Journal of Neurophysiology

Group III Metabotropic Glutamate Receptors (mGluRs) Modulate Transmission of Gustatory Inputs in the Brain Stem

Robert M. Hallock, Christopher J. Martyniuk, Thomas E. Finger


Glutamate is the principal neurotransmitter at the primary sensory afferent synapse in the medulla for the taste system. At this synapse, glutamate activates N-methyl-d-aspartate (NMDA) and non-NMDA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid [AMPA] and kainate) ionotropic receptors to effect a response in the second-order neurons. The current experiment is the first to examine the role of metabotropic glutamate receptors (mGluRs) in the transmission of taste information. In an in vitro slice preparation of the primary vagal gustatory nucleus in goldfish, primary gustatory afferent fibers were stimulated electrically, whereas evoked dendritic field potentials were recorded in the sensory layers. Recordings were made before, during, and after bath application of mGluR agonists for various mGluR groups and subtypes. Whereas l-AP4, a group III agonist, reduced the field potential, group I and group II agonists had no effect. Furthermore, the selective mGluR4 agonist ACPT-III and mGluR8 agonist PPG were effective at reducing the field potential, whereas agonists selective for mGluR6 and 7 were not. MAP4, a group III mGluR antagonist, attenuated frequency-dependent depression, indicating that endogenous glutamate binds to presynaptic mGluRs under normal conditions. Furthermore, polymerase chain reaction showed that mRNA for mGluR4 and 8 is expressed in the vagal ganglia, a prerequisite if those receptors are expressed presynaptically in the vagal lobe. Collectively, these experiments indicate that mGluR4 and 8 are presynaptic at the primary gustatory afferent synapse and that their activation inhibits glutamatergic release.


The rostral portion of the nucleus of the solitary tract (nTS) is a complex assembly of subnuclei in the caudal medulla that receive gustatory and visceral information from the periphery (Loewy 1990). Gustatory information is carried from taste receptors on the tongue to the nTS by the chorda tympani nerve (CN VII), whereas both gustatory and visceral information are carried to the nTS by the glossopharyngeal (CN IX) and vagus (CN X) nerves. Whereas the chorda tympani nerve innervates the anterior two thirds of the tongue and projects to the rostral nTS, the glossopharyngeal and vagus nerves innervate the posterior tongue and oral cavity and project to the caudal portion of the nTS (Hamilton and Norgren 1984; Norgren 1985). The primary neurotransmitter at the primary afferent terminals of gustatory nerves is glutamate (Bradley and Grabauskas 1998; Li and Smith 1997; Smeraski et al. 1999; Smith et al. 1998). Glutamate has three types of receptors: N-methyl-d-aspartate (NMDA) and non-NMDA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid [AMPA] and kainate) ionotropic receptors and metabotropic receptors (mGluRs). Although the ionotropic receptors are well characterized in the taste system (Smeraski et al. 1999; Wang and Bradley 1995), little attention has been paid to the possible actions of mGluRs at the primary gustatory synapse, despite findings in other glutamate- mediated systems that show the importance of mGluRs in regulation of transmitter release (Best et al. 2005; Browning and Travagli 2007; Chen et al. 2002; Glaum and Miller 1993; Han et al. 2006; Neugebauer et al. 2000; Nistri et al. 2006; Slattery et al. 2006; Thomas et al. 2001; Travagli et al. 2006).

In the present study, we take advantage of the taste system of the common goldfish (Carassius auratus) to determine how mGluRs are involved in gustatory processing at primary afferent nerve terminals of the vagus nerve. The goldfish has a highly specialized taste nucleus, the vagal lobe, which is homologous to the vagal gustatory portion of the nTS (Finger 2008; Finger and Dunwiddie 1992; Morita and Finger 1985, 1987). Afferent nerve fibers from general visceral and gustatory ganglia are anatomically segregated in the vagal complex of goldfish with gustatory fibers projecting to distinct laminae in the vagal lobe (Morita and Finger 1985, 1987) (see Fig. 1). This is in contrast to the dominant rodent models where visceral centers occur in nuclei adjacent to gustatory centers (Altschuler et al. 1989; Broussard and Altschuler 2000), making the two components difficult to parse. The organization of the vagal lobe facilitates anatomical and physiological studies of taste processing and represents a distinct advantage over rodent models in that it allows us to solely examine gustatory processing separate from general visceral components.

FIG. 1.

A: a photograph of a dissected goldfish brain, showing the main parts of the brain as well as branches of the vagus nerve. Vertical bar over the vagal lobe represents the plane of B. B: a photomicrograph of a transverse section of a vagal lobe that shows the vagal nerve as well as the sensory, fiber, and motor layers. Primary afferent fibers are labeled by horseradish peroxidase (HRP) injection into the palatal nerve (preparation by Y. Morita) and demonstrate the distinctive laminar organization of the system. Also shown are positions of the stimulating and recording electrodes. The stimulating electrode is placed on the vagal nerve fibers, whereas the recording electrode is placed within the sensory layers of the vagal lobe where the vagus nerve terminates. C: a typical vagal lobe field potential recording, showing the stimulus artifacts (Pulse 1, Pulse 2) from a pair of electrical pulses that were separated by 30 ms and subsequent negative potentials (N1, N2, and N3). The amplitudes of N2 and N3 were measured as indicated.

First, we tested whether cells in the vagal lobe could respond to agonists specific to the broad groups of mGluR and then tested selective mGluR agonists to narrow down which mGluRs subreceptors were present in the lobe.

Second, we performed additional electrophysiological studies to demonstrate whether these receptors are presynaptically expressed. Third, we tested whether mGluRs are activated by synaptically released glutamate. Specifically, these studies demonstrated how mGluRs modulate activity in the lobe and contribute to the processing of taste responses in the caudal brain stem. Last, we used polymerase chain reaction (PCR) to test whether the receptor mRNA was present in the sensory ganglion of the vagus nerve, a necessary condition if the mGluRs are on the afferent nerve terminals.


