In certain Aplysia neurons, glutamate, GABA, and acetylcholine (ACh) all elicit desensitizing Cl-dependent responses. This fact and the finding that the glutamate and GABA responses “cross-desensitize” led to the suggestion (Swann and Carpenter, 1975; King and Carpenter, 1987) that the responses to these transmitters were mediated by the same receptor–channel complex.
This hypothesis is incompatible with the demonstration given here that the GABA- and glutamate-gated channels are clearly distinct; the GABA channel, but not the glutamate channel, shows outward rectification (Matsumoto, 1982; King and Carpenter, 1987, 1989) and is selectively blocked by intracellular sulfate. Exploiting these distinctive characteristics and the independent expression of the receptors in some cells, we have been able to reevaluate the so-called cross-desensitization by analyzing the ability of GABA, glutamate, and other agonists to interact with each of the receptor molecules.
The cross-desensitization was found to be exclusively attributable to the ability of GABA to interact with the glutamate receptor (Oyama et al., 1990). The GABA receptor is unaffected by glutamate. Nevertheless, in cells expressing both receptors, glutamate can reduce the GABA response by auto-desensitizing the part of the response that is mediated by the glutamate receptor. No interactions were observed between ACh-induced responses and either of the responses elicited by the amino acids.
The invertebrate glutamate-gated Cl channels that have been cloned resemble the vertebrate glycine receptor (Vassilatis et al., 1997). Our pharmacological evaluation of the molluscan glutamate receptor points in the same direction.
- chloride channel
- pharmacological characteristics
A number of studies on vertebrate neurons have described interactions between inhibitory Cl responses elicited by GABA, glycine, and β-alanine, e.g., between responses to GABA and glycine (Barker and McBurney, 1979; Hamill et al., 1983; Baev et al., 1992; Lewis and Faber, 1993; Trombley et al., 1999), between responses to β-alanine and either GABA or glycine (Choquet and Korn, 1988), and between responses to glycine and β-alanine (Krishtal et al., 1988). The demonstration that recombinant homomeric glycine receptors (formed from subunits cloned from zebrafish) can be activated by GABA (Fucile et al., 1999) might well explain certain of the interactions observed in vertebrate neurons between GABA- and glycine-induced responses.
In invertebrates, many different neurotransmitters, including glutamate but not glycine, gate Cl channels. In cells in which different transmitters activate Cl conductances, interactions observed between the responses have led to the suggestion that different transmitters share the same receptor molecule. For example, in crayfish muscle, a single Cl channel is activated by similar concentrations of glutamate, GABA, and acetylcholine (ACh) (Franke et al., 1986; Zufall et al., 1988). In certain identified Aplysia neurons, these same three transmitters often increase Cl conductance in the same cell, but studies evaluating interactions between these responses have yielded conflicting conclusions. King and Carpenter (1987) observed bi-directional “cross-desensitization” between the GABA and glutamate responses and suggested that, as in crayfish muscle, these two amino acids share the same receptor molecule. However, Ikemoto and Akaike (1988) failed to observe any interactions, and Oyama et al. (1990) observed that, whereas the glutamate response was diminished in the presence of GABA, only a distinctive slowly desensitizing, high-threshold element of the GABA response was diminished by glutamate. They proposed that this element of the GABA response was attributable to GABA activation of the glutamate receptor. In contrast, similar studies on other invertebrate ganglionic neurons revealed no interactions between GABA and glutamate Cl-dependent responses (lobster, Cleland and Selverston, 1998; crab, Duan and Cooke, 2000).
In the present study, we have reexamined the interactions between the glutamate-, GABA-, and ACh-activated responses in Aplysianeurons. By exploiting our finding that the GABA-gated channel has distinguishing characteristics and the fact that, in certain cells, only one of the relevant receptors is expressed, we have shown that all three of these transmitters open Cl channels by binding to distinct and independent molecular entities. Nevertheless, interactions were shown to exist between the GABA and glutamate responses. These seemingly bi-directional interactions can be entirely explained by the ability of GABA to interact with the glutamate receptor (Oyama et al., 1990).
MATERIALS AND METHODS
Experimental preparation. The experiments described in this paper were performed on cells from either the buccal, pleural, or cerebral ganglia of Aplysia.
The ganglia were prepared as described previously (Kehoe, 1985). For the study of the coexpressed GABA, glutamate, and ACh receptors mediating Cl-dependent responses, two series of cells were used: the medial cells of the pleural ganglia (Kehoe, 1972b) and selected cells from the buccal ganglia (including B3 and B6) (Gardner and Kandel, 1977). For examining glutamate-induced Cl-dependent responses in cells having no GABA receptor, we used small unpigmented cells from the pleural ganglia (Ascher et al., 1978), whereas, for the evaluation of GABA-induced Cl-dependent responses in cells having no glutamate-induced Cl response, we chose small, unpigmented cells located on the base of the midline of the dorsal surface of the cerebral ganglion.
