A brainstem map for visceral sensations

Abstract

The nervous system uses various coding strategies to process sensory inputs. For example, the olfactory system uses large receptor repertoires and is wired to recognize diverse odours, whereas the visual system provides high acuity of object position, form and movement^1-5. Compared to external sensory systems, principles that underlie sensory processing by the interoceptive nervous system remain poorly defined. Here we developed a two-photon calcium imaging preparation to understand internal organ representations in the nucleus of the solitary tract (NTS), a sensory gateway in the brainstem that receives vagal and other inputs from the body. Focusing on gut and upper airway stimuli, we observed that individual NTS neurons are tuned to detect signals from particular organs and are topographically organized on the basis of body position. Moreover, some mechanosensory and chemosensory inputs from the same organ converge centrally. Sensory inputs engage specific NTS domains with defined locations, each containing heterogeneous cell types. Spatial representations of different organs are further sharpened in the NTS beyond what is achieved by vagal axon sorting alone, as blockade of brainstem inhibition broadens neural tuning and disorganizes visceral representations. These findings reveal basic organizational features used by the brain to process interoceptive inputs.

Fig. 1. Tuning of NTS neurons to gut and upper airway stimuli.

a, Cartoons depicting NTS two-photon calcium imaging (left) and gastric balloon distension (right). b, Representative two-photon NTS images of H2B-jRGECO1a fluorescence at baseline and during stomach stretch, transverse plane. Green arrowheads indicate responsive neurons. M, medial; R, rostral. c, Heatmaps depicting all observed NTS neuron responses (peak ΔF/F normalized) to a 40-s stomach stretch (600 μl). k-means clustering defined slowly adapting neurons (top: 548) or rapidly adapting neurons (bottom: 696) from 19 mice. d, Heatmap depicting NTS responses (188 neurons, 2 mice) to gastric distension (150, 300, 600 and 900 μl) before and after subdiaphragmatic vagotomy. e, Heatmap depicting all observed NTS neurons responses (1,133 neurons, 6 mice) to a stimulus series of oral balloon distension (200 μl), laryngeal water perfusion, stomach distension (150, 300, 600 and 900 μl), duodenum distension (90, 115 and 140 μl), jejunum distension (90, 115 and 140 μl) and caecum distension (100, 250 and 350 μl). f, Correlation coefficient matrix of maximal ΔF/F for each stimulus pair across all neurons responsive to any stimulus in e. g, Heatmap depicting all observed NTS neuron responses (484 neurons, 4 mice) to successive laryngeal perfusion of high salt (10× PBS), water or citric acid (25 mM, pH 2.6). Responses are of two trials averaged. h, Correlation coefficients (R) of maximal ΔF/F for each stimulus pair across all responsive neurons in g (left, 6 × 6 box) or for stimulus pairs from e that included laryngeal water and the highest stretch stimulus in each organ across all neurons responsive to those stimuli (right, 1 × 6 column). Scale bar, 100 μm (b). White bars, 10 s (c–e,g). Source data.

Fig. 2. A sensory map within the NTS.

