cAMP is a second messenger implicated in sensory transduction for taste. The identity of adenylyl cyclase (AC) in taste cells has not been explored. We have employed RT-PCR to identify the AC isoforms present in taste cells and found that AC 4, 6, and 8 are expressed as mRNAs in taste tissue. These proteins are also expressed in a subset of taste cells as revealed by immunohistochemistry. Alterations of cAMP concentrations are associated with transduction of taste stimuli of several classes. The involvement of particular ACs in this modulation has not been investigated. We demonstrate that glutamate, which is a potent stimulus eliciting a taste quality termed umami, causes a decrease in cAMP in forskolin-treated taste cells. The potentiation of this response by inosine monophosphate, the lack of response to d-glutamate, and the lack of response to umami stimuli in nonsensory lingual epithelium all suggest that the cAMP modulation represents umami taste transduction. Because cAMP downregulation via ACs can be mediated through Gαi proteins, we examined the colocalization of the detected ACs with Gαi proteins and found that 66% of AC8 immunopositive taste cells are also positive for gustducin, a taste-specific Gαi protein. Whether AC8 is directly involved in signal transduction of umami taste remains to be established.
- taste transduction
membrane-associated adenylyl cyclases (AC) are enzymes that are central in the regulation of intracellular cAMP levels. ACs, comprising a family of nine isoforms (AC1–9), are integral membrane proteins, composed of two cytosolic catalytic domains, C1 and C2, each preceded by six transmembrane domains (16, 34). The catalytic domains are highly conserved across all ACs and even across many guanylate cyclases and are responsible for the synthesis of cAMP using ATP as a substrate. Physiologically, G protein-coupled receptors (GPCRs) regulate ACs through heterotrimeric G protein subunits, specifically the stimulatory Gαs, the inhibitory Gαi, and/or certain Gβγ-subunits (9, 45). All AC isoforms can be stimulated by Gαs and by forskolin (FSK), a diterpene commonly used in cAMP analysis. Inhibition by Gαi is limited to a subset of AC isoforms. Inhibitory Gαisubunits are able to overcome the FSK-mediated activation of AC1, 3, 5, 6, 8, and 9, presumably because FSK binds near the catalytic site, distant from the Gαi binding site (16, 55). ACs also differ in their response to modulators such as Ca2+ and calmodulin. Indeed, a unified classification of ACs incorporates both such regulatory properties and their sequence similarities.
Taste buds are heterogeneous clusters of taste receptor cells, each of which responds to subsets of taste stimuli and contains distinct populations of cell signaling molecules. Modulation of cAMP has been demonstrated in the sensory transduction of stimuli that elicit sweet and bitter tastes (43, 52). However, the effect of glutamate on cAMP modulation in taste tissue has not been described previously. Glutamate elicits a taste quality termed “umami” that is potentiated by the simultaneous presence of the monophosphates of guanosine and inosine nucleotides. The potentiation is seen in electrophysiological measurements of afferent nerve responses, as well as in behavioral studies (12, 32, 38, 39, 51). Patch-clamp recordings of rodent taste buds suggest that the electrophysiological response to glutamate is mediated by a decrease in cAMP (5,22). Candidate GPCRs for the sensory transduction of umami taste have been proposed and include a truncated metabotropic glutamate receptor, taste-mGluR4, and the heterodimeric receptor T1R1/T1R3 (10, 21, 31). Many metabotropic glutamate receptors are known to couple to an inhibitory cAMP cascade (44). In Chinese hamster ovarian (CHO) cells, the decrease in cAMP is thought to be mediated by Gαi and can overcome FSK-stimulated AC activation. Taste-mGluR4 expressed in heterologous cells also exhibited Gαi-mediated inhibition of FSK-stimulated cAMP (10), but signaling downstream of T1R1/T1R3 has not been established.
Gustducin is a Gα-subunit that is found at high concentration in a subset of taste cells (28). Other Gα-subunits prominently expressed in taste cells include Gαs and Gαi2 (20). The involvement of gustducin in taste responses to many bitter and sweet compounds is supported through a variety of biochemical, genetic, and electrophysiological analyses (14, 24, 26). Recent behavioral experiments with knockout mice suggest that gustducin may also be involved in umami taste transduction (17, 37).
