Mapping of the Sensory Innervation of the Mouse Lung by Specific Vagal and Dorsal Root Ganglion Neuronal Subsets

Abstract The airways are densely innervated by sensory afferent nerves, whose activation regulates respiration and triggers defensive reflexes (e.g., cough, bronchospasm). Airway innervation is heterogeneous, and distinct afferent subsets have distinct functional responses. However, little is known of the innervation patterns of subsets within the lung. A neuroanatomical map is critical for understanding afferent activation under physiological and pathophysiological conditions. Here, we quantified the innervation of the mouse lung by vagal and dorsal root ganglion (DRG) sensory subsets defined by the expression of Pirt (all afferents), 5HT3 (vagal nodose afferents), Tac1 (tachykinergic afferents), and transient receptor potential vanilloid 1 channel (TRPV1; defensive/nociceptive afferents) using Cre-mediated reporter expression. We found that vagal afferents innervate almost all conducting airways and project into the alveolar region, whereas DRG afferents only innervate large airways. Of the two vagal ganglia, only nodose afferents project into the alveolar region, but both nodose and jugular afferents innervate conducting airways throughout the lung. Many afferents that project into the alveolar region express TRPV1. Few DRG afferents expressed TRPV1. Approximately 25% of blood vessels were innervated by vagal afferents (many were Tac1+). Approximately 10% of blood vessels had DRG afferents (some were Tac1+), but this was restricted to large vessels. Lastly, innervation of neuroepithelial bodies (NEBs) correlated with the cell number within the bodies. In conclusion, functionally distinct sensory subsets have distinct innervation patterns within the conducting airways, alveoli and blood vessels. Physiologic (e.g., stretch) and pathophysiological (e.g., inflammation, edema) stimuli likely vary throughout these regions. Our data provide a neuroanatomical basis for understanding afferent responses in vivo.


Introduction
The mammalian lung is densely innervated by sensory afferent nerves, whose activation modulates respiratory rhythms and triggers reflex changes in airway function (Coleridge and Coleridge, 1984;Mazzone and Undem, 2016). Most airway sensory nerves are projected from neurons residing in the vagal ganglia, but there is also a small component projected from the dorsal root ganglia (DRGs) neurons. Airway sensory nerves are heterogenous with respect to size, myelination, conduction velocity, stimuli sensitivity, and neuropeptide content.
Critically, studies have identified biochemical and genetic markers of airway sensory subsets that, when activated, evoke specific physiological responses (Ricco et al., 1996;Ho et al., 2001;Undem et al., 2004;Kubin et al., 2006;Nassenstein et al., 2010;Lieu et al., 2011;Chang et al., 2015;Nonomura et al., 2017;Wang et al., 2017;Mazzone et al., 2020;Taylor-Clark, 2021). Lung sensory nerves have been broadly characterized into two groups: (1) myelinated fast conducting Ab fibers that express the mechanosensitive Piezo2 channel and are activated by physiologically-relevant mechanical forces (inflation and deflation). Activation of these afferents contributes to homeostatic regulation of breathing (so-called "lowthreshold mechanoreceptors"); (2) unmyelinated slow conducting C fibers that are activated by noxious stimuli, triggering defensive reflexes such as cough, dyspnea, mucus secretion and bronchospasm (so-called "nociceptors"). The capsaicin-sensitive transient receptor potential vanilloid 1 channel (TRPV1) is a marker of nociceptive neurons and is expressed on almost all airway C fibers (Ricco et al., 1996;Ho et al., 2001;Kollarik et al., 2003;Undem et al., 2004;Hooper et al., 2016). Excessive activation of airway afferents, in particular C fibers, is responsible for many symptoms and outcomes of airway disease caused by inflammation or infectious agents. Indeed, ablation of TRPV1-expressing vagal afferents reduces airway hyperreactivity associated with allergic asthma and increases survival and bacterial clearance in a model of pneumonia (Tränkner et al., 2014;Baral et al., 2018). Thus, airway sensory nerves and their evoked reflexes are therapeutic targets.
Importantly, embryological origin also determines physiological function (Taylor-Clark, 2021), e.g., jugular C fiber activation evokes cough and tachykinin-dependent neurogenic bronchospasm, whereas nodose C fiber activation decreases cough reflex sensitivity and fails to induce neurogenic bronchospasm (Ellis and Undem, 1994;Muroi et al., 2013;Chou et al., 2018). However, heterogeneity in other reflex responses is not so easily explained: in anesthetized guinea pigs, inhalation of capsaicin causes tachypnea, whereas application of capsaicin to the trachea causes bradypnea and intravenous capsaicin causes tachypnea followed by bradypnea (Chou et al., 2008). Furthermore, both the inhalation and intravenous injection of the irritant allyl isothiocyanate Nesuashvili et al., 2013) causes parasympathetic-mediated reflex bradycardia in rats, but only the inhalationevoked reflex is abolished by anesthesia (Hooper et al., 2019). As such, in addition to the jugular versus nodose paradigm that determines nociceptive function, there appears to be further complexity which may depend on anatomically distinct subsets.
Despite extensive characterization of the electrophysiological properties of airway afferent subsets, there is a gap in our understanding of their innervation patterns and how this impacts their sensitivity to physiological and pathophysiological stimuli. Immunohistochemistry (IHC) of lung afferents has yielded conflicting data (Taylor-Clark, 2021) and is hampered by a lack of selective markers and the difficulty in labeling varicosities. Recently, genetically-defined afferents have been identified in the lung using cre-lox reporter systems (Nonomura et al., 2017;Su et al., 2021). Here, we have used the cre-lox system to systematically map the afferent innervation of conducting airways, blood vessels and alveolar regions by genetically-defined vagal and DRG subsets using Pirt Cre (all afferents), TRPV1 Cre (nociceptors), Tac1 Cre (tachykinergic afferents), and 5HT 3 Cre (nodose afferents).
Purification of DNA for genotyping from ear punches was performed using the HotSHOT procedure (Truett et al., 2000). PCR was performed using HotStarTaq DNA Polymerase (QIAGEN). For each TRPV1 Cre , Tac1 Cre , and Pirt Cre PCR, 5.8 ml of DNAase/RNase-free distilled H 2 O was mixed with 1.2 ml 10Â PCR buffer (QIAGEN), 1 ml of 25 mM MgCl 2 , 1 ml of deoxynucleotide triphosphate mixture (dNTPs; Takara), 0.5 ml for each of the two primers (20 mM) and 2 ml of purified DNA. Separate reactions were used for mutant and wild-type alleles. TRPV1 Cre wild-type primers were TTC AGG GAG AAA CTG GAA GAA (forward) and TAG TCC CAG CCA TCC AAA AG (reverse), yielding a 490-bp product. TRPV1 Cre mutant primers were GCG GTC TGG CAG TAA AAA CTA TC (forward) and GTG AAA CAG CAT TGC TGT CAC TT (reverse), yielding a 102-bp product. The common reverse primer for Tac1 Cre was GCA TAT TTG GCT TTT ACT CTG G; the wild-type forward primer was GCA TGT TTC CTG TTT CGT GA, yielding a 362-bp product; the mutant forward primer was TGG TGG CTG GAC CAA TGT, yielding a 510bp product. The common reverse primer for Pirt Cre was TCC CTG GGA CTC ATG ATG CT; the wild-type forward primer was CAA CTT TGT GGT ACC CGA AG, yielding a 194-bp product; the mutant forward primer was ATC CGT AAC CTG GAT AGT GAA, yielding a 277-bp product. For each 5HT 3 Cre PCR, 17.5 ml of DNAase/RNase-free distilled H 2 O was mixed with 2.5 ml 10Â PCR buffer, 1 ml of 25 mM MgCl 2 , 1 ml of dNTPs, 0.5 ml for each of the two primers (20 mM) and 2 ml of purified DNA. 5HT 3 Cre transgene primers were GTC TGC CTG GGA CAT GAG GTT G (forward) and CGG CAA ACG GAC AGA AGC ATT (reverse), yielding a 208-bp product. Following an initial 3-min denaturation at 95°C, the DNA was amplified for 30 cycles of denaturation at 94°C for 30 s, followed by annealing at 62°C (TRPV1 Cre and Tac1 Cre ) or 55°C (Pirt Cre and 5HT 3 Cre ) for 30 s and then extension at 72°C for 1 min. The final extension period was increased to 5 min; 4 ml of product was mixed with 6 ml of DNAase/RNase-free distilled H 2 O and 2.5 ml 5Â DNA Loading Buffer Blue (Bioline) then run on a 1.5% agarose gel with 100-bp Hyperladder. Bands were visualized with a Biorad GelDoc.

