To generate Tas2r143-CreERT2 transgenic mice expressing CreERT2 under the control of the Tas2r143 promoter, the BAC clone RP23-316O11 (CHORI, CA, USA) from mouse chromosome 6 was used, containing the Tas2r143, Tas2r135, and Tas2r126 genes. The genomic sequence between the start codon of Tas2r143 and the stop codon of Tas2r126 (35.2 kb) on the BAC was replaced by a cassette carrying the CreERT2 cDNA followed by a polyadenylation (pA) signal and a FRT-flanked ampicillin resistance cassette using Red/ET recombination kit (Gene Bridges). Correct targeting was verified by restriction digestion (BaeI, HpaI, and SwaI) and DNA sequencing. After Flp-mediated excision of the ampicillin resistance cassette and linearization (NotI), the recombined BAC was injected into C57BL/6 oocytes. Targeted offspring was genotyped for BAC insertion by genomic PCR. Following primers were used: P1 (forward): 5′-CAG GAG TCA TTG AAC TGG GAG-3′; P2 (reverse): 5′-CAG CAT CCA CAT TCT CCT TTC TGA-3′; P3 (forward): 5′-GCA TCG CAT TGT CTG AGT AGG T-3′; P4 (reverse): 5′-CAG ACA TGA AAG GAA CAG GAC AT-3′; PCR with P1/P2 and P3/P4 amplified a 614 bp and a 418 bp fragment, respectively, representing the correctly inserted CreERT2. Three independently generated Tas2r143-CreERT2 lines were crossed with the Cre-reporter line Rosa26^{flox−mT−stop−flox−mG} (Jackson Lab, Stock 007576) and analyzed, showing similar expression pattern. Nuclear translocation of the CreERT2 fusion protein was induced in animals by intraperitoneal injection of 1 mg tamoxifen (Sigma T5648) per day for 5 consecutive days. Five days after the last tamoxifen injection the animals were analyzed. In the absence of tamoxifen treatment, all Tas2r143-CreERT2 lines did not show any basal recombination. We therefore in some cases used the Rosa26^{flox−mT−stop−flox−mG} mouse line as a control.

Animals were kept on a C57BL/6 background and housed under a 12 h light-dark cycle with free access to food and water and under pathogen-free conditions. The Animal Ethics Committee of the Regierungspräsidium Darmstadt approved all animal procedures (Protocol No. B2/1016, B2/1112).

Tissues and organs were fixed in 4% PFA in PBS at 4°C for 12 h, subsequently incubated in 30% sucrose overnight for cryopreservation and embedded in O.C.T. tissue freezing medium (Sakura®, The Netherlands) and stored at −80°C before sectioning. For immunofluorescence analysis, cryosections (8–10 μm) were fixed in 4% PFA for 5 min or in acetone at −20°C for 10 min, washed three times in PBS, blocked with 5% BSA + 0.1% Triton X-100 in PBS for 30 min at room temperature. The blocked sections were incubated overnight at 4°C with following primary antibodies: rabbit anti-DCLK1 (ab31704, Abcam), goat anti-ChAT (AB144P, Millipore), goat anti-Gα gust (I-20; sc-395, Santa Cruz), rabbit anti-PLCß2 (sc-206, Santa Cruz), rat anti-CD31 (550274, BD Bioscience), rabbit anti-K5 (ab53121, Abcam), rabbit anti-K10 (905401, BioLegend), rabbit anti-K18 (clone SP69, M3690, Spring Bio), biotin anti-αSMA (ab125057, Abcam), Alexa Fluor® 647 anti-mouse CD3 (100209, BioLegend). After primary antibody incubation, sections were washed three times in PBS and incubated in the dark for 1 h at room temperature with appropriate secondary antibodies: goat anti-rabbit IgG Alexa Fluor® 647 (A32733), donkey anti-goat IgG Alexa Fluor® 647 (A21447), goat anti-rat lgG Cy5® (A10525), and streptavidin Alexa Fluor® 647 (S21374) from Life Technologies. Cell nuclei were labeled with DAPI (D3571, Invitrogen). After washing three times in PBS, sections were mounted and images were taken by SP5 confocal microscope (Leica). DAPI was detected at a wavelength of 405 nm, EGFP at 488 nm, Tomato at 561 nm and Alexa Fluor® 647 and Cy5® at 633 nm. Images were analyzed by Leica software (LAS AF Lite).

