The lung is a major barrier surface that interfaces with the environment and is often prone to infection. A highly coordinated immune response protects the respiratory tract from pathogens and other external insults. The role of the nervous system in regulating pulmonary host defense is not well defined. Pulmonary infections and lethal pneumonia are major public-health problems frequently causing death in children and immunocompromised and elderly people1. S. aureus is a Gram-positive human bacterial pathogen that is the leading cause of hospital-acquired infections, particularly respiratory-tract infections and ventilator-associated pneumonia1,2,3,4. The increased prevalence of multidrug-resistant bacteria including methicillin-resistant S. aureus (MRSA) strains necessitates nonantibiotic approaches to treatment. Targeting neuroimmunological signaling may be a novel approach to boost host immunity against lung pathogens.

The trachea, bronchi, and airways are innervated by peripheral sensory afferents originating from vagal and spinal sensory neurons, whose cell bodies reside within the vagal ganglia (VG) and dorsal root ganglia (DRG), respectively5,6,7. Nociceptor neurons are the subset of these neurons that respond to noxious stimuli including heat, protons, ATP, mechanical injury, inflammation, and chemical irritants8. Upon activation, nociceptors induce pain, coughing, and bronchoconstriction5,8,9,10. Recent work has shown that nociceptors cross-talk with immune cells in the respiratory tract, thereby driving allergic responses and bronchoconstriction in mouse models of asthma5,11,12. Here, we investigated a previously unexplored role of sensory neurons in pulmonary host defenses against bacterial invasion and lethal pneumonia.


TRPV1+ neurons mediate survival and bacterial clearance in pneumonia

We hypothesized that lung-innervating nociceptors are poised to detect bacterial invasion and to coordinate pulmonary immunity. The Transient receptor potential vanilloid 1 (TRPV1) ion channel responds to capsaicin, protons, and heat stimuli8,13. TRPV1 is expressed by many C fibers, including nociceptors that mediate thermal nociception and inflammatory hyperalgesia14,15,16. TRPV1+ neurons have been found to drive allergic airway hypersensitivity5.

We first used a genetic approach to determine the role of TRPV1+ neurons in host defense5,16. Trpv1-Dtr mice express the human diphtheria-toxin receptor (DTR) under control of mouse TRPV1 regulatory sequences16. Mouse cells are normally resistant to diphtheria toxin (DT)-induced apoptosis but are rendered susceptible by expression of DTR. We performed daily injections of DT into 5- to 7-week old Trpv1-Dtr mice to selectively ablate TRPV1+ neurons5,16. DT treatment, compared with PBS treatment, significantly ablated TRPV1+ neurons in both the DRG and VG in Trpv1-Dtr mice (Supplementary Fig. 1). CGRP is expressed by many peptidergic C-fiber nociceptors16,17. There were significantly fewer CGRP+ neurons in DT-treated Trpv1-Dtr mice than in PBS-treated controls (Supplementary Fig. 1). In contrast, the proportion of NF-200+ neurons, which include A fibers, was higher in the DT-treated Trpv1-Dtr mice. In DT-treated compared with PBS-treated Trpv1-Dtr mice, we also observed a loss of CGRP+ nerves around the airways (Supplementary Fig. 2) and decreased noxious-heat responses in hot-plate and tail-flick assays (Supplementary Fig. 3).

Next, we asked whether TRPV1+ neurons might affect pulmonary host defenses. Trpv1-Dtr mice recovered 7 d after DT or PBS treatment and were subsequently intratracheally inoculated with a lethal dose of the MRSA strain USA300 (1.3 × 108 to 1.4 × 108 colony-forming units (CFU); Fig. 1a). Trpv1-Dtr mice treated with DT, compared with those treated with PBS, showed significantly longer survival and better maintenance of core body temperature after MRSA pneumonia (Fig. 1b). DT-treated Trpv1-Dtr mice, compared with PBS-treated controls, also exhibited tenfold-lower bacterial burdens recovered from lungs at 12 h postinfection (Fig. 1c).

Figure 1: TRPV1 neurons regulate survival and the outcome of lethal S. aureus pneumonia.
figure 1

(a) For genetic ablation of TRPV1+ neurons, Trpv1-Dtr mice 5 to 7 weeks of age were treated with DT (200 ng/mouse intraperitoneally (i.p.)) daily for 21 d. Mice were rested 7 d before intratracheal inoculation with S. aureus USA300 (1.3 × 108 to 1.4 × 108 CFU/mouse). (b) Left, survival curves of PBS-treated Trpv1-Dtr mice (n = 11) and DT-treated Trpv1-Dtr mice (n = 13). Log-rank (Mantel–Cox) test (P = 0.01). Right, measurements of core body temperature (temp.) over time in PBS-treated (n = 7) and DT-treated Trpv1-Dtr mice (n = 8). Two-way repeated (RM) analysis of variance (ANOVA) with Bonferroni post tests (***P = 0.001; *P = 0.014). (c) Lung bacterial burdens 12 h after infection in PBS-treated (n = 13) and DT-treated Trpv1-Dtr mice (n = 12). Two-tailed unpaired t test (P = 0.0042). (d) Resiniferatoxin (RTX)-mediated chemical ablation of TRPV1+ neurons. WT mice 4 weeks of age were injected subcutaneously daily with three escalating doses of RTX or vehicle. Mice were rested for 4 weeks before intratracheal inoculation with S. aureus USA300 (0.8 × 108 to 1 × 108 CFU/mouse). (e) Left, survival curves of vehicle-treated (n = 20) and RTX-treated mice (n = 18). Log-rank test (P < 0.0001). Right, core-body-temperature measurements in vehicle-treated (n = 5) and RTX-treated mice (n = 5). Two-way RM ANOVA with Bonferroni post tests (****P < 0.001). (f) Lung bacterial-load recovery 12 h after S. aureus infection in vehicle-treated (n = 13) and RTX-treated mice (n = 13). Two-tailed unpaired t tests (P = 0.0035). Data were pooled from two (b) or three (c,e,f) independent experiments. Data are shown as mean ± s.e.m. (b,e, core body temperature) and mean (c,f).

We used Resiniferatoxin (RTX) as a second strategy to target TRPV1+ neurons. RTX is a high-affinity TRPV1 ligand that can be used to chemically denervate and ablate nociceptors15,18. Mice were subcutaneously treated with RTX at 4 weeks of age for consecutive days with escalating doses (30, 70, and 100 μg/kg) according to established protocols19,20. RTX-treated mice, compared with vehicle-treated mice, showed increased latency to noxious heat in hot-plate and tail-flick assays, and loss of TRPV1+ and CGRP+ neurons in the DRG and VG (Supplementary Figs. 3 and 4). At 4 weeks after RTX injection, mice were intratracheally inoculated with MRSA (0.8 × 108 to 1 × 108 CFU; Fig. 1d). Whereas most vehicle-treated mice succumbed to pneumonia (80% mortality), most RTX-treated mice survived (Fig. 1e). RTX-treated mice, compared with vehicle-treated mice, showed improved maintenance of core body temperature (Fig. 1e) and less lung bacterial burden (Fig. 1f). Trpv1-Dtr-mediated ablation and RTX treatment enhanced protection in mice infected with a sublethal dose of S. aureus (2 × 107 to 4 × 107 CFU), as measured by bacterial-load recovery (Supplementary Fig. 5). Because nociceptors may regulate the peripheral resistance of the cardiovascular and pulmonary systems to infection, we measured vital signs. However, the oxygen saturation, heart rate, perfusion, and respiratory rates did not differ between RTX-treated and vehicle-treated mice at steady state; the respiratory rates also did not differ postinfection (Supplementary Fig. 6).

