Glutamate-Induced Cobalt UptakeReveals Non-NMDA Receptors in Rat

Taste Cells

ALEJANDRO CAICEDO,1* KYUNG-NYUN KIM,1,2AND STEPHEN D. ROPER1,3

1Department of Physiology and Biophysics,University of Miami School of Medicine, Miami, Florida 33101

2Department of Oral Physiology, College of Dentistry, Kangnung National University,Kangnung, Korea

3Rocky Mt. Taste and Smell Center, University of Colorado Health Sciences Center,Denver, Colorado 80262

ABSTRACTTaste receptor cells are chemical detectors in the oral cavity. Taste cells form synapses

with primary afferent neurons that convey the gustatory information to the central nervoussystem. Taste cells may also synapse with other taste cells within the taste buds. Further-more, taste cells may receive efferent connections. However, the neurotransmitters at thesesynapses have not been identified. Glutamate, a major excitatory neurotransmitter in othersensory organs, might act at synapses in taste buds. We used a cobalt staining technique todetect Ca21-permeable glutamate receptors in taste buds and thus establish whether theremight be glutamatergic synapses in gustatory end organs.

When 500 mm slices of foliate and vallate papillae were briefly exposed to 1 mMglutamate in the presence of CoCl2, a subset of spindle-shaped taste cells accumulated Co21.Cobalt uptake showed concentration-dependency in the range from 10 mm to 1 mM gluta-mate. Interestingly, higher glutamate concentrations depressed cobalt uptake. Thisconcentration-response relation for cobalt uptake suggests that synaptic glutamate receptors,not receptors for glutamate taste, were activated. Sensory axons and adjacent non-sensoryepithelium were not affected by these procedures. Glutamate-stimulated cobalt uptake intaste cells was antagonized by the non-NMDA receptor antagonist CNQX. Depolarizationwith 50 mM K1 and application of NMDA (300 mM) did not increase the number of stainedtaste cells. This pharmacological characterization of the cobalt uptake suggests that non-NMDA receptors are present in taste cells. These receptors might be autoreceptors at afferentsynapses, postsynaptic receptors of a putative efferent system, or postsynaptic receptors atsynapses with other taste cells. J. Comp. Neurol. 417:315–324, 2000. © 2000 Wiley-Liss, Inc.

Indexing terms: taste buds; foliate papilla; vallate papilla; gustatory epithelium; tongue

Taste buds are peripheral sensory organs specialized todetect and analyze chemicals in the oral cavity. Receptorcells in taste buds form synaptic connections with sensoryafferent fibers. Taste cells may also synapse with othercells within the taste bud. Furthermore, there may beefferent inputs to taste buds from the nervous system(Roper, 1992). However, the identity of neurotransmittersin taste buds is not yet known. This is mostly due totechnical limitations. For instance, there is no ready ac-cess to taste cells and primary sensory axons within tastebuds in the intact preparation. Thus, it has not yet beenpossible to apply transmitter candidates focally to test andcharacterize synaptic responses directly. Consequently,

taste lags behind other senses with respect to the identi-fication of synaptic mechanisms involved in the peripheralsensory organ.

Glutamate, a major excitatory neurotransmitter in thebrain and sensory organs, by analogy might also act at syn-

Grant sponsor: NIH/NIDCD; Grant numbers: 1PO1 DC00244, 2PO1DC00374.

*Correspondence to: Alejandro Caicedo, Department of Physiology andBiophysics, University of Miami School of Medicine, PO Box 016430, Mi-ami, FL 33101. E-mail: alejoc@aol.com

Received 24 March 1999; Revised 6 October 1999; Accepted 26 October1999

THE JOURNAL OF COMPARATIVE NEUROLOGY 417:315–324 (2000)

© 2000 WILEY-LISS, INC.

apses in taste buds. If so, one would anticipate that synapticglutamate receptors (GluRs) would be present on afferentterminals, taste cells, or both. Synaptic ionotropic receptorsfor glutamate (iGluRs) are commonly divided into three fam-ilies. One group is selectively activated by N-methyl-D-aspartate (NMDA). The second group is activated bya-amino-3-hydroxy-5-methyl-4-isoxazole (AMPA). The thirdgroup shows high affinity for kainate. GluRs of the secondand third groups are collectively termed non-NMDA recep-tors because they are insensitive to NMDA. However, little isknown about the role, if any, of any of these three classes ofiGluRs at synapses in taste buds. Furthermore, the functionof iGluRs in taste buds is complicated by the action of glu-tamate as a taste stimulus. As such, glutamate presumablyactivates apical chemosensitive receptors on taste bud cells,and these receptors may differ fundamentally from synapticiGluRs (Chaudhari et al., 1996).

Glutamate is present in high concentration in a smallproportion of taste cells and in afferent axons innervatingtaste buds in Necturus (Jain and Roper, 1991). Also, glu-tamate is taken up by some taste cells in Necturus tastebuds as well as by glial cells surrounding afferent nervefibers (Nagai et al., 1998). These data are consistent with,but do not prove the existence of glutamatergic synapsesin taste buds. Isolated taste buds and taste cells from miceand rats respond to bath-applied glutamate, but the un-derlying receptors have not been well characterized (Ha-yashi et al., 1996; Lin et al., 1996; Sugimoto, 1994, 1996;Bigiani et al., 1997). More importantly, these glutamateresponses are believed to reflect activation of apical tastereceptors, in contrast with basolateral synaptic receptors.Lastly, Chaudhari et al. (1996) identified specific metabo-tropic and ionotropic GluRs in lingual epithelium fromfoliate and vallate papillae in rats. Although one of themetabotropic GluRs appeared to represent an apical tastereceptor, the functional significance of the iGluRs was notexplored.

