Pharmacology of ionotropic and metabotropic glutamate receptors on neurons involved in feeding behavior in the pond snail, Helisoma trivolvis

Central pattern generators (CPGs), or rhythmically active neuronal circuits, directly underlie many identifiable behaviors in invertebrates. Rhythmic feeding behavior in the pond snail Helisoma trivolvis is controlled by a CPG that exists entirely within the paired buccal ganglia. Feeding occurs through a tripartite mechanism involving the protraction of a toothed radula through the oral opening, followed by radular retraction, which includes scraping the substrate, and then radular tensing, which releases food particles into the esophagus (Quinlan and Murphy, 1991). These movements are under the control of three distinct sets of motor neurons that are, in turn, thought to be under the control of a buccal CPG. Three groups of neurons in the CPG, referred to as the S1, S2 and S3 subunits, are hypothesized to be independent conditional oscillators, and cause their respective motor neurons to fire bursts of action potentials (APs) (Quinlan et al.,1995). The subunits can operate independently or in concert to drive a variety of feeding behaviors. The normal sequence of activity underlying feeding behavior is thought to be S1–S2–S3, with excitatory connections from S1 causing depolarization in S1 motor neurons,such as neurons B6, B7 and B8, and excitation of the S2 subunit. S2 activity is then thought to excite S2 motor neurons (e.g. B27), while at the same time inhibiting motor neurons associated with the S1 and S3 subunits. Thus activity of the S2 subunit seems to be central to the control of the `standard pattern'of feeding behavior in Helisoma(Quinlan and Murphy, 1991; Quinlan et al., 1995).

Glutamate has been shown to be a major neurotransmitter controlling the feeding behavior of Helisoma. It both excites S2 motor neurons and inhibits S1 and S3 motor neurons of the buccal ganglia, effects that mimic stimulation of the S2 interneuron B2. In addition, B2 neurons are labeled by anti-glutamate antibodies (Quinlan et al.,1995). These studies suggest that B2 neurons are glutamatergic,and the responses of motor neurons to B2 stimulation are likely to be mediated by glutamate receptors. The excitatory response of an S2 motor neuron (B27) in the buccal ganglia is mimicked by application of the ionotropic glutamate receptor (iGluR) agonist kainate (KA), and is inhibited by 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), an antagonist of the mammalian excitatory amino acid receptors (Quinlan and Murphy, 1991). A receptor (Lym-eGluR2) with similar pharmacology has been cloned from the nervous system of Lymnaea stagnalis and shows 44–48% amino acid sequence homology to vertebrate GluR5, 6 and 7 KA receptor subunits(Stuhmer et al., 1996). Lym-eGluR2, when expressed in oocytes, is activated by glutamate and KA, and these currents in turn are blocked by CNQX. The pharmacology of the receptors and their localization to neurons of the buccal ganglia suggest a role for these receptors in feeding. It is hypothesized, therefore, that the excitatory glutamate receptors present in Helisoma B27 buccal neurons(Quinlan and Murphy, 1991) may be similar to these.

The receptors responsible for hyperpolarization by glutamate in S1 and S3 motor neurons in Helisoma, and in invertebrate neurons in general,are much less well understood. These receptors are not sensitive to any of the antagonists associated with mammalian ionotropic glutamate receptors and the inhibition is mimicked by application of quisqualic acid(Quinlan and Murphy, 1991), an analog of glutamate that also activates inhibitory responses in the pulmonate molluscs Helix, Lymnaea, Aplysia and Planorbarius(Bolshakov et al., 1991; Walker, 1976; Katz and Levitan, 1993). Some of the inhibitory actions of glutamate and quisqualic acid are thought to be mediated by the opening of an integral chloride channel(Cleland, 1996) and are often activated by ibotenic acid. However, a glutamate-induced reduction in AP frequency was not blocked by the Cl^– channel antagonist niflumic acid in Helisoma B5 and B19 neurons(Kleckner et al., 2003). Other inhibitory glutamate actions are thought to be due to an increase in K⁺ conductance, most likely through activation of metabotropic glutamate receptors (mGluRs). The mGluR agonist(1S,3R)-1-aminocyclopentane-1,3-dicarboxylic acid (ACPD) was effective in activating the K⁺ conductance in Aplysiabuccal neurons (Katz and Levitan,1993), but was ineffective in altering membrane potential or AP frequency in Helisoma B19 neurons(Quinlan and Murphy, 1991). In Planorbarius, at least two separate receptors are thought to mediate an increase in K⁺ conductance, each with a different agonist pharmacology (Bolshakov et al.,1991). Preliminary studies have shown that hyperpolarizations caused by glutamate in Helisoma B5, a neuron sporadically active during S1, and B19, an S3 motor neuron, are blocked by both pertussis toxin and Ca²⁺ and Ba²⁺, suggesting a role for G_i/oG proteins and K⁺ channels in the response(Rafferty and Kleckner, 2005). Therefore, it is hypothesized that the receptors mediating inhibition in these neurons are similar to vertebrate mGluRs that mediate inhibition, such as group II or III receptors (Conn and Pin,1997; Sharon et al.,1997; Knoflach and Kemp,1998).

