Bitter taste signaling in tracheal epithelial brush cells elicits innate immune responses to bacterial infection

Constant exposure of the airways to inhaled pathogens requires efficient early immune responses protecting against infections. How bacteria on the epithelial surface are detected and first-line protective mechanisms are initiated are not well understood. We have recently shown that tracheal brush cells (BCs) express functional taste receptors. Here we report that bitter taste signaling in murine BCs induces neurogenic inflammation. We demonstrate that BC signaling stimulates adjacent sensory nerve endings in the trachea to release the neuropeptides CGRP and substance P that mediate plasma extravasation, neutrophil recruitment, and diapedesis. Moreover, we show that bitter tasting quorum-sensing molecules from Pseudomonas aeruginosa activate tracheal BCs. BC signaling depends on the key taste transduction gene Trpm5, triggers secretion of immune mediators, among them the most abundant member of the complement system, and is needed to combat P. aeruginosa infections. Our data provide functional insight into first-line defense mechanisms against bacterial infections of the lung.


Intratracheal administration of bitter substances and vascular permeability in the airways
Animals were anesthetized via intraperitoneal injection with 10% urethane (Sigma-Aldrich) or ketamine (85 mg/kg) / xylazine (15 mg/kg) (both obtained from Serumwerk Bernburg AG, Bernburg, Germany). Twenty minutes after the onset of anesthesia, 200 µl Evans blue (20 mg/kg) was injected into the systemic circulation through the retro-orbital venous sinus. A cricothyrotomy surgical approach was used for intratracheal administration of various substances to the lower airways. An incision of the skin was made in the ventromedial cervical region, and the salivary glands were positioned laterally to the trachea. The right infrahyoidal muscles were moved gently to the side of the trachea. Then, the median cricothyroid ligament of the larynx was separated by a small incision and a small breathing cannula was inserted into the tracheal lumen. 4 µl of vehicle or the following substances were administered into the trachea via the cannula: denatonium benzoate (1, 10 or 20 mM; Sigma-Aldrich), 3-Oxo-C12homoserine lactone (3-Oxo-C12-HSL, 1 or 10 mM; Sigma-Aldrich), Pseudomonas quinolone signal (PQS, 1, 10 or 50 µM), as well as supernatants from three different strains of Pseudomonas aeruginosa (NH57388A, PA103 or D8A6) (D8A6 was kindly provided by Lutz Wiehlmann, Clinic for Pediatric Pneumology, Allerology and Neonatology, Hannover Medical School) or two different strains of Streptococcus pneumoniae (D39 or PN36). A detailed description on the preparation of the bacterial supernatants can be found below. Animals were sacrificed 30 min after treatment using isoflurane (Abbott, Wiesbaden, Germany) and transcardially perfused with 4% paraformaldehyde (PFA). Tracheal segments (1-5 cartilage ring of the trachea, 6-11 and 12-bifurcation) were dissected, postfixed in 4% PFA, washed, incubated with 18% sucrose and shock-frozen in liquid nitrogen.
Trpa1-tauGFP-DTR mice were injected intraperitoneally with 20 ng/kg diphtheria toxin (DT) on three consecutive days. Intratracheal administration of denatonium or vehicle was performed as described above on the seventh day after the last DT injection. The NK1 receptor antagonist CP96345 (2.5 µg/g mouse body weight, Tocris) was administered intraperitoneally 30 min and the CGRP receptor blocker CGRP8-37 (800 ng, GenScript Biotech, Leiden, The Netherlands) 24 h and 2 h before the intratracheal stimulation with denatonium. These doses of CP96345 and CGRP8-37 were previously proven to be effective (CGRP: (7), CP: (8)(9)(10)). The combination of mecamylamine and atropine (1 mM for each substance) was applied intratracheally 5 min prior to denatonium application. 30 min after intratracheal instillation of the substances (denatonium, QSM or supernatants), the animals were sacrificed using isoflurane and perfused transcardially with 4% PFA. Tracheal segments (1-5 cartilage rings of the trachea, 6-11 and 12-bifurcation) were dissected, post-fixed in 4% PFA for 1 h and washed for 2 h with 10 mM phosphate buffer.
Finally, the tissue was incubated in a solution of 18% sucrose/10 mM phosphate buffer for cryoprotection for 4 h and shock-frozen in liquid nitrogen.

Cryosectioning and Evans blue extravasation
Evans blue fluorescence was evaluated on 10 µm-thick serial cryosections of the middle part of the trachea (segment 6-11) using a Texas Red filter (585/29 nm excitation; 624/40 nm emission) on an epifluorescence microscope (Axioplan 2 imaging, Zeiss). The distance between two sections was 100 µm and altogether 5 sections per animal (n = 3-5) and condition were included in the evaluation. Tracheal ring images were acquired and Evans blue intensity of the experimental groups was recorded with the same exposure time in each total tracheal ring for all groups. The fluorescence of each ring was then quantified using the ImageJ software by measuring the area of the fluorescence of a composite picture of the tracheal ring.

