Nasal chemosensory cells use bitter taste signaling to detect irritants and bacterial signals

Abstract

The upper respiratory tract is continually assaulted with harmful dusts and xenobiotics carried on the incoming airstream. Detection of such irritants by the trigeminal nerve evokes protective reflexes, including sneezing, apnea, and local neurogenic inflammation of the mucosa. Although free intra-epithelial nerve endings can detect certain lipophilic irritants (e.g., mints, ammonia), the epithelium also houses a population of trigeminally innervated solitary chemosensory cells (SCCs) that express T2R bitter taste receptors along with their downstream signaling components. These SCCs have been postulated to enhance the chemoresponsive capabilities of the trigeminal irritant-detection system. Here we show that transduction by the intranasal solitary chemosensory cells is necessary to evoke trigeminally mediated reflex reactions to some irritants including acyl–homoserine lactone bacterial quorum-sensing molecules, which activate the downstream signaling effectors associated with bitter taste transduction. Isolated nasal chemosensory cells respond to the classic bitter ligand denatonium as well as to the bacterial signals by increasing intracellular Ca²⁺. Furthermore, these same substances evoke changes in respiration indicative of trigeminal activation. Genetic ablation of either Gα-gustducin or TrpM5, essential elements of the T2R transduction cascade, eliminates the trigeminal response. Because acyl–homoserine lactones serve as quorum-sensing molecules for Gram-negative pathogenic bacteria, detection of these substances by airway chemoreceptors offers a means by which the airway epithelium may trigger an epithelial inflammatory response before the bacteria reach population densities capable of forming destructive biofilms.

The nasal cavity houses two major chemoreceptive modalities potentially capable of detecting foreign organisms and substances: the extended olfactory system and the trigeminal chemesthetic system. The extended olfactory system, including the main olfactory epithelium, septal organ, and vomeronasal organ, contains bipolar chemoreceptor neurons that send axons directly to the forebrain and that trigger behavioral responses to sexual and social signals from conspecifics, territorial marking, and general odorants including food odors and odors of predators. The trigeminal chemesthetic system relies instead on free nerve endings and solitary chemosensory cells (also called solitary chemoreceptor cells, or SCCs) (1). The respiratory epithelium of the nasal cavity houses a significant population of SCCs (2–4) (Fig. 1), which respond to a wide variety of irritants to activate trigeminal nerve-mediated protective airway reflexes such as sneezing, coughing, or apnea (5) as well as local neurogenic inflammatory responses (6, 7). The SCCs express many elements of the bitter taste signaling pathway, including the Tas2R family of receptors, Gα-gustducin, phospholipase Cβ2 (PLCβ2), and the monovalent-selective cation channel TrpM5 (1, 3, 4). Receptor–ligand interaction results in a transduction cascade involving activation of Gα-gustducin, PLC-mediated release of Ca²⁺ from intracellular stores, Ca²⁺-dependent activation of TrpM5, membrane depolarization, and, presumably, neurotransmitter release and subsequent excitation of capsaicin-sensitive trigeminal pain fibers that richly innervate the chemosensory cells (Fig. 1 B and C; see also refs. 1 and 4). Studies to date have used a variety of artificial and natural compounds to activate the SCCs, but identification of natural ligands has remained elusive. Because naturally occurring plant-derived lactones activate bitter taste receptors (8), Sbarbati and coworkers (9) have suggested that the airway chemosensory cells might be capable of responding to acyl–homoserine lactones (AHLs) produced by Gram-negative bacteria (10) as quorum-sensing signals. For example, the opportunistic respiratory pathogen Pseudomonas aeruginosa signals its population density using small molecules including AHLs (reviewed in refs. 11–13). When population density is low, the concentration of AHLs is low; as the population density increases, the AHL concentration also increases, reaching levels over 200 μM (14). This triggers downstream signaling cascades in the bacteria through specific AHL-binding transcriptional regulatory proteins leading to the production of pathogenic biofilms and destructive virulence factors (15). Accordingly, we investigated whether AHL signaling molecules can activate the epithelial chemosensory cells at micromolar concentrations, and if so, whether such activation is sufficient for activation of the trigeminal airway responses.

Fig. 1.