Unilateral injection of AAV into the vagal ganglia
Using an intraganglionic injection procedure that has been described elsewhere (Kim et al., 2020b), ;2 cm of incision was made over a shaved superficial portion of the masseter muscle area. The skin was retracted, and the vagus nerve was located. The vagus nerve was separated from the common carotid artery and the anterior laryngeal nerve using a cotton tip. The vagal nodose ganglia was carefully exposed. The virus microinjection assembly consisted of a pulled glass micropipette (;20-mm tip diameter) attached to a 1-ml syringe via plastic tubing. Micropipettes were filled with AAV9flex-GFP (1 ml) using capillary force. The tip of the micropipette was gently inserted into the vagal ganglia and then injected by depressing the plunger (;0.5 Pounds per square inch). For co-injections of AAV9-flex-GFP, AAV9-flex-tdT, and AAV9-Flex-Ruby2sm-Flag, 1 ml of each virus were premixed in a tube, and then a final volume of 1 ml was injected into the ganglia. Three to six weeks later, mice were euthanized by CO 2 asphyxiation and the ipsilateral and contralateral vagal ganglia and lungs were collected (see below).

Intraganglionic injection of AAV into the thoracic DRGs
Approximately 4 cm of incision was made over a shaved portion of the mouse back region, just around the neck area. The skin flaps were opened, and the neck muscles were identified. The C7 and T1 vertebra were used as visual guides. We then separated muscle fibers to get to the T1 and T2 intervertebral space. We followed the spinal nerves and identified the beginning of the ganglia. A pulled glass micropipette (;20-mm tip diameter) was prefilled with 0.75 ml of AAV9-flex-tdT then injected unilaterally into the T1-T3 DRGs. Three to six weeks later, mice were euthanized by CO 2 asphyxiation and the ipsilateral and contralateral DRG and lungs were collected (see below).
Intratracheal instillation of rAAV2 for airway-specific nerve tracing Using a procedure that has been described elsewhere (Kim et al., 2020b), 30 ml of rAAV2-flex-tdT was diluted with 20 ml of minimum essential medium (MEM; Invitrogen) for lung instillation via endotracheal intubation. The mouse was placed on a vertical stand and tongue was gently pulled to find the intubation path. Using an otoscope attached with a speculum, a 20-gauge intravenous catheter (1.5 inches, BD Insyte) was inserted into the trachea. Successful intubation was confirmed by observing respiration-evoked changes in the liquid level in an attached syringe. The syringe was removed and 50 ml of virus/MEM mixture was pipetted into the catheter. The lung was then inflated with a 1-ml syringe filled with 300 ml of air to ensure instillation of the entire volume of the virus/MEM mixture. Three to six weeks later, mice were euthanized by CO 2 asphyxiation and the vagal ganglia, DRG, brainstem, and lungs were collected (see below).

Labeling of pulmonary vasculature using liquid latex
Following euthanasia with CO 2 inhalation, the mouse was perfused with ice-cold PBS through left ventricle. The pulmonary artery was ligated with 5-0 monofilament suture. On the distal side, the artery was cannulated using a 22-gauge blunt needle connected to polyethylene tubing. The inferior vena cava was clamped. Another 22-gauge blunt needle connected to polyethylene tubing was placed in the left ventricle to target the pulmonary vein. A 1-ml syringe was filled with pink and blue colored liquid latex (VWR, for the pulmonary artery and vein, respectively. The liquid latex for both the artery and vein was simultaneously injected until the liquid latex was visually observed to fill the lung blood vessels. The lung was then dissected out and dropped in ice-cold PBS. After 10 min or washing, the lung was postfixed in ice-cold 3.7% formaldehyde (FA) overnight. The tissue was then washed with ice-cold PBS three times for 30 min. The lung was inflated with 2% low melting agarose via the trachea. After the agarose solution solidified, the lung lobes were separated and 200-mm slices were collected using a vibratome. Lung slices were stained and mounted on a glass slide with VECTASHIELD Antifade Mounting Medium with DAPI (H-1200, Vector Laboratories) for imaging. Brightfield images were obtained using a Keyence microscope (BZ-X700), and fluorescent images were obtained using an Andor Dragonfly spinning disk confocal microscope.

Tissue collection and IHC
Mice were euthanized by CO 2 inhalation and transcardially perfused with ice-cold PBS. Vagal ganglia, DRG from T1 to T6 and brainstem were dissected out. Vagal ganglia and DRG were postfixed for 1 h, and the brainstem was postfixed for 4 h in FA at 4°C. These tissues were processed and immunostained as described previously (Kim et al., 2020b). The tissue was washed in PBS to remove residual FA and transferred to 20% sucrose solution for cryoprotection. The tissue was mounted in optimal cutting temperature (OCT) compound and snap frozen in dry ice. Vagal ganglia and DRG were sectioned in 20-mm slices, and brainstem was sectioned in 40-mm slices using a cryostat, and those were collected onto Superfrost plus slides. Slides were air-dried at room temperature in the dark overnight. Slides were washed with PBS three times for 10 min, and tissue was permeabilized with 0.3% Triton X-100 in PBS (PBSTx) for 15 min followed by blocking for 1 h with 1% bovine serum albumin (BSA)/10% donkey serum (DS)/0.3% PBSTx. Tissue was incubated with primary antibodies (Table 1) diluted in blocking buffer overnight at 4°C. After washing with 0.2% Tween 20 in PBS (PBST) three times for 10 min, the tissue was incubated with secondary antibodies (Table 1) in 1% BSA/5% DS in 0.2% PBST for 1 h. The tissue was washed with 0.2% PBST three times for 10 min and then, in some cases, counterstained with NeuroTrace fluorescent Nissl Stain for 1 h at 1:300 dilution in PBS. After washing with PBS, slides were air-dried and mounted with DPX mounting medium (Sigma).
Lungs were collected and postfixed in 3.7% FA overnight at 4°C with gentle agitation. Lungs were washed with ice-cold PBS three times for 30 min, followed by cryoprotection in 30% sucrose solution until the lung sank to the bottom of the tube. The lungs were flushed with PBS three times and inflated with 2% low melting agarose solution. After the agarose solution had solidified, the lung lobes were separated and snap frozen with OCT compound for cryosection; 80-mm lung slices were collected in cryoprotectant filled well-plates and stored in À20°C. Only ipsilateral lung lobes were included in the analyses in experiments with unilateral intraganglionic AAV injections. Using identical blocking solutions, permeabilizing solutions and antibody solutions (see above), lung slices was washed in PBS three times for 10 min and permeabilized for 20 min. Lung slices were then blocked for 1.5 h followed by primary antibody incubation overnight at 4°C. The slices were then washed three times for 20 min and incubated with secondary antibodies for 2 h. The slices were washed again and mounted onto glass slides with VECTASHIELD Antifade Mounting Medium with DAPI.

