A vagal-brainstem interoceptive circuit for cough-like defensive behaviors in mice

Abstract

Coughing is a respiratory behavior that plays a crucial role in protecting the respiratory system. Here we show that the nucleus of the solitary tract (NTS) in mice contains heterogenous neuronal populations that differentially control breathing. Within these subtypes, activation of tachykinin 1 (Tac1)-expressing neurons triggers specific respiratory behaviors that, as revealed by our detailed characterization, are cough-like behaviors. Chemogenetic silencing or genetic ablation of Tac1 neurons inhibits cough-like behaviors induced by tussive challenges. These Tac1 neurons receive synaptic inputs from the bronchopulmonary chemosensory and mechanosensory neurons in the vagal ganglion and coordinate medullary regions to control distinct aspects of cough-like defensive behaviors. We propose that these Tac1 neurons in the NTS are a key component of the airway-vagal-brain neural circuit that controls cough-like defensive behaviors in mice and that they coordinate the downstream modular circuits to elicit the sequential motor pattern of forceful expiratory responses.

Extended Data Fig. 1 |. Molecularly heterogenous neurons in the NTS controls diverse breathing responses.

a, UMAP plot of 5523 high-quality NTS cells showing 8 different cell populations. b, 8 clusters of the NTS cells are annotated based on published dataset of the nervous system. c, UMAP plot of the putative neurons showing 19 molecularly distinct subclusters. d, g, j, m, Violin plots showing the expression of four markers, Slc17a6 (Vglut2) in d, Slc32a1 (Vgat) in g, Grp in j, and Tac1 in m in 19 clusters of NTS neurons. e, h, k, n, Optogenetic activation of NTS neuron subsets elicits diverse breathing responses in freely-moving mice: 15 ms laser activation of Vglut2-Cre neurons induces an ectopic inspiratory peak in e. 15 ms laser activation of Vgat-Cre neurons induces apnea in h. 15 ms laser activation of Grp-Cre neurons elicits a sigh in k. Activation of Tac1-Cre neurons induces a stereotyped breathing pattern in n. f, Individual (gray) and average (black) traces for ectopic inspiratory peaks induced by optogenetics of Vglut2 neurons (n = 30 events, 3 mice). Traces aligned by laser onset. i, Individual (gray) and average (black) traces for apnea induced by optogenetics of Vgat neurons (n = 30 events, 3 mice). Traces aligned by laser onset. l, Individual (gray) and average (black) traces for sighs induced by optogenetics of Grp neurons (n = 30 events, 3 mice). Traces aligned by laser onset. Please note that the response induced by activation of Grp-Cre neurons could exhibit latency, so only the events with immediate responses were included here. o, Individual (gray) and average (black) traces for the stereotyped breathing pattern induced by optogenetics of Tac1 neurons (n = 30 events, 3 mice). Traces aligned by laser onset.

Extended Data Fig. 2. |. Classification of respiratory defensive behaviors in freely behaving mice.

