Differential Expression of RNA and Protein of the Three Pore-forming Subunits of the Amiloride-sensitive Epithelial Sodium Channel in Taste Buds of the Rat

Abstract

Salt taste signals from the rat anterior tongue are probably transduced via epithelial sodium channels (ENaCs) residing in the apical cellular pole of taste cells. The signals are blocked by mucosal amiloride in low μM concentrations. In contrast, the rat vallate papilla does not contribute to amiloride-blockable salt taste. Two approaches were used to probe for the three subunits of ENaC in the anterior and posterior tongue of the rats in sodium balance. (a) Immunohistochemistry with antibodies against ENaC subunits and against amiloride binding sites. In the anterior tongue, reactivity for α-, β-, and γ-subunits was present in taste buds and lingual epithelium. In the posterior tongue vallate papilla, reactivity for α-subunit and for amiloride binding sites was easily demonstrable, whereas that for β-subunit and especially for γ-subunit was weaker than in the anterior tongue. (b) RT-PCR techniques were used to probe for the presence of ENaC subunit mRNA. In isolated taste buds of the anterior tongue, mRNA of all three subunits was found, whereas in isolated taste buds of the vallate papilla only mRNA of the α-subunit was easily detectable. That of β- and γ-subunits was much less abundant. RNA of all three subunits was abundant only in taste buds of the anterior tongue. Therefore, subsets of elongated taste cells do express ENaC, but regional differences exist in the transcription and expression of subunits. The regional differences suggest that amiloride-sensitive salt taste, which requires all three subunits, is present in the anterior but not the posterior tongue of rats, as functional studies indicate.

In rodents, the salt taste which is mediated by the anterior tongue is specific for Na⁺ and Li⁺ ions and is sensitive to amiloride in low μM concentrations (Heck et al. 1984; Ye et al. 1993, 1994; Spector et al. 1996). Whole-cell patch-clamp experiments with isolated taste cells have shown Na⁺ inward currents blockable by amiloride with inhibition constants in the sub-micromolar range (Doolin and Gilbertson 1996), and recordings from taste pores have demonstrated apical Na⁺ inward currents that were 50% blocked at 1 μM amiloride (Avenet and Lindemann 1991). These results suggested that an amiloride-sensitive epithelial Na⁺ channel, most likely a member of the ENaC family (Canessa et al. 1993, 1994; Lingueglia et al. 1993; Lingueglia et al. 1995; Waldmann et al. 1995a,b, 1997) is present in the apical plasma membranes of taste receptor cells, and provides the “receptor” element for the Na⁺ component of salt taste. Previous in situ hybridization has attempted to discriminate between ENaC localized in the lingual epithelium surrounding the taste papillae and ENaC actually present in taste cells (Li et al. 1994). Although this work clearly showed the presence of ENaC RNA in the lingual epithelium, the demonstration of ENaC in taste cells was less successful. The difficulty may be that a low abundance of ENaC RNA in the taste cells causes a poor spatial resolution of the in situ images. We reinvestigated the occurrence of ENaC in taste cells by using immunocytochemistry as well as molecular techniques applied to taste bud-enriched sheet preparations and to taste buds totally separated from the surrounding epithelium. The taste-cell-specific G-protein α-gustducin was used as a reference molecule, providing an additional way of discriminating taste tissue from nontaste tissue (McLaughlin et al. 1992).

Testing for the presence of Na⁺ channel RNA with a reverse transcriptase-PCR protocol, we found that RNA of α-, β-, and γ-subunits of ENaC was present in taste buds of fungiform papillae (anterior tongue), whereas predominantly α-ENaC was found in the vallate papilla (posterior tongue). In addition, immunore-activity for α-, β- and γ-subunits was found in anterior tongue taste buds and reactivity for α- and β-subunits in the foliate and vallate papillae. Therefore the results obtained with the two independent methods point to differences in RNA abundance and in expression of the subunit proteins when areas of the tongue are compared.

A portion of our results concerning the vallate papilla was presented at the 1997 ISOT XII Symposium in San Diego (Lindemann et al. 1998).

Materials and Methods

Isolation of Taste Tissues

Tongues were isolated from freshly sacrificed rats (Sprague-Dawley, 180–240 g, in sodium balance) and stored in ice-cold Tyrode's solution for at least 10 min. Three hundred μl of Tyrode's solution [in mM: NaCl 140, KCl 5, CaCl₂ 1, MgCl₂ 1, HEPES 10 (N-(2-hydroxyethyl)piperazine-N′-(2-ethanesulfonic acid)), adjusted to pH 7.4 with NaOH] containing 1 mg dispase I/ml (Boehringer; Mannheim, Germany), 1 mg collagenase A/ml (Boehringer), and 1 mg trypsin inhibitor/ml (Sigma; Deisenhofen, Germany) was injected under the anterior lingual epithelium (aLE), the vallate papilla (VP), and the surrounding epithelium of the posterior tongues (pLE), using 1-ml insulin syringes with needles of 0.4-mm diameter. Tongues were incubated at room temperature (RT) for 25 min. When the epithelia had loosened from the tongue muscle layer, they were peeled with fine forceps and washed in Tyrode's solution.

Isolation of Taste Buds

Anterior Tongue. We followed the published procedure (Béhé et al. 1990) with small modifications. Briefly, the peeled anterior epithelium was placed in a Petri dish, the bottom of which was covered with cured Sylgard, and pinned down with the inner side facing upwards. The tissue was covered with a low-Ca²⁺-Tyrode's solution [containing no added Ca²⁺ and Mg²⁺ but 1 mM BAPTA (1,2-bis(2-aminophenoxy)ethane N,N,N′,N′-tetraacetic acid)] for 10-20 min at RT. Taste buds were individually removed from fungiform papillae by suction with a transfer pipette (a glass pipette having an opening of 100 μm). Up to 90 buds could be isolated from one anterior tongue. They were pipetted into a collecting dish containing Tyrode's. After removal of all buds from a given epithelial sheet, its area was measured and the tissue processed as “aLE.” Histological controls have shown that buds isolated from the anterior tongue as described above often included some epithelial cells that were attached to the vicinity of the taste pore. Thus these isolated buds were not entirely free of epithelial contamination.

