Oxytocin modulates respiratory heart rate variability through a hypothalamus-brainstem-heart neuronal pathway

Abstract

The variation in heart rate in phase with breathing is called respiratory heart rate variability (RespHRV). Relaxation and positive socio-emotional states can amplify RespHRV, yet the underlying mechanism remains largely unknown. Here we identify a hypothalamus-brainstem neuronal pathway in rodents through which oxytocin (OT) amplifies RespHRV during calming behavior. OT neurons from the caudal paraventricular nucleus in the hypothalamus regulate the activity of a subgroup of inhibitory neurons in the pre-Bötzinger complex, the brainstem nucleus that generates the inspiratory rhythm. Specifically, OT enhances the glycinergic input from OT-receptor-expressing neurons in the pre-Bötzinger complex to cardiac-innervating parasympathetic neurons in the nucleus ambiguus during inspiration. This leads to amplified respiratory modulation of parasympathetic activity to the heart, thereby enhancing RespHRV. We show that OT neurons participate in the restoration of RespHRV amplitude during recovery from stress in mice, indicating that OT acts centrally to regulate cardiac activity during a calming behavior.

Fig. 1. OT fibers from the caudal PVN to the preBötC/nA can amplify RespHRV, decrease mHR and increase respiratory frequency in adult freely moving mice at rest.

a, Immunohistochemical labeling of OT fibers in the preBötC and nA, using FluoroGold (FG) retrograde tracing (Extended Data Fig. 1a,b) and classical anatomical landmarks (counts in n = 6 mice for nA, n = 6 mice for preBötC; Extended Data Fig. 1c,d). Scale bars, 50 µm. b, Retrograde tracing to localize OT neurons that project onto preBötC/nA neurons (n = 2, bilateral injections) (Extended Data Fig. 1e–g). 3V, third ventricle. Scale bars: large image, 100 µm; inset, 50 µm. c, ChR2-tdTomato expression in OT neurons (Extended Data Fig. 2a; n = 1) and in fibers in the preBötC/nA (n = 6 mice). ChAT, choline acetyltransferase. Scale bars: left, 100 µm; right, 50 µm. d, Strategy for bilateral photoexcitation of OT fibers in the preBötC/nA (verified a posteriori in n = 2 mice). NK1-R, neurokinin 1 receptor. Scale bar, 500 µm. e, Photoexcitation (1 min, 50 Hz, 10 ms pulses) of OT fibers in the preBötC/nA of freely moving mice (respiration measured by plethysmography; HR measured with ECG telemetry probes). Adult Oxt-Cre; Ai27(Rosa26-LSL-ChR2-tdTomato) (Cre⁺, n = 8) and Ai27(Rosa26-LSL-ChR2-tdTomato) female littermates (Cre⁻, n = 4) were used. Gray shaded areas on the left trace correspond to the recordings shown in the middle trace. bpm, beats per minute. f, Effects of the photoexcitation of OT fibers in the preBötC/nA of Cre⁺ mice on RespHRV, mHR, respiratory frequency (freq) and respiratory amplitude (ampl) (n = 8 mice). Repeated-measures one-way ANOVA, Tukey’s multiple comparison. cpm, cycles per minute; a.u., arbitrary units. g, Quantification of the relative effects (delta (Δ) in % or bpm for mHR) induced by photoexcitation of OT fibers in Cre⁺ (from absolute values shown in f) and Cre⁻ (absolute values shown in source data file) mice. Intra-group Δ changes (photoexcitation vs pre-photoexcitation, shown in f), repeated-measures one-way ANOVA with Tukey’s multiple comparison. Inter-group comparison (Cre⁺ vs Cre⁻), unpaired two-sided t-test. In violin plots, the dashed lines indicate the median and the dotted lines represent the quartiles. h, Correlation analysis between the Δ changes in RespHRV amplitude and mHR with the pre-photoexcitation values of these parameters. Pearson two-sided correlation analysis, simple linear regression plotted. Detailed statistics are presented in Supplementary Table 1. Source data

Fig. 2. PVN^OT fibers can amplify RespHRV by releasing OT in the preBötC/nA in adult anesthetized mice.