a, Hierarchical clustering separates all expiratory events (n = 313 events) induced by either capsaicin or compound 48/80 into 8 clusters. Each cluster was assessed for the percentage of events occurring after AEN denervation over the total number in the cluster. Clusters with a lower percentage indicate that the breathing patterns predominantly occur in mice with intact AEN and diminish after AEN denervation, designating it as sneezes (class 2). Clusters with a higher percentage indicate that the breathing patterns are not dependent on intact AEN and are designated as coughs (class 1). Based on this assessment and the similarity between clusters, breathing patterns were classified into two classes. b, Individual (gray) and average (red) traces characterized in class one (n = 218 events). Aligning at the transition between compressive and expulsive phases makes it harder to appreciate the inspiration. c, Individual (gray) and average (black) traces characterized in class two (n = 95 events). d, PCA plot of all events in b and c. Please note that linear support vector machine (linear SVM) demonstrates an accuracy of 98.08% for these two classes. e, The confusion matrix of the linear SVM method that is used to distinguish between the two clusters moving forward. f, g, Difference in duration (f, p = 0.0001, two-tailed Mann-Whitney test) and amplitude (g, p = 0.0109, two-tailed Mann-Whitney test) between cough-like (n = 218 events) and sneeze-like behaviors (n = 95 events) of the compressive phase. Mean ± SEM. h, Cough-like (p = 0.92, two-tailed paired t-test) and sneeze-like behaviors (p = 0.02, one-tailed paired t-test) in a 10-minute compound 48/80 assay in mice (n = 5 mice) before (pre) and after (post) an anterior ethmoidal nerve (AEN) denervation. Mean ± SEM. i, Cough-like (p = 0.26, two-tailed paired t-test) and sneeze-like behaviors (p = 0.16, one-tailed Wilcoxon matched-pairs signed rank test) in a 6-minute capsaicin assay in mice (n = 5 mice) before (pre) and after (post) an anterior ethmoidal nerve (AEN) denervation. Mean ± SEM. j, Quantification of the occurrence of cough-like and sneeze-like behaviors in 6-minute capsaicin assay. n = 93 events, 10 mice, p = 0.0039 two-tailed Wilcoxon matched-pairs signed rank test. Mean ± SEM.

Extended Data Fig. 3 |. Characterization of cough-like behavior.

a, Simultaneous recordings of a cough with the box flow by WBP, intrapleural pressure by telemetry, and sound recording. Cough is composed of three phases: (1) inspiratory (blue), (2) compressive (purple) and (3) expulsive (red). In the inspiratory phase, the intrapleural pressure decreases and air is inhaled into the lungs. The box flow is negative (flow out of the chamber) due to the increased temperature and humidity of the inhaled air in the animal body. In the compressive phase, the glottis is closed and the muscles compress air in the lungs, resulting in a rapid increase of the intrapleural pressure. Due to the compression of the body, the pressure in the chamber decreases and air flows into the chamber (positive box flow). In the expulsive phase, the glottis is opened and intrapleural pressure is released. Sound is produced as air rushes past the glottis. Because the air with high pressure is released from the lungs to the chamber, the air flows out of the chamber (negative box flow). b, Simultaneous recordings of an expiratory reflex with the box flow by WBP, intrapleural pressure by telemetry, and sound recording. The expiratory reflex is made up of the last two phases of the cough: compressive (purple) and expulsive (red). c, Stereotyped breathing waveform and sound spectrum of a cough evoked by capsaicin. d, Repetitive coughs evoked by a tussive challenge. e, Top: raster plots of coughs across ten mice during 3-minute nebulization of saline (gray) and 3-minute following nebulization. Bottom: average events rate; shaded area, SEM. f, Coughs in a 6-minute cough assay with different concentrations of capsaicin (n = 15 mice). Mean ± SEM. g, Cough-like behaviors in a 6-minute assay with different concentrations of citric acid (n = 15 mice). Mean ± SEM. h, Coughs in a 6-minute capsaicin assay before (pre) and after (post) unilateral vagotomy (n = 5 mice, p = 0.001, two-tailed paired t-test). Mean ± SEM. i, Coughs in a 6-minute cough assay in mice with AEN denervation (n = 8) compared to mice with a sham surgery (n = 12). p = 0.63, two-tailed unpaired t-test. Mean ± SEM.

Extended Data Fig. 4 |. Characterization of other behavioral events during a tussive challenge.

a, A sigh is characterized by a biphasic and augmented inspiration. b, A sneeze is defined as double expiratory peaks with two audio peaks. c, An expiratory peak is defined as an expiratory peak with no obvious inspiratory phase. d, Multiple inspiration event is characterized by multiple inspiration peaks with continuous and loud sound signal. Note that the sound trace is cropped at 0.03. e, A sound only event is a regular breathing pattern with an audio signal that is greater than or equal to 0.03. Note that all these events in a-e are correlated to audio peaks comparable or greater than that of a cough-like response (compare to Figs. 1c and 1d). f, Quantification of various respiratory and audio events in a 6-minute tussive challenge assay. n = 5 mice. Mean ± SEM. Please note that the vertical scale on the audio graphs displays the amplitude. a.u., arbitrary unit.