In early work, AC activity was demonstrated in taste tissue (42,43), although which AC isoforms are expressed remained to be identified. The balance between the activities of ACs and phosphodiesterases (PDEs) determines the level of cAMP in cells. Several types of PDE have been identified in taste tissue, and taste tissue exhibits high basal PDE activity (36). The molecular identities of taste-expressed PDEs have been difficult to confirm, although recent studies have shown the presence of PDE3 (28) and two isoforms of PDE 1A (29). Pharmacological studies also suggest the presence of PDE4 in taste tissue (19). Whether the cAMP modulation in response to taste stimuli is achieved by regulation of AC or of PDE or both has not been studied in detail.
We employed RT-PCR and immunohistochemistry to identify the ACs in taste tissue and demonstrate the presence of AC4, AC6, and AC8 in taste receptor cells. We also demonstrated modulation of cAMP in rat taste epithelium ex vivo by measuring cAMP levels in the presence of FSK, a broad AC activator, and IBMX, a broad PDE inhibitor. We further tested whether cAMP levels are altered when taste buds are stimulated with taste-effective concentrations of glutamate. We found that glutamate decreased cAMP. Downregulation of cAMP via ACs can be mediated through Gαi proteins. We examined the colocalization of detected ACs with Gαi proteins and found that 66% of AC8-immunopositive taste cells are also positive for gustducin, a Gα taste-specific protein, which belongs to a Gi/o/t/zsubfamily.
RNA was isolated from brain, circumvallate (CV), and foliate taste papillae, nontaste lingual tissue, as well as from enzymatically delaminated taste and nontaste epithelia (15). Poly(A+)RNA was isolated by affinity chromatography on oligo (dT) cellulose (FastTrack kit, Invitrogen), whereas total RNA was isolated by Nanoprep kit (Stratagene, La Jolla, CA). These RNAs were used as template for cDNA synthesis using Thermoscript RT or Superscript II (Life Technologies, Bethesda, MD). Degenerate primers were designed from the most conserved region of ACs, corresponding to the sequences LGDCYYC (5′-T/CTIGGIGAC/TTGC/TTACTACTG-3′) and KIKTIG (5′-C/G,T/CICCIATG/AGTC/TTTG/AATCTT-3′) in the large cytoplasmic loop and COOH-terminal domain, respectively. The PCR reactions were carried out using a “touch-up” protocol with annealing temperatures rising at 1°C/cycle from 37–47°C. RT-PCR products were concentrated by ethanol precipitation before digestion with XbaI and StuI (Life Technologies). Specific primers for AC2, AC3, AC4, AC5, AC6, AC7, AC8, and AC9 and the amplification conditions were as previously described (3). For amplification of AC1, primers were designed in the region conserved between mouse and human cDNA sequences (Table 1).
Frozen sections (30 μm) of rat CV (previously fixed in 4% paraformaldehyde and cryoprotected in 30% sucrose overnight at 4°C) were blocked in 5% BSA, 2% normal donkey serum, and 0.025% Triton X in phosphate-buffered saline (PBS) for 1 h at room temperature. Before blocking, endogenous peroxidase activity was quenched with 30% H2O2 for 30 min at 20°C. Sections were incubated overnight at 20°C with 1:300 diluted polyclonal antibodies against AC3, AC4, AC5/6, or AC8 (all from Santa Cruz Biotechnology, Santa Cruz, CA). The closely related isoforms AC5 and AC6 are detected by a common antibody (designated AC5/6). Goat anti-rabbit horseradish-peroxidase-labeled secondary antibody (1:1,000; Sigma) was used for AC3, AC4, and AC5/6, whereas donkey anti-goat antibody conjugated to Alexa 488 (1:1,000; Molecular Probes) was used to detect AC8. The signal for AC3, AC4, and AC5/6 was amplified and detected with tyramide-cyanine 3 (Perkin-Elmer, Boston, MA). For double immunostaining, rabbit anti-gustducin (1:4,000), mouse anti-Gαi2 (1:1,000), and goat anti-Gαi/o/t/z (1:1,000) were used, all from Santa Cruz Biotechnology. Secondary antibodies were donkey anti-rabbit conjugated to Alexa 594 for gustducin (1:1,500), donkey anti-mouse conjugated to Alexa 488 (1:1,000), and rat anti-mouse Cy3 antibody (1:150, Jackson ImmunoResearch), both for Gαi2 and donkey anti-goat conjugated to Alexa 488 (1:1,000) for Gαi-total, all from Molecular Probes (Eugene, OR) unless otherwise stated. The Gαi-total antibody (Gi/o/t/z) detects all members of the Gαi family. Images were obtained on an Olympus Fluoview confocal microscope. We confirmed using immunoblotting that each of the anti-ACs reacted against an antigen of the appropriate size in brain membrane extracts.