Experimental groups
1. Pirt-Ai9 mice. These mice were divided into two groups. The first group of five mice, which were harvested without latex labeling of the pulmonary vasculature, yielded 10 lung slices. Three lung slices were immunostained for a smooth muscle actin and E-Cadherin, three lung slices were immunostained for E-Cadherin and DAPI and four lung slices were immunostained for CGRP and DAPI. In all 10 slices, native tdTomato was imaged without immunostaining. The second group of three mice, which were harvested following latex labeling of the pulmonary vasculature, yielded six slices. These were immunostained for tdTomato and DAPI. The data from both groups was combined to quantify the overall innervation by tdTomato1 fibers in the Pirt-Ai9 mice. 2. Three Pirt Cre mice were administered rAAV2-flex-tdT via intratracheal instillation, yielding seven lung slices. These were immunostained for tdTomato and DAPI. 3. Three Pirt Cre mice received injection of AAV9-flex-tdT into the DRG and AAV9-flex-GFP into the vagal ganglia, yielding seven lung slices. These were immunostained for E-Cadherin, GFP, tdTomato, and DAPI. 4. Seven 5HT 3 -Ai9 were harvested, yielding 19 lung slices. Six lung slices were immunostained for a smooth muscle actin and E-Cadherin, three lung slices were immunostained for E-Cadherin and DAPI, and 10 lung slices were labeled for DAPI alone. In all 19 slices, native tdTomato was imaged without immunostaining. 5. Tac1-Ai9 mice. These mice were divided into two groups. The first group of six mice, which were harvested without injection of the vagal ganglia, yielded 13 lung slices, which were immunostained for E-Cadherin and DAPI. The second group of five mice, which received injection of AAV9-flex-GFP into the vagal ganglia, yielded 14 lung slices. These were immunostained for GFP and DAPI. For all Tac1-Ai9 lung slices, native tdTomato was imaged without immunostaining. The data from both groups was combined to quantify the overall innervation by tdTomato1 fibers in the Tac1-Ai9 mice. 6. Three Tac1 Cre mice received injection of AAV9-flex-tdT into the DRG, yielding five lung slices. These were immunostained for E-Cadherin, tdTomato, and DAPI. 7. TRPV1 Cre mice receiving AAV injection into the vagal ganglia were divided into two groups. The first group of seven mice received vagal injection of AAV9-flex-GFP and these yielded 13 lung slices. Eight lungs slices were immunostained for GFP, E-Cadherin, and DAPI, and five lung slices were immunostained for GFP, CGRP, E-Cadherin, and DAPI. The second group of three mice received vagal injection of AAV9flex-GFP, AAV9-flex-tdT, and AAV9-Flex-Ruby2sm-Flag, and these yielded six lung slices that were immunostained for GFP, tdTomato, Flag, and DAPI. The data from both groups were combined to quantify the overall innervation by GFP1 fibers in the TRPV1 Cre mice receiving AAV injection into the vagal ganglia. 8. Four TRPV1 Cre mice were administered rAAV2-flex-tdT via intratracheal instillation, yielding nine lung slices. These were immunostained for tdTomato and DAPI. 9. Three TRPV1 Cre mice received injection of AAV9-flex-tdT into the DRG, yielding three lung slices. These were immunostained for E-Cadherin, tdTomato, and DAPI. 10. Controls: three lung slices from 5HT 3 -Ai9 mice were used as controls for immunostaining for E-Cadherin and a smooth muscle actin (primary antibody was omitted). In addition, three lung slices from three TRPV1 Cre mice that had not received AAV9 injections were immunostained for GFP, tdTomato, and Flag expression.
Imaging and data analysis Images were taken with Andor Dragonfly spinning disk confocal microscope equipped with a Zyla 4.2 PLUS sCMOS camera (2048 Â 2048 pixels with 6.5-mm pixel size). The pinhole size was 25 mm. We used either a 10Â UPLSAPO (0.4 NA), a 20Â UPLSAPO (0.75 NA), or a 40Â UPLSAPO (1.25 NA, silicone oil immersion) objective was used, depending on the study. Fluorophores were excited by laser wavelengths at 405, 488, 561, or 637 nm. Zstacked multi-tile images were stitched using either Fusion software or Imaris Stitcher. All 3D images were further processed using Imaris software. Nonconsecutive Research Article: New Research lung slices were imaged. In most cases the entire slice was imaged. In many cases, staining of E-Cadherin, a marker of airway epithelial cell adherens junctions, was used to identify conducting airways from blood vessels. Nevertheless, conducting airways and vessels could be easily distinguished by DAPI staining: conducting airways are comprised of a compact monolayer of epithelial cells surrounded by lamina propria and smooth muscle compared with diffused endothelial and muscle cells in vessels. All identified conducting airways and vessels had their diameter measured (using the shortest length across the lumen). We then assessed the presence of reporter1 fibers innervating each airway/vessel. This analysis did not distinguish between the density of the fibers or branches. We also determined whether the reporter1 fiber projected away (.20 mm) from the conducting airway or blood vessel into the alveolar region. The maximal straight-line distance projected by the fiber was recorded. Efforts were made to identify nerve terminals within the physical slice, but terminals were not systematically quantified. For some analyses, airways and vessels were grouped into three categories based on diameter: small (0-175 mm), medium (176-375 mm), and large (376 mm and greater). Control experiments (see study group 10) indicated that reporter expression required AAV-mediated or ROSA26mediated expression. For experiments involving unilateral intraganglionic AAV injections (vagal, DRG, or both), lung data were only included in the analysis if the injections induced selective and widespread reporter expression in neurons within the respective ipsilateral ganglia. Brightfield images (Keyence) of lung slices from Pirt-Ai9 mice with liquid latex injected into the pulmonary artery (pink) and pulmonary vein (blue) were overlayed with their corresponding confocal fluorescence images (Andor) to determine the Pirt1 fiber innervation of identified arteries and veins. In CGRP-stained lung slices, we identified CGRP1 epithelial cell clusters, termed neuroepithelial bodies (NEBs). The diameter of the conducting airway at the location of the NEB and the number of CGRP1 (and, in some cases, Pirt1) cells within the cluster was recorded. In addition, we analyzed the presence of CGRP1 and reporter1 fibers innervating the NEB. This was defined as a nerve fiber within 10 mm of the NEB. All data were analyzed in GraphPad Prism 9. Groups were compared with the nonparametric Mann-Whitney two-tailed U test, p , 0.05 was considered significant.

Comparison of lung structures innervated by specific sensory nerve subsets
We have used a series of cre recombinase-expressing mouse strains (Pirt Cre , 5HT 3 Cre , Tac1 Cre , and TRPV1 Cre ) to identify specific sensory nerve subsets innervating the mouse lung. Cre expression within specific neuronal populations within the sensory ganglia for each of these strains has been previously characterized (Gong et al., 2003;Cavanaugh et al., 2011b;Patel et al., 2011;Harris et al., 2014;Kim et al., 2020b;Su et al., 2021). Cell-specific expression was visualized by Cre-driven fluorescent reporter expression (typically GFP or tdTomato) under the control of either the endogenous ROSA26 gene (Ai9 mouse) or AAV instilled into the lungs or unilaterally injected into sensory ganglia (vagal ganglia or DRG; Fig.  1A). Ipsilateral lung slices were sectioned, stained for specific markers and imaged. In some experiments the expression of the sensory neuropeptide CGRP was also determined using IHC. Conducting airways (bronchi and bronchioles) and blood vessels .25 mm in diameter were analyzed (Fig. 1B) and stratified into three groups: 0-175 mm (small), 176-375 mm (medium), and .376 mm (large; Fig. 1C). Each structure was analyzed for the presence of (1) reporter1 fibers within close proximity (,40 mm) to the epithelium/endothelium, and (2) reporter1 fibers that project away from the structure into the alveolar regions. Generally, .75% of conducting airways had reporter1 fibers (Fig. 1D), although this proportion decreased for the smaller conducting airways (Fig. 1E). Only a small subset of conducting airways had reporter1 fibers that projected into the alveolar region ( Fig. 1D), and this only occurred for conducting airways ,375 mm in diameter (Fig. 1F). There was little difference between the distance that re-porter1 fibers from the different strains projected into the alveolar region (Fig. 1G). A small subset of blood vessels had reporter1 fibers (Fig. 1D), and this proportion tended to decrease for the smaller blood vessels (Fig. 1H). Of the 1619 blood vessels analyzed, none had reporter1 fibers that projected into the alveolar regions, and so this analysis was not presented. For some unilateral intraganglionic AAV injections, contralateral lung slices were studied. While some reporter1 fibers were observed, these were very limited in number and not systematically analyzed further.