a–c, Unilateral photoexcitation of PVN^OT fibers in the preBötC/nA of anesthetized Oxt-Cre; Ai27(Rosa26-LSL-ChR2-tdTomato) adult female and male mice (a) induces RespHRV amplification and mHR decrease (b, representative traces; c, quantifications) (n = 15 Cre⁺ females, n = 15 Cre⁺ males, n = 5 Cre⁻ females and males). Intra-group Δ changes (photoexcitation vs pre-photoexcitation); repeated-measures one-way ANOVA with Tukey’s multiple comparison. Inter-group comparison (Cre⁺ females vs Cre⁺ males vs Cre⁻), one-way ANOVA with Tukey’s multiple comparison, except for respiratory amplitude, Kruskal–Wallis test with Dunn’s multiple comparison. In violin plots, the dashed lines indicate the median and the dotted lines represent the quartiles (c). Raw data are shown in Extended Data Fig. 3a. IntC EMG, intercostal electromyography. d, Correlation analysis between the Δ changes in RespHRV amplitude and mHR with the pre-photoexcitation values of these parameters. Pearson two-sided correlation analysis; simple linear regression plotted. e–g, Ipsilateral (ipsi) injection of a selective OT-R antagonist ((d(CH₂)₅¹,Tyr(Me)²,Thr⁴,Orn⁸,des-Gly-NH₂⁹)-Vasotocin; 200 nl at 1 µM) in the preBötC/nA (e) abolishes the photoexcitation-induced RespHRV amplification but not the mHR decrease (f, representative traces obtained in the conditions shown above in e; g, quantifications). Photoexcitation following ipsilateral vehicle injection (g, traces shown in Extended Data Fig. 4a), and photoexcitation in the contralateral (contra) preBötC/nA, induced similar RespHRV and mHR effects as ipsilateral pre-injection photoexcitation (n = 9 for ipsi and OT-R antagonist + ipsi; n = 6 for ipsi and OT-R antagonist + ipsi and OT-R antagonist + contra; n = 5 for ipsi and vehicle + ipsi). Intra-condition Δ changes (photoexcitation vs pre-photoexcitation), repeated-measures one-way ANOVA with Tukey’s multiple comparison. Inter-condition comparisons for OT-R antagonist (ipsi vs OT-R antagonist + ipsi vs OT-R antagonist + contra), repeated-measures mixed-effects analysis with the Geisser–Greenhouse correction, Tukey’s multiple comparison. Inter-condition comparisons for vehicle (ipsi vs vehicle + ipsi), paired two-sided t-test. Violin plots are represented in gray, dashed lines indicate the median and dotted lines represent the quartiles (g). Raw data for RespHRV and mHR, and data for the effects on respiratory frequency and respiratory amplitude are shown in Extended Data Fig. 3c–e. Localization of injection spots is shown in Extended Data Fig. 4b. Detailed statistics are presented in Supplementary Table 1. Source data

Fig. 3. Anatomical and neurochemical characterization of OT-R⁺ cells in the preBötC/nA.

a, Crossing of Oxtr-Cre mice with Ai14(Rosa26-LSL-tdTomato) mice to express tdTomato only in OT-R⁺ cells. FG was injected intraperitoneally in adult mice to label nA and DMV neurons (n = 6 mice for nA counts; n = 5 mice for DMV counts). nA^cardiac neurons were detected as nA FG⁺ neurons that do not express the calcitonin gene-related peptide (CGRP⁻). Arrows indicate labeled neurons, with white arrows showing double-labeled OT-R⁺FG⁺ neurons. Individual data are shown in Extended Data Fig. 7a. Scale bars, 100 µm. b, Unilateral preBötC boundaries were mapped using contralateral preBötC FG injection (Extended Data Fig. 1a,c) and using classical anatomical landmarks of the preBötC location. PreBötC^OT-R (OT-R⁺) cells were tested for their contralateral projection phenotype (FG⁺; n = 4 mice), neuronal vs astrocytic phenotype (NeuN⁺ vs glutamine synthetase (GS)⁺, respectively; n = 4 mice), and for the expression of two classical markers of preBötC neurons (NK1-R⁺, n = 4 mice; µ-opioid receptor (µO-R⁺), n = 6 mice). Arrows indicate labeled neurons, with white arrows showing double labeling between OT-R⁺ and the cyan marker. Individual data are shown in Extended Data Fig. 7b,c. Scale bar for preBötC^OT-R and FG only, 100 µm; other scale bars, 20 µm. c, OT-R⁺ cells are densely present in the preBötC, and minimally present in the adjacent Bötzinger complex (BötC) (n = 4 mice, data presented as means; error bars, s.e.m.). nA_sc, semi-compact formation of the nA; nA_c, compact formation of the nA. Scale bars, 100 µm. d–e, RNAscope in situ hybridization in adult wild-type mice to detect RNA transcripts encoding the expression of the OT-R (highest expression density in the DMV (d(i))) and of a glycinergic or glutamatergic phenotype (n = 4 mice). Representative examples from the preBötC of two mice, showing OT-R transcript expression (magenta arrows) only in glycinergic neurons (d(ii), cyan arrows), or in both a glycinergic and a glutamatergic neuron (e, cyan and yellow arrows, respectively). White arrows indicate dually labelled neurons. Scale bars: full image in d, 200 µm; d(i) inset, 50 µm; d(ii) inset, 20 µm; e, 20 µm. f, Relative quantification of the phenotype of preBötCOT-R neurons. Individual data are shown in Extended Data Fig. 7d. Detailed statistics are presented in Supplementary Table 1. Source data

Fig. 4. PreBötC^OT-R neurons control RespHRV amplitude and project onto nA^cardiac neurons.

