Excessive extracellular accumulation of the excitatory neurotransmitter, glutamate, and subsequent excitotoxic neuronal death have been implicated in the pathogenesis of hypoxia-ischemic neuronal death (Benveniste et al., 1984; Simon et al., 1984; Meldrum, 1985; Choi and Rothman, 1990; Albers et al., 1992). More recently, attention has also focused on the possibility that the inhibitory neurotransmitter, 
-aminobutyric acid (GABA), may serve as a brake on overexcitation and thus reduce the pathogenesis of ischemic brain damage. Extracellular GABA levels are also increased during hypoxia-ischemia in vivo (Hagberg et al., 1985; Globus et al., 1988). The idea that this increase may counteract the toxic effects of glutamate led Globus et al. (1991) to propose that an "excitotoxicity index," calculated as [glutamate] 
[glycine]/[GABA] in the extracellular space, might better predict ischemic brain damage than glutamate levels alone. GABA inhibition is mediated both by GABAA receptors, which open membrane chloride channels and stabilize the membrane potential below firing threshold, and GABAB receptors, which act via G proteins to reduce transmitter release from presynaptic terminals (Kaila, 1994).
Recent studies from several laboratories have examined enhancement of GABA receptor activation as a strategy for reducing brain damage in animal models of ischemia. Several such strategies reduce hippocampal CA1 neuronal loss following global ischemia in gerbils: (a) preischemic administration of muscimol, baclofen, or diazepam (Sternau et al. 1989); (b) pre- or postischemic administration of chlormethiazole (Cross et al., 1991); (c) pre- and postischemic intraventricular infusion of a GABA-transaminase inhibitor or muscimol (Shuaib et al., 1992, 1993; see also Abel and McCandless, 1992 which utilized preischemic administration); and (d) post-ischemic administration of diazepam (Schwartz et al., 1994). In rats, pre- and postischemic administration of GABA uptake inhibitor or diazepam (Johansen and Diemer, 1991) or postischemic administration of diazepam (Schwartz et al., 1995) reduced hippocampal CA1 cell loss following global ischemia. Neurological function following multifocal ischemia in rats was improved by postischemic administration of muscimol (Lyden and Hedges, 1992; Lyden and Lonzo, 1994), and chlormethiazole pretreatment reduced infarct volume in a rat model of transient focal ischemia (Sydserff et al., 1995). Protective effects of barbiturates have also been observed in both focal and global ischemia models (see review by Spetzler and Hadley, 1989), although the multiplicity of known barbiturate actions hinders interpretation of these observations strictly in context of GABAA receptors.
In contrast to the consistent neuroprotective effects of GABAmimetic drugs in animal models of ischemia, both protective and injurious effects have been observed in studies of excitotoxicity in vitro. In our murine cortical cell cultures, neither GABA nor diazepam reduced the rapidly triggered excitotoxic neuronal death induced by a brief intense glutamate exposure (Koh and Choi, 1987b). Using a more prolonged and weaker N-methyl-D-aspartate (NMDA) exposure, Okuma et al. (1994) found that muscimol reduced NMDA neurotoxicity in cortical cultures. However, Erdo et al. (1991) observed that 1 mM GABA or 10–100 
M 1,2,3,4-tetrahydroisoxasolopyridinol (THIP), a GABAA agonist, accelerated neuronal death in rat cortical cultures exposed continuously to NMDA or kainate.
While this last observation could simply reflect distortions induced by cell culture methods, it is also possible that cell culture isolates an important deleterious effect of GABAA receptor activation that is masked in vivo by more powerful beneficial effects. Thus, we set out to characterize in detail the effects of both GABAA and GABAB receptor agonists on the neuronal death induced by excitotoxins or oxygen-glucose deprivation in cortical cell cultures. Abstracts have appeared elsewhere (Monyer et al., 1990; Muir et al., 1994; Muir and Choi, 1995).