Immunohistochemistry
For immunofluorescent staining of tracheal and lung slices of Trpm5-tauGFP, Trpm5 -/-, ChAT-eGFP or C57Bl6/J mice the tissue samples were blocked for 1 h and then incubated over night with primary antisera (single or in combinations) against Trpm5 (11)(12)(13) Table 3). Sections were then incubated for 1 h with the respective secondary antisera coupled to either Cy3 or Cy5, washed one time with PBS and incubated for 10 min with DAPI followed by three washing steps in PBS. The specificity of GNAT3, DCLK1, CGRP and SP antibodies was tested by pre-absorption of the antibodies with the corresponding peptides. Trpm5 antibody specificity was verified on Trpm5 -/mouse tissue (Supplemental Figure 16 and 17). Images from flagellin stained lung sections obtained from infection experiments with wt (n=9) and Trpm5 -/mice (n=8) were acquired at the same image settings using an Axio Scan.Z1 slide scanner (Zeiss). The mean fluorescence intensity in left lung lobe coronal sections was analyzed by the ZEN3.0 software (Zeiss). For whole-mount stainings employing CD31, CGRP, and SP antibodies, tracheae were processed as described previously (14).
For neutrophil staining, tracheal sections were incubated for 1 h at room temperature with 0.1% bovine serum albumin, 10% horse serum, and 0.2% Tween-20 to block non-specific protein binding sites followed by an incubation step in anti-Ly6G-FITC antiserum (Invitrogen, Thermo Fisher Scientific, 11-5931-85, RRID: AB_465315) at room temperature. Numbers of infiltrated neutrophils in tracheal ring sections of denatonium treated and controls as well as from animals   treated with 3-oxo-C12-HSL, PQS, NH57388A, PA103, D8A6, D39 or PN36 supernatants were counted manually using an epifluorescence microscope (Supplemental Figure 3B). Gramstaining of lung cryo-sections was conducted with a commercial kit according to the manufacturer's protocol (Carl Roth, Karlsruhe, Germany). were then incubated for 1h with the respective secondary antisera, counter-stained with DAPI and mounted with a cover-slip.

Confocal Laser Scanning Microscopy Analysis
Z-stacks of either vehicle or 1 mM Denatonium treated tracheal whole-mount preparations were taken with a Leica TCS SP5 Confocal Laser Scanning Microscope (Leica Microsystems, Wetzlar, Germany). Each stack was scanned with an XY-resolution of 512 × 512 and a size of 387.5 x 387.5 x 30 µm, respectively, starting from the luminal side of the tracheal epithelium.
Analysis from one to four image stacks from randomly chosen regions of interest from the membranous part of the tracheae were performed using Imaris 9.9.0 (Oxford Instruments) or ImageJ (NIH) software.

Quantification of brush cell-nerve fiber contact sites and volume analysis of tracheal sensory nerve fibers
For the quantification of brush cellnerve fiber contact sites image stacks containing at least 15 brush cells were examined with ImageJ. Within a stack all brush cells were manually examined whether they have a contact site with either CGRP+ or SP+ sensory nerve fibers. The number of brush cells with contacts to sensory nerve fibers was then normalized to the total number of brush cells in each stack. Volumes of CGRP+-and SP+-nerve fibers were quantified from the 3D reconstructed surfaces. Each surface volume was normalized to its corresponding stack volume.

Calcium imaging experiments
Trpm5-GCaMP3 mice were killed by inhalation of an overdose of isoflurane followed by aortic exsanguination and the trachea was dissected and cut into three pieces and opened longitudinally. Tracheal pieces were placed into the recording chamber (Warner Instruments RC-27LD, Hamden, USA) coated with ELASTOSIL ® RT 601 A (Wacker Silicones, Burghausen, Germany) and continuously perfused with experimental solutions pre-heated to 33°C. A confocal imaging system microscope (Zeiss LSM710, Jena, Germany) with a water immersion objective (W Plan Apochromat 20x/1.0 DIC VIS-IR, Zeiss) and an argon laser (488 nm) was used to excite and an emission filter at 493-598 nm to collect fluorescent signals from tracheal cells expressing GCaMP3 with a 2 Hz frame rate. Solutions were based on standard Tyrode III solution consisting of (in mM): 130 NaCl, 10 HEPES, 10 Glucose, 5 KCl, 1 MgCl2,   8 CaCl2, 10 sodium pyruvate, 5 NaHCO3 in which 1 mM, 10 mM or 20 mM denatonium or 1 mM C12-Oxo-HSL were freshly dissolved before the experiment. Denatonium was applied for 5 min followed by a 5 min washing step with Tyrode solution. C12-Oxo-HSL was applied for 3 min followed by a washing step with Tyrode solution and application of 100 µM ATP.
Further analysis was performed using ImageJ and Prism (GraphPad).
For calcium imaging experiments of isolated tracheal brush cells, tracheae were treated and measurements were performed as previously described (12, 15).