Extended Data Fig. 5 |. NTS Tac1 cell body optogenetics analysis.

a, Brain slice showing the expression of ChR2-eYFP and optic fiber implantation in the NTS of a Tac1-Cre mouse. b, Quantification of the number of ChR2-eYFP+ cells in the NTS compared to the immediate surrounding areas (area postrema, cuneate nucleus and dorsal motor nucleus of the vagus nerve) (n = 6 mice; p = 0.002, two-tailed Mann-Whitney test). Mean ± SEM. c, Individual (gray) and average (black) traces for cough-like behaviors induced by a 50 ms laser pulse (n = 30 events, 3 mice). Traces aligned by laser onset. Blue, laser on. d, Individual (gray) and average (black) traces for cough-like behaviors induced by a 100 ms laser pulse (n = 30 events, 3 mice). Traces aligned by laser onset. Blue, laser on. e, Individual (gray) and average (black) traces for cough-like behaviors induced by a 200 ms laser pulse (n = 30 events, 3 mice). Traces aligned by laser onset. Blue, laser on. f, Respiratory trace for an event induced by a 10 Hz optogenetic cell body activation. Blue, laser on. g, Respiratory trace for an event induced by a 20 Hz optogenetic cell body activation. Blue, laser on. h, Individual (gray) and average (black) traces for cough-like behaviors induced by a 15 ms laser pulse (n = 33 events, 3 mice). Traces aligned by compressive peaks. Laser hit during eupneic inspiration. i, Individual (gray) and average (black) traces for cough-like behaviors induced by a 15 ms laser pulse (n = 33 events, 3 mice). Traces aligned by compressive peaks. Laser hit during eupneic expiration. j, Individual (gray) and average (black) traces for all events in h and i. Traces aligned by compressive peaks. k, respiratory events in h, now aligned by laser onset. (n = 33 events, 3 mice). l, respiratory events in i, now aligned by laser onset. (n = 33 events, 3 mice). m, respiratory events in j, now aligned by laser onset. (n = 33 events, 3 mice).

Extended Data Fig. 6 |. Distinct characteristics of respiratory defensive behaviors in anesthetized mice.

a, An EMG recording of the styloglossus muscle during optogenetic activation of a cell body activation of NTS Tac1 neurons. b, An EMG recording of the styloglossus muscle during a stimulation of the superior laryngeal nerve (SLN), which induces a fictive cough. c, An EMG recording of the styloglossus muscle during a stimulation of the anterior ethmoidal nerve (AEN), which induces a fictive sneeze. d, Quantification of panel a of the amplitude of the styloglossus muscle EMG (p = 0.49, two-tailed paired t-test, n = 3 events). e, Quantification of panel b of the amplitude of the styloglossus muscle EMG (p = 0.42, two-tailed paired t-test. n = 3 events). f, Quantification of panel c of the amplitude of the styloglossus muscle EMG (p = 0.011, two-tailed paired t-test, n = 3 events). g, The normalized distance between vocal cords during optogenetic stimulation of NTS Tac1 neurons compared to basal breathing (Ctrl). (n = 11 trials; p = 0.0001, two-tailed paired t-test). h, Normalized distance between vocal cords during SLN stimulation compared to basal breathing (Ctrl). (n = 3 trials; p = 0.0176, two-tailed paired t-test). i, Normalized distance between vocal cords during AEN stimulation compared to basal breathing (Ctrl). (n = 3 trials; p = 0.25, two-tailed Wilcoxon matched-pairs signed rank test).