Male Sprague-Dawley rats 6- to 8-wk-old were used in all experiments. Epithelial sheets from CV papillae and adjacent nonsensory lingual surface were prepared by subepithelial injection of a protease cocktail containing 1 mg/ml collagenase D, 2.5 mg/ml dispase II, and 1 mg/ml trypsin inhibitor (15). Taste epithelial sheets containing taste buds were trimmed to remove most of the nontaste, surrounding epithelium, and von Ebner's glands, as described in detail (42). The sheets were then cut in the midline to yield equal left and right halves, which served as paired control and treated samples (Fig. 3 A). Pairs of epithelial sheets containing taste buds were treated with FSK (1 or 10 μM) in Tyrode's buffer for 9.5 min at 30°C, followed by 30 s of either Tyrode (control) or taste stimuli [glutamate and/or inosine 5′-monophosphate (IMP)], both in the continuing presence of FSK. The reaction was terminated and cAMP was extracted from tissue using 70% perchloric acid as described (19). cAMP was then quantified using an enzyme immunoassay kit (Amersham Pharmacia). In Fig. 4, cAMP concentration in taste-stimulated epithelium is expressed as a fraction of the cAMP in the paired control taste epithelial sheet. Protein levels in the acid-precipitated tissue were measured using a NanoOrange Protein Quantitation kit (Molecular Probes) and were used to normalize the absolute value of cAMP before comparing treated vs. control samples. As an additional control, we used nontaste epithelial sheets (devoid of taste buds) and measured cAMP, with and without FSK.
We designed degenerate PCR primers to two conserved sequences (LGDCYYC and KIKTIG) within the C1 and C2 domains of types AC1–9 from mouse and rat. The RT-PCR product from brain RNA yielded a heterogeneous product that included an ≈1,720-bp band (the size expected for AC1 and 8), a broad band of 1,790–1,836 bp (AC2–7), and a band at 2,010 bp (AC9). This is consistent with the known presence of all ACs in the brain (27). Poly (A) RNA from CV and foliate papillae was reverse transcribed, used in a parallel amplification reaction, and yielded bands in the 1,721- to 1,836-bp range (Fig. 1 A). To identify ACs present in taste receptor cells, we used diagnostic restriction digestion of the degenerate PCR product. When the taste amplification product was cut with XbaI, a diagnostic band of 1,356 bp was detected (Fig. 1 B, arrowhead). A unique restriction site for XbaI is found only in AC4, demonstrating the presence of this AC in taste tissue. The presence of DNA not digested by XbaI indicated the presence of additional ACs. A similar analysis with StuI digestion suggested the presence of AC3/AC8 (∼1,575 bp) and AC5/AC6 (∼1,550 bp).
Taste papillae contain many cell types in addition to taste receptor cells. To minimize the presence of extraneous cell types, we extracted RNA from trimmed taste epithelium, delaminated from underlying lingual tissue. This RNA was reverse transcribed and used in RT-PCR. Using primer pairs specific for each AC isoform, we confirmed the presence of all nine ACs in brain RNA. By contrast, RNA from CV epithelium consistently revealed strong RT-PCR bands for AC4, AC6, and AC8. RT-PCR products corresponding to AC2, AC5, and AC9 were occasionally detected in some taste samples, whereas RT-PCR products for AC1, AC3, and AC7 could not be detected.