a, Viral-mediated expression of the somBiPOLES optogenetic construct in preBötC^OT-R neurons of Oxtr-Cre mice (Extended Data Fig. 7i) (n = 10 mice). Viral injections in Cre⁻ mice did not induce somBiPOLES expression (Extended Data Fig. 7h; n = 4 mice). An anti-OT-R antibody was used for immunohistochemical labeling. Scale bars: full image, 100 µm; inset, 20 µm. b, Effects of bilateral photomodulations of preBötC^OT-R neurons in an anesthetized mouse (Cre⁺) on HR and inspiratory activity (IntC EMG), before and after intraperitoneal injection of the muscarinic receptor antagonist atropine (10 mg kg⁻¹, 600 µl), and compared to a control (Cre⁻) mouse. c,d, Expanded traces from the gray shaded areas in b, and individual data showing bidirectional cardiorespiratory effects of photoexcitation (c) vs photoinhibition (d) of preBötC^OT-R neurons (n = 8 Cre⁺ mice for photoexcitation and n = 10 Cre⁺ mice for photoinhibition without treatment; n = 6 Cre⁺ mice for photoexcitation or photoinhibition before vs after atropine; n = 4 Cre⁻ mice). Intra-group Δ changes (photoexcitation vs pre-photoexcitation), repeated-measures one-way ANOVA with Tukey’s multiple comparison except for Cre⁻ RespHRV photoinhibition, Friedman test with Dunn’s multiple comparison. Cre⁺ before and after atropine comparison, paired two-sided t-test except for respiratory amplitude photoexcitation, Wilcoxon two-sided matched-pairs signed rank test. Inter-group comparison (Cre⁺ no treatment vs Cre⁺ atropine vs Cre⁻ no treatment), Kruskal–Wallis test with Dunn’s multiple comparison. Violin plots are represented in gray, dashed lines indicate the median and dotted lines represent the quartiles. Raw data are shown in Extended Data Fig. 7j,k. e, Viral-mediated expression of tdTomato and synaptophysin-enhanced GFP (SypEGFP) in preBötC^OT-R neurons (n = 4 mice). Scale bar, 100 µm. f,g, PreBötC^OT-R neurons project to (tdTomato⁺ fibers) and make putative pre-synaptic contacts with (SypEGFP⁺ puncta) nA^cardiac neurons (FG⁺BCHE⁺ and FG⁺Calb⁺ neurons; Extended Data Fig. 7l; n = 4 mice). BCHE, butyrylcholinesterase; Calb, calbindin. Scale bars: in f, 50 µm; in g, 20 µm. Detailed statistics are presented in Supplementary Table 1. Source data

Fig. 5. TGOT increases the frequency of preBötC inspiratory bursts, depolarizes preBötC^OT-R neurons that are inspiratory bursting and amplifies the glycinergic post-synaptic currents occurring in nA neurons during inspiratory bursts in rhythmic slices in vitro.

a–c, TGOT (0.5 µM) increases the frequency of preBötC inspiratory bursts in rhythmic preBötC slices from neonatal wild-type mice (n = 12 mice; ⎰preBötC, integrated extracellular recordings; V_m, whole-cell current-clamp recordings). Gray shaded areas in (a) are expanded in (b). Wilcoxon two-sided matched-pairs signed rank test (c). The violin plot is represented in gray, dashed line indicates the median and dotted lines represent the quartiles (c). d,e, TGOT (0.5 µM) increases the frequency of preBötC inspiratory bursts in preBötC ‘island’ preparations (n = 5 mice; d, representative trace; e, quantifications). Paired two-sided t-test. The violin plot is represented in gray, dashed line indicates the median and dotted lines represent the quartiles (e). f–h, PreBötC^OT-R neurons recorded in rhythmic preBötC slices from neonatal Oxtr-Cre; Ai14(Rosa26-LSL-tdTomato) mice display an inspiratory bursting pattern of activity (f), and TGOT (0.5 µM) depolarizes preBötC^OT-R neurons following the blockade of fast synaptic transmission (bicuculline (Bic), 10 µM; strychnine (Stry), 5 µM; CNQX, 20 µM; n = 7 mice) (g, representative traces; h, quantifications). Paired two-sided t-test. The violin plot is represented in gray, dashed line indicates the median and dotted lines represent the quartiles (h). i, Example of a recorded preBötC^OT-R neuron (dye⁺ and OT-R;tdTomato⁺, white arrows) located in the NK1-R-expressing preBötC region (n = 4 neurons recorded that were labeled and localized a posteriori). Scale bars: full image, 40 µm; inset, 20 µm. j, All respiratory-modulated nA neurons recorded (n = 10 wild-type mice) were hyperpolarized during the preBötC inspiratory bursts, causing inhibition of their spiking activity. k,l, nA neurons showed volleys of inhibitory currents during preBötC inspiratory bursts in the control condition (I_m, whole-cell voltage-clamp recordings; V_h, −40 mV holding voltage), which were amplified by TGOT (0.5 µM) application (n = 10 mice; k, representative traces; l, quantifications). During TGOT application, bicuculline (10 µM) application had no effect, while strychnine (5 µM) application either abolished all currents (n = 5 mice, top traces in k) or exposed the presence of volleys of excitatory currents during preBötC inspiratory bursts (n = 3 mice, bottom traces in k, further characterized in Extended Data Fig. 8d). Control vs TGOT, paired two-sided t-test. Δ changes following TGOT, bicuculline and strychnine applications, Friedman test with Dunn’s multiple comparison. Violin plots are represented in gray, the dashed lines indicate the median, the dotted lines represent the quartiles and the green line represents the limit above which the recorded currents are inhibitory and under which they are excitatory (l). m, Example of a recorded nA neuron (arrows indicate the dye⁺ recorded neuron that is Phox2b⁺ and ChAT⁺; n = 5 neurons recorded that were labeled and localized a posteriori). Scale bars: full image, 40 µm; inset, 20 µm. Detailed statistics are presented in Supplementary Table 1. Source data