Extended Data Fig. 7 |. NTS Tac1 neurons are required for coughing in mice.

a, Sigh rate in normoxia (p = 0.75) and hypoxia (p = 0.31) conditions in mice with (Ablation, n = 5) and without (Control, n = 6) ablation of Tac1 neurons. Two-tailed Mann-Whitney test. b, Sneezes observed in a 6-minute tussive challenge of capsaicin of mice before (Pre) and after (Post) ablation (n = 11 mice; p = 0.03, two-tailed Wilcoxon matched-pairs signed rank test). c, Quantification of sneezes in a 6-minute tussive challenge of capsaicin in mice with (+, n = 5, p = 0.9999, two-tailed Wilcoxon matched-pairs signed rank test) and without (−, n = 6, injected with AAV-DIO-mCherry, p = 0.75, two-tailed Wilcoxon matched-pairs signed rank test) hM4Di expression in Tac-1 neurons (Tac1-hM4Di) treated with (+) and without (−, vehicle only) CNO injection. d, Coughs in a 6-minute tussive challenge of citric acid in mice with hM4Di expressed in NTS Tac-1 neurons treated with (+) and without (−, vehicle only) CNO injection (n = 13, p = 0.0156, two-tailed Wilcoxon matched-pairs signed rank test). e, Sneezes in a 6-minute tussive challenge of citric acid in mice with hM4Di expressed in NTS Tac-1 neurons treated with (+) and without (−, vehicle only) CNO injection (n = 5; p = 0.9999, two-tailed Wilcoxon matched-pairs signed rank test). f, Coughs in a 10-minute challenge of an internasal injection of 20% polyvinyl alcohol in mice with hM4Di expressed in NTS Tac-1 neurons treated with (+) and without (−, vehicle only) CNO injection (n = 6; p = 0.0325, two-tailed Wilcoxon matched-pairs signed rank test). g, Sneezes in a 10-minute challenge of an internasal injection of 20% polyvinyl alcohol in mice with hM4Di expressed in NTS Tac-1 neurons treated with (+) and without (−, vehicle only) CNO injection (n = 6; p = 0.16, two-tailed paired t-test). h, Breathing frequency in normoxia and hypoxia conditions in mice with (+, n = 5) and without (−, n = 6, Tac1-Cre mice injected with AAV-DIO-mCherry) hM4Di expression in Tac-1 neurons (Tac1-hM4Di) treated with (+) and without (−, vehicle only) CNO injection. For control mice, p = 0.39, two-tailed paired t-test (normoxia) and p = 0.80, two-tailed paired t-test (hypoxia); for experimental mice, p = 0.60, two-tailed paired t-test (normoxia) and p = 0.12, two-tailed Wilcoxon matched-pairs signed rank test (hypoxia). All data presented as Mean ± SEM.

Extended Data Fig. 8 |. NTS Tac1 neurons are innervated by pulmonary vagal sensory neurons that express mechanoreceptor Piezo2.