In lung slices of Pirt-Ai9 (n = 8 animals, n = 16 lung slices), tdTomato1 fibers were found across multiple structures (Fig. 1D). Surprisingly, a small subset of airway smooth muscle cells also expressed tdTomato (Fig. 2B), which to a certain extent obscured the visualization of the tdTomato1 fibers within the airways. A total of 210 of the 219 conducting airways (96%) had tdTomato1 fibers (Fig.  1D,E), and these were found in a complicated plexus of fibers and identified terminations (Fig. 2C,D) beneath the epithelium. Typically, individual fibers were found just beneath the epithelium, whereas bundles of multiple fibers were found on the outside surface of the encircling smooth muscle cells (Fig. 2E). A total of 71 of the 219 conducting airways (32%) had tdTomato1 fibers that projected into the alveolar region, but this only occurred for Figure 1. Comprehensive quantification of lung innervation patterns by specific sensory nerve subsets. A, Four approaches used to label specific sensory populations with the fluorescent reporters GFP or tdTomato. B, Histogram of the diameters of conducting airways and blood vessels analyzed. Dotted lines denote 175 and 375 mm. C, Relative quantification of the diameters of conducting airways and blood vessels clustered into "small," "medium," and "large" groups. D, Overall quantification of innervation by specific sensory nerve subsets: % of conducting airways with fibers (black), with fibers that project out to the alveolar region (red), and blood vessels with fibers (green). E, % of conducting airways with fibers of specific afferent subsets, clustered by airway diameter. F, Percentage of conducting airways with fibers of specific afferent subsets that project out to the alveolar region, clustered by airway diameter. G, Distance projected from the conducting airway into the alveolar region by fibers of specific afferent subsets (red airways ,375 mm in diameter (Figs. 1D,F, 2F). A total of 86 of the 270 blood vessels (32%) had tdTomato1 fibers and this innervation was much less prevalent for small blood vessels (Figs. 1D,H, 2G,H). For some mice (n = 3 animals, n = 6 lung slices), we also injected colored latex beads into the pulmonary artery (pink) and pulmonary vein (blue) before dissection to selectively label these different vessels (Fig. 2I-L). Almost all vessels .25 mm contained either pink or blue latex, consistent with the lack of bronchial circulation in the mouse intrapulmonary airways (Mitzner et al., 2000). Consistent with other reports, pulmonary arteries invade the lung in conjunction with the bronchopulmonary tree, whereas pulmonary veins were often isolated. A total of 45 of the 101 pulmonary arteries (45%) had tdTomato1 fibers, and arteries without innervation were smaller than those with innervation (Mann-Whitney two-tailed U test, p = 0.0045; Fig. 2M). Only 26 of the 104 pulmonary veins (25%) had tdTomato1 fibers, but again there was a considerable correlation of innervation and diameter (Mann-Whitney two-tailed U test, p , 0.0001), with all large veins having Pirt1 innervation but only 9% of small vessels having Pirt1 innervation (Fig.  2N). tdTomato1 fibers were observed to penetrate the muscle layer surrounding the innervated arteries and veins. Some mouse intrapulmonary veins have a variable and discontinuous cardiomyocyte coat in close proximity to the vascular smooth muscle layer (Mueller-Hoecker et al., 2008), but these cells were not investigated in the current study.
To better visualize the Pirt1 innervation of the lung without the obscuring muscle cell tdTomato expression observed in Pirt-Ai9, we instilled rAAV2-flex-tdTomato into the lungs of Pirt Cre mice (n = 3 animals, n = 7 lung slices; Fig. 3A). Within the vagal ganglia, AAV-mediated expression of tdTomato was noted in 113 out of 3500 vagal neurons (3.2%; Fig. 3B). These airway-specific vagal neurons projected tdTomato1 central terminations to the brainstem nTS (and the neighboring area postrema; Fig. 3C-F), but no tdTomato1 fibers were found within the paratrigeminal complex (Pa5; Fig. 3C,D). Interestingly, we found no tdTomato1 neurons within the DRG (out of 4522 neurons), indicating that these neurons were not labeled by lung instillation with rAAV2. In the lung, 107 of the 127 conducting airways (84%) had tdTomato1 fibers (Figs. 1D,E, 3G). Almost no tdTomato1 fibers penetrated the airway epithelium of conducting airways (Fig. 3H). Instead tdTomato1 fibers were found within the lamina propria and smooth muscle layers, as well as on the outside surface of the smooth muscle (Fig. 3H). The tdTomato1 fibers within the lamina propria appeared thinner than the fibers on the surface of the smooth muscle, but tracing indicated that these were the same fiber populations, which got progressively thinner when they penetrated the smooth muscle layer. Confirmed tdTomato1 terminations were also occasionally found within the conducting airways (Fig. 3I). A total of 44 of the 127 conducting airways (35%) had tdTomato1 fibers that projected into the alveolar region, but again this only occurred for airways ,375 mm in diameter (Figs. 1D,F, 3J). Only 17 of the 134 of blood vessels (13%) had tdTomato1 fibers and, unlike the Pirt-Ai9 dataset, none of the large blood vessels were innervated (Figs. 1D,H, 3K-M). For blood vessels ,375 mm in diameter, the tdTomato1 fibers appeared to innervate multiple layers of the muscle, in some cases penetrating close to the endothelial layer ( Fig. 3K-M).
The simultaneous injection of AAV9-flex-GFP into the vagal ganglia and AAV9-flex-tdTomato into the thoracic DRG of Pirt Cre mice (n = 3 animals, n = 7 lung slices; Fig.  4A) produced robust and selective reporter expression in neurons within these ganglia (Fig. 4B,C). In the lung, 194 of the 229 conducting airways (85%) had GFP1 fibers (vagal), whereas only 46/229 (20%) had tdTomato1 fibers (DRG; Figs. 1D, 4D-L). There was a considerable correlation of airway diameter and DRG Pirt1 innervation, with tdTomato1 fibers in 100% and 12% of large and small airways, respectively (Figs. 1E, 4D). The GFP1 and tdTomato1 fibers were part of the same loose plexus surround the conducting airways, beneath the epithelium (Fig. 4H). A total of 114 of the 229 conducting airways (50%) had GFP1 fibers (vagal) that projected out into the alveolar region, whereas none of the tdTomato1 fibers (DRG) projected into the alveolar region (Figs. 1D,F, 4D-L). A total of 31 of the 131 blood vessels (24%) had GFP1 fibers (vagal) compared with just 10/131 (8%) with tdTomato1 fibers (DRG; Figs. 1D, 4M-P). Although both innervations were more prevalent for larger blood vessels, this correlation was much more extreme for DRG Pirt1 fibers which almost exclusively innervated only large blood vessels (Fig. 1H). Qualitative analysis of the blood vessel innervation suggested that GFP1 fibers (vagal) innervate the outer muscle layers, whereas tdTomato1 fibers (DRG) sometimes project through the muscle layers to come in close apposition to the endothelial layer ( Fig. 4M-P).

Tac1 Cre
Tac1, the gene for preprotachykinin (precursor for the tachykinin neuropeptide substance P), is expressed in jugular and DRG neurons, but is expressed in very few nodose neurons innervating the airways (Ricco et al., 1996;Undem et al., 2004;Nassenstein et al., 2010;Usoskin et al., 2015;Kim et al., 2020b). Here, we investigated Tac11 innervation of the lungs using three approaches: Tac1-Ai9 (Fig. 6A), expressing tdTomato in all Tac1-expressing cells; the Tac1-AAV-GFP Vagal (Fig. 7A), expressing GFP in vagal afferents expressing Tac1; and the Tac1-AAV-tdT DRG (Fig. 8A), expressing tdTomato in DRG afferents expressing Tac1. In the lungs of Tac1-Ai9 mice (n = 11 animals, n = 27 lung slices), tdTomato1 fibers were noted in both conducting airways and blood vessels (Fig. 6B). In addition, tdTomato was expressed by a subset of intrinsic cells found in the vasculature and occasionally the continued (green) fibers innervating the conducting airways and projecting into the alveolar region (white arrows). M, Lung slice stained for DAPI (blue) showing tdTomato1 (red) and GFP1 (green) fibers innervating a blood vessel. N, Higher magnification of white box in M, showing tdTomato1 fibers (gray arrow) extending to the inner muscle layers of the vessel whereas GFP1 fibers innervate the outer layers only (white arrows). O, Lung slice stained for DAPI (blue) lacking GFP1 (green) fibers but showing a tdTomato1 (red) fiber in close proximity to a DAPI1 cell (presumed endothelial cell) adjacent to the blood vessel lumen. P, Lung slice stained for DAPI (blue) lacking tdTomato1 (red) fibers but showing GFP1 (green) fibers (white arrows) innervating outer muscle (m) layers of a blood vessel. In some images, lumens are denoted by "L." Scale bars denote 500 mm conducting airways. A total of 228 of the 281 conducting airways (81%) had tdTomato1 fibers (Figs. 1D,E, 6B-H). Smaller conducting airways were less likely to have Tac11 innervation (67%) compared with medium and large airways (96% and 100%, respectively; Fig. 1E). tdTomato1 fibers and their confirmed terminations were noted in the plexus beneath the epithelial layer of the conducting airways (Fig. 6B-H). Strikingly, no tdTomato1 fibers in the Tac1-Ai9 mouse lung projected into the alveolar region (Figs. 1D,F, 6B-H). 59 of the 259 blood vessels (23%) had tdTomato1 fibers (Figs. 1D,H, 6B,D). Tac11 innervation was particularly prevalent in large blood vessels (80%; Fig. 1H).