The purpose of the experiments described here was to characterize the excitatory and inhibitory glutamate receptors that respond to glutamate in three buccal neurons, B5, B27 and B19, associated with the S1, S2 and S3 phases of the feeding motor pattern, and influenced by release of glutamate from B2 interneurons. Unexpectedly, we found that all three neurons are inhibited by bath application of glutamate and excited by bath application of KA, suggesting that either each neuron has at least one excitatory and one inhibitory type of glutamate receptor, or that synaptic and electrotonic coupling of certain neuron pairs is responsible for these effects. Receptors mediating inhibition have a unique pharmacology compared with vertebrate receptors.

Albino pond snails (Helisoma trivolvis Say 1816) were kept in the laboratory at room temperature and fed trout chow and lettuce. Originally,they were provided by Patsy Dickinson (Bowdoin College, Brunswick, MA,USA).

De-shelled snails were place in 20% Listerine for 10 min before being pinned in a dissection dish filled with Helisoma saline (HS; 51.3 mmol l^–1 NaCl, 1.7 mmol l^–1 KCl, 1.5 mmol l^–1 MgCl₂, 4.1 mmol l^–1CaCl₂, 5 mmol l^–1 Hepes, pH 7.8). The central nervous system of the snail, which includes the central nerve ring and the buccal ganglia, was removed through a longitudinal incision on the dorsal surface of the snail. Each ganglia preparation was pinned loosely to a small Sylgard-covered dish, bathed in 0.5% protease (Sigma, St Louis, MO, USA) in saline for 1.5–2 min, and rinsed three times with saline. Ganglia preparations were then tightly re-pinned caudal side up with 0.1 mm minuten pins. A map of some of the neurons in the buccal ganglia (see Quinlan et al., 1995) allowed for identification of neurons B5, B19 and B27 based on size and location relative to each other and to the processes exiting the buccal ganglia(Fig. 1A).

Intact preparations of the central ganglia were obtained by removing the entire central nerve ring, including the buccal ganglia, as indicated above. The central nerve ring was incubated in 0.2% trypsin in Helisomadefined media (DM; 50% Leibowitz L-15 medium modified by addition of salts to adjust the ionic composition to that of Helisoma saline; Gibco,special order) for 35 min followed by 20 min of incubation in 0.2% trypsin inhibitor in DM. The ganglia were pinned to a dish, caudal side up. Identified neurons were isolated and cultured with a modification of the technique of Hadley et al. (Hadley et al.,1985). Pinned ganglia were placed in high osmolarity DM (DM supplemented with 30 mmol l^–1 glucose). A slit was made in the overlying sheath with an electrolytically sharpened tungsten microknife. Neuron B19 somata were removed from each buccal ganglion with a fire-polished glass pipette attached to a syringe micrometer. Neurons were plated in 2 ml of Helisoma DM onto sterile 35 mm Petri dishes (Falcon no. 1008) in the absence of neurite promoting factors. These conditions allow neurons to adhere to the polylysine-coated coverslip or plastic dish with minimal extension of processes. Neurons were used within 24 h of plating. All data were collected from isolated neurons that could produce APs in response to the injection of depolarizing current.

Membrane potential changes in B5, B19 and B27 were measured with standard intracellular recording techniques. Intracellular glass electrodes filled with 3 mol l^–1 potassium acetate with resistances between 10 and 40 MΩ were used to measure the membrane potential of neurons during perfusion with pharmacological agents. Preparations were perfused with varying concentrations of the following glutamate agonists and antagonists, all obtained from Sigma or Tocris (Ellisville, MO, USA): glutamate,(S)-α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid(AMPA)/KA receptor agonist KA, AMPA/KA receptor antagonist CNQX, non-specific mGluR agonists quisqualate and ACPD, group I mGluR agonist 3,5-dihydrophenylglycine (DHPG), group II mGluR agonist(2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl)glycine(DCG-IV), and group III mGluR agonist l-2-amino-4-phosphonobutyric acid (l-AP4). Between agonist perfusions, the ganglia were perfused with Helisoma saline for several minutes to ensure that the previous drug did not remain on the preparation. Membrane potentials were measure with a Neuroprobe amplifier (A-M Systems, Sequim, WA, USA) and permanent recordings were stored and analyzed with a Datapac 2K2 data acquisition and analysis system (Run Technologies, Mission Viejo, CA, USA).