SCCs in the nose are innervated by trigeminal sensory nerves. (A) Midline section through the head of a mouse showing the nasal airway passages. The area of respiratory epithelium indicated by the blue rectangle contains a high density of SCCs. (B and C) Double-label images show the SCC cells (exhibiting green TrpM5-driven GFP fluorescence) richly innervated by pain fibers of the trigeminal nerve, which contain the neuropeptides substance P (red fluorescence in B and B′′) and CGRP (red fluorescence in C and C′′). (D) Area marked by blue rectangle in A in a TrpM5–GFP mouse showing green fluorescent SCCs in the nasal epithelium.

Previously, we have shown that SCCs respond to the presence of denatonium using a PLC-mediated cascade to generate an increase in intracellular Ca²⁺ (2). To test whether bacterially produced AHLs can stimulate the solitary chemosensory cells similarly, we took advantage of two lines of transgenic mice in which either the Gα-gustducin or TrpM5-promoter drives expression of green fluorescent protein (GFP) (1, 4). In both of these lines, fluorescent SCCs are readily apparent in the anterior nasal epithelium (Fig. 1) as reported previously (1, 2, 4). The SCCs are heavily innervated by peptidergic [calcitonin gene-related peptide (CGRP)/substance P-immunoreactive] nerve fibers of trigeminal origin that also express TrpV1, the capsaicin receptor (16).

Results

Responses of SCCs to Bacterial Quorum-Sensing Molecules.

Fluorescent cells (Fig. 1D) obtained from dissociated epithelium were loaded with the calcium indicator dye fura-2 acetoxymethyl ester (AM), and the intracellular Ca²⁺ level was monitored before, during, and after application of a variety of potential stimuli. Our previous experiments with this system showed that a large percentage of the GFP-labeled (either gustducin-driven or TrpM5-driven) SCC cells responded robustly to 10–20 mM denatonium, a classic bitter-tasting ligand (2). We used similar methods to test whether denatonium-responsive SCCs respond to either bacterially produced or synthetic AHLs.

Of the SCCs tested, the large majority also responded with an increase in intracellular Ca²⁺ to one or more of the AHL samples (Fig. 2 A–C). The AHLs produced by the P. aeruginosa LasI enzyme expressed in Escherichia coli (17) include 3-oxo-C12-homoserine lactone (HSL) along with related long-chain (3-oxo-C10-C14) HSLs (Fig. 2F; also see Fig. S1) (18), whereas 3-oxo-C6-HSL and C6-HSL were produced (18, 19) in E. coli by the T140A mutant of EsaI enzyme (referred to herein as EsaI). Over 90% (34/36) of the denatonium-responsive cells also showed robust responses to 60- to 120-μM concentrations of the LasI- or EsaI-produced AHLs as well as to the synthetic AHLs (Fig. 2 A–C). Few chemosensory cells exhibited responses to the C4-HSL signal from expression of the P. aeruginosa AHL synthase RhlI. There was essentially no response to extracts from bacteria lacking all AHL-synthase genes (containing the pViet plasmid only) (17), indicating that the response was due to the bacterially produced AHL rather than to any natural metabolites or components of the E. coli itself. In nearly all cases (13/14), responses to the bacterially produced 3-oxo-C12-HSL and 3-oxo-C6-HSL/C6-HSL were eliminated or significantly reduced (to <50%) by treatment of the SCCs with the irreversible PLC inhibitor U73122 (Fig. 2 A and B). The inactive form of that drug, U73343, had no effect on the responses (Fig. 2E), establishing the specificity of the drug effect. Likewise, responses to the synthetic 3-oxo-C12-HSL (Sigma) were significantly reduced or eliminated by U73122 (Fig. 2E). These findings strongly suggest that SCCs respond to bacterial AHL products using a GPCR/PLC-mediated signaling cascade. Responsiveness to the AHLs was restricted to the GFP-labeled epithelial cells because non-GFP-labeled cells showed no calcium responses to these ligands (Fig. 2C). Using the synthetic AHLs, we find an EC₅₀ for both the 3-oxo-C12-HSL and 3-oxo-C6-HSL to be 50–80 μM (Fig. 2D). Because bacterially produced biofilms generate levels of AHL in excess of 200 μM (14), this detection level is well within the range of concentrations predicted for an expanding bacterial population.

Fig. 2.