To investigate the presence and possible functional sig-nificance of iGluRs in taste buds, we have used the cobaltstaining technique introduced by Pruss et al. (1991). Thistechnique involves the pharmacological stimulation ofiGluRs in the presence of CoCl2, which leads to the uptakeof Co21 through Ca21-permeable iGluRs. Subsequently,one can visualize activated cells by precipitating histo-chemically and intensifying the intracellular Co21 (seeFulton et al., 1992; Zhou et al., 1995; Caicedo et al., 1998).This technique serves to identify cells that possess Ca21-permeable iGluRs of the non-NMDA type. Depolarization-activated Ca21 channels do not contribute to Co21 signalswith this methodology (Pruss et al., 1991). Our results onlingual slices show that a subset of taste cells take upCo21 after stimulation with glutamate. Pharmacologicalcharacterization of this glutamate-induced Co21 uptake intaste cells shows that synaptic iGluRs of the non-NMDAclass are most probably involved. An abstract of thesefindings has been presented (Caicedo and Roper, 1998).

MATERIALS AND METHODS

All experimental protocols were approved by the Uni-versity of Miami Animal Care and Use Committee.

Preparation of slices

Tongue slices were obtained from 74 young adultSprague Dawley rats of both sexes. Rats were killed in a

closed chamber containing CO2, followed by cervical dis-location. Tongues were quickly removed and immersed incold, oxygenated, low-sodium, high–sucrose cobalt uptakebuffer (hereafter referred to as high-sucrose Tyrode’s) con-taining (in mM): 139 sucrose, 57.5 NaCl, 5 KCl, 1 CaCl2, 2MgCl2, 10 HEPES, and 12 glucose, at pH 7.4, 300–310mosm. This buffer has been widely used in cobalt uptakestudies because it limits excitotoxic injury. Blocks of tissuecontaining vallate and foliate papillae were removed fromthe tongue and mounted on a tissue slicer (Stoelting,Wood Dale, IL). Vallate and foliate papillae slices (500-mm-thick) were cut transversally with respect to their longaxis and incubated for 1 hour at 37°C in oxygenated high-sucrose Tyrode’s.

Co21

uptake

For glutamate-induced Co21 uptake, slices were trans-ferred into 6-well dishes containing oxygenated high-sucrose Tyrode’s at room temperature (19–22°C). Aftertwo 10-minute washes in high-sucrose Tyrode’s, sliceswere incubated in high-sucrose Tyrode’s containing 5 mMCoCl2 plus glutamate and/or other drugs for 5 minutes atroom temperature. Pilot experiments showed that 5 min-utes were sufficient to reliably induce Co21 staining reli-ably while reducing non-specific background staining. Incontrol experiments, the non-NMDA receptor antagonist6-cyano-7-nitro-quinoxaline-2,3-dione (CNQX; 100 mM;n 5 60 slices) or the broad spectrum GluR blocker kynure-nate (1 mM; n 5 7 slices) was added to the stimulussolution (e.g., glutamate) before, during, and after expo-sure to Co21. After pharmacological stimulation, sliceswere rinsed for 5 minutes in high-sucrose Tyrode’s, thenfor 5 minutes in high-sucrose Tyrode’s containing 2 mMEDTA to remove non-specifically bound Co21, followed by5 minutes in high-sucrose Tyrode’s. Co21 was then precip-itated by incubating slices for 5 minutes in high-sucroseTyrode’s containing 1.2% (NH4)2S. Slices were againwashed in high-sucrose Tyrode’s and fixed overnight in 4%paraformaldehyde in 0.01 M phosphate-buffered saline(PBS, pH 7.4) at 4°C. Subsequently, slices were cryopro-tected in 20% sucrose in PBS and resectioned in the trans-verse plane at 30 mm on a cryostat (CM 1900, Leica Mi-crosystems, Inc., Deerfield, IL). Up to 10 to 15 sectionscontaining taste buds could be cut from each lingual slice.Silver enhancement of the cobaltous sulfide precipitatewas carried out on floating sections with the intenSEM kitfrom Amersham (Arlington Heights, IL). Sections weredipped in 0.5% gelatin, mounted on slides, counter-stainedwith methylene blue, dehydrated, cleared, and cover-slipped.

For depolarization experiments, Na1 was substitutedwith K1 in a modified Tyrode’s containing (in mM): 139sucrose, 12.5 NaCl, 50 mM KCl, 1 mM CaCl2, 2 mMMgCl2, 10 mM HEPES, and 12 mM glucose. For experi-ments using high glutamate concentrations, sucrose wassubstituted with glutamate. Osmolarity and pH were keptconstant at 300–310 mosm and pH 7.4 in all substitutionexperiments.