MATERIALS AND METHODS
Cell culture
Primary cortical cell cultures containing both neuronal and glial elements were prepared from fetal mice at 15–16 days of gestation as previously described (Rose et al., 1993). Dissociated cortical cells were plated at approximately 2.75 hemispheres per 24-multiwell (15-mm) vessels (Primaria; Falcon) on a previously established glial bed (see below) in modified Eagle's minimal essential medium (MEM, Earle's salts, supplied glutamine-free) supplemented with 5% heat-inactivated horse serum, 5% fetal calf serum, glutamine (2 mM). and glucose (total 25 mM). Cultures were maintained at 37°C in a humidified incubator containing 5% CO2. Nonneuronal cell division was halted after 5–7 days in vitro by a 1- to 3-day exposure to 10 M cytosine arabinosine. Cultures were subsequently fed twice weekly with a medium identical to the plating medium except lacking fetal serum, and used for experiments between 13 and 16 days in vitro. Glial cultures were prepared similarly from 1- to 3-day-old mice neocortices plated at 0.5–1 hemisphere per 24-multiwell vessel in a plating medium similar to that described above, but containing 10% heat-inactivated horse serum, 10% fetal calf serum, and 10 ng/ml epidermal growth factor.
Excitatory amino acid exposure
Rapidly triggered excitotoxicity induced by a 5-min exposure to 100 M NMDA was carried out at room temperature in a HEPES-buffered salt solution containing (millimolar): 120 NaCl, 5.4 KCl, 0.8 MgSO4, 1.8 CaCl2, 20 HEPES, 15 glucose, and 0.01 glycine (pH 7.4). GABA or related drugs were applied concurrently with NMDA exposure. Cultures were then returned to MEM containing 10 M glycine. Slowly triggered excitotoxicity was induced by a 24-h exposure to NMDA (15 M), 
-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA, 10 M), or kainate (35–40 M) in MEM supplemented with 10 M glycine at 37°C. MK-801 (10 M) was included with AMPA or kainate to block secondary activation of NMDA receptors.
Oxygen-glucose deprivation
Cultures were placed in an anaerobic chamber (Forma Scientific, Marietta, OH, U.S.A.) containing a gas mixture of 5% CO2, 10% H2, 85% N2 (<0.2% O2) (Goldberg and Choi, 1993). Culture medium was replaced by thorough exchange with a deoxygenated, glucose-free balanced salt solution (BSS) containing (millimolar): 116 NaCl, 5.4 KCl, 0.8 MgSO4, 1 NaH2PO4, 26.2 NaHCO3, 1.8 CaCl2, and 0.01 glycine. Cultures were then placed in a 37°C humidified incubator within the chamber. Following 40–45 min oxygen-glucose deprivation, the exposure medium was exchanged with oxygenated MEM containing glucose and cultures returned to a normoxic incubator. Specified drugs were added during the period of oxygen-glucose deprivation only.
Cyanide exposure
Cultures were exposed to 1 mM potassium cyanide in oxygenated, glucose-free BSS at 37°C. Exposure was terminated by washout with oxygenated MEM containing glucose and cells returned to an incubator for 20–24 h. Specified drugs were added during the period of cyanide exposure only.
Assessment of neuronal injury
Neuronal cell death was estimated by examination of cultures under phase-contrast microscopy and quantitated by measurement of lactate dehydrogenase (LDH) released by damaged or destroyed cells in the bathing media 20–24 h after insult initiation (Koh and Choi, 1987a). LDH-mediated conversion of pyruvate to lactate was monitored by measuring levels of reduced nicotinamide adenine dinucleotide (NADH), by absorbance at 340 nm, using a spectrophotometric microplate reader (UVmax, Molecular Devices, Menlo Park, CA, U.S.A.). LDH was converted to units per milliliter by fitting the rate of change of absorbance of each sample to a LDH standard curve (Sigma Enzyme Control 2-E). The LDH signal corresponding to near-complete neuronal death without glial death was measured in sister cultures exposed to 300 M NMDA for 24 h (total neuronal LDH). Since absolute total LDH levels may vary among plates obtained from different dissections, LDH values were normalized to the levels associated with control injury in sister cultures. Basal LDH levels determined in sister cultures subjected to sham wash (generally <15% of total LDH) were subtracted from values obtained in experiments to yield the signal specific to experimental injury.
