Background
The airway epithelium is composed of various cell types. Their identities and functions have more recently been defined using sequencing studies of mouse and human airway epithelium [
1‐
4]. Among those are the rare tuft cells, formerly referred to as brush cells in the lower airways and as solitary chemosensory cells in the upper airways [
5]. These cells have over decades merely been described morphologically by the presence of their apical tuft of microvilli [
6]. They are classified as a rare population of cells, given that they constitute approx. 1% of the total epithelial cell population in mice [
7,
8]. Despite their scarcity tuft cells in the airways are chemosensory and execute essential functions as triggers of fundamental innate immune processes as this has become evident more recently [
5]. However, most functional studies on tuft cells have been performed in mouse models, and little is known about their distribution and function in human airways.
One hallmark of these cells is their expression of choline acetyltransferase (ChAT), an enzyme responsible for the synthesis of acetylcholine (ACh) [
1,
9,
10]. Additionally, these cells express members of the bitter taste signalling cascade in mice, which is known to convey detection of bitter substances in taste buds. Activation of tuft cells with bitter substances, such as denatonium or bacterial quorum-sensing molecules with bitter character, leads to an increase in the intracellular Ca
2+ level [
11,
12]. The increased intracellular Ca
2+-level then activates the transient receptor potential melastatin 5 (Trpm5) cation channel, which leads to downstream effects, such as the release of acetylcholine (ACh) from mouse tracheal epithelial tuft cells [
11,
13]. Alternatively, Trpm5-dependent ACh release may be induced by other bacterial products, such as formyl peptides or succinate, independent of bitter taste receptors [
14,
15]. In mice, the released ACh acts in a paracrine manner by inducing protective innate immune responses such as stimulation of mucociliary clearance [
11,
14] as well as evoking protective neurogenic inflammation [
12,
16], which proved to be essential to overcome bacterial infections with
Pseudomonas aeruginosa [
12].
However, in humans, tuft cells are less well characterised, since most studies, in particular functional ones, have been conducted in mice. Shah et al. reported an increase in ciliary beat frequency (CBF) after stimulation of bitter taste receptors with the bitter substance denatonium in primary airway epithelial cell cultures from human trachea and bronchi [
17]. In sinonasal primary epithelial cell cultures
P. aeruginosa quorum-sensing molecules were shown to stimulate CBF dependent on the bitter taste signalling component PLC
β2 and on nitric oxide (NO) [
18]. Yet, these effects observed on CBF after stimulation of bitter taste signalling in the lower and upper airways were attributed to bitter taste receptors present in ciliated cells and the role of tuft cells was not addressed in both studies. Moreover, presence of characteristic markers in human airway tuft cells have mostly been described on RNA levels by RNAseq studies and not on protein level [
2‐
4]. Here, we investigated the presence of tuft cells along the lower respiratory tract in human body donor specimens. Additionally, we characterised the cells and assessed the occurrence of gene transcripts in humans from molecules involved in tuft cell functions in mice. Moreover, we delineated the impact of tuft cells on mucociliary transport in native human tracheal epithelial preparations.
Methods
Human and mouse sample preparation
All body donors had previously given their consent and all procedures were approved by the Ethics Committee of Saarland Medical Association. For histological evaluation, samples were prepared from lungs dissected from body donors within 24 h after death. Prior to their death, all body donors had signed an agreement to bequeath their bodies to the Institute of Anatomy and Cell Biology after passing away. Only body donors who died of causes of death unrelated to airway diseases such as pneumonia or COPD were included in our study. Tissue specimens from different regions (trachea, main bronchi, lobar bronchi, segmental bronchi, bronchioles as well as alveolar regions) were collected and further processed for fixation (see below).
Tracheal samples were collected for particle transport speed (PTS) measurements, nucleic acid and protein extraction as follows. The upper 3–4 cartilage rings of the trachea located distal the cricoid cartilage were dissected and immediately transferred into cold media (MEM: minimal essential medium (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) supplemented with L-glutamine (Gibco) and penicillin/streptomycin (Gibco). The epithelial layer was carefully separated from the conductive tissue, cut into small pieces and incubated at 37 °C and 5% CO2 in MEM until PTS measurements. Human tissue samples for nucleic acid (RNA, DNA) and protein extraction were collected within 18 h after death in RNeasy sample buffer, DNA lysis buffer, and RIPA-buffer, respectively. For this purpose, the epithelial layer was separated from the remaining tracheal wall by a rubber cell scraper (Sarstedt, Nümbrecht, Germany).