a, Quantification of cough-like behaviors in a 6-minute capsaicin challenge of Tac1-Cre, Vglut2 fl/fl mice (n = 4) and their littermate controls (n = 14) (p = 0.0007, two-tailed Mann-Whitney test). Mean ± SEM. b, Quantification of cough-like behaviors in a 6-minute capsaicin challenge of Tac1 −/− mice (n = 13) and their littermate controls (n = 7; p = 0.50, two-tailed unpaired t-test). Mean ± SEM. c, Quantification of sneeze response to a 6-minute capsaicin tussive challenge in Trpv1−/− (n = 8) and control (n = 8) mice (p = 0.75, two-tailed unpaired t-test). Mean ± SEM. d, Quantification of sigh response to a 6-minute capsaicin tussive challenge in Trpv1−/− (n = 8 mice) and control (n = 8) mice (p = 0.13, two-tailed unpaired t-test). Mean ± SEM. e, Quantification of cough-like behaviors to a 6-minute citric acid tussive challenge in Trpv1−/− (n = 13) and control (n = 5) mice (p = 0.09, two-tailed Mann-Whitney test). Mean ± SEM. f, Quantification of the cough-like behaviors to a saline tussive challenge in Trpv1−/− (n = 13) and control (n = 9) mice (p = 0.95, two-tailed Mann-Whitney). Mean ± SEM. g, Quantification of the vagal ganglion neurons labeled by rabies virus that are in nodose (gray) and jugular (white) ganglia. (n = 177). h, Quantification of the vagal ganglion neurons labeled by both rabies virus and fast blue (FB) that are positive and negative for Trpv1 expression (gray). (n = 31). i, Multiplex in situ hybridization of vagal ganglion slice of Tac1 mouse injected with a rabies tracing method. The slice is probed for Egfp (green), Piezo2 (gray) and Phox2b (red). Blue, fast blue. j, Individual neurons from g are labeled by fast blue (FB), Egfp (green), Piezo2 (gray) and Phox2b (red). k, Quantification of the vagal ganglion neurons labeled by both rabies virus and fast blue (FB) that are positive and negative for Piezo2 expression (gray). (n = 17).

Extended Data Fig. 9 |. Trpv1 vagal neurons connect the airway to the NTS.

a, Scheme for projection tracing of the vagal Trpv1 neurons. b, Vagal ganglion slice stained with the nodose marker, Phox2b. c, Quantification of the labeled neurons that are in nodose ganglion (gray). (n = 285). d, Section of lung tissue shows tdTomato positive processes. e, Serial sections of the NTS showing the central projection of vagal Trpv1 neurons.

Extended Data Fig. 10 |. NTS Tac1 neurons integrate different downstream motor circuit modules to control distinct motor aspects of an expiratory effort.

a, Para-NA region with expression of ChR2-eYFP and optic fiber implantation. b, c, Individual (gray) and average (black) traces for 15 ms NA terminal activations (n = 25, 3 mice) aligned by compressive peak (b) and laser onset (c). d-f, Individual (gray) and average (black) traces for NA terminal activations for 50 ms (d), 100 ms (e), and 200 ms (f) (n = 30, 3 mice each). Blue, laser on. g, cVRG region with expression of ChR2-eYFP and optic fiber implantation. h, i, Individual (gray) and average (black) traces for cVRG terminal activations (n = 25, 3 mice) aligned by downward peak (h) and laser onset (i). j-l, Individual (gray) and average (black) traces for NA terminal activations for 50 ms (j), 100 ms (k), and 200 ms (l). Blue, laser on. m, Duration difference between the activation of the NTS Tac1 neurons (CB) (n = 30) and the activation of the NTS Tac1 terminals into the NA (NA) (n = 25) of the compressive phase (p = 0.19, two-tailed Mann-Whitney test). Mean ± SEM. n, Amplitude difference between CB (n = 30) and NA (n = 25) of the compressive phase (p = 0.0001, two-tailed unpaired t-test). Mean ± SEM. o, Duration difference between CB (n = 30), NA (n = 25) and the activation of the NTS Tac1 neurons into the cVRG (cVRG) (n = 24) of the expulsive phase (p = 0.02 NA vs. CB, p = 0.99 cVRG vs. CB; Kurskal-Wallis test, Dunn’s multiple comparisons test). NA and CB expulsive phases are the downward peak, while the cVRG expulsive peak is the upward peak. Mean ± SEM. p, Amplitude difference between CB (n = 30), NA (n = 25) and cVRG (n = 24) of the expulsive phase (p = 0.0001 NA vs CB, p = 0.01 cVRG vs CB; Kurskal-Wallis test, Dunn’s multiple comparisons test). NA and CB expulsive phases are the downward peak, while the cVRG expulsive peak is the upward peak. Mean ± SEM.