The Rosa26^{flox−mT−stop−flox−mG} mouse line expresses either EGFP or Tomato in a mutually exclusive fashion, and both fluorescent proteins are present in the cell in a membrane-associated form (Muzumdar et al., 2007). Therefore, in confocal images with a relatively large slice thickness, the overlay of the fluorescent images obtained at 488 nm and 561 nm in some cases wrongly suggests colocalization. Colocalization was only concluded when the signal from EGFP colocalized with an immunohistochemical signal obtained at 633 nm in an area which was not restricted to the direct cell-cell contact site.

Tongues (n = 4) were sectioned longitudinally and 10–15 cryosections were analyzed for each mouse. Taste buds were identified based on their onion-like morphology visualized by Tomato. Number of total and EGFP-positive taste buds was counted manually throughout the section using the 20x magnification field. Hearts (n = 5) were sectioned longitudinally and 12–15 independent images were taken at 20x magnification per mouse. Muscularized vessels were identified by αSMA. Number of total and EGFP-positive muscularized vessels was counted manually. Tracheas (n = 4) between the larynx and the bifurcation were dissected and sectioned longitudinally. Twenty to twenty-four independent images were taken at 20x magnification for each mouse. Cell counting was performed on the basis of nuclear staining with DAPI and cell markers. EGFP-positive cells, DCLK1-positive tuft cells, and total epithelial cells were counted manually. Stomachs (n = 4) were opened along the greater curvature, embedded flatly and sectioned as reported previously (Eberle et al., 2013). Images were taken at 20x magnification and gastric units, which could be viewed longitudinally, were analyzed. EGFP-positive cells, DCLK1-positive tuft cells and total epithelial cells were counted manually in 110–150 longitudinal gastric units for each mouse. Female urethra (n = 3) were sectioned longitudinally and 20–22 independent images were taken at 20x magnification for each mouse. EGFP-positive cells were counted manually and epithelial cells were counted by ImageJ (NIH) based on nuclear staining with DAPI in the epithelial layer.

The isolation protocol was reported previously (Voigt et al., 2012). Briefly, the tongue was removed and an enzyme mix containing 2.5 mg/ml dispase II (D4693, Sigma), 1 mg/ml collagenase A (10103578001, Roche) and 0.5 mg/ml DNase I (A3778,0100, AppliChem) in PBS was injected under the epithelial layer. The injected tongue was incubated for 15 min at room temperature and the epithelium was peeled, washed in PBS and prepared for RNA isolation.

The isolation protocol was reported previously (Rock et al., 2009). Briefly, tracheas were cut into pieces, incubated in 16 U/ml dispase (17105-041, Gibco) in PBS for 30 min at room temperature and centrifuged at 350 g for 5 min. The digested tracheas were resuspended in 0.1% trypsin (25200056, Gibco), 1.6 mM EDTA in PBS and incubated for 20 min at 37°C. Trypsin was inactivated by adding DMEM with 5% FBS. The cell suspension was passed through a 40-μm cell strainer (431750, Corning). The filtered cells were centrifuged at 350 g for 5 min and the cell pellet was resuspended in DMEM with 5% FBS for FACS analysis.