Some neurons were injected with Lucifer Yellow (LY; Sigma) to verify accurate electrode placement. In these cases, electrode tips were filled with LY (5% in 0.1 mol l^–1 LiCl) and shanks were backfilled with 0.1 mol l^–1 LiCl. Following pharmacological experiments, as described above, neurons were injected with LY by application of –5.0 nA of current to the electrode for 3–5 min. Following approximately 1 h of incubation in Helisoma saline to allow for diffusion of the dye,ganglia were preserved by incubation overnight (12–16 h) in 4%paraformaldehyde in phosphate-buffered saline (PBS; 0.1 mol l^–1, pH 7.3) at 4°C. Ganglia were rinsed in PBS and dehydrated in an ethanol series (50%, 70%, 80%, 90%, 100%, 100%) at room temperature for 10 min each, followed by incubation in methyl salicylate for 30 min at room temperature. Ganglia were mounted between two coverslips in methyl salicylate, and visualized with a Nikon 80i (with differential interference contrast, DIC) or TE2000U (with Hoffman modulation contrast, HMC)microscope (Melville, NY, USA) equipped with epifluorescence and a QImaging Retiga fast monochrome camera (Surry, BC, Canada) and MetaView imaging system(Molecular Devices, Downington, PA, USA), or a Spot Insight 2 megapixel camera and Spot Advanced imaging software (Diagnostic Instruments, Sterling Heights,MI, USA). Separate fluorescence and either HMC or DIC images were combined using layers with a linear dodge, and processed using levels and curves to control contrast and brightness as well as unsharp mask to increase sharpness in Adobe Photoshop CS2. In some cases, the fluorescence image was created in Helicon Focus (Helicon Soft, Kharkov, Ukraine) from a stack of separate images to maximize the depth of field.

Datapac2K2 was used to measure changes in AP frequency in response to drug treatments. Firing rates (Hz) were determined from 30–40 s intervals during the application of each drug or saline and are presented as mean AP frequency ±s.e.m. Student's two-tailed paired t-tests were used to compare control and (one) drug-induced firing rates, and repeated measures one-way ANOVA was performed, as was appropriate, to compare the effects of the application of drug combinations or multiple concentrations of drugs on neuronal firing rate. Neuman–Keuls multiple comparison tests were performed when ANOVA indicated the effect of drug application on firing rate was significant (P<0.05).

The caudal side of the buccal ganglia of Helisoma contains several large neurons known to be part of the feeding motor circuitry, including the readily accessible B5, B19 and B27 neurons(Fig. 1A). B5 is the largest,most prominent neuron on the caudal surface and is active sporadically during phase I of feeding (Quinlan and Murphy,1996). The neurotransmitter that activates B5 is unidentified, but it is inhibited by applied glutamate(Quinlan and Murphy, 1991)(see below). B5 sends its main process out through the ipsilateral esophageal trunk (ET; Fig. 1B)(Murphy, 2001). Neuron B19(Fig. 2A,B) is another giant neuron that is active during the third phase of feeding due to stimulation of the S3 interneuron N3a (Quinlan and Murphy, 1996). Glutamate also inhibits B19, but it is not clear whether the receptors that mediate this effect are the same as those in B5. B27 (Fig. 3A,B) is active during the second phase of feeding and was shown to be stimulated by applied glutamate and KA (Quinlan and Murphy,1991), as well as stimulation of the glutamatergic interneuron B2(Quinlan et al., 1995). The purpose of these experiments was to further characterize the glutamate receptors in these neurons.

To confirm the presence of glutamate receptors on B5, B19 and B27 neurons,three concentrations of glutamate (30, 100 and 300 μmol l^–1) were applied to whole-ganglia preparations. Dose-dependent inhibition in response to glutamate was observed in all three neurons, including B27 (Fig. 1C,D; Fig. 2C,D; Fig. 3C,D).