L-glutamate, D-glutamate, kainate, and kynurenatewere purchased from Sigma Chemical Co. (St. Louis, MO),NMDA and CNQX from Tocris (Ballwin, MO).

Data analysis

Sections were examined and photographed on a ZeissAxioplan microscope with brightfield illumination and No-

316 A. CAICEDO ET AL.

marski differential interference contrast optics. Photo-graphic slides were scanned with a Nikon LS-1000 scan-ner (Tokyo, Japan) and compiled using Adobe Photoshop4.0.1 (Adobe Systems Inc., San Jose, CA). Figures werenot altered from the original scans except to adjust con-trast and brightness for uniform tone within a singlefigure. Specific Co21 staining was defined as an intracel-lular, homogeneous, brownish precipitate within identifi-able cells. To quantify Co21 uptake we counted well-stained cells (Fig. 1). Although the number of stained cellsis an indirect measure for the pharmacological effects ofiGluR activation, it proved to be a useful parameter todetermine differences among the treatments. Owing tothe variability in the number of stained cells (see below),however, subtle pharmacological effects could probablynot be resolved and may not have been detected in thisstudy. Well-stained cells were counted blindly and inde-pendently by two investigators. We included only thosecells in which the cobalt stain was intense and uniformand where the cell nucleus and at least one of the elongateprocesses were clearly visible. The incidence of Co21 up-take was expressed as the number of stained cells pertaste bud. The number of stained cells per taste bud wasdetermined by dividing the number of stained cells by thenumber of taste buds counted in two 30 mm-sections fromeach 500 mm-thick slice. For data collected from vallateand foliate papillae, this means that the average numberof stained cells per taste bud was derived from approxi-mately 40 and 28 taste buds, respectively. Because therewas variability in staining from section to section within aslice, the two 30 mm sections with the highest incidence ofCo21-stained cells were chosen from each slice. Typically,these sections were not from the cut faces of the 500 mmslices. The mean number of stained cells per taste bud fora given experimental group was then determined by pool-ing the values from several individual slices.

Although we attempted to keep all methodological pa-rameters constant, variability in Co21 uptake and non-specific staining could not always be avoided. Possiblesources of variability and artifactual staining included

anoxia, cellular damage near each face of the tissue slices,and diffusion barriers. To compensate for experimentalvariability and to allow comparisons between experimen-tal groups, we included at least three slices per condition.Also, slices with high, non-specific background stainingwere excluded from the analysis because specificallystained cells could not be reliably identified. Lastly andimportantly, as stated previously, all cell counts wereconducted blindly by two investigators.

In a first set of experiments we compared glutamate-stimulated Co21 uptake in high-sucrose Tyrode’s with up-take in different control treatments (CNQX, CNQX plusglutamate, and high-sucrose Tyrode’s alone). Because thenumbers of stained cells per taste bud in the CNQX con-trol group were not normally distributed (due to many 0values; see Fig. 5), we used a one way ANOVA on ranks(Kruskal-Wallis) followed by pairwise multiple compari-sons (Dunn’s method) to make comparisons within thisdata set. For statistical comparisons between two experi-mental groups (glutamate vs. CNQX in vallate and foliatepapillae), we used a Mann-Whitney test. In a second set ofexperiments, we compared the effects of glutamate withthat of other pharmacological treatments (depolarization,NMDA, and kainate). Comparisons in this group weremade with a one way ANOVA followed by pairwise mul-tiple comparison procedures (Student-Newman-Keulsmethod). All data are expressed as mean 6 SEM.

RESULTS

Glutamate-induced Co21

uptake intaste cells

When stimulated for 5 minutes with 1 mM glutamate,taste cells in vallate and foliate papillae showed strongCo21 uptake (Figs. 2, 3). Homogeneous Co21-stainingfilled the cytoplasm of taste cells, delineating the cell bodyand its processes but sparing the nucleus (Fig. 2C,D).Cobalt-stained cells were spindle-shaped with clear,round nuclei and thin processes extending to the apicaland basal ends of the taste bud. These characteristics maycorrespond best to light cells (Pumplin et al., 1997), but wemade no detailed attempt to categorize taste cells by theirmorphotype. Afferent sensory axons were not distinctlystained. No staining could be found in the surroundingnon-sensory epithelium (Fig. 2B). Muscle fibers below theepithelium showed non-specific Co21 uptake becausethese cells were transected during the slicing procedure.

After glutamate stimulation in high-sucrose Tyrode’s,there were 3.9 6 0.4 stained cells per taste bud in foliatepapillae (n 5 21 slices; Fig. 4). In the absence of gluta-mate stimulation (high-sucrose Tyrode’s alone), thenumber of stained cells was lower (2.2 6 0.5 cells pertaste bud, n 5 7 slices; Figs. 2E, 4). These differenceswere statistically significant (see below). Adding thecompetitive non-NMDA receptor antagonist CNQX (100mM) to the high-sucrose Tyrode’s during incubation ap-peared to further reduce the number of stained cells pertaste bud (1.1 6 0.2; n 5 31 slices; Fig. 4) but thedifferences were not statistically significant (see below).Similar results were obtained with the broad-spectrumGluR antagonist kynurenate (1 mM; data not shown).The results with CNQX and kynurenate treatmentssuggest that glutamate may be present in the intersti-tial spaces in tissue slices, possibly leakage from dam-

Fig. 1. Schematic summary of the experimental protocol.