Vallate Papilla. The epithelium around the VP was removed, yielding the epithelial preparation pLE. From the VP, the superficial part, which has only few taste buds, was cut off and discarded. It is known that the number of buds present in a rat vallate papilla increases from 260 (3 weeks old) to 610 (13 weeks old) (Hosley et al. 1987; Norgren et al. 1989). For the deep part of the cleft, which has the highest density of buds, we counted in serial sections 280 taste buds for our animals. This part was cleaned from the attached ducts of von Ebner glands, yielding the taste bud-enriched sheet preparation VPt (Striem et al. 1991). Isolated taste buds were prepared from the VPt by incubation in the low-Ca²⁺-Tyrode's (see above) for 10-30 min. The VPt was then grabbed with a small forceps, transferred to a collecting dish containing Tyrode's, and shaken until the taste buds fell off (Bernhardt et al. 1996). The isolated buds were visible as small “specks” under darkfield illumination. Alternatively, the buds were individually removed from the tissue using the 100-x m transfer pipette (see above).

Counting and Lysis of Buds. Using a binocular microscope and a regular light microscope, buds were inspected and those that appeared fragmented were picked out with a transfer pipette (having an opening of 100 μm) and discarded. The remainder were counted, individually transferred onto a clean piece of parafilm, and counted again. Excess medium was removed from the paraffin with the transfer pipette and the buds covered with 100 μl of lysis buffer (4 M guanidine thiocyanate, 25 mM tri-sodium citrate, pH 7.0, 0.5% lauryl-sarcosyl, 0.72% 2-mercaptoethanol). Using a 200-μl pipette, the lysate was transferred to Eppendorf tubes, vortexed, frozen, and stored at −20C.

RNA Preparation

Yeast transfer RNA (1 μg) was used as a carrier. Sequentially, 0.1 volume of 2 M Na-acetate, pH 4.0, 1 vol of acidic phenol, and 0.2 vol of chloroform was added, the mixture being vortexed after each addition. After incubation at 4C for 15 min and centrifugation at high speed, the aqueous phase was recovered and RNA was precipitated with 1 vol of isopropanol. After centrifugation, the pellet was resuspended in 100 μl of lysis buffer and RNA was reprecipitated with isopropanol. After two washings with 75% cold ethanol, the pellet was dried, dissolved in water, and either stored at −20C or used directly for RT-PRC. At the end of the purification, pLE, and VPt RNA, available in higher amounts, were quantified by optical density and the integrity was checked on a 1% agarose gel.

Polymerase Chain Reaction

For the lingual epithelium, 100 ng of total RNA was used for RT, performed either with mouse Moloney virus reverse transcriptase (Pharmacia; Uppsala, Sweden), or RT Superscript II (Gibco; Gaithersburg, MD). To compare the signal in fungiform and vallate papillae, PCR reactions were carried out with an equivalent of one taste bud per reaction (one VPt contained about 280 taste buds; see above). Genomic DNA did not contribute to the signal as suggested by two protocols. In the first, RNA was treated in parallel in the presence and absence of reverse transcriptase, and the material was then used for PCR. In the absence of reverse transcriptase there was only minimal amplification of fragments having the expected size. Subsequently, information about the genomic organization of the human genes encoding α-, β-, and γ-ENaC revealed that the three human genes are characterized by a very similar intron-exon organization (Thomas et al. 1996) and (Alexandre Persu and Xavier Jeunemaitre, personal communication). Assuming a similar genomic organization in the rat, oligonucleotides belonging to distinct exons were used for further controls. PCR reactions led in this case to the amplification of two bands, either specific for the genomic DNA (characterized by a longer size owing to the presence of at least one intron) or for the reverse-transcribed complementary DNA.

The primers used for DNA amplification were as follows.

For the α-subunit AR1306sens: 5′-ATTACGGCGACTGTACTGAGAA-3′, located in exon 7, and AR1618asns: 5′-CAATCCTGGGACTTCACAGATG-3′, located at the junction between exon 10 and exon 11. This pairing amplified a 312-bp fragment (as the anti-sense oligonucleotide AR1618asns, located in the middle of an intron-exon boundary, was unable to amplify genomic DNA at the annealing temperature used).

For the β-subunit. BR1197sens: 5′-CAACACGACCTATTCCATCC-3′) was located in exon 7. BR1663sens: 5′-TCGGTGCTGTGCCTCATTG-3′ and BR1921asns: 5′-AGCCTCAGGGAGTCATAGT-3′ were both located in exon 13. The pairing (BR1663sens-BR1921asns) amplified a 258-bp fragment but could not be used when contamination by genomic DNA was observed, because both sequences were located in the same exon. The pairing (BR1197sens-BR1921asns) amplified a single 724-bp fragment specific for the cDNA. No genomic DNA could be amplified under the experimental conditions used here, owing to the large intronic sequences present between the two primers.

For the γ-subunit. GR1616sens: 5′-ACAAAGACCTGAACCAAAG-3′ was located in exon 12, and GR1902asns: 5′-GCATAGCAGAGGTAAAAGT-3′ was located in exon 13. They were amplifying a 286-bp fragment of cDNA and a 1200-bp fragment of genomic DNA.

For β-actin. ACTINsens: 5′-TCCTAGCACCATGAAGATC-3′ and ACTINasns: 5′-AAACGCAGCTCAGTAACAG-3′ amplified a 313-bp genomic fragment and one fragment of about 200 bp transcribed from RNA.