RT-PCR results suggested that ACs are present in taste cells. Next, we wanted to confirm the presence of AC isoforms in taste cells at the protein level. We carried out immunohistochemical analysis of the prominently detected isoforms, AC4, AC5/6, and AC8, as well as AC3, which was seen in RT-PCR from taste papillae (but not from delaminated epithelium). AC3 immunoreactivity was confined to the subepithelial-stroma and probably represents nerve fibers based on the elongate profiles visible (Fig.2 A). No immunoreactivity for AC3 was detected in taste buds, in agreement with our RT-PCR results with delaminated epithelium (Fig. 1 C). Immunohistochemistry with an antibody specific for AC4 revealed that a small subset of spindle-shaped taste receptor cells express this isoform. The majority of taste buds in each section contained two to five stained cells in our 30-μm sections. Antibodies specific for the closely related sequences, AC5 and AC6, and for AC8 similarly showed subsets of taste receptor cells stained (Fig. 2, C–F). Although immunostaining for AC8 was readily detected using fluorescent secondary antibody, the signal for AC4 and AC5/6 required a tyramide-based amplification for stronger visualization. In all experiments, control sections processed in parallel showed no signal if primary AC-specific antibodies were omitted (Fig. 2, G andH) or if such antibodies were preincubated with the immunogenic peptide (not shown). The cells immunopositive for AC4 and for AC5/6 were spindle-shaped and slender. Cells immunoreactive for AC8 were broad and had rounded large nuclei (Fig. 2 F, arrowhead). Morphologically distinct taste receptor cell types have been observed to express distinct markers (53). Although we have not conducted detailed morphometric analyses, our results suggest that the various AC isoforms may exist in different taste receptor cell types.
Taste cells are functionally heterogeneous and may contain transduction components for responding to different sets of stimuli. Consequently, when taste tissue is disrupted for biochemical analyses, nonphysiological combinations of G proteins (which are found in both membrane and soluble compartments), receptors, and effector enzymes may yield responses not found in intact tissue. Thus we stimulated intact taste cells before measuring cAMP. We measured cAMP in isolated epithelium of the CV papillae and also in the adjacent nonsensory epithelium (devoid of taste cells). Basal cAMP level in CV epithelium was 20.5 ± 3.8 fmol/μg protein (n = 8), consistent with previous experiments on unstimulated CV epithelia (19, 42). Isolated epithelium from CV papillae was cut in half along the midline (Fig.3 A), and pieces were treated for 9.5 min with either FSK, to raise baseline cAMP levels, or with Tyrode's (control). In taste epithelium, 1 and 10 μM FSK increased cAMP approximately threefold (to 59.9 ± 7.2 fmol/μg protein,n = 6) and 10-fold (to 203 ± 41.8 fmol/μg protein, n = 5), respectively (Fig. 3 B). When taste epithelia were incubated with both IBMX and FSK, cAMP levels increased to 1,830 ± 611 fmol/μg protein. Such an increase confirms the presence of high basal PDE activity in CV epithelium, as previously suggested (36).
A candidate taste receptor for umami stimuli, taste mGluR4, showed glutamate-dependent cAMP decrease in FSK-pretreated cells (10). Other mGluRs also show similar coupling to an inhibitory cAMP cascade (44). Hence, we measured cAMP in native taste epithelia pretreated with FSK. After FSK pretreatment, half of a CV epithelium (Fig. 3 A) was stimulated with glutamate for 0.5 min while the other half was not, and served as a control. The range of glutamate concentrations used was based on the known threshold for glutamate preference in rodent behavioral studies (32). Within the range of glutamate concentrations tested (1–20 mM), the response was greatest at 20 mM glutamate (cAMP concentration was 65.7 ± 8.1% of that in paired control taste epithelium treated with only 1 μM FSK; Fig.4 A). The response was taste tissue specific, insofar as glutamate (at all tested concentrations) did not induce a decrease in cAMP in nontaste epithelium. cAMP levels upon glutamate treatment in nontaste epithelium sheets were 117.6 ± 11.7% of paired control, nontaste epithelium (n = 16, not significantly different from 100%).