Some of the Tac1-Ai9 mice received an AAV9-flex-GFP injection into their vagal ganglia to compare the overall Tac11 innervation (tdTomato) with the vagal-specific Tac11 innervation (GFP; n = 5 animals, n = 14 lung slices; Fig. 7A). GFP1 terminations were found in both conducting airways and blood vessels (Fig. 7B-G). As expected, all GFP1 fibers also expressed tdTomato, but some tdTomato1 fibers lacked GFP (Fig. 7C,F,G), suggesting these particular Tac11 fibers were not projected from the vagal ganglia (possibly projected from DRG). A total of 162 of the 208 conducting airways (78%) had GFP1 fibers within the plexus surrounding the conducting airways (Figs. 1D,E, 7B-E). tdTomato1/GFP-negative fibers were more common in large conducting airways compared with smaller airways. Confirmed vagal Tac11 terminations were noted (Fig. 7E). No GFP1 fibers projected from the conducting airways into the alveolar region (Figs. 1D,F, 7B-E). A total of 34 of the 188 blood vessels (18%) had GFP1 fibers, and like the tdTomato1 (overall) Tac11 innervation, this was more common in large blood vessels. Comparison of Tac11 fibers innervating large blood vessels indicated that the confirmed vagal Tac11 fibers (tdTomato1/GFP1) innervated only the outer muscle layers, whereas the presumed nonvagal Tac11 fibers (tdTomato1/GFP-negative) projected through the muscle layers to within close proximity with the endothelial layer (Fig. 7G).
A lack of GFP expression in Tac11 fibers following vagal injection with AAV9-flex-GFP is not definitive evidence that the fiber does not project from vagal neurons (as transfection is rarely 100%). As such, we investigated the DRG Tac11 innervation by injecting the thoracic DRG of Tac1 Cre mice with AAV9-flex-tdTomato (n = 3 animals, n = 5 lung slices; Fig. 8A). Only 12 of the 153 conducting airways (8%) had tdTomato1 fibers (Figs. 1D, 8B-D). DRG Tac11 innervation was almost exclusively restricted to the large airways, of which 71% had tdTomato1 fibers (Fig. 1E). As expected, no tdTomato1 fibers projected into the alveolar region. Only eight of the 118 blood vessels (7%) had tdTomato1 fibers (Figs. 1D, 8E,F) and this innervation was restricted to medium and large blood vessels. In some cases, DRG Tac11 fibers projected through the muscle layers to come into close contact with the endothelial cells (Fig. 8F).
Instillation of rAAV2-flex-tdTomato into the lungs of TRPV1 Cre mice (n = 4 animals, n = 9 lung slices; Fig. 10A) resulted in AAV-mediated expression of tdTomato in 113 out of 3298 vagal neurons (3.4%) within the vagal ganglia Figure 7. Mapping the lung innervation by vagal Tac11 nerves. A, Approach for labeling all Tac11 afferents with tdTomato and vagal Tac11 afferents with GFP. B, Lung slice stained for E-cadherin (red) and DAPI (blue) showing GFP-expressing (green) nerves innervating conducting airways (white arrows). C, Lung slice stained for DAPI (blue) showing a large conducting airway trench innervated by GFP-expressing (green) nerves (white arrow) and tdTomato-expressing (red) nerves (gray arrow). D, Lung slice stained for DAPI (blue) showing a conducting airway innervated by fibers expressing both GFP (green) and tdTomato (red). E, Higher magnification of white box in D, with identified tdTomato1 nerve terminal (white arrow). F, Lung slice stained for DAPI (blue) showing a large blood vessel (slightly folded) innervated by GFP-expressing (green) nerves and tdTomato-expressing (red) nerves. G, Individual zplanes (1-3) of the white box in F, at higher magnification, showing fibers expressing only tdTomato innervating the inner muscle layers (gray arrows), whereas GFP-expressing fibers (white arrow) innervate only the outer muscle layer. In some images, lumens are denoted by "L." Scale bars denote 100 mm (B, C, D, F), 50 mm (G), or 20 mm (E). (Fig. 10B). These airway-specific vagal TRPV11 neurons projected tdTomato1 central terminations to the brainstem nTS and neighboring area postrema (Fig. 10C,D), but not to the Pa5 (Fig. 10C). Based on our Pirt Cre studies, we did not assess tdTomato expression in the DRG of these mice. In the lungs, 133 of the 161 conducting airways (83%) had tdTomato1 fibers (Figs. 1D,E, 10E). Furthermore, 70 of the 161 conducting airways (43%) had tdTomato1 fibers that projected into the alveolar region (Figs. 1D,F, 10E,F). Once again this only occurred for small and medium diameter airways (Fig. 1F). Only nine of the 139 blood vessels (6%) had tdTomato1 fibers, and such TRPV11 innervation was completely absent from large vessels (Fig. 1D,H).
Injecion of AAV9-flex-tdTomato into the thoracic DRG of TRPV1 Cre mice (n = 3 animals, n = 3 lung slices; Fig. 10G) produced robust tdTomato expression in a subset of DRG neurons (Fig. 10H). Nevertheless, very few structures within the lungs of these animals were innervated by tdTomato1 fibers. Only two of the 82 conducting airways (2%) had DRG TRPV11 innervation, and this was exclusively restricted to large airways (Figs. 1D,E, 10I). The very sparse tdTomato1 fibers were found within the airway smooth muscle layer (Fig. 10I). Similarly, only one of the 57 blood vessels (2%) had tdTomato1 fibers, and this too was exclusively restricted to large blood vessels (Fig. 1D,H).