All animal procedures were performed in accordance with National Institutes of Health guidelines and all protocols were approved by the Institutional Animal Care and Use Committee at the University of Colorado Denver Health Sciences Center.



Forty-seven goldfish (10–20 cm, Carassius auratus) were prepared for electrophysiological recordings as described previously (Finger and Dunwiddie 1992; Smeraski et al. 1999). Briefly, fish were anesthetized with 0.15 g/l tricaine methanesulfonate (MS 222; Sigma), quickly decapitated, and their brains immediately removed. The vagal lobe was then blocked in the transverse plane, fixed to a platform with a tissue acrylate (Vetbond; 3M, St. Paul, MN), embedded with a 2% low-temperature agar (Agarose, Type IX; Sigma), and then cut in 400-μm slices on a vibratome. Slices were then placed in aerated artificial cerebrospinal fluid (aCSF) for ≥90 min until electrophysiological recordings began.

aCSF was a standard solution containing (in mM) 131 NaCl, 20 NaHCO3, 2 KCl, 1.25 monobasic KH2PO4, 2 MgSO4, 2.5 anhydrous CaCl2, and 10 dextrose. aCSF was then oxygenated (95% O2-5% CO2) for 10 min and then pH balanced to 7.40 ± 0.01 (Mathieson and Maler 1988).


Slices were transferred to a recording chamber that was constantly superfused with aerated aCSF at a rate of 2 ml/min. A Teflon-coated nichrome bipolar electrode, insulated except for the tips, was placed over the vagal nerve root. This electrode was coupled to an A-360 stimulus-isolation unit (World Precision Instruments, Sarasota, FL) and controlled by a pCLAMP-8 system coupled to a Digidata 1320 (Axon Instruments, Foster City, CA) analog to digital converter (Axon Instruments). The recording electrode was a 3 M NaCl-filled glass micropipette (∼5 MΩ, 0.86 mm ID, 1.5 mm OD) and was placed in layers VI to VIII of the vagal lobe, within the sensory layers of the lobe (Morita and Finger 1985). Electrical stimulation was delivered in pairs of pulses with an interpulse interval (IPI) of 30 ms until a reliably evoked field potential was found. Pulses were 0.2 ms in duration. After a stable evoked response was recorded, pairs of pulses were delivered at a current level that elicited a response that was 75% of the maximal response and a pair of pulses was delivered every 30 s. Data were digitized at 10 kHz and stored for later off-line analyses. For each recording session, average waveforms were computed by pCLAMP-8 software. Electrophysiological signals were amplified with a DAM 80-I isolated Bio-amplifier (World Precision Instruments). pCLAMP software calculates an average waveform for a user-defined series of sweeps. In a typical field potential, the artifact from electrical stimulation is followed by three negative-going potentials: N1, N2, and N3 (see Fig. 1C). N1 represents evoked activity from vagal fibers in the lobe; N2 reflects monosynaptic connections; and N3 represents responses from higher-order cell populations. Finger and Dunwiddie (1992) established the nature of these potentials; they showed that removal of extracellular Ca+2 from the bath solution in these recordings eliminated N2 and N3, but not N1. Removal of calcium eliminates synaptic transmission, whereas axonal transmission remains unaffected. Thus N1 was shown to reflect axonal activation of afferent fibers, whereas N2 and N3 reflect synaptically mediated events within the cell layers. The single peak at N1 in Ca-free Ringer's strongly suggests that the incoming fiber volley consists of a single wave of activity reflecting a uniform diameter to the incoming primary afferent fibers; i.e., the conduction velocity is similar across all primary afferent fibers. Roughly 3 ms following N1, N2 occurs, and this represents monosynaptic transmission at the primary afferent synapse. This time course for monosynaptic transmission is also consistent with synaptic currents seen in whole cell patch-clamp recordings in vagal lobe cells (unpublished observations). Subsequent potentials (N3), which occur about 3 ms after N2, thus represent polysynaptic connections. The amplitudes of N2 and N3 were measured by subtracting the amplitude of the negative-going potential from the amplitude of the preceding inflection point (Sharp and Finger 2002). N2s after the first and second pulses are denoted as N2/p1 and N2/p2, respectively, whereas N3 after the second pulse is denoted as N3/p2. N3 after the first pulse was rarely present and was not measured. The measurements were taken with Axon software. These average amplitudes for each part of the recording session were then scaled to the baseline level, which was set to 100. Relative increases or decreases in waveform amplitude were then compared between recordings in different slice preparations, as is typical of these field potential studies due to the positioning of both the recording and stimulating electrodes (Sharp and Finger 2002).


mGluR agonists were infused into the flow of aCSF (2 ml/min) at 0.02 ml/min (1:100 dilution). This low flow rate did not change the overall flow rate in the recording chamber or the pH of the aCSF. Highly reproducible results across agonist and antagonist drug applications suggest that the concentration of drugs reach a steady state during delivery, but slight fluctuations in the concentration gradient of drugs during applications cannot be ruled out. A full list of the agonists and their final diluted concentrations in the recording chamber are reported in Table 2 with results from those drug applications. (S)-2-Amino-2-methyl-4-phosphonobutanoic acid (MAP4), an mGluR antagonist, required a final bath concentration of 0.5 mM. It was impossible to create a concentration of this drug 100 times >0.5 mM to deliver into the flow of aCSF. Therefore the flow of aCSF was shut down for experiments with this drug and a solution of this drug was administered directly to the bath solution with the slice. All drugs were obtained from Tocris Bioscience (Ellisville, MO).


Application of mGluR agonists. Initial experiments assessed a wide range of available mGluR agonists to test whether any changed the evoked field potential. Each mGluR agonist was applied to three to five slice preparations. A baseline recording consisted of about 10 pairs of pulses. Next, the mGluR agonist was delivered for about 10 min and then the recording continued for around 45 min. If any drug had no significant effect on field potentials, l-2-amino-4-phosphonobutanoate (l-AP4), a group III mGluR agonist that proved effective in reducing the amplitude of N2 and N3 of the field potential (see Fig. 3), was delivered to verify the viability of the slice. In these cases, recovery from l-AP4 was also recorded. For each of the three critical potentials (N2/p1, N2/p2, and N3/p2), a ratio was taken between the amplitude of the potential under influence of the drug compared with the amplitude of the potential during baseline activity. A one-sample t-test was used to test whether this ratio was significantly different from 1.0. To account for the three comparisons performed on each data set, a Bonferroni correction lowered the alpha criterion for significance from 0.05 to 0.0167. All statistics were performed on these ratios, although the data are graphically represented as percentages.