a, Representative two-photon NTS images of H2B-jRGECO1a fluorescence (transverse view, images are similarly oriented), with neurons colour-coded based on maximal responses to laryngeal water (100 μl), stomach stretch (150, 300, 600 and 900 μl) or duodenal stretch (90, 115 and 140 μl). DCN, dorsal column nuclei; AP, area postrema. Scale bar, 100 μm (a). b, Neuron positions are charted by response type. Axis origin: centroid for stomach stretch-responsive neurons. The following number of neurons and mice were analysed: oral–stomach: 1,204 neurons, 11 mice; larynx–stomach: 10,610 neurons, 57 mice; duodenum–stomach: 28,556 neurons, 107 mice; jejunum–stomach: 13,050 neurons, 66 mice; caecum–stomach: 415 neurons, 9 mice. Colour scale depicts neuron density (Methods). Axis length, 150 μm. M, medial; R, rostral. c, Schematic depicting the relative location of different organ response domains in the NTS. d, Neuron positions are charted by response type. Axis origin: centroid of neurons responsive to duodenal glucose (300 mM, HBSS). n = 425 neurons (left) or 97 neurons (right), 2 mice. e, Quantification of the segregation of neurons in d responsive to duodenum glucose from neurons responsive to duodenum or stomach stretch. Mean ± s.e.m., **P < 0.0001, two-tailed Mann–Whitney test. See Methods for segregation index calculation. f, Cartoon depicting sites of balloon distension in the gastrointestinal tract. g, Neuron positions are charted by response type and axis origin: centroid of duodenum stretch-responsive neurons. The following number of neurons and mice were analysed: left: 14,981 neurons, 71 mice; middle: 2,846 neurons, 53 mice; right: 280 neurons, 7 mice. Axis length, 150 μm. h, Quantification of the segregation of neurons in g responsive to duodenum stretch from neurons responsive to other stimuli. Mean ± s.e.m., ****P < 0.0001, Dunn’s multiple comparisons test following Kruskal–Wallis test of significance. Source data.

Fig. 3. Inhibition sharpens NTS responses.

a, Vgat-ires-cre;Rosa26-lsl-Gfp-L10a mice were injected in the NTS with AAV8-Syn-DIO-HA-hM3dq-ires-mCitrine and AAV1-Syn-H2b-jRGECO1a. NTS imaging was performed on GFP-negative excitatory neurons before and after CNO administration (bottom). Control mice (top) lacked expression of hM3Dq and were instead injected with saline. Heatmaps depict responses to stomach stretch (150, 300, 600 and 900 μl) and duodenum stretch (90, 115 and 140 μl). n = 869 neurons, 3 mice (top) or 673 neurons, 5 mice (bottom). b, Average change in peak ΔF/F of neurons from a after injection of saline or CNO. Samples sizes (left to right, top): 726 neurons, 3 mice; 507 neurons, 5 mice; (left to right, bottom): 94 neurons, 3 mice; 92 neurons, 5 mice. ****P < 0.0001, two-tailed Mann–Whitney test. c, Heatmaps depicting all observed NTS responses without (top left: 16,222 neurons, 103 mice; top right: 7,919 neurons, 73 mice) or with (bottom left: 1,168 neurons, 5 mice; bottom right: 1,442 neurons, 5 mice) NTS-localized administration of the GABA_A receptor antagonist bicuculline. Responses were measured for gastric distension (150 and 300 μl), duodenum distension (90, 115 and 140 μl) and laryngeal water perfusion (100 μl), with the peak ΔF/F of each responding neuron normalized to 100%. Neurons in each heatmap are sorted by their response bias to stomach stretch (top) over duodenum stretch/laryngeal water (bottom). d, Correlation coefficient matrices of maximal ΔF/F for each stimulus pair across all responsive neurons from c (top: 11,795 neurons, 73 mice; bottom: 1,806 neurons, 5 mice). e, Responses (maximal ΔF/F above thresholds; Methods) of NTS neurons from c. n = 400 randomly depicted neurons per condition, 2 out 1,600 out of range. White bars, 10 s (a,c). Source data.

Fig. 4. NTS response suppression by paired inputs.

a, Calcium responses were recorded in NTS neurons during successive application of stomach stretch (600 μl), duodenal stretch (115 μl) and simultaneous stomach–duodenal stretch, with the stimulus series repeated and responses averaged. Neurons that selectively detected duodenal stretch (left, 176 neurons) or stomach stretch (right, 544 neurons) were classified into two groups that exhibited inhibition from the simultaneous stomach–duodenal stretch (mix < single; left: 51 neurons; right: 94 neurons) or did not (mix ≈ single; left: 125 neurons; right: 450 neurons, 5 mice). Average traces (top) or heatmaps (bottom) of responding neurons in each class are separately depicted, with the peak ΔF/F of each responding neuron normalized to 100%. White bar, 10 s. b, Representative traces depicting normalized ΔF/F over time for individual neurons from a that showed responses to the six stimuli described (left) and another series of eight duodenal stretches with simultaneous application of 0, 300, 600, 900, 900, 600, 300 and 0 μl stomach stretch (right).