In addition to CGRP expression in nerve fibers, CGRP was robustly expressed in clusters of epithelial cells termed NEBs (Fig. 11A-F). A total of 112 CGRP1 NEBs were found throughout the 296 conducting airways. The number of CGRP1 cells within a particular NEB varied Figure 9. Mapping the lung innervation by vagal TRPV11 nerves. A, approach for labeling vagal TRPV11 afferents with GFP. B-E, lung slice stained for E-cadherin (red) and DAPI (blue) showing GFP-expressing (green) nerves innervating conducting airways (white arrows) and blood vessels (gray arrows). Note that GFP1 fibers are found within the smooth muscle layer of the large conducting airways in E. F, Lung slice stained for E-Cadherin (blue) showing a GFP-expressing (green) nerve projecting from a small conducting airway into the alveolar region (white arrow denotes identified terminal). G, Higher magnification of white box in F, with DAPI (white), showing GFP1 fiber intercalating (white arrow) with smooth muscle (sm) cells surrounding the conducting airway. H, Vagal ganglia of TRPV1 Cre mouse following vagal injection of triple cocktail of AAV9-flex-reporters: tdTomato (red), GFP (green), and FLAG (blue). I-L, lung slice of mouse from H. I, Lung slice stained for DAPI (blue) showing multiple individual fibers innervating large conducting airway including GFP1 fibers (green arrow), GFP1/tdTomato1 fibers (yellow arrows), and tdTomato1 fibers (red arrows). J, Lung slice stained for DAPI (white) showing individual GFP1 (green), tdTomato1 (red), and FLAG1 (blue) fibers innervating small conducting airways (red arrow, cyan arrow and yellow arrow). K, Lung slice showing a single GFP1/tdTomato1/FLAG1 fiber innervating a small blood vessel (identified terminal denoted by white arrow, parental axon denoted by gray arrow). L, 3-dimensional rotation of K to visualize the circumferential structure of the TRPV11 terminal innervating the vessel. In some images, lumens are denoted by "L." Scale bars denote 200 mm (B, C), 100 mm (H), 50 mm (D, E, F, I, J, K, L), or 20 mm (G). from 2 to 88 (mean of 12.8 6 1.1), solitary CGRP1 cells were excluded from this analysis. There did not appear to be any correlation between the number of cells within an NEB and the diameter of the conducting airway (Fig. 11L). In the lungs of Pirt-Ai9, 45 of the 74 CGRP1 NEBs (60%) unexpectedly contained cells that also expressed tdTomato (Fig. 11B,C), including all NEBs comprised of more than three CGRP1 cells. The mean number of tdTomato1 cells within an NEB was 5.7 6 2.5. Of the 112 CGRP1 NEBs studied in total, 62 (55%) were innervated by CGRP1 fibers (Fig. 11M). In the lungs of Pirt-Ai9, 53 of the 72 CGRP1 NEBs (74%) were innervated by tdTomato1 fibers (Fig. 11M). NEBs were innervated by both CGRP1 and CGRP-negative fibers expressing tdTomato. In the lungs of TRPV1 Cre mice following vagal injection of AAV9-flex-GFP, 18 of the 40 (45%) CGRP1 NEBs were innervated by GFP1 fibers (Fig.  11M). Almost all the GFP1 fibers (i.e., vagal TRPV11 fibers) innervating the NEBs also expressed CGRP. Importantly, multiple NEBs were innervated by GFPnegative/CGRP1 fibers (Fig. 11D,E). There were no differences in the diameters of conducting airways with NEBs that were or were not innervated by Pirt1, CGRP1, and GFP1 (vagal TRPV11) fibers (Mann-Whitney two-tailed U test, p = 0.49, p = 0.74, and p = 0.09, respectively; Fig. 11N). Nevertheless, the number of CGRP1 cells within an innervated NEB was significantly greater than the number within noninnervated NEBs (Mann-Whitney two-tailed U test, p , 0.0001, p = 0.0005, and p = 0.0264 for Pirt1, CGRP1, and GFP1, respectively; Fig. 11O): for example, in the Pirt-Ai9 lung: 41 of the 47 (87%) NEBs comprised of more than six CGRP1 cells were innervated by tdTomato1 (i.e., Pirt1) fibers, whereas only 12 of the 25 (48%) NEBs comprised of less than or equal to six CGRP1 cells were innervated. Furthermore, 13 of the 19 (68%) NEBs without tdTomato1 innervation had less than or equal to six CGRP1 cells. We also identified five solitary CGRP1 epithelial cells in the 296 airways analyzed, only one of which was innervated by CGRP1 fibers.

Discussion
Electrophysiological, biochemical, and transcriptomic studies of the vagal sensory innervation to the airways have identified distinct afferent subsets (Coleridge and Coleridge, 1984;Mazzone and Undem, 2016;Taylor-Clark, 2021). Selective activation or interruption of  . Mapping the lung innervation by CGRP. A-C, Pirt-Ai9 lung slice stained for CGRP (green), E-Cadherin (blue), and tdTomato1 pirt-expressing fibers/cells (red, in B, C). A, B, CGRP1 fibers (white arrows) innervating a conducting airway. Note the cluster of CGRP1 epithelial cells (gray arrow) comprising an NEB. C, Higher magnification of white box in B, showing the NEB with both CGRP1/Pirt1 cells (white arrow) and CGRP1/Pirt-negative cells (gray arrow). Note the gain of the red and green channels have been decreased to better visualize the NEB cells. Thus, less bright CGRP1 and Pirt1 fibers are no longer visible. D-H, lung slice of TRPV1 Cre mouse following vagal injection with AAV9-flex-GFP, stained for CGRP (red), E-Cadherin (blue), and GFP1 TRPV1-expressing vagal fibers (green). D, F, GFP1/CGRP1 (yellow arrows), GFP1/CGRP-negative (green arrows) and GFP-negative/CGRP1 (red arrows) fibers innervate the conducting airways. Also shown are CGRP1 NEBs (white arrows). E, Higher magnification of white box in D, showing an NEB with associated innervation by both GFP1/CGRP1 (yellow arrows) and GFP-negative/ CGRP1 (red arrows) fibers. G, H, Higher magnification of white box in F showing GFP1 fibers innervating a conducting airway and projecting into the alveolar region (green arrow, in G) and CGRP1 fibers innervating a conducting airway but not projecting into the alveolar region (red arrow, in H). I-K, Quantification the % of conducting airways with fibers (I), the % of conducting airways with specific subsets has demonstrated their role in regulating respiratory and autonomic responses (Mazzone and Canning, 2002;Mazzone et al., 2005Mazzone et al., , 2009Muroi et al., 2011Muroi et al., , 2013Tränkner et al., 2014;Chang et al., 2015;McAlexander et al., 2015;Nonomura et al., 2017;Baral et al., 2018;Chou et al., 2018). There is surprising heterogeneity in airway nociceptive reflexes (Chou et al., 2008(Chou et al., , 2018Muroi et al., 2013;Hooper et al., 2019), some of which can be explained by the jugular versus nodose paradigm (Taylor-Clark, 2021), but there appears to be further complexity which may depend on anatomically distinct subsets. The current gap in our knowledge of the innervation patterns of specific subsets hinders our understanding of how pathologic insults such as inflammation, infection or edema trigger reflexes that can contribute to disease morbidity.
Using Cre-mediated reporter expression to map genetically-defined sensory subsets has advantages over IHC, which may be hampered by the absence or limited expression of specific markers and poor antibody selectivity (or lack of validation). Here, cell-specific expression of GFP or tdTomato was robustly driven by the endogenous ROSA26 gene or by CAG promoters in AAV, thus providing high signal-to-noise ratios for the detection of even the thinnest fibers. Nevertheless, there are some caveats to our approach. First, tdTomato expression driven by the ROSA26 gene reflects Cre expression at any point in the neuron's lineage, not necessarily current Cre expression in the adult neuron (Cavanaugh et al., 2011a;Kim et al., 2020b). Whereas AAV-mediated reporter expression only occurs if Cre is actively expressed. Second, injection/instillation with AAV vectors is unlikely to transfect all sensory neurons/fibers within that locale. For example, lung instillation of rAAV2-flex-tdTomato unexpectedly failed to label large blood vessels, although these structures likely had Cre1 fibers.
Here, we opted to image 80-mm-thick lung sections. Thus, in many cases reporter-expressing fibers were found to enter in/out of the physical plane making comprehensive structural analysis impossible. As such, it is likely that the projecting fiber length is an underestimate. Nevertheless, confirmed terminations for specific fibers were identifiable if they occurred between the upper and lower limits of the z-stacked image.
