluster 0) and POMC (Cluster 3). Cluster sizes: AGRP n=650, POMC n=271, BNC2 n=35. Relative LepR expression: AGRP 1.32, POMC 0.52, BNC2 1.09.
• Marker co-expression (mouse ARC, ISH): Among Bnc2+ cells, 90.4% co-express Lepr; 96.6% co-express Slc32a1 (GABAergic); only 2.6% and 3.3% co-express Agrp or Pomc, respectively. In humans, Lepr/Bnc2 co-localization was significant (P=0.0218), and 82.1% (23/28) of Lepr/Bnc2 neurons did not co-localize with Agrp, Npy, or Pomc.
• Leptin and feeding activate BNC2 neurons: Refeeding or leptin increased pSTAT3 and FOS in BNC2 neurons. Patch clamp showed leptin depolarized and increased firing in BNC2 neurons; responsiveness was higher after fasting (responders: 16/17 fasted vs 7/12 fed), and effects were reversible upon washout.
• Fast sensory responses: In fasted mice, BNC2 neuronal activity increased within seconds to food presentation (chow, PB), scaled with palatability (PB>chow), and further increased during consumption. In fed mice, responses were attenuated (PB still effective). Sensory-only exposure (caged food) elicited significant but smaller/shorter responses. Naïve exposure to novel palatable foods did not elicit pre-ingestive activation until consumption began. Removal of food rapidly extinguished activity, while continued presence maintained activation.
• Behavioral control of feeding: Chemogenetic activation (hM3Dq) of BNC2 neurons reduced food intake and body weight; inhibition (hM4Di) increased both. Optogenetic activation (ChR2, 20 Hz, 20 min) acutely reduced intake in fasted mice with suppression persisting up to 20 min after stimulation ended; optogenetic inhibition (stGtACR2) increased intake during light cycle. Locomotion was unaffected by activation.
• Positive valence in hunger: Real-time place preference revealed preference for the BNC2-activation–paired chamber in fasted mice, but not in sated mice.
• Monosynaptic inhibition of AGRP/NPY: CRACM showed BNC2→AGRP/NPY inhibitory monosynaptic connections in ~81% of AGRP/NPY neurons (25/31) with short latency (~5.12 ms). oIPSCs were abolished by TTX, restored by 4-AP, and blocked by GABAA antagonist picrotoxin, indicating direct GABAergic inhibition. Conversely, AGRP/NPY→BNC2 connections were not detected (0/25).
• Mediating sensory suppression of AGRP/NPY: Inhibiting BNC2 neurons blunted the typical reduction in AGRP/NPY calcium activity upon chow presentation, indicating BNC2 neurons convey part of the sensory input that suppresses AGRP/NPY activity during refeeding.
• LepR in BNC2 neurons is required for energy balance: CRISPR knockout of Lepr in BNC2 neurons increased body weight and fat mass on chow, raised daily intake, increased EE (not significant after body weight regression), increased RER, and did not alter locomotion. On HFD, sgLepr mice gained more weight and fat mass and ate more. Before obesity onset, leptin reduced intake less in sgLepr than sgCtrl mice, showing BNC2 LepR is required for full anorectic leptin response. Ablation of BNC2 neurons diminished leptin’s anorectic effect but did not alter the anorectic effect of semaglutide.
• Glucose homeostasis: BNC2 LepR knockout mice had elevated fasting glucose, impaired GTT, and reduced insulin sensitivity. Chemogenetic activation of BNC2 neurons acutely decreased glucose and improved GTT/ITT; inhibition increased glucose and impaired GTT/ITT, indicating BNC2 neurons acutely regulate peripheral glucose independent of body weight.

The findings identify BNC2-expressing, LepR-positive, GABAergic neurons in the ARC as a previously unrecognized node in the leptin-regulated feeding circuitry. These neurons are rapidly activated by leptin and food-related sensory and ingestive signals, directly inhibit AGRP/NPY neurons via GABAA receptors, and bidirectionally modulate food intake and glucose metabolism. Their rapid, sustained satiety effect after activation mirrors the sustained hunger caused by AGRP/NPY activation, but in the opposite direction. Unlike POMC neurons, BNC2 neuron activation robustly and acutely suppresses feeding and confers positive valence during hunger, and LepR deletion restricted to BNC2 neurons produces marked hyperphagia and obesity. In vivo, BNC2 neurons partially mediate the food-cue–induced suppression of AGRP/NPY activity, suggesting they integrate associative sensory inputs with interoceptive leptin signals; however, direct sensory inputs to AGRP/NPY likely also contribute. The distinction from TRH ARC neurons (which are GLP-1–regulated rather than leptin-regulated) underscores functional heterogeneity among non-AGRP, non-POMC ARC LepR neurons. The data position BNC2 neurons as critical fast-acting satiety neurons that fill the hypothesized missing counterpart to AGRP/NPY neurons in leptin’s control of energy balance.

This work discovers and functionally characterizes BNC2-expressing LepR neurons in the ARC as a fast-acting, leptin-responsive population that acutely suppresses feeding by monosynaptically inhibiting AGRP/NPY neurons. Activation produces positive valence during hunger and improves glucose homeostasis; inhibition or LepR deletion causes hyperphagia, obesity, and impaired glucose control. BNC2 neurons thus constitute a key cellular component of leptin’s homeostatic regulation of energy balance and metabolism. Future research should: (1) map afferent sensory circuits driving BNC2 neurons; (2) dissect the contribution of candidate neuropeptides co-expressed in BNC2 neurons alongside GABA; (3) perform Bnc2 gene–specific manipulations within BNC2/LepR neurons to test roles in development/function; (4) determine downstream projection targets beyond AGRP/NPY; and (5) explore pharmacologic strategies to selectively activate BNC2 neurons to treat obesity and mitigate the negative valence of hunger.

• Species and translational scope: Findings are primarily in mice; human co-localization data are correlative and limited.
• Cell numbers: BNC2 cluster size was relatively small (n=35), which may limit detection of heterogeneity within this population.
• Circuit incompleteness: While BNC2→AGRP/NPY inhibition is established, the upstream sensory sources and broader efferent targets of BNC2 neurons remain unmapped.
• Partial mediation of sensory effects: Inhibiting BNC2 only partially blunted food-cue suppression of AGRP/NPY activity, indicating parallel pathways not defined here.
• Marker versus effector: Bnc2 was used as a marker gene; its causal role in neuronal development/function is not yet determined.
• Temporal dynamics: The duration of sustained satiety beyond ~20 minutes post-activation was not fully characterized.
• Chemogenetic/optogenetic constraints: Artificial activation/inhibition may not perfectly mimic physiological patterns; potential off-target effects of CNO or light cannot be completely excluded.