In B5 neurons, glutamate produced a significant, dose-dependent inhibition of AP firing (Fig. 1C,D;one-way ANOVA, F_3,6=3.403, P<0.05, N=7). In pairwise comparisons, 300 μmol l^–1glutamate significantly reduced the firing rate from 0.76±0.34 Hz to 0.0±0.0 Hz (Neuman–Keuls post-hoc comparison, P<0.05, N=7), while the reductions in firing rate in response to 30 and 100 μmol l^–1 glutamate to 0.39±0.19 and 0.09±0.06 Hz (N=7) were not significant. These results confirm that inhibitory glutamate receptors are present on B5 neurons.

B5 neurons had a variable response to KA, with a reduction in AP frequency at 10 μmol l^–1 (Student's two-tailed t-test, P<0.05, N=9; Fig. 1E,F), and a trend toward increased firing rate at 30 μmol l^–1 (N=9). One-way ANOVA indicated a significant effect on firing rate of 10 μmol l^–1 KA, 50 μmol l^–1 CNQX and KA plus CNQX (F_3,8=3.801, P<0.05, N=9; Fig. 1F).

Neuron B19 is a motor neuron active in phase III of feeding, and its processes extend across the buccal commissure and out through both ventral buccal nerves (VBNs) and the ipsilateral lateral buccal nerve (LBN; Fig. 2A,B)(Murphy, 2001). A significant,dose-dependent inhibition of AP frequency was observed in B19 neurons following bath application of glutamate (one-way ANOVA, F_3,8=10.03, P=0.0002, N=9), confirming the presence of inhibitory glutamate receptors on these neurons(Fig. 2C,D). Glutamate at 100 and 300 μmol l^–1 caused significant reductions in firing rate from the control rate of 2.88±0.68 Hz to 0.43±0.32 Hz(P<0.05) and 0.31±0.29 Hz (P<0.05,Neuman–Keuls post-hoc test), respectively. Application of 30μmol l^–1 glutamate reduced firing rate to 2.00±0.77 Hz (N=9).

KA has been shown to stimulate B19 neurons, suggesting the presence of a second type of glutamate receptor in these cells(Quinlan and Murphy, 1991). To further characterize this effect, 10 and 30 μmol l^–1 KA were applied while recording from B19 neurons. KA (10 μmol l^–1) did not significantly increase the firing rate (control,1.54±0.50 Hz; 10 μmol l^–1 KA, 2.24±0.47 Hz;paired t-test, P=0.068, N=9), while 30 μmol l^–1 KA significantly increased the firing rate from 1.55±0.44 Hz to 2.59±0.46 Hz (paired t-test, P<0.03, N=13). CNQX (50 μmol l^–1) had no effect on firing rate in the presence or absence of 10 μmol l^–1 KA (Fig. 2F; one-way ANOVA, F_3,8=1.666, P=0.20, N=9). Lower concentrations of CNQX, or 50 μmol l^–1 CNQX applied with higher concentrations of KA also had no effect on the KA-induced firing rate in any cell type tested (data not shown).

Neuron B27 is a motor neuron that extends processes out through both LBNs(Fig. 3A,B)(Murphy, 2001). It is active during phase II of feeding and is known to be excited by glutamatergic interneuron B2 (Quinlan et al.,1995). Contrary to expectations, therefore, a significant,dose-dependent inhibitory effect of glutamate was observed in B27 neurons(F_3,7=9.451, P=0.0004, N=8). Similar to the effects observed in B19, increasing glutamate concentration caused a greater decrease in firing rate (Fig. 3C,D), with both 100 and 300 μmol l^–1glutamate producing significant reductions. In B27, the firing rate decreased from 3.94±0.51 Hz to 1.71±1.48 Hz (P<0.01, N=8) in response to 100 μmol l^–1 glutamate, to 1.22±1.667 Hz (P<0.01, N=8) in response to 300μmol l^–1 glutamate and to 3.61 Hz (P>0.05, N=8) in response to 30 μmol l^–1 glutamate(Neuman–Keuls post-hoc comparisons). These results suggest the presence of inhibitory glutamate receptors on B27, a neuron that is thought to be excited by synaptic release of glutamate in vivo.