45Ca2+ accumulation
Cultures were exposed to NMDA for 5 min or deprived of oxygen and glucose as described above with the inclusion of specified drugs and 45Ca2+ (New England Nuclear, Wilmington, DE, U.S.A.; final activity, 2 Ci/ml) in the exposure medium (Hartley et al., 1993; Goldberg and Choi, 1993). Exposure was terminated by thoroughly washing cells of extracellular 45Ca2+. Subsequently, the extracellular medium was removed, cells were lysed with 0.2% sodium dodecyl sulfate, and an aliquot was added to scintillation fluid for counting. Since absolute neuronal density may vary among plates obtained from different dissections, experimental values were normalized to the levels associated with control injury (= 100) in sister cultures. Previous study has determined that 45Ca2+ accumulation is mostly neuronal (Hartley et al., 1993).
High-performance liquid chromatography (HPLC)
Samples of the bathing medium collected prior to the termination of oxygen-glucose deprivation were assayed for glutamate by using phenylisothiocyanate (PITC) derivatization, HPLC separation using a Hypersil-ODS reverse-phase column (Hewlett Packard, Wilmington, DE, U.S.A.), and ultraviolet detection at a wavelength of 254 nm (Cohen et al., 1986). A 200 l of sample medium was derivatized with 100 l of PITC, methanol, and triethylamine (2:7:4) and dried under vacuum. The samples were then reconstituted in solvent consisting of 0.14 M sodium acetate, 0.05% triethylamine, 6% acetonitrile, and brought to pH 6.4 with glacial acetic acid, prior to the HPLC run. The above solvent was used as the mobile phase with the column being washed with 60% acetonitrile/40% water between each sample run. Media glutamate concentrations were calculated by normalizing to glutamate standards. Glutamate measurements were found to be linear over the range 0.1–10 M, with 0.1 M being the threshold for reliable detection.
Reagents
Culture medium was purchased from Gibco (Grand Island, NY, U.S.A.) as a 10x concentrated stock lacking bicarbonate and glutamine; serum was obtained from Hyclone Laboratories Inc. (Logan, UT, U.S.A.). NMDA, GABA, muscimol hydrobromide, picrotoxin, and phenylisothiocyanate were obtained from Sigma Chemical (St. Louis, MO, U.S.A.). (R)-(+)-Baclofen hydrochloride, guvacine hydrochloride, bicuculline methiodide, nipecotic acid, and MK-801 were purchased from Research Biochemicals Inc (Natick, MA, U.S.A.).
RESULTS
Rapidly triggered excitotoxicity
Consistent with an earlier study (Choi et al., 1988), exposure of mixed neuron/glial cortical cell cultures to 100 M NMDA for 5 min resulted in degeneration of a majority (50–75%) of the neuronal population 20–24 h later. Concurrent application of either 100–1000 M GABA (Fig. 1A), or 3–100 M muscimol (Fig. 1B), a specific GABAA receptor agonist, produced a partial, concentration-dependent reduction in NMDA-induced neuronal death. In contrast, the specific GABAB receptor agonist, baclofen, had no effect on NMDA-induced neuronal death over the tested concentration range of 1–100 M (Figs. 1C). Application of 1 mM GABA, muscimol, or baclofen alone for 24 h was not toxic.
Figure 1.
GABAA receptor agonists attenuate NMDA-mediated excitotoxicity. Cultures were exposed to 100 M NMDA for 5 min in the absence or presence of increasing doses of GABA agonists. LDH was measured 20–24 h later. Data represent mean LDH 
SD scaled to the mean LDH measured in cultures exposed to 100 M NMDA alone (=100). A: GABA, n = 8 cultures per condition pooled from two experiments; control NMDA injury = 68 5% of the total neuronal population. B: Muscimol, n = 8–20 cultures per condition pooled from five experiments; control NMDA injury = 59 19% of the total neuronal population. C: Baclofen, n = 16 cultures per condition pooled from four experiments; control NMDA injury = 68 19% of the total neuronal population. An asterisk indicates values significantly different from NMDA alone by ANOVA followed by Dunnett's t test (p < 0.05).