Extended Data Fig. 1. Characterizing the calcium sensor, AAV injection time, and anesthetic used during in vivo NTS imaging.

a, Mice were injected in the NTS with AAVs encoding H2B-jRGECO1a and GCaMP6m, and NTS imaging was later performed to measure gastric distension-evoked calcium transients. A heat map depicting time-resolved responses of all 74 NTS neurons (2 mice) that displayed both H2B-jRGECO1a (left) and GCaMP6m (right) responses to a series of gastric distensions (green bars of increasing thickness: 150, 300, 600, and 900 μl), white bar: 10 s. b, Representative traces depicting normalized ΔF/F over time for individual neurons (same neurons on left and right), scale bar: 10 s. c, Maximal ΔF/F of NTS neurons responsive to 900 μl gastric distension. d, Mice were injected in the NTS with an AAV encoding H2B-jRGECO1a at 2 (left) and 5 (right) weeks of age, and NTS imaging was later performed. Representative two-photon images of H2B-jRGECO1a fluorescence in the NTS (transverse view). Neurons are color-coded based on their relative peak response amplitudes above threshold to laryngeal water (blue, 100 μl), stomach stretch (green, peak response to 150, 300, 600, or 900 μl), or duodenal stretch (red, peak response to 90, 115, or 140 μl), scale bar: 100 μm. e, Positions of neurons selectively responsive to stimuli indicated are charted with axis origin corresponding to the centroid for stomach-stretch responsive neurons, top left: 695 neurons, 4 mice, bottom left: 1106 neurons, 4 mice, top right: 2331 neurons, 8 mice, bottom right: 2215 neurons, 6 mice, color scale depicts neuron density (see methods), axis length: 150 μm, R: rostral, M: medial. f, Quantifying the percentage of neurons singly tuned to either gastric distension, duodenum distension, or laryngeal water, dots: individual mice, n: 4 (left), 8 (right), mean ± sem. g, Representative two-photon images of H2B-jRGECO1a fluorescence in the NTS (transverse view) of mice anesthetized with urethane (2.2–2.6 mg/mg) or isoflurane (1.5%–2.5%). Neurons are color-coded based on their relative peak response amplitudes above threshold to laryngeal water (blue, 100 μl), stomach stretch (green, peak response to 150, 300, 600, or 900 μl), or duodenal stretch (red, peak response to 90, 115, or 140 μl), scale bar: 100 μm. h, Positions of neurons selectively responsive to stimuli indicated are charted with axis origin corresponding to the centroid for stomach-stretch responsive neurons, top left: 1136 neurons in 8 mice randomly selected from 49 mice, bottom left: 2525 neurons in 9 mice randomly selected from 98 mice, top right: 1010 neurons, 8 mice, bottom right: 1707 neurons, 9 mice, color scale depicts neuron density (see methods), axis length: 150 μm, R: rostral, M: medial. i, Quantifying the percentage of neurons singly tuned to either gastric distension, duodenum distension, or laryngeal water, dots: individual mice, n: 63 (left), 10 (right), mean ± sem. Source data

Extended Data Fig. 2. Two classes of NTS neurons respond to gastric distension with different adaptation rates.