For α-gustducin. GU401sens: 5′-CAATCCGAGAAGTAGAGAGG-3′ and GU869asns: 5′-GCTGTTGAAGAGGTGAAGAC-3′. They were amplifying a 468 bp fragment of cDNA (Höfer et al. 1996).

After reverse transcription, samples equivalent to one taste bud (about 1 ng total RNA), or to 100 buds (for vallate tissue or epithelium) were denatured for 120 sec at 95C. Twenty-five, 30, 35, or 40 cycles of amplification were performed with the following parameters: denaturation at 95C for 30 sec; annealing at 55-60C for 60 sec; extension at 72C for 60 sec. Reactions were performed in the presence of 300 ng of each oligonucleotide and 0.08 U of Taq DNA polymerase (Promega; Madison, WI). Fifty percent of the amplified product was analyzed on a 2% agarose gel, either by Southern blot hybridization with specific probes or by direct ethidium bromide staining. The amplified PCR products were sequenced and the expected partial sequences of α-, β-, and γ-ENaC were found.

Semiquantitative PCR

During the exponential phase of a PCR experiment, the signal measured after n cycles is given by

${\text{A}}_{\text{n}}={\text{A}}_0*{(1+\text{E})}^{\text{n}}$

where A₀ represents the initial amount of template and E the efficacy of the amplification reaction (with 0<E<1). In preliminary experiments performed with pure plasmids, it was checked that the different pairings used for α-, β-, and γ-ENaC were characterized by similar efficacies. E was estimated to 0.69 ± 0.10, 0.64 ± 0.11, and 0.82 ± 0.10 for α-, β-, and γ-ENaC, respectively. No statistical difference was found among the three values. Therefore, under the experimental conditions used, the relative abundance of the three species of RNA in a given tissue could be calculated from the signal measured during any cycle of the exponential phase of the PCR reaction.

Southern blot hybridizations were quantified on a Fuji phosphor imager, using analytical software developed by the company. Quantification of some experiments by direct ethidium bromide staining was also performed. In that case, a standardized protocol was developed to ensure that the same volume of agarose was cast, that an equal amount of ethidium bromide was added to the gel, and that a similar time of migration on the gels and of exposure of the films was used. Quantification was then performed using the NIH Image software (http://rsb.info.nih.gov/nih-image/download.html).

Detection of Splice Variants of ENaC

To probe for the splice variant described for rat vallate papilla (Li et al. 1995), two oligonucleotides were used: AR1327S: 5′-TGTGCATTCACTCCTGCT-3′, located in exon 7, and AR1592asns: 5′-CAATCCTGGGACTTCACA-3′, located in exon 10. PCR was conducted as described above.

Detection of Other Members of the ENaC Gene Superfamily

The presence of two other members of the ENaC-FaNaCh-degenerin gene superfamily was probed by PCR using the following pairs of oligonucleotides. For ASIC1-7S: 5′-ATTGCTCTTCCCATCTCTAT-3′ and for ASIC1-10AS: 5′-TTCAAGGCCCATACCTAAGT-3′ (Waldmann et al. 1997). For MDEGI-S: 5′-CTGTCGGATCGACTGTGAGA-3′ and for MDEGI-AS: 5′-CCAAGTAAGGCAGCAACTTC-3′ (Waldmann et al. 1996). PCR was conducted as described above.

Immunohistochemistry in Sections

Tongues were removed and tissue containing taste buds (anterior tongue, foliate papillae, circumvallate papilla) was excised and immersed in Bouin's fixative (prepared freshly). After 3 days of fixation at 4C, tissues were dehydrated and embedded in paraffin (Vogel; Giessen, Germany) of 60C and were sectioned to a thickness of 7 μm with a standard microtome. Incubation with primary antibody was overnight. Polyclonal antibodies were raised in rabbits against peptides of sequences specific for one of the subunits of ENaC. The sequence Q₄₄-G₅₈ was used for α-rENaC (Lingueglia et al. 1994; Renard et al. 1994, 1995), G₆₅₅-P₆₆₉ for β-hENaC (Voilley et al. 1998), and Y₁₂₇-S₁₄₃ for γ-rENaC (Renard et al. 1995). The first two antigens are located at the cytosolic aspects and the third antigen at the extracellular aspect of their respective subunits. Specificity of the antibodies was checked with Western blots and immunoprecipitation using homogenates of cells after heterologous ENaC expression (data not shown; and Renard et al. 1994). Secondary antibody (goat anti-rabbit, diluted 1:50 and used for 2 hr at RT) was conjugated to peroxidase (Dako; Glostrup, Denmark). Substrate for peroxidase was 3,3′-diaminobenzidine (DAB free base; Sigma, Berlin, Germany) and H₂O₂, reaction time 4-16 min at RT. This reaction yielded a brown stain.

In addition to the three anti-peptide antibodies targeting ENaC subunits, we also used an antibody directed against α-gustducin, generously provided by Dr. R.F. Margolskee (McLaughlin et al. 1992), and one monoclonal antibody directed against amiloride binding sites. This anti-idiotypic antibody, RA 6.3, was generously provided by Dr. T. R. Kleyman (Kleyman et al. 1991; Lin et al. 1994a, b). Finally, a polyclonal antibody directed against histaminase (diamine oxidase), an amiloride binding protein unrelated to ENaC, was used as a corollary to RA 6.3 (Novotny et al. 1994).

The primary antibodies were used at the following final dilution: anti-α-rENaC 1:100, anti-β-hENaC 1:400, anti-γ-rENaC 1:400, RA6.3 1:50, anti α-gustducin 1:1000, and anti-histaminase 1:100.