317GLUTAMATE RECEPTORS IN RAT TASTE BUDS

aged muscle cells, and may stimulate from a nominalCo21 uptake. Therefore, we used data from CNQX-treated slices as our baseline and hereafter refer tothese slices as controls. When slices were stimulatedwith glutamate (1 mM) in the presence of CNQX (100mM), the number of stained cells (1.1 6 0.6; n 5 7 slices;Figs. 3, 4) was the same as in control slices, indicatingthat the effects of added glutamate could be antago-nized by blocking non-NMDA receptors.

One way analysis of variance on ranks (Kruskal-Wallis)showed that the differences in the median values amongthe glutamate, high-sucrose Tyrode’s, and CNQX groupswere significant (H 5 22.6 with 3 degrees of freedom, P 50.00005). Pairwise multiple comparisons (Dunn’s method;P , 0.05) showed that glutamate significantly increased

the number of stained cells per taste bud over that of theother experimental groups. However, high-sucrose Ty-rode’s was not different from CNQX alone.

The incidence of Co21-stained cells was higher in foliatetaste buds than in vallate taste buds (Figs. 2, 3, 5). Afterglutamate stimulation (n 5 8 slices), the mean number ofCo21-stained cells per vallate taste bud was 2.2 6 0.6,while it was 3.9 6 0.4 in foliate papillae (Student’s t-test P, 0.05). Because glutamate-stimulated Co21 uptake wasstronger in foliate taste buds, most of the experiments ofthis study were conducted on slices of foliate papillae.

When stimulated with glutamate, all foliate taste budsappeared to contain Co21-stained cells (Fig. 6). There wasno evidence for selective Co21 uptake in some taste budsand not in others. That is, the number of stained cells per

Fig. 2. Glutamate stimulated Co21 uptake in taste bud cells. Low(A) and high-power (B–E) photomicrographs illustrating Co21 stain-ing in 30 mm sections of taste papillae in the rat. Lingual tissues wereincubated with 1 mM glutamate (glu) for 5 minutes, as described inMaterials and Methods. Glutamate stimulated strong Co21 uptake in

a subset of taste cells within taste buds of foliate (A,C) and vallate (D)papillae. Co21 uptake was totally absent from the surrounding non-sensory epithelium (B). The incidence of Co21 uptake was low intongue slices incubated without glutamate (E). Scale bars 5 100 mmin A; 50 mm in E (applies to B–E).

318 A. CAICEDO ET AL.

taste bud was uniformly distributed. In taste buds of con-trol slices, the incidence of stained cells was markedlyshifted to lower values, with most taste buds (ca. 50%)lacking stained cells (Fig. 6). These results indicated thatthere was no preferential clustering of glutamate-stimulated Co21 uptake in a subset of taste buds and that

cells showing glutamate-induced Co21 uptake were evenlydistributed throughout the gustatory papilla. Based oncounting the total number of cells in taste buds inspectedin 30 mm sections, we estimate that the population ofCo21-stained cells corresponds to about 10% of the totalcells in a taste bud.

Fig. 3. Glutamate-induced (1 mM) uptake of Co21 was significantly antagonized by the non-N-methyl-D-aspartate (NMDA) antagonist 6-cyano-7-nitro-quinoxaline-2,3-dione (CNQX, 100 mM) in val-late (A,B) and foliate papillae (C,D). Note that the foliate papilla had more stained cells than the vallatepapilla. Scale bar 5 50 mm.

319GLUTAMATE RECEPTORS IN RAT TASTE BUDS

Pharmacological characterization of Co21

uptake in gustatory cells

In the previous section, we showed that glutamate-induced Co21 uptake was antagonized by blocking non-NMDA receptors. To test receptor specificity underlyingglutamate-stimulated Co21 uptake further, we treatedslices with kainate, a specific agonist for non-NMDA re-ceptors. Co21 uptake induced by incubating slices 5 min-utes with 300 mM kainate (3.2 6 0.5 cells per taste bud;n 5 7 slices; Figs. 7A, 8) was similar to that induced by 1mM glutamate.

Although several reports state that Co21 influx does notoccur through voltage-gated Ca21 channels or NMDA re-ceptors (e.g., Pruss et al., 1991; Fulton et al., 1992; Zhou etal., 1995), some studies have reported NMDA-stimulatedCo21 uptake in the hippocampus and spinal cord (Wil-liams et al., 1992; Nagy et al., 1994). Thus, we testedwhether activation of NMDA receptors could elicit Co21

uptake in taste cells. Lingual slices were incubated inNMDA (300 mM), glycine (1 mM), and CNQX (100 mM) inthe absence of Mg21 (n 5 4 slices). This concentration ofNMDA and these conditions are in the range reported toinduce Co21 uptake in the hippocampus and the spinalcord (Williams et al., 1992; Nagy et al., 1994). The highglycine and NMDA concentrations were needed to over-come blockage of the glycine-binding and agonist-bindingsites of NMDA receptors by CNQX. However, in taste budstreated with NMDA, Co21 uptake did not differ from con-trol slices (0.9 6 0.5 cells per taste bud; Figs. 7B, 8). Wealso examined if Co21 influx could occur through voltage-

