Report

ATP Signaling Is Crucial for Communication from Taste Buds to Gustatory Nerves

See allHide authors and affiliations

Science  02 Dec 2005:
Vol. 310, Issue 5753, pp. 1495-1499
DOI: 10.1126/science.1118435

Abstract

Taste receptor cells detect chemicals in the oral cavity and transmit this information to taste nerves, but the neurotransmitter(s) have not been identified. We report that adenosine 5′-triphosphate (ATP) is the key neurotransmitter in this system. Genetic elimination of ionotropic purinergic receptors (P2X2 and P2X3) eliminates taste responses in the taste nerves, although the nerves remain responsive to touch, temperature, and menthol. Similarly, P2X-knockout mice show greatly reduced behavioral responses to sweeteners, glutamate, and bitter substances. Finally, stimulation of taste buds in vitro evokes release of ATP. Thus, ATP fulfils the criteria for a neurotransmitter linking taste buds to the nervous system.

Taste buds transduce chemical signals in the mouth into neural messages transmitted to gustatory nerve fibers of the facial and glossopharyngeal nerves. Despite recent progress in delineating the molecular mechanisms for taste transduction, the identity of the neurotransmitter from taste buds to the nerve fibers is unknown. Elucidation of the neurotransmitter between taste cells and nerve fibers is complicated by several factors. Each taste bud contains a variety of cell types, likely communicating with each other as well as with the gustatory nerve fibers. Hence, the mere presence of a potential transmitter substance in the taste bud does not imply that a particular substance is used to activate the gustatory nerve fibers.

Recent studies have suggested that serotonin [5-hydroxytryptamine (5-HT)] is released by taste buds (1) and acts on 5-HT3 receptors on the gustatory nerve fibers to convey the taste message (2). To test this hypothesis, we examined the taste behavior of 5-HT3A knockout (KO) mice. The 5-HT3A subunit is crucial for the function of both homomeric and heteromeric 5-HT3 receptors (3). These KO mice have depressed behavioral thresholds to pain, in which serotonin acts as both a peripheral activator and a central modulator (4). In contrast, the taste behavior of the 5-HT3A KO mice is identical to that of wild-type controls for each taste quality tested (fig. S1). Thus, serotonin does not act on neural 5-HT3 receptors to transmit taste information from taste buds to the gustatory nerve.

Taste nerves express two ionotropic P2X receptor subunits (P2X2 and P2X3) (5), which suggests that adenosine triphosphate (ATP) may serve as a neurotransmitter in this system. P2X2 and P2X3 receptor subunits can form homomeric P2X2, homomeric P2X3, and heteromeric P2X2/3 receptors (6). ATP acting on P2X2 receptors (likely homomeric P2X2) appears crucial to transmission in the carotid body, which has structural similarities to taste buds (7). In order to test whether the P2X receptors are required for transmission of taste information, we recorded neural responses to chemical stimulation of the oral cavity in P2X2/P2X3 double-knockout [P2X2/P2X3Dbl–/– (KO)] mice (8), C57BL/6J mice, and P2X2/P2X3 double wild-type [P2X2/P2X3Dbl+/+ (WT)] mice. No taste-evoked activity was seen in either the chorda tympani or glossopharyngeal nerves of P2X2/P2X3Dbl–/– (KO) mice. In contrast, stimulation by touch, temperature, and menthol solutions elicited robust neural responses (Fig. 1). Menthol activates fine-caliber somatosensory nerve fibers via direct action on neural transient receptor potential (TRP) channel receptors and does not involve taste buds (9). Thus, although the gustatory nerves of P2X2/P2X3Dbl–/– (KO) mice are highly responsive to somatosensory stimuli, they are not responsive to taste. These findings indicate that P2X2 and/or P2X3 receptors are essential for activation of the gustatory nerves.

Fig. 1.

(A and B) (Top) Gustatory nerve recordings from P2X2/P2X3Dbl–/– (KO) and P2X2/P2X3Dbl/+/+ (WT) mice. (Bottom) Comparison of response magnitude of control mice (blue bars) and P2X2/P2X3Dbl–/– (KO) (red bars) to a variety of taste, tactile, and thermal stimuli. The responses of wild-type siblings do not differ significantly (ANOVA, P > 0.01) from responses recorded from C57BL/6J mice. The control group for the CT recordings includes data from both P2X2/P2X3Dbl/+/+ (WT) and C57BL/6J mice. Blue asterisks indicate significant differences (P < 0.05) between control and KO groups (t test with Bonferroni correction for repeated measures).