Fig. 6. TGOT injection in the preBötC induces RespHRV amplification owing to amplified respiratory modulation of cardiac parasympathetic activity in adult anesthetized rats and in in situ WHBP of juvenile rats.

a–c, The bilateral injection of TGOT (50 nl at 0.5 µM) in the preBötC of adult anesthetized Wistar rats (a) induced an amplification of RespHRV, a decrease in mHR and an increase in respiratory amplitude but no change in respiratory frequency (b,c). In violin plots, dashed lines indicate the median and dotted lines represent the quartiles (c). Mapping of injection spots (a) in Extended Data Fig. 9c (n = 15 TGOT and n = 10 vehicle). Scale bar, 300 µm (a). Before vs after TGOT injection (n = 15, c), paired two-sided t-test (respiratory frequency) or Wilcoxon two-sided matched-pairs signed rank test. Δ effects induced by TGOT or vehicle (n = 10, absolute values in Extended Data Fig. 9b) injections, Mann–Whitney two-sided test or unpaired two-sided t-test (mHR). d–j, Bilateral TGOT injection in the preBötC of in situ WHBPs (d) of juvenile Wistar rats (P21–P30) induced an amplification of RespHRV (e,f) and an increase in the respiratory modulation of the cardiac vagal branch activity (CVBA) (h,i), with no change in mHR, amplitude or frequency of the phrenic nerve activity (PNA), amplitude of the cervical vagal nerve activity (VNA), mean thoracic sympathetic nerve activity (tSNA) or mean perfusion pressure (PP). Before vs after TGOT injection (f,i) (RespHRV, n = 16; mHR, n = 18; PNA frequency, n = 19; PNA amplitude, n = 18; VNA, n = 7; tSNA, n = 8; mPP, n = 18; CVBA, n = 7), paired two-sided t-test or Wilcoxon two-sided matched-pairs signed rank test. Δ effects induced by TGOT or vehicle injections (absolute values for vehicle injections in Extended Data Fig. 9e; RespHRV, n = 8; mHR, n = 8; PNA frequency, n = 9; PNA amplitude, n = 8; VNA, n = 9; tSNA, n = 8; mPP, n = 8), Mann–Whitney two-sided test or unpaired two-sided t-test. In violin plots, dashed lines indicate the median and dotted lines represent the quartiles (f,i). Pearson two-sided correlation analysis, simple linear regression plotted (g). Mapping of injection spots (j) in Extended Data Fig. 9h (n = 19 TGOT and n = 4 vehicle). Scale bars: large image, 300 µm; inset, 100 µm. Detailed statistics are presented in Supplementary Table 1. Source data

Fig. 7. Chemogenetic inhibition of all OT neurons slows down the recovery of RespHRV amplification following a restraint stress test.

a, Expression of the HA-tagged inhibitory chemogenetic actuator hM4Di in all OT neurons in adult transgenic mice (n = 4 mice). 3V, third ventricle. Scale bar, 100 µm. b, Whole-cell current-clamp recordings of PVN^OT neurons expressing tdTomato and hM4Di in slices from Cre⁺ mice, showing that application of the DREADD agonist C21 (10 µM, for 5 min) inhibits action potentials. C21 had no effect on neurons recorded in the PVN in Cre⁻ mice (n = 3 Cre⁺ and n = 2 Cre⁻ P20 mice, at least three recorded neurons per mouse, each from a different slice). Repeated-measures one-way ANOVA with the Geisser–Greenhouse correction, Tukey’s multiple comparison. c, Experimental strategy to test the effect of chemogenetic inhibition of OT neurons on respiratory and cardiovascular parameters, before and after a restraint stress test in a ventilated cylinder. The mice were implanted with blood pressure telemetry probes to enable measurements of mean blood pressure (mBP) and HR. They were placed in a plethysmography chamber (to which they were habituated for 1 week) during the pre-stress and post-stress conditions, to enable measurement of their respiratory activity. d–f, Effects induced by the intraperitoneal injection of the hM4Di agonist C21 or vehicle in adult Cre⁺ mice (n = 7, three trials averaged for each condition (C21 and vehicle) in each mouse; C21 or vehicle trials spaced by at least 24 h). Traces on the right for both the C21 and vehicle conditions are enlargements of the grayed areas on their respective traces on the left (d). Chemogenetic inhibition of OT neurons did not alter RespHRV amplitude at rest (e), but slowed down the recovery of RespHRV amplification following the restraint test (orange traces in d) (f). Pre-stress conditions, C21 vs vehicle; paired two-sided t-test (e). Stress and post-stress Δ (in % or bpm) effects compared to the pre-stress condition, C21 vs vehicle, repeated-measures two-way ANOVA with Šidák’s multiple comparison (f). Cre⁻ data are shown in Extended Data Fig. 10. Violin plots are represented in gray, dashed lines indicate the median and dotted lines represent the quartiles (f). Detailed statistics are presented in Supplementary Table 1. Source data