Double Staining. Sections were first incubated overnight with the polyclonal antibody directed against α-rENaC. The staining sequence was continued (above) using goat anti-rabbit linked to alkaline phosphatase as the secondary antibody and Fast Blue RR (Sigma) as the chromogen. The sections were covered with gelatin and coverglass and the distribution of blue precipitate was imaged. Then the coverglass and gelatin were removed and the section was exposed overnight to the monoclonal antibody RA6.3 directed against amiloride binding sites. The staining sequence was continued, using goat anti-mouse linked to peroxidase as the secondary antibody. The new staining pattern, i.e., the brown color of the peroxidase-DAB reaction, was also imaged. Because the two primary antibodies used for double labeling were from different species, crossreactions of secondary antibodies were not a problem. The two primary antibodies were occasionally added simultaneously. In this case, too, the two staining reactions were executed in sequence, the peroxidase reaction being the last of the two. Inhibition of the phosphatase reaction by substrates of the peroxidase was thus avoided.

Controls Using Peptides. Preabsorption experiments were conducted to test for specificity of the antibodies. The antibodies (working dilution) directed against α-rENaC and γ-rENaC were preincubated with their corresponding peptide (up to 2 mg/ml) at RT for 24 h. The staining procedure was then carried out as described above.

Immunohistochemistry in Sheet Preparations

Freshly peeled anterior epithelia (see above) had areas up to 9 × 9 mm and contained up to 90 fungiform papillae with taste buds that could easily be seen when viewed with a binocular microscope from the “inner” side. In a Petri dish with a cast silicon bottom, the sheets were stretched out with needles and covered with Bouin's fixative (prepared on the same day) for 48 hr at 4C. Sheets were cut into pieces having about 10 fungiform papillae each and were transferred to a beaker containing 200 ml of working buffer [TRIS (tris(hydroxymethyl)aminomethane) buffer, pH 7.4, containing 5 mM NaN₃ and 1% fish gelatin (Sigma)] which was magnetically stirred overnight at 4C. The sheets were individually placed in 1 ml of working buffer containing primary antibody and were stored for 24 hr at 4C, then washed for 8 hr in 25 ml of working buffer at 4C. Sheets were incubated individually in 1 ml of working buffer containing secondary antibody and stored overnight at 4C, then washed for 5 hr at 4C. Primary antibodies were used at the following final dilution: anti-α-rENaC 1:250, anti-β-hENaC 1:250, anti-β-rENaC 1:250, RA6.3 1:50, anti α-gustducin 1:250. Secondary antibodies were affinity-purified goat anti-rabbit and goat anti-mouse conjugated to the fluorescent dye Cy3 (Jackson ImmunoResearch; West Grove, PA). Alternatively, secondary antibodies conjugated to peroxidase (Dako) were used and staining performed as described for sections.

After staining, the whole mounts were placed between two large coverglasses with a drop of working buffer and were inspected in a fluorescence microscope. The objective was a Nikon Fluor X40 (oil immersion) of 1.3 NA. The fluorescence of individual buds was first documented from the inner side in an optical plane that intersected the long diameter of the bud roughly in the middle. The sheet was then turned around for imaging of taste pores. Each plane was also documented with transmitted light. Stacks of images were taken by changing the focal plane of the microscope in steps of 0.5 or 1 μm, using a feedback-controlled motor (RS8, HA-400; Harmonic Drive, Limburg, Germany) activated by the imaging software (T.I.L.L. Photonics; Planegg, Germany). For controls, the fixed and washed whole mounts were incubated with preimmune serum instead of primary antibody and the subsequent steps performed as described above. In addition, whole mounts were imaged with a confocal microscope (Bio-Rad; München, Germany) to confirm the intracellular location of the immunoreactivity.

Results

Immunohistochemistry with Sections

A survey of immunoreactivity for α, β-, and γ-subunits of ENaC is shown in Figure 1. In the anterior tongue (upper row), α- and γ-subunit-like reactivity was easily observed in the taste buds and in the epithelium of the fungiform papillae, whereas β-subunit-like reactivity was restricted to a few taste cells. In the foliate papillae (center row), α- and β-subunit-like reactivity was easily seen whereas γ-subunit-like reactivity was weak. In the vallate papilla (bottom row) α-subunit-like reactivity was easily observed. Intensity of staining for the β-subunit was less, provided that incubation times for the primary antibody were 24 hr. However, the intensity became more comparable to that of α-subunit when incubation times were extended to 48 hr. The reactivity for γ-subunit was very low or absent in the vallate papilla, even at long incubation times (Figures 1I and 2D).

Figure 1

Vertical sections through taste papillae of the rat, showing α-, β-, and γ-ENaC-like immunoreactivity. (A-C) Fungiform papillae (anterior tongue). (D-F) Foliate papillae. (G-I) Vallate papillae. In each row, the sequence is α-, β-, γ- from left to right. In the bottom row (vallate papilla), the reaction time for peroxidase required for the β-signal (H) was increased fourfold over than that for the α-signal (G), to yield the staining intensities shown. In contrast, immunoreactivity for γ-subunit failed to highlight taste buds or taste cells in the vallate papilla even when long reaction times were used.

Control experiments with antibodies that had been preincubated with their antigenic peptide for 24 hr (see Materials and Methods) showed increasingly weaker immunostaining when the peptide concentration was increased. At 2 mg peptide/ml, reactivity was minimal (Figures 2B and 2E). Controls omitting primary antibody are shown in Figures 2C and 2F. Furthermore, positive controls were performed with sections from rat kidney and rat submandibular gland. The expected targets (Duc et al. 1994; Renard et al. 1995) were stained, i.e., the outer medullary collecting duct (Figures 2G-2I) and the striated ducts of the salivary gland (Figures 2J-2L). Unexpected targets, such as cellular nuclei and nerve fibers, remained unstained in salivary gland, kidney, and taste tissues. Interestingly, the straight segments of the proximal nephron showed α-subunit immunoreactivity localized in the brush border (data not shown). However, we do not consider this target unexpected because of the recent demonstration of ENaC mRNA in the straight proximal segment (Willmann et al. 1997).