Experiments on the paired-pulse ratio.

One common way to evaluate whether changes are due to presynaptic mechanisms is to examine the paired-pulse ratio (PPR). A low PPR is evident when the amplitude of a potential after a second pulse in a pair of pulses is lower than the amplitude after the first pulse and it is thought that this is mediated presynaptically (Gingrich and Byrne 1985; Patil et al. 1998). If mGluRs modulate activity via a presynaptic mechanism, then a low PPR should occur when pulses are delivered at short IPIs and the PPR would become higher (to 1.0) as the duration between the two pulses is increased sufficiently to where the first pulse has no effect on the second pulse. Furthermore, the PPR should increase toward 1.0 in the presence of an effective mGluR agonist, even at short IPIs. Physiology experiments used l-AP4, a drug that was effective in the first experiment, to test whether it was effective in increasing the PPR. This PPR experiment was conducted using l-AP4 with IPIs that ranged from 10 to 1,000 ms and the PPR was calculated, as defined in Fig. 5. An ANOVA with treatment (baseline, with l-AP4, and recovery) and time (seven levels of IPIs) as main factors was conducted on these data. Planned (a priori) contrasts were conducted on the PPR for individual IPIs. These linear contrasts compared the PPR during l-AP4 administration to the average PPR of baseline and recovery at each of the seven IPIs.

Studies with an mGluR antagonist.

The ability of a group III mGluR antagonist (MAP4) to reverse the effects of the group III agonist was tested. In these experiments, a field potential was isolated as described earlier and l-AP4 was added directly to the bath solution (final bath concentration = 0.25 μM). After a decrease in amplitude of the field potential was observed (∼5 min), MAP4 (final bath concentration = 0.5 mM) was added to the bath solution. The stimulation protocol and analyses were identical to those described earlier for the experiments on mGluR agonists. These data were analyzed with two-tailed t-test that compared the amplitude of potentials during baseline activity to the amplitudes during both applications of l-AP4 and l-AP4 plus MAP4. A Bonferroni correction was used to account for multiple tests.

We tested the ability of MAP4 to attenuate frequency-dependent synaptic depression. The protocol we followed for electrical stimulation and subsequent analyses was identical to that previously used (Chen et al. 2002). This protocol delivers trains of stimulation at different frequencies, such that frequency is directly correlated with glutamate depletion and inversely proportional with mGluR binding. If mGluRs are endogenously activated and subsequently depress postsynaptic activity, then blocking those receptors with an antagonist will lead to less of a decrease in synaptic activity (Chen et al. 2002). If, on the other hand, mGluRs are not endogenously activated in this system, then there will be no change in the field potential in the presence of the antagonist.

In this paradigm, baseline responses to trains of stimulation were assessed and then these patterns were repeated in the presence of the group III mGluR antagonist MAP4. A baseline response to a set of 10 single pulses delivered at 0.2 Hz was recorded and was then followed by a randomly chosen train of stimulation (0.4, 3, 9, 24, and 48 Hz; 30 pulses for each train); this continued until all five trains were delivered. Then, MAP4 was applied to the slice and sets of baseline and stimulation patterns were randomly delivered until all five stimulation patterns were delivered. The average response of the last five stimulations of the baseline session was taken. Similarly, the last 10 responses for each train of stimulation were averaged. Then, the ratio of the amplitude of N2 after the stimulation train to the amplitude of N2 at the preceding baseline session was taken at each frequency before and after application of MAP4. A two-factor repeated-measures ANOVA with drug (2 levels) and frequency (5 levels) was performed on these data.


If mGluRs are located presynaptically, then the relevant ganglion cells should express the appropriate mRNA. To test for mRNA expression of mGluRs, we used PCR on the vagal ganglion that innervates the pharyngeal taste buds; in goldfish, this ganglion is entirely separate from that providing general visceral innervation to the heart and digestive tract. mRNA expression of any mGluR is obligatory if those receptors are present in the vagal afferent nerve terminals in the vagal lobe.


Sequence information for mGluRs was obtained from the National Center for Biotechnology Information (NCBI) and aligned using ClustalW (EMBL-EBI; Homologous sequence information for mGluRs was taken from Danio rerio, Takifugu rubripes, Ictalurus punctatus, and Rattus rattus, when available. These sequences were aligned and conserved regions were used for primer design. Primer3 (Rozen and Skaletsky 2000) was used to design gene-specific or degenerate primers. Primer sequences are given in Table 1 along with the amplicon size for each receptor subunit. PCR was performed to determine whether transcripts for mGluR1, 4, 6, 7, and 8 were expressed in the whole brain, vagal ganglia, kidney, and liver of the goldfish. The positive control was β-actin, which is widely present in these tissues (Maruyama et al. 2006). All partial mGluR sequences for goldfish have been submitted to NCBI GenBank.

View this table:

Primers used to sequence goldfish mGluRs and β-actin


Standard protocols for RNA extraction, reverse transcription, and PCR from the resultant cDNA were used. RNA extraction was performed using the TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Whole brain, the oropharyngeal ganglia of the vagus nerve, kidney, liver, olfactory bulbs, and lateral line nerve were harvested from three adult goldfish. Tissue was immediately frozen under liquid nitrogen, cut on a cryostat in 10- to 16-μm sections with a clean blade, and immediately placed in TRIzol. These tissues were homogenized by passing them with TRIzol through a sterile 1-cm2 syringe and 22-gauge needle and were then ready for RNA extraction.