Fig. 8. Schematic model of the preBötC^OT-R→nA^cardiac neuronal circuit that controls RespHRV at rest and its modulation by PVN^OT neurons for RespHRV amplification during a calming behavior.

a, At rest, OT is not involved in the regulation of RespHRV amplitude. Yet expression of the OT-R in a subgroup of preBötC neurons, the group that generates the inspiratory rhythm, defines a subpopulation of cells that participates in the control of RespHRV. PreBötC^OT-R neurons are active during inspiration and induce glycinergic inhibition of the activity of nA^cardiac neurons by putative monosynaptic contacts. This results in a decrease in cardiac parasympathetic activity that is concomitant with each inspiratory burst in the phrenic nerve and at the effector level in an increase in HR during inspiration. b, PVN^OT neurons are involved in the recovery of RespHRV amplitude during calming behavior following a stress. The release of OT by PVN^OT neurons raises the excitability of preBötC^OT-R neurons, leading to the amplification of the preBötC^OT-R→nA^cardiac glycinergic connectivity during inspiration. This induces an amplification of the respiratory modulation of parasympathetic activity to the heart, and thus an amplification of RespHRV. OT released endogenously in the preBötC induces little modulation of respiratory activity, which is not correlated with the RespHRV amplitude. Inhibition of PVN^OT neurons slows the recovery of RespHRV amplitude following a stress.

Extended Data Fig. 1. OT fibers are predominantly present in the preBötC compared to the nA, and OT neurons from the caudal PVN form a dense and regionally-specific cluster projecting to preBötC/nA neurons.

a, FluoroGold injection in the preBötC unilaterally labels a subgroup of contralateral preBötC neurons, enabling delineation of the preBötC boundaries. PreBötC injections were guided by extracellular recordings made with the injection pipette, to localize the peak pre-inspiratory/inspiratory bursting activity on the rostral side of the neuronal cell column presenting inspiratory bursting activity. b, Intraperitoneal (IP) injection of the retrograde tracer FluoroGold labels nucleus ambiguus neurons. c, Coronal sections at three rostro-caudal levels (insets) with immunohistochemical labeling of OT fibers, and labeling of either contralaterally-projecting preBötC neurons (top, n = 6 mice) or nA neurons (bottom, n = 6 mice) with FluoroGold (FG). The preBötC contains a greater density of OT fibers compared to the nA. Scale bars, 100 µm. d, Average number of OT fibers counted in three brainstem sections (50 µm thick) per animal (n = 6 for preBötC, n = 6 for nA). Paired two-sided t-test for preBötC vs. nA. e-g, Bilateral injection of the retrograde tracer cholera toxin B in the preBötC/nA. OT neurons that project to the preBötC/nA are restricted to the dorso-caudal PVN. The most rostral level showing OT neurons co-labelled with cholera toxin B is at ~ Bregma -0.94 mm (g). The level with the highest proportion of OT neurons co-labelled with cholera toxin B is at ~ Bregma -1.06 mm (e, f). At this level, only three structures are labeled with cholera toxin B (e), the cortex (I in e), the PVN (II in e), and the central amygdala (III in e), similarly to what was shown previously using monosynaptic retrograde viral tracing from preBötC neurons. Counts of the number of neurons expressing OT and/or cholera toxin B (CTB) (inset in e), expressed as relative proportions (%), were performed in the dorso-caudal PVN where cholera toxin B labeling was found (n = 2 mice). No cholera toxin B labeling was found in the supraoptic nucleus (SON) (f). Scale bars, 500 µm (whole section in e), 100 µm (I, II and III in e), 200 µm (f, g). Detailed statistics are presented in Supplementary Table 1. Source data

Extended Data Fig. 2. Specificity of ChR2-tdTomato expression in the OT neurons of OT::Cre;Ai27(LSL-ChR2-tdTomato) mice.

a, The selective expression in OT neurons of the blue light gated cation channel ChR2 fused to the fluorescent protein tdTomato is induced by crossing OT::Cre mice with Ai27(LSL-ChR2-tdTomato) mice (n = 1 mouse). White arrows in insets indicate neurons counted as OT⁺ and ChR2-tdTomato⁺. A magenta arrow indicates a neuron counted as OT⁻ and ChR2-tdTomato⁺, therefore representing one of the rare neurons counted as non-specific expression (3% of OT⁻ ChR2⁺ neurons, Fig. 1c). This very low non-specific expression is likely overestimated by our count, because of the limited sensitivity of OT immunohistochemical labeling leading to an underestimation of the number of OT neurons, and because we counted neurons showing faint OT labeling as OT⁻ neurons (such as the one indicated by the magenta arrow). Scale bar, 200 µm. b, Littermates that did not express the Cre recombinase (Cre⁻), but were genotyped positive for the LSL-ChR2-tdTomato transgene, obtained from crossings between OT::Cre mice with Ai27(LSL-ChR2-tdTomato) mice, were used as control mice. No ChR2-tdTomato expression was found in these mice (n = 1 mouse). Scale bar, 200 µm.