The isolation protocol was reported previously (Sakata et al., 2012). Briefly, stomachs were inverted, inflated with ~1 ml PBS, and digested with 0.05 mg/ml Liberase TM (05401119001, Roche) in DMEM while being slowly shaken at 37°C for 90 min. The cell suspension was centrifuged at 350 g for 5 min and the cell pellet was resuspended in 0.25% trypsin–EDTA for 5 min at 37°C. Trypsin was inactivated by adding DMEM with 10% FBS. The cell suspension was passed through a 100-μm cell strainer (431752, Corning). The filtered cells were centrifuged at 350 g for 5 min and the cell pellet was resuspended in DMEM with 10% FBS for FACS analysis.

Cells isolated from reporter mice were stained with EpCAM Alexa Fluor® 647 (118211, BioLegend) to label epithelial cells. Cells isolated from Rosa26^{flox−mT−stop−flox−mG} mice were not stained and served as EGFP/EpCAM-negative control. Single live cells were detected by gating on SSC/FSC and selecting DAPI-negative cells. EGFP-positive cells were identified and sorted based on EGFP-positive/DAPI-negative expression. Tomato-positive cells were identified and sorted based on Tomato-positive/EpCAM-positive/EGFP-negative/DAPI-negative expression. Same number of EGFP- and Tomato-positive cells was isolated from the tamoxifen-treated reporter mice and sorted directly in lysis buffer for RNA isolation.

RNA was prepared using RNAeasy Kit (74004, Qiagen) and treated with TURBO™ DNase (AM1907, Ambion) to remove genomic DNA. First-strand cDNA was synthesized using ProtoScript II Reverse Transcriptase (M0368, New England BioLabs). Real-time qPCR primers were designed with the online tool provided by Roche, and quantification was performed using the LightCycler 480 Probe Master System (Roche). Genomic DNA was used as a universal standard to calculate gene copy number per ng of RNA. Relative expression levels were obtained by normalization to Actb mRNA. Following real-time qPCR primers were used: Tas2r143 (forward): 5′-CAG GCA TCT TTT TGA ACT CCA-3′; Tas2r143 (reverse): 5′-TCT TCA GGG CCT TTC TCA GT-3′; Tas2r135 (forward): 5′-CCA TCA TGT CCA CAG GAG AA-3′; Tas2r135 (reverse): 5′-TCA GTA GTC TGA CAT CCA AGA ACT GT-3′; Tas2r126 (forward): 5′-GTG TGT GGG ATT GGT CAA CA-3′; Tas2r126 (reverse): 5′-GCT CCC GGA GTA CTC AAC C-3′; Actb (forward): 5′-AAA TCG TGC GTG ACA TCA AA-3′; Actb (reverse): 5′-TCT CCA GGG AGG AAG AGG AT-3′.

The isolation protocol of adult mouse cardiomyocytes was reported previously (Takefuji et al., 2012). Briefly, the heart was removed quickly and cannulated from the aorta with a blunted 27G needle to allow retrograde perfusion of the coronary arteries. The heart was first washed with 50 ml perfusion buffer (113 mM NaCl, 4.7 mM KCl, 0.6 mM KH₂PO₄, 1.2 mM MgSO₄, 12 mM NaHCO₃, 10 mM KHCO₃, 10 mM HEPES, 30 mM Taurine, 10 mM 2,3-Butanedione monoxime, 5.5 mM Glucose, pH 7.46), then digested with 75 ml digesting buffer (0.05 mg/ml Liberase DH (05401089001, Roche) and 12.5 μM CaCl₂ in perfusion buffer). The heart was removed from the perfusion apparatus and minced with a forceps in digesting buffer. Undissociated clumps were removed by filtration through 100 μm nylon mesh. Cardiomyocytes were enriched by three times centrifugation at 50 g for 1 min. RNA was isolated by RNeasy Kit (Qiagen). NanoString analyses were performed as described previously (Khan et al., 2011). In brief, 500 ng RNA was applied in a total volume of 30 μl in the assay. Barcodes were counted for ~1,150 fields of view per sample. Counts were first normalized to the geometric mean of the positive control spike count, then a background correction was done by subtracting the mean + two standard deviations of the eight negative control counts for each lane. Values that were <20 were fixed to background level. Relative expression levels were obtained by normalization to reference gene Gusb.