Experiments on mice were performed using wild type or
ChAT-eGFP [
19] mice older than 12 weeks of both sexes. All animal procedures were conducted in accordance with the German guidelines for care and use of laboratory animals. Mouse tracheas were prepared as described previously [
11] and used for RNA extraction or PTS measurements. For RNA extraction and immunoblotting analyses of the epithelium, the epithelium was scraped off with a rubber cell scraper, and collected in respective buffers (see above).
RT-PCR analysis
Human and mouse tissue samples were harvested, and RNA was extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. DNA was isolated by tissue lysis and subsequent nucleic acid precipitation using 2-propanol as has been described previously [
20]. All extracted nucleic acid preparations (i.e., RNA and DNA) were quantified using a NanoDrop One (Thermo Scientific, Waltham, MA, USA). DNaseI treatment of isolated RNA as well as subsequent cDNA synthesis was performed as follows: 1 µg of isolated total RNA, or in case of lower concentrations a maximal volume of 8 µl, were incubated with 1 µl 10x DNaseI Reaction buffer and 1 µl DNaseI (amplification grade, both Thermo Scientific) in a total volume of 10 µl (add DNase- and RNase-free water if applicable) for 15 min at 37 °C. To assure thorough DNA digestion, reactions were further supplemented with another 1 µl of DNaseI and incubated for additional 45 min before enzymatic reaction was terminated by addition of 1 µl 25 mM EDTA solution and incubation at 65 °C for 10 min (all reagents were from Thermo Scientific). In the next step, DNaseI-treated RNA samples were subject to reverse transcription following the manufacturer’s protocol using the SuperScript II reverse transcription kit (Thermo Scientific). Briefly, 10 µl of the DNaseI-treated RNA samples were incubated with 1 µl Oilgo dT
18 Primer and 1 µl dNTPs (10 mM, each; both Thermo Scientific) for 5 min at 65 °C, before adding 4 µl First Strand Buffer (5x) and 2 µl DTT (0.1 M). Samples were incubated for 2 min at 42 °C before the reverse transcription was initiated by adding 1 µl of SuperScript II Reverse Transcriptase. The enzyme was substituted by 1 µl RNase/DNase-free water in control reactions. Reactions were incubated for 50 min at 42 °C and eventually terminated by an incubation at 72 °C for 15 min. All reactions yielded a total volume of 20 µl and were subsequently used for PCR analysis.
Gene transcription was assessed using the SYBR®Green-based, two-primer detection method. Primer pairs for the detection of cDNA of gene transcripts from mice have been published before [
11] and were purchased from (Eurofins Scientific, Luxembourg City, Luxembourg). Primer pairs for the detection of human gene transcripts have been purchased from IDT (Integrated DNA Technologies, Coralville, IO, USA) employing the company’s respective PCR design tools (
https://eu.idtdna.com/pages/tools). A list of the primer sequences and the access numbers of the underlying nucleotide sequence from which they derived can be found in Suppl. Table 1. All PCR reactions were run using the Bio-Rad CFX Connect™ RealTime System with its software (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Sample preparation for PCR was set up as follows: 10 µl iTaq Universal SYBR®Green Supermix (Bio-Rad), 8 µl of nuclease-free water, 1 µl of cDNA (or control) of the previous step (alternatively 1 µl of 5 ng/µl genomic DNA sample) and 1 µl of a gene-specific primer mix. Primer mixes were prepared by the combination of gene-specific forward and reverse oligonucleotides at a final concentration of 10 mM. Reactions of a total volume of 20 µl were loaded on 96-well plates suitable for the CFX Connect™ RealTime System. Reactions were run according to the manufacturer’s protocol, i.e., initial denaturation at 95 °C for 30 s followed by a repetition of 40 cycles at 95 °C for 5 s and 60 °C for 30 s. PCR products were supplemented with OrangeG-colored loading-dye, separated on a 2% agarose gel and documented using a ChemiDoc™ XRS + with Image LabTM Software (Bio-Rad).