Fig. 1 |. Optogenetic activation of the NTS Tac1 neurons induces cough-like behaviors.

a, Schematic showing the WBP chamber. b, Top, raster plots of coughs during 3-min capsaicin nebulization (gray) and 3 min after. Bottom, cough rate; shaded area, s.e.m. n = 10 mice. c, Respiratory trace and audio of a capsaicin-evoked cough (red tick mark in b). d, Magnified depiction of c showing the three cough phases: inspiratory (I), compressive (C) and expulsive (E). e, Individual (gray) and average (black) traces for coughs induced by capsaicin (n = 30, three mice). f, Schematic for optogenetic activation of NTS Tac1 neurons. g, Raster plot of coughs after photostimulation (blue triangle, 15 ms); respiratory trace (red tick mark) depicted in h (n = 20, five mice). h, Respiratory trace and audio of a cough upon activation of NTS Tac1 neurons. i, Magnified depiction of h. j, Individual (gray) and average (black) traces for cough-like behaviors induced by optogenetics (n = 30, three mice). k, Time from laser onset to compressive peak when laser hits during eupneic inspiration (n = 32, three mice) versus eupneic expiration (n = 79, three mice; P = 0.0002, two-tailed Mann–Whitney test). Duration (l) and amplitude (m) of compressive phases for coughs induced by capsaicin (cough, n = 85), opto-activation of NTS Tac1 neurons (opto-activation, n = 30), compared to eupneic expiration (n = 30) (P = 0.0191 cough versus opto-activation, P = 0.0001 cough versus eupneic expiration, P = 0.0001 opto-activation versus eupneic expiration in l; P = 0.2109, 0.0001, 0.0001 in m). n,o, Duration (n) and amplitude (o) of compressive phases for coughs induced by capsaicin (cough, n = 85), opto-activation of NTS Tac1 neurons (opto-activation, n = 30), compared to eupneic inspiration (n = 30) (P = 0.0913, P = 0.0491, P = 0.0005 in n; P = 0.4665, P = 0.0001, P = 0.0001 in o). For l–n, Kurskal–Wallis test and Dunn’s multiple comparisons test were used. p, Control mice showed no response to laser stimulation. q, Individual (gray) and average (black) traces after light activation in control mice (n = 30, three mice). r, The probability of inducing a respiratory response in mice injected with ChR2 (n = 20, five mice) and eYFP (n = 20, three mice) (P = 0.0001, Mann–Whitney test). Data are presented as mean ± s.e.m. NS, not significant.

Fig. 2 |. NTS Tac1 neurons are activated by a tussive challenge.

a,b, Multiplex single-molecule in situ hybridization (RNAscope) of the NTS brain region from mice under control in a and tussive challenge in b probed for Tac1 (green) and c-Fos (red). The box regions are enlarged and shown in a′ and b′. Blue, DAPI. Arrows, cells positive for Tac1 and c-Fos. AP, area postrema. c, Quantification of the c-Fos-positive neurons in control (n = 4 mice) versus experimental (n = 4 mice) mice (P = 0.02, two-tailed unpaired t-test). Mean ± s.e.m. d, Quantification of c-Fos and Tac1 double-positive neurons in control (n = 4 mice) versus experimental (n = 4 mice) mice (P = 0.02, two-tailed unpaired t-test). Mean ± s.e.m. e, Percentage of Tac1 neurons that are c-Fos positive in control (n = 4 mice) versus experimental (n = 4 mice) mice (P = 0.0003, two-tailed unpaired t-test). Compared to controls, experimental mice had a statistically significant increase in c-Fos and Tac1 double-positive cells. Mean ± s.e.m. f, Quantification of Tac1-negative cells that are c-Fos positive. Note that there is no difference between control (n = 4 mice) and experimental (n = 4 mice) mice (P = 0.07, two-tailed unpaired t-test). Mean ± s.e.m. NS, not significant.