According to electrophysiological evidence (Avenet and Lindemann 1991), the ENaC is expected on the apical membrane of the anterior tongue taste buds. However, the upper row of Figure 1 shows that the reactivity was, to a great extent, intracellular. Occasionally the reactivity for α- or β-subunit emphasized cellular borders (Figures 1B, 1E, and 1G), suggesting that the signal was at or near the basolateral membrane of the cells. Whether apical reactivity was present in addition was difficult to ascertain because the sections seldomly passed through the taste pore. However, when sections were cut tangentially through the anterior lingual epithelium, taste pores were sometimes revealed. They contained immunoreactivity for ENaC subunits, as shown in Figure 3. (An alternative procedure, which used epithelial sheet preparations, is described below.)

Figure 2

ENaC-immunostaining for (A) α- and (D) γ-subunits in rat VP. Corresponding negative controls using preabsorption with peptide-immunogens at 2 mg/ml for 24 h (B,E). Negative controls omitting primary antibody (C,F). Positive controls of ENaC immunoreactivity in the rat kidney outer medullary collecting duct (G-I) and in striated ducts of the rat submandibular gland (J-L) using anti-α (G,J), anti-β (H,K) and anti-γ (I,L).

The monoclonal antibody RA 6.3, directed against amiloride binding sites, labeled a subset of elongated taste cells. The label was mostly intracellular, as was the pattern obtained with the anti α-subunit antibody. This suggested that the intracellularly located subunits of ENaC are able to bind amiloride. To check for the presence of alternative binding sites of amiloride, we used a polyclonal antibody directed against epithelial histaminase (Novotny et al. 1994). The reaction center of this enzyme has affinity to amiloride (Chassande et al. 1994) and might be responsible for the binding of RA 6.3. Only weak histaminase-like reactivity was found in taste cells, and its subcellular distribution was different from that of RA 6.3 reactivity (data not shown).

Figure 3

Tangential section through fungiform papillae, passing through the taste pores. (A) Immunoreactivity for α-subunit of rENaC. (B) Immunoreactivity for β-subunit of rENaC.

To further probe the nature of the ENaC-subunit immunoreactivity, double staining with anti α-subunit antibody and RA 6.3 was carried out as described in Materials and Methods. This was especially interesting to do with the vallate papilla, in which the functional role of ENaC is unknown. At the cellular level, there was good coincidence between the reactivity of α-rENaC and of amiloride binding sites (Figure 4). Furthermore, Figure 4 shows rather clearly that the reactivity was limited to a subset of elongated cells within a bud and that at least some of these cells had slim extensions pointing towards the base of the bud. These cells may therefore be of Type I, a subtype of elongated cells that maintain contact with the basement membrane (reviewed in Lindemann 1996).

Figure 4

Longitudinal section through a taste bud of the vallate papilla. Double staining technique. (A) Immunoreactivity for α-rENaC, using horseradish peroxidase-linked secondary antibody (goat anti-rabbit) and DAB as the chromogen (brown stain). (B) Immunoreactivity for amiloride binding sites (RA6.3 was the primary antibody), using alkaline phosphatase-linked secondary antibody (goat anti-mouse) and Fast Blue RR as the chromogen (dark blue stain). Note the partial co-localization of α-rENaC and amiloride binding sites in taste cells.

In addition, double staining of vallate papillae was carried out with antibodies directed against α-ENaC and against α-gustducin, a G-protein specific for taste tissue (McLaughlin et al. 1992). α-Gustducin was previously localized to a subset of Type II elongated taste cells (Boughter et al. 1997). We found that some elongated taste cells had both ENaC and gustducin reactivities, whereas others had only one of them (data not shown). Therefore, the population of cells expressing ENaC may not have been homogeneous.

Immunohistochemistry with Sheet Preparations

On the anterior tongue, fungiform papillae are distributed randomly, most of them containing only one taste bud. In consequence, the buds were difficult to find, especially in paraffin-embedded material, and only seldom did a section contain a taste pore. As an alternative, therefore, we prepared sheets (whole-mounts) of the anterior tongue epithelium (see Materials and Methods). With this approach, it was easy to subject many anterior taste buds to immunostaining and subsequently to image their taste buds and taste pores. Immunoreactivity of anterior taste buds was found for α-, β-, and γ-ENaC in the sheet preparations, as shown in Figure 5. The labeling was further checked by confocal microscopy. These experiments confirmed the mainly intracellular location of the ENaC-like reactivity. When in control experiments preimmune serum was used instead of primary antibody, staining became insignificant (Figure 5C).

Figure 5

ENaC immunohistochemistry of whole-mount fungiform papillae imaged from the interstitial side. Left column: appearance in transmitted light. The focal plane passed approximately through the center of the apicobasal diameter of the elongated taste cells. Therefore in-focus structures represent cross-sections through the taste buds. They are surrounded by a rim of out-of-focus structures from the top of the fungiform papilla (the connective tissue within the papilla having been removed after collagenase treatment). More peripherally, the focal plane passes through the lingual epithelium encasing the papilla. Right column: in the same focal plane, the fluorescence due to immunostaining for α-, β-, and β-subunits of ENaC was imaged. In the lowermost pair of images, preimmune serum was used instead of the anti-γ-antibody as a control for nonspecific staining. Bar = 20 μm.

Taste pores were searched for possible fluorescent labeling of the apical microvilli of taste cells. It turned out that such labeling was not seen at the level of the outer opening of a taste pore but at 0.5-1 downward where, indeed, the apical poles of taste receptor cells are to be expected. Figure 6 depicts apical immunoreactivity for α-, β-, and γ-ENaC. Labeling of apical cellular poles close to the taste pores was generally evident, even though it was not possible to distinguish the fluorescence of microvilli from that of other less superficial structures.