EGFP- and Tomato-positive cells were sorted by FACS and pooled from three reporter mice of one of the transgenic lines. RNA was isolated from EGFP- and Tomato-positive cells using the RNeasy micro Kit (Qiagen) combined with on-column DNase digestion (DNase-Free DNase Set, Qiagen) to avoid contamination by genomic DNA. RNA and library preparation integrity were verified with a BioAnalyzer 2100 (Agilent) or LabChip Gx Touch 24 (Perkin Elmer). RNA amount was adjusted on number of isolated cells by FACS and ~250 ng of total RNA was used as input for SMART-Seq® v4 Ultra® Low Input RNA Kit (Takara Clontech) for cDNA pre-amplification. Obtained full length cDNA was checked on LabChip and fragmented by Ultrasonication by E220 machine (Covaris). Final Library Preparation was performed by Low Input Library Prep Kit v2 (Takara Clontech). Sequencing was performed on the NextSeq500 instrument (Illumina) using v2 chemistry, resulting in minimum of 28 M reads per library with 1 × 75 bp single end setup. The resulting raw reads were assessed for quality, adapter content and duplication rates with FastQC (Andrews, 2010). Reaper version 13–100 was employed to trim reads after a quality drop below a mean of Q20 in a window of 10 nucleotides (Davis et al., 2013). Only reads between 30 and 150 nucleotides were cleared for further analyses. Trimmed and filtered reads were aligned vs. the Ensembl mouse genome version mm10 (GRCm38) using STAR 2.4.0a with the parameter “—outFilterMismatchNoverLmax 0.1” to increase the maximum ratio of mismatches to mapped length to 10% (Dobin et al., 2013). The number of reads aligning to genes was counted with featureCounts 1.4.5-p1 tool from the Subread package (Liao et al., 2014). Only reads mapping at least partially inside exons were admitted and aggregated per gene. Reads overlapping multiple genes or aligning to multiple regions were excluded. The Ensemble annotation was enriched with UniProt data (release 06.06.2014) based on Ensembl gene identifiers (Activities at the Universal Protein Resource, UniProt). The relative expression of genes in EGFP-positive cells compared to Tomato-positive cells was calculated as log₂(EGFP/Tomato).

The murine bitter taste receptor genes Tas2r143, Tas2r135, and Tas2r126 are located in close proximity to each other on chromosome 6 without any other known genes or coding sequences located within this area. NanoString analysis confirmed that these receptors were expressed in the heart (Figure 1A), suggesting that these three receptors may be transcribed under a common regulatory element (Foster et al., 2015). To analyze the expression of this cluster in mice we generated a BAC based Tas2r143-CreERT2 transgenic mouse line by inserting the cDNA encoding CreERT2 between the start codon of Tas2r143 and the stop codon of Tas2r126 (35.2 kb; Figure 1B). Transgenic mice were genotyped by genomic PCR to identify correct insertion of CreERT2 (Figures 1C,D).

To analyze the recombination function of Tas2r143-CreERT2 transgenic mice we crossed these mice with the Cre-reporter mouse line Rosa26^{flox−mT−stop−flox−mG} to generate Tas2r143-CreERT2;Rosa26^{flox−mT−stop−flox−mG} reporter mice (hereafter Tas2r143-reporter mice; Figure 2A). In this reporter strain the red fluorescent protein Tomato is ubiquitously expressed at baseline and switches to enhanced green fluorescent protein (EGFP) in all cells that have undergone Cre-mediated recombination. Activation of Cre recombinase was induced by tamoxifen application in adult mice for 5 days, and these mice were analyzed 5 days after the last injection (Figure 2B). EGFP-positive cells were detected in 47 ± 7% (n = 4, mean ± SD) of all taste buds in the tongue, and these cells were positive for the taste receptor cell marker PLCß2 (Roper, 2013; Figure 2C). We isolated the epithelium from the tongue and confirmed the expression of Tas2r143, Tas2r135, and Tas2r126 by performing real-time qPCR (Figure 2D). Beyond the tongue, EGFP-positive cells were detected in the trachea, stomach, jejunum, urethra, vagina, cervix, and thymus, as well as in the heart (Figure 2E). To validate and characterize the EGFP-positive cells we took a closer look at various organs.