Quinlan and Murphy (Quinlan and Murphy,1991) found that bath application of 1 mmol l^–1glutamate inhibited B19 and excited B27, contrary to this study's findings that 100 and 300 μmol l^–1 glutamate inhibited both neurons. This contradiction indicated a possible differential effect of glutamate depending on its concentration, so the effects of a higher glutamate concentration on B27 neurons were examined. The results were highly variable. Glutamate at 3 mmol l^–1 increased the firing rate of three neurons, depolarized two others that were not firing, and hyperpolarized and reduced the firing rate of three others. No significant effect on firing rate was observed when the six firing neurons were combined (data not shown). This variability in responsiveness indicates that additional unidentified factors may modulate the effect of glutamate on B27, and that there may be individual variability in the relative sizes of inhibitory and excitatory glutamate receptor populations within a given type of neuron. However, excitation in B27 was never observed in response to lower concentrations of glutamate (30, 100 and 300 μmol l^–1), indicating that the direction of effect of glutamate on this neuron may vary with concentration.

As expected, KA increased the firing rate of B27 neurons(Fig. 3E). KA at 10 μmol l^–1 did not have a significant effect on firing rate (paired t-test; control, 1.48±0.66 Hz; 10 μmol l^–1KA, 4.31±1.22 Hz; P=0.058, N=9), whereas 30 μmol l^–1 KA produced a significant increase from 3.47±0.69 Hz to 6.86±0.81 Hz (paired t-test, P<0.0002, N=10). There was a significant effect of 10 μmol l^–1 KA, 50 μmol l^–1 CNQX and 10 μmol l^–1 KA/50 μmol l^–1 CNQX on firing rate in B27 neurons (one-way ANOVA, F_3,8=4.60, P=0.0112, N=9). In pairwise comparisons, 10 μmol l^–1 KA significantly increased AP frequency compared with both control and CNQX(Neuman–Keuls post-hoc comparison, P<0.05; data the same as Fig. 3E). However, CNQX in the presence of KA did not significantly reduce AP frequency compared with KA-induced excitation (4.31±1.22 Hz to 2.21±0.98 Hz, P>0.05, N=9; Fig. 3G). Pairwise comparison also indicated no significant difference between control and KA/CNQX firing rates, suggesting that KA was ineffective at increasing AP frequency in the presence of CNQX.

Because of the unexpected inhibitory effects of glutamate on B27 neurons and the excitatory effects of KA on B19 neurons, the electrotonic coupling between neurons B27 and B19 was examined. During simultaneous intracellular recordings of B27 and B19 neurons, hyperpolarizing current injected into neuron B19 produced a hyperpolarization in the membrane potential of B27(Fig. 4A). This demonstrates electrotonic coupling between these neurons and suggests that the apparent hyperpolarization and reduction of AP frequency of B27 neurons by glutamate,and the depolarization and increase in AP frequency of B19 by KA might be indirect.

To further test this hypothesis, the preliminary effects of the application of glutamate and KA were tested on individually isolated B19 neurons in cell culture. Only neurons that were removed from the buccal ganglia with partial neurites intact were used for experimentation. This criterion increases the probability that peripheral receptors would be retained in cultured neurons. Individually isolated identified Helisoma neurons retain electrogenic capabilities, and can release transmitter and respond to chemical receptor stimulation in culture (Haydon and Man-Son-Hing, 1988; Syed et al., 1993). All data were collected from isolated neurons that could produce APs in response to injection of depolarizing current.

KA and glutamate had differential effects on isolated B19 neurons(Fig. 4B). Bath application of 10 μmol l^–1 KA had no apparent effect on the membrane potential of an individually isolated B19 neuron. This same isolated neuron produced a robust hyperpolarization in response to application of 1 mmol l^–1 glutamate (N=3). The hyperpolarizing effect of glutamate application on B19 neurons was completely reversible. These data suggest that the effect of KA on B19 neurons in whole-ganglion preparations is the result of electrotonic coupling with B27 neurons.

The inhibition of firing by glutamate application in all three neurons suggests the presence of either glutamate-gated chloride channels(GluCl^–) or mGluRs coupled to potassium channels. Because our lab found no evidence for the role of chloride in glutamate-mediated inhibition (Kleckner et al.,2003), the presence of mGluRs that might couple to potassium channels was investigated with agonists for vertebrate mGluRs.

Quisqualate and ACPD, non-selective agonists of most mGluRs(Conn and Pin, 1997), were applied to investigate the presence of glutamate receptors resembling vertebrate mGluRs on B5, B19 and B27 neurons. Additionally, the more selective vertebrate mGluR agonists DHPG, DCG-IV and l-AP4 were applied to more specifically classify inhibitory glutamate receptors as resembling vertebrate group I, II and III mGluRs, respectively, in B5, B19 and B27 neurons.