NMDA-stimulated uptake of 45Ca2+, which correlates with subsequent neuronal death (Hartley et al., 1993), was partially reduced by 10–100 M muscimol (Table 1). 45Ca2+ accumulation was completely blocked by the NMDA antagonist, MK-801 (10 M).
To demonstrate that muscimol's protective action was mediated by GABAA receptor activation, specific antagonists for the receptor were utilized. The competitive GABAA receptor antagonist, bicuculline (10–100 M), antagonized the protective effect of muscimol on NMDA-induced neuronal death (Fig. 2A). A similar antagonism was demonstrated with the noncompetitive GABAA receptor antagonist, picrotoxin (10–100 M, Fig. 2B). Neither antagonist increased NMDA-induced neuronal death when applied alone (data not shown).
Figure 2.
GABAA receptor antagonists block muscimol's protective action on NMDA excitotoxicity. Cultures were exposed to 100 M NMDA for 5 min in the absence or presence of the indicated drugs. LDH was measured 20–24 h later. Data represent mean LDH SD scaled to the mean LDH measured in cultures exposed to 100 M NMDA alone (=100). A: Antagonism by bicuculline, n = 20 cultures per condition pooled from five experiments; control NDMA injury = 62 19% of the total neuronal population. B: Antagonism by picrotoxin, n = 16 cultures per condition pooled from four experiments; control NMDA injury = 48 12% of the total neuronal population. Data were analyzed by ANOVA followed by Student-Newman-Keuls multiple comparisons test. An asterisk indicates values significantly different (p < 0.05) from NMDA alone; a pound sign indicates values significantly different from NMDA plus muscimol.
Slowly triggered excitotoxicity
In addition to reducing rapidly triggered excitotoxicity, muscimol also reduced the slowly triggered excitotoxic neuronal loss induced by a 24-h exposure to 15 M NMDA, 10 M AMPA, or 35–40 M kainate (Table 2). A concentration-dependent reduction in neuronal death was observed with each agonist, although the protective effect against AMPA- or kainate-induced death was weaker than that against NMDA-induced death (Table 2, Fig. 1B). Reduction of NMDA-induced death was also observed with 1 mM guvacine or nipecotic acid, inhibitors of GABA uptake, but not against AMPA or kainate injury (Table 2). Addition of baclofen did not affect NMDA-, AMPA-, or kainate-induced neuronal death (data not shown).
Table 2 - Effect of muscimol or GABA uptake inhibitors on slowly triggered excitotoxicity by NMDA, kainate, or AMPA.
Full table Oxygen-glucose deprivation
Consistent with earlier studies (Goldberg and Choi, 1993; Koh et al., 1995), a 40- to 45-min period of oxygen and glucose deprivation produced a low level (10–30%) of neuronal death 20–24 h later. In striking contrast to the protective effect observed for excitotoxicity, addition of 10–100 M GABA or 1–100 M muscimol (but not baclofen) during this oxygen-glucose deprivation period produced a concentration-dependent increase in immediate neuronal swelling and an exacerbation of neuronal death 20–24 h later (Fig. 3). With 100 M GABA or muscimol, the same 40- to 45-min period of oxygen-glucose deprivation induced the death of a majority of neurons. This GABA- or muscimol-induced potentiation of neuronal death after oxygen-glucose deprivation was blocked by 10 M MK-801 (Fig. 3). An antagonist of AMPA/kainate receptors, NBQX (30 M) had no effect.
Figure 3.
GABAA receptor agonists potentiate neuronal injury induced by oxygen-glucose deprivation. Cultures were deprived of oxygen and glucose for 40–45 min in the absence or presence of the indicated drugs. LDH was measured 20–24 h later. Data represent mean LDH SD expressed as percentage of total neuronal LDH. A: GABA, n = 3–16 cultures per condition pooled from four experiments. B: Muscimol, n = 13–24 cultures per condition from six experiments. C: Baclofen, n = 10–14 cultures per condition from four experiments. Data were analyzed by ANOVA followed by Student-Newman-Keuls multiple comparisons test. An asterisk indicates values significantly different (p < 0.05) from 40- to 45-minute deprivation alone; a pound sign represents a significant decrease of the GABAA receptor agonist-induced potentiation.