Immunoblot analysis
Tissue samples were collected in RIPA-buffer (10 mM sodium phosphate buffer, 40 mM sodium fluoride, 2 mM EDTA, 0.1% sodium dodecyl sulfate, 1% TritonX-100, 0,1% sodium deoxycholate) supplemented with a proteinase inhibitor cocktail (cOmplete, Roche Diagnostics GmbH, Mannheim, Germany) and homogenised using a speed mill. Samples were incubated for 2 h at 4 °C to allow protein solubilization and centrifuged at 4000x g for 10 min at 4 °C. The supernatant was removed and its protein concentration was determined using a BCA-Protein Assay Kit (Thermo Scientific). A total of 40 µg protein per sample was diluted in SDS-sample buffer, loaded and separated by SDS-gel electrophoresis on a 10% acrylamide gel. Subsequently, gels were blotted on a nitrocellulose membrane and blocked in 5% milk powder dissolved in TBS-T buffer (Tris-buffered saline: 8.5 mM Tris-HCl, 1.7 mM Tris, 50 mM NaCl, 0.1% Tween-20) for 1 h. Membranes were further incubated in blocking buffer supplemented with respective primary antibodies at 4 °C overnight. On the following day, membranes were washed three times for 10 min in TBS-T buffer before they were incubated in blocking buffer supplemented with the secondary antibody for 1 h at room temperature. After another round of three washing steps in TBS-T buffer for 10 min at room temperature, membranes were subject to enhanced chemiluminescence reaction using the Supersignal West Pico Plus ECL Kit (Thermo Scientific) according to manufacturer’s recommendation. ECL-reaction was documented using a Bio-Rad geldoc system. The following antibodies have been used: CHAT (host rabbit monoclonal 13H9L16; concentration 1:1000; Merck), POU2F3 (host rabbit polyclonal AV32537; concentration 1:1000; Sigma-Aldrich, St. Louis, MO, USA) and anti-rabbit horseradish peroxidase-coupled secondary antibody (anti-rabbit IgG; concentration, 1:20,000; Sigma-Aldrich, Steinheim, Germany).
Histology and immunofluorescence analysis
Fixation of the tissue and processing for immunofluorescence analysis have been performed as described previously [
12]. Briefly, dissected tissues were subject to fixation in Zamboni fixative, cryo-preserved in Tissue Tek und stored at -20° C until use. Samples for bright field documentation upon classical hematoxylin/eosin staining (HE) were subject to dehydration and paraffine-embedding. Paraffine-embedded samples were sectioned at 10 μm, collected on glass slides, hydrated, processed for HE-staining, dehydrated, mounted in Mowiol (Sigma-Aldrich), and documented by bright field microscopy (Fig.
1b/c). Prior to blocking of unspecific antigens antigen-retrieval was performed with cryosections in a boiling 10 mM sodium citrate buffer (pH 6) for 5 min. The following primary antibodies have been used to detect CHAT (host, goat; concentration, 1:800; Merck Millipore, Temecula, CA, USA), DCLK1 (host, rabbit; concentration, 1:1600; abcam, Cambridge, UK), POU2F3 (host, rabbit; 1:1600; Sigma-Aldrich, St. Louis, MO, USA), PGP9.5 (host, rabbit; concentration 1:800; abcam) and TRPM5 (host rabbit, 1:200; Proteintech, Manchester, UK for human tissues and Trpm5 794 [
11] host, rabbit, 1:800, for mouse tissues) followed by the application of secondary antibodies, Cy3-donkey anti-rabbit (concentration, 1:500; Merck Millipore), Cy3-donkey anti-goat (concentration, 1:400; Merck Millipore), Cy5-donkey anti-rabbit (concentration, 1:250; Jackson ImmunoResearch, Cambridge, UK). All sections were mounted using Mowiol (Sigma-Aldrich) and evaluated using a Zeiss epifluorescence microscope (AxioImager M2 with Axio-Cam 512 color, Zeiss, Oberkochen, Germany) equipped with ZEN software (Zeiss) for documentation.
Measurements of particle transport speed in human tracheal preparations
For PTS experiments, tracheal epithelial preparations were pinned into a Sylgard-coated (Dow Corning GmbH, Wiesbaden, Germany) delta t-dish (Bioptechs, Butler, PA, USA) with the luminal side upwards. The delta t-dish dish contained 1.5 ml pre-heated buffer solution consisting of (in mM): 136 NaCl, 5.6 KCl, 10 Glucose, 10 HEPES, 1 MgCl2, 2.2 CaCl2, pH 7.4. To perform the measurement, the delta t-dish was mounted under an Eclipse 80i microscope (Nikon, Tokio, Japan) equipped with a SMX16E1M camera (Sumix, Oceanside, CA, USA) and 1–3 µl dynabeads (Invitrogen, Thermo Fisher) were added to the dish. Videos were recorded using the Streampix 7 software (Norpix Inc, Montreal, Canada) at every two minutes, starting from minute 0 until minute 29. Baseline was recorded until minute 7 and at minute 8 the first substance was applied. ATP (Sigma-Aldrich) was added in each experiment at minute 24 to assess the viability of the tissue. The following substances were used to investigate their influence on the transport speed of the dynabeads (particles): denatonium (1 mM, Sigma-Aldrich), quinine (100 µM, Sigma-Aldrich) and ATP (100 µM). The following inhibitors were applied: TPPO (100 µM, Sigma-Aldrich), atropine (50 µM, Sigma-Aldrich), mecamylamine (100 µM, Sigma-Aldrich) and L-Name (20 µM, Enzo Life Sciences, Lörrach, Germany). PTS was evaluated by tracking each particle in the videos using the ImageProPremier 9.3 software (Media Cybernetics Inc, Rockville, MD, USA).