Next, we asked whether nociceptors might modulate host defense against bacterial pathogens other than S. aureus. RTX-treated and vehicle-treated mice were infected with lethal doses of Streptococcus pneumoniae, Klebsiella pneumoniae, or Pseudomonas aeruginosa. The RTX-treated mice and vehicle-treated mice showed similar decreases in core body temperature after infection with the three pathogens (Supplementary Fig. 7). Nociceptor deficiency showed a modest but nonsignificant protective effect (P = 0.13) in survival during S. pneumoniae infection (Supplementary Fig. 7). Nociceptor deficiency did not affect death caused by K. pneumoniae or P. aeruginosa pneumonia (Supplementary Fig. 7).

Nav1.8 is a voltage-gated sodium channel expressed by a large subset of nociceptors that overlap with but are distinct from TRPV1+ neurons16,21. Nav1.8-cre+/− mice were bred with diphtheria toxin A (DTA) reporter mice to generate animals deficient in Nav1.8-lineage neurons (Nav1.8-Cre+/−; Dta)21. After MRSA infection, we observed a trend toward higher survival (P = 0.09) and lower bacterial burden (P = 0.07) in Nav1.8-Cre+/−; Dta mice than in control littermates (Supplementary Fig. 8). However, the beneficial effects of Nav1.8-lineage neuron ablation (Supplementary Fig. 8) were considerably smaller than those observed for TRPV1 neuron ablation (Fig. 1).

TRPV1 ion channel does not mediate pulmonary host defense

We next determined whether the TRPV1 ion channel itself was involved in host defense. Trpv1−/− mice have previously been found to have exaggerated physiologic responses in a model of polymicrobial sepsis22. After S. aureus lung infection, we did not observe significant differences in survival, core-body-temperature measurements, or lung bacterial burdens in Trpv1−/− mice compared with Trpv1+/− or Trpv1+/+ control littermates (Supplementary Fig. 9). The postinfection induction of cytokines (IL-17A, IL-6, and IL-23) in lung lysates of Trpv1−/− mice was similar to that in control littermates (Supplementary Fig. 10). We also examined the role of TRPA1, which mediates airway inflammation in a mouse model of asthma11. Trpa1−/− mice, compared with Trpa1+/+ littermates, did not show differences in bacterial burdens after lethal or sublethal S. aureus infection (Supplementary Fig. 11).

TRPV1 and Nav1.8 neurons regulate bacterial dissemination

We next determined whether nociceptors mediated the spread of bacterial pathogens from the lung to extrapulmonary sites. DT-treated Trpv1-Dtr mice showed higher numbers of bacteria in the blood (P = 0.01) after lethal MRSA infection than did controls (Supplementary Fig. 12). RTX-treated mice also showed greater blood dissemination than did vehicle-treated controls (Supplementary Fig. 12). At a sublethal dose of infection, both Trpv1-Dtr ablation and RTX treatment increased MRSA dissemination to the blood and spleen (Supplementary Fig. 13). In Nav1.8-Cre+/−; Dta mice, compared with control littermates, we also observed significantly greater bacterial dissemination to the blood, which was accompanied by greater spleen size (Supplementary Fig. 13). We investigated whether nociceptor ablation affected lung-barrier permeability. RTX-treated mice, compared with vehicle-treated mice, showed greater leakage of fluorescein isothiocyanate (FITC)–dextran to the blood after intratracheal inoculation, thus suggesting a role of nociceptors in maintaining barrier integrity (Supplementary Fig. 14).

TRPV1 neurons regulate lung inflammation and cytokine induction

We performed histological analysis of lungs at different time points to analyze pulmonary inflammation after S. aureus infection. RTX-treated mice, compared with vehicle-treated mice, showed greater immune-cell influx in the lungs at 12 h and 24 h postinfection, as determined by H&E staining (Fig. 2a). We hypothesized that early increases in immune cells might correlate with improved bacterial clearance in RTX-treated mice. Brown and Brenn staining showed many Gram-positive bacterial colonies in vehicle-treated lungs at 12 h and 24 h postinfection (Fig. 2b). In contrast, RTX-treated mice showed few bacterial colonies (Fig. 2b). We next determined whether nociceptors regulated proinflammatory-cytokine production. At an early time point of infection (6 h), RTX-treated mice showed higher induction of total inflammatory-protein levels in the bronchoalveolar lavage fluid (BALF) (Fig. 2c), as well as levels of the cytokines TNF-α and the chemokine CXCL-1 (Fig. 2d). By 12 h postinfection, TNF-α, IL-6, and CXCL-1 levels were lower in RTX-treated mice (Fig. 2d), as were levels of the cytokines IL-1β and MCP-1 (Supplementary Fig. 15). These data indicated that TRPV1+ neuron ablation leads to faster induction and resolution of cytokine levels during infection.

Figure 2: TRPV1 neurons suppress lung inflammation and cytokine induction during S. aureus infection.
figure 2

(a) H&E-stained lung sections from vehicle-treated and RTX-treated mice at 12 h or 24 h after S. aureus infection. Representative images were chosen from 15 lung lobes imaged from 3 mice in each group (vehicle (veh) and RTX). Scale bars: black, 500 μm; red, 100 μm. (b) Brown and Brenn Gram-stained images of bacterial colonies in lung sections from vehicle-treated and RTX-treated mice, 12 h and 24 h after infection. Representative images were chosen from 15 lung lobes imaged from 3 mice in each group (vehicle and RTX). Inset, purple bacterial cocci are S. aureus colonies. Red scale bars, 50 μm. (c) Quantification of total protein levels in BALF at different time points after S. aureus infection (1 × 108 CFU/mouse). Vehicle-treated group, 0 h (n = 6), 2 h (n = 8), 6 h (n = 8), 12 h (n = 9); RTX-treated group, 0 h (n = 6), 2 h (n = 7), 6 h (n = 6), 12 h (n = 10). Two-way RM ANOVA with Bonferroni post tests (****P < 0.001). (d) Levels of IL-6, TNF-α, and CXCL-1 in BALF from mice at different time points after S. aureus infection (1 × 108 CFU/mouse). TNF-α: vehicle-treated group, 0 h (n = 6), 2 h (n = 10), 6 h (n = 10), 12 h (n = 10); RTX-treated group, 0 h (n = 6), 2 h (n = 7), 6 h (n = 6), 12 h (n = 10). IL-6: vehicle-treated group, 0 h (n = 6), 2 h (n = 10), 6 h (n = 10), 12 h (n = 10); RTX-treated group, 0 h (n = 6), 2 h (n = 7), 6 h (n = 6), 12 h (n = 10). CXCL-1: vehicle-treated group, 0 h (n = 6), 2 h (n = 8), 6 h (n = 8), 12 h (n = 10); RTX-treated group, 0 h (n = 6), 2 h (n = 7), 6 h (n = 6), 12 h (n = 10). Statistical analysis for all cytokines by two-way RM ANOVA with Bonferroni post tests (**P = 0.0013; ****P < 0.0001). Data from one experiment with multiple biological replicates are shown in a and b; data pooled from two independent experiments are shown in c and d. Data are shown as mean ± s.e.m. in c and d.

TRPV1 neurons suppress recruitment of neutrophils

We next used FACS analysis to analyze the kinetics of immune-cell influx into inflamed lungs during S. aureus infection. RTX-treated mice, compared with vehicle-treated mice, displayed greater CD11b+Ly6G+ lung neutrophil recruitment at 6 h and 12 h postinfection (Fig. 3a,b). Trpv1-Dtr-neuron-ablated mice showed greater neutrophil recruitment than did PBS-treated controls at 12 h postinfection (Fig. 3c). Because neutrophils are critical for bacterial clearance23, we hypothesized that neuronal modulation of neutrophil recruitment might play a major role in MRSA pneumonia. We depleted neutrophils in RTX-treated mice by using low-dose anti-GR1 antibody treatment, which, compared with control IgG treatment, eliminated CD11b+Ly6G+ neutrophils in infected lungs (Fig. 3d). Anti-GR1 also decreased CD11b+Ly6Chi monocytes but did not affect CD11b+Ly6Clo monocytes (Supplementary Fig. 16). Neutrophil depletion in RTX-treated mice significantly increased their susceptibility to MRSA pneumonia: whereas 100% of anti-GR1-treated RTX mice succumbed to infection, 0% of the control IgG-treated RTX mice died from infection (Fig. 3e). These experimental results were confirmed in an independent cohort of mice (Supplementary Fig. 17). Neutrophils were also required for baseline protection against MRSA pneumonia (Fig. 3e). These data suggested that RTX-mediated enhancement of lung immunity requires neutrophils.