Electrodes, voltage clamp, recording procedures, and treatment of data. For experiments illustrated in Figures2 A–C, 3 B, 4 B,6 C, and 10 A, whole-cell patch-clamp methods (Hamill et al., 1981) were used (for details, see Kehoe, 1994). Pipettes of 200–500 kΩ ensured a good exchange, and data were retained only from cells for which the access resistance remained constant throughout the experiment. In all other experiments, recordings were made in conventional microelectrode voltage clamp using hand-pulled microelectrodes made with a de Fonbrune microforge and filled with 0.5 mK2SO4. Although cells with intact axons were used in these experiments, and obviously the voltage in much of the axon was not controlled by the somatically placed electrodes, clamp error cannot explain the changes in response amplitude studied in these experiments. No evidence of a failure to control somatic voltage was seen in the voltage traces, and the control of the somatic voltage was evident from the unchanging amplitude of one of the two agonist-induced responses in experiments in which the other agonist response was markedly reduced. Activation of axonal membrane was minimized by the somatic application of the agonist and by its rapid removal by the flow from the control tube, as well as by a continual superfusion of the entire ganglion. For an additional description of the characteristics of the two-electrode voltage clamp, see Kehoe (1985).
Continuous recordings were made on a Servogor 340 paper recorder and on a digital audio tape recorder. Records of currents elicited by agonist applications were digitized and sampled on-line on a Dell (Round Rock, TX) computer, via a Cambridge Electronic Design (Cambridge, UK) 1401 interface, using the Whole Cell Eectrophysiology program from Strathclyde Electrophysiological Software.
External and internal solutions. The artificial seawater (ASW) bathing the ganglion contained (in mm): 480 NaCl, 10 KCl, 10 CaCl2, 50 MgCl2, and 10 Na-HEPES, pH 7.8. The internal solution used in the whole-cell clamp experiments illustrated here contained (in mm): 411 K2SO4, 8.3 Na2SO4, 1 CaCl2, 2 MgCl2, and 1.5 EGTA, buffered with 20 K-HEPES to pH 7.4. In a few nonillustrated experiments, a KCl-based solution was used to show that the relevant factor causing the diminution of the GABA response in whole-cell perfusion experiments came from the presence of sulfate ions in the internal solution. The KCl-based internal solution contained (in mm): 496 KCl, 10 NaCl, 2 MgCl2, 1 CaCl2, 1.5 EGTA, and 20 K-HEPES, pH 7.4.
Fast application system. The fast application system used here has been described previously in detail (Johnson and Ascher, 1987;Kehoe, 1994). Three glass tubes held together with heat-shrink tubing were moved laterally so that the appropriate (ASW or agonist-containing) tube was directly in front of the cell being studied. The central tube, called the “control tube,” contained ASW that continually bathed the cell under study except when the tubes were moved laterally to permit a 2 sec application of one of the agonists. Solenoid-driven valves were placed in the line of the tubing feeding each of the agonist-containing glass tubes, permitting the flow of agonist only when the relevant tube was directly facing the cell. Although the concentration jump occurring at the tip of the tubes is very rapid, the relevant measurement is of course the speed at which the solutions flowing from the tube envelop the cell. An estimation of this exchange was made by switching from normal ASW in the control tube (containing 10 mm K) to high-K seawater (containing 154 mm K). The time to reach 67% of the maximal current induced by the high-K seawater was found to be on the order of 70–80 msec.
The duration of agonist application in all experiments was 2 sec, and the interval between applications was 1 min. Because in most experiments there was an alternation between two different agonists (e.g., GABA and glutamate), for a given agonist, a 2 min interval separated two successive applications. With the agonist concentrations used here, this regimen yielded control responses of a constant amplitude. The antagonists were applied only through the control tube (see exception in Fig. 9 A). This implies that, during the 2 sec application of the agonist, a partial wash of the preapplied antagonist occurs. The application of desensitizing agonists through the control tube resulted in a similar partial wash during the 2 sec agonist application. This led, at some concentrations, to particularities in the recorded currents that are discussed in the relevant text describing the figures.
Drugs. GABA, l-glutamic acid (monosodium salt), ACh, β-alanine, tubocurarine chloride (TC), strychnine hydrochloride, and pentobarbital were all obtained from Sigma (St. Louis, MO); methyllycaconitine citrate (MLA) was obtained from Research Biochemicals (Natick, MA); andcis-4-aminocrotonic acid (CACA),trans-4-aminocrotonic acid (TACA), hypotaurine, and bicuculline methochloride were obtained from Tocris Cookson (Bristol, UK).
Experimental n values, statistical analyses, and experimental controls. Each conclusion that has been illustrated in this paper was drawn from a minimum of three experiments (with the exception of that illustrated in Fig. 2 B, for which only two experiments were performed). Statistical analyses using either the ANOVA test (Figs. 1 C, 8) or the Student'st test (other figures) quantified the effects of the various manipulations on the responses to the different agonists.