Quisqualate and ACPD reduced AP frequency significantly in B5 neurons. Quisqualate (10 μmol l^–1) reduced AP frequency in B5 neurons from 0.56±0.13 Hz to 0.008±0.007 Hz (paired t-test, P=0.0007, N=15; Fig. 5B). Likewise, 100 μmol l^–1 ACPD significantly reduced AP frequency, from 1.03±0.31 Hz to 0.33±0.25 Hz (paired t-test, P=0.0061, N=13). None of the selective mGluR agonists, DHPG(100 μmol l^–1), DCG-IV (30 μmol l^–1)or l-AP4 (100 μmol l^–1) had significant inhibitory effects on AP firing in B5. These results suggest that the agonist binding sites on mGluRs in Helisoma buccal neurons, while capable of binding to glutamate and non-selective agonists, do not closely resemble the binding conformations on vertebrate group I, II and III mGluRs.

In response to 10 μmol l^–1 quisqualate, a reduction in AP frequency occurred in B19 neurons, verifying that inhibitory mGluRs are present on these buccal neurons. Quisqualate caused AP frequency to decrease from 1.179±0.16 Hz to 0.11±0.10 Hz (paired t-test, P<0.0001, N=20; Fig. 5C). Neither ACPD nor any of the selective mGluR agonists, DHPG,DCG-IV and l-AP4, had significant effects on the AP firing rate in B19 neurons (Fig. 5C). These results indicate that inhibitory receptors responsive to quisqualate are present on B19 neurons, but they are insensitive to ACPD, and do not resemble vertebrate group I, II or III mGluRs with respect to agonist binding.

Quisqualate (10 μmol l^–1) produced a significant inhibitory effect on AP frequency in B27 neurons(Fig. 5D). AP frequency decreased from 3.21±0.81 Hz to 1.3±0.74 Hz (paired t-test, P=0.046, N=9). ACPD (100 μmol l^–1), DHPG (100 μmol l^–1), DCG-IV (30μmol l^–1) and l-AP4 (100 μmol l^–1) did not affect AP frequency in B27 neurons(Fig. 5D). These results suggest the presence of an mGluR on B27 neurons, but it is not sensitive to ACPD or the selective agonists for vertebrate mGluR subtypes used in this study.

The results of this study suggest that each of the neurons tested, B5, B19 and B27, may contain at least two types of glutamate receptor, one that causes depolarization and increases firing rate, and another that causes hyperpolarization and a reduction in firing rate. Electrotonic coupling between neurons B27 and B19, however, suggests that some of the effects of glutamate agonists may be mediated indirectly, by the activity of receptors on a coupled neuron.

A surprising finding that emerged from the data was that glutamate, at concentrations between 100 and 300 μmol l^–1, reduced AP frequency in B27 neurons. These neurons have previously been found to be activated by stimulation of the phase II glutamatergic interneuron B2, and by perfusion of 1 mmol l^–1 glutamate(Quinlan et al., 1995), so glutamate was expected to stimulate the cells. Results with a higher concentration, 3 mmol l^–1, of glutamate were highly variable. These data suggest that different B27 neurons might have different relative numbers of excitatory and inhibitory glutamate receptors, leading to an inconsistent response to high concentrations of perfused glutamate. Nevertheless, if we assume the presence of a significant number of inhibitory glutamate receptors on B27 neurons, it is interesting to speculate on why stimulation of interneuron B2 consistently causes depolarization and increased firing rate (Quinlan and Murphy,1991). The most likely interpretation is that ionotropic glutamate receptors, which cause depolarization, are located in synapses, and metabotropic receptors, which are hypothesized to cause hyperpolarization, are located extrasynaptically. Thus, ionotropic glutamate receptors are exposed to high concentrations of glutamate in the synapse, and metabotropic receptors to lower concentrations. This organization has been observed in many vertebrate systems (Conn and Pin, 1997). High frequency activation of presynaptic neurons is thought to be required to cause the spill-over of glutamate and subsequent activation of extra-synaptic glutamate receptors. Perhaps this organization of receptors on buccal motor neurons in the feeding CPG of Helisoma would provide modulation during times of high activity.

An alternative explanation for the inhibitory effects of glutamate on B27 neurons is provided by evidence for electrotonic coupling between B19 and B27 neurons. Glutamate actions on mGluRs located on B19 neurons could hyperpolarize B27 and counteract the effects of glutamate on excitatory receptors present on B27. The similar effects of metabotropic agonists on B19 and B27 neurons (see below) supports this explanation. A full explanation of the inhibitory effects of glutamate on B27 neurons awaits the results of experiments with isolated B27 neurons, currently in progress.