Figure 3: TRPV1 neurons suppress recruitment of Ly6G+ neutrophils essential for host defense against lethal pneumonia.
figure 3

(a) Representative FACS plots of neutrophils (CD11b+Ly6G+, out of CD45+ cells) in vehicle-treated and RTX-treated mice in BALF collected at 6 h postinfection with S. aureus (0.8 × 108 CFU/mouse). Representative FACS plots were chosen from 4 mice in each group (vehicle and RTX). (b) Time course of CD11b+Ly6G+ neutrophil recruitment in the BALF of RTX-treated mice compared with vehicle-treated mice after lethal S. aureus infection. n = 4 mice in each group (vehicle and RTX) for 2 h, 6 h, and 12 h data sets. Two-way ANOVA with Bonferroni post tests (***P = 0.001). (c) Representative FACS plots (left) and quantification data (right), showing neutrophils (Gr1hiCD11b+ cells) in the lung homogenates of DT-treated (n = 4) and PBS-treated Trpv1-Dtr mice (n = 4) 12 h after S. aureus infection (1.3 × 108 CFU/mouse). Statistical analysis by two-tailed t test. (d) FACS plots showing anti-GR1 antibody–mediated ablation of lung neutrophils in vehicle (n = 3) and RTX-treated (n = 3) mice 12 h after S. aureus infection (0.8 × 108 CFU/mouse). (e) Left, Kaplan–Meier survival curves after S. aureus lung infection (1.2 × 108 to 1.35 × 108 CFU/mouse) of vehicle + rat IgG mice (n = 10), vehicle + anti-GR1 mice (n = 10), RTX + rat IgG mice (n = 9), and RTX + anti-GR1 mice (n = 10). Statistical analysis by log-rank test, P < 0.0001 (RTX + rat IgG versus RTX + anti-GR1), P < 0.0001 (vehicle + rat IgG versus vehicle + anti-GR1), and P = 0.002 (vehicle + rat IgG versus RTX + rat IgG). Right, core body temperature, measured after S. aureus infection with or without neutrophil depletion. Vehicle + rat IgG (n = 5); vehicle + anti-GR1 (n = 5); RTX + rat IgG (n = 4); RTX + anti-GR1 (n = 5). Statistical analysis by two-way RM ANOVA with Bonferroni post tests (P = 0.008, RTX + rat IgG versus RTX + anti-GR1). Data from two independent experiments (b,e) or one experiment with multiple biological replicates (c) are shown. Data are shown as mean ± s.e.m. in b, c (neutrophil proportions), and e (core body temperature).

TRPV1 neurons regulate pulmonary neutrophil surveillance

Within lungs, neutrophils perform endothelial and parenchymal surveillance for pathogens24,25. Using intravital microscopy, we analyzed the subpleural vascular bed to assess whether neutrophil kinetics and patrolling of tissues were regulated by nociceptors. Compared with control mice, RTX-treated mice recruited significantly more neutrophils to pulmonary capillaries during early S. aureus pneumonia (Fig. 4a,b and Supplementary Video 1). We observed less GFP–S. aureus in the lungs of nociceptor-depleted animals, in agreement with previous CFU data. Dynamic neutrophil behavioral phenotypes were analyzed for tethering, crawling, and firm adhesion within the vasculature. Tethering is a rapid, transient neutrophil interaction with the vessel wall, which requires limited cellular activation. Adhesion indicates a more advanced state of activation mediated by integrins before tissue emigration. Crawling is a complex intravascular behavior that requires upregulation of β2-integrin, which occurs rapidly during lung host defense24,25. Neutrophils behaved differently during host defense in control versus RTX-treated mice. In RTX-treated mice, a significant proportion of neutrophils demonstrated vascular crawling (Fig. 4c), and tracking individual pulmonary vascular neutrophils revealed significantly greater crawling distances (Fig. 4d,e), a phenotype consistent with cellular activation and host defense against bacterial pathogens. Therefore, live imaging supports the contention that enhanced neutrophil function in the absence of nociceptors aids in the eradication of bacterial pneumonia.

Figure 4: Neutrophil dynamics are altered in TRPV1-neuron-ablated mice.
figure 4

(a) In vivo imaging of neutrophils in the lungs after lethal GFP–S. aureus USA300 lung infection in live vehicle-treated and RTX-treated mice at 4 h postinfection. (b) Total neutrophils, determined per field of view. (c) Neutrophil behavior, phenotyped and quantified as tethering, adhesion, or crawling. NS, nonsignificant. (d) Crawling tracks, displayed for individual neutrophils. (e) Distances for individual neutrophils. n = 3 individual experiments performed for vehicle-treated mice and 4 individual experiments for RTX-treated mice. For statistical analyses, values for each parameter (b,c,e) were averaged for each animal, and two-tailed unpaired t tests were performed, comparing vehicle-treated and RTX-treated mouse groups. Data are shown as mean ± s.e.m. in b, c, and e.

Nociceptor neurons regulate lung-resident γδ T cells

We next asked whether TRPV1+ neurons might alter the lung-resident immune-cell populations in naive mice, thus setting the stage for subsequent inflammatory responses. We first examined immunological-transcriptome data sets at the Immunological Genome Project ( and found that Trpv1 expression was absent across immune-cell types (Supplementary Fig. 18). Analysis of a second transcriptional data set26 showed that Trpv1 was absent in immune cells but was highly expressed in DRG (Supplementary Fig. 18). We next purified CD4+ T cells, B cells, neutrophils, and γδ T cells from mouse lungs, and performed quantitative PCR analysis for Trpv1 compared with that in sensory ganglia. Whereas Trpv1 was highly expressed in the VG and DRG, it was undetectable in all lung immune cells analyzed (Supplementary Figs. 19 and 20). These data indicated that Trpv1-Dtr or RTX-mediated ablation should specifically target nociceptors but should not have direct effects on immune cells. We examined spleens of RTX-treated and vehicle-treated mice and did not observe differences in the populations of B220+ B cells, NK1.1+ cells, CD4+ T cells, CD8+ T cells, or γδ T cells; moreover, Trpv1-Dtr-neuron-ablated mice showed similar results (Supplementary Fig. 21).

We next examined whether lung-resident immune-cell types differed in nociceptor-ablated mice at steady state. CD11b+ dendritic cells (CD11b+SiglecFCD24+CD103F4/80MHCII+), CD103+ dendritic cells (CD11bSiglecFCD103+MHCII+), alveolar macrophages (SiglecF+CD11c+CD64+F4/80+), and interstitial macrophages (CD11b+CD24CD64+F4/80+) did not differ between RTX-treated and vehicle-treated mice (Supplementary Fig. 22 and Fig. 5a). B cells, natural killer cells, and CD8+ T cells also did not differ; however, CD4+ T cells were slightly higher in RTX-treated mice than in controls (Supplementary Fig. 23 and Fig. 5b).