We also carried out stimulations on CV epithelium preincubated with 10 μM FSK. Again, 20 mM glutamate was found to trigger a decrease in cAMP levels to 75.2 ± 3.6% of FSK-treated control (n = 8; Fig. 4 B). Higher concentrations of glutamate led to high variability in cAMP concentrations in both taste and nontaste samples, perhaps stemming from osmotic or other nonspecific effects. These were not pursued further.
In electrophysiological and behavioral assays, the taste response to glutamate is potentiated by the simultaneous presence of monophosphate nucleotides of inosine and guanosine. We measured cAMP in CV epithelium after stimulation with 20 mM glutamate, 0.5 mM IMP, and a mixture of both (Fig. 4 B). Stimulation with IMP alone did not lead to an appreciable change in cAMP concentration. Compared with glutamate or IMP alone, the mixture caused a significantly greater decrease in cAMP. In contract, in nontaste epithelium stimulated with glutamate and IMP, cAMP levels were 94 ± 9% of paired control, nontaste samples (n = 9). The stereoisomer d-glutamate does not elicit umami taste. We found that 20 mM d-glutamate also did not modulate cAMP in CV epithelium at a concentration that produces a robust signal with l-glutamate (Fig.4 C).
Taste cells may mediate decreases of cAMP by regulating AC activity through Gαi proteins. In taste cells, the prominent Gα-subunits of the Gαi family are Gαi2and Gα gustducin. Hence, to determine whether ACs detected in taste cells colocalize with Gαi proteins, we used double immunohistochemistry.
Gustducin expression is well characterized in taste buds (28). Because antibodies against gustducin, AC4, and AC5/6 were all raised in rabbit, we used antibodies against Gαi2 and against the entire class of Gαiproteins (i.e., Gαi/o/t/z) in addition to anti-gustducin. The mRNA for Gαi2 is found in all gustducin-positive cells (1). We confirmed that this overlapping expression also applies for Gαi2 protein by double labeling immunohistochemistry for gustducin and Gαi2 (Fig. 5,A–C). We also confirmed that a common set of cells is labeled with antibodies against Gαi2 and Gαi-total (Fig. 5,D–F). Hence, we conclude that all three antibodies against the Gαi proteins detect the same cells and can be used interchangeably.
The expression profiles of AC4, AC5/6, and AC8 with Gαiproteins are presented in Fig. 6. AC4-immunopositive cells did not colocalize with Gαi2(Fig. 6 C). Although the overlay panel shows some yellow, this appears to reflect separate cells that overlap in this section. Note the distinct cellular shapes in the green and red panels where each antigen is viewed separately (Fig. 6, A andB). Taste cells strongly labeled with anti-AC5/6 did not overlap with Gαi-total immunostaining. We noted that some cells weakly stained with anti-AC5/6 did show robust signal for Gαi-total (Fig. 6, D–F). The staining for AC5/6 was not sufficiently robust to quantify the overlap of expression. In contrast, immunoreactivity for AC8 appeared to colocalize with immunoreactivity for gustducin. To quantify this colocalization, immunopositive cells that were well defined, with visible nuclei, were counted in CVs from three rats. From each CV, 12 sections were randomly selected and three taste buds from one crypt of each section were counted. Out of 200 cells positive for AC8, 66 ± 19% were also strongly positive for gustducin. Conversely, out of 203 cells immunopositive for gustducin, 57 ± 8% were also strongly positive for AC8. In the CV, only 10–20% of all taste cells are gustducin positive (7). Hence, the overlap between gustducin and AC8 appears at a probability higher than chance. We also observed a similar colocalization pattern of AC8 vs. Gαi2 (data not shown).
Numerous AC isoforms exist that are subject to distinct regulatory controls and exhibit subtype-selective sensitivity toward physiological modulators. These ACs enable cells to fine tune cAMP levels and integrate signals between transduction pathways. Chemosensory systems are known to express several AC isoforms. The essential role of AC3 in olfactory neurons (49) and the presence of AC2 in vomeronasal neurons (4) have been described. The present study demonstrates for the first time that AC4, AC6, and AC8 are prominently expressed in taste tissue, both at the mRNA and protein level.