To characterize taste-related behaviors of the P2X2/P2X3Dbl–/– (KO) mice, two-bottle preference tests were used (10). Compared with P2X2/P2X3Dbl/+/+ (WT) mice, the P2X2/P2X3Dbl–/– (KO) mice were not responsive to artificial sweeteners (Fig. 2, A and C), sucrose (Fig. 2B), or monosodium glutamate (Fig. 2D) and were largely unresponsive to many bitter substances including quinine hydrochloride (Fig. 2E) and denatonium benzoate (Fig. 2F). Single-KO mice, lacking either the P2X2 or P2X3 subunit, also were assessed in two-bottle preference tests for certain key tastants. The P2X3-only KO animals were significantly impaired in their responses to SC45647, saccharin, and denatonium (fig. S2), whereas the P2X2-only KO mice were only marginally impaired to denatonium. The single-KO strains were, however, at least an order of magnitude more responsive than the P2X2/P2X3Dbl–/– (KO) line. Thus, loss of either P2X2 or P2X3 alone resulted in only a moderate change in taste-mediated behaviors in contrast to the profound deficit seen in P2X2/P2X3Dbl–/– (KO) animals. This suggests that neither homomeric P2X2 nor homomeric P2X3 receptors suffice for normal function in this system.

Fig. 2.

(A to H) Behavioral data showing the P2X2/P2X3Dbl–/– (KO) mice (red) do not respond to most taste substances and exhibit near-absent behavioral responses to other tastants including quinine and denatonium (bitter). For the artificial sweetener SC45647 (C), P2X2-only KO mice are not significantly different from WT mice. In contrast, P2X3-only KO mice are significantly impaired compared with WT but are significantly less impaired than the P2X2/P2X3Dbl–/– (KO) animals (fig. S2) (ANOVA, P < 0.01). Red boxes indicate points that are significantly different between P2X2/P2X3Dbl–/– (KO) and P2X2/P2X3Dbl/+/+ (WT) mice (t test with Bonferroni correction, P < 0.05).

The near-total loss of neural and behavioral responses to taste stimuli suggests a peripheral origin to the defect. We compared the morphology and innervation of the lingual taste buds in the P2X2/P2X3Dbl–/– (KO) mice with P2X2/P2X3Dbl/+/+ (WT) animals. As in P2X2/P2X3Dbl/+/+ (WT) mice, the P2X2/P2X3Dbl–/– (KO) mice showed a normal complement of taste cells in fungiform, foliate, and circumvallate papillae (Fig. 3, A to D). Cells within the taste buds express the T1R taste receptors as assessed by in situ hybridization (Fig. 3, A to D for T1R1 and T1R2) and display roughly normal proportions of cells expressing gustducin, phospholipase C–β2 (PLC–β2), or serotonin, all markers of taste cells in normal mice. Thus, the peripheral taste apparatus appears intact both structurally and molecularly.

Fig. 3.

Taste buds in P2X2/P2X3Dbl–/– (KO) mice (A and C) are normal in terms of cell morphology and histochemistry compared with wild-type mice (B and D). (A and B) Taste buds in circumvallate papillae showing expression of gustducin (green) and T1R2 (red). (C and D) Palatal taste buds showing expression of gustducin (green) and T1R1 (red). (E) Laryngeal solitary chemoreceptor cells are densely innervated in wild-type mice. Gustducin immunoreactivity (green) in a solitary chemoreceptor cell showing dense innervation as revealed by immunoreactivity with the pan-neuronal marker PGP9.5 (red). (F) Laryngeal taste buds are innervated by purinoceptive nerve fibers expressing P2X2 (red). Gustducin-expressing taste cells are green. This suggests that laryngeal taste buds, like lingual taste buds, rely on ATP as a transmitter. (G) Laryngeal solitary chemoreceptor cells (green) in a WT mouse are not innervated by P2X-expressing nerve fibers (red), although such fibers do innervate nearby epithelium. This indicates that nerve fibers that innervate laryngeal SCCs utilize a different neurotransmitter and/or receptor system. Compare this image with (E). (H and I) c-Fos immunoreactivity in the laryngeal portion of the nucleus of the solitary tract in P2X2/P2X3Dbl–/– (KO) mice. Activation of c-Fos in numerous cells (dark spots indicated by red circles) of this area indicates that quinine stimulates the laryngeal nerve, which sends information to this caudal portion of the nucleus. AP, area postrema; NTS, nucleus of the solitary tract; nXII, hypoglossal nucleus; v, 4th ventricle.

Despite the lack of neural response to any applied tastant in the P2X2/P2X3Dbl–/– (KO) mice, the animals do exhibit near-normal avoidance to caffeine (bitter to humans) (Fig. 2G) and citric acid (sour) (Fig. 2H). These clear behavioral responses are unexpected, given the total lack of chorda tympani and glossopharyngeal gustatory nerve response to these same substances. Two possibilities may explain the behavioral results. Either these substances are being detected by chemoreceptors that are not taste buds, including those of the larynx, pharynx, esophagus, or gut, or nonlingual taste buds use different neurotransmitters than do lingual taste buds.