a, Heat map depicting the time-resolved responses (color coded based on the percentage increase in ΔF/F of jRGECO1a fluorescence from baseline) of all 81 (1 mouse) responding NTS neurons to ten consecutive gastric distensions (green bars). Each row shows the response of an individual neuron, with rows sorted by response amplitude, white bar: 10 s. b, A rank-ordered plot of peak ΔF/F for all 81 neurons depicted in Fig. 1c, gray circles: peak response amplitudes of individual trials; red circles: means of peak response amplitudes across trials. c, Representative traces depicting normalized ΔF/F over time for individual neurons to repeated stomach stretch (600 μl, distension rate: ~25 μl/s for trials #1 and #3, and ~300 μl/s for trials #2 and #4), scale bar: 10 s. d, A histogram depicting response duration (full width at quarter maximum) of slowly adapting (green) and rapidly adapting (yellow) neurons in Fig. 1c. e, A representative two-photon image of H2B-jRGECO1a fluorescence in the NTS (transverse view) with neurons pseudocolored based on their slow (green) or rapid (yellow) adaptation to stomach stretch, scale bar: 100 μm. f, Positions of neurons that adapt slowly (green, 544 neurons) or rapidly (yellow, 650 neurons) to stomach stretch, with axis origin corresponding to the centroid for slowly adapting neurons, 19 mice, color scale depicts neuron density (see methods), axis length: 200 μm, R: rostral, M: medial. g, Percentage of slow-adapting (green, 289 neurons in 13 mice) and fast-adapting (yellow, 493 neurons in 13 mice) stomach stretch-activated neurons that also responded to duodenum stretch (90, 115, or 140 μl). h, Responses of slow-adapting (green, 316 neurons in 13 mice) and fast-adapting (yellow, 518 neurons in 13 mice) neurons to various magnitudes of gastric distension; maximal ΔF/F normalized to 100% for each neuron, mean ± sem. Source data

Extended Data Fig. 3. Response properties of NTS neurons.

a, Representative traces depicting normalized ΔF/F over time for individual neurons from Fig. 1e. b, The percentage of neurons from Fig. 1e that selectively respond to different organ inputs or display multi-tuned responses. c, Pie chart depicting neurons from Fig. 1e responsive to stimuli from 1 (singly-tuned) or > 1 (multi-tuned) organ. d, The percentage of multi-tuned neurons in Fig. 1e to various stimulus pairs. e, The percentage of singly tuned neurons in each imaged mouse (circles) of Fig. 1e, mean ± sem. f, Responses (maximal ΔF/F above thresholds) of neurons that responded to any stomach stretch, and/or any stimulation in other organs. Each chart depicts responses of 300 neurons randomly selected from 3815 neurons, 21 mice (oral vs. stomach), from 18895 neurons, 73 mice (larynx vs. stomach), from 35120 neurons, 113 mice (duodenum vs. stomach), from 21653 neurons, 84 mice (jejunum vs. stomach), and from 4414 neurons, 22 mice (cecum vs. stomach). g, Responses (maximal ΔF/F above thresholds) of neurons that responded to 600 ml or 900 ml stomach stretch (300 neurons randomly selected from 27125 neurons, 103 mice). h, Representative traces depicting normalized ΔF/F over time for 13 individual neurons from Fig. 1g, scale bar: 10 s. i, Responses (maximal ΔF/F above thresholds, see methods) of 362 (left), 404 (middle), and 467 (right) responsive NTS neurons from Fig. 1g. Peak responses were from one trial (left) or were the larger response from two trials (middle, right). Source data

Extended Data Fig. 4. NTS responses to intestinal stimuli.

a, Heat maps depicting the time-resolved responses of all responding NTS neurons to various magnitudes of duodenum stretch (top, 271 neurons) and jejunum stretch (bottom, 118 neurons), 6 mice, white bar: 10 seconds. b, Peak ΔF/F of neurons in a that responded to stimuli indicated, F test for non-zero slope: ****P < 0.0001. c, Heat map depicting the time-resolved responses of all 74 responding NTS neurons (2 mice) to duodenal perfusion (24 s) of glucose (300 mM in HBSS, purple) and saline (HBSS, olive), scale bar: 10 s. d, Heat map depicting the time-resolved responses of all 492 responding NTS neurons (2 mice) to duodenal perfusion of glucose (300 mM, dark purple), duodenum stretch (red: 140 μl), and stomach stretch (600 μl, green), white bar: 10 s. Source data