RNA (5 μl) was DNAse treated by adding 1 μl of DNAse I (Roche), followed by incubation at 37°C for 30 min and 70°C for 10 min to inactivate the enzyme. Next, 1 μl of oligo dT-primer (Invitrogen: 500 ng/μl) was added and then incubated at 65°C for 5 min, followed by cooling to room temperature for 15 min. For reverse transcription, a mixture of 10 μl 5× SS II-RT buffer (Invitrogen), 1 μl SS II-RT (Invitrogen), 4 μl DTT (Invitrogen), 1 μl RNAse inhibitor, and 5 μl dNTPs (10 mM) per reaction was added to the above-cited product (i.e., cDNA, DNAse I, and oligo dT-primer). The solution was mixed and incubated at 42°C for 60 min, followed by 10 min at 70°C. The samples were then stored at −20°C or used immediately for PCR.

PCR was performed with HotMaster taq (Eppendorf). For each reaction, 2 μl of 10× HotMaster Taq buffer, 10 μl dNTPs mix (10 mM), 1 μl of each of the forward and reverse primers (see following text for details), 1 μl of cDNA from reverse transcription, 0.2 μl of HotMaster Taq (Eppendorf), and 14.3 μl of RNA-free distilled water was used. Thermal cycling was done in an Eppendorf Master-cycler gradient PCR machine and conditions were as follows: enzyme activation at 95°C for 5 min, followed by 30 cycles of 95°C for 1 min, primer annealing at optimal temperature for 30 s, and extension at 72°C for 45 s. After cycling, there was an additional 5 min at 72°C. PCR products were run on a 2% agarose gel via electrophoresis in concert with a low molecular weight ladder and then visualized by ethidium bromide staining. Last, successful PCR products were purified using a ProbeQuant G-50 Purification Kit (GE Healthcare, Piscataway, NJ) and then sequenced at the University of Colorado Denver core facility. All PCR products reported herein matched the mGluR sequences previously identified.




As previously reported, electrical stimulation of vagal afferent fibers in an in vitro slice preparation of the vagal lobe evokes a dendritic field potential with multiple components (Finger and Dunwiddie 1992; Smeraski et al. 1999). In the field potential, N1 represents evoked activity from vagal fibers in the lobe, N2 reflects primary afferent transmission, and N3 represents responses from polysynaptically connected cell populations (Finger and Dunwiddie 1992) (see Fig. 1C). Administration of aCSF instead of the drug (n = 3) did not change any of the potentials from baseline levels (P > 0.25 for each comparison) (see Fig. 2A). Stable field potential recordings were made for ≤90 min, with no rundown in signal or evidence of desensitization (data not shown). This also indicated that there was no significant change in activity over the time course of the recording. To test for the effectiveness of mGluRs, a broad-spectrum mGluR agonist was used. In mammals, the broadest mGluR agonist is (±)-1-aminocyclopentane-trans-1,3-dicarboxylic acid (ACPD). At 50 μM, it is a strong agonist to all group I and group II receptors and also a weak agonist to the group III subreceptors mGluR 4, 6, and 8; it is inactive to mGluR7 (Cartmell and Schoepp 2000). Bath application of this drug (50 μM, n = 5) resulted in a significant decrease of N2/p1 (P = 0.005) and N3/p2 (P < 0.01). N2/p2 was not reduced by application of ACPD (P > 0.25). Washout of the drug (∼30 min) resulted in a partial return of activity to baseline levels (see Fig. 2). See Table 2 for the percentage change of field potentials for each of the three potentials for various drug conditions.

FIG. 2.

A: the recordings are stable across time and the administration of artificial cerebrospinal fluid (aCSF) does not change the field potential. See above for recording conditions and a description of the graph. Bars represent SE; n = 3 slices. B: a bar graph depicting the effects of (±)-1-aminocyclopentane-trans-1,3-dicarboxylic acid (ACPD) on the 3 components of the field potential. In this figure, the amplitude of N2 after pulse 1 and pulse 2, and N3 after pulse 2 are each shown during drug and recovery (Recov) and are represented as a percentage of baseline activity, which is depicted by the dotted line at 100%. ACPD (50 μM) significantly decreased the amplitudes of N2/p1 and the N3/p2. The amplitudes of these potentials returned to baseline levels after recovery, when the drug was washed out. Bars represent SE; n = 4 slices. *P < 0.05; **P < 0.01. C: average traces comparing ACPD (dark line) to baseline (lighter trace). Scale bar represents 5 ms and 100 μV and applies to both C and D. Each trace is the average of ≥5 individual sweeps. D: average traces comparing recovery (dark line) to baseline (lighter line). Notice they are nearly overlapping traces.

View this table:

Effect of mGluR agonists on the three components of the fEPSP

Specific group I, group II, and group III mGluR agonists were then used to determine which groups were present. Application of the broad group I agonist 3,5-dihydroxyphenylglycine (DHPG; 100 μM, n = 3) and group II agonist (2R,4R)-4-aminopyrrolidine-2,4-dicarboxylate (APDC; 10 μM; n = 3) had no effect on the field potential (P > 0.25 for all comparisons), despite that these drugs are effective in fish at these concentrations (Gafka et al. 1999; Poli et al. 1999; Zhang and Schmidt 1999) (see Fig. 3, A and B).

FIG. 3.

Group I and group II agonists do not change the field potential, but a group III agonist does. Administration of a group I metabotropic glutamate receptor (mGluR) agonist, 100 μM 3,5dihydroxyphenylglycine (DHPG, A) and a group II agonist, 10 μM (APDC, B), did not change any component of the field potential. The group III agonist, 2.5 μM l-2-amino-4-phosphonobutanoate (l-AP4), significantly attenuated the amplitude of N2 after the first pulse and N3 after the second pulse. Bars represent SE; n = 3 slices for each group I and group II agonists, but 5 for the group III agonist. **P < 0.01. D: a log-plot dose–response curve for the inhibitory effects of l-AP4 on the amplitude of N2 after the 1st pulse. This graph shows that a low concentration (0.025 μM) does not decrease the amplitude of N2, but 2.5 μM inhibits N2 by about 50%. Bars represent SE; n = ≥3 cases for each data point. E: field potential comparing baseline traces (light gray line) with l-AP4 (darker line). F: field potential from same recording in E, comparing baseline (light gray) with recovery (black). Each trace is the average of at least 5 individual sweeps.