Addition of 1–100 M muscimol also markedly increased accumulation of glutamate in the extracellular medium, and intracellular 45Ca2+ uptake, measured immediately after a 40- to 45-min period of oxygen-glucose deprivation (Fig. 4). Muscimol itself did not alter intracellular 45Ca2+ uptake in sham-washed control cells. MK-801 at 10 M attenuated 45Ca2+ accumulation but did not significantly decrease accumulation of extracellular glutamate (Fig. 4). Bicuculline (1–100 M) only weakly reduced the muscimol-induced potentiation of neuronal death after oxygen-glucose deprivation (Fig. 5A), probably because it alone produced a concentration-dependent, MK-801-sensitive, increase in this neuronal death (Fig. 5B). Similar results were obtained using picrotoxin (data not shown).
Figure 4.
Effect of muscimol on glutamate release and 45Ca2+ accumulation induced by oxygen-glucose deprivation. Cultures were deprived of oxygen and glucose for 40–45 min in the presence of the indicated concentration of muscimol. A: Data represent mean SD glutamate concentration in the bathing medium at the end of oxygen-glucose deprivation, n = 8–12 cultures per condition pooled from three experiments. Note that bath glutamate concentrations are much lower than observed in vivo (see for example, Benveniste et al., 1984) due to the larger dilution volume; however, local concentrations in synaptic regions are likely much higher. B: Data represent mean cpm SD scaled to the mean cpm measured in the condition without muscimol (=100); n = 10–18 cultures per condition pooled from five experiments. Note that 45Ca2+ accumulation in the condition without muscimol was very low, similar to that found in cultures exposed to sham-wash alone. Data were analyzed by ANOVA followed by Student-Newman-Keuls multiple test. An asterisk indicates values significantly different (p < 0.05) from 40- to 45-min deprivation alone; a pound sign represents a significant decrease of the muscimol-induced potentiation.
Figure 5.
Effect of bicuculline on neuronal injury induced by oxygen-glucose deprivation in the presence and absence of muscimol. Cultures were deprived of oxygen and glucose for 40–45 min in the absence or presence of the indicated drugs. LDH was measured 20–24 h later. Data represent mean LDH SD expressed as percentage of total LDH. A: Effect of bicuculline on muscimol (10 M) potentiation of neuronal injury, n = 8–11 cultures per condition from three experiments. B: Effect of bicuculline alone, line alone, n = 16 cultures per condition pooled from four experiments. Data were analyzed by ANOVA followed by Student-Newman-Keuls multiple comparisons test. An asterisk indicates values significantly greater (p < 0.05) than oxygen-glucose deprivation alone; a pound sign indicates values significantly different from (A) muscimol-induced potentiation of oxygen-glucose deprivation or (B) bicuculline-induced potentiation of oxygen-glucose deprivation.
Cyanide exposure
As an additional neuronal injury paradigm triggered by energy depletion, cultures were exposed to cyanide. GABA or muscimol addition also potentiated the neuronal death induced by energy-depletion due to a 40- to 45-min exposure to 1 mM potassium cyanide in glucose-free media (Fig. 6).
Figure 6.
GABAA receptor agonists potentiate neuronal injury induced by cyanide exposure. Cultures were exposed to potassium cyanide for 40–45 min in the absence or presence of the indicated drugs. LDH was measured 20–24 h later. Data represent mean LDH SD expressed as percentage of total LDH; n = 16 cultures per condition pooled from four experiments. An asterisk indicates values significantly greater from cyanide exposure alone as determined by ANOVA followed by Dunnett's t test (p < 0.05).
DISCUSSION
Present data document an unexpected contrast between the protective effect of GABAA agonists on the cortical neuronal death induced by exogenous NMDA addition and the exacerbating effect of the same drugs on the neuronal death induced by either oxygen-glucose deprivation or cyanide exposure. The latter is unexpected for two reasons: (a) there is convincing evidence that GABAmimetic drugs are neuroprotective in several animal models of global or focal ischemia (see above); and (b) prior study in our system has indicated that the death induced by short periods of oxygen-glucose deprivation (such as used here) is mostly mediated by NMDA receptors (Goldberg and Choi, 1993).