Ca²⁺ responses in SCCs show responsiveness to AHLs. More than 50% (16/31) of TrpM5–GFP cells and 75% (36/47) of Gα-gustducin-GFP-–labeled cells respond to 10–20 mM denatonium (Den) and to the bacterially produced AHLs but not to the control plasmid (pViet). (A and B) Responses are eliminated by treatment with the PLC inhibitor U73122, indicating that these Ca²⁺ responses are mediated by a G-protein–coupled receptor system. (C) Only GFP-labeled cells respond to denatonium (Den) and AHL molecules (C6, 3-oxo-C6-HSL; C12, 3-oxo-C12-HSL); neither GFP-labeled nor unlabeled epithelium cells show Ca²⁺ responses to stimulation by capsaicin (Cap). (D) Dose–response curves relating the magnitude of the Ca²⁺ response to the concentration of the applied synthetic AHL. (E) Cell responses (Mean ± SEM) to denatonium (Den) and 3-oxo-C12-HSL (C12) applied in the presence of the control (noneffective) inhibitor (U73343) and the active PLC inhibitor U73122 normalized to the response with no inhibitor (normalized response = 1.0). The PLC blocker U73122 significantly (** denotes P < 0.001, one-sample t test with Bonferroni correction, n = 7 cells) reduces the Ca²⁺ response of isolated TrpM5–GFP SCCs to both denatonium (Den) and C12 synthetic AHL whereas the control drug has no effect (P > 0.05). (F) Structure of the principal AHL compounds produced by each of the bacterial genes used. See Fig. S1 for a complete analysis of the bacterial products.

Respiratory Responses to Intranasal Stimulation by Bitter Compounds.

Although Ca²⁺ imaging of isolated SCCs adequately tests the cellular-level responses to particular substances, it does not give any insights into whether these cellular responses are transmitted to the nervous system. To assess the effectiveness of AHLs in evoking activation of the trigeminal nerve, we tested whether these substances would give rise to trigeminal-nerve-mediated respiratory reflexes. Activation of the trigeminal nerve by other irritants produces neurogenic reflexes, including local inflammation and respiratory depression (1, 5, 20), but whether the AHLs could activate the system adequately required direct testing. For these experiments, we relied on a modification of the traditional Alarie assay for respiratory irritation, which measures changes in respiratory rate as a function of stimulus concentration. The rapid time course of the respiratory change in this assay is indicative of a neurally mediated reflex and correlates strongly with trigeminal nerve activation (1, 5). As was shown previously for rats (1), 20 mM denatonium, a known T2R ligand and SCC activator, caused significant decrease in spontaneous respiration (Fig. 3A). Similarly, both the LasI (3-oxo-C12-HSL) and EsaI (3-oxo-C6-HSL/C6-HSL) samples as well as synthetic AHLs evoked significant reductions in respiratory rate (13/13, 7/7, and 8/8 mice tested, respectively) (Fig. 3). These results indicate that those AHLs that produce a Ca²⁺ surge in SCCs also are capable of evoking responses indicative of nasal trigeminal irritants. Although we measure only the respiratory component of these reflexes, activation of the trigeminal peptidergic fibers that innervate the SCCs will also evoke local neurogenic inflammatory changes (6), which are likely to be crucial in the immediate epithelial and/or local immune response to the presence of the bacterial agents.

Fig. 3.

Respiratory changes evoked by AHL stimulation in wild type (blue) and TrpM5–KO (red) mice. Gα-gustducin–KO mice show a similar lack of responsiveness. (A–D) Graphs (Upper) and respiratory records (Lower) showing changes in relative breath duration following stimulation [stimulus application indicated by a caret (“_^”)] in wild-type (WT) and KO strains. Values >1 indicate a slowing in breathing with higher numbers indicating a longer pause. The lack of respiratory response to denatonium, LasI, or EsaI products in the KO mice implicates the SCCs in the detection of these substances. The continued responsiveness of the KO mice to capsaicin (D) confirms that the trigeminal nerve is directly responsive to capsaicin and does not require functioning of the SCCs. (E) Summary graph (mean ± SEM) comparing relative breath duration following stimulation with irritants and control solutions in WT, Gα-gustducin–KO, and TrpM5–KO animals. Application of denatonium, LasI, or EsaI products and the synthetic HSLs each resulted in a highly significant (two-tailed t test: ***P < 0.001; *P < 0.05) pause in respiration in the WT but not in the KO mice whereas application of capsaicin and acetic acid elicits responses in WT as well as in KO lines. Application of 1% methanol (vehicle control) or extract from bacteria carrying the control pViet plasmid showed no such responses.

Respiratory Responses in Mice Lacking Gα-gustducin and TrpM5.