The second messenger, cAMP, is implicated in transduction of sweet and bitter taste qualities. For instance, sucrose increased cAMP levels in CV taste buds (42, 43), whereas denatonium and strychnine, two bitter substances, induced rapid reductions in cAMP (52). Electrophysiological studies in mammalian taste buds have suggested that modulation of cAMP levels may play a role in sour transduction (25). The role of cAMP in glutamate taste transduction is suggested by the damping of membrane currents when cAMP analogs were introduced into taste cells (23). We show here that rat taste cells undergo a decrease in cAMP levels when stimulated with 1–20 mM glutamate, a concentration range compatible with behavioral and afferent nerve responses in this species (41, 51). This decrease is taste specific because it is not present in nontaste epithelium. The potentiation of this response by IMP, the lack of response to d-glutamate, and the lack of response to umami stimuli in nonsensory epithelium all suggest that the cAMP modulation represents umami taste transduction.
The functional properties of AC4, AC6, and AC8 suggest that each could readily be involved in transduction pathways in taste buds to signal one or more taste qualities. In taste cells, increases of cAMP could be mediated via AC4, AC6, or AC8, depending on their colocalization with Gαs. Furthermore, AC4 can be activated by Gβγ-subunits (45, 46) to increase cAMP in a Gs-independent manner downstream of Gi-, Go-, or Gq-coupled receptors (47). Also, AC8, principally a brain-expressed isoform (40), can be stimulated by calcium/calmodulin (50). The presence of calcium-sensitive AC8 in taste cells suggests a means of resolving the dual signaling pathways that have been noted for sweet transduction (26). In gustducin-positive cells, the βγ-partners of gustducin lead to activation of PLCβ2, increased IP3, and a consequent elevation of intracellular calcium. In our experiments, AC8 colocalized with gustducin. This would explain how saccharin increases both IP3 and cAMP (30). AC8 is stimulated via capacitative calcium entry (13), a mechanism that is also compatible with the presence of a Trp channel (35) and capacitative calcium currents in taste cells (33). Recently defined splicing variants of AC8 are known to be differentially sensitive to such regulation (8). However, our RT-PCR primers could not discriminate between these types.
Decreases of cAMP have been noted in response to certain taste stimuli, but attributing these to particular ACs is difficult. Direct inhibition of AC4 by Gαi2 proteins has not been described; hence, it is not surprising that colocalization of AC4 and Gαi2 was not observed. In many tissues, the closely related AC5 and AC6 are responsible for downregulation of cAMP. For instance, AC6 is inhibited by low concentrations (<1 μM) of calcium (11) and by Gαi (54). We found that taste cells that stained prominently for AC5/6 did not strongly stain for Gαi-subunits. However, AC6 is also inhibited by certain Gβγ combinations, notably β1γ2 (2). Because the functional properties of Gβ1γ2 are very similar to those of Gβ1γ13 (6) and the latter dimer is present in many taste cells (18), these G proteins may serve to inhibit AC6. Although inhibitory regulation of AC8 has not been examined, we postulate by analogy with the homologous AC1 that AC8 might be inhibited by Gβγ-subunits released from Giproteins (48). The substantial overlap between AC8 and both gustducin and Gαi2 suggests that AC8 may be the principal target for inhibitory regulation in taste cells. Whether such inhibition arises from the Gα or the Gβγ-subunits remains an open question.
We thank Laura Cooney for technical assistance with cAMP measurements, Elizabeth Pereira for assistance with immunohistochemistry, and Dr. Alejandro Caicedo for early help on confocal microscopy.
This work was supported by a grant from National Institutes of Health/NIDCD DC-03013 to N. Chaudhari.
Address for reprint requests and other correspondence: T. Abaffy, Dept. of Physiology and Biophysics (R430), Univ. of Miami School of Medicine, 1600 NW 10th Ave., Miami, FL 33136 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published February 26, 2003;10.1152/ajpcell.00556.2002
- Copyright © 2003 the American Physiological Society