In earlier studies, neither we (Koh and Choi, 1987b; Monyer et al., 1990) nor Erdo and Michler (1990) observed a neuroprotective effect of GABA agonists against maximal levels of rapidly triggered, NMDA receptor-mediated excitotoxicity. Detection of a partial protective effect of GABA agonists in the present study probably reflects employment of lower NMDA concentrations, which produced submaximal neuronal death. Our observations are consistent with the finding by Ohkuma et al. (1994) that muscimol reduced the neuronal injury of cultured cortical neurons induced by low (10 M) concentrations of NMDA. Taken together, available data suggest that GABAA receptor stimulation produces a partial protective effect against submaximal NMDA receptor-mediated toxicity that can be overridden by intense NMDA receptor activation.
The protective effect of muscimol and GABA versus excitatory amino acid toxicity may be due to the ability of GABAA receptor activation to drive the membrane potential towards ECl, which is usually near the resting potential. This should increase the voltage-dependent Mg2+ block of the NMDA receptor-gated channel (Mayer et al., 1984; Nowak et al., 1984) and thereby reduce NMDA receptor-mediated toxic Ca2+ entry (Fig. 7). In addition, membrane hyperpolarization can be expected to reduce Ca2+ influx mediated via the Na+-Ca2+ exchanger or mediated via voltage-gated Ca2+ channels (Choi, 1988). Consistent with the idea that GABAA receptor activation reduces NMDA-induced net neuronal Ca2+ influx, it decreased cellular accumulation of extracellular 45Ca2+ induced by a 5-min exposure to 200 M NMDA. Reduction of depolarization-induced Ca2+ influx may also explain the ability of GABAA receptor activation to attenuate the slowly triggered excitotoxic neuronal death induced by a 24-h exposure to NMDA, AMPA or kainate.
Figure 7.
Proposed effect of GABAA receptor activation on NMDA receptor-mediated current (I) in energy-competent and energy-depleted neurons. This graph depicts a typical NMDA receptor I/V curve, exhibiting voltage-dependent Mg2+ block (modified from MacDermott and Dale, 1987). In energy-competent neurons (only partially depolarized by exposure to NMDA), the hyperpolarizing effect of GABAA receptor activation may move the membrane potential from point 1 to point 2, thereby strengthening the Mg2+ block. However, in energy-depleted neurons, the same GABAA receptor activation may move membrane potential from point 3 to point 4, producing no effect on the Mg2+ block, but still increasing the inward driving force for Ca2+.
Full figure and legend (53K)In slowly triggered excitotoxicity paradigms, addition of either muscimol or drugs that inhibit GABA uptake produced a greater protective effect on NMDA-induced neuronal death than on AMPA- or kainate-induced death. Perhaps this is because the effect of membrane hyperpolarization on the Mg2+ block of NMDA receptor-gated channels is not shared with AMPA or kainate receptor-gated channels. The observed lack of protective effect of GABA uptake inhibitors against AMPA or kainate toxicity likely reflects lower levels of GABAA receptor stimulation achieved via this route (compared to that achieved during NMDA exposure), perhaps because of less GABA release.
Further study will be required to elucidate the mechanisms underlying the unexpected potentiating effect of GABAA receptor activation on oxygen-glucose deprivation-induced neuronal death. Data presented here indicate that this potentiated death is associated with enhanced Ca2+ accumulation and is still mediated by NMDA receptors. One intriguing possible explanation can be developed from the idea that membrane hyperpolarization has opposing effects on NMDA receptor-mediated Ca2+ influx: increasing the Mg2+ block of the NMDA receptor-gated channel, but also increasing the electrochemical gradient favoring Ca2+ entry. Thus, in energy-competent neurons with hyperpolarized membrane potentials (e.g., neurons exposed to exogenously added NMDA), GABAA receptor activation may move the membrane potential towards ECl enough that the Mg2+ block is enhanced and Ca2+ influx through the NMDA receptor-gated channel falls (Fig. 7). However, in energy-depleted neurons with collapsed membrane potentials (e.g., neurons exposed to oxygen-glucose deprivation or cyanide; see Hansen, 1985), GABAA receptor activation may hyperpolarize neuronal membranes enough to increase the inward Ca2+ driving force, but not enough to bring in the Mg2+ block (Fig. 7). Increased neuronal Ca2+ entry may also lead to greater glutamate release and help explain the observed increase in extracellular glutamate accumulation observed in cultures exposed to oxygen-glucose deprivation in the presence of muscimol. Increased extracellular glutamate itself would be expected to increase neuronal death.