Fig. 3 |. NTS Tac1 neurons are required for tussive-agent-evoked cough-like behaviors.

a, Schematic showing the genetic ablation of Tac1 neurons in the NTS of a Tac1-Cre mouse. b, Mouse NTS brain slice shows the absence of Ai9 neurons in the Tac1-Cre mouse after the genetic ablation. Note that the labeling of neurons ventral to the NTS is a result of the developmental expression of Tac1-Cre (also see Supplementary Fig. 1). c, Quantification of cough-like behaviors of Tac1-Cre mice in a 6-min tussive challenge of capsaicin before (pre) and after (post) the ablation (n = 11 mice; P = 0.0002, two-tailed paired t-test). Mean ± s.e.m. d, Quantification of cough-like behaviors in a 6-min tussive challenge of citric acid before (pre) and after (post) the ablation (n = 3 mice; P = 0.04, two-tailed unpaired t-test). Mean ± s.e.m. e, Respiratory rate of animals with the ablation of NTS Tac-1 neurons (Ablation, n = 11) and control animals (Control, n = 7) in normoxia (21% O₂ balanced by N₂) and hypoxia (10% O₂ balanced by N₂). For normoxia, P = 0.86, two-tailed unpaired t-test. For hypoxia, P = 0.92, two-tailed Mann–Whitney test. Mean ± s.e.m. f, Schematic showing the hM4Di expression in Tac1 in the NTS of a Tac1-Cre mouse. g, Mouse NTS brain slice shows the expression of hM4Di-mCherry in the Tac1-Cre mouse. h, Quantification of cough-like behaviors in a 6-min tussive challenge of capsaicin in mice with (+, n = 12) and without (−, n = 6, Tac1-Cre mice injected with AAV-DIO-mCherry) hM4Di expression in Tac-1 neurons (Tac1-hM4Di) treated with (+) and without (−, vehicle only) CNO injection. For control mice, P = 0.65, two-tailed paired t-test. For experimental mice, P = 0.002, Wilcoxon matched-pairs signed-rank test. Mean ± s.e.m. i, Sigh rate in normoxia and hypoxia conditions in mice with (+, n = 5) and without (−, n = 6) hM4Di expression in Tac-1 neurons (Tac1-hM4Di) treated with (+) and without (−, vehicle only) CNO injection. For control mice, P = 0.50, two-tailed paired t-test in normoxia and P = 0.42 in hypoxia, two-tailed paired t-test; for experimental mice, P = 0.54 in normoxia, two-tailed paired t-test, and P = 0.13, two-tailed paired t-test in hypoxia. Mean ± s.e.m. AP, area postrema; NS, not significant.

Fig. 4 |. Trpv1-positive bronchopulmonary vagal ganglion neurons connect the airway to the NTS Tac1 neurons.

a, Quantification of cough-like behaviors in response to a 6-min tussive challenge in Trpv1^−/− (n = 13) and control (n = 9) mice (P = 0.0001, two-tailed Mann–Whitney test). Mean ± s.e.m. b, Scheme for retrograde tracing of vagal Trpv1 neurons that project to NTS. c, Vagal ganglion slice showing the retrogradely labeled neurons (tdTomato) in the nodose ganglion. Many of these neurons are labeled by a nodose ganglion marker, Phox2b. d, Quantification of the number of AAVretro-labeled cells in the nodose (gray) and jugular (white) (n = 795 neurons). e, A cross-section of the trachea showing the processes of Trpv1 neurons that project to the NTS. e′, Enlarged boxed region from e. f, A whole-mount image of a lobe of the lung with innervation of Trpv1 neurons (red). f′, Enlarged boxed region from f. g, Innervation of Trpv1 neurons (red) in the parenchyma of the lung. h, The process of the Trpv1 neurons (red) in the distal airway. i, Scheme for tracing the vagal bronchopulmonary neurons that are presynaptic to NTS Tac1 neurons. j, Quantification of the number of FB-labeled cells expressing Phox2b (gray) and those not expressing Phox2b (white) (n = 99 neurons). k, Multiplex in situ hybridization of vagal ganglion slice probed for Egfp, Trpv1 and Phox2b. Blue, FB. l,m, Individual neurons from k are labeled by FB, Egfp, Trpv1 and Phox2b.