The effectiveness of the AHLs in evoking a trigeminal reflex does not, by itself, implicate the SCCs in the detection of these substances. The trigeminal nerve fibers themselves express a variety of chemically sensitive channels and respond directly to many irritants (21–23), including capsaicin (via TrpV1), acetic acid (via acid-sensing ion channels [ASICs] and/or TrpV channels), nicotine (via nicotinic receptors), and mustard oil (via TrpA1) (24, 25). To test whether SCC function is required for AHL-induced respiratory responses, we tested respiratory reflexes in Gα-gustducin and TrpM5 knockout mice. In these animals, the TrpM5 transduction channel or the Gα subunit required for T2R function is absent, so that the SCCs are incapable of transmitting transduction information to the afferent nerve fibers (Figs. 3 and 4). Because TrpM5 and Gα-gustducin are expressed by SCCs and not nerve fibers (Fig. S2), any lack of response in the knockout mice must be attributable to the SCCs and not to the nerve fibers. Both TrpM5-KO and Gα-gustducin–KO mice failed to exhibit respiratory reflex responses to SCC-mediated stimuli (denatonium and AHLs; Fig. 3) although they still showed robust respiratory responses to substances known to activate the trigeminal sensory terminals directly: capsaicin and acetic acid (Figs. 3 and 4). Because some olfactory receptor neurons and other cells of the olfactory epithelium also express TrpM5 (27, 28), absence of a response in TrpM5 KO animals is not conclusive. However, Gα-gustducin is expressed only by SCCs; hence the lack of responsiveness in Gα-gustducin–KO mice strongly implicates these cells and transmission to the trigeminal nerve processes as necessary components in this response.

Fig. 4.

Summary diagram of the SCC trigeminal system. The SCCs synapse on the same trigeminal nerve fibers that provide sensory innervation to the surrounding epithelium in the form of free nerve endings. The nerve fibers express several chemosensitive ion channels, including TRPs and ASICs. These nerve endings terminate below the level of the epithelial tight junctions (26) and therefore can respond only to stimuli capable of crossing this barrier. The SCCs extend above the barrier of tight junctions and so have access to all stimuli in the overlying mucus layer. Our results show that certain stimuli, including denatonium and AHLs, must act via the SCCs to activate the trigeminal nerve.

Discussion

We offer here a functional demonstration that nasal solitary chemosensory cells are necessary for the initiation of trigeminally mediated respiratory reflexes for certain classes of substances. The intraepithelial free nerve endings of the trigeminal nerve are capable of detecting a variety of irritants via the intrinsic Trp channels including TrpV1 (capsaicin), TrpA1 (mustard oil, etc.) and TrpM8 (mints) (21, 22) as well as ASICs (for responsiveness to acids). Because the trigeminal free nerve endings terminate just below the level of the epithelial tight junctions (26), the type and number of stimuli that can penetrate the junctional barrier to access the nerve endings is limited. The presence of SCCs, which extend beyond the junctional barrier, offers a means whereby the irritant-detecting system can respond to substances restricted to the mucus layer overlying the epithelium. Further, the expression of T2R receptor molecules by the SCCs (1) expands the potential molecular repertoire of the system beyond that of the chemosensitive channels or receptors on the nerve fibers, e.g., Trp channels or nicotinic receptors (29).

Although T2Rs and their related downstream signaling elements were first identified in taste cells (30, 31), the T2R receptor cascade is present in numerous cells throughout the body, including the gastrointestinal tract (3, 32) and the respiratory tree (1, 3, 4, 10, 33, 34). Most of these reports describe a limited population of specialized epithelial cells similar in morphology and molecular signature to the solitary chemosensory cells that we describe. Such cell populations express T2Rs, gustducin, PLCβ2, and TrpM5 as well as other markers of taste cells. Despite this similarity in molecular expression, these diverse cells in different organ systems are not identical despite their sometimes being lumped together under the term “brush cells” (35). For example, the gustducin-positive cells that we have studied in the nasal cavity are heavily innervated by polymodal nociceptors (1) (Fig. S3 B and F) whereas similar-appearing cells lower in the respiratory tree are seldom innervated (Fig. S3 A, C, and E). Likewise, the T2R-expressing cells of the gut are a class of enterochromaffin cell that is not necessarily innervated (36) (Fig. S3D). The noninnervated enteroendocrine cells release bioactive peptides rather than neurotransmitters in response to T2R activation (37, 38). Despite these differences, the T2R/gustducin/TrpM5-expressing epithelial cells in the different tissues are similar in extending microvillous processes to the top of the epithelium where the cells will have access to the luminal content of each particular organ.