Application of the broad group III mGluR agonist, l-AP4 (0.25 or 2.5 μM) significantly attenuated N2 and N3 potentials similar to the broad mGluR agonist ACPD (see Fig. 3, CF). The low concentration of l-AP4 (0.25 μM; n = 8) significantly attenuated N2/p1 and N3/p2 (P < 0.01), but N2/p2 was not reduced (P > 0.25). During recovery, all the affected potentials reached baseline levels after washout of the drug (∼20 min). A high concentration of l-AP4 (2.5 μM; n = 6) had more profound effects on the field potential. Here, all three components of the field potential were significantly attenuated (P < 0.01 for each comparison). Furthermore, activity after washout of the drug (∼30 min) only partially recovered; the amplitude of N2/p2 did not recover after washout of the drug (P < 0.001). The effectiveness of l-AP4 was dose dependent (Fig. 3D), with an IC50 (for N2/p1) of 4.33 μM.

Next, selective agonists for group III mGluR subreceptors were tested at concentrations found effective in other studies. These included agonists for mGluR4a (Acher et al. 1997), mGluR6 (Brauner-Osborne et al. 1996), mGluR7 (Mitsukawa et al. 2005), and mGluR8 (Gasparini et al. 1999). Drugs selective for mGluRs 4a and 8 attenuated the field potential, whereas drugs for mGluRs 6 and 7 had no effect (see Fig. 4). The mGluR4a agonist (3RS, 4RS)-1-aminocyclopentane-1,3,4-tricarboxylic acid (ACPT-III; 40 μM; n = 4) significantly attenuated N2/p1 (P = 0.001) and N3/p2 (P < 0.005). Additionally, the mGluR8 agonist (RS)-4-phosphonophenylglycine (PPG; 1 μM; n = 4) attenuated the field potential. Although N2/p1 (P = 0.065) and N2/p2 (P > 0.15) were unchanged, N3/p2 was significantly attenuated (P < 0.005).

FIG. 4.

Within group III mGluRs, mGluR4a [(3RS,4RS)-1-aminocyclopentane-1,3,4-tricarboxylic acid (ACPT-III)] significantly reduced N2/p1 and N3/p2 (n = 4), similar to the effects of both the broad mGluR agonist and the general group III mGluR agonist. mGluR8 (1 μM PPG) agonist was effective only in reducing N3/p2 (n = 3), whereas agonists for mGluR6 [100 μM 2-amino-4-(3-hydroxy-5-methylisoxazole-4-yl)butyric acid (homo AMPA)] (n = 5) and mGluR7 [1 μM N,N′-bis(diphenylmethyl)-1,2-ethanediamine dihydrochloride (AMN-082)] (n = 3) agonists were not effective in reducing any component of the field potential. Bars represent SE. **P < 0.01.

Conversely, the mGluR6 agonist 2-amino-4-(3-hydroxy-5-methylisoxazole-4-yl) butyric acid (homo AMPA; 100 μM; n = 3) did not reduce any component of the field potential (P > 0.05 for each comparison). This concentration is effective in other systems; the EC50 for homo AMPA is 82 μM (Brauner-Osborne et al. 1996). Additionally, the mGluR7 agonist N,N′-bis(diphenylmethyl)-1,2-ethanediamine dihydrochloride (AMN-082; 1 μM; n = 3) had no effect on the field potential (P > 0.05 for each comparison). This drug is an allosteric modulator and activates signaling via a site within the transmembrane domain, not within the typical glutamate binding site (Mitsukawa et al. 2005). Allosteric modulators often have no effect on their own; rather, they potentiate endogenous activity (Conn and Niswender 2006). AMN-082 was also ineffective in our preparation in the presence of a low concentration of l-AP4 (data not shown). The concentration of AMN-082 we used is effective in other systems; Mitsukawa et al. (2005) found that the EC50 for this drug was 300 nM.

mGluRs are presynaptic and activated by endogenously released glutamate

In our electrophysiology experiments, application of l-AP4, a group III mGluR agonist, decreases the amplitude of the field potential. To address whether these changes were mediated presynaptically, we examined the PPR. Under baseline recordings, the average PPR increased logarithmically with an increase in IPI; the average PPR was 0.34 at 10-ms IPI, and 0.85 at 1,000-ms IPI (Fig. 5; n = 7). l-AP4 (2.5 μM) significantly increased the PPR at IPIs ≤50 ms. An ANOVA with treatment (baseline, with l-AP4, and recovery) and time (seven levels of IPIs) as main factors revealed a significant main effect of treatment [F(2,79) = 25.17, P < 0.0001]. The PPR during l-AP4 was significantly higher than the PPR of baseline (0.57 ± 0.05) and recovery (0.64 ± 0.05). There was also a significant effect of time [F(6,79) = 20.15, P < 0.0001]. The PPR was significantly lower at short IPIs than that at long IPIs. Linear contrasts revealed that the PPR during l-AP4 application was significantly greater than baseline and recovery at 10 ms (t = 5.01, P < 0.0001), 20 ms (t = 3.10, P = 0.0027), 30 ms (t = 2.36, P = 0.021), and 50 ms (t = 2.90, P = 0.005) IPI. No differences were present at higher IPIs (all associated P values were >0.25). This significance in PPR at short IPIs is attributable entirely to a decrease in the amplitude of N2/p1 during l-AP4 application. That is, the amplitude of N2/p1 was reduced so that it was roughly equivalent to the amplitude of N2/p2, which was largely unchanged by drug application (see Fig. 5, inset). Thus l-AP4 caused an increase in the PPR from baseline levels by inhibiting the amplitude of N2 following the first pulse. This experiment suggests that the group III mGluRs exhibit their effects via a presynaptic mechanism.