Inhibitory effects of glutamate in invertebrates have been shown to be mediated by both glutamate-gated chloride channels (GluCl^–)and mGluRs, sometimes in the same cells. GluCl^– are found in many invertebrate systems (Bolshakov et al., 1991; Cleland and Selverston, 1995; Cully et al.,1994; Katz and Levitan, 1983), and chloride has been implicated along with potassium as the mediators of inhibitory currents in Helisoma (Jones et al.,1987). Previous evidence from our laboratory indicates that the chloride channel blocker niflumic acid is unable to prevent the hyperpolarization and reduced AP frequency observed when glutamate is applied to B5 and B19 neurons (Kleckner et al.,2003). In addition, Ca²⁺ and Ba²⁺ have been shown to block glutamate-induced hyperpolarization of isolated B5 and B19 neurons (Rafferty and Kleckner,2005), suggesting the effect is mediated through K⁺channels. Vertebrate inhibitory mGluRs, which include the group II and III mGluR subtypes, are often coupled with G protein-coupled inwardly rectifying K⁺ channels (GIRKs). Following the binding of glutamate to group II or III mGluRs, a G protein belonging to the G_i/o class may activate a GIRK, in addition to inhibition of adenylyl cyclase, resulting in the efflux of K⁺ to produce hyperpolarization(Knoflach and Kemp, 1998). Supporting the role of this pathway in glutamate-induced inhibition in Helisoma, pertussis toxin was shown to block hyperpolarization of B5 and B19 neurons by glutamate (Rafferty and Kleckner, 2005). Similar receptors, responsive to quisqualate,ACPD and l-AP4, have also been found in Drosophila nervous system (Parmentier et al.,1996).

Evidence supporting the presence of mGluRs in Helisoma buccal neurons includes the reduction of AP frequency by quisqualate, an agonist that interacts with all three groups of mGluR, in all three neurons tested. However, the mGluR agonist ACPD, also relatively non-selective, reduced AP frequency only in B5 neurons, indicating that at least two types of mGluR exist on buccal neurons. A lack of sensitivity of B19 and B27 neurons to ACPD despite activation by quisqualate suggests that group III mGluRs might be involved, since these receptors have a lower affinity for ACPD (>300μmol l^–1) than receptors in groups I or II(Conn and Pin, 1997). However,neither B19 nor B27 neurons were inhibited by the selective group III agonist l-AP4, suggesting that these receptors have a unique agonist pharmacology. This pharmacology also distinguishes the receptors in B19 and B27 from those cloned from Drosophila (DmGluRA)(Parmentier et al., 1996),which did respond to l-AP4. Invertebrate glutamate receptors with unique pharmacology compared with vertebrate receptors have been found in networks of the Pacific spiny lobster (Panulirus interruptus)stomatogastric ganglia, which are largely regulated by glutamate. These neurons did not respond to quisqualate, KA, AMPA or NMDA(Cleland and Selverston,1995).

Interestingly, a small but not significant trend toward excitation in response to DHPG occurred in B27, suggesting the possible presence of receptors resembling group I mGluRs on this neuron. In vertebrates, group I mGluRs are excitatory, causing slow depolarization through the activation of phospholipase C. G_q proteins activate phospholipase C, which then activates diacylglycerol and inositol triphosphate as second messengers to open ion channels. G_q proteins may also open ion channels directly without acting through second messengers(Meldrum, 2000; Sharon et al., 1997; Slattery et al., 2006). Perhaps robust excitation of B27 within the feeding CPG occurs as a result of both excitatory iGluRs resembling KA receptors and excitatory mGluRs resembling vertebrate group I mGluRs. However, the possible presence of group I mGluR-like excitatory receptors does not account for the inhibitory effect of quisqualate on B27, which suggests that a population of inhibitory mGluRs with unique pharmacology is also present.

Thus, it is difficult to classify the inhibitory receptor populations in these buccal neurons based on any known receptor pharmacology. Future classification of receptors in buccal neurons should be done in isolated neurons to prevent issues of synaptic and electrotonic coupling, and should include the motor neurons B6, B7 and B8 since these neurons are driven more directly by S1 interneurons (Quinlan and Murphy, 1991; Quinlan and Murphy, 1996) and are therefore more likely to fire consistently.