Recently, Shah and colleagues (34) described the expression of T2Rs and gustducin in the ciliated cells of the pulmonary epithelium in cultured airway epithelial cells from humans. In this situation, the T2Rs and associated downstream elements are not restricted to a scattered cell population in the epithelium, but are expressed widely by the ciliated respiratory epithelium. Further, the activation of the T2R cascade in the ciliated cells results in increases in intracellular Ca²⁺, which alters the ciliary beat frequency. In this system, then, the T2R signaling appears to act in a cell-autonomous fashion; i.e., the receptor cell is also the effector cell. We do not see any expression of T2Rs (in situ hybridization data in ref. 1) or gustducin or TrpM5-driven GFP in the ciliated cells of the nasal epithelium in mice. Whether ciliated pulmonary epithelial cells in mice express elements of the T2R transduction cascade in vivo remains to be determined.

The T2R family contains some dozens of different receptors in each species. In humans, the different T2Rs respond to a variety of natural ligands, including lactones and diterpenoids (8). The most obvious source of lactones for the taste system is the multitude of plants that produce these products. For the respiratory system, T2R ligands are likely to enter the airways in the form of aerosols and dusts. In addition, aerosols produced by the coughing or sneezing of individuals infected with Gram-negative bacteria potentially bring these pathogens into the airway, leading to further accumulation of AHLs through the growth of biofilms (39).

Various mammalian cells can respond to AHLs (reviewed in ref. 40). For example, AHLs can influence inflammatory responses (41–43), chemotaxis (44), induction of apoptosis (45–47), and mucus secretion (10). All of these responses are relatively slow and involve changes in gene regulatory networks or secretory mechanisms. In contrast, our study reveals a fast response in the time frame of sensory systems, i.e., milliseconds to seconds. This implicates the trigeminal sensory fibers as the afferent limb of a central reflex impacting on brainstem respiratory centers (48).

Sensory detection of bacteria by complex animals is not without precedent. In Caenorhabditis elegans, specific chemosensory cells can detect bacterially generated AHLs that serve as a chemoattractant for feeding (49). Even in rodents, the extended olfactory system, i.e., the vomeronasal organ, responds to by-products of bacterial infection, e.g., formyl peptides (50, 51), which may indicate the health status of conspecifics. These behavioral responses, however, require integrative activity unlike the reflex activation of trigeminal airway protective systems. Our study, then, not only offers a previously undescribed demonstration of a clear function for nasal SCCs, but also establishes that a mammalian sensory system can directly detect the presence of bacteria on an epithelial surface. The consequent activation of trigeminal capsaicin-sensitive nerve fibers (52, 53) releases CGRP and substance P into the surrounding mucosa, resulting in local inflammation, including microvascular leakage and activation of the innate immune system, to combat the bacterial invasion (7).

Materials and Methods

Preparation of Bacterially Generated AHLs.

AHL bacterial signaling molecules were produced in E. coli expressing AHL synthase genes (EsaI and LasI) and were purified using solid-phase extraction procedures as described previously (18, 19). The purified extracts were analyzed (Fig. S1) by means of mass spectrometry. Similarly purified bacterial extracts from E. coli that harbor the empty vector pViet were found to contain no AHLs. After purification and quantitation of the AHLs, the resulting solutions of AHLs in 1% methanol were used in functional studies.

Cell Isolation and Ca Imaging.

SCCs were isolated from the anterior nasal epithelium of TrpM5–GFP or Gα-gustducin–GFP mice using papain in Ca–Mg-free Tyrode’s solution as described previously (2). Individual cells or small cell clusters were plated onto polylysine-coated coverslips and loaded with fura-2 AM Ca-indicator dye. Calcium levels were determined as a ratio of fluorescence emissions at F350/F380.

Respiratory Monitoring.

Wild-type, TrpM5–KO, and Gα-gustducin–KO mice were anesthetized with urethane and subjected to tracheotomy. A breathing tube containing a thermistor to monitor respiratory activity was inserted into the pulmonary end of the trachea. The upper end of the trachea was cannulated to permit continuous retronasal flow of Tyrode’s solution. Test substances were injected into the retronasal stream and allowed to flow through the nasal cavity. Respiratory rate was determined as instantaneous interbreath interval. See SI Materials and Methods for detailed description of experimental procedures.