Figure 5: TRPV1 neurons regulate lung γδ T cells, which mediate host protection against S. aureus pneumonia.
figure 5

(ac) To detect differences at steady state, lung tissues from nociceptor-depleted (RTX treated, n = 8) and nondepleted mice (vehicle treated, n = 8) were analyzed by flow cytometry for myeloid immune cells (a), lymphoid immune cells (b), and γδ T cell populations (c). Statistical analysis in ac by two-tailed unpaired t tests. Alv. mac, alveolar macrophages; int. mac, interstitial macrophages; NK, natural killer. (d) Representative FACS plots showing γδ T cells in WT or Tcrd−/− mice (RTX or vehicle treated) either uninfected (uninf.) or 12 h postinfection (inf.) with S. aureus (1 × 108 CFU/mouse). Vehicle-treated groups: WT uninfected (n = 8), WT infected (n = 4), Tcrd−/− infected (n = 3); RTX-treated groups: WT uninfected (n = 8), WT infected (n = 4), Tcrd−/− infected (n = 3). (e) Left, survival curves for vehicle-treated WT mice (n = 7), vehicle-treated Tcrd−/− mice (n = 6), RTX-treated WT mice (n = 5), and RTX-treated Tcrd−/− mice (n = 6) after lethal S. aureus infection (1 × 108 CFU/mouse). Statistical analysis by log-rank test (P = 0.01, RTX-treated WT versus RTX-treated Tcrd−/−; P = 0.08, vehicle-treated WT versus vehicle-treated Tcrd−/−). Right, core-body-temperature measurements in vehicle-treated WT mice (n = 5), vehicle-treated Tcrd−/− mice (n = 5), RTX-treated WT mice (n = 5), and RTX-treated Tcrd−/− mice (n = 4). Two-way RM ANOVA with Bonferroni post tests (P < 0.0001, RTX-treated WT versus RTX-treated Tcrd−/−). (f) Representative FACS plots of CD11bSiglecF+ alveolar macrophages after intratracheal administration of CLL or PBS-L in vehicle-treated and RTX-treated mice 12 h postinfection. Vehicle-treated groups: uninfected (n = 8), PBS-L infected (n = 3), CLL infected (n = 3); RTX-treated groups: uninfected (n = 8), PBS-L infected (n = 4), CLL infected (n = 3). (g) Left, survival curves after S. aureus lung infection (1 × 108 CFU/mouse) with or without alveolar macrophage depletion. Vehicle + PBS-L (n = 10), vehicle + CLL (n = 10), RTX + CLL (n = 10), RTX + PBS-L (n = 8). Log-rank test (P = 0.37, RTX + PBS-L versus RTX + CLL; P = 0.22, vehicle + PBS-L versus vehicle + CLL). Right, core-body-temperature measurements. Vehicle + PBS-L (n = 4), vehicle + CLL, (n = 4), RTX + CLL (n = 4), and RTX + PBS-L (n = 4). Two-way RM ANOVA with Bonferroni post tests. One experiment with multiple biological replicates was performed for e, and two independent experiments were performed for ad and fg. Data shown in ac, e (core body temperature), and g (core body temperature) are mean ± s.e.m.

We observed higher absolute numbers of lung-resident γδ T cells, in contrast to most other immune-cell types, in RTX-treated mice compared with vehicle-treated mice (Fig. 5c). Further subset analysis revealed that this increase was specific to Vγ1+ cells and Vγ1Vγ2 cells, but not Vγ2+ cells (Fig. 5c and Supplementary Figs. 24 and 25). We found a similar increase in γδ T cells in Nav1.8-Cre+/−; Dta mice compared with control littermates (Supplementary Fig. 24). γδ T cells reside within epithelial layers of the lungs, skin, and gut, where they act as first responders to infection27. We next used γδ T cell–deficient Tcrd−/− mice to investigate the role of these cells in neuroimmunological suppression. Wild-type (WT) or Tcrd−/− mice were treated with RTX to ablate TRPV1+ neurons, and then S. aureus pneumonia was induced (Fig. 5d,e). The absence of γδ T cells was confirmed in Tcrd−/− mice through flow cytometry (Fig. 5d). Tcrd deficiency led to a loss of protection against MRSA infection and abrogated the survival enhancement due to RTX treatment (Fig. 5e). This reversal of protective immunity correlated with an imbalance in core body temperature (Fig. 5e) and greater bacterial burdens in BALF isolated from RTX-treated Tcrd−/− mice compared with RTX-treated WT mice (Supplementary Fig. 26). Tcrd−/− also showed defective baseline immunity against MRSA pneumonia (Fig. 5e). Whereas IL-6 levels were unaffected by Tcrd deficiency, levels of IL17A, a cytokine mediating protection against MRSA28, were significantly lower in Tcrd−/− mice (Supplementary Fig. 26). Neutrophil recruitment did not differ in the lungs of RTX-treated Tcrd−/− mice and RTX-treated WT mice (Supplementary Fig. 26), thus suggesting that γδ T cells and neutrophils are separately regulated.

We next determined whether alveolar macrophages mediated neuronal regulation of the host defense (Fig. 5g). Mice were intratracheally instilled with clodronate-laden liposomes (CLL) to kill alveolar macrophages through phagocytosis-dependent apoptosis. PBS-encapsulated liposomes (PBS-L) were used as a control treatment. CLL treatment specifically eliminated alveolar macrophages (SiglecF+CD11c+CD64+F4/80+) but not interstitial macrophages or dendritic cells (Fig. 5f and Supplementary Fig. 27). Alveolar-macrophage depletion did not alter the greater survival or core-body-temperature maintenance of RTX-treated compared with vehicle-treated mice (Fig. 5g). Together, our results suggested that the RTX-treatment-mediated enhancement of MRSA immunity requires both γδ T cells and neutrophils but not alveolar macrophages.

Ablation of vagal TRPV1 neurons improves host defense

The vagus nerve provides the major source of sensory innervation of the lung. The cell bodies of vagal afferents reside in fused ganglia at the base of the skull, controlling coughing, breathing, and bronchoconstriction5,6. Our previous experimental approaches targeted all TRPV1+ cells, including both DRG and VG neurons. We hypothesized that vagal TRPV1+ neurons might include the subset regulating neuroimmunological suppression. To specifically target these neurons, we performed bilateral intraganglionic DT injections into the VG in 5- to 9-week-old Trpv1-Dtr mice. Immunostaining showed that vagal but not DRG TRPV1+ neurons were specifically ablated (Fig. 6a and Supplementary Fig. 28). Vagal TRPV1+ neuron ablation did not alter the heart rate, oxygen saturation, perfusion, or respiratory rate (Supplementary Fig. 6). Mice were rested 2 weeks after intraganglionic injections, then infected with a lethal dose of MRSA. We observed a striking survival benefit in vagal DT-injected Trpv1-Dtr mice compared with vagal PBS-injected mice, and better maintenance of core body temperature (Fig. 6b). All mice from the PBS-injected group died within 72 h after infection, whereas a 90% survival rate was observed among mice lacking vagal TRPV1 neurons (Fig. 6b). The greater survival in vagal DT-treated mice correlated with higher neutrophil recruitment and lower lung bacterial burdens (Supplementary Fig. 29).

Figure 6: Vagal TRPV1 neurons and the neuropeptide CGRP regulate S. aureus pneumonia.
figure 6