Activation of vertebrate KA receptors, most often of the GluR5 or GluR6 type, has been shown to activate inhibitory metabotropic glutamate receptors in a variety of vertebrate systems(Grabauskas et al., 2007; Partovi and Frerking, 2006; Schmidt et al., 1999), so,given the inhibition of B5 neurons by 10 μmol l^–1 KA in this study, the possibility exists that Helisoma has similar KA receptors and mGluRs to cause this effect. GluR5/6-containing KA receptors have an affinity for KA in the nmol l^–1 range(Egeberg et al., 1991),consistent with activation by a relatively low concentration of KA. At higher concentrations, KA acting at AMPA receptors, which have higher μmol l^–1 affinities (30–100 μmol l^–1)(Hollmann et al., 1989), might be responsible for depolarization and the increased frequency of firing. Immunoreactivity to a vertebrate GluR5/6/7 antibody has been observed in both Helisoma buccal ganglia (Marya et al., 2007) and leech CNS(Thorogood et al., 1999). However, B5 was not among the neurons labeled in Helisoma. Similar to the experiments with metabotropic agonists, B5 neurons had highly variable firing rates, even under control conditions, so solid conclusions about the nature of the receptors present in these neurons might need to await the cloning, expression and localization of the receptors or characterization in other neurons inhibited by glutamate (e.g. B6, 7, 8).

KA-mediated reduction in firing rate might also have been due to the actions of KA at KA/AMPA receptors located on other neurons, which make synapses onto B5 neurons to cause inhibition. This effect would most likely be due to the presence of GluR5 or 6 receptors with relatively high affinities. Higher concentrations of KA might then act directly on AMPA receptors on B5 neurons to depolarize them. Experiments with isolated neurons or with blockers of synaptic transmission are needed to determine the precise mechanism.

Unlike the results in B5 neurons, 30 μmol l^–1 KA consistently stimulated B19 and B27 neurons, suggesting the presence of ionotropic receptors resembling vertebrate AMPA or KA receptors. Given that KA was not able to depolarize isolated B19 neurons, it is likely that the effects of KA on both neurons are mediated through receptors on B27 neurons. However,the vertebrate AMPA/KA receptor antagonist CNQX partially blocked the KA-induced increase in firing rate in B27 neurons, but not in B19 neurons. Importantly, only a concentration of CNQX that is five to ten times the effective concentration against most vertebrate AMPA/KA receptors(Honore et al., 1989) caused inhibition of KA stimulation, and even then only at low KA concentrations. The ability of 30 μmol l^–1 KA to stimulate AP firing suggests similarity to vertebrate AMPA receptors, which have EC₅₀ values for KA of 30 to 110 μmol l^–1 for cloned receptors from rat;these values contrast with those of vertebrate KA receptors, which have nmol l^–1 to low μmol l^–1 affinities for KA(reviewed in Hollmann and Heinemann,1994). Thus, it appears that glutamate and KA binding affinities may have changed very little over evolutionary time, whereas antagonist binding affinities have changed. Given the lack of selective pressure to maintain the additional sites required for binding of a modern synthetic antagonist like CNQX, this finding is not altogether surprising.

Several types of glutamate receptor are likely to exist on buccal neurons of the pond snail Helisoma trivolvis. Excitatory glutamate receptors in buccal neuron B27 resemble vertebrate glutamate receptors, perhaps of the AMPA receptor subtype, in their excitatory responses to KA. B19 neurons likely respond to KA through electrotonic coupling with B27 neurons. Variable responsiveness to KA in B5 neurons suggests either the presence of more than one subtype of excitatory receptor, perhaps including KA receptors that couple to mGluRs, or synaptic coupling between neurons with a relatively high affinity KA receptor and B5 neurons with a lower affinity AMPA receptor.

Inhibition by glutamate is most likely to occur through mGluRs whose signal transduction pathways most closely resemble group II or III vertebrate receptors. Multiple subtypes with differential sensitivity to ACPD are likely to be present in different neurons, and the pharmacology of receptors in each type of neuron does not match the known receptor pharmacology of vertebrate receptors (Conn and Pin, 1997)or receptors from Drosophila nervous system(Parmentier et al., 1996). Because of the lack of effective antagonists for each of these receptors, more detailed pharmacological characterization may await cloning and expression of receptors, and localization to buccal neurons through in situhybridization. Experiments are also underway in which glutamate receptor agonists are applied to isolated identified buccal neurons.

This project was supported by two Howard Hughes Medical Institute grants(E.S. and C.A.D.) and NIH grant number P20 RR-016463 from the INBRE Program of the National Center for Research Resources (N.W.K.). Thanks to Will Ash from the Bates College Imaging and Computing Center for assistance with the Lucifer Yellow images and to Anne Kate Perry for her willingness to initiate this project.