Fig. 4. CNQX added to the bath reduced glutamate-induced Co21

uptake. Quantification of results of Co21 uptake for foliate taste buds. Inthe absence of glutamate, there was a low-level, basal Co21 uptake (e.g.see Figure 2) (shaded bar, “Tyr”; see Fig. 1E). Basal Co21 uptake wasreduced by CNQX (100 mm). The incidence of Co21 uptake is expressedas the number of stained cells per taste bud. Glu 5 1 mM glutamate.Numbers in parentheses, number of preparations (slices) included in theanalyses. Differences among the median values of the experimentalgroups are significant (Kruskal-Wallis one way ANOVA on ranks; H 22.6with 3 degrees of freedom; P 5 0.00005). The median value of theglutamate group is significantly different from that of the other groups.The median value of the Tyrode’s group is not different from that ofeither CNQX group (pairwise multiple comparisons, Dunn’s method; P ,0.05).

Fig. 5. The incidence of glutamate-induced Co21 uptake washigher in foliate taste buds than in vallate taste buds (Student’s t-test,P , 0.05). CNQX (100 mM) induced a significant 3- to 4-fold reductionin glutamate-induced Co21 uptake in both foliate and vallate papillae(Mann-Whitney test, P , 0.05). The incidence of Co21 uptake isexpressed as the number of stained cells per taste bud. Numbers inparentheses, number of preparations (slices) included in the analyses.

Fig. 6. Distribution of Co21 uptake in taste buds of foliate papillae.Frequency histograms of the number of stained cells per taste budshow that glutamate-induced Co21 uptake in taste cells was uni-formly distributed (solid bars). In CNQX-treated slices (open bars),most taste buds lacked Co21-stained cells. Data were taken from 260(glutamate) and 453 (CNQX) taste buds pooled from all slices in eachgroup.

320 A. CAICEDO ET AL.

gated Ca21 channels. Depolarizing slices with K1 (50 mM)in the presence of CNQX (100 mM; n 5 8 slices) did notincrease Co21 uptake over control values (1.2 6 0.5 cellsper taste bud; Figs. 7C, 8). One way analysis of varianceshowed that the differences in the mean values among the

glutamate, NMDA, kainate, and depolarization groupswere significant (F 5 7, P 5 0.0008). Pairwise multiplecomparisons (Student-Newman-Keuls method, P , 0.05)showed that stimulating with glutamate differed signifi-cantly from NMDA stimulation and depolarization, butnot from kainate treatment.

To test whether the effects of glutamate were mediatedby concentration-dependent, receptor-agonist interactions

Fig. 9. Co21 uptake induced by glutamate in foliate taste buds wasconcentration-dependent. For glutamate concentrations ranging from10 mM to 1 mM, Co21 uptake increased with increasing glutamateconcentration. Above 1 mM glutamate, Co21 uptake was stronglyreduced. The incidence of Co21 uptake is expressed as the number ofstained cells per taste bud. Dashed line, control value (as in Figure 8).Numbers in parentheses, number of preparations (slices) included inthe analyses.

Fig. 7. Co21 uptake in foliate taste buds was stimulated by kainate (300 mM; A) but not by NMDA(300 mM with 1 mM glycine and 0 mM Mg21; B), or by high K1 (50 mM, C). Scale bar 5 50 mm.

Fig. 8. Depolarizing foliate taste buds with K1 (50 mM) or treatingpreparations with NMDA (300 mM plus 1 mM glycine and 0 mMMg21) did not stimulate Co21uptake (open bars). The dashed lineshows the mean number of stained cells per taste bud in controlpreparations (see Figure 4, Tyr1CNQX). Kainate (300 mM; shadedbar) induced Co21 uptake similar to that elicited by 1 mM glutamate(solid bar; cf. Figs. 8, 9). The incidence of Co21 uptake is expressed asthe number of stained cells per taste bud. Numbers in parentheses,number of preparations (slices) included in the analyses. Differencesamong the experimental groups were significant (one way ANOVA,F 5 7, P 5 0.0008). The mean values of the glutamate group weresignificantly different from those of the K1 and NMDA groups, but notfrom those of the kainate group (pairwise multiple comparisons,Student-Newman-Keuls method; P , 0.05).

321GLUTAMATE RECEPTORS IN RAT TASTE BUDS

rather than non-specifically through other mechanisms,we tested Co21 uptake with a range of glutamate concen-trations. Co21 uptake was concentration-dependent andincreased from 10 mM to 1 mM glutamate (Fig. 9). Thus, atconcentrations that are reached during activity at centralglutamatergic synapses (' 1 mM, Clements et al., 1992),glutamate induced Co21 uptake in taste bud cells. How-ever, at higher concentrations (10 and 100 mM), gluta-mate did not induce Co21 uptake and even appeared todepress it, suggesting that additional mechanisms may beactivated at concentrations higher than 1 mM. Addition-ally, we examined the concentration dependency for thenon-NMDA receptor agonist kainate. Kainate (30–300mM) also increased Co21 uptake in a concentration-dependent manner (Fig. 10).