(a) DT or PBS alone was injected bilaterally into VG in Trpv1-Dtr mice. Quantification of proportions of TRPV1+ and CGRP+ neurons in VG and DRG (T1-T9) of vagal DT-injected mice (n = 3) and PBS-injected littermate-control (n = 3) mice. Statistical analysis by two-tailed unpaired t tests. (b) Left, survival curves of PBS VG-injected Trpv1-Dtr mice (n = 9) compared with DT vagal-injected Trpv1-Dtr mice (n = 8) after lethal S. aureus lung infection (1.3 × 108 to 1.4 × 108 CFU/mouse). Log-rank test, P = 0.0003. Right, core body temperature in PBS vagal-injected Trpv1-Dtr mice (n = 4) compared with DT vagal-injected Trpv1-Dtr mice (n = 4) after infection. Two-way RM ANOVA with Bonferroni post tests (****P < 0.0001). (c) CGRP levels in the BALF in uninfected mice or mice 12 h after infection with WT USA300 S. aureus (0.8 × 108 CFU/mouse) or Δagr USA300 S. aureus (0.8 × 108 CFU/mouse); uninfected (n = 5), WT infected (n = 5), and Δagr infected (n = 5). Statistical analysis by one-way ANOVA with Bonferroni post tests. (d) CGRP levels in the BALF in RTX-treated mice (n = 5) and vehicle-treated mice (n = 5); or CGRP levels in vagal DT-treated (n = 5) or PBS-treated Trpv1-Dtr mice (n = 4) 12 h after WT S. aureus infection (1.1 × 108 to 1.4 × 108 CFU/mouse). Statistical analysis by two-tailed unpaired t test. (e) Production levels of TNF-α, IL-6, and CXCL-1 by whole lung cell cultures after infection with S. aureus at a multiplicity of infection (MOI) of 2 at 2 h, 6 h, 12 h, and 20 h postinfection with or without CGRPα (100 nM). Statistical analysis by two-way ANOVA with Bonferroni post tests. (**P < 0.01; ***P < 0.001; ****P < 0.0001). (f) The CGRP-receptor antagonist CGRP8–37 was administered systemically (i.p.) at 800 ng (256 pmol) per dose dissolved in PBS, at −24 h, −2 h, 12 h, 24 h, 36 h and 48 h relative to intratracheal S. aureus infection (0 h). Control mice received i.p. PBS injections at the same time points. Left, survival curves for mice (PBS treated, n = 8; CGRP8–37 treated, n = 8) after S. aureus lung infection. Log-rank test, P = 0.04. Right, core-body-temperature measurements (PBS treated, n = 3; CGRP8–37 treated, n = 5) after S. aureus infection. Two-way RM ANOVA with Bonferroni post tests (**P = 0.004; ***P = 0.0002). (g) CGRP8–37 was administered i.p. at 7.5 μg (2.4 nmol) per dose in PBS at 4-h, 16-h and 24-h time points relative to intratracheal S. aureus infection (0 h). Control mice received i.p. PBS injections at the same time points. Left, survival curves for PBS-treated (n = 20) and CGRP8–37-treated (n = 19) mice. Log-rank test, P = 0.03. Right, core body temperature of PBS-treated (n = 12) and CGRP8–37-treated (n = 11) mice. Two-way RM ANOVA with Bonferroni post tests (**P = 0.002 (16 h); **P = 0.004 (24 h)). One experiment was performed with multiple biological replicates for a, d, and f; two independent experiments were performed for b, c, and e; three independent experiments were performed for g.

Nociceptive neuropeptide CGRP modulates lung antimicrobial defense

Nociceptor neurons actively communicate with the immune system through their release of neuropeptides stored within peripheral nerve terminals29. The nociceptive neuropeptide CGRP inhibits TNF-α production in macrophages and suppresses lymph node hypertrophy in skin bacterial infection29,30. We hypothesized that CGRP might mediate neuroimmunological signaling during lethal bacterial pneumonia. We found that CGRP levels significantly increased in the BALF after S. aureus infection (Fig. 6c). Infection with an S. aureus strain mutant for agr, a key bicomponent quorum-sensing regulator of virulence-factor expression31, did not induce CGRP release into the BALF (Fig. 6c). S. aureus also directly induced cultured neuronal release of CGRP in vitro, in a manner dependent on agr (Supplementary Fig. 30). TRPV1+ cells mediated CGRP release in the lungs, because CGRP levels were significantly lower in the BALF in RTX-treated and vagal DT-treated Trpv1-Dtr mice than in control-infected mice at 12 h postinfection (Fig. 6d). CGRP levels were also significantly lower in the BALF at steady state in nociceptor-depleted mice than in control nondepleted mice (Supplementary Fig. 31).

We next asked whether CGRP might play a role in MRSA pneumonia. Quantitative PCR analysis showed that lung γδ T cells and neutrophils expressed Ramp1 and Calcrl, which encode the cognate CGRP receptor (Supplementary Fig. 32). We found that CGRP treatment inhibited lung-cell production of TNF-α and CXCL1, but not IL-6, in response to MRSA infection (Fig. 6e). Increasing concentrations of CGRP also inhibited intracellular killing of S. aureus by mouse neutrophils (Supplementary Fig. 33). To explore the involvement of CGRP in host defense, we treated mice with the competitive CGRP peptide antagonist CGRP8–37 at time points before and after S. aureus lung infection (Fig. 6f). This treatment, compared with vehicle treatment, improved survival and core-body-temperature maintenance (Fig. 6f). We next found that CGRP8–37 administered at time points after S. aureus infection also significantly improved survival and core-body-temperature maintenance (Fig. 6g). These data showed that nociceptors mediate CGRP release during lung infections and that postinfection blockade of CGRP signaling may aid in treatment of bacterial pneumonia.


Nociceptor neurons and immune cells play key roles in protecting organisms from environmental dangers. It is potentially advantageous that interactions between these cell types coordinate host responses to pathogen invasion. We found that TRPV1+ afferents in the VG played a critical role in modulating innate immune responses against MRSA lethal pneumonia. Targeting these neurons through Trpv1-Dtr-mediated ablation (or RTX treatment) improved survival, neutrophil and γδ T cell responses, and bacterial clearance. Nav1.8-cre; Dta mice, in which an overlapping though distinct nociceptor subset is targeted, showed a milder protective phenotype. Both strategies paradoxically resulted in increased bacterial dissemination. These data suggested that differences in phenotypes (lung clearance versus barrier function) are mediated by distinct neuronal subsets. A recent study has shown that Trpv1 expression in the adult DRG is mainly restricted to CGRP+ and substance P (SP)+ C fibers14. In contrast, Nav1.8 has been found to be expressed in myelinated A fibers as well as C fibers32. Another study has shown that Nav1.8-Cre; Dta mice still possess CGRP+ neurons expressing TRPV1 (ref. 21). Therefore, future experiments using more refined genetic tools should help to distinguish the functional contributions of individual TRPV1+ and/or Nav1.8+ neuronal subsets in pulmonary immunity and barrier function.

Our work adds to recent studies showing major physiological roles for neuroimmunological interactions at peripheral barrier tissues33. In the respiratory tract, nociceptors actively cross-talk with immune cells, thereby mediating allergic airway inflammation5,11,12. Skin-innervating nociceptors drive inflammation and immunological activation in mouse models of psoriasis20 and contact dermatitis34. In the gut, sympathetic neurons regulate macrophage tissue programming at homeostasis and during Salmonella infection35.

We found that nociceptors suppressed pulmonary γδ T cell– and neutrophil-mediated host defense during MRSA lung infections. A recent study has found that nociceptors drive dendritic-cell IL-23 production and γδ T cell activation during skin invasion by the fungal pathogen Candida albicans19. The observed phenotypic difference in that study compared with our study is interesting, because differential interactions of vagal versus somatosensory sensory neurons may occur with immune-cell types at different barrier sites. Diverse γδ T cell populations seed mucosal and epithelial sites27. In the skin, epidermal γδ T cells are mostly Vγ5+ cells that mediate barrier integrity, whereas dermal γδ T cells do not express Vγ5, but 40% of them are Vγ4+ and are involved primarily in IL-17A production36. IL-17 production by γδ T cells has also been found to mediate host defense against S. aureus skin infections28,37. Heterogeneous subsets of γδ T cells are found in the respiratory tract, including Vγ1+, Vγ2+, and Vγ6+ populations38.

It is striking that vagal sensory afferents, which comprise fewer than 5,000 neurons, are able to potently regulate antimicrobial immunity. Distinct vagal afferents control physiological functions including breathing and nutrient sensation6,39. It would be interesting to ascertain how neuronal subsets differentially cross-talk with immune cells. Immune cells may utilize nerves as tracts for migration, as has been observed for dendritic-cell interactions with Nav1.8+ nociceptors in the skin20. Lung-resident immune cells may be proximal to vagal nerve afferents and may consequently able to set up local neuroimmunological responses. Recently, the neuropeptide NMU has been found to drive ILC2-mediated inflammation in the gut and lungs40,41,42. We found that nociceptors released the neuropeptide CGRP into the airways during infection and downregulated immunity. CGRP has previously been linked to vasodilation and vascular permeability43. CGRP suppresses CXCL1, an important chemokine for lung-neutrophil chemoattraction44. Furthermore, CGRP antagonism improves survival outcomes in MRSA-infected mice and is therefore a potential target for clinical application in pneumonia.