Pirt is a TRP channel regulator that is expressed in almost all sensory neurons in the vagal ganglia and DRG but not in other cell types (Patel et al., 2011;Patil et al., 2019;Kim et al., 2020b). Pirt1 fibers innervated almost all conducting airways but only ;30% of blood vessels, indicating that afferents project to specific structures. The vagal-specific Pirt1 innervation was almost as widespread as the overall Pirt1 innervation. The conducting airways were also densely innervated by TRPV11, 5HT 3 1, and Tac11 fibers. TRPV1 is expressed by almost all mammalian C fibers projected from the nodose and jugular ganglia to the lungs (Ho et al., 2001;Undem et al., 2004;Hooper et al., 2016), and their activation evokes defensive reflexes (Coleridge and Coleridge, 1984;Mazzone and Undem, 2016;Taylor-Clark, 2021). In the guinea pig, TRPV11 fibers have been identified by IHC innervating the conducting airways (Watanabe et al., 2005(Watanabe et al., , 2006. In the mouse, however, TRPV1 expression is restricted to C fibers that conduct action potentials ,0.75 m/s (Kollarik et al., 2003). Mouse TRPV1-expressing vagal afferents are critical for the airway hyperreactivity associated with allergic asthma and bacterial clearance in a model of pneumonia (Tränkner et al., 2014;Baral et al., 2018). Our data indicate that nociceptive vagal fibers innervate the vast majority of conducting airways, intercalating with the smooth muscle layer, and coming into close proximity with the epithelium. Although we did not co-label the vagal TRPV11 fibers with specific nodose or jugular markers, we also observed vagal Tac11 fibers and 5HT 3 1 fibers innervating and terminating within the same smooth muscle layer of the conducting airways as TRPV11 fibers. Tac1, the gene that encodes preprotachykinin (precursor to the tachykinin neuropeptide substance P), is expressed in jugular TRPV11 neurons but not nodose TRPV11 neurons (Undem et al., 2004;Nassenstein et al., 2010;Kim et al., 2020b). Whereas 5HT 3 is expressed exclusively in nodose neurons (Chuaychoo et al., 2005;Kim et al., 2020b). Thus, our data suggest that both nodose C fibers and jugular C fibers innervate the conducting airways through to the smallest airways. This is consistent with receptive field mapping of lung C fibers to punctate stimulation (Undem et al., 2004). The density of 5HT 3 1 fibers in the large and medium diameter airways appeared greater than for the vagal TRPV11 or vagal Tac11 innervation. Indeed, the density of 5HT 3 1 innervation in these airways resembled the density of Pirt1 innervation. Previously, tdTomato in the 5HT 3 -Ai9 was observed in both TRPV11 and TRPV1-negative nodose neurons (Kim et al., 2020b), and thus much of the 5HT 3 1 and Pirt1 fiber density is likely because of the presence of TRPV1-negative fibers. The extent of the TRPV1-negative innervation was surprising given that TRPV1-negative A fibers represent only ;10-20% of lung afferents (Coleridge and Coleridge, 1984; continued fibers that project out to the alveolar region (J), and the % of blood vessels with fibers (K) for CGRP1 fibers and TRPV11 (GFP1) fibers. Data in I-K are derived from three TRPV1 Cre mice with AAV9-flex-GFP vagal injections (186 conducting airways, 195 blood vessels). L, Correlation of the number of CGRP1 cells within an NEB and the diameter of the conducting airway. M, Percentage of NEBs innervated by Pirt1, CGRP1, and TRPV11 (GFP1) fibers. N, Median (with interquartile range) airway diameter of conducting airways at locations of individual NEBs, grouped by innervation (or lack of) by Pirt1, CGRP1, and TRPV11 (GFP1) fibers. O, Median (with interquartile range) number of CGRP1 cells within the NEBs, grouped by innervation (or lack of) by Pirt1, CGRP1, and TRPV11 (GFP1) fibers. Data in L-O are derived from three TRPV1 Cre mice with AAV9-flex-GFP vagal injections and three Pirt-Ai9 mice (in total, 112 NEBs). * denotes significant difference between NEBs with and without innervation (Mann-Whitney two-tailed U test, p , 0.05, see text for precise values). In some images, lumens are denoted by "L." Scale bars denote 100 mm (A, B, F), 50 mm (D, G, H), 30 mm (C), or 20 mm (E). Undem, 2016), suggesting that the arborization of these afferents or C fibers conducting .0.75 m/s is extensive compared with TRPV11 fibers. Here, we noted that multiple TRPV11 fibers innervated the same large conducting airway, but each fiber had a relatively simple arbor.
While all studied vagal subsets innervated the conducting airways, only Pirt1, TRPV11, and 5HT 3 1 fibers projected into the alveolar region. We therefore conclude that only nodose fibers, many (if not all) being TRPV11, innervate these regions. Such projections only occurred from conducting airways (not blood vessels) that were ,375 mm in diameter. We found no evidence that neural crest-derived sensory neurons (jugular or DRG) project into the alveolar region. These data are consistent with a recent report using vagal injections of AAV-flexreporter vectors in TRPV1 Cre and Tac1 Cre mice (Su et al., 2021), although substance P1 fibers were occasionally observed innervating the lung parenchyma in the guinea pig (Kummer et al., 1992).
Sensory innervation of blood vessels was limited compared with conducting airways, consistent with other reports in mice (Su et al., 2021), although another study in guinea pig suggests widespread vessel innervation by substance P1 fibers (Haberberger et al., 1997). Large blood vessels were much more likely to be innervated by Pirt1 vagal fibers than smaller vessels. Although there was some 5HT 3 1 innervation of blood vessels (indicating nodose fibers), the vagal Pirt1 innervation pattern was replicated by vagal Tac11 innervation (likely jugular fibers). Although we did not assess Tac11 and TRPV11 expression in the same tissue, fewer large blood vessels were innervated by vagal TRPV11 fibers (36%) compared with vagal Tac11 fibers (64%), suggesting the existence of a vagal Tac11/TRPV1-negative population. In rat and guinea pig studies, capsaicin toxicity has been shown to eliminate both substance P immunoreactivity and vagal stimulation-evoked tachykinergic-mediated (i.e., neurogenic) bronchospasm and vascular leakage, suggesting that substance P is solely expressed in TRPV11 afferents (Lundberg et al., 1983(Lundberg et al., , 1984Undem, 1990, 1992;Baluk et al., 1992). However, not all mouse lung C fibers express TRPV1 (Kollarik et al., 2003), and others have identified Tac11/TRPV1-negative populations within the mouse vagal ganglia (Surdenikova et al., 2012;Kupari et al., 2019;Kim et al., 2020b). The identity and function of these putative vagal Tac11/TRPV1-negative fibers is unclear.
Using latex beads, we identified the pulmonary arteries and pulmonary veins in Pirt-Ai9 mice. Pirt1 fiber innervation was more common for both larger arteries and larger veins. Indeed Pirt1 fibers innervated almost all medium and large diameter veins but none of the smaller veins. Although we did not perform the latex labeling in the other strains, it is likely that much of the sensory innervation of large veins is Tac11 (but not necessarily TRPV11).
Importantly, sectioning or ablation of vagal afferents does not eliminate afferent fibers within the lung (Lundberg et al., 1983;Baluk et al., 1992). Conventional retrograde tracing (e.g., DiI, fast blue) in guinea pig (Kummer et al., 1992;Oh et al., 2006), rat (Groth et al., 2006), and mouse (Dinh et al., 2004;Nassenstein et al., 2010) indicate that DRG afferents also project to the lungs, as do a few neurons within the esophageal myenteric plexus (Fischer et al., 1998). Using AAV injected into the DRG, we demonstrate direct evidence that DRG Pirt1 fibers innervate conducting airways of all sizes, but such innervation is much more common in larger airways. DRG Pirt1 fibers also innervated blood vessels, but this was largely restricted to those .376 mm in diameter. The overwhelming predominance of vagal versus DRG innervation observed here is consistent with retrograde tracing in multiple species (Kummer et al., 1992;Dinh et al., 2004;Nassenstein et al., 2010;McGovern et al., 2012). Nonetheless, the density of DRG innervation of large airways was particularly striking. A total of 70% of large airways and 10% of medium airways were also innervated by DRG Tac11 fibers, but the DRG TRPV11 innervation was very sparse. Similarly, more blood vessels were innervated by DRG Tac11 fibers compared with DRG TRPV11 fibers. Unfortunately, there is a lack of consensus regarding protein expression in DRG lung afferents. TRPV1 appears widespread in guinea pig and rat DRG lung afferents (Groth et al., 2006;Oh et al., 2006), but TRPV1 expression has been reported in ;10% of mouse DRG lung afferents by IHC (Dinh et al., 2004) and ;34% by single neuron RT-PCR (Nassenstein et al., 2010). Our data suggest there is a population of DRG Tac11/ TRPV1-negative fibers innervating the mouse lung, and this is consistent with the innervation of other visceral organs (Surdenikova et al., 2012;Hockley et al., 2019). The contribution of fibers projected from esophageal myenteric plexus neurons (Fischer et al., 1998) to the fibers labeled in the Pirt-Ai9 or Tac1-Ai9 lung is currently unknown.
Interestingly, DRG innervation of blood vessels by Pirt1 or Tac11 fibers was more likely than vagal innervation to penetrate through the muscle layers to come into close proximity to the endothelium. Instead, vagal Pirt1 or vagal Tac11 fibers seemed to innervate the outer muscle layers. The vascular smooth muscle layer of some intrapulmonary veins is surrounded by a discontinuous coat of cardiomyocytes (Mueller-Hoecker et al., 2008). It is currently unresolved whether such cardiomyocytes receive differential sensory innervation compared with the vascular smooth muscle.