Previous studies demonstrated solitary chemosensory cells in respiratory airways and bitter taste receptor expression in the trachea (Krasteva et al., 2011; Tizzano et al., 2011; Voigt et al., 2015). We analyzed the lower respiratory tract, which consists of the trachea, bronchi, bronchioles, and lungs. EGFP-positive cells were detected in the epithelium of trachea, primary bronchi, and rarely in the epithelium of smaller bronchi as well as bronchioles (Figure 3A; Supplementary Figure 3). The number of EGFP-positive epithelial cells in the trachea was assessed from imaging, and we found that about 4% of all epithelial cells were EGFP-positive (Table 1). The majority of EGFP-positive epithelial cells colocalized with DCLK1 (microtubule-linked protein kinase 1), which is a tuft cell marker (Gerbe et al., 2009). About 85% of tuft cells were EGFP-positive epithelial cells and about 90% of EGFP-positive epithelial cells were tuft cells (Table 2). In order to test whether EGFP-positive cells express members of the Tas2r143/Tas2r135/Tas2r126 cluster, we isolated mRNA from EGFP-positive cells purified from the tracheas of the reporter mice by FACS (Supplementary Figure 1) and performed RNA-seq analysis. The enrichment of EGFP-positive cells was confirmed by analyzing the expression of EGFP, CreERT2 and Tomato compared with Tomato-positive cells purified from the tracheas of the reporter mice (Figure 3B). EGFP-positive cells expressed Tas2r135, Tas2r126 and other bitter taste receptors including Tas2r108, Tas2r105, Tas2r138, Tas2r106, Tas2r118, Tas2r115, and Tas2r136 (Figure 3C). In addition, EGFP-positive cells demonstrated high expression levels of genes encoding proteins in the bitter taste transduction system (Gnat3, Plcb2, Trmp5), acetylcholine synthesis (Chat; Roper, 2013) as well as cytokines responsible for type II immunity (Il25 and Tslp; Gerbe and Jay, 2016; Figure 3D). Besides Dclk1, EGFP-positive cells highly expressed tuft cell markers Ptgs1 and Pou2f3 (Gerbe et al., 2016; von Moltke et al., 2016) as well as the ciliated cell marker Tuba1a, which is also enriched in tuft cells (Bezencon et al., 2008), but showed poor expression of marker genes for neuroendocrine cells, goblet cells, basal cells and club cells (Kotton and Morrisey, 2014; Figure 3E).

Solitary chemosensory cells have been reported in the stomach and intestine (Chandrashekar et al., 2006; Finger and Kinnamon, 2011). Recently, these cells were shown to function as critical sentinels in the gut epithelium by promoting type II immune response (Gerbe et al., 2016; Howitt et al., 2016; von Moltke et al., 2016).