SP and CGRP measurements
Mice were euthanized with an overdose of isoflurane followed by aortic exsanguination.
Tracheae were explanted and collected in 200 µl buffer consisting of (in mM): 136 NaCl, 5.6 KCl, 10.7 Glucose, 10 Hepes, 1 MgCl2, 2.2 CaCl2. Tracheae were stimulated with 1 mM or 20 mM denatonium for 5 min at 37°C and then kept on ice. Tissues were homogenized using a tissue ruptor (Qiagen, Hilden, Germany) and centrifuged at 1500 g for 5 min at room temperature. The clear supernatant was further processed using commercial Substance P and CGRP ELISA kits (R&D systems, Biotechne, Abingdon, UK and Bertin Pharma, Montigny-le-Bretonneux, France, respectively) according to the manufacturers' protocols. Immunoassay plates were then assessed photometrically with a microplate reader (Bio-Rad, Laboratories GmbH, Feldkirchen, Germany) and results were presented in pg/ml.

Intravital imaging
Mice were anaesthetized with urethane (1.5 g/kg i.p., Sigma-Aldrich), placed on a heated microscope stage (37°C) and the trachea was exposed as described previously (16 Supernatant was collected after centrifugation and stored at -80°C. and incubated for 30 min at room temperature with shaking followed by three more washing steps. Then wells were incubated with 50 µl streptavidin-PE for 10 min at RT with shaking and again washed three times. Finally, 125 µl resuspension buffer was added and the plate was shaken for 30 sec, followed by data acquisition.

Mass Spectrometry Analysis of Secretome and Proteome
To investigate the secretome of denatonium-stimulated tracheae, mice were euthanized using

Quantitative real-time PCR analysis
Quantitative real-time PCR was performed as previously described (17 an incompletely elucidated function. The ORF 5PG21 in the genomic island PAGI-5 is inactivated in this mutant (23). This ORF is not part of the P. aeruginosa core genome. PAGI-5 is characterized by a set of several dozen conserved genes ORFs which are also annotated as type IV secretion system-like components. The conserved ORFs are supposed to be involved in mobilisation, transfer and genomic integration of the genomic island with regard to their descendance from mobile DNA elements. The question, how the knock-out of such a gene can affect the quorum sensing system and cause the described phenotypes, is still not answered, but an surprising influence of this D8A6/5PG21 gene from the accessory genome on this core regulatory system can be postulated. The mutation of the 5PG21-ORF as the actual cause of the observed effects was proven by subsequent generation of target mutants.

Characterization of the Trpa1-DTR mouse model
In order to characterize the Trpa1 + neurons, we performed Ca 2+  Moreover, CGRP + /SP + sensory nerve fibers in the trachea were almost completely depleted in Trpa1-tauGFP-DTR mice after DT treatment (Supplemental Figure 9). Taken together, this shows that the Trpa1-DTR/DT mouse model is suitable for studying sensory nerve-mediated responses from the respiratory system.  corresponding to the EC 50 ) is significantly reduced after DT treatment of Trpa1-DTR mice compared to Trpa1 +/+ mice. J) Representative images of DRG tissue sections from Trpa1 +/+ (wt) and a Trpa1-DTR mouse treated with DT stained with a TRPV1 antibody (orange). Arrows indicate TRPV1 + neurons. K) TRPV1 + neurons were reduced from 43% in wt (Trpa1 +/+ ) mice to approximately 10% in Trpa1-DTR mice after treatment with DT. C, F-I, K) Data are shown as single values and mean ± SEM and were analyzed with one-way ANOVA followed by Bonferroni's multiple comparison (C) or the unpaired Student's t-test (F-I, K).  Figure 13: Evans blue extravasation and neutrophil recruitment in response to bacterial bacterial culture supernatants. A) Application of the Pseudomonas aeruginosa Pa103 supernatant increased Evans blue intensity significantly in wild-type mice. Supernatants of the Pseudomonas aeruginosa strain D8A6, a quorum-sensing molecule-deficient strain, significantly increased Evans blue intensity, but Evans blue intensity was significantly less than in the Pa103 strain. B) Pa103 and D8A6 supernatants did not influence intra-or extraepithelial neutrophil numbers in wild-type mice. C) Application of supernatants of cultures of two Streptococcus pneumoniae strains (D39, PN36) significantly increased Evans blue intensity in wild-type and Trpm5 -/mice. The Evans blue intensity induced by PN36 and D39 was significantly reduced in Trpm5 -/mice compared to wild-type controls. The vehicle controls were not significantly different from each other. D) D39 and PN36 supernatants did not influence intra-epithelial neutrophil numbers in wild-type or Trpm5 -/mice. D39 supernatants increased extraepithelial neutrophils in wild-type mice, but not in Trpm5 -/mice. Supernatants of PN36 did not influence extraepithelial neutrophil numbers in wild-type or Trpm5 -/mice. A-D) Data are presented as single values and mean ± SEM and were analyzed with one-way ANOVA followed by Bonferroni's multiple comparisons. *: p<0.05, **: p<0.01, ***: p<0.001 Conc. [pg/ml]