Notably, other mechanisms beside CGRP signaling could mediate nociceptor immunological signaling. Nociceptors release glutamate, ATP, and other neuropeptides including SP, neurokinin A, and VIP. They also upregulate cytokines including CCL2 (ref. 45) and CSF-1 after nerve injury46. Vagal afferents may also induce sensoriautonomic neuroimmunological reflexes including a 'cholinergic antiinflammatory reflex', which acts through vagal autonomic efferents and downregulates peripheral macrophage TNF-α production47.

The role of nociceptors in host defense may vary depending on the type of pathogenic invasion. Whereas increased neutrophil influx confers host protection against MRSA pneumonia, the same responses could lead to immunopathology in other infections. For example, influenza virus and severe acute respiratory syndrome coronavirus cause pathology through overactivation of lung inflammation48,49. In pneumonia caused by K. pneumoniae, Escherichia coli, and S. pneumoniae, bacterial dissemination is a primary cause of sepsis and mortality50,51. For MRSA-induced pneumonia, lethality is mainly mediated by damage to the lungs by secreted exotoxins (Hla and PVL) rather than bacterial dissemination3,52. These differences in bacterial pathogenesis may explain our observed differences in responses to distinct pathogens in nociceptor-ablated mice.

Our study demonstrates that nociceptors play a critical role in regulating pulmonary immunity and the outcomes of bacterial lung infections. Targeting neuroimmunological communication through CGRP or other molecular mechanisms may be an effective approach to enhance host protection against pneumonia.



All animal experiments were approved by the Harvard Medical School Institutional Animal Care and Use Committee (IACUC) or by the University of Calgary Animal Care Committee. Mice were housed in a specific-pathogen-free animal facility at Harvard Medical School or the University of Calgary. C57BL/6J, B6.Trpv1−/−, B6.Tcrd−/−, B6.Dta+/+, and B6129.Trpa1−/− mice were purchased from Jackson Laboratories. Trpv1-Dtr mice16 were provided by M. Hoon (NIH). Nav1.8-cre mice21 were originally from J. Wood (University College London). Nav1.8-cre+/− mice were bred with B6.Dta+/+ mice to generate Nav1.8-cre+/−; Dta mice and control littermates (Nav1.8-cre−/−; Dta). For Trpv1 and Trpa1 experiments, heterozygous mice were bred to produce WT, heterozygous, and knockout littermates. Age-matched 8- to 14-week-old male and female mice were used for experiments.

Bacterial strains and cultures.

The MRSA strain USA300/LAC53 was provided by M. Otto (NIH). For infection, USA300/LAC was grown overnight (O/N) at 37 °C in tryptic soy broth (TSB, Sigma) at 250 r.p.m. and was subcultured at a 1:100 dilution for 3.5 h in TSB to mid-log phase. K. pneumoniae strain 43816 serotype 2 was purchased from American Type Culture Collection and was grown O/N at 37 °C in TSB at 250 r.p.m. for infection. S. pneumoniae WU-2 strain from R. Malley (Boston Children's Hospital) was grown at 37 °C under 5% CO2 without shaking for 18 h in Todd's Hewitt broth (THB, Sigma) with 0.5% yeast extract and was subcultured at a 1:10 dilution for 8 h in fresh THB with 0.5% yeast extract to reach mid-log phase for infection. P. aeruginosa strain PA01V from G. Pier (Brigham and Women's Hospital) was grown O/N at 37 °C in TSB at 250 r.p.m. and was subcultured at a 1:100 dilution for 4 h in TSB for infection. For all strains, cultures were centrifuged at 5,000 r.p.m. for 5 min, and bacterial pellets were washed and resuspended in phosphate-buffered saline (PBS). The OD600 was measured to estimate bacterial density, and serial plating was performed on tryptic soy agar (TSA) plates to quantify CFU values. For intravital imaging, a GFP–MRSA S. aureus transgenic bacterium was used, whose construction has been previously reported25.

Bacterial lung infections.

For all bacterial infections, age-matched 8- to-14-week-old male and female mice, weighing between 19 and 30 g, were studied. For lethal infections, 50 μl containing 0.8 × 108 to 1.6 × 108 CFU S. aureus in PBS was intratracheally inoculated per mouse. Control animals were intratracheally infused with 50 μl PBS only. For sublethal infections, 2 × 107 to 4 × 107 CFU of S. aureus was used per mouse. For S. pneumoniae infections, 106 CFU in 50 μl PBS was intratracheally inoculated. For K. pneumoniae infections, 104 CFU in 50 μl PBS was intratracheally inoculated. For P. aeruginosa infections, 7 × 106 CFU in 50 μl PBS was intratracheally inoculated. Mice were monitored twice daily for morbidity and mortality. In some experiments, CGRP8–37 (Genscript) was administered i.p. at 800 ng (256 pmol) or 7.5 μg (2.4 nmol) per dose in 200 μl PBS, at different time points relative to infection (0 h). Control mice received 200 μl PBS only.

Vital-sign measurements.

Heart rate, oxygen saturation, and perfusion were measured under isoflurane anesthesia by Pulse Oximetry with the Kent Scientific PhysioSuite (Kent Scientific Corporation). For accuracy, measurements were performed three independent times on different days for the same mice, and values represent the average of three measurements. Pulse oximetry could not be used on MRSA-infected mice because they could not survive isoflurane anesthesia, and thus the measurements were performed only at steady state. Respiratory rates were determined by manually recording the number of breaths per minute and were averaged over three measurements. Core body temperature was measured with a rectal thermal probe (Bioseb).

Genetic and chemical ablation of TRPV1+ nociceptors.

Trpv1-Dtr mice were treated with DT as previously described5. Mice were injected i.p. with 200 ng of DT (Sigma Aldrich) dissolved in 100 μl PBS or with 100 μl PBS (vehicle) daily for a 21-d period. 5- to 7-week-old male and female mice were used for these experiments. For chemical ablation of TRPV1+ neurons, C57BL/6 mice 4 weeks of age were treated with RTX (Sigma) as previously described19,20. Mice were injected subcutaneously in the flank on consecutive days with three increasing doses of RTX (30, 70, and 100 μg/kg) dissolved in 2% DMSO with 0.15% Tween 80 in PBS. Control mice were treated with vehicle alone. For intravital imaging experiments, the same dosage for RTX treatment was used in 4-week-old mice, except the vehicle for dissolution was DMSO (without Tween 80). RTX was diluted into DMSO (1 μg/ìl) and subsequently into saline before injections. For VG -targeted ablation, we performed bilateral intraganglionic injections of DT or PBS into Trpv1-Dtr mice as previously described5. 20 ng DT in 120 nl PBS containing 0.05% Fast Green was injected into nodose/jugular/petrosal VG with a nanoinjector (Drummond Scientific Company). Mice were anesthetized with 1–3% isofluorane with oxygen. The vagal ganglion was exposed after a midline incision in the neck (1.5 cm in length). DT was gently injected; this process was repeated for the vagal ganglion on the other side of the body.

Bronchoalveolar lavage fluid (BALF) analysis.

Mice were euthanized by CO2 inhalation, and the trachea were exposed and cannulated with a 20-gauge catheter (BD Insyte Autoguard). BALF was collected two times by instilling 0.8 ml of cold PBS containing heparin and dextrose, then centrifuged at 4,000 r.p.m. for 7 min at 4 °C, and the cell pellet was separated from the supernatant. Total BALF leukocytes were counted after red-blood-cell lysis (RBC lysis buffer, eBioscience) and subjected to flow cytometry. Cell-free BALF supernatant was filtered through a 0.22-μm filter and mixed with protease/phosphatase-inhibitor cocktail, then kept at −80 °C for protein and cytokine analysis.