Extended Data Fig. 5. Responses of NTS inhibitory neurons.

a-c, Vgat-ires-Cre mice were injected in the NTS with AAV9-Cag-Flex-Gfp and AAV1-Syn-H2b-jRGECO1a and prepared for in vivo two-photon imaging. a, A representative image of native fluorescence from H2B-jRGECO1a (magenta) and GFP (green, depicting Vgat neurons) is shown, scale bar: 100 μm. Nonlinear image adjustments were made to ensure visualization of all green cells. b, Heat map depicting the time-resolved responses of all 1933 Vgat-negative (left) and 206 Vgat-positive (right) responding NTS neurons in 7 mice to laryngeal water perfusion (dark blue), stomach stretch (green, increasing thickness: 150, 300, 600, and 900 μl) and duodenum distension (red, increasing thickness: 90, 115, and 140 μl), white bar: 10 s. c, Pie charts depicting the percentages of Vgat-negative and Vgat-positive neurons responsive to stimuli indicated. d-f, Vgat-ires-Cre mice were crossed with Rosa26-lsl-Gfp-L10a mice and prepared for in vivo two-photon imaging. d, A representative image of native fluorescence from H2B-jRGECO1 (magenta) and GFP (green, depicting Vgat neurons) is shown, scale bar: 100 μm. Nonlinear image adjustments were made to ensure visualization of all green cells. e, Heat map depicting the time-resolved responses of all 766 Vgat-negative (left) and 402 Vgat-positive (right) responding NTS neurons in 5 mice to laryngeal water perfusion (dark blue), stomach stretch (green, increasing thickness: 150, 300, 600, and 900 μl) and duodenum distension (red, increasing thickness: 90, 115, and 140 μl), white bar: 10 s. f, Pie charts depicting the percentages of Vgat-negative and Vgat-positive neurons responsive to stimuli indicated.

Extended Data Fig. 6. Responses of Cre-defined NTS neuron subtypes.

a, NTS imaging was performed on mice containing a Cre-dependent Gfp allele and Cre alleles indicated. Representative images of native H2B-jRGECO1 (magenta) and GFP (green) fluorescence, nonlinear image adjustments made to ensure visualization of all green cells, scale bar: 100 μm. b, Positions of GFP-positive, H2B-jRGECO1a-positive (green) and GFP-negative, H2B-jRGECO1a-positive neurons (magenta), with axis origin corresponding to the centroid of GFP-negative, H2B-jRGECO1a-positive neurons. Numbers of green/red neurons from 5 Th-Cre mice: 354/11748, 3 Cartpt-ires-Cre mice: 106/8235, 4 Calcr-ires-Cre mice: 756/6911, 3 Pdyn-ires-Cre mice: 1 /6747, 7 Tac1-ires-Cre mice: 456/15016, 4 Penk-ires-Cre mice: 1326/8635, 3 Gcg-Cre mice: 478/11122, 3 Sst-ires-Cre mice: 812/7149, and 5 Crhr2-ires-Cre mice: 398/14377, color scale depicts neuron density (see methods), axis length: 200 μm, R: rostral, M: medial. c, Pie charts depicting the percentages of responsive GFP-negative, H2B-jRGECO1a-positive (top) and GFP-positive, H2B-jRGECO1a-positive (bottom) neurons responsive to stomach stretch, duodenum stretch, or both (multi-tuned). d, An enrichment index quantifies the relative composition of Cre-defined NTS cell types responsive to each stimulus (see Methods). e, Uniform manifold approximation and projection (UMAP) plot indicating NTS neuron subtypes based on published single-cell transcriptome data. f, Dot plots showing normalized expression of genes indicated across 24 NTS neuron subtypes from e. Source data