Measurements of particle transport speed in mouse trachea
Mouse tracheal PTS measurements were performed as described previously [
11]. Mice were sacrificed by inhalation of an overdose of isoflurane followed by aortic exsanguination and tracheae were immediately dissected and opened longitudinally. The measurement was performed as described above for human PTS measurements. Videos were recorded starting 30 min after exsanguination. The following substances were used: denatonium (1 mM), diphenidol (200 µM, Sigma-Aldrich), chlorpheniramine (300 µM, Sigma-Aldrich), and TPPO (100 µM).
Statistical analyses
PTS measurements were repeated at least with n = 4 samples from a minimum of 3 body donors. Immunohistochemistry was performed on tracheal and bronchial preparations from at least 3 different body donors. To assess statistical differences in the frequency of airway tuft cells, data were subjected to one-way ANOVA followed by Tukey’s post-hoc analyses for multiple comparisons. To assess statistical differences in PTS, the paired Student’s t-test was applied after the data were subjected to a Kolmogorov-Smirnov analysis to test for normal distribution. P values < 0.05 were considered statistically significant. All statistical analyses were performed using the GraphPad Prism 9 software (GraphPad Software, Boston, MA, USA).
Discussion
Tuft cells in the airways have emerged as crucial regulators of innate immune processes in mice [
5]. While tuft cells have also been described in recent RNAseq studies of human airway epithelium [
2‐
4], it remained yet unclear whether human tuft cells also display chemosensory and cholinergic traits (as their mouse homologs) and whether they have similar functions in innate immunity. We here detected tuft cells throughout the airways and delineated a cholinergic function for tuft cells in the regulation of mucociliary clearance, a fundamental innate immune process of the airways.
In human airways, tuft cells have long been described morphologically in the trachea and bronchi [
27,
28]. Under pathophysiological conditions they have further been detected in the alveoli of children [
29,
30]. We identified tuft cells by the expression of their marker protein POU2F3 [
4] throughout the conductive lower airways from the trachea to the bronchioles. They appeared to be absent in the human alveolar regions under physiological conditions. Transcripts of IRAG2, POU2F3, DCLK1, and TRPM5 were present in human tracheal epithelium, and protein expression of POU2F3, DCLK1 and CHAT, established markers for mouse tuft cells, has been confirmed by immunofluorescence analysis in various segments of the lower respiratory tract. These data delineate the presence of tuft cells with cholinergic features in the human lung.
Evidence suggesting the presence of chemosensory cells in human airways solely by means of RNAseq needs to be considered with precaution. In one study conducted with human bronchial primary epithelial cultures tuft cells clustered with neuroendocrine cells [
2], while in another study using human tracheal samples tuft cells, neuroendocrine cells and ionocytes emerged as distinct cell populations. In the latter study, tuft cells expressed POU2F3 [
4] and therefore we consequently used POU2F3 for the identification of tuft cells in our study. Moreover, we found that CHAT expressing cells represented POU2F3 + and DCLK1 + cells given the overlap in the immunolabelling for these proteins. Strikingly, these cells were not identical to neuroendocrine cells, since CHAT and PGP9.5 immunostaining did not overlap. This further strengthens the assumption that tuft cells and neuroendocrine cells represent two distinct cell populations with most probably different functions. Since Chat is a marker for mouse tracheal epithelial cells [
7,
9,
11], and POU2F3 is a marker for human tuft cells [
7,
9,
11] it is likely that the CHAT
+ cells co-expressing POU2F3 in the human tracheal samples are also tuft cells. In addition to the paracrine effects of ACh on ciliated cells, the presence of mAChR1 and 3 transcripts in human tracheal epithelium as well as in mouse tracheal tuft cells opens the possibility for a potential autocrine signalling as this has been previously described for mouse tuft cells [
11,
14]. Since cholinergic neurons are located in the tracheal adventitia, and this was not included in our samples, we can exclude that intraepithelial nerves contribute to the mAChR transcripts detected in our study. In mice, ACh is released from tuft cells in a Trpm5-dependent manner and elicits important paracrine effects such as an increase in mucociliary clearance or an induction of neurogenic inflammation [
11,
12,
14]. In support of this observation, we could here demonstrate that tuft cell stimulation by bitter substances in the human trachea plays a role in stimulating mucociliary clearance and involves cholinergic signalling.