Other mechanisms can also be considered. For instance, although GABAA receptor stimulation usually inhibits central neuronal firing, it can, under some circumstances, be excitatory (Michelson and Wong, 1991; Kaila, 1994). In particular, while normally inhibitory, dendritic GABAA receptors can become excitatory during intense activation due to an activity-dependent shift in the GABAA reversal potential away from ECl and toward EHCO3 (Staley et al., 1995). Furthermore, it is conceivable that prolonged exogenous application of GABA or muscimol desensitizes GABAA receptors sufficiently that a paradoxical loss of inhibitory tone results. The apparent voltage-dependence of this desensitization (Frosch et al., 1992) raises the possibility that desensitization might be greater when the agonists are applied prior to membrane depolarization (as would be the case in the oxygen-glucose deprivation and cyanide paradigms), compared with when the agonists are applied together with a depolarizing stimulus (as would be the case in the excitotoxicity paradigms).
The weak antagonism of bicuculline or picrotoxin on the injury-potentiating effect of muscimol in cultures deprived of oxygen and glucose is well explained by the observed injury-potentiating effect exhibited by these GABAA antagonist drugs when applied alone. This intrinsic injury-potentiating effect may reflect an effect of GABAA receptor antagonists to increase circuit excitability, enhancing ATP depletion and glutamate release. In exogenous excitotoxicity induced by addition of NMDA, endogenous glutamate release is likely swamped by exogenously added NMDA.
In contrast to the observed modulatory effects of GABAA receptor activation upon cortical neuronal death induced by exposure to excitotoxins or oxygen-gluocse deprivation, GABAB receptor activation with baclofen lacked effect. The picture appears different in CNS white matter, which has few glutamate receptors and is not vulnerable to excitotoxic injury. Fern et al. (1995) recently reported that anoxia-induced injury of the rat optic nerve was attenuated by GABA or baclofen, but not GABAA receptor agonists.
The present in vitro experiments should not be viewed as conflicting with existing in vivo data indicating that GABAA agonists can reduce ischemic brain injury. The final weighting of different variables can only be accomplished in vivo, and nothing demonstrated here argues against a final predominance of beneficial effects. However, present results provide novel isolation of a potential deleterious effect of GABAA receptor activation upon energy-depleted neurons. Strategies aimed at eliminating this deleterious action-for example, addition of NMDA receptor blockade (see for example, Lyden and Lonzo, 1994)-might further enhance the neuroprotective effects of GABAA agonists in the ischemic brain.