Extended Data Fig. 7. Spatial patterning of NTS responses to gut and upper airway stimuli.

a, Representative two-photon image of H2B-jRGECO1a fluorescence in the NTS (transverse view), with neurons color-coded based on their relative peak response amplitudes above threshold to leg pinch (red) and stomach stretch (green, successive 150, 300, 600, and 900 μl distensions), scale bar: 100 μm. b, Positions of neurons selectively responsive to stimuli indicated are charted as in Fig. 3b, with analysis here involving 9 randomly selected mice from Fig. 3b to equalize sample size, oral/stomach: 925 neurons, larynx/stomach: 1128 neurons, duodenum/stomach: 2347 neurons, jejunum/stomach: 1153 neurons, cecum/stomach: 415 neurons (same data as Fig. 3b), color scale depicts neuron density (see methods), axis length: 150 μm, R: rostral, M: medial. c, Representative two-photon images at various depths of H2B-jRGECO1a fluorescence in the NTS (transverse view), with neurons color-coded based on their relative peak response amplitudes above threshold to stomach stretch (green, successive 150, 300, 600, and 900 μl distensions) and duodenum stretch (red, successive 90, 115, or 140 μl distensions), scale bar: 100 μm. d, Positions of neurons at various depths selectively responsive to stimuli as described in c are charted with axis origin corresponding to the centroid for stomach-stretch responsive neurons, 0 μm: 455 neurons, 17 mice, 80 μm: 2517 neurons, 46 mice, 160 μm: 4909 neurons, 53 mice, 240 μm: 4445 neurons, 52 mice, 320 μm: 2204 neurons, 39 mice, color scale depicts neuron density, axis length: 150 μm, R: rostral, M: medial. e, Quantifying spatial segregation of neurons in d, mean ± sem, see methods for Segregation Index. f, Pairwise distances between neurons responsive to the same stimulus (colored) or different stimuli (black). Data from Fig. 3b (real) were compared with data from a simulation where neuron responses were randomly assigned based on the observed response frequency in each field of view (shuffled) to control for regional variation in neuron density, oral/stomach: 1204 neurons, 11 mice, larynx/stomach: 10610 neurons, 57 mice, duodenum/stomach: 28556 neurons, 107 mice, jejunum/stomach: 13050 neurons, 66 mice, cecum/stomach: 415 neurons, 9 mice, mean ± sem, ^####P < 0.0001, significant interaction in two-way analysis of variance between responder types and shuffling, ****P < 0.0001, Šídák multiple comparisons test. g, Cartoon depicting sites of balloon distension in the gastrointestinal tract. h, Positions of neurons responsive to stimuli indicated are charted relative to the centroid (coordinate origin) of neurons responsive to distention of stomach site 2, left: 1155 neurons, 4 mice, right: 927 neurons, 4 mice, axis length: 150 μm, R: rostral, M: medial. i, Quantifying spatial segregation of neurons in h responsive to stimuli indicated, mean ± sem, ****P < 0.0001, two-tailed Mann-Whitney test, see methods for Segregation Index. Source data

Extended Data Fig. 8. Comparing the positions of vagal axons and responsive NTS neurons.