FIG. 5.

Bar graph showing effects of l-AP4 on paired-pulse ratio (PPR) as a function of IPI. Pulses with various (10 to 1,000 ms) interpulse intervals (IPIs) were randomly presented before (baseline), during l-AP4, and after washout of the drug (Recov). The average PPR was taken for each IPI, and the means presented here represent the average PPR from 7 separate slice recordings. Application of l-AP4 significantly increased the PPR at IPIs ≤50 ms, and the PPR returned to baseline levels after washout of the drug. Bars represent SE. *P < 0.05; **P < 0.01. Inset: results from a typical session showing overlapping traces of baseline recording (black) and during mGluR agonist administration (gray). This graph shows that the drug decreased the amplitude of N2 after the first pulse without changing the amplitude of N2 after the second pulse. Consequently, this leads to an increase in the PPR, which is due solely to a decrease in the amplitude of N2 after the 1st pulse. Also, N3 after the 2nd pulse is largely eliminated with l-AP4. The formula for calculating the PPR is also given.

The group III mGluR antagonist MAP4 blocked l-AP4-induced depression of the field potentials (see Fig. 6, AC). These results replicated the finding that l-AP4 significantly reduced the amplitudes of N2/p1 and N3/p2 (P values <0.05) in the field potential, as well as confirmed drug specificity. Addition of MAP4 eliminated the l-AP4–induced depression for the amplitude of N2/p1 (P < 0.05) and N3/p2 (P < 0.01), increasing the amplitude of these two potentials to baseline levels.

FIG. 6.

(S)-2-Amino-2-methyl-4-phosphonobutanoic acid (MAP4) reverses the effects of l-AP4 and reveals that group III mGluRs are activated in the vagal lobe by endogenously released glutamate. In a typical recording, l-AP4 causes a decrease in evoked activity in the vagal lobe (A), but subsequent application of MAP-4 reverses this decrement (B). C: the average decreases in the 3 potentials of the waveform across slices. n = 3; bars represent SE. *P < 0.05. D: group data showing the amplitude of N2 during the last 10 stimulations of a train relative to the amplitude of N2 during baseline stimulation, during a series of frequencies, in aCSF and in the presence of MAP4; n = 6. A 2-factor repeated-measures ANOVA with drug (2 levels) and frequency (5 levels) was performed on these data. *P < 0.05.

The vagal afferent fibers were stimulated with a series of five frequencies of electrical stimulation under baseline levels and then with the application of MAP4. These results showed that MAP4 attenuated frequency-dependent synaptic depression (see Fig. 6D). An ANOVA showed that there was a main effect of frequency [F(4,20) = 26.47, P < 0.001]. The degree of depression significantly increased with an increase in frequency. There was also a main effect of drug (no drug vs. MAP4) [F(1,5) = 14.59, P = 0.01]. The average decrease in activity without the drug was greater than the decrease observed under MAP4. Last, the amount of depression across frequencies was significantly different between the drug conditions, as evidenced by a significant interaction between these factors [F(4,20) = 3.144, P = 0.04] (see Fig. 6D).

Partial mGluR sequences of goldfish mGluRs

A number of partial sequences for mGluRs were cloned in this study. The partial goldfish mGluR4 sequence (783 bp; NCBI accession number EU147495) shows 95% amino acid identity with zebrafish mGluR4 (XM_690474.2) and 75% amino acid identity with human mGluR4 (BC130528.1). The partial goldfish mGluR6 sequence (677 bp; accession number EU147497) shows 99% amino acid identity with zebrafish clone (NM_001080020.1) and 95% amino acid identity with human mGluR6 (NP_000834.2). The partial goldfish mGluR7 sequence (747 bp; accession number EU147496) shows 87% amino acid identity with zebrafish clone (XM_692122.2) and 72% amino acid identity with human mGluR7 (AF458054_1). The partial goldfish mGluR8 sequence (388 bp; accession number EU346360) shows 98% amino acid identity with zebrafish clone (XP_001334979.1) and 90% amino acid identity with human mGluR7 (EAL24322).

mGluR mRNAs in the vagal lobe

Results from the pharmacological experiments suggested the likelihood of presynaptic group III mGluRs on primary afferent terminals. If an mGluR is present in the central terminals of a ganglion cell, then the mRNA should be detectable in the ganglion. PCR demonstrated that the mRNA for mGluRs 4 and 8 is present in the vagal ganglion. Conversely, no mRNA for the mGluRs was found in the lateral line nerve, the negative control tissue. Mixed results were found in other tissues. For example, mGluR4 is detectable in the liver, mGluR8 in the olfactory bulb, and mGluR6 in the kidney. β-Actin, a ubiquitous protein, was detected in all six tissues (Fig. 7). The presence of mGluR4 and mGluR8 in the vagal ganglion is consistent with the pharmacological results, suggesting a presynaptic mode of action for these receptors. However, despite a relatively faint PCR band for mGluR7 in the ganglionic tissue, there was no pharmacological effect to the mGluR7 agonist we used.

FIG. 7.

Polymerase chain reaction (PCR) results from group III mGluRs 4, 6, 7, and 8 in the liver (L), vagal ganglia (G), brain (B), kidney (K), lateral line nerve (LL), and olfactory bulb (Olf). PCR product for mGluR4 and 8 is evident in the ganglion (G). β-Actin, a ubiquitous protein, was found in each tissue for each reaction. A low-mass DNA ladder (Invitrogen) was run with each reaction and is present in the middle of each panel. The molecular weights of this ladder are 2,000, 1,200, 800, 400, 200, and 100 base pairs, with the largest lanes at the top of each gel. The β-actin control represents 338 base pairs. Arrows represent bands of mGluR4 and 8 mRNA in the vagal ganglia, whereas arrowheads represent bands of expression for these receptors in brain.