Tas2r143-reporter mice showed that EGFP-positive cells were clustered in the gastric groove, which is a tissue fold located at the boundary between fundus and corpus, and which is enriched by chemosensory cells (Eberle et al., 2013). EGFP-positive cells could also be detected in the epithelium of the corpus close to the gastric groove but rarely in other regions of the stomach (Supplementary Figure 3). EGFP-positive cells in both gastric groove and epithelium were to 100% colocalized with the tuft cell marker DCLK1 (Figure 4A; Table 2). In addition, EGFP-positive cells in the gastric groove were positive for α-gustducin and ChAT (choline acetyltransferase), the latter is indicating a cholinergic nature of these cells (Supplementary Figure 2). Quantitative assessment of the stomach corpus area showed that about 0.3% of total epithelial cells were EGFP-positive and about 15% of all DCLK1-positive tuft cells were EGFP-positive (Tables 1, 2). In order to test whether EGFP-positive cells express members of the bitter taste receptor cluster, we isolated mRNA from EGFP-positive cells purified from the stomachs of the reporter mice by FACS (Supplementary Figure 1) and performed RNA-seq analysis. The enrichment of EGFP-positive cells was confirmed by analyzing the expression of EGFP, CreERT2, and Tomato compared with Tomato-positive cells purified from the stomachs of the reporter mice (Figure 4B). EGFP-positive cells only expressed Tas2r126 (Figure 4C) and demonstrated high expression levels of genes encoding proteins in the bitter taste transduction system (Gnat3, Plcb2, Trmp5), acetylcholine synthesis (Chat; Roper, 2013) as well as the cytokine Il25 responsible for type II immunity (Gerbe and Jay, 2016; Figure 4D). In addition to Dclk1, EGFP-positive cells highly expressed gastrointestinal tuft cell markers Ptgs1, Pou2f3, and Gfi1b (Gerbe et al., 2016; von Moltke et al., 2016), but showed poor expression of marker genes for enteroendocrine cells, Paneth cells, parietal cells, chief cells, isthmus cells, and pit cells (Quante and Wang, 2009; Stange et al., 2013; Figure 4E).

In the intestine, EGFP-positive cells were rarely found in the epithelium, but all of the EGFP-positive epithelial cells were DCLK1-positive. In addition, EGFP-positive cells were detected in the lamina propria of duodenum, jejunum, and colon, and the majority of these cells were positive for ChAT (Figure 5).

Since Tas2r143-reporter mice showed EGFP-positive cells in the epithelium of airways and the gastrointestinal tract, we investigated the epithelium in other organs and found obvious EGFP expression in the urethra, vagina and cervix. In the transitional epithelium of the urethra, EGFP-positive cells appeared to be in all layers of the epithelium and a subset of cells colocalized with mitotically active basal epithelial cell marker K5 or terminally differentiated epithelial cell marker K10. Quantitative assessment from the imaging showed that about 3% of total epithelial cells were EGFP-positive in the urethra (Table 1). In the vagina and cervix, which are lined by stratified squamous epithelium, the majority of EGFP-positive cells were found in the K5-positive mitotically active basal epithelial layer, and a small fraction of cells colocalized with terminally differentiated epithelial cell marker K10 (Figure 6).

Cortical and medullary thymic epithelial cells coordinate the development and repertoire of T cells in the thymus (Abramson and Anderson, 2017). Recently, cholinergic chemosensory cells have been reported in the thymus in ChAT- and Tas2r131-reporter mice (Panneck et al., 2014; Soultanova et al., 2015). Also in Tas2r143-reporter mice, EGFP-positive cells were detected in the thymus. These cells localized in the thymus medulla, but were negative for K5 and the T cell marker CD3. EGFP-positive cells in the thymus medulla were positive for ChAT and partially colocalized with cortical epithelial cell marker K18 and terminal differentiated epithelial cell marker K10 (Figure 7).

A previous study (Foster et al., 2013) and our NanoString analysis suggested Tas2r143/Tas2r135/Tas2r126 expression in the heart. Interestingly, EGFP-positive cells were detected in cardiac blood vessels in Tas2r143-reporter mice. These cells were adjacent to endothelial cells marked by CD31, and were positive for αSMA, indicating that these cells were vascular smooth muscle cells (Figure 8). Quantitative assessment from the imaging showed that 47 ± 6% (n = 5, mean ± SD) of cardiac muscularized vessels contained EGFP-positive cells.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.