Extended Data Fig. 3. Individual data showing the respiratory and cardiac effects of the photoexcitation of PVN^OT fibers in the preBötC/nA, before and after injection of an OT-R antagonist.

a, Individual data from the photoexcitation of OT fibers unilaterally in the preBötC/nA of adult anesthetized OT::Cre;Ai27(LSL-ChR2) mice (n = 15 Cre⁺ females, n = 15 Cre⁺ males, n = 5 Cre⁻ females and males). Cre⁺ female vs. male mice, pre-photoexcitation vs. photoexcitation vs. post-photoexcitation periods, repeated-measures two-way ANOVA, Sidak’s multiple comparison. Cre⁻ mice, repeated-measures one-way ANOVA, Tukey’s multiple comparison. freq, frequency; ampl, amplitude. b, In the mice subgroups (mixed sexes) shown in a, an OT-R antagonist ((d(CH₂)₅¹,Tyr(Me)²,Thr⁴,Orn⁸,des-Gly-NH₂⁹)-Vasotocin, 200 nl at 1 µM) or vehicle was then injected unilaterally in the preBötC/nA (n = 9 for OT-R antagonist injections, n = 5 for vehicle injections). Pre-injection (Control) vs. post-injection (OT-R antagonist or Vehicle), paired two-sided t-test or Wilcoxon two-sided matched-pairs signed rank test. c-d, Following the OT-R antagonist or vehicle injections, photoexcitations were repeated at the same unilateral site (OT-R antagonist + ipsilateral data in c, vehicle + ipsilateral in d, compared to pre-injections ipsilateral data) (n = 9 for ipsilateral and OT-R antagonist + ipsilateral in c, n = 5 for ipsilateral and vehicle + ipsilateral in d). In a subgroup of mice injected with the OT-R antagonist, photoexcitations in the contralateral preBötC/nA were made after the ipsilateral photoexcitations (n = 6, OT-R antagonist + contralateral in c). Repeated-measures one-way ANOVA, Tukey’s multiple comparison. e, Delta (Δ) changes in respiratory frequency and respiratory amplitude, from the pre-photoexcitation and photoexcitation data shown in c and d. Intra-condition Δ changes (photoexcitation vs. pre-photoexcitation), repeated-measures one-way ANOVA, Tukey’s multiple comparison. OT-R antagonist comparisons (ipsi vs. OT-R antagonist + ipsi vs. OT-R antagonist + contra), repeated-measures mixed-effects analysis with the Geisser-Greenhouse correction, Tukey’s multiple comparison. Vehicle comparisons (ipsi vs. vehicle + ipsi), paired two-sided t-test. Detailed statistics are presented in Supplementary Table 1. Source data

Extended Data Fig. 4. Raw traces for the photoexcitation of PVN^OT fibers in the preBötC/nA before and after vehicle injection, localization of the injection spots of the OT-R antagonist and vehicle, and control experiments for photoexcitations in the DMV.

a, Vehicle injection in the preBötC/nA did not alter the effects induced by the photoexcitation of PVN^OT fibers (individual data in Fig. 2g and Extended Data Fig. 3d). b, Localization of the injection spots of a selective OT-R antagonist (200 nl) and of vehicle (200 nl) in the preBötC/nA, for data shown in Fig. 2e-g, Extended Data Fig. 3b-e, and Extended Data Fig. 5a-f. Each spot corresponds to a single injection in one mouse. c, In Cre⁻ mice, blue light illumination in the DMV induced no cardiorespiratory changes (n = 5 mice). Pre-photoexcitation vs. photoexcitation vs. post-photoexcitation periods, RM one-way ANOVA, Tukey’s multiple comparison, except for mHR, Friedman test, Dunn’s multiple comparison. Detailed statistics are presented in Supplementary Table 1. Source data

Extended Data Fig. 5. PVN^OT fibers can decrease mHR via DMV connectivity in adult anesthetized mice.

a-f, Photoexcitation of PVN^OT fibers either in the preBötC/nA or in the DMV of anesthetized OT::Cre;Ai27(LSL-ChR2) adult mice induce the same effects, both before or after injection of the OT-R antagonist ((d(CH₂)₅¹,Tyr(Me)²,Thr⁴,Orn⁸,des-Gly-NH₂⁹)-Vasotocin, 200 nl at 1 µM) in the preBötC/nA (n = 6 for preBötC/nA and DMV before OT-R antagonist, n = 7 for preBötC/nA and DMV after OT-R antagonist). Traces in (c) were obtained in the conditions shown above in (a). b, Fluorescent microbeads in the injectate solutions enabled the localization of the injection spots for the OT-R antagonist or vehicle a posteriori (n = 9 mice for OT-R antagonist injections and n = 5 mice for vehicle injections, mapped in Extended Data Fig. 4b). Scale bar, 500 µm. d, repeated-measures one-way ANOVA, Tukey’s multiple comparison. e, repeated-measures one-way ANOVA, Tukey’s multiple comparison. f, Δ changes (photoexcitation vs. pre-photoexcitation), repeated-measures one-way ANOVA, Tukey’s multiple comparison. Comparisons before and after OT-R antagonist injection (preBötC/nA vs. DMV), paired two-sided t-test. g, CUBIC clearing of thick (1 mm) coronal brainstem slices at the level of the preBötC/nA in OT::GFP mice (n = 3 mice) shows OT fibers crossing both the preBötC/nA and the DMV. Scale bar, 500 µm. Detailed statistics are presented in Supplementary Table 1. Source data