In human nasal airway epithelium, TAS2R are present in chemosensory cells and possibly also in ciliated cells [
18,
31]. In the human tracheal epithelium we were able to detect TAS2R4 transcripts, the orthologue for the mouse tuft cell marker Tas2r108, whereas transcription of TAS2R38, the receptor detected in human nasal ciliated cells [
18], was less explicit. Detection of all potential bitter taste receptor transcripts in our study might be limited by the quality of RNA extracted from the body donors. Previous studies found bitter taste receptors also in ciliated epithelial cells obtained from tracheal and bronchial samples and ciliary beat frequency was increased upon stimulation with the bitter substance denatonium [
17]. Denatonium has been shown to act on several human bitter taste receptors, among them the TAS2R4, an orthologue to the mouse Tas2r108, the hallmark receptor for mouse tracheal tuft cells [
22,
26]. Interestingly, the observed stimulation of mucociliary clearance after application of denatonium to human tracheal epithelium from body donors, was dependent on TRPM5, which we detected on protein level. Expression of Trpm5 has been attributed exclusively to chemosensory cells in mouse airway epithelia [
1,
7,
9‐
11,
32]. The scarcity of TRPM5 staining in our human tracheal samples suggests that TRPM5 expression is limited to a rare epithelial cell type such as tuft cells. In support, TRPM5 was listed as a marker for human airway tuft cells (Suppl. Material in Deprez et al. [
4]). Our data provide evidence for functional TRPM5 in human tracheal epithelium and point towards its role in activation of mucociliary clearance after stimulation of the bitter taste signalling cascade as this has recently been demonstrated in mice [
11]. The underlying mechanism involves NO in addition to ACh, as the denatonium-induced increase in mucociliary clearance was reduced upon inhibition of NOS or antagonising AChRs.
In mouse trachea, the activation of nAChRs and mAChRs is well established as a stimulator of mucociliary clearance, measured as an increase in PTS [
33,
34]. Moreover, we have shown previously, that ACh released from a non-neuronal source in the mouse trachea modulates mucociliary clearance by activating transepithelial ion transport processes via mAChRs and nAChRs [
35‐
38]. Recently, mouse tracheal brush cells have been identified as the source for the non-neuronal cholinergic regulation of transepithelial ion transport [
15]. Here, we observed that the regulation of the mucociliary clearance in the human trachea was also dependent on cholinergic signalling, thus it is likely that this effect was mediated by cholinergic tuft cells that are equipped with the bitter transduction cascade. These findings, together with the presence of CHAT in POU2F3 + cells, suggests that human tuft cells are able to synthesize ACh, which then acts in a paracrine manner to stimulate mucociliary clearance. In human upper airways ACh is a known regulator of mucociliary clearance [
24,
39,
40], yet the source of ACh-release was not addressed in the studies. Besides ACh, a production of NO in the epithelium leads to an increase in mucociliary clearance [
18,
24,
25]. Since this was attributed to ciliated cells [
18], it is tempting to speculate that tuft cell-released ACh acts on ciliated cells in a paracrine manner, thereby stimulating NOS in these cells. Supportive of this hypothesis is a recent finding in mouse tracheas showing that activation of tuft cells leads to release of ACh, which excites neighbouring cells and initiates a Ca
2+ wave through gap junction signalling, reaching also distant ciliated and secretory cells [
15]. Since NOS can be activated by binding of Ca
2+ to calmodulin [
41], tuft cell-released ACh might lead to NO production by an increase of intracellular Ca
2+ in ciliated cells. Taken together, our data support, a tuft cell-dependent stimulation of mucociliary clearance that was mediated by NO and ACh signalling. This is suggestive for a role of human tracheal epithelial cells in the induction of innate immune processes, since an increase in mucociliary clearance is a crucial step in preventing infections by transporting bacteria out of the airways.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.