Figure 6

ENaC immunohistochemistry of whole-mount fungiform papillae imaged from the mucosal side. Left column: one taste pore per papilla in transmitted light (arrow). Right column: corresponding fluorescence (arrow) after immunostaining for α-, β, and γ-subunits of ENaC. (A) The focal plane passing through the outer orifice of the taste pore shows little fluorescence. When the focal plane was moved down by 3 μm, now presumably passing through the apical cellular poles of the same taste bud, the fluorescence was stronger (B). In C and D, the focal mode was as in B. Bar = 20 μm.

PCR of Isolated Buds

Although the anti-ENaC antibodies were raised against peptides with no significant homology to other known proteins, the possibility exists that they reacted non-specifically with non-ENaC proteins. However, if ENaC proteins exist in a cell, then the corresponding RNA must also be present. Therefore, we used RT-PCR to corroborate expression of the three subunits in taste cells. To minimize epithelial contamination, taste buds were physically isolated from the anterior tongue and the vallate papilla and were pooled into groups of 30-50 buds. Total RNA was extracted, reverse-transcribed, and subjected to the PCR procedure. The quality of the RNA was extensively checked, as described in Materials and Methods.

Occurrence of ENaC mRNA in Posterior Tongue and in Anterior Tongue

A single VPt preparation (i.e., vallate tissue enriched in taste buds but containing some epithelium) yielded about 900 ng of total mRNA (Lindemann et al. 1998). α-Gustducin (McLaughlin et al. 1992), and all three ENaC subunits were detected in the preparation, although the contribution of the included epithelium to VPt signals remained unclear (data not shown).

Figure 7

mRNA of ENaC subunits and α-gustducin found by RT-PCR in pools of isolated taste buds of fungiform and vallate papillae. From a pool of 30 buds, about 5% was used per assay. The PCR signal for β-actin was comparable in all samples. V, F, vallate and fungiform papillae; neg, pos, negative and positive controls.

Subsequently, pools of 30 isolated buds were analyzed (Figure 7). Vallate buds (labeled V) were strongly positive for α-gustducin and for α-ENaC mRNA. Signals for β- and γ-ENaC mRNA were weaker and were only demonstrable after hybridization with specific probes. Because the efficiency of the three oligonucleotide pairings used was similar (see Materials and Methods) and because the material was quantified during the exponential phase of the PCR reaction (although not detectable by staining with ethidium bromide), the relative abundance of the RNAs encoding α-ENaC, β-ENaC, and γ-ENaC was directly deduced from data like those shown in Figures 7 and 8, yielding α-ENaC ≫ β-ENaC > γ-ENaC for taste buds from the vallate papilla.

The lingual epithelium (pLE) posterior to the vallate papilla was also subjected to RT-PCR, as described in Materials and Methods. The amount of each PCR product obtained from vallate papilla isolated buds and from pLE was quantified by scanning on a phosphor imager and was normalized to 1 ng of total RNA. Ratios were then computed for the abundance of ENaC RNA found in the taste buds over ENaC RNA found in the posterior epithelium in the same assay. The ratios for α-, β-, and γ-ENaC were 3.3, 2.7, and 0.7, respectively. This result clearly shows the differential expression of subunits when taste buds and lingual epithelium of the posterior tongue are compared.

In pooled anterior tongue taste buds, a high abundance was noted for all three species of ENaC mRNA (Figure 7). The differential occurrence of the three subunits in anterior vs vallate papilla was confirmed by the semiquantitative approach described below.

Semiquantitative Estimation of RNA Abundance

The specific signal in isolated fungiform buds and VPt was measured for each subunit and for gustducin after 25, 30, 35, and 40 cycles of amplification (Figure 8). In the anterior tongue taste buds, α-, β-, and γ-ENaC were detectable by staining with ethidium bromide after only 30 cycles of PCR, and the relative amounts of the three amplicons did not vary with the number of cycles used. In contrast, in the vallate papilla, despite some contamination by epithelium included in the VPt-preparation, the signals for β-ENaC and γ-ENaC were much lower than the signal for α-ENaC. At 30 cycles of amplification, only α-ENaC was detected, whereas β-ENaC and γ-ENaC were absent. Detection of β-ENaC required 35 cycles and γ-ENaC 40 cycles. This comparison between vallate papilla and the fungiform papilla shows that abundance of α-ENaC RNA was similar, whereas β-ENaC and γ-ENaC RNA were more abundant in anterior taste buds than in vallate papilla taste buds. The abundance of α-gustducin RNA was consistently stronger in the vallate papilla than in the anterior tongue.

Splice Variants of ENaC

Using the RT-PCR with material from rat VP and rat lingual epithelium, Li and co-workers(1995) have identified two alternatively spliced transcripts of α-rENaC. These splice variants, α-rENaC-a and α-rENaC-b, have deletions of nucleotides that introduce a premature stop codon and may result in proteins shortened by 199 and 216 amino acids, respectively, at the carboxy terminus. The presence of the splice variants in taste buds of the vallate papilla and the anterior tongue was investigated using oligonucleotide AR1327S, located in exon 7, and oligonucleotide AR1592asns, located in exon 10 (see Materials and Methods). In α-hENaC, β-hENaC, and γ-hENaC, identical sizes of 118 bp and 79 bp have been described for the eighth and ninth exons, respectively. A greater phylogenetic distance exists between α-ENaC and β-ENaC or γ-ENaC than between α-rENaC and α-hENaC. It is therefore quite likely that the same genomic organization exists also for the three rat genes. RT-PCR with oligonucleotides AR1327S and AR1592asns can theoretically, in that case, amplify several additional bands. In addition to amplification of a 265-bp band, which corresponds to the normal form of the transcript, a deletion of the eighth exon would provide an additional band at 147 bp. A deletion of the ninth exon would provide an additional band at 186 bp (observed by Li et al). A deletion of both exons would provide an additional band at 68 bp. According to Li et al. (1995), an additional band at 242 bp is also to be expected, even though this band cannot be explained by classical alternative splicing.