Extended Data Fig. 6. Correlation analysis between RespHRV and mHR in basal states and as delta changes in various experimental conditions.

a, In freely moving mice (Fig. 1e-h), RespHRV amplitude and mHR are not correlated, in both pre-stim (basal) states and as Δ changes during photoexcitation of OT fibers in the preBötC/nA (n = 8 mice). b, In anaesthetized mice (Fig. 2a-d and Extended Data Fig. 4a), RespHRV amplitude and mHR are not correlated during pre-stim (basal) state, but their Δ changes during photoexcitation of OT fibers in the preBötC/nA are correlated (n = 15 female and n = 15 male mice). c, In a subgroup of mice shown in (b), in experimental conditions presented in Fig. 2e-g, Extended Data Fig. 3c-e and Extended Data Fig. 5a-f, RespHRV amplitude and mHR are not correlated during the pre-stim (basal) state (c1, n = 14 mice), but their Δ changes during photoexcitation of OT fibers in the preBötC/nA are correlated (c2, n = 14 mice). No correlation is evident in other experimental conditions (photostimulations in the DMV (c3, n = 6 mice), or photostimulations following OT-R antagonist injection in the preBötC/nA (c4, n = 9 mice; c5, n = 6 mice; c6, n = 7 mice)). d, In anaesthetized mice (Fig. 4b-d and Extended Data Fig. 7i-k, n = 10 mice), RespHRV amplitude and mHR are not correlated during the pre-stim (basal) state, but their Δ changes during photomodulations of preBötC^OT-R neurons are correlated (orange and blue backgrounds represent data from photoexcitations and photoinhibitions, respectively). e, In anesthetized rats (Fig. 6a-c), RespHRV amplitude and mHR are not correlated, in both the pre-TGOT (basal) state (n = 19 rats) and as Δ changes after TGOT injection in the preBötC (n = 14 rats). f, In rat in situ Working Heart-Brainstem Preparations (Fig. 6d-i), RespHRV amplitude and mHR are not correlated, in both the pre-TGOT (basal) state and as Δ changes after TGOT injection in the preBötC (n = 18 rats). a-f, Pearson two-sided correlation analysis, simple linear regression plotted. Detailed statistics are presented in Supplementary Table 1. Source data

Extended Data Fig. 7. Individual data for the anatomo-functional characterization of OT-R⁺ cells in the preBötC/nA, and absence of somBiPOLES expression in Cre⁻ mice.

a-c, Individual data for the averaged counts shown in Fig. 3a-b. Each point represents the total number of cells counted unilaterally in one mouse (n = 6 mice for nA counts and n = 5 mice for DMV counts in a; n = 4 mice in b; n = 4 mice for FG⁺ counts, n = 4 mice for NK1-R⁺ counts and n = 6 mice for µO-R⁺ counts in c; data presented as mean ± s.e.m. ). d, Individual data for the averaged counts shown in Fig. 3f. Each point represents the total number of cells counted bilaterally in three brainstem coronal sections containing the preBötC of each mouse (n = 4 mice, data presented as mean ± s.e.m.). e-f, Crossing of OT-R::Cre;Ai14(LSL-tdTomato) mice with GAD67::GFP mice (e) or GlyT2::GFP mice (f) to label OT-R⁺ cells (tdTomato⁺) and GABAergic or glycinergic neurons (GFP⁺), respectively. Scale bars: full images, 100 µm; insets, 50 µm. g, Individual data for the averaged counts shown in e and f (n = 4 mice for GABAergic counts, n = 4 mice for glycinergic counts, data presented as mean ± s.e.m.). h, Absence of somBiPOLES expression in Cre⁻ mice following viral injection in the preBötC (n = 4 mice). Scale bar, 100 µm. i, Schematic representation of the somBiPOLES construct. j-k, Individual data from the photoexcitation (j) and photoinhibition (k) of preBötC^OT-R neurons bilaterally in adult anesthetized mice (n = 8 Cre⁺ mice for photoexcitation and n = 10 Cre⁺ mice for photoinhibition with no treatment, n = 6 Cre⁺ mice for photoexcitation/photoinhibition after atropine, n = 4 Cre⁻ mice). Repeated-measures one-way ANOVA, Tukey’s multiple comparison. The pre-photomodulation and photomodulation data were used to calculate the delta changes for each measured parameter represented in Fig. 4c and d. freq, frequency; ampl, amplitude. l, Number of butyrylcholinesterase⁺ (BCHE + ) and calbindin⁺ (Calb + ) nA^Ccardiac neurons counted bilaterally (n = 4 mice, data presented as mean ± s.e.m.) with viral-mediated expression of tdTomato and synaptophysin-eGFP in preBötC^OT-R neurons (Fig. 4e-g). Detailed statistics are presented in Supplementary Table 1. Source data