Comparison of glutamate- and GABA-gated Cl responses
Glutamate, GABA, and ACh each elicits a rapidly activated inhibitory response in a number of neurons in Aplysiacentral ganglia. In a given experiment and at a given time, the reversal potential of this response is the same for all three of the transmitters and has been shown, for the response to ACh, to be equal to the Cl equilibrium potential (Kehoe, 1972a). Such cells offer an easily controlled comparison of these receptor-operated channels and provide a good preparation for the evaluation of any possible interactions between them. In this paper, only the glutamate and GABA responses will be described in detail, but some data will be presented concerning the ACh-induced Cl-dependent responses (Kehoe and McIntosh, 1998) that coexist with the glutamate and GABA Cl responses in the medial cells of the pleural ganglia, as well as in selected cells of the buccal ganglia (e.g., B3 and B6) (Gardner and Kandel, 1977).
A given molluscan neuron often expresses more than one receptor type for a given transmitter. That is the case for glutamate in the medial cells studied here, because, in addition to gating a Cl channel in those cells, glutamate also activates a nonspecific cationic channel, thereby eliciting a two-component response (King and Carpenter, 1987). Because the cationic response desensitizes very rapidly, it does not interfere with the peak of the glutamate-activated increase in Cl conductance and has consequently not been pharmacologically eliminated for these experiments.
Differential voltage dependence of the glutamate- and GABA-gated Cl conductances
Examples of the responses to glutamate and GABA in the two different cell types are shown in Figure1. In the medial cells (Fig.1 A), it can be seen that the Cl current elicited by glutamate (outward at −26 mV, inward at −66 mV, with the reversal potential in this cell at −46 mV) is preceded by a very rapidly desensitizing cationic inward current. The only response elicited by GABA in the same cells is a Cl-dependent one, hence no change in current is seen at ECl (−46 mV). Figure1 B illustrates records from one of the selected buccal cells in which both glutamate and GABA activate only Cl-dependent responses. (Given the rapidity of the desensitization of the glutamate cationic response, such a response might be missed in these large cells, which are often a few hundred micrometers in diameter. For cells of this size, the fast application tubes must be pulled back further from the cell surface, and the speed of the concentration jump at the level of the membrane is thereby reduced.)
An examination of the outward- and inward-going responses to the two amino acids reveals that, when the holding potential is fixed on either side of (but equidistant from) ECl, the inward and outward currents elicited by glutamate are of equal amplitude, whereas the inward current elicited by GABA is always smaller than the outward current. This was shown to be true in both cell types. In Figure 1, A and B, the response recorded at a holding potential more negative than ECl has been multiplied by −1 and is represented by a dashed line to permit an easy comparison with the amplitude of the outward current. The vertical dashed line crossing the recordings indicates the end of the agonist application and the return to ASW flowing from the control tube.
A statistical evaluation of the differential voltage dependence of the GABA- and glutamate-induced conductances is presented in Figure1 C, and the results obtained from an ANOVA of these data are presented in the legend to Figure 1. The outward rectification of the GABA-induced conductance has been described previously (Matsumoto, 1982; King and Carpenter, 1987, 1989).
Differential effect of intracellular sulfate ions on the glutamate and GABA responses
The differential voltage dependence of the Cl conductances activated by glutamate and GABA suggests that the two transmitters gate different types of Cl channels.
This impression was reinforced by the differential effect of intracellular sulfate ions observed here on the responses to GABA and glutamate. The experiments illustrated in Figure2 A–C were performed in whole-cell patch-clamp mode, so the solution in the pipette was constantly exchanging with the cytoplasm, hence altering the internal Cl concentration and thereby provoking changes in ECl. Consequently, in Figure 2, A andB, periodic adjustments throughout the experiments were made to ensure that the holding potential remained at a constant distance from ECl. Under these conditions, the amplitude of the glutamate response remained constant (mean of six experiments was 99.1% of control; ς = 2.13%), whereas that to GABA decreased progressively to an average of 37.7% of the control value (ς = 8.33%) after ∼20 min whole-cell perfusion (n = 6; p ≤ 0.0001). Figure2 B shows that the blocking effect of sulfate on the GABA response is completely independent of both membrane voltage and the direction of current flow. In this experiment made at ±15 mV on either side of ECl, the inward/outward current ratio was maintained at 68%, despite a sulfate-induced reduction in the GABA response amplitude to 36% of the initial value. In the second such experiment (data not shown) performed at ±10 mV on either side of ECl, the inward/outward current ratio was maintained at 78% throughout the sulfate-induced diminution of the absolute response amplitude, which reached 57% of the initial amplitude after 20 min whole-cell recording.
Figure 2 C illustrates another protocol for studying the differential effect of internal sulfate ions on the GABA and glutamate responses. In this experiment, the holding potential was held constant while ECl became more negative because of the whole-cell perfusion. Because of the resulting increase in driving force for Cl ions, the response to glutamate increased, whereas, despite the change in driving force, the response to GABA decreased. This simplified protocol, which of course leads to the same conclusion, was used in experiments to be described later evaluating differential effects of intracellular sulfate on responses elicited by various agonists.
