A brainstem to hypothalamic arcuate nucleus GABAergic circuit drives feeding

Abstract

The obesity epidemic is principally driven by the consumption of more calories than the body requires. It is therefore essential that the mechanisms underpinning feeding behavior are defined. Neurons within the brainstem dorsal vagal complex (DVC) receive direct information from the digestive system and project to second-order regions in the brain to regulate food intake. Although γ-aminobutyric acid is expressed in the DVC (GABA^DVC), its function in this region has not been defined. In order to discover the unique gene expression signature of GABA^DVC cells, we used single-nucleus RNA sequencing (Nuc-seq), and this revealed 19 separate clusters. We next probed the function of GABA^DVC cells and discovered that the selective activation of GABA^DVC neurons significantly controls food intake and body weight. Optogenetic interrogation of GABA^DVC circuitry identified GABA^DVC → hypothalamic arcuate nucleus (ARC) projections as appetite suppressive without creating aversion. Electrophysiological analysis revealed that GABA^DVC → ARC stimulation inhibits hunger-promoting neuropeptide Y (NPY) neurons via GABA release. Adopting an intersectional genetics strategy, we clarify that the GABA^DVC → ARC circuit curbs food intake. These data identify GABA^DVC as a new modulator of feeding behavior and body weight and a controller of orexigenic NPY neuron activity, thereby providing insight into the neural underpinnings of obesity.

Keywords: body weight; dorsal vagal complex; food intake; gamma-aminobutyric acid; hunger; hypothalamus; neuropeptide Y; nucleus of the solitary tract; obesity.

Graphical abstract

Figure 1. GABA^DVC activation reduces food intake and body weight

(A) tSNE plot of Slc32a1⁺ neurons in the DVC colored by cluster. (B) Dot plot showing scaled expression of marker genes for each of the 19 Slc32a1⁺ clusters. (C) tSNE plot of the Slc32a1⁺ nuclei colored by nutritional status. (D) Schematic of the DVC with diagram showing AP, NTS, and DMV subregions. (E) Representative photomicrograph (scale bars, 500 μm) of GABAergic cell distribution in the medial brainstem (top) and in the DVC (bottom) using Vgat^Cre:tdTom mice. (F) Quantification of GABAergic cells expressing c-Fos following 16 h fasting and fasting + 2 h refeeding (n = 3, two-way ANOVA, bregma level: F_(13,34) = 2.63; p = 0.012; nutritional state: F_(1,4) = 8.69; p = 0.042). (G) Representative micrograph (scale bars, 200 μm) of Vgat^tdTom cells (red) expressing c-Fos (green) in mice fasted 16 h (top) and re-fed for 2 h (bottom) and magnifications (scale bars, 100 μm). (H) Schematic of AAV-DIO-hM3Dq/hM4Di/mCherry (see STAR Methods viral vectors) infused into NTS of Vgat^Cre mice. (I) Representative photomicrograph, right, (scale bars, 200 μm) of c-Fos (green) expression in GABA^DVC:hM3Dq-mCherry (red)-expressing cells in the NTS of Vgat^Cre mice treated with clozapine-n-oxide (CNO, 1 mg/kg intraperitoneally [i.p.]) and magnification, left (scale bar, 100 μm). White arrows indicate double labelled cells. (J) 6 h cumulative (left) and 2 h and 6 h total (right) food intake (n = 6, t(11) = 2.663, and p =0.0238) in GABA^DVC:hM3Dq CNO-injected mice compared with GABA^DVC:mCherry. (K) 2 h cumulative and (J) 2 h total food intake (n = 5, t(8) = 3.205, and p = 0.0125) in GABA^DVC:hM4Di CNO-injected mice compared with GABA^DVC:mCherry mice. (L–O) (L) 10-day treatment protocol schematic. Twice-daily CNO treatment significantly reduced (M) food intake (RM two-way ANOVA F_(1,11) = 20.57; p = 0.0008), (N) body weight (RM two-way ANOVA (F_(1,11) = 21.96; p = 0.0007), and (O) body weight change in GABA^DVC:hM3Dq mice compared with control GABA^DVC:mCherry mice (t(11) = 5.252, p = 0.0003). (C and D) n = 4/group, (H–K) n = 6, (M–O) n = 6 GABA^DVC:mCherry and n = 7 GABA^DVC:hM3Dq. Data are represented as mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001. AP, area postrema; CC, center canal; DMV, dorsal motor nucleus of the vagus; NTS, nucleus of the solitary tract; tSNE, t-distributed stochastic neighbor embedding. See also Figure S1 and Data S1 and S2 for more information.