Little is known of the physiological role of DRG afferents innervating the mouse lungs. Activation of TRPV11 DRG afferents in guinea pigs and rats evokes sympathetic reflexes to the lung and heart (Oh et al., 2006;Shanks et al., 2018). Inhibition of capsaicin-evoked sympathetic reflexes then reveals a minor DRG-mediated neurogenic bronchospasm (Saria et al., 1985). As mentioned above, most DRG lung afferents do not express TRPV1 in the mouse, thus it is not clear whether they too would evoke similar reflexes.
CGRP is another sensory neuropeptide that is preferentially expressed in jugular neurons compared with nodose neurons (Springall et al., 1987;Helke and Hill, 1988;Zhuo et al., 1997). Similar to Tac1, CGRP1 fibers innervated almost all conducting airways and some blood vessels. However, unlike Tac11 fibers, some CGRP1 fibers projected into the alveolar region. All of these CGRP1 fibers also expressed TRPV1. We should note, however, that the majority of TRPV11 fibers that projected to the alveolar regions lacked CGRP. Given that only nodose fibers projected into the alveolar region, we conclude that some of these nodose TRPV11 fibers express CGRP but not substance P, consistent with reports that CGRP does not match Tac1 expression in mice (Wang et al., 2017;Kupari et al., 2019). However, this conclusion conflicts with tracing studies that suggest ;1% of airway-specific nodose neurons express CGRP in the guinea pig and rat (Springall et al., 1987;Kummer et al., 1992;Undem et al., 2004).
CGRP is also a marker of specialized epithelial cells, whose clusters are termed NEBs. NEBs occur in multiple mammalian species and are innervated by multiple sensory nerve subtypes (Brouns et al., 2003(Brouns et al., , 2009Adriaensen et al., 2015). NEBs are thought to act as part of the sensory system, although their precise role has been debated for decades (Adriaensen et al., 2015;Mazzone and Undem, 2016;Sui et al., 2018). Overall, we found ;74% of NEBs were innervated, similar to previous studies in mice (Chang et al., 2015) and rats (Larson et al., 2003). Consistent with previous reports, we found NEBs were innervated by at least two types of (Pirt1) sensory fibers: CGRP1 and CGRP-negative fibers (Brouns et al., 2003(Brouns et al., , 2009). These CGRP-negative NEB-innervating afferents are likely myelinated fibers that express the markers P2ry1 (Chang et al., 2015) or Calbindin 1 (Su et al., 2021). Because of their reported (Brouns et al., 2003(Brouns et al., , 2009) lack of expression of the nodose marker P2X 2 (Kim et al., 2020a), CGRP1 fibers innervating NEBs were originally thought to be projected from DRG neurons, but our data suggest that many are TRPV11 fibers projected from the vagal ganglia. As such, we suggest many CGRP1 fibers innervating NEBs are from jugular neurons.
Lung instillation of rAAV demonstrated that both Pirt1 and TRPV11 airway afferents have central projections to the nTS and the area postrema. Importantly, no airway afferents projected to the Pa5. This is consistent with other reports using rAAV lung instillation in mice (Kim et al., 2020b;Su et al., 2021), despite observations that vagal injection of AAV9 causes reporter expression in jugular afferents projecting to the nTS and the Pa5 (Kim et al., 2020b;Su et al., 2021). Furthermore, airway instillation of herpes simplex viral vectors in guinea pigs and rats have traced airway jugular (but not nodose) afferent projections to the Pa5 (McGovern et al., 2012Driessen et al., 2015) and functional imaging has demonstrated evidence of this pathway in humans (Farrell et al., 2020). It is possible that only jugular afferents innervating organs other than the lungs project to the Pa5 in mice. Nevertheless, it has been postulated that the AAV instillation in mice fails to transfect the trachea and large conducting airways where the majority of jugular afferents are presumed to be primarily located (Su et al., 2021). But here we provide evidence that Tac11 afferents (likely jugular) are found in great abundance throughout the conducting airways, and it is probable that even a low number of reporter1 terminals would be detectable in the Pa5 using confocal microscopy. Alternatively, it is possible that the lack of airway-Pa5 circuitry in the mouse is because of AAV tropism. Details are presently lacking for the airway-Pa5 circuitry, but evidence suggests that tropism is a factor for AAV-mediated labeling of DRG lung afferent innervation. We have found that DRG injection of Pirt Cre with AAV9flex-tdTomato revealed a substantial tdTomato1 DRG innervation of large conducting airways. Nevertheless, lung Figure 12. Overview of the vagal and DRG sensory innervation of the mouse lung. Schematic identifies three lung structures [conducting airways (brown), blood vessels (blue), and alveoli (gray)] and their innervation by vagal Pirt1, TRPV11, Tac11 (predominantly jugular only) and 5HT31 (nodose only) fibers (green) and DRG Pirt1, TRPV11, Tac11 fibers (red). Airways and blood vessels are subcategorized into large (L, .376 mm in diameter), medium (M, 176-375 mm in diameter) and small (S, 0-175 mm in diameter) groups. Innervation is semi-quantified for specific structures: 1 denotes common occurrence, (1) denotes rare occurrence, -denotes absence, and n.d. denotes not determined.
instillation of Pirt Cre with rAAV2-flex-tdTomato (which induced robust tdTomato expression in vagal neurons innervating the large airways) traced no fibers back to the DRG. This is consistent with other rAAV2 studies in the mouse lung that failed to trace to DRG neurons (Su et al., 2021), despite the fact that conventional retrograde tracers such as fast blue or DiI identify many lung-labeled DRG afferents (Kummer et al., 1992;Dinh et al., 2004;Oh et al., 2006;Nassenstein et al., 2010). AAV tropism for afferent subsets may be dependent not only on the intrinsic properties of the vector and the neuron (Mason et al., 2010;Jacques et al., 2012;Kudo et al., 2021) but also perhaps on the properties of the afferent terminal within the specific lung tissue. Given the widespread use of AAVs, more work is needed to explore afferent subsetspecific tropism in situ.
In summary, our data indicate that sensory innervation of structures within the lung is complex (Fig. 12). We have identified substantial differences in the innervation patterns of nodose and jugular afferents, particularly for fibers projecting into the alveolar region. We have also identified specific innervation patterns for the arterial and venous vessels in the pulmonary circulation. Lastly, we have identified a robust DRG innervation of large conducting airways and blood vessels, which partially expresses Tac1, but lacks TRPV1 expression. As discussed above, the afferent innervation of the mouse lung is somewhat different to that of other mammals such as the guinea pigin particular the contribution by TRPV1-negative C fibers. Mice also lack a major nodose Ad fiber subset that innervates the trachea (Hennel et al., 2018;Kim et al., 2020a), and mice lack a cough reflex to canonical tussive stimuli (Plevkova et al., 2021) and have limited neurogenic inflammatory responses (Baluk et al., 1999). Nevertheless, the genetic tools available for the mouse provide anatomically relevant information regarding distinct nociceptive and non-nociceptive subsets whose activation evokes specific respiratory and cardiovascular reflexes (Coleridge and Coleridge, 1984;Mazzone and Undem, 2016;Taylor-Clark, 2021).
Here, we have focused on the nodose-jugular paradigm for identifying distinct subsets, but there remains the possibility that other gene expression patterns may govern anatomically distinct subsets currently grouped together, e.g., nodose TRPV11 fibers innervate the conducting airways, the alveolar region and some blood vesselsare these the same functional fiber type or three distinct subtypes? After all, exogenous and endogenous stimuli (including physical stretch, pollutants, irritants, inflammatory mediators, autacoids, and edema) would be expected to differ greatly across these regions. In addition, the expression of many sensory nerve markers, including Tac1 and TRPV1, and the lung fiber innervation patterns themselves are plastic and can be modulated by inflammation and infection (Hoyle et al., 1998;Hunter et al., 2000;Carr and Undem, 2001;Zhang et al., 2008;Lieu et al., 2012). More work is needed to determine the role of specific vagal and DRG subsets innervating distinct anatomic sites in evoking reflex responses in health and disease.