Bacterial-load and cytokine measurements.

Lungs and spleen tissues were homogenized in 1 ml sterile water with BB beads (Daisy Outdoor Products) in a Tissue Lyzer II (Qiagen). Lung, spleen homogenates, blood, or BALF was serially diluted in PBS and plated on TSA plates. The bacterial CFU were enumerated after overnight incubation of TSA plates at 37 °C. Cytokine levels in lung homogenates and BALF were measured through enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturer's instructions (Biolegend).

Behavior tests.

For behavioral assays, observers were blinded to the treatment group and genotype. To measure heat sensitivity, mice were placed on a hot plate set at 52 °C (IITC Life Science). Latency to hindpaw lifting, licking or flinching was recorded, and stopped at a maximum of 60 s. For tail-flick assays, mice were kept vertically in a relaxed fashion with their tails immersed in a temperature-controlled water bath maintained at 52 °C. The latency to a tail flick was recorded, with a maximum of 60 s.

Immune-cell depletion.

For neutrophil depletion, we followed an established protocol54. Mice were injected i.p. with 125 μg of anti-GR1 (clone RB6-8C5, BioXCell, NH) (in 200 μl) per mouse 24 h before lung infection. Control mice received 125 μg of rat IgG (Jackson Immunoresearch). For depletion of alveolar macrophages, 100 μl CLL (purchased from was delivered intratracheally into mice 2 d before infection. Control mice received an equal volume of PBS-L.

Lung-barrier permeability assay.

FITC–dextran (Mol Wt 4,000, Sigma) was intratracheally inoculated in 50 μl/mouse at 20 mg per kg body weight. Control mice were inoculated with PBS. Four hours later, mice were euthanized, and blood was collected by cardiac puncture. Blood was allowed to coagulate for 30 min at room temperature in the dark and centrifuged at 2,500g for 15 min. Fluorescence in serum was recorded with a plate reader (BioTek Synergy) and normalized to FITC–dextran standards (1.56–100 μg/ml).

Flow cytometry.

Lung tissues were mechanically separated and minced, then digested in DMEM (Life Technologies) containing 2% FBS and 1.5 mg/ml collagenase D (Roche) at 37 °C for 1 h at 250 r.p.m. The cell mixture was passed through an 18-gauge needle three times and filtered through a 70-μm cell strainer (BD). Red blood cells were lysed with RBC lysis buffer (eBioscience), treated with Fc Block (Biolegend), and resuspended in FACS buffer (PBS, 2% FBS, and 1 mM EDTA). For splenocytes, spleens were mashed and filtered through a 70-μm strainer (BD). Red blood cells were lysed with RBC lysis buffer (eBioscience), treated with Fc Block, and resuspended in FACS buffer. Incubations with antibody cocktails were conducted on ice for 30 min, and samples were subjected to two washes and resuspension in PBS with 2% PFA and 1 mM EDTA before flow cytometry. Antibodies used for staining included: anti-CD11b-Brilliant Violet 605 (clone M1/17, BioLegend), anti-CD45-APC/Cy7 (30-F11, BioLegend), anti-Ly-6G-Alexa Fluor 488 (1A8, BioLegend), anti-Ly-6C-PerCP/Cy5.5 (HK1.4, BioLegend), anti-Gr1-FITC (RB6-8C5, BioLegend), anti-CD4 Pac Blue (GK1.5, BioLegend), anti-CD8α-PE/Cy7 (53-6.7, BioLegend), anti-CD11c-APC (N418, BioLegend), anti-CD64-Brilliant Violet 421 (X54-5/7.1, BioLegend), anti-SiglecF-Alexa Fluor 488 (E50-2440, BD Bioscience), anti-CD103 PE (2E7, BioLegend), anti-TCR γδ-PE (GL3, BioLegend), anti-F4/80-FITC (BM8, BioLegend), anti-NK1.1-PerCP/Cy5.5 (NK-1.1, BioLegend), anti-B220-APC (RA3-6B2, BioLegend), anti-CD3α-PE/Cy7 (17A2, BioLegend), anti-CD24 Brilliant Violet 510 (M1/69, BioLegend), anti-TCR-β Brilliant Violet 421 (H57-597, BioLegend), anti-TCR Vγ1.1-APC (2.11, BioLegend), and anti-TCR Vγ2-PE-Cy7 (UC3-10A6, eBioscience). Flow cytometry was conducted on an LSRII flow cytometer (BD). Data were collected with BD DIVA software, and files were analyzed with FlowJo (Treestar, version 10.0.8r1). A live-cell stain (eFluor 450, ebioscience) was used to exclude dead cells. Positive staining and gates for each fluorescent marker was defined by comparing full stain sets with fluorescence minus one (FMO) control stain sets.

Fluorescence-activated cell sorting of lung immune cells.

For FACS purification of lung-resident populations, we used antibodies against CD3ɛ (clone 145-2C11), CD4 (clone RM4-5), CD11b (clone M1/70), Ly6G (clone 1A8), CD19 (clone 6D5), CD45 (clone 30-F11), TCRβ (clone H57-597), and TCRγ/δ (clone GL3) from BioLegend. 7AAD was from BD Pharmingen. Single-cell suspensions were generated from lungs of two or three C57Bl/6J mice with a lung-dissociation kit (Miltenyi Biotec). Single-cell suspensions were incubated with CD90.2 MicroBeads (Miltenyi Biotec) and separated into CD90.2-positive and CD90.2-negative fractions. Both fractions were stained on ice with surface antibodies and live/dead marker 7AAD, and sorted on a BD FACSAria (BD Biosciences). Different cell types were identified through the following gating strategies: B cells (7AADCD45+CD19+) and neutrophils (7AADCD45+CD11b+Ly6G+) were sorted from the preenriched CD90.2 cell fraction, whereas CD4+ T cells (7AADCD45+CD3+TCRβ+CD4+) and TCRγδ T cells (7AADCD45+CD3+TCRβTCRγδ+) were sorted from the preenriched CD90.2+ cell fraction.

Quantitative real-time PCR.

An RNeasy Plus Mini Kit (Qiagen) was used to isolate RNA, which was reverse transcribed to cDNA with an iScript cDNA Synthesis Kit (Bio-Rad). Relative gene expression was determined by quantitative real-time PCR on a ViiA7 System (Thermo Fisher Scientific) with TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) with the following primer/probe sets: Trpv1 (Mm01246300_m1), Ramp1 (Mm00489796_m1), Calcrl (Mm00516986_m1), and Actb (Applied Biosystems). Expression values relative to Actb (detected in the same sample by duplex qPCR) were calculated.

Neuronal cultures and CGRP analysis.

DRG neuron cultures were grown as previously described30. In brief, total DRG were dissected from 8- to 12-week-old mice and digested in HEPES-buffered saline (Sigma) containing 1 mg/ml collagenase A and 3 mg/ml dispase II (Roche Applied Sciences) for 60 min at 37 °C. The cell suspension was triturated with fire-polished Pasteur pipettes, then centrifuged over a 12% BSA (Sigma) gradient. The top layer of debris was discarded, and cell pellets were resuspended in neurobasal (NB) medium containing B27 (Life Technologies). Neurons were plated on laminin-coated 96-well culture dishes in NB medium containing B27, 50 ng/ml nerve growth factor (Life Technologies), and penicillin/streptomycin (Life Technologies). The medium was changed every other day. At day 7, DRG neurons were stimulated with S. aureus or 500 nM capsaicin (Sigma) for 30 min, and supernatant was collected. CGRP levels in the culture supernatant, BALF, or lung homogenates were determined with a CGRP EIA kit according to the manufacturer's instructions (Cayman Chemical).