Figure 2. GABA^DVC projections to the ARC release GABA at NPY^ARC neurons

(A and B) (A) Illustration depicting DVC injection of AAV2-EF1a-DIO-ChR2(E123T/T159C)-mCherry (top) and representative photomicrograph (scale bars, 1 mm) of injection (bottom) in Vgat^Cre mice and (B) ChR2-mCherry-containing fibers in the ARC (scale bars, 200 μm). (C and D) (C) Representative photomicrograph (scale bars, 100 μm) and magnifications (scale bars, 50 and 10 μm) of ChR2-mCherry fibers in the ARC close to NPY^hrGFP-expressing cells but not (D) POMC-expressing cells in Vgat^Cre mice bilaterally injected with AAV-DIO-ChR2-mCherry. (E–J) CRACM study in Vgat^Cre::Npy^hrGFP and Vgat^Cre::Pomc^DsRed mice injected with AAV-DIO-ChR2-mCherry into the DVC. (E) Representative membrane potential response of NPY^hrGFP and POMC^DsRed cells after light stimulation (blue shading) of GABA^DVC → ARC mCherry-containing terminals. 4/27 NPY^hrGFP cells and 0/19 cells POMC^DsRed cells responded. (F) Representative membrane current in NPY^hrGFP cell recorded during a voltage clamp experiment from –100 mV to –10 mV in 10-mV steps. (vertical line, light pulse; green cross, base line; red cross, peak value.) (G) Representative light-induced postsynaptic current during voltage clamp experiment showing peak conductance, G_s, and reversal potential, E_rev. (H) Individual values and median of (H) E_rev, (I) conductance, G_s and (J) latency of all responsive cells. (K) Diagram of outside-out patch technique in Vgat^Cre:Npy^hrGFP mice bilaterally injected with AAV-DIO-ChR2-mCherry into the DVC. (L) Representative channel opening recordings of control (left), gabazine (middle), and washout (right) recordings. (M) Quantification of receptor opening time (n = 4, RM two-way ANOVA F_(2,6) = 16.14; p = 0.0039, Bonferroni adjusted p = 0.0051 control vs. gabazine and p = 0.0156 gabazine vs. washout. Data in (M) are expressed as mean ± SEM. *p < 0.05; **p < 0.01. See also Figure S2.

Figure 3. Activation of GABA^DVC → ARC projections reduce food intake without inducing aversion

(A) Schematic of DVC injection of AAV2-EF1a-DIO-ChR2(E123T/T159C)-mCherry in Vgat^Cre mice and optic fiber placement for in vivo stimulation of GABA^NTS → ARC terminals. (B) Protocol of stimulation. (C–E) (C) Representative 1 min resolution time plot of 3 h food intake measurement. GABA^DVC → ARC stimulation significantly reduces (D) cumulative 3 h food intake and (E) 60 min food intake (n = 7, day 1 [OFF] vs. day 2 [ON] t test t(6) = 3.035, p = 0.0229 and day 2 [ON] vs. day 3 [OFF] t(6) = 3.283, p = 0.0168). (F–H) (F) Diagram illustrating real-time place preference (RTPP) task. GABA^DVC → ARC does not alter (G) time spent (representative heat map of time) or (H) distance traveled in stimulated and non-stimulated zones. (I) Diagram illustrating an object interaction task. (J) Interactions with novel, known nutritional, or novel palatable objects and a representative path track around each object (red circle). GABA^DVC → ARC stimulation reduced interaction with known nutritional item (n = 6, t(5) = 3.043, and p = 0.0287). Data are expressed as individual values and as mean ± SEM, n = 6 mice. *p < 0.05. See also Figure S3.

Figure 4. Selective chemogenetic activation of GABA^DVC → ARC neurons supresses feeding

(A) Diagram illustrating the two-virus intersectional strategy to express hM3Dq only in GABA^DVC neurons projecting to the ARC (GABA^DVC → ARC:hM3Dq mice). Representative photomicrograph (scale bars, 500 and 50 μm) depicting mCherry-expressing cells in the caudal NTS (red), c-Fos (green), and co-labeled (yellow) following treatment with clozapine-n-oxide (CNO, 1 mg/kg i.p.). (B and C) (B) CNO significantly reduced cumulative food intake over 6 h (0–6 h F_(1,15) = 7.418, p = 0.0157), (C) 1 h (t(15) = 2.648, p = 0.0183), and 3 h (t₍₁₅₎ = 2.568, p = 0.0214) compared with saline in GABA^DVC → ARC:hM3Dq mice. (D–K) (D) CNO did not alter 12 h or (E) 1, 3, 6, or 12 h quantification of ambulation; (F) 12 h RER or (G) AUC quantification; (H) 12 h or (I) 1, 3, 6, and 12 h heat production per mouse compared with saline in GABA^DVC → ARC:hM3Dq mice. (J and K) CNO significantly reduced 3 h food intake following overnight fasting compared with control saline treatment (J, two-way ANOVA F_(1.13) = 6.139, p = 0.0277; and K, t(13) = 2.320, p = 0.0372) in GABA^DVC → ARC:hM3Dq mice. (L and M) CNO attenuated ghrelin hyperphagia over 2 h (RM ANOVA F_(3,16) = 11.12, p = 0.0003, Bonferroni adjusted p = 0.0042 control vs. ghrelin, p = 0.0003 CNO vs. ghrelin, and p = 0.002 ghrelin vs. ghrelin+CNO) in GABA^DVC → ARC:hM3Dq mice. (N–P) (N) CNO reduced meal size (t(14) = 2.256, p = 0.0406) and (O–P) decreased meal number (t(14) = 2.824, p = 0.0135) compared with control saline treatment. Data are expressed as individual values and as mean ± SEM. *p < 0.05; **p < 0.01, ***p < 0.001.