It is known that GABA responses in some mammalian neurons show rundown when intracellular ATP is washed out by whole-cell perfusion (Shirasaki et al., 1992). This however was not the explanation for the observation made here, because the results were unchanged by the inclusion of ATP and GTP in the K2SO4-based internal solution. Furthermore, when KCl-based, ATP-free solutions were used (three experiments), GABA and glutamate responses changed proportionately as intracellular perfusion progressed (data not shown). Finally, the same differential blockade of the GABA response by sulfate ions was seen (Fig. 2 D) when 0.5mK2SO4 was simply injected by pressure into a cell in which recordings were made with conventional two-electrode voltage clamp, thereby avoiding any possibility of washout associated with whole-cell perfusion (see figure legend for a statistical résumé of the relevant experiments).
Pharmacological distinctions between the glutamate and GABA responses
One of the factors that has held back the identification of glutamatergic and GABAergic synapses in molluscan nervous systems is the paucity of pharmacological tools that have been found to be selective for the different Cl-dependent responses activated by these different transmitters. In an effort to distinguish further between the glutamate- and GABA-induced Cl responses and to provide compounds that could eventually assist in identifying synapses using these neurotransmitters, a number of potential agonists and antagonists were evaluated.
β-alanine (1 mm) was found to elicit in both the medial and buccal cells a Cl-dependent response that reflects the activation of the glutamate rather than of the GABA receptor. The response to β-alanine was, on average, 45% of the peak amplitude of the response to 1 mm glutamate in the same cells (n = 5; ς = 5.59%). Like the glutamate-activated conductance, that activated by β-alanine showed no voltage dependence (Fig.3 A). Furthermore, the response to β-alanine, like the glutamate response, was not blocked by intracellular sulfate. In a cell held at −30 mV (Fig. 3 B), the increase in outward-going currents in response to both glutamate and β-alanine as whole-cell perfusion proceeds reflects the perfusion-induced change in ECl. The two responses show the same percent increase. The mean difference in increases for the two responses in such experiments was, forn = 3, 1% (ς = 0.8%). Another manifestation of the fact that β-alanine activates the glutamate receptor preferentially can be seen in Figure 3 C in which 50 μm β-alanine is shown to selectively desensitize–block the glutamate Cl-dependent response (forn = 5, mean glutamate response in β-alanine was 8% of control with ς = 3.67%; mean GABA response in β-alanine was 81.8% of control with ς = 8.93%; p ≤ 0.0002). This selectivity of β-alanine is not maintained at higher concentrations (see below) (Ikemoto et al., 1988a).
β-alanine is primarily known for its preferential action on vertebrate glycine receptors (Wu et al., 1993; Jonas et al., 1998). No glycine-induced Cl-dependent responses have as yet been reported in invertebrate neurons, and even at 10 mm, glycine failed to increase Cl conductance in any of the cell types used in this study (three experiments for each cell type; data not shown).
In addition to activating a Cl conductance, glutamate has been shown to act on two other pharmacologically and ionically distinct receptors: one mediating a cationic excitatory response (see the rapid inward current in the medial cells) (Kehoe, 1978; King and Carpenter, 1987), and the other, a K-dependent inhibitory response (Kehoe, 1978, 1994;Katz and Levitan, 1993). β-alanine neither activated nor blocked the glutamate receptors mediating these two responses (results from three experiments studying each of the two other glutamate receptors).
The GABA analog TACA activates a conductance that, like that activated by GABA, is voltage-dependent (Fig.4 A) and is blocked by intracellular sulfate ions (Fig. 4 B). Furthermore, TACA (100 μm) preferentially desensitizes–blocks the GABA response (Fig. 4 C) (Fig.4 A–C, see legend for statistical analyses of the findings illustrated).
Hypotaurine, an endogenous amino acid shown to have inhibitory effects in cerebellar slices (Okamoto and Sakai, 1981), had no agonist effect on any of the cells tested. However, it did preferentially block the Cl-dependent response to glutamate, having only a very slight effect on the Cl-dependent response to GABA (Fig.5 A, see legend for statistical evaluation of relevant experiments). Hypotaurine had no blocking effect on either the cationic or the K-dependent glutamate response (three experiments on each response type; data not shown).
CACA, an agonist of GABAC vertebrate receptors (Johnston, 1996), induced only a weak current in the cells studied here, but, like TACA, preferentially desensitized–blocked the GABA Cl-dependent response, having only a very weak blocking effect on the glutamate response (Fig. 5 B, see legend for statistical analysis of three experiments of this type).
A number of other compounds (TC, strychnine, bicuculline, and pentobarbital) were tested at 100 μm on the glutamate- and GABA-induced Cl-dependent responses (data not shown). TC and strychnine were without effect on GABA responses but caused a small reduction in the glutamate response. The glutamate response in 100 μm strychnine was reduced to an average of 86% of the control value (n = 3; ς = 2%). Bicuculline caused a slight diminution in both the glutamate and GABA responses, whereas pentobarbital reduced the GABA response (Ikemoto et al., 1988a) to ∼20–30% of the control amplitude (three experiments) but had practically no effect on the glutamate response. Even with 1 and 10 μm concentrations, the effect of pentobarbital was to reduce rather than potentiate the GABA response.
Differential expression of GABA and glutamate receptors in different cells
Cells in which the glutamate but not the GABA receptor is expressed reveal that GABA can activate the glutamate receptor
In a group of small identifiable cells in the right pleural ganglion (Ascher et al., 1978) in which glutamate activates a marked Cl-dependent response, GABA was shown to elicit a very weak response (Fig. 6 A) that does not share the characteristics of the GABA responses seen in the medial or buccal cells. The GABA-activated Cl conductance in the small pleural cells was shown to be independent of voltage (Fig.6 B, response to 10 mm GABA) and unaffected by intracellular sulfate ions (Fig. 6 C). Furthermore, like the response to glutamate in all of the cells studied, it was markedly reduced by 50 μmβ-alanine (Fig. 7 A), weakly blocked by 100 μm TACA (Fig. 7 B), markedly reduced by 1 mm hypotaurine (Fig.7 C), and only slightly affected by 200 μm CACA (Fig. 7 D). In all of these respects, the current elicited by GABA in these cells resembles the response to glutamate seen in the medial and buccal cells (see figure legend for a statistical evaluation of these findings). This suggests that it is mediated by the glutamate-sensitive Cl channel and that the GABA-gated Cl receptor–channel complex is not expressed in these cells.
As was seen in Figure 6 A, GABA activation of the glutamate receptor yields a response that is much smaller than that elicited by the same concentration of glutamate (for n= 3, the average response to 1 mm GABA is 3.83% that of the response to 1 mm glutamate). Nevertheless, there appears to be little difference in the concentrations of the two amino acids needed to elicit a threshold response via the glutamate receptor. When testing 20 and 50 μm glutamate and 20 and 100 μm GABA on the pleural cells, no response could be elicited with 20 μm of either agonist. A response was elicited by 50 μmglutamate in all five cells (average of 1268 pA), whereas 100 μm GABA elicited a response (average of 142 pA) in four of the five cells. The fifth cell responded to 200 μm, the next higher concentration tested.
GABA activation of the glutamate receptor explains the observation of cross-desensitization
Cross-desensitization between transmitter-activated Cl channels has been described frequently in both vertebrate and invertebrate preparations (invertebrate, Franke et al., 1986; Zufall et al., 1988;King and Carpenter, 1987, 1989; vertebrate, Barker and McBurney, 1979;Choquet and Korn, 1988; Krishtal et al., 1988; Baev et al., 1992; Lewis and Faber, 1993; Trombley et al., 1999). In view of the finding that GABA can activate the glutamate receptor mediating the Cl-dependent response, it is not surprising to see that GABA has a desensitizing–blocking effect on the response to glutamate. Even at low concentrations (e.g., 1 and 10 μm, which induce no visible current in cells having only the glutamate receptor), GABA significantly reduces the glutamate response in all cells expressing the glutamate receptor (pleural, medial, and buccal cells) (Fig.8, recordings from a pleural cell). After application of a given GABA concentration, the glutamate response was allowed to stabilize, and all measurements describing the diminution of the glutamate response by GABA refer to such “stabilized” responses. The diminution in the glutamate response increased with increasing concentrations of GABA (Fig. 8). No statistical differences were seen in the percent diminution in these stabilized responses for a given concentration of GABA, whether or not that concentration had been preceded by another, weaker concentration or applied singly (t tests comparing such experiments yieldedp = 0.385 for 10 μm GABA andp = 0.305 for 100 μm GABA). Consequently, for describing the diminution in the glutamate response as a function of GABA concentration, all data for a given concentration have been lumped together. The response to 1 mmglutamate in 1 μm GABA was reduced to 85.1% of the control amplitude (ς = 7.16%; n = 7), in 10 μm GABA to 70.2% (ς = 11.8%;n = 7), and in 100 μm GABA to 34.9% (ς = 4.22%; n = 8).
It should be mentioned at this point that changes in ECl cannot account for the GABA-induced diminutions in the amplitude of the glutamate response: (1) ECl was continually monitored, and evaluations of the glutamate response were always made at a holding potential of a constant distance from ECl; (2) application of GABA had the same desensitizing effect on the glutamate response at ECl and at less negative potentials; (3) GABA had a marked desensitizing effect at concentrations that elicited no current in the cell in which the glutamate responses were being evaluated; and (4) in many neurons (e.g., the medial cells), whereas a Cl current gated by the GABA receptor blocked the glutamate response, a similar Cl current gated by the glutamate receptor failed to affect the GABA response.
Likewise, it should be pointed out that an action of GABA on presynaptic neurons can be rejected as a possible source for the cross-desensitization effect. A number of similar experiments (data not shown) performed here on isolated cells in culture confirmed completely the findings from cells in the intact ganglion. Furthermore, the experiments of King and Carpenter (1987) and Oyama et al. (1990), both of which showed similar GABA-induced diminution in glutamate responses, were performed on isolated cells.
Although GABA clearly desensitizes–blocks the glutamate-induced Cl-dependent response, it has no effect on either the glutamate-induced cationic or the glutamate-induced K-dependent response (three experiments made for each response type; data not shown).
Cells in which the GABA but not the glutamate receptor is expressed reveal that glutamate cannot activate the GABA receptor
There are identifiable cells in the cerebral ganglion in which GABA, but not glutamate, elicits a Cl-dependent response. The response to glutamate in these cells is a pure cationic one and that to GABA is a pure Cl-dependent one, as illustrated in Figure9 A, first column. Although, as has been shown above, GABA can effectively activate (Figs.6, 7) and block (Fig. 8) the glutamate Cl-dependent response, the opposite is not true. In Figure 9 A, glutamate added to the control tube completely desensitized the glutamate excitatory response in the cerebral cells but had no effect on the GABA Cl-dependent response (Oyama et al., 1990). Even at a 10 mmconcentration, glutamate failed to elicit an increase in Cl conductance in these cells and failed to diminish the GABA-induced Cl-dependent response (mean response to 1 mm GABA in the presence of 10 mm glutamate for four experiments was 100% of control; ς = 2.36%).
The failure of glutamate to block the Cl-dependent response to GABA can also be seen in the medial cells in which both glutamate and GABA receptors mediating Cl-dependent responses are expressed (Fig.9 B) (mean response to 1 mm GABA in the presence of 1 mm glutamate for four experiments was 102% of control; ς = 2.46%). See the figure legend of Figure 9, A and B, for a more detailed description.
It was mentioned above when discussing the agonist activity of β-alanine on the glutamate receptor that β-alanine could, at higher concentrations, activate the GABA receptor. The cerebral cells used for Figure 9 A permitted the confirmation of this activation, because β-alanine in these cells elicited a weak voltage-dependent response that was unaffected by 1 mm glutamate and markedly blocked by 100 μm TACA, all of these findings consistent with the hypothesis that β-alanine in these cells was activating, albeit weakly, the GABA receptor. Because the response to 1 mm β-alanine in these cells was very weak (∼3% of that to 1 mm GABA), the analyses of this response were made using 10 mmβ-alanine. The amplitude of the response in the cerebral cells to 10 mm β-alanine was ∼21% of that to 1 mm GABA (n = 3; ς = 4.77%). The mean percent inward to outward current (n= 3) for β-alanine in these cells was indistinguishable from that for GABA (51.3 and 52.3%, with ς = 2.52% and 1.15%, for β-alanine and GABA, respectively, measured at 20 mV ± ECl). Comparing these results with the β-alanine response in cells having only the glutamate receptor (Fig.3) reveals the preferential, but incompletely selective, activation by β-alanine of the glutamate receptor.
In cells in which the glutamate receptor is highly expressed, the response to 1 mm GABA often contains a detectable component mediated by the glutamate receptor
In light of the finding that GABA can activate the glutamate receptor (Figs. 6, 7), it is interesting to turn our attention to the differences in kinetics of the response to 1 mm GABA in the medial versus buccal neurons. The expression of the glutamate-gated Cl channel is much greater in the buccal than in the medial cells, with the response to 1 mm glutamate being typically approximately twice that to 1 mm GABA in the buccal cells, and vice versa in the medial cells.
In the medial cells, the decay of the response to 1 mm GABA is almost always smooth, with no change in rate of decay manifested at the end of the 2 sec GABA pulse (Figs. 1 A,9 B). At this concentration, the decay appears to depend exclusively on the “off rate” of the desensitized GABA receptor. That is not the case for the response to glutamate, which shows a sharp change in decay rate when the control flow replaces the agonist flow.
In the buccal cells, on the other hand, the response to 1 mm GABA often shows an “atypical” decay (Figs.1 B, 9 C). As can be seen in Figure9 C′, the “hump” sometimes seen in the GABA response of the buccal cells disappears with a “desensitizing” application of 100 μm glutamate in the control tube, leaving a decay typical of the GABA response mediated by the voltage-dependent, sulfate-sensitive GABA receptor. The transformation in the GABA response in the presence of 100 μm glutamate was evaluated by approximating the integrals of the GABA responses in the presence and absence of 100 μm glutamate and subtracting these integrals. The mean reduction in the control GABA response thus obtained was 10.3% (n = 6; ς = 3.94%). This 10% diminution by 100 μmglutamate of the GABA response in some buccal cells should be compared with the total failure of even 10 mm glutamate to affect the GABA response in the cerebral cells, which are lacking the glutamate receptor. The element of the response desensitized by 100 μm glutamate (Fig. 9 C′,Difference trace) can hence be attributed to a current resulting from the activation by GABA of the glutamate receptor.
Comparison of the amino acid receptors with the two ACh receptors mediating Cl-dependent responses
It was shown recently by Kehoe and McIntosh (1998) that two pharmacologically distinct receptors mediate the ACh-induced increase in Cl conductance. One receptor mediates a rapidly desensitizing response, and the other mediates a sustained response. These Cl-dependent cholinergic responses are seen, alongside the glutamate- and GABA-induced Cl responses, in the medial cells and in selected buccal cells but are not always associated with them in other cells. For example, on the small pleural cells that express no GABA receptor but do express the glutamate receptor mediating a Cl-dependent response, ACh elicits a pure cationic response (Ascher et al., 1978).
The ACh-activated conductance, like that activated by glutamate, fails to show a voltage dependence over the range of holding potentials studied here (Kehoe and McIntosh, 1998), and neither of the two Cl-dependent responses elicited by ACh was affected by intracellular sulfate ions (Fig.10 A). Whereas the response to ACh after perfusion with a sulfate-based pipette solution remained essentially unchanged (98.8% of control; ς = 3.4%), that to GABA was markedly reduced (35.7% of control; ς = 19%;n = 3; p = 0.005).
The pharmacology of the Cl-dependent ACh responses has been described in detail by Kehoe and McIntosh (1998) in which the effects of 100 μm TC and strychnine can be seen. MLA, like α-bungarotoxin (Kehoe and McIntosh, 1998), blocked both of the cholinergic Cl-dependent responses but had no effect on either the glutamate- or GABA-induced responses (Fig. 10 B, records taken from a buccal cell). The ACh response in MLA, measured at the time corresponding to the peak of the control ACh response, was only 1% of the control (n = 3; ς = 0.5%); the responses to glutamate and GABA in the presence of MLA were 103% (ς = 0.58%) and 102% (ς = 3.4%) of control, respectively, for the two sets of responses. Neither glutamate nor GABA desensitized–blocked the ACh responses (Fig.10 C,C′), and ACh, in turn, had no desensitizing–blocking effect on the responses to the two amino acids (Fig. 10 D,D′) (Ikemoto and Akaike, 1988). These latter experiments, however, had to be performed in cells that did not respond to ACh with a strong nondesensitizing Cl current (Kehoe and McIntosh, 1998). In a cell in which such persistent outward ACh-induced currents exist, the application of either glutamate or GABA solutions (which did not contain ACh) would cause a change in current because of the washing off of ACh, giving a false estimation of either the GABA- or glutamate-activated current. Statistical analyses of the results illustrated in Figure 10 are presented in the figure legend.
The experiments described here have shown that the receptor–channel complexes mediating the Cl-dependent responses to glutamate, GABA, and ACh in Aplysia neurons are independent entities. Nevertheless, cross-desensitization does occur between the GABA and glutamate responses and was shown to result exclusively from the ability of GABA to interact with the glutamate receptor.
These findings are compatible with the results obtained by King and Carpenter (1987) but not with their conclusion that the glutamate and GABA responses gate the same receptor–channel complex. On the other hand, our results are incompatible with those of Ikemoto and Akaike (1988), who saw no interactions between the GABA and glutamate responses. These same authors participated in a later study (Oyama et al., 1990) in which interactions were seen, and GABA was shown to elicit both a low-threshold and a high-threshold response. In the 1990 study (Oyama et al.), the authors proposed that the high-threshold GABA response they observed was attributable to the activation, by GABA, of the glutamate receptor, a hypothesis confirmed by the results presented here.
Glutamate and GABA activation of the glutamate receptor
In our study, the concentrations of GABA and glutamate required for threshold activation of the glutamate receptor were found to be remarkably similar. That is, to consistently obtain a discernible response to glutamate, a concentration of 50 μm was needed; GABA was shown to activate the same receptor at only twice that concentration. In contrast, the average currents elicited by 1 and 10 mm GABA in cells having only the glutamate receptor (Fig.6 A,B) were only ∼4 and 36%, respectively, of the response of the same cell to 1 mm glutamate.
The low concentrations of GABA that are able to affect the sensitivity of the glutamate receptor (Fig. 8) suggest that, if both GABA and glutamate receptors are coexpressed on the same subsynaptic membrane, liberation of GABA on that membrane would be able to cross-desensitize the glutamate receptor. The protocol used here did not permit an evaluation of the kinetics of the GABA-induced desensitization. However, from the work of Oyama et al. (1990), it is clear that a 10 sec application of 300 μm GABA is sufficient to cause a very marked diminution in the response to the same concentration of glutamate.
Although the data strongly suggest that direct activation by GABA of the glutamate receptor mediates the cross-desensitization process, the molecular mechanism by which GABA and β-alanine desensitize the glutamate receptor remains to be established. The fact that the response to glutamate shows auto-desensitization suggests that there exists a “desensitized state” of the glutamate receptor, and GABA and β-alanine may act by favoring the transition toward this state. However, the strength of the cross-desensitization appears to be correlated with neither the strength of activation by the “inhibiting” compound nor the rate of desensitization it induces in the “inhibited” receptor. The inhibition induced by GABA and β-alanine may be, to some degree, a partial agonist effect. That is, GABA and β-alanine may compete with glutamate for binding to the same site and inhibit the glutamate response because they induce a much lower probability of opening of the channels.
Relationship between the invertebrate glutamate receptor and the vertebrate glycine receptor
The invertebrate glutamate receptors gating Cl channels that have thus far been cloned show greater resemblance to the vertebrate glycine receptor than to any other receptor (Cully et al., 1994; Vassilatis et al., 1997). Furthermore, no one has reported evidence for a glycine-gated Cl channel in invertebrate neurons. Conversely, the only glutamate-gated Cl channels thus far seen in vertebrates appear to be associated with glutamate transporters (for review, see Fairman and Amara, 1999). It thus appears reasonable to relate the invertebrate glutamate receptors to the vertebrate glycine receptors. Despite the failure of the glutamate receptor studied here to be blocked by 100 μm strychnine, its sensitivity to β-alanine reinforces this parallel (Wu et al., 1993; Jonas et al., 1998).
A recent study evaluating the differential activation by glycine and GABA of a recombinant homomeric glycine receptor revealed a 200- to 300-fold difference in the EC50 values of the activation curves for the two amino acids (Fucile et al., 1999). In contrast, in a study measuring the activity of GABA and glycine on a recombinant homomeric human GABA ρ receptor, the difference in sensitivity to the two amino acids was much greater, on the order of 10,000-fold (Calvo and Miledi, 1995).
In previously published experiments on isolated Aplysianeurons, concentration–response curves were obtained for the glutamate response (Ikemoto et al., 1988b) and for the high-threshold GABA response hypothesized to be mediated by the glutamate receptor (Oyama et al., 1990) and yielded EC50 values of 130 and ∼6 mm, respectively. The difference in EC50 values is higher (∼46-fold) than that between the threshold values referred to above (twofold), probably because of a smaller Hill coefficient of the GABA response. Yet the difference is surprisingly small compared with the failure of even very high concentrations of glutamate (10 mm) to activate the GABA receptor, for which the EC50for activation by GABA was estimated to be 64 μm (Ikemoto et al. (1988a).
These studies together suggest a high selectivity of both the vertebrate and invertebrate GABA receptor, whereas both the vertebrate glycine receptor and the invertebrate glutamate receptor show some sensitivity to GABA.
However, many molecular variations of GABA and glycine receptors exist in vertebrate neurons, and this diversity may explain the wide range of findings concerning cross-desensitization between these ligand-gated Cl channels. For example, whereas Choquet and Korn (1988) found no interaction between GABA and glycine responses in cultured chick spinal neurons, Baev et al. (1992), studying lamprey spinal cord neurons, found that there was total cross-desensitization between the responses to the same amino acids. They concluded that only one receptor–channel complex, responding to both amino acids, was expressed on these neurons. Recently, Trombley et al. (1999) described all possible cross-desensitization profiles between GABA and glycine responses in rat olfactory bulb neurons in primary culture, with the profile depending on the neuron studied. Their findings suggest that more than one type of receptor for each amino acid is expressed in these cells.
A similar diversity must exist in invertebrate neurons because, whereas the glutamate receptor of molluscs is desensitized by even low concentrations of GABA, no interaction between the two amino acid responses is seen in some invertebrate ganglionic neurons (lobster,Cleland and Selverston, 1998; crab, Duan and Cooke, 2000).
The ACh-activated receptor–channel complex
Although we have no data that permit us to discriminate between the properties of the ACh-gated channel(s) (Kehoe and McIntosh, 1998) and those of the glutamate-gated channel, the fact that the ACh and glutamate receptor channel complexes are not consistently coexpressed permits us to conclude that, even if the channels per se are identical, the receptor–channel complexes for the different transmitters are distinct molecular entities.
GABAA or GABAC?
The object of the pharmacological evaluation of the GABA receptor in this study was primarily to differentiate it from the receptor underlying the glutamate response. However the finding that TACA is a more effective agonist than CACA suggests that the molluscan GABA receptor resembles the GABAC rather than the GABAA receptor (Johnston, 1996), a conclusion likewise drawn from previous studies showing that the molluscan receptor is not blocked by bicuculline and is not potentiated by either diazepam (Ikemoto et al., 1988a; Kim and Takeuchi, 1990) or pentobarbital (Ikemoto et al., 1988a) (but see Kim and Takeuchi, 1990). The molluscan GABA receptor desensitizes rapidly, a characteristic associated with the GABAA receptor. However, rapidly desensitizing GABAC receptors have been described recently in vertebrates (Han et al., 1997; Boué-Grabot et al., 2000).
Sulfate blockade of the GABA channel
The observation made here that intracellular sulfate ions selectively block the GABA response was exploited in these experiments as a “tool” for differentiating the GABA receptor–channel complex from the glutamate and ACh receptor–channel complexes, and the means by which this blockade occurs have not as yet been examined. However, the block is clearly independent of voltage over the range tested (Fig.2 B). Furthermore, a differential permeability of the GABA receptor channel to sulfate ions cannot explain its differential sensitivity to internal sulfate. It was shown previously (Kehoe, 1972a) that sulfate cannot pass through the channels opened by ACh, and it was shown here that partial substitution of intracellular Cl by sulfate alters the inversion potential of all three agonist-induced responses to the same extent.
This research was supported by the Centre National de la Recherche Scientifique (Unité Mixte de Recherche 8544) (to J.K.) and by Université Pierre et Marie Curie Visiting Professorship and INTAS-93–3269 (to C.V.).
Correspondence should be addressed to JacSue Kehoe, Laboratoire de Neurobiologie, Ecole Normale Supérieure, 46, rue d'Ulm, Paris, 75005, France. E-mail:.