Fig. 5 |. NTS Tac1 neurons coordinate downstream circuit modules to elicit distinct motor outputs of cough.

a, Processes (green) and presynaptic terminals (red) of Tac1 neuron in para-NA. ChAT, choline O-acetyltransferase, a marker for nucleus ambiguus. b, Schematic for optogenetic activating para-NA processes of NTS Tac1 neurons. c,d, Snapshots of vocal cords before (c) and after (d) NA terminal activation. Red line, measured distance. e, Normalized distance between vocal cords before (off) and during (on) laser activation (n = 6; P = 0.0001, two-tailed paired t-test). f, Respiratory trace and audio of a breathing trace upon a 15-ms photostimulation of para-NA processes. Upward peak represents compression because of the closure of the glottis. g, Magnified depiction of f. h, Processes (green) and presynaptic terminals (red) of Tac1 neuron in cVRG. i, Schematic for optogenetic activating cVRG processes of NTS Tac1 neurons. j,k, Snapshots of vocal cords before (j) and after (k) cVRG activation. l, Normalized distance between vocal cords before (off) and during (on) laser activation (n = 10; P = 0.82, two-tailed paired t-test). m, Respiratory trace and audio of a breathing trace upon a 15-ms photostimulation of cVRG processes. n, Magnified depiction of m. Upward peak represents an expiration due to the glottis remaining open. o, The time between laser onset to the initial upward peak, for NA terminal activations (n = 17, three mice) and cVRG terminal activations (n = 26, three mice). p, Probability of inducing a respiratory response in ChR2-injected mice for NA terminal activations (n = 14, three mice) and cVRG terminal activations (n = 15, three mice). q, The number of coughs (P = 1, two-tailed paired t-test), sneezes (P = 0.75, two-tailed Wilcoxon matched-pairs signed-rank test) and sighs (P = 0.99, two-tailed Wilcoxon matched-pairs signed-rank test) of experimental mice (n = 3) in a 6-min tussive challenge of capsaicin before (pre) and after (post) surgery. r, Schematic for retrograde circuit tracing. s, Brainstem slice of the NTS displaying Tac1-Cre neurons by retrograde AAV injected into the cVRG (green) and para-NA (red). Filled arrowheads, neurons labeled by both AAVs; open arrowheads, neurons labeled by only one AAV. AP, area postrema; 10N, the dorsal motor nucleus of the vagus. t, Quantification of the ratios of double-labeled neurons over total neurons labeled by either AAV (n = 4 mice). Data are presented as mean ± s.e.m. NS, not significant.

Fig. 6 |. A vagal–brainstem circuit that controls cough-like behaviors.

a, Model for the cough control circuit. The tussive stimuli in the airways activate the Phox2b-positive vagal ganglion sensory neurons (Trpv1 positive or Piezo2 positive) that directly synapse on the Tac1-positive neurons in the NTS, which, in turn, project to the para-NA region and the cVRG region in the ventrolateral medulla to release glutamate to elicit a cough-like behavior. b, NTS receives different interoceptive stimuli through the vagal afferents. Tac1 neurons in the NTS are activated by both chemo-tussive and mechano-tussive stimuli and coordinate different medullary circuit modules (that is, para-NA and cVRG) to control the distinct motor phases of a cough-like behavior. The cough-like behaviors contain the compressive phase and the expulsive phase. The Tac1 neurons–para-NA circuit controls the compressive phase, whereas the Tac1 neurons–cVRG circuit controls the expulsive phase.