a, Cartoon depicting NTS two-photon imaging in Vglut2-ires-Cre mice injected with AAV1-Cag-Flex-Synaptophysin-Gfp in the larynx (top), stomach (middle), and duodenum (bottom), and AAV1-Syn-H2b-jRGECO1a in the NTS. b, Left: representative two-photon images (transverse view) of NTS neuronal nuclei (red, jRGECO1a) and boutons of vagal axons from organs indicated (green, Synaptophysin-GFP). Right: the same images with NTS neurons color-coded based on their relative peak response amplitudes above threshold to laryngeal water (100 μl, blue), stomach stretch (green, successive 150, 300, 600, and 900 μl distensions), and duodenal stretch (red, successive 90, 115, and 140 μl distensions), R: rostral, M: medial, scale bar: 100 μm. c, Positions of vagal axon boutons from the larynx (top row, dark blue, 489 boutons, 3 mice), stomach (middle row, dark green, 966 boutons, 3 mice), and duodenum (bottom row, dark red, 891 boutons, 5 mice) are charted along with the positions of NTS neurons selectively responsive to 100 μl laryngeal water (left column, blue), any distension (150, 300, 600, or 900 μl) of stomach (2nd left column, green), any distension (90, 115, or 140 μl) of duodenum (2nd right column, red), or any distension (90, 115, or 140 μl) of jejunum (right column, orange). Neurons depicted top to bottom: larynx 24, 54, 57, stomach 209, 364, 555, duodenum 91, 92, 109, jejunum 69, 59, 58, axis origin: centroid for stomach-stretch responsive neurons, axis length: 150 μm, R: rostral, M: medial. d, Positions of vagal axon boutons from c with axis origin corresponding to the centroid for stomach-stretch responsive neurons, axis length: 150 μm, R: rostral, M: medial. e, The percentages of vagal axon boutons from the larynx (top), stomach (middle), and duodenum (bottom) in c that are located within a variable radius from any neuron responsive to stimuli indicated, mean ± sem. f, Distances between NTS neurons responsive to stimuli indicated and the closest boutons of vagal axons from the larynx (top), stomach (middle), and duodenum (bottom), mean ± sem. g, Percentages of NTS neuron responsive to stimuli indicated that are located in the area of innervation (see methods) of vagal axons from different organs. h, Left: representative two-photon images (transverse view) of NTS neuronal nuclei (red, jRGECO1a) and boutons of vagal axons (green, Synaptophysin-GFP) visualized by injecting AAV1-Cag-Flex-Synaptophysin-Gfp into the stomach of Vglut2-ires-Cre (top), Glp1r-ires-Cre (middle), or Gpr65-ires-Cre (bottom) mice, scale bar: 100 μm. Right: the same images with NTS neurons colored in green to indicate neurons responsive to stomach distension. i, Positions of axonal boutons (dark green) from stomach neurons labeled in Vglut2-ires-Cre (top, 966 boutons, 3 mice, same data as c), Glp1r-ires-Cre (middle, 1276 boutons, 3 mice), or Gpr65-ires-Cre (bottom, 243 boutons, 3 mice) mice and compared to the positions of NTS neurons responsive to stomach stretch, axis origin: centroid for stomach-stretch responsive neurons, top: 364 neurons (same data as c), middle: 539 neurons, bottom: 244 neurons, axis length: 150 μm, R: rostral, M: medial. j, Positions of axonal boutons from i. k, Distances between NTS neurons responsive to stomach stretch and the closest boutons of vagal axons from the stomach of Vglut2-ires-Cre (same data as in f), Glp1r-ires-Cre, and Gpr65-ires-Cre mice, mean ± sem, **P < 0.01, ****P < 0.0001, Dunn’s multiple comparisons test following Kruskal-Wallis test of significance. l, Pairwise distances were calculated between NTS neurons responsive to stomach stretch and axonal boutons of stomach neurons from Cre lines indicated (black) and compared with an average of bouton-bouton pairwise distances and NTS neuron-NTS neuron pairwise differences (green). Real data were compared with data from a simulation where identity of bouton or NTS neuron was randomly assigned based on observed frequency in each field of view (shuffled) to control for regional variation in neuron or bouton density, mean ± sem, ^###P < 0.001, ^####P < 0.0001, significant interaction in two-way analysis of variance between responder types and shuffling, ****P < 0.0001, Šídák multiple comparisons test. Source data

Extended Data Fig. 9. Inhibition shapes responses of both NTS excitatory and inhibitory neurons.