The chemical profile (acids and some bitter-tasting substances) of compounds detected by P2X2/P2X3Dbl–/– (KO) mice is similar to the response profile of laryngeal chemoreceptors innervated by the superior laryngeal nerve (11, 12). The larynx is replete with both taste buds and solitary chemoreceptor cells (SCCs) (13) lying within a specific laryngeal sensory epithelium (14). To determine whether either or both of these laryngeal chemoreceptors rely on purinergic neurotransmission, we used immunocytochemistry to localize P2X2 and P2X3 receptor subunits in transgenic mice in which green fluorescent protein (GFP) is expressed in gustducin-expressing taste cells and SCCs. Laryngeal taste buds were clearly innervated by nerve fibers immunoreactive for P2X2 (Fig. 3F, red) and P2X3 receptor subunits. In contrast, laryngeal SCCs, identified by gustducin-driven GFP expression, were not innervated by P2X2 (Fig. 3G, compare Fig. 3E) or P2X3 subunit–expressing nerve fibers. These results suggest that laryngeal taste buds may utilize purinergic neurotransmission, whereas laryngeal SCCs do not.

In order to assess whether avoidance of sour and bitter tastants in P2X2/P2X3Dbl–/– (KO) mice is mediated by the gustatory system, we relied on tastant-induced expression of c-Fos within the brainstem. Activation of the gustatory system by strong tastants quickly activates expression of immediate early genes such as c-fos within the primary taste nucleus (NTS; nucleus of the solitary tract) in the brainstem (15, 16). The lingual gustatory nerves terminate within the rostral and intermediate parts of the NTS, so presentation of tastants to the oral cavity evokes prominent c-Fos expression within the rostral and/or intermediate region of the NTS in both wild-type mice and rats. In contrast, in P2X2/P2X3Dbl–/– (KO) mice, quinine (but not water) evokes little c-Fos activation in rostral or intermediate portions of NTS, but does evoke significant c-Fos expression in more caudal portions of the nucleus, where the superior laryngeal nerve and general visceral branches of the vagus nerve terminate (Fig. 3, H and I). These findings suggest that the behavioral avoidance seen in P2X2/P2X3Dbl–/– (KO) mice is mediated by the superior laryngeal nerve or general visceral branch of the vagus conveying signals to the caudal NTS.

Finally, to test for tastant-evoked release of ATP, we used a standard luciferin-luciferase bioluminescence assay to detect ATP release from stripped epithelial preparations. For these experiments, taste bud–bearing epithelia of C57BL/6J mice were stripped from the vallate and foliate taste fields, as well as from nongustatory epithelium devoid of taste buds. Stimulation of the apical surface with buffer resulted in a basal level of ATP release as measured by the luminometer. When a mixture of two bitter compounds was applied to the apical surface of taste bud–bearing epithelium, the luminous flux was significantly greater, which indicated an increase in ATP concentration (Fig. 4). In contrast, stimulation of nongustatory epithelium with the bitter mixture evoked little or no ATP release. These findings demonstrate that ATP is released from taste epithelium when it is exposed to appropriate taste stimuli.

Fig. 4.

Release of ATP from taste epithelium when stimulated with a bitter mixture containing denatonium and quinine. Nongustatory epithelium (non-TB) and taste bud–bearing epithelial sheets containing either circumvallate (CV) or foliate papillae were placed in an Ussing-type chamber that permits selective application of taste stimuli to the apical membrane. ATP released from the basolateral compartment was collected in the luciferase assay buffer and transferred to the luminometer for measurement of relative light units, which were converted into ATP concentration. Stimulation of taste epithelia with the bitter mixture significantly increases ATP release (mean ± SEM) from CV and foliate tissues relative to non-TB tissue (P < 0.05, t test).

Our results strongly suggest that ATP serves as a key neurotransmitter linking taste buds to sensory nerve fibers. Criteria for a neurotransmitter include release, the presence of specific receptors, and a mechanism for clearance. Our luminometer data demonstrate tastant-evoked release of ATP. The immunocytochemical studies by Bo et al. (5), as well as our own results, show the presence of postsynaptic P2X receptors, whereas the physiological and behavioral studies demonstrate that the P2X receptors are necessary for sensory transmission in this system. Finally, extracellular ATP is rapidly degraded by ecto-ATPases known to be abundantly present in taste buds (1719). Thus, ATP meets all the essential criteria for being the major neurotransmitter of the peripheral taste system. Other neuropeptides and transmitters observed within taste buds (1, 2, 20, 21) likely play a modulatory role or may be crucial for intragemmal communication among the different types of taste cells.

Supporting Online Material

www.sciencemag.org/cgi/content/full/310/5753/1495/DC1

Materials and Methods

Figs. S1 and S2

References and Notes

References and Notes

View Abstract

Navigate This Article