Gene expression analysis.

We analyzed transcript levels in mouse transcriptome data sets deposited at the Immunological Genome Project55 (GEO GSE15907). Data sets for CD4+, CD8+ T cells, B cells, γδ T cells, NK cells, macrophages, dendritic cells, and neutrophils were analyzed. Trpv1 expression was also analyzed in the Mouse Gene Atlas MOE430 transcriptome data set26 (GEO GSE1133). Microarray data were background corrected and normalized with the robust multiarray average (RMA) algorithm in GenePattern (Broad Institute). A heat map for average transcript values was plotted with Morpheus (Broad Institute), Trpv1 levels were also plotted with Prism (GraphPad).

Lung histology.

Whole lungs were dissected from mice after euthanasia, fixed and stored in 10% formalin (Sigma Aldrich). Samples were embedded, sectioned, and stained with H&E or with a Brown and Brenn stain for Gram-positive bacteria by the Harvard Rodent Histopathology Core. Light microscopy of histological sections was conducted on a Nikon Ti-E microscope.

Immunofluorescence and microscopy.

For immunostaining, mice were perfused with PBS followed by 4% paraformaldehyde (PFA) in PBS. VG and thoracic DRG (T1-T13) were dissected and postfixed for 2 h in 4% PFA/PBS at 4 °C, incubated O/N at 4 °C with 30% sucrose/PBS, embedded in optimal cutting temperature compound (OCT, Tissue-Tek, PA) and stored at −80 °C. 12-μm cryosections were cut and immunostained with the following antibodies: guinea pig anti-TRPV1 (Millipore, AB5566, dilution 1:1,000), rabbit anti-CGRP (Sigma, C8198, dilution 1:5,000), mouse anti-NF-200 (Millipore, MAB5266, dilution 1:1,000), rabbit anti-βIII-tubulin (Tuj1) (Abcam, ab18207, dilution 1:1,000), and mouse anti-βIII-tubulin (Abcam, ab7751, dilution 1:500). Secondary antibodies included DyLight-488 donkey anti-rabbit IgG (Abcam, 1:500), CF-488A goat anti–guinea pig IgG (Sigma, 1:500), Alexa 488 donkey anti-mouse IgG (Abcam, 1:500), Alexa 594 donkey anti-mouse IgG (Abcam, 1:500) and Alexa 594 donkey anti-rabbit IgG (Abcam, 1:500). Sections were mounted in Vectashield and imaged with a Nikon Ti-E microscope with 10× magnification with NIS Elements software (Nikon, version AR3.22.08). For quantification of VG and DRG neuronal populations, images were analyzed by observers blinded to genotypes and treatment groups. β-tubulin-III was used as a general neuronal marker. Multiple fields were captured from three mice per group.

For immunostaining of lung sections, lungs were perfused with cold PBS and gravity inflated with 4% PFA/PBS. After overnight fixation at 4 °C, lungs were incubated 2 d in 30% sucrose/PBS, cryoembedded in OCT, and stored at −80 °C until sectioning. 40-μm cryosections were blocked for 4 h in PBS with 10% donkey serum, 2% bovine serum albumin (BSA), and 0.8% Triton X-100. Sections were incubated with rabbit anti-CGRP (C8198, Sigma) at a 1:5,000 dilution in blocking solution (PBS with 2% donkey serum, 2% BSA, and 0.3% Triton X-100) for 16–18 h at 4 °C, then incubated with secondary antibody (Alexa 594 donkey anti-rabbit IgG H&L, Abcam, ab150076) at a 1:500 dilution in blocking solution at 4 °C. Stained specimens were imaged on an inverted laser scanning confocal microscope (Olympus Fluoview FV1000).

Adobe Photoshop (Adobe) was used to quantify CGRP+ airway innervation in lung sections. The lasso tool was used to trace the circumference of each inner lung airway on the basis of DAPI staining; The area enclosed (in pixels) was derived from the histogram tool. The traced circumference was expanded by 162 pixels (100 μm) for each airway. The magic-wand tool was used to select the CGRP+ pixels within the circumscribed area and quantified out of the entire airway border, which is the percentage area covered by CGRP+ nerve fibers.

Neutrophil isolation and bacterial killing.

Mouse bone marrow neutrophils were isolated with an EasySep Mouse Neutrophil Enrichment Kit according to the manufacturer's instructions (StemCell Technologies). A Diff-Quick Stain kit (Thermo scientific) was used to confirm purity, which was found to be >95% neutrophils. To perform bacterial killing assays, 2.5 × 105 neutrophils were cocultured with 5 × 105 CFU of S. aureus (MOI of 2) in DMEM (Gibco) containing 5% FBS. CGRP (Genescript) was added at different concentrations, and cells were incubated at 37 °C for the indicated times. Neutrophils were then centrifuged at 600 r.p.m. for 2 min, and the pellet was washed twice with PBS and incubated for 30 min with 200 μg/ml gentamicin. Neutrophils were lysed with 0.1% Triton X-100, and dilutions were plated onto TSA plates to measure intracellular, viable bacteria. Lung cells were also isolated from C57BL/6J mice as described above in the flow cytometry section and cocultured with S. aureus (MOI of 2) with or without CGRP (100 nM) at the indicated time points for measurement of IL-6, TNF-α, and CXCL-1 levels in culture supernatants with specific ELISA kits (BioLegend).

Pulmonary intravital microscopy.

After mice were anesthetized (ketamine, 100 mg/kg; xylazine, 10 mg/kg), they received a jugular vein catheter for administration of fluorescent antibodies or additional anesthetics. To visualize neutrophils and endothelium, 3.5 μg Alexa Fluor 594–conjugated anti-Ly6G antibody (clone 1A8, BioLegend) and 5 μg Alexa Fluor 647–conjugated anti-CD31 antibody (clone MEC13.3, BioLegend) were injected intravenously. Mice were placed on a heating pad maintained at 37 °C and connected to mechanical ventilation (Harvard Apparatus) after tracheostomy was performed. Mice were kept in a right lateral decubitus position, and the left lung was exposed after thoracotomy and rib resection. A portion of the lung was immobilized via a gentle vacuum chamber with a glass slide fitted on top. A resonant-scanner confocal microscope (Leica SP8) equipped with a white-light laser and three HyD spectral detectors and a 25×/0.9 water objective lens were used for intravital microscopy. Images were acquired every 10 s for a total of 10 min, and three to five fields of view were observed. All videos and images were processed and analyzed with Leica software and Volocity software. For neutrophil behavior analysis, tethering was defined as rapid movement with blood flow and stop for less than 30s. Adhesion was defined as neutrophils that remained stationary for 30 s or longer. Crawling was defined as polarized cells that maintained continuous contact with the endothelium while changing physical location for at least 30 s. Random neutrophils were tracked manually in 10-min videos in Volocity software.

Sample size and statistical analysis.

For survival studies, we used animal numbers between 5 and 20 mice per experimental group/genotype. For core-body-temperature measurements, we used animal numbers between 3 and 5 mice per group/genotype. For survival studies and core-body-temperature measurements, experiments were performed at least twice, and data from individual mice were pooled from all experiments. For bacterial-load recovery and FACS analyses, 4–18 mice per group/genotype were used. For measurement of cytokine and CGRP levels, 3–12 mice per group/genotype were used. Survival data were analyzed with the log-rank test; bacterial-load recovery, behavioral data, core body temperature, FACS and cytokines were compared with two-way ANOVA with Bonferroni post tests, one-way ANOVA with Bonferroni post tests, two-tailed unpaired t tests for parametric analyses, or Mann–Whitney test for nonparametric analyses. Data were plotted in Prism (GraphPad).

Life Sciences Reporting Summary.

Further information on experimental design is available in the Life Sciences Reporting Summary.

Data availability.

All relevant data are readily available upon reasonable request to the corresponding author.

Additional information

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.