In this study we have shown that 1) exogenously applied group III mGluR agonists inhibit the activity of second-order gustatory neurons in the vagal lobe, 2) the mRNA for group III mGluRs 4 and 8 is expressed in the ganglia of the vagal nerve, 3) group III mGluRs are expressed presynaptically, and 4) these receptors are activated by synaptically released glutamate. Together, these data indicate that these receptors likely play an important functional role in the synapse and contribute to the brain stem processing of gustatory information.

Only activation of group III mGluRs modulates synaptic transmission

mGluR agonists decreased the amplitude of synaptic potentials of the field potential in our in vitro slice preparation. A broad mGluR agonist, ACPD, decreased the amplitudes of N2 and N3 of the field potential, where N2 represents the synaptic potential at the primary afferent synapse and N3 at a higher-order synapse. Specific group I and group II mGluR agonists had no effect on synaptic potentials, but a group III agonist, l-AP4, attenuated N2 and N3 more effectively than did ACPD. This is probably because ACPD is a weaker agonist for group III mGluRs than l-AP4 (Conn and Pin 1997). Group I mGluRs are generally located postsynaptically, are positively coupled to phospholipase C, lead to inositol triphosphate production, and produce neuronal excitation (Cartmell and Schoepp 2000; Conn and Pin 1997). Conversely, group II and group III mGluRs are primarily located presynaptically, are negatively coupled to adenylyl cyclase, and inhibit cAMP formation (Cartmell and Schoepp 2000; Conn and Pin 1997). Within the group III mGluRs, agonists selective for either mGluR6 or mGluR7 were not effective, whereas agonists for mGluRs 4 and 8 were effective. The mGluR4 agonist reduced the amplitudes of N2/p1 and N3/p2, similar to the broad group III agonist. The mGluR8 agonist, on the other hand, significantly decreased only the amplitude of N3/p2, which is likely due to activity of higher-order synapses. Thus we posit that mGluR4 and, possibly, mGluR8 are present at the primary afferent terminals in the vagal lobe. This logic, however, is contingent on the assumption that these subreceptor-selective drugs are indeed selective and that they have similar affinities to the receptors in fish as in mammals. We are therefore unable to definitively conclude which subreceptors are responsible for the observed effects, but are nonetheless confident that group III mGluRs are responsible for the effects seen in these experiments.

We also found that mGluR agonists attenuate the N3 component of the field potential more robustly than N2. For example, PPG attenuated N3/p2 by an average of 83.1%, but decreased N2/p1 by an average of only 23.6%. The amplitude of N3 increases nonlinearly with respect to N2 as the current level of electrical stimulation is increased (unpublished observations). Thus we believe the effects of mGluR agonists on N3/p2 reflect intrinsic response properties of the higher-order synapses and are only partially attributed to the decrease in transmission at the earlier (N2) synapse. Consistent with this notion, Smeraski et al. (1999) asserted that the N3 component of the evoked field potential, which is largely NMDA-mediated, functions to amplify incoming activity from the primary afferent nerve. Our results further support this notion and suggest that this component of the response has a high threshold for activation and, once activated, potentiates an afferent signal.

Group III mGluRs are presynaptic

Three electrophysiological and molecular experiments provide evidence that group III mGluRs are located presynaptically. First, the paired-pulse experiment suggests that mGluRs are presynaptic. A low PPR was evident in baseline recordings when the IPIs were ≤50 ms. This effect is traditionally explained by neurotransmitter depletion (Takeuchi 1958; Thies 1965) because less neurotransmitter is available for release by the second pulse after the releasable pool has been depleted by the first pulse. More recently, specific possible mechanisms of decreasing the PPR have been described (Gingrich and Byrne 1985; Patil et al. 1998). These include: 1) a decrease in calcium channel activation after the first pulse and 2) a decrease in the efficacy of the synaptic machinery. Importantly, both these putative mechanisms lead to less neurotransmitter release from the presynaptic terminal. In our preparation, application of l-AP4 depressed the magnitude of the initial response to match that of the second response, resulting in a PPR of roughly 1.0. Second, the group III antagonist attenuated frequency-dependent depression of synaptic transmission and this effect is mediated presynaptically (Chen et al. 1999; Felder and Heesch 1987; Galarreta and Hestrin 1998; Mifflin and Felder 1988; Schild et al. 1995).

Third, results from PCR indicate that mRNA for mGluRs 4 and 8 is expressed in the vagal gustatory ganglion, which is a requirement if functional mGluRs are present on the gustatory afferent fiber terminals. Results for mGluRs 6 and 7, the other group III mGluRs, suggest that they are not responsible for the effects seen with the broad group III mGluR agonist. In our preparation, the agonist for mGluR6 was ineffective in vagal lobe slices and PCR failed to detect the mRNA transcript in the vagal ganglia. mGluR6 is thought to be expressed exclusively on the dendritic tip of bipolar cells in the outer plexiform layer of the retina (Nakajima et al. 1993), although the mRNA for the receptor has also been found centrally in brain areas important in visual processing (Ghosh et al. 1997). Conversely, the transcript for mGluR7 was detected in the vagal ganglia, although the selective agonist was ineffective in altering evoked responses in the slice preparation. The possibility that the drug is not effective in goldfish cannot be ruled out.

In summary, PCR data show that the mRNA for mGluRs 4, 7, and 8 is present in the ganglion, but pharmacology implicates only mGluRs 4 and 8 in presynaptic functions. Interestingly, mGluR8 immunostaining is located in the medial, rostral-central, and ventral subdivisions of the rat nTS (King 2003). These areas are primarily gustatory in nature (Harrer and Travers 1996) and the ventral subdivision is an area that sends an output to brain stem oromotor centers (Beckman and Whitehead 1991). Therefore mGluRs 4 and 8 are likely presynaptically present in the vagal gustatory system.