Figure 8

Semiquantitative comparison of ENaC and α-gustducin RNA expression in one isolated taste bud from the anterior tongue vs the equivalent of one taste bud from vallate papilla. The lanes differ by the number of cycles used for the PCR process (abscissa).

With our set of specific oligonucleotides, only a single unique band of 265 bp was amplified. It corresponded to the normal form of the transcript (data not shown). This suggests that the two splice variants were not present in our material, or at least that their levels of expression were below the threshold of detection of our assay. In any case, it appears unlikely that these spliced forms, expressed at a low level, can affect the function of the normal proteins and modulate sodium perception. A more trivial explanation of the results obtained by Li et al. is that their use of degenerate oligonucleotides has caused the amplification of artifactual products. A similar problem was encountered during the cloning of β-hENaC (Voilley et al. 1995).

Occurrence of MDEGs and ASICs

The presence of further members of the ENaC-degenerin-FaNaCh gene superfamily was investigated using the two pairs of oligonucleotides specified under Materials and Methods. The pairing ASIC1-7S and ASIC1-10AS amplified specifically a 500-bp fragment of ASIC, which encodes an H⁺-gated cation channel (Waldmann et al. 1997). The pairing MDEGI-S and MDEGI-AS amplified specifically a 1200-bp fragment of MDEG, a protein closely related to ASIC, which encodes an H⁺-gated cation channel (Waldmann et al. 1996). For both, no specific signal was found in taste buds of the vallate papilla and of the anterior tongue, suggesting that these two proteins may not be involved in the taste perception of the rat (data not shown).

Discussion

Sensory cells contained in taste buds of the anterior tongue respond to increases in mucosal Na⁺ concentration with inward current and firing of action potentials. This is especially true in rodents, in which the responses are blockable by mucosal amiloride in low micromolar concentrations, suggesting the involvement of the epithelial Na⁺ channel ENaC in the Na⁺-(and Li⁺)-specific component of salt taste (Avenet and Lindemann 1988, 1991; Gilbertson et al. 1992, 1993; Doolin and Gilbertson 1996; Miyamoto et al. 1996). In one case, the anterior tongue of the rat, it was possible to show directly that amiloride-blockable Na⁺ currents flowed inward through the apical membranes of taste cells (Avenet and Lindemann 1991). These results, as well as results obtained by recording from taste nerve fibers (Contreras et al. 1984; Heck et al. 1984; Boudreau et al. 1985; Frank et al. 1988; Ninomiya and Funakoshi 1988; Hettinger and Frank 1990; Ye et al. 1991, 1994) have led to a working model for Na⁺ taste transduction in which mucosal Na⁺ ions, if present in sufficient concentration, will flow into the receptor cell via Na⁺-specific channels located at the apical membrane. According to the model, the current loop thus established depolarizes the basolateral membrane and causes the firing of receptor cell action potentials, followed by release of transmitter (reviewed in Lindemann 1996, 1997).

In contrast to the anterior tongue, the salt taste signals arising from the posterior tongue (vallate papilla) of rats in sodium balance are not Na⁺- and Li⁺-specific and are not blockable by amiloride (Formaker and Hill 1991; Doolin and Gilbertson 1996). It is surprising, therefore, that Na⁺ channel-like immunoreactivity was previously found in taste buds of the vallate papilla of rat and dog (Simon et al. 1993; Lin et al. 1997; Lindemann et al. 1998) and that in situ hybridization appeared to show the presence of ENaC mRNA in the lingual epithelium and in taste buds of the vallate papilla (Li et al. 1994a,b, 1995). However, in situ hybridization, as used in these studies, did not separate mRNA signals arising from the epithelium and from the taste buds (Li et al. 1994a, b). Therefore, there is clearly a need for molecular data showing ENaC mRNA in vallate papilla taste cells and comparing it to ENaC mRNA in the anterior tongue taste cells.

In this study we physically isolated taste buds from the vallate papilla and used purification of mRNA followed by RT-PCR to investigate the presence of ENaC-type message in taste buds isolated from the anterior and posterior tongue. In this way we hoped to reduce or overcome the contamination with ENaC stemming from the lingual epithelium. In addition, we used antibodies against peptides derived from the three subunits of ENaC (Canessa et al. 1993, 1994; Lingueglia et al. 1993, 1994) to search for ENaC protein in taste buds obtained from different areas of the tongue.

ENaC-RNA in Taste Buds from Anterior and Posterior Tongue

RT-PCR revealed message for ENaC subunits in isolated buds derived from the vallate papilla and in buds from the anterior tongue. These results provide the first molecular evidence for the presence of ENaC subunits in taste buds that had reduced, presumably almost negligible, epithelial contamination. The relative abundance of subunit message varied with lingual area. In the anterior buds, α-, β-, and γ-ENaC RNA was readily detected. The same was true for the anterior and posterior lingual epithelium. In posterior buds, however, only α-ENaC RNA appeared to be abundant, whereas β- and especially γ-ENaC RNA was less easily detected.