DISCUSSION

The identification of putative neurotransmitters at thefirst gustatory synapse, taste cells to sensory axons, hasmainly been limited by technical problems. Access of phar-macological substances to taste buds in the intact tongueis limited. A detailed physiological characterization oftaste cell synaptic mechanisms has not yet been achieved.As shown in the present study, the Co21 staining tech-nique represents a valuable new approach for visualizingglutamate-activated synapses in taste buds. The presentstudy has shown that stimulating taste buds with gluta-mate induces Co21 uptake in a subset of taste cells infoliate and vallate papillae of the rat. No glutamate-induced Co21 uptake was observed in postsynaptic pri-mary sensory axons. The pharmacological characteriza-tion of the glutamate-stimulated Co21 uptake in tastecells indicates that uptake occurs via ionotropic glutamatereceptors (iGluRs) of the non-NMDA type. Furthermore,Co21 uptake was stimulated at glutamate concentrationsthat are effective at glutamatergic synapses in the centralnervous system. These results suggest that a subpopula-

tion of taste cells express non-NMDA GluRs and thatthese iGluRs are involved in synaptic transmission intaste buds.

Since the original report by Pruss et al. (1991), severalstudies have used Co21 uptake to study neural popula-tions with divalent cation-permeable iGluRs (e.g., Fultonet al., 1992; Williams et al., 1992; Leinders-Zufall et al.,1994; Allcorn et al., 1996; Caicedo et al., 1998). In accor-dance with most studies on the brain (e.g., Pruss et al.,1991; Fulton et al., 1992), NMDA application does notinduce Co21 uptake in taste cells. Also, Co21 influxthrough voltage-gated Ca21 channels can be excluded,because no Co21 uptake was induced by high levels ofextracellular potassium (Pruss et al., 1991; Hack and Bal-azs, 1995; Zhou et al., 1995; present results). We showedthat the competitive non-NMDA antagonist CNQX antag-onizes glutamate-induced Co21 uptake and that kainatestimulates Co21 uptake in a concentration-dependentmanner. These findings are consistent with Co21 influxoccurring through Ca21-permeable iGluRs of the non-NMDA type. The GluRs of the non-NMDA type mediatemost of the excitatory neurotransmission in the brain andthus, by analogy, may be serving a synaptic function intaste buds. Moreover, glutamate activated iGluRs in tastecells at concentrations that are consistent with glutama-tergic synaptic mechanisms in the central nervous system.

Given that non-NMDA receptors of the AMPA type de-sensitize upon exposure to glutamate but not to kainate, itwas somewhat surprising that kainate did not induce ahigher incidence of Co21 uptake than glutamate. How-ever, when glutamate or kainate activate kainate recep-tors, rapid desensitization occurs (Lerma et al., 1997).Thus, a similar incidence of Co21 uptake after kainate andglutamate stimulation in taste cells might suggest thatkainate receptors, and not AMPA receptors, were acti-vated. Taste epithelium indeed expresses kainate receptorsubunits but not AMPA receptor subunits (Chaudhari etal., 1996).

Glutamate is a taste stimulus and presumably stimu-lates taste receptors (as opposed to synaptic receptors) onthe apical chemosensitive tips (Chaudhari and Roper,1998). However, glutamate as a taste stimulus is onlyeffective in concentrations above ' 1 mM (and up to 300mM in rats). Thus, our concentration-response relation-ship for glutamate-stimulated Co21 uptake is not consis-tent with activation of taste receptors for glutamate. Thisreinforces our interpretation that glutamate-stimulatedCo21 uptake reflects activation of synaptic receptors intaste buds. As a taste stimulus, glutamate presumablyacts on a different type of GluR (e.g., a metabotropic GluR,Chaudhari et al., 1996) than those underlying the Co21

uptake shown in this report.

Localization of non-NMDA receptors intaste buds

Cobalt-stained cells had a similar shape in vallate andfoliate papillae. These cells had a central, enlarged cellbody containing a large round nucleus and processes ex-tending towards the apical pore and the basal portion ofthe taste bud. They further had a regular outline andlacked lateral cytoplasmic projections. Although a moredetailed morphological characterization would be neces-sary for an unambiguous identification, our results sug-gest that Co21-stained cells are light cells, not dark cells(Pumplin et al., 1997). We estimated that the population

Fig. 10. Co21 uptake induced by kainate in foliate taste buds wasconcentration dependent. The incidence of Co21 uptake is expressedas the number of stained cells per taste bud. Dashed line, controlvalue. Numbers in parentheses, number of preparations (slices) in-cluded in the analyses.

322 A. CAICEDO ET AL.

of Co21-stained cells corresponds roughly to 10% of thetotal cells in a taste bud. Light cells represent approxi-mately 20% of the taste cell population (Lindemann,1996). This might imply that glutamate stimulates Co21

uptake in about half of the light cells in a taste bud.Sensory afferent axons did not show glutamate-induced

Co21 uptake. This is puzzling because, in analogy to othersensory systems, one might expect synaptic non-NMDAreceptors to be present in sensory axons. However, previ-ous studies in the brainstem and the cochlea have shownthat axonal Co21 uptake is weak and almost undistin-guishable from Co21 staining in the granular neuropil(Caicedo et al., 1998; A. Caicedo, J. Ruel, and J.-L. Puel,unpublished observations). Such a finding would result ifGluR densities on sensory axons were too low to permitdetectable Co21 influx, or if iGluRs present on sensoryaxons had very low Ca21 permeability. The proceduresused here probably do not enable us to detect Co21 influxinto sensory axons in taste buds if it occurs. Furtherexperiments using different approaches are necessary todetermine if GluRs are present on sensory axons.