References
| 1. | Abel MS & McCandless DW. (1992) Elevated  -aminobutyric acid levels attenuate the metabolic response to bilateral ischemia. J Neurochem 58: 740−744. |
| 2. | Albers GW, Goldberg MP & Choi DW. (1992) Do NMDA antagonists prevent neuronal injury? Yes. Arch Neurol 49: 418−420. |
| 3. | Benveniste H, Drejer J, Schousboe A & Diemer NH. (1984) Elevation of the extracellular concentrations of glutamate and aspartate in rat hippocampus during transient cerebral ischemia monitored by intracerebral microdialysis. J Neurochem 43: 1,369−1,374. |
| 4. | Choi DW. (1988) Calcium-mediated neurotoxicity: relationship to specific channel types and role in ischemia damage. TINS 11: 465−469. |
| 5. | Choi DW & Rothman SM. (1990) The role of glutamate neurotoxicity in hypoxic-ischemic neuronal death. Annu Rev Neurosci 13: 171−182. |
| 6. | Choi DW, Koh J & Peters S. (1988) Pharmacology of glutamate neurotoxicity in cortical cell culture: attenuation by NMDA antagonists. J Neurosci 8: 185−196. |
| 7. | Cohen SA, Bidlingmeyer BA & Tarvin TL. (1986) PITC derivatives in amino acid analysis. Nature 320: 769−770. |
| 8. | Cross AJ, Jones JA, Baldwin HA & Green AR. (1991) Neuroprotective activity of chlormethiazole following transient forebrain ischaemia in the gerbil. Br J Pharmacol 104: 406−411. |
| 9. | Erdo SL & Michler A. (1990) GABA does not protect cerebro-cortical cultures against excitotoxic cell death. Eur J Pharmacol 1982: 203−206. |
| 10. | Erdo SL, Michler A & Wolff JR. (1991) GABA accelerates excitotoxic cell death in cortical cultures: protection by blockers of GABA-gated chloride channel. Brain Res 542: 254−258. |
| 11. | Fern R, Waxman SG & Ransom BR. (1995) Endogenous GABA attenuates CNS white matter dysfunction following anoxia. J Neurosci 15: 699−708. |
| 12. | Frosch MP, Lipton SA & Dichter MA. (1992) Desensitization of GABA-activated currents and channels in cultured cortical neurons. J Neurosci 12: 3,042−3,053. |
| 13. | Globus MY-T, Busto R, Dietrich WD, Martinez E, Valdes I & Ginsberg MD. (1988) Intra-ischemic extracellular release of dopamine and glutamate is associated with striatal vulnerability to ischemia. Neurosci Lett 1: 36−40. |
| 14. | Globus MY-T, Ginsberg MD & Busto R. (1991) Excitotoxic index-a biochemical marker of selective vulnerability. Neurosci Lett 127: 39−42. |
| 15. | Goldberg MP & Choi DW. (1993) Combined oxygen and glucose deprivation in cortical cell culture: calcium-dependent and calcium-independent mechanisms of neuronal injury. J Neurosci 13: 3,510−3,524. |
| 16. | Hagberg H, Lehmann A, Sandberg M, Nystrom B, Jacobson I & Hamberger A. (1985) Ischemia-induced shift of inhibitory and excitatory amino acids from intra- to extracellular compartments. J Cereb Blood Flow Metab 5: 413−419. |
| 17. | Hansen AJ. (1985) Effect of anoxia on ion distribution in the brain. Physiol Rev 65: 101−148. |
| 18. | Hartley DM, Kurth MC, Bjerkness L, Weiss JH & Choi DW. (1993) Glutamate receptor-induced 45Ca2+ accumulation in cortical cell culture correlates with subsequent neuronal degeneration. J Nuerosci 13: 1,993−2,000. |
| 19. | Johansen FF & Diemer NH. (1991) Enhancement of GABA neurotransmission after cerebral ischemia in the rat reduces loss of hippocampal CA1 pyramidal cells. Acta Neurol Scand 84: 1−6. |
| 20. | Kaila K. (1994) Ionic basis of GABAA receptor channel function in the nervous system. Prog Neurobiol 42: 489−537. |
| 21. | Koh J & Choi DW. (1987a) Quantitative determination of glutamate mediated cortical neuronal injury in cell culture by lactate dehydrogenase efflux assay. J Neurosci Methods 20: 83−90. |
| 22. | Koh JY & Choi DW. (1987b) Effect of anticonvulsant drugs on glutamate neurotoxicity in cortical cell culture. Neurology 37: 319−322. |
| 23. | Koh JY, Gwag BJ, Lobner D & Choi DW. (1995) Potentiated necrosis of cultured cortical neurons by neurotrophins. Science 258: 573−575. |
| 24. | Lyden PD & Hedges B. (1992) Protective effect of synaptic inhibition during cerebral ischemia in rats and rabbits. Sroke 23: 1,463−1,470. |
| 25. | Lyden PD & Lonzo L. (1994) