Bicuculline/Baclofen-Insensitive Gaba Response in Crustacean Neurones in Culture

The first experiments demonstrating the inhibitory action of y-aminobutyric acid (GABA) were performed on invertebrate neurones. Florey (1954) showed that a substance extracted from mammalian brain (Factor I), now identified as GABA (Bazemore et al. 1956), had an inhibitory effect on the stretch receptor neurone in the crayfish Astacus astacus. GABA was also demonstrated to be active on the isolated cardiac ganglion of the horseshoe crab Limulus polyphemus (Burgen and Kuffler, 1957). It is now generally accepted that GABA functions as an inhibitory transmitter in the peripheral and central nervous system of invertebrates (Robinson and Olsen, 1988). In vertebrates, GABA is thought to be the predominant inhibitory transmitter in the central nervous system, where two well-characterized classes of GABA receptors can be distinguished. GABA_A receptors are ligand-gated Cl⁻ channels that are antagonized by bicuculline and picrotoxin and are allosterically modulated by barbiturates and benzodiazepines (reviewed by Johnston, 1986). In contrast, GABA_B receptors regulate K⁺ and Ca²⁺ channels through GTP-binding proteins and intracellular messenger pathways (reviewed by Bowery et al. 1991). It is becoming clear that, in invertebrates, GABA receptors present similarities in pharmacological profile with the mammalian ones without fitting precisely into the classification developed for mammalian brain. Invertebrates have specific binding sites similar to the GABA/muscimol sites of mammalian GABA_A receptors, but most of them differ from those of vertebrates by a markedly reduced sensitivity to bicuculline and by a distinctly different benzodiazepine pharmacology and a weaker affinity binding to picrotoxin (Lunt, 1991).

Recent observations indicate that the vertebrate retina contains a new type of GABA receptor (Feigenspan et al. 1993; Qian and Dowling, 1993), with unusual pharmacological properties, as proposed earlier for GABA_C receptors (Johnston et al. 1975; Drew et al. 1984). These receptors are activated by an analogue of GABA in a folded conformation, cis-4-aminocrotonic acid (CACA), but are insensitive to the GABA_A antagonist bicuculline and to the GABA_B agonist baclofen. Moreover, the GABA response in these receptors is not modulated by barbiturates and benzodiazepines (Feigenspan et al. 1993; Qian and Dowling, 1993). Here we report that GABA receptors in lobster thoracic neurones in culture have a pharmacological profile that resembles that of the GABA_C receptor in the vertebrate retina. Part of this work has been published in the form of abstracts (Jackel et al. 1993).

Egg-bearing female lobsters (Homarus gammarus) were obtained from commercial suppliers (Aiguillon-Marée, Arcachon) and maintained in laboratory tanks with circulating sea water at a temperature of 13°C and a 12 h:12 h day:night cycle. Electrophysiological studies were performed on isolated neurones from adult lobsters of both sexes and from embryos at 60–100% of development. Lobster embryos were staged by measuring the size of the pigmented area of the eye with an ocular micrometer (Perkins, 1972).

Embryos were removed from the egg and pinned to the bottom of a Sylgard-lined dish. Muscles surrounding the nerve cord were removed and the five thoracic ganglia dissected. Thoracic ganglia from adult lobster were desheathed prior to enzyme treatment. All following steps were identical for the two preparations.

The method we used to culture neurones from marine crustaceans was adapted from the method originally developed for crayfish (Krenz et al. 1990). Briefly, thoracic ganglia were incubated for 1–2 h in either collagenase/dispase (Boehringer Mannheim) or subtilisin (Sigma), both 2 mg ml⁻¹ (collagenase/dispase in Ca2+/Mg2+-free medium). The enzyme was removed by rinsing in sterile medium containing 10% foetal calf serum and gentamycin (50 μg ml⁻¹, Sigma) and streptomycin (100 μg ml⁻¹, Sigma). The following steps were carried out under sterile conditions in a laminar flow hood and with sterile-filtered solutions. Ganglia were washed again by transferring them through three volumes of medium. In the final culture dish, the cells were removed from the ganglia by gentle suction through fire-polished micropipettes. Culture dishes were coated with concanavalin A (500 μg ml⁻¹, Sigma) in order to facilitate cell adhesion. The medium contained equal parts of Leibovitz’s L15 medium and double-strength saline. The final composition of the salts in the culture medium was (in mmol l⁻¹): 490.55 NaCl, 15.42 KCl, 14.3 CaCl₂, 10.41 MgSO₄, 0.49 MgCl₂, 0.67 Na₂HPO₄, 0.22 KH₂PO₄, 3.91 Na₂SO₄, 10.0 Hepes. The pH was adjusted to 7.45 with NaOH and osmolality was 1042 mosmol l⁻¹.

After 24 h, the culture medium was replaced with medium containing 10% foetal calf serum. The cultures were stored for 3–5 days before experimentation. Under these conditions many neurones survived and grew processes (Fig. 1).

The solutions used for electrophysiological experiments were made up as follows. Standard external solution contained (in mmol l⁻¹): 521 NaCl, 12.74 KCl, 13.67 CaCl₂, 10 MgSO₄, 3.91 Na₂SO₄, 5 Hepes (pH adjusted to 7.45 with HCl). Standard internal solution contained (in mmol l⁻¹): 48.37 NaCl, 20.63 KCl, 139.68 K₂SO₄, 0.5 CaCl₂, 5 MgCl₂, 5 sodium ATP, 5 maleic acid, 11 Tris base, 5 EGTA, 550 D-mannitol (pH adjusted to 7.45 with NaOH). In some experiments this medium was supplemented with 0.2 mmol l⁻¹ sodium GTP.

The intracellular solution for high Cl⁻ contained (in mmol l⁻¹): 48.37 NaCl, 300 KCl, 0.5 CaCl₂, 5 MgCl₂, 5 NaATP, 5 maleic acid, 11 Tris base, 5 EGTA, 430 D-mannitol (pH adjusted to 7.45 with KOH). For low Cl⁻, the composition of the intracellular solution was (in mmol l⁻¹): 8.37 NaCl, 40 sodium isethionate, 20.63 KCl, 139.68 K₂SO₄, 0.5 CaCl₂, 5 MgCl₂, 5 sodium ATP, 5 maleic acid, 11 Tris base, 5 EGTA, 550 D-mannitol (pH adjusted to 7.45 with KOH).

Neurones were investigated under current-and voltage-clamp using the patch-clamp technique in the whole-cell configuration (Hamill et al. 1981). Patch pipettes were pulled from borosilicate glass capillaries with filament (GC150TF-10, Clark Electromedical Instruments) on a vertical pipette puller (L/M-3P-A, List Medical). Fire-polished pipettes were filled with standard intracellular solution (described above) and had resistances of 5–8 MΩ. High-resistance seals formed easily when gentle suction was applied after touching the cell membrane. Rupture of the patch for whole-cell recording was obtained either by applying a brief pulse of suction or a brief pulse of current.

Experiments were performed with a patch-clamp amplifier (Axopatch 1D, Axon Instruments). Cells were clamped at a holding potential of −50 mV, unless otherwise stated. Series resistance values were obtained for each experiment from the patch-clamp amplifier settings after compensation and varied between 6 and 20 MΩ. Stimulations and data acquisition were performed with the pClamp software package (Axon Instruments, USA, version 5.5.1) running on a 20 MHz 386 microcomputer (Packard-Bell, 320SX) equipped with a Labmaster DMA acquisition system (TL-1 interface, Axon Instruments). Data were stored on the computer hard disk and processed later with data analysis programs of the pClamp program package. Curves were fitted using the Sigmaplot software (Jandel Scientific, version 5.0).

In order to determine liquid junction potentials in the presence of different external solutions, the pipette tip (containing the internal standard solution) was first placed into a bath filled with standard external solution and the junction potential between the pipette and the bath was compensated to zero. The external solution was then replaced by modified extracellular solutions and the deviation for each solution was measured. All present data were corrected for junction potentials, which varied between 1 and 4 mV. Results are presented as individual values or means ± standard error (S.E.M.), unless otherwise stated in the text. All experiments were performed in standard extracellular solution (unless otherwise noted) at a temperature of about 17°C.

Drugs used in the present experiments were γ-aminobutyric acid (GABA), muscimol, 3-aminopropylphosphonic acid (3-APA), bicuculline and bicuculline methiodide, all purchased from Sigma. Phaclofen, (−)-baclofen-(b-p-chlorophenyl)GABA (baclofen), 2-(3-carboxypropyl)-3-amino-6-(4-methoxy)-phenyl pyridazinium bromide (SR 95531) and isoguvacine–HCl were purchased from RBI. cis-4-aminocrotonic acid (CACA) was purchased from Tocris Neuramin, Bristol, UK. Diazepam and phenobarbital were a gift from the Cooperation Pharmaceutique Française. All drugs were either dissolved in the external solution or, when insoluble in saline, first dissolved in methanol at a concentration of 10 mmol l⁻¹ and further diluted in saline to their final concentration. Drugs were stored at −20°C and thawed just before use, except for bicuculline which was prepared fresh.

Application of drugs was performed by pressure ejection (Picospritzer II, General Valve Corp., USA) from glass pipettes pulled on the vertical pipette puller. In order to minimize desensitization of the GABA receptor during repetitive application, drugs were applied at 90 s intervals. Comparisons between agonists were undertaken with equivalent pipettes the first of which contained GABA and the second muscimol, isoguvacine or CACA. Both pipettes were positioned at a distance of 40–70 μm from the cell body. Drugs were applied at a concentration of 1 mmol l⁻¹ with brief pulses of pressure (68.94 kPa) of increasing duration. Several experiments with Fast Green added to the pipette solution showed that the ejected bolus of drug did mix with the stream of saline before hitting the cell. For studies of the effect of antagonists and modulators on the GABA-induced current we used both bath application and pressure ejection (see Table 1). For pressure ejection we used the same pipette configuration as described above. The antagonists were applied for 30 s, followed immediately by a brief pulse of GABA (50 ms).

Under current-clamp conditions, application of GABA suppressed the firing of action potentials (Fig. 2A). In this experiment, the cell fired overshooting action potentials at a membrane potential of −33 mV. Each action potential (AP) is followed by an after-hyperpolarization potential (AHP) of 12 mV. Application of GABA at a concentration of 1 mmol l⁻¹ for 50 ms (Fig. 2A, arrow) slightly hyperpolarized the cell and transiently inhibited firing.

The response to GABA was then analyzed under voltage-clamp conditions. Fig. 2B shows the response of a cell held at a potential of −60 mV, a potential at which GABA evoked a large transient depolarization in unclamped neurones. Applying GABA at a concentration of 1 mmol l⁻¹ evoked an inward current of 250 pA. Fig. 2C shows the same cell as in Fig. 2B but, in addition to the holding potential of −60 mV, hyperpolarizing pulses of -20 mV were applied repetitively in order to monitor variations in membrane conductance. Pressure ejection of GABA increased the amplitude of the current, indicating that the GABA response is associated with an increase in membrane conductance.

In order to determine the reversal potential for the GABA-induced current, GABA was applied in 50 ms pulses at various holding potentials. Fig. 3A shows the original recordings from one such experiment. In Fig. 3B the peak currents are plotted against the holding potentials. The solid line is the linear regression through the data points. The I–V relationship is linear within the range of voltages applied. In this cell, the GABA response reversed at a membrane potential of −46 mV. The same type of experiment was repeated in 23 other cells from embryos and in five cells from adults. Mean reversal potential values were −42.7±4.9 mV for embryonic cells and −49.6±1.8 mV for adult ones (no significant difference; Student’s t-test, P>0.05). Both values are close to the equilibrium potential for Cl⁻ calculated from the Nernst equation (E_Cl=-49 mV). The observations that GABA produced membrane responses that inverted at a potential close to the equilibrium potential for Cl⁻ and that it induced an increase in membrane conductance suggest that opening of Cl⁻ channels underlies the GABA-induced response. To test this hypothesis, we modified the intracellular and extracellular concentrations of Cl⁻.

Both low and high intracellular chloride solutions were used. In the first case, the normal Cl⁻ concentration was decreased by 50% by replacement with sodium isethionate to obtain a final Cl⁻ concentration of 40 mmol l⁻¹ (see Materials and methods). This shifted the equilibrium potential for Cl⁻ to −66 mV according to the Nernst equation. For the high-chloride solution, the intracellular concentration of Cl⁻ was modified by increasing [Cl⁻] to 359 mmol l⁻¹ with KCl (see Materials and methods) thus shifting the equilibrium potential for Cl⁻ to −11 mV.

The recordings in Fig. 4Ai show GABA-evoked currents at different voltages when the imposed equilibrium potential for Cl⁻ was −66 mV, −49 mV and −11 mV. The interpolated reversal potentials are indicated by the arrows. Fig. 4Aii illustrates the current–voltage relationships obtained from the above data. The I–V relationships remained linear over the range of membrane potentials tested and were only shifted along the voltage axes in parallel with the modified intracellular Cl⁻ concentrations. In these experiments, the GABA current reversed at −48 mV under control conditions (circles). The reversal potential was −11 mV for high intracellular [Cl⁻] (squares) and −63 mV for low intracellular [Cl⁻] (triangles). Mean values for embryonic cells determined from five experiments of this kind gave reversal potentials of −59±1.33 mV for low-chloride and −6.1±2.2 mV for high-chloride solutions. These values coincide well with the Cl⁻ equilibrium potentials calculated from the Nernst equation. This result confirms the involvement of Cl⁻ in the GABA response.

In order to test whether the equilibrium potential of the GABA response in embryonic cells was also affected by changing the extracellular concentration of Cl⁻, alterations were made with the appropriate substitutions of NaCl by sodium methanesulphonic acid in the superfusion solution. Fig. 4B demonstrates the effect of substitution of 50% of the Cl⁻. The lines drawn through data are linear regressions for each condition. Data points are mean values ± S.E.M. of two GABA applications. Again, the I–V relationships are linear and are shifted along the voltage axes with modified extracellular Cl⁻ concentration. Under control conditions, the GABA-evoked current reversed at −42.6 mV in this cell (E_Cl=−49 mV). In the modified external solution, the reversal potential for the GABA-induced current in the experiment shown in Fig. 4B was shifted to −31.1 mV (ΔV=-11.5 mV). In most experiments, however, the change in reversal potential was only 9.76±0.47 mV (N=4). This value represents only half the change in reversal potential predicted by the Nernst equation (−18 mV).

Since ionotropic cation-selective GABA receptors have been described in crustaceans (see Discussion), we wished to assess whether other ions might also be involved in the GABA response studied here. Therefore, we measured the response of embryonic cells to GABA when the extracellular concentrations of Na⁺, K⁺ and Ca²⁺ were modified.

In 50% of the cells tested (N=10), the reversal potential of the GABA current was not displaced when the extracellular K+ concentration was either increased 2.5-fold (K×2.5) or decreased to half of the standard concentration (K/2). An example of one such response is shown in Fig. 5A. When the peak values were plotted against the different membrane potentials, a linear relationship was observed. The reversal potential of the GABA-induced current in this cell was −41 mV for all three conditions. Mean values of five experiments where cells responded in the same way to GABA applications were −40.86±0.29 mV for controls, −40.6±0.29 mV for increased [K⁺] (29.84 mmol l⁻¹) and −42.05±0.55 mV for decreased [K⁺] (6.37 mmol l⁻¹). In the other 50% of the cells, we observed a slight change of the reversal potential of the GABA-evoked current (data not shown). In these cells, the GABA current reversed at a membrane potential of −46.2±2.2 mV under control conditions. In the presence of saline containing twice the normal K⁺ concentration, the reversal potential was found to be −40.3±1.6 mV.

Superfusion of the cells with K/2 shifted the reversal potential of the GABA current to −52.4±3.4 mV. These values represent changes of only one-third to one-quarter of those predicted by the Nernst equation for K⁺, namely 5.9 mV instead of 23 mV for K×2.5 and 6.2 mV instead of 18 mV for K/2. In these experiments the GABA current always reversed at a membrane potential closer to the chloride equilibrium potential (−49 mV) than to the K⁺ equilibrium potential (E_Kcontrol=−79 mV, E_K×2.5=−56 mV, E_K/2=−97 mV). These findings further suggest that the GABA current is not carried by K⁺.

Fig. 5B shows the results when the extracellular Na⁺ concentration was modified. GABA was applied at various membrane potentials and the peak current was measured under control conditions and when 50% of the Na⁺ was replaced by glucamine. Solid lines were obtained from linear regressions through the data points. In all tested cells (N=5) the equilibrium potential of the GABA response was not affected by Na⁺ replacement. Furthermore, the slopes of the solid lines did not change, showing that the amplitude of the response for all membrane potentials was not modified. These results led us to conclude that Na⁺ is not involved in the GABA-induced current.

In order to determine whether the GABA response was partially dependent upon Ca²⁺, the potent calcium-channel blockers Mn²⁺ and Cd²⁺ were added to the external saline at a concentration of 5 mmol l⁻¹. Fig. 5C illustrates one experiment where the extracellular solution contained Mn²⁺. In this cell, the reversal potential of the GABA-induced current was at −37 mV (open squares) and was unaffected by the presence of Mn²⁺ (filled squares). Furthermore, the amplitude of the GABA current was the same for controls and for saline containing Mn²⁺. Cd²⁺ was also without effect on the GABA current (N=5, data not shown). From the above results, it can be concluded that the current induced by GABA is not mediated by Na⁺ or Ca²⁺, but that Cl⁻ is the main carrier of the GABA-induced current. We cannot, however, exclude some involvement of K⁺ in the GABA response but, if this is so, its influence is small.

Further evidence that the GABA current is mediated by chloride ions was obtained from experiments with picrotoxin (PTX). PTX is thought to block the Cl⁻ ionophore either directly or following binding to a closely located site in the vertebrate GABA_A receptor (Barker et al. 1983). Fig. 6A shows the effect of PTX on a single cell clamped at a potential of −60 mV. As can be seen, PTX reduced the amplitude of the GABA-induced current by about 60%. This effect was only partially reversible, even following prolonged superfusion with control saline. Fig. 6B illustrates the current–voltage relationship for the same cell in control saline and in saline containing PTX. The reduction of the GABA current by PTX was proportional at all membrane potentials, so that a linear relationship was maintained. That the blockade is independent of voltage is further illustrated in Fig. 6C, which summarizes the results from five experiments performed on embryonic cells. Normalized currents remaining after application of PTX are represented as a percentage of the control value, recorded at three different membrane potentials. The percentage of blockade is almost identical at the three potential values. Mean values of PTX block were 73.8±7.3% for embryonic (N=5) and 64.5±2.3% for adult cells (N=3). This difference is not statistically significant (Student’s t-test, P>0.05). Higher doses of PTX did not increase the percentage of blockade. 50 μmol l⁻¹ PTX produced 59.3±3.6% of blockade in three adult cells. The difference between the effects of 10 and 50 μmol l-1 PTX is not statistically significant (Student’s t-test, P>0.05).

Having established that GABA induces a membrane current carried mainly by chloride ions in Homarus gammarus thoracic neurones in culture, we attempted to determine the pharmacological properties of the GABA receptor activating the chloride conductance. The results are summarized in Table 1.

As a first step, we tested several specific GABA_A receptor agonists on embryonic and adult neurones. Muscimol and isoguvacine are known to be agonists at the GABA_A receptor in vertebrates (Barker and Mathers, 1981). When applied by pressure ejection onto cells clamped at a holding potential of −60 mV, muscimol and isoguvacine induced inward currents. Fig. 7A shows the response of one embryonic cell to GABA and isoguvacine and Fig. 7B shows the response to GABA and muscimol in another cell. Note that the amplitudes of the currents were different, although the agonists were applied with identical ejection parameters. The muscimol-induced current was larger than the GABA-evoked current. Isoguvacine evoked the smallest current and the duration of the response was also shorter (see below). I–V curves for muscimol and isoguvacine demonstrated that the current inverted at the same potential as the GABA-induced current (data not shown).

In order to determine whether the GABA-and muscimol-evoked responses are mediated by the same receptor, both substances were applied to the same voltage-clamped neurones with a short interval of 1 s. Fig. 7C shows the response of one embryonic cell when first GABA and then muscimol were applied. Fig. 7D shows the same cell when muscimol was applied first. In both cases, the amplitudes of the responses were not additive, indicating that both muscimol and GABA bound to and activated a common population of receptors.

The graphs in Fig. 8A,B also indicate that in these two cells GABA, muscimol and isoguvacine are characterized by different T₅₀ values (the concentration of an agonist necessary to produce 50% of the maximal response). Although the T₅₀ values are not actual concentrations, they give an indication of the affinity for the receptor (see Discussion). In the first cell, the T₅₀ value for muscimol-induced current (T₅₀=13 ms) is smaller than that for the GABA current (T₅₀=22 ms), indicating a higher affinity of muscimol for the receptor. The highest T₅₀ value was found for isoguvacine (T₅₀=32 ms), suggesting that its affinity for the receptor is the smallest in the cells tested. Dose–response relationships for all agonists measured in embryonic cells are summarized in Fig. 8C. The peak currents of each cell were expressed as a percentage of the maximal response to GABA and averaged. Mean values for each agonist (N=10 for GABA, N=5 for isoguvacine, N=5 for muscimol) are plotted as a function of pulse duration. The mean T₅₀ values determined from the averaged dose–response curves were 17 ms for muscimol, 20 ms for GABA and 40 ms for isoguvacine.

Although GABA_A agonists activated GABA receptors in lobster thoracic neurones, the latter were completely insensitive to GABA_A antagonists. Bicuculline is a specific competitive antagonist at the vertebrate GABA_A receptor. We tested bicuculline as well as its water-soluble analogue bicuculline methiodide (Fig. 9A) for effects on the GABA-evoked currents at concentrations of up to 100 μmol l⁻¹. In 10 experiments on embryonic neurones and in six on adult neurones, the GABA-induced current was completely unaffected by these two compounds (Fig. 9A). The synthetic GABA_A receptor antagonist SR 95531 was also without effect up to concentrations of 100 μmol l⁻¹ (Fig. 9B, N=5).

The vertebrate GABA_A receptor is associated with binding sites for benzodiazepines and barbiturates, compounds which enhance the action of GABA. We tested two drugs, phenobarbital and diazepam, for modulatory effects on the GABA response in embryonic and adult neurones. Both substances were without effect: the GABA response in the presence of the modulators was not significantly different from that in control saline in embryonic (N=10) and adult (N=4) cells. Together with the insensitivity to GABA_A antagonists, these findings indicate that the GABA receptors of thoracic neurones in culture differ from vertebrate GABA_A receptors.

We next tested whether GABA_B receptors might be involved. We examined the effects of the two specific GABA_B agonists, namely baclofen (Hill and Bowery, 1981) and 3-APA (both tested in embryonic and adult cells, N=13). In all experiments, neither agent induced a response up to a concentration of 1 mmol l⁻¹ (tested over a holding potential range of −70 to −10 mV). Phaclofen, a highly selective antagonist of the vertebrate GABA_B receptor (Kerr et al. 1987; Dutar and Nicoll, 1988) failed to inhibit the GABA-induced current up to a concentration of 1 mmol l⁻¹ (N=5). We thus conclude that GABA_B receptors are either not expressed in lobster thoracic neurones in culture, or, if they are present, do not affect any conductance under our experimental conditions.

Insensitivity to bicuculline and baclofen, as demonstrated in our cells, is a characteristic of a third class of GABA receptors recently described in vertebrates, designated GABA_C (Drew et al. 1984; Johnston, 1986). This type of receptor is thought to be specifically activated by folded GABA analogues such as cis-4-aminocrotonic acid (CACA). Thus, we examined the response of our cells to CACA.

In standard saline, CACA, like GABA, induced inward currents when applied at holding potentials more negative than −50 mV (Fig. 10A). Fig. 10B shows the reversal potentials of the CACA-induced current determined with different solutions in the patch pipette, i.e. standard solution (E_Cl=−49 mV), low-chloride (E_Cl=−11 mV) and high-chloride (E_Cl=−66 mV) internal solutions. The data points (filled circles, means ± S.E.M., N=3) correlate well with theoretical values for the chloride equilibrium potential calculated from the Nernst equation (dotted line). This indicates that the CACA-induced current, like the GABA response (see Fig. 4), is carried mainly by Cl⁻. Moreover, Fig. 10C indicates that CACA and GABA act on the same population of receptors. When CACA and GABA were applied successively, the responses of the two agonists were not additive, the summated response being of the same amplitude as the response for GABA alone.

Similarities between CACA and GABA responses are further illustrated in Fig. 10D,E, which shows that, like the GABA response, the CACA-induced current is substantially decreased in the presence of PTX (50 μmol l⁻¹ in Fig. 10D), but completely unaffected by bicuculline (100 μmol l⁻¹ in Fig. 10E). In three tested cells, PTX reduced the CACA-induced current by 68.6±9.3% (mean ± S.E.M.), while bicuculline was without effect on the CACA response (N=4). Finally, desensitization of the response was slow during prolonged application of CACA as well as GABA. In the experiment documented in Fig. 11A, desensitization over 30 s was only 18.6% for the CACA-induced current and 26% for the GABA-induced current. In four cells evaluated in the same way, the mean values for desensitization were 20.25±6.5% for GABA and 12.2±4.4% (mean ± S.D.) for CACA. Furthermore, in three cells in which we investigated the rate of desensitization to GABA and CACA, we did not detect any significant difference between the two ligands (not shown).

Despite their similarities, however, the CACA and GABA responses showed quantitative differences. First, during prolonged applications as shown in Fig. 11A, the CACA-evoked current took 2–3 times as long (2.75±0.52 s, N=7) as the GABA current (0.96±0.25 s, N=7) to reach its peak. Second, as shown by Fig. 11A and by the dose–response curves for one adult cell in Fig. 11B, the maximal amplitude for the CACA-evoked current (758 pA) was smaller than that for GABA (847 pA). In three tested thoracic neurones, the mean value of the CACA response was 68±16% of the GABA response. Finally, Fig. 11B shows that the t₅₀ value was much smaller for GABA (68 ms) than for CACA (445 ms), indicating a lower affinity for the receptor for CACA than for GABA. Thus, the rank order of potency is extended to muscimol>GABA>isoguvacine>CACA.

Although many invertebrate GABA receptors appear to be related to vertebrate GABA_A receptors in that they consist of a chloride channel and respond to the GABA_A agonist muscimol, other characteristics of their pharmacological profile do not fit into the conventional GABA_A and GABA_B receptor classification. In the present study, we have shown that the pharmacological characteristics of GABA receptors on thoracic neurones from adult and embryonic lobster in culture are similar, instead, to those of a novel type of receptor described in vertebrates as GABA_C (Johnston, 1986). So far, membrane current responses from individual neurones that present this novel type of pharmacology have been reported only in vertebrate retinal cells (Feigenspan et al. 1993; Qian and Dowling, 1993). Our work presents the first example of such responses in an invertebrate. We found that all neurones tested responded to GABA, and that the receptors investigated were remarkably homogeneous with respect to their ionic mechanism and pharmacological profile. Moreover, we did not detect any significant difference between the GABA receptors in neurones cultured from embryonic and from adult lobsters.

We have shown that GABA evoked, in lobster thoracic neurones in culture, an inward current accompanied by an increase in membrane conductance. The GABA-evoked current reversed close to the equilibrium potential for Cl⁻. When the intracellular concentration of Cl⁻ was modified by changing the pipette solution, the reversal potential of the GABA-induced response coincided with the equilibrium potential of Cl⁻ calculated from the Nernst equation. When the external chloride concentration was decreased, the change in reversal potential of the GABA-induced current was less than predicted by the Nernst equation. However, the difference between calculated and measured values may be due to changes in the internal Cl⁻ concentration resulting from a substantial resting Cl⁻ conductance. This has been reported for neurones of the stomatogastric nervous system (Marder and Paupardin-Tritsch, 1978), in which the resting Cl⁻ conductance can be 2.7 times the resting K⁺ conductance (Golowasch, 1990). Assuming a high Cl⁻ conductance, the E_Cl in cultured thoracic neurones would tend to equilibrate to the resting membrane potential or, in our experimental conditions, to the holding potential when the external chloride concentration is changed, resulting in a smaller than expected shift in reversal potential of the GABA response. Alternatively, active extrusion of chloride, which has been demonstrated for the crayfish stretch receptor (Aickin et al. 1982) and for a variety of neurones in the vertebrate central nervous system (Thompson et al. 1988), could also compensate partially for any strong transmembrane Cl⁻gradient in our preparation.

In crustacean nervous systems, the reversal potential for Cl⁻-mediated GABA responses may vary depending on the type of neurone. For instance, this potential is −70 mV in the abdominal stretch receptor cell of the crayfish (Kuffler and Eyzaguirre, 1955), about −50 mV in neurones of the stomatogastric ganglion (Marder and Paupardin-Tritsch, 1978) and −35 mV in crayfish primary afferent terminals of a leg proprioceptor (El Manira and Clarac, 1991; El Manira, 1992). In the latter case, as well as in crayfish primary afferent terminals of tail mechanoreceptors (Kennedy et al. 1980), the reversal potential for the Cl⁻-mediated GABA response is clearly more positive than the resting potential, and GABA provokes a depolarization of the terminal. Under our whole-cell recording conditions, both the reversal potential of the GABA response and the resting membrane potential depend on ion concentrations in the pipette, so the type of GABA response of an intact thoracic neurone cannot be determined. A GABA-induced depolarization, however, has been reported in vivo for leg motoneurones in crayfish thoracic ganglia (El Manira, 1992).

In invertebrate preparations, GABA can activate other conductances as well. In crustacean neuromuscular junction, it gates both a Cl⁻¹ and a K⁺ conductance (Fuchs and Getting, 1980), as it does in the stomatogastric nervous system (Marder and Paupardin-Trisch, 1978). In molluscan neurones, GABA receptors may control conductances for sodium, potassium and chloride (Yarowsky and Carpenter, 1978). In our experiments, changing the extracellular concentration of Na⁺, as well as blocking Ca²⁺ channels, had no effect on the reversal potential of the GABA response. Changing the extracellular concentration of K⁺ was also without effect in 50% of the tested cells. From these results, we conclude that, as in the vertebrate GABA_A and GABA_C receptors and in many invertebrate GABA receptors, the GABA-induced current described here is mediated by an increase in Cl⁻ conductance. In some of the thoracic cells, however, changing the extracellular concentration of K⁺ induced slight shifts in the reversal potential of the GABA current. One possible explanation for this would be the existence of a K⁺/Cl⁻ cotransporter, as reported, for instance, for CA3 pyramidal cells in organotypic hippocampal slice cultures, where it causes a hyperpolarizing shift in E_GABA when the external concentration of K⁺ is decreased (Thompson and Gähwiler, 1989). An alternative possibility is a partial involvement of K⁺ itself in the GABA response, assuming some heterogeneity in the GABA receptor type in thoracic neurones. However, preliminary results obtained with single-channel recordings did not indicate any such diversity in the ionic bases of the GABA response analyzed here (C. Jackel, W.-D. Krenz and F. Nagy, in preparation).

GABA receptors in the vertebrate nervous system are generally classified into two main families: GABA_A and GABA_B receptors. The former constitute a Cl⁻ channel thought to be associated directly with binding sites for GABA and with modulatory sites for benzodiazepines, barbiturates and steroids (for a review, see Grayson et al. 1991; Ticku, 1991). The GABA_B receptors, in contrast, regulate various conductances (K⁺ and Ca2+) via intracellular second-messenger systems (Bowery et al. 1991).

Pharmacological profiles of invertebrate GABA receptors do not fit well into this vertebrate scheme. Sensitivity for the vertebrate GABA_A agonists muscimol and isoguvacine has been demonstrated, with varying potency orders, in a variety of invertebrate preparations, such as insect somata (Robinson and Olson, 1988; Rauh et al. 1990; Sattelle, 1990), molluscan neurones (for a review, see Nistri and Constanti, 1979), Limulus polyphemus heart (Benson, 1989), crustacean primary afferents (El Manira and Clarac, 1991) and crustacean muscle (for a review, see Nistri and Constanti, 1979). However, insensitivity to bicuculline, the blocking effect of which is a characteristic property of the vertebrate GABA_A receptor, is common in invertebrate GABA-gated chloride channels. It has been reported, for example, for insect neuronal somata (Lees et al. 1987; Neumann et al. 1987; Benson, 1988; Satelle, 1990), for central nervous system neurones (Walker and Roberts, 1982) and for heart muscle (Benson, 1989) of Limulus polyphemus and in Ascaris suum muscle (Holden-Dye et al. 1988). Inhibition of GABA responses by bicuculline was shown in lobster (Shank et al. 1974; Constanti, 1978) and in crayfish muscle (Takeuchi and Onodera, 1972). Finally, picrotoxin (PTX), which is a potent inhibitor of GABA-mediated inhibition in vertebrates (for a review, see Simmonds, 1983), thought to act by blocking the GABA-dependent Cl⁻ channel (Takeuchi and Takeuchi, 1969), has variable effects depending on the invertebrate preparation under study (for a review, see Lunt, 1991). In crustaceans, for instance, blockage of the GABA response by PTX has been reported in lobster and crayfish muscle (Takeuchi and Takeuchi, 1969; Shank et al. 1974; Constanti, 1978) and in crayfish primary afferents (El Manira and Clarac, 1991). In the stomatogastric ganglion of the crab, however, PTX has no effect on GABA-induced currents and has been shown, instead, to block glutamate responses (Marder and Paupardin-Tritsch, 1978). Also, PTX is not active on the Limulus polyphemus heart GABA-receptor (Benson, 1989).

In the present study, the current induced by activation of the GABA receptors described was also blocked by up to 70% by PTX. Moreover, the lobster thoracic neurones investigated here were sensitive to the GABA_A agonists muscimol and isoguvacine. The dose–response relationships presented for the different agonists were based on the duration of the ejection pulse. Because the ejected volume is linearly related to the pulse duration (McCaman et al. 1977; Sakai et al. 1979), it was possible to demonstrate that the GABA-and agonist-induced currents were dose-dependent and that the Michaelis–Menten equation, with t₅₀ values replacing C₅₀ values, could be fitted to the data points to yield the familiar sigmoid dose–response curve. In this way, the relative potencies of the different ligands could be compared, although the absolute C₅₀ values were not available. In our preparation, the rank order of agonist potency was muscimol>GABA>isoguvacine. This is in accordance with most cases in both vertebrates (e.g. Nakagawa et al. 1991) and invertebrates (e.g. Benson, 1989), although in some preparations, such as in Ascaris suum muscle cells, GABA has the highest potency when applied to the receptor (Holden-Dye et al. 1988).

Besides its sensitivity to the chloride ionophore blocker PTX and to a number of GABA_A receptor agonists, the GABA receptor described here shows a pharmacological profile resembling that of a novel type of receptor preferentially expressed in vertebrate retina. This receptor was characterized in oocyte expression studies (Cutting et al. 1991; Polenzani et al. 1991; Shimada et al. 1992; Woodward et al. 1992) and was recently studied in situ (Feigenspan et al. 1993; Qian and Dowling, 1993). The retinal receptor shows unusual pharmacological properties as proposed earlier for GABA_C receptors (Drew et al. 1984; Johnston, 1986) and shares these with the lobster receptor described in the present paper.

First, the current induced by GABA, muscimol and isoguvacine was entirely unaffected by application of the GABA_A antagonists bicuculline, bicuculline methiodide and the synthetic SR 95531. Second, receptors in the thoracic neurones were totally unresponsive to GABA_B agonists such as baclofen and 3-APA, and the GABA_B antagonist phaclofen had no effect on the GABA-evoked current. Note that, for these experiments, we supplemented the intracellular solution with sodium GTP since the action of baclofen is mediated via interactions with GTP-binding proteins (Bowery et al. 1991). Third, The GABA-evoked chloride currents in the thoracic neurones are not modulated by benzodiazepines and barbiturates. Fourth, receptors of the lobster thoracic neurones are sensitive to CACA, a folded GABA analogue with restricted conformation, which is thought to activate specifically the GABA_C receptor (Johnston et al. 1975). CACA produced membrane responses that inverted at the same potential as GABA-evoked currents and were reduced to the same extent by PTX, suggesting that CACA, like GABA, increases a Cl⁻ conductance. CACA yields a dose–response curve with a T₅₀ value about ten times longer and a lower maximal response than those of GABA, indicating a lower affinity of CACA for the receptor. This is consistent not only with properties of the GABA responses recorded in retinal neurones, where GABA was reported to be about ten times as potent as CACA (Feigenspan et al. 1993; Qian and Dowling, 1993) but also with those of the GABA responses recorded extracellularly from neurones of the frog optic tectum (Sivilotti and Nistri, 1989). Fifth, both the GABA-and the CACA-induced responses described in the present paper show only moderate desensitization, a feature also encountered in the vertebrate retina.

Thoracic ganglia are made up of different types of interneurones and motoneurones, and the dissociated thoracic neurones in our preparation are necessarily a heterogeneous population. Their GABA receptors, however, appeared remarkably similar with respect to their ionic mechanism and pharmacological profile in all neurones tested, suggesting strongly the existence of only one type of GABA receptor in lobster thoracic neurones. This hypothesis is reinforced by preliminary results obtained with single-channel recordings indicating that bicuculline-resistant GABA and CACA responses are mediated by the same Cl⁻channel (C. Jackel, W.-D. Krenz and F. Nagy, in preparation). This is in strong contrast with the vertebrate retina, where both bicuculline-resistant and bicuculline-sensitive receptors are present in different neurones (Qian and Dowling, 1993) or in the same neurones associated with two populations of channels with different unitary conductances (Feigenspan et al. 1993). This raises the question of how widespread the bicuculline-resistant GABA_C-like receptors are in crustaceans. To our knowledge, sensitivity to CACA has not been investigated so far in any other crustacean preparation. CACA has been tested only in a few invertebrate excitable cells, such as the Limulus polyphemus heart (Benson, 1989) and Ascaris suum muscle (Holden-Dye et al. 1988), where it was shown to be inactive.

At present, the functional significance of a GABA receptor with the characteristics described in this study is a matter of speculation. Weak and slow desensitization may enable sustained inhibition on continued release of GABA. The fact that we did not detect any significant difference between the GABA responses in neurones from embryonic and adult lobsters could also indicate the involvement of an embryonic type of receptor. This receptor might be expressed in adult neurones after isolation in culture, in a way similar to the appearance of the embryonic nicotinic acetylcholine receptor in adult vertebrate skeletal muscle following denervation (Mishina et al. 1986; Witzemann et al. 1991).

This work was supported by CNRS grant 1126. We would like to thank Professor M. Moulins for support and helpful discussions and for careful reading of the manuscript.