It appears likely that the functional channel contains all three subunit proteins, for example, in the proportion α, β, α, γ (Firsov et al. 1998; Kosari et al. 1998). However, the abundance of the three subunit RNAs is strikingly different in different tissues. In the airways, for example, the mRNA abundance is α ≍ β ≍ γ in the bronchiolar epithelium, whereas in the nasal and tracheal epithelium the relationship is α > γ ≫ β (Farman et al. 1997). Even though translational efficiencies were not determined, the variation of these ratios across tissues is likely to reflect a tissue-specific mechanism that controls the expression of functional channel protein having a standard stoichiometry like α, β, α, γ. For example, only one of the RNA species may be under hormonal control and may therefore determine the rate of overall channel synthesis. In the cortical collecting duct, aldosterone controls α-subunit mRNA, whereas in the distal colon aldosterone controls β- and γ-mRNA (Renard et al. 1995; Asher et al. 1996; Farman 1997). Our finding that mRNA of all three subunits of ENaC is abundant only in taste buds of the anterior tongue may mean that (in rats in sodium balance) the functional channel is present only in anterior buds, in agreement with functional measurements (see above). In contrast, receptor cells of the vallate papilla may lack functional channels when the animal is in sodium balance, because the low abundance of γ- and/or β-mRNA does not allow the assembly of the multimeric complex. Interestingly, a recent study performed parallel to ours describes that treatment of Na⁺-deprived rats with aldosterone enhanced immunoreactivity to ENaC subunits in the rat vallate papilla (Lin et al. in press).

Other Channels of the ENaC Family

ENaC is homologous to a large number of proteins which are expressed in epithelial as well as excitable cells (reviewed by Barbry and Hofman 1997; Garty and Palmer 1997). The proteins MDEG and ASIC (both encoding H⁺ -gated ion channels) are members of the gene family which might possibly have a role in salt and acid perception (Waldmann et al. 1995a, b, 1996, 1997). Using specific oligonucleotides, we probed for the two transcripts in vallate papilla and anterior tongue. Because no specific signal was found in the taste buds, these two proteins are probably not involved in taste perception at the level of receptor cells.

ENaC Immunoreactivity in Taste Buds from Anterior and Posterior Tongue

α- and β-ENaC-like reactivity was found in taste cells of the vallate papilla, whereas γ-like reactivity was scarce or absent. In buds of the foliate papilla, γ-like reactivity was weak even though not absent. In contrast, α-, β-, and γ-reactivity was clearly present in taste cells of the anterior tongue. In taste cells of all lingual areas, some of the immunoreactivity was at the apical cellular pole, either at microvilli or at apical and basolateral membranes close to the tight junction. However, the major part of the immunoreactivity was not apical but intracellular. The same result was obtained with the RA6.3 antibody directed against amiloride binding sites. Double staining showed that those taste cells that had ENaC-like reactivity also had RA6.3 reactivity.

Localization of the Immunoreactivity

Most of the ENaC-like reactivity was located intracellularly. Although this may be surprising for a channel expected to reside in the apical membrane, it is entirely compatible with reports by others. Using an antibody against the native (at the time, putative) channel protein (Sorscher et al. 1988; Brown et al. 1989; Tousson et al. 1989), Simon et al. (1993) found intra-cellular labeling in taste cells of the dog vallate papilla together with some emphasis on cellular borders. Furthermore, Stewart et al. (1995), using the same antibody, found intracellular labeling in the rat fungiform papilla and the lateral membranes of taste cells. Similar results were also obtained by Lin et al. using a different set of antibodies, directed against peptides derived from ENaC subunits (Lin et al. 1997 and in press). Taking these results together with ours, we can state that seven different antibodies, targeting the native channel protein, amiloride binding sites and the α-, β-, and -γ-subunit of ENaC, all indicated a predominantly intracellular location of the amiloride-blockable Na⁺ channel in taste cells.

Similar observations made with other membrane proteins are on record. For example, the α-subunit of the taste-specific G-protein gustducin, expected to be located close to receptor proteins at the apical membrane, was found in addition intracellularly, associated with intermediate filaments (Takami et al. 1994; Tabata et al. 1996). In addition, immunohistochemistry localized odorant receptor protein, expected on apical cilia, to the soma of the olfactory neurons (Menco et al. 1997). Subunits of T-cell antigen receptor protein expressed in fibroblasts were found in ER, Golgi, and lysosomal pathways (Lippincott-Schwartz et al. 1988), and immature CFTR protein was abundant in the ER and was subject to ubiquitination and degradation without ever reaching the surface membrane (Ward et al. 1995).

We can only speculate about the significance of ENaC immunoreactivity at intracellular locations. One attractive explanation is provided by recent insights into the protein editing performed by the endoplasmic reticulum. Incorrectly folded proteins and heteromeric complexes that remained incomplete were “edited out” in the ER (pre-Golgi degradation) and in the lysosomal pathway of the Golgi system (Bonifacino et al. 1989; Klausner 1989; Klausner and Sitia 1990). In a recent elegant study, ENaC subunits expressed in oocytes matured and assembled slowly and large amounts were subjected to pre-Golgi degradation. In consequence, the amount of protein appearing at the surface membrane was only a small fraction of the amount translated (Valentijn et al. in press). Therefore, excess subunit protein piling up in ER and degradation pathways may have contributed to the intracellular immunoreactivity we observed in taste cells.

In addition, in the fungiform papilla, part of the intracellular immunoreactivity may represent ENaC units in transit to the apical membrane (Brown and Stow 1996). A large capacity of transit compartments will be especially important for the ENaC, which is a substrate for extracellular proteases (Loo et al. 1983; Vallet et al. 1997; Chraibi et al. 1998) and has a fast turnover (May et al. 1997; Staub et al. 1997). For the vallate papilla, this possibility would exist if the apical α-subunit associates with a new ENaC subunit, as yet unknown, that confers a low amiloride affinity to the complex. An ENaC-like channel located on the lateral membrane would cause a strong depolarization of the cell unless the channels had a low open probability, i.e., were practically closed. The interesting possibility that basolateral ENaC is closed at rest but opens in response to cellular or extracellular signals will have to be considered in future experimentation.