Postsynaptic mGluRs may also exist at this synapse, but the primary focus of this report has been on presynaptic mGluRs. Interestingly, mRNA for group II mGluRs (mGluRs 2 and 3) are not present in the vagal lobe of goldfish (Poli et al. 1999). It remains to be seen whether group I mGluRs, which are thought to be primarily postsynaptic (Conn and Pin 1997), may modulate glutamatergic transmission in the goldfish taste system. Additionally, it is noteworthy that two of the drugs used in this study (AMN-082 and ACPT-III) are known to be allosteric modulators. These drugs do not directly bind to the receptor site, but rather some other location that in turn influences the receptor. Thus results from these drugs may lead to an interaction between several binding sites that obfuscate the results from these drug applications.

mGluRs are endogenously activated

The experiment with MAP4, the group III mGluR antagonist, demonstrates that mGluRs are endogenously activated when the system responds to afferent nerve stimulation. Importantly, we stimulated the nerve with frequencies that are typical of taste responses in peripheral gustatory nerves (Frank 1973, 1991). We stimulated the nerve and observed frequency-induced depression and then found that the addition of a group III mGluR antagonist attenuated this depression, albeit by a small but significant amount. It is notable that this magnitude of attenuation is comparable to that of previous results with this technique (Chen et al. 2002). Specifically, we found that mGluRs were endogenously activated at the highest frequencies (>9 Hz) of electrical stimulation. This finding supports the idea that group III mGluRs may function as overflow detectors of synaptic glutamate and function to attenuate transmission after taste afferent fibers have been activated, as suggested in other systems (Best et al. 2005; Corti et al. 2002; Gereau and Conn 1995; Glaum and Miller 1993; Grueter and Winder 2005; Hirasawa et al. 2002; Hoang and Hay 2001; Lacey et al. 2005; Liu et al. 1998; Page et al. 2005; Zhai et al. 2002). The perisynaptic location of group III mGluRs in other systems (Shigemoto et al. 1997) supports this notion.

Functionally, presynaptic mGluRs underlie habituation to sensory stimuli (Best et al. 2005; Dolan and Nolan 2002; Simmons et al. 2002). In the caudal portion of the nTS of the rat (responsible for cardiovascular, respiratory functions, gastrointestinal, and feeding behaviors), group II and group III mGluR agonists act presynaptically to bind overflow glutamate and subsequently depress synaptic transmission (Browning and Travagli 2007; Chen et al. 2002; Glaum and Miller 1993; Hay and Lindsley 1995; Liu et al. 1998). In nociception, mGluR agonists act presynaptically to attenuate transmission of pain information, resulting in analgesia (Dolan and Nolan 2002; Simmons et al. 2002). In olfaction a broad mGluR agonist significantly attenuated field potential recordings from the anterior piriform cortex in in vitro recordings in rats and, when a group III mGluR antagonist was administered, the rats failed to habituate to the odor (Best et al. 2005).

Our studies establish that group III mGluRs, most likely mGluRs 4 and 8, presynaptically modulate afferent gustatory information. These receptors, once activated by excess glutamate in the synaptic cleft, inhibit subsequent glutamatergic release from the presynaptic terminal. The greater the afferent nerves stimulation, the more glutamate is released from the terminals, which leads to more glutamate binding to these group III mGluRs. This decreases excitatory input to the dendritic spines of second-order gustatory cells in the vagal lobe and attenuates the response from these cells to taste stimuli. We posit that mGluRs act as overflow detectors and attenuate transmission after afferent taste fibers have been activated. In this way, mGluRs may contribute to adaptation or habituation in the gustatory system.


This work was supported by National Institute on Deafness and Other Communication Disorders Grants R01-DC-00147 to T. E. Finger, F32-DC-009158 to R. M. Hallock, and P30-DC-04657 to P. Restrepo and T. E. Finger.


We thank Drs. Nathan Schoppa and Jozsef Vigh for constructive comments on the manuscript and N. Shultz and Dr. Marco Tizzano for technical assistance with molecular techniques.


  1. Acher et al. 1997.
  2. Beckman and Whitehead 1991.
  3. Best et al. 2005.
  4. Bradley and Grabauskas 1998.
  5. Brauner-Osborne et al. 1996.
  6. Browning and Travagli 2007.
  7. Cartmell and Schoepp 2000.
  8. Chen et al. 1999.
  9. Chen et al. 2002.
  10. Conn and Niswender 2006.
  11. Conn and Pin 1997.
  12. Corti et al. 2002.
  13. Dolan and Nolan 2002.
  14. Felder and Heesch 1987.
  15. Finger 2008.
  16. Finger and Dunwiddie 1992.
  17. Frank 1973.
  18. Frank 1991.
  19. Gafka et al. 1999.
  20. Galarreta and Hestrin 1998.
  21. Gasparini et al. 1999.
  22. Gereau and Conn 1995.
  23. Ghosh et al. 1997.
  24. Gingrich and Byrne 1985.
  25. Glaum and Miller 1993.
  26. Grueter and Winder 2005.
  27. Han et al. 2006.
  28. Harrer and Travers 1996.
  29. Hay and Lindsley 1995.
  30. Hirasawa et al. 2002.
  31. Hoang and Hay 2001.
  32. King 2003.
  33. Lacey et al. 2005.
  34. Li and Smith 1997.
  35. Liu et al. 1998.
  36. Maruyama et al. 2006.
  37. Mathieson and Maler 1988.
  38. Mifflin and Felder 1988.
  39. Mitsukawa et al. 2005.
  40. Morita and Finger 1985.
  41. Morita and Finger 1987.
  42. Nakajima et al. 1993.
  43. Neugebauer et al. 2000.
  44. Nistri et al. 2006.
  45. Page et al. 2005.
  46. Patil et al. 1998.
  47. Poli et al. 1999.
  48. Rozen and Skaletsky 2000.
  49. Schild et al. 1995.
  50. Sharp and Finger 2002.
  51. Shigemoto et al. 1997.
  52. Simmons et al. 2002.
  53. Slattery et al. 2006.
  54. Smeraski et al. 1999.
  55. Smith et al. 1998.
  56. Takeuchi 1958.
  57. Thies 1965.
  58. Thomas et al. 2001.
  59. Travagli et al. 2006.
  60. Wang and Bradley 1995.
  61. Zhai et al. 2002.
  62. Zhang and Schmidt 1999.
View Abstract