Functional considerations

Sensory information in taste buds is transmitted fromtaste cells to afferent axons. Evidence for efferent controlof taste cells is sparse (for a review, see Roper, 1989). Inthis respect, the presence of iGluRs in the supposed pre-synaptic elements, taste cells, seems enigmatic. However,a typical ultrastructural feature of many synapses in tastebuds is the lack of a clear polarity. Indeed, clusters ofvesicles are present at active sites of afferent axons, andsubsynaptic cisternae, which are characteristic of efferentsynapses in other sensory organs (e.g., outer hair cells inthe cochlea), can be seen near some synapses in taste cells(Zahm and Munger, 1983). In addition, the presynapticmarker SNAP-25 is present in sensory axons in taste buds(Yang et al., 1998). Conceivably, some synapses in tastebuds are efferent or bidirectional. In this context, it shouldbe noted that the release of neuroactive substances, in-cluding glutamate, by sensory afferent neurons is welldocumented (for a review, see Maggi, 1991). For instance,peripheral branches of primary afferent sensory axons canrelease neuropeptides, such as substance P. A dualsensory-efferent function for capsaicin-sensitive nocicep-tive axons has also been reported (e.g., Amann et al., 1990;Mayer et al., 1990; Myers and Undem, 1991). Most impor-tantly, excitatory amino acids, such as glutamate, arereleased from somato-sensory afferent axons (Jeftinija etal., 1991; Jackson et al., 1995). Lastly, there are sugges-tions that axon collaterals from sensory afferent fibersthat innervate one taste bud may exert local reflex effer-ent synaptic feedback onto an adjacent taste bud (Mu-rayama, 1988). Therefore, we speculate that gustatorysensory afferent axons could have a dual sensory-efferentfunction and exert a local efferent regulation of taste cellsby iGluRs.

There is further, albeit indirect, support for the hypoth-esis that gustatory sensory neurons provide glutamatergicinput to taste cells. Gustatory sensory neurons releaseglutamate at their central axonal terminals in the nucleusof the solitary tract (Bradley et al., 1996; Li and Smith,1997) and thus they have the potential to release gluta-mate at their peripheral projections. Glutamate has beenshown to be present in the nerve supply to taste buds inNecturus (Jain and Roper, 1991; Lu and Roper, 1993).

Moreover, glutamate is taken up by glial cells that accom-pany gustatory nerve fibers and by certain cells in tastebuds (Nagai et al., 1998). Taken together, our presentfindings and those cited above suggest that there is aglutamatergic efferent regulation of taste bud function.

ACKNOWLEDGMENTS

We thank Dr. Orlando Gomez-Marın for his help withthe statistical analyses. We also appreciate Dr. SamirJafri’s comments on the manuscript. Dr. Kyung-NyunKim was supported by Kangnung National University.

LITERATURE CITED

Allcorn S, Catsicas M, Mobbs P. 1996. Developmental expression andself-regulation of Ca21 entry via AMPA/KA receptors in the embryonicchick retina. Eur J Neurosci 8:2499–2510.

Amann R, Donnerer J, Lembeck F. 1990. Activation of primary afferentneurons by thermal stimulation. Influence of ruthenium red. Naunyn-Schmiedeberg Arch Pharmacol 341:108–113.

Bigiani A, Delay RJ, Chaudhari N, Kinnamon SC, Roper SD. 1997. Re-sponses to glutamate in rat taste cells. J Neurophysiol 77:3048–3059.

Bradley RM, King MS, Wang L, Shu X. 1996. Neurotransmitter andneuromodulator activity in the gustatory zone of the nucleus tractussolitarius. Chem Senses 21:377–385.

Caicedo A, Roper SD. 1998. Glutamate receptors in taste buds in rats. SocNeurosci Abstr 24:1887.

Caicedo A, Kungel M, Pujol R, Friauf E. 1998. Glutamate-induced Co21

uptake in rat auditory brainstem neurons reveals developmentalchanges in Ca21 permeability of glutamate receptors. Eur J Neurosci10:941–954.

Chaudhari N, Roper SD. 1998. Molecular and physiological evidence forglutamate (umami) taste transduction via a G protein-coupled recep-tor. Ann NY Acad Sci 855:398–406.

Chaudhari N, Yang H, Lamp C, Delay E, Cartford C, Than T, Roper SD.1996. The taste of monosodium glutamate: membrane receptors intaste buds. J Neurosci 16:3817–3826.

Clements JD, Lester RAJ, Tong G, Jahr CE, Westbrook GL. 1992. The timecourse of glutamate in the synaptic cleft. Science 258:1498–1501.

Fulton BP, Burne JF, Raff MC. 1992. Visualization of O-2A progenitor cellsin developing and adult rat optic nerve by quisqualate-stimulatedcobalt uptake. J Neurosci 12:4816–4833.