Extended Data Fig. 8. OT-R⁺ neurons outside of the preBötC do not discharge in phase with preBötC inspiratory bursts, the neurochemical phenotype of preBötC^OT-R neurons is similar in P0 mice compared to adults, and characterization of the excitatory post-synaptic currents in nA neurons during inspiratory bursts.

a, All OT-R⁺ neurons recorded outside of the preBötC (V_m, whole-cell current-clamp recordings) showed activity patterns that were not in phase with the preBötC inspiratory-phase discharging population activity (n = 6 neurons, 3 tonic and 3 silent). b-c, In P0 triple transgenic mice, preBötC^OT-R neurons show the same predominantly GABAergic (b) or glycinergic (c) phenotype as in adult mice (Extended Data Fig. 7e-g; arrows indicate OT-R and GAD67 or GlyT2 co-labelling). Scale bars: full images, 100 µm; insets, 50 µm. d, Same experiment as in Fig. 5k on a rhythmic preBötC slice from a neonatal wild-type mouse, here showing successive effects induced by the application of TGOT (0.5 µM), bicuculline (GABA_A receptor antagonist, 10 µM), strychnine (glycine receptors antagonist, 5 µM), and CNQX (AMPA/kainate receptor antagonist, 20 µM), on preBötC population activity (⎰preBötC, integrated extracellular recordings) and on the currents recorded from a respiratory-modulated nA neuron that showed volleys of excitatory currents during inspiratory bursts following strychnine application (I_m, whole-cell voltage-clamp recordings; V_h, -40 mV holding voltage). These excitatory currents were blocked by CNQX application, showing that they are induced by glutamatergic transmission (n = 3 mice).

Extended Data Fig. 9. Vehicle injections in the preBötC of adult anesthetized rats and in situ Working Heart-Brainstem Preparations of juvenile rats, maps of preBötC injections, and functional characterization of CVBA recordings.

a-b, Bilateral injection of vehicle in the preBötC of adult anesthetized Wistar rats had no effect on HR and inspiratory activity (IntC EMG). Absolute values are represented for the conditions before and after vehicle injection (n = 10). Paired two-sided t-test for respiratory frequency, respiratory amplitude and mHR, Wilcoxon two-sided matched-pairs signed rank test for RespHRV. c, Schematic coronal maps showing the distribution of all TGOT and vehicle injection spots in adult anesthetized Wistar rats. d-e, Effects of vehicle injection in the preBötC of in situ Working Heart-Brainstem Preparations (WHBP) of juvenile Wistar rats (P21-30) (HR, n = 8; PNA frequency, n = 9; PNA amplitude, n = 8; VNA, n = 9; tSNA, n = 8; mPP, n = 8). Absolute values are represented for the conditions before and after vehicle injection (e). Paired two-sided t-test or Wilcoxon two-sided matched-pairs signed rank test. f, Effects of bilateral TGOT vs. vehicle injection in the preBötC of in situ WHBP of juvenile Wistar rats on the two components of tSNA, the phasic burst of respiratory modulation of tSNA (RespSNA) and the tonic SNA (n = 8 TGOT and n = 8 vehicle). Before vs. after injection conditions, paired two-sidedt-test. Comparison of the relative effects of TGOT and vehicle (Δ in %), unpaired two-sided t-test for RespSNA, Mann-Whitney two-sided test for tonic SNA. g, Peripheral chemoreflex (potassium cyanide (KCN) injection) and baroreflex (increased perfusion flow rate) challenges were used to confirm functionally the validity of the CVBA recordings (data shown from the same preparation). h, Schematic coronal maps showing the distribution of all TGOT and vehicle injection spots in the in situ WHBP of juvenile Wistar rats. Detailed statistics are presented in Supplementary Table 1. Source data

Extended Data Fig. 10. The DREADD agonist Compound 21 is mostly inert on cardiovascular and respiratory functions in OT::Cre;R26-LSL-hM4Di-DREADD Cre⁻ mice, except for mHR.

a-c, Effects induced by the intraperitoneal injection of the hM4Di agonist Compound 21 (C21) or vehicle, before, during and after a restraint stress test in a ventilated cylinder, in Cre⁻ mice from the OT::Cre;R26-LSL-hM4Di-DREADD transgenic mouse line (n = 5, three trials averaged for each condition, C21 and vehicle, in each mouse; C21 or vehicle trials spaced by at least 24 h). There were no off-target effects of C21 in Cre⁻ mice on RespHRV amplitude, mBP, respiratory frequency and respiratory amplitude during the pre-stress condition (b), nor on the effect of stress and during the recovery from stress (c), but mHR was increased during the pre-stress condition (b). Pre-stress conditions, C21 vs. vehicle, paired two-sided t-test (b). Stress and post-stress Δ (in % or bpm) effects compared to the pre-stress condition, C21 vs. vehicle, repeated-measures two-way ANOVA, Sidak’s multiple comparison (c). Detailed statistics are presented in Supplementary Table 1. Source data