a, Vgat-ires-Cre; Rosa26-lsl-Gfp-L10a mice were injected in the NTS with AAV1-Syn-H2b-jRGECO1a, and in vivo NTS imaging was performed with (bottom) or without (top) NTS-localized administration of bicuculline. Heat maps depict the time-resolved responses of Vgat-positive (left) and Vgat-negative (right) neurons to laryngeal water (dark blue, 100 μl), stomach stretch (green, increasing thickness: 150, 300, 600, and 900 μl) and/or duodenum distension (red, increasing thickness: 90, 115, and 140 μl). Top: 402 Vgat-positive and 766 Vgat-negative neurons, 5 mice, same data as Extended Data Fig. 5e; bottom: 365 Vgat-positive and 851 Vgat-negative neurons, 2 mice, white bar: 10 s. b, Pie charts depicting the percentages of Vgat-positive (left) and Vgat-negative (right) neurons from c responsive to stimuli indicated with (bottom) or without (top, same data as Extended Data Fig. 5f) bicuculline administration. c, The percentage of Vgat-positive and Vgat-negative neurons from a responsive to stimuli from > 1 organ (multi-tuned), ****P < 0.0001, two-tailed χ² test. Source data

Extended Data Fig. 10. Characterizing NTS response suppression by multiple input pairs.

a, Positions of neurons from Fig. 4a selectively responsive to duodenum stretch (top) or stomach stretch (bottom), with differential coloring based on sensitivity to lateral inhibition, axis origin: centroid for uninhibited neurons, axis length: 150 μm, R: rostral, M: medial. b, Response suppression in stimulus pairs indicated measured on a per-animal basis in neurons that received dual stimulation-evoked inhibition (Mix < single, right) or did not (Mix ≈ single, left), data points: average response changes of all neurons in a single mouse, 5 mice. c, Average traces (top) or heat maps (bottom) of normalized, time-resolved NTS neuron responses to successive application of stomach stretch (green, 600 μl), jejunal stretch (yellow, 115 μl), and both. Neurons that selectively detected jejunum stretch (left, 378 neurons) or stomach stretch (right, 1183 neurons) were classified into two groups were dual stimulation evoked inhibition (Mix < single; left: orange, 125 neurons, right: light green, 124 neurons) or did not (Mix ≈ single; left: gray, 253 neurons, right: dark green, 1059 neurons, 8 mice), white bar: 10 s. d, Positions of neurons from c selectively responsive to jejunum stretch (left) or stomach stretch (right), with differential coloring based on sensitivity to lateral inhibition, axis origin: centroid for uninhibited neurons, axis length: 150 μm, R: rostral, M: medial. e, Response suppression in stimulus pairs indicated measured on a per-animal basis in neurons that received dual stimulation-evoked inhibition (Mix < single, right) or did not (Mix ≈ single, left), data points: average response changes of all neurons in a single mouse, 7 mice. f, Average traces (top) or heat maps (bottom) of normalized, time-resolved NTS neuron responses to successive application of stomach stretch (green, 600 μl), laryngeal water (blue), and both. Neurons that selectively detected laryngeal water (left, 119 neurons) or stomach stretch (right, 243 neurons) were classified into two groups were dual stimulation evoked inhibition (Mix < single; left: light blue, 13 neurons, right: light green, 37 neurons) or did not (Mix ≈ single; left: dark blue, 106 neurons, right: dark green, 206 neurons, 3 mice), white bar: 10 s. g, Positions of neurons from f selectively responsive to laryngeal water (left) or stomach stretch (right), with differential coloring based on sensitivity to lateral inhibition, axis origin: centroid for uninhibited neurons, axis length: 150 μm, R: rostral, M: medial. h, Response suppression in stimulus pairs indicated measured on a per-animal basis in neurons that received dual stimulation-evoked inhibition (Mix < single, right) or did not (Mix ≈ single, left), data points: average response changes of all neurons in a single mouse, 3 mice, mean ± sem, *P < 0.05, **P < 0.01, ****P < 0.0001, Šídák multiple comparisons test. Source data