Hack N, Balazs R. 1995. Properties of AMPA receptors expressed in ratcerebellar granule cell cultures: Ca21 influx studies. J Neurochem65:1077–1084.

Hayashi Y, Zviman MM, Brand JG, Teeter JH, Restrepo D. 1996. Mea-surement of membrane potential and [Ca21 ]i in cell ensembles: appli-cation to the study of glutamate taste in mice. Biophys J 71:1057–1070.

Jackson DL, Graff CB, Richardson JD, Hargreaves KM. 1995. Glutamateparticipates in the peripheral modulation of thermal hyperalgesia inrats. Eur J Pharmacol 284:321.

Jain S, Roper SD. 1991. Immunocytochemistry of gamma-aminobutyricacid, glutamate, serotonin, and histamine in Necturus taste buds.J Comp Neurol 307:675–682.

Jeftinija S, Jeftinija K, Liu F, Skilling SR, Smullin DH, Larson AA. 1991.Excitatory amino acids are released from rat primary afferent neuronsin vitro. Neurosci Lett 125:191.

Leinders-Zufall T, Rand MN, Waxman SG, Kocsis JD. 1994. Differentialrole of two Ca21-permeable non-NMDA glutamate channels in ratretinal ganglion cells: kainate-induced cytoplasmic and nuclear Ca21

signals. J Neurophysiol 72:2503–2516.Lerma J, Morales M, Vicente MA, Herreras O. 1997. Glutamate receptors

of the kainate type and synaptic transmission. Trends Neurosci 20:9–12.

Li CS, Smith DV. 1997. Glutamate receptor antagonists block gustatoryafferent input to the nucleus of the solitary tract. J Neurophysiol77:1514–1525.

Lindemann B. 1996. Taste reception. Physiol Rev 76:719–766.Lin W, Ogura T, Kinnamon SC. 1996. Responses to monosodium glutamate

and amiloride occur in single rat fungiform taste cells. Chem Senses21:632–633.

323GLUTAMATE RECEPTORS IN RAT TASTE BUDS

Lu KS, Roper SD. 1993. Electron microscopic immunocytochemistry ofglutamate-containing nerve fibers in the taste bud of mudpuppy (Nec-turus maculosum). Microscop Res Techniques 26:225–230.

Maggi CA. 1991. The pharmacology of the efferent function of sensorynerves. J Autonom Pharmacol 11:173–208.

Mayer EA, Koelbel CB, Snape WJ Jr, Eysselein V, Ennes H, Kodner A.1990. Substance P and CGRP mediate motor response of rabbit colon tocapsaicin. Am J Physiol 259:G889–G897.

Murayama N. 1988. Interaction among different sensory units within asingle fungiform papilla in the frog tongue. J Gen Physiol 91:685–701.

Myers AC, Undem BJ. 1991. Functional interactions between capsaicin-sensitive and cholinergic nerves in the guinea pig bronchus. J Phar-macol Exp Therap 259:104–109.

Nagai T, Dealy RJ, Welton J, Roper SD. 1998. Uptake and release ofneurotransmitter candidates, [3H]serotonin, [3H]glutamate, and [3H]g-aminobutyric acid, in taste buds of the mudpuppy, Necturus maculo-sus. J Comp Neurol 392:199–208.

Nagy I, Pabla R, Matesz C, Dray A, Woolf CJ, Urban L. 1994. Cobaltaccumulation in neurons expressing ionotropic excitatory amino acidreceptors in young rat spinal cord: morphology and distribution.J Comp Neurol 344:321–335.

Pruss RM, Akeson RL, Racke MM, Wilburn JL. 1991. Agonist-activatedcobalt uptake identifies divalent cation-permeable kainate receptors onneurons and glial cells. Neuron 7:509–518.

Pumplin DW, Yu C, Smith DV. 1997. Light and dark cells of rat vallatetaste buds are morphologically distinct cell types. J Comp Neurol378:389–410.

Roper SD. 1989. The cell biology of vertebrate taste receptors. Annu RevNeurosci 12:329–353.

Roper SD. 1992. The microphysiology of peripheral taste organs. J Neuro-sci 12:1127–1134.

Sugimoto K. 1994. Electrophysiological properties and chemically-inducedresponses of mammalian taste bud cells (Abstract). In: Kawamura Y, KareMR, editors: Olfaction and taste. Vol XI. New York: Dekker. p 111.

Sugimoto K. 1996. Patch-clamp study of taste transduction mechanisms(Abstract). Chem Senses 21:81–82.

Williams LR, Prezenger JF, Oostven JA. 1992. Induction of cobalt accu-mulation by excitatory amino acids within neurons of the hippocampalslice. Brain Res 581:181–189.

Yang R, Crowley HH, Rock ME, Kinnamon JC. 1998. Immunocytochemicalcharacterization of SNAP-25 in rat taste buds (Abstract). Chem Senses23:592.

Zahm DS, Munger BL. 1983, Fetal development of primate chemosensorycorpuscles. I. Synaptic relationships in late gestation. J Comp Neurol213:146–162.

Zhou N, Taylor DA, Parks TN. 1995. Cobalt-permeable non-NMDA receptorsin developing chick brainstem auditory nuclei. NeuroReport 6:2273–2276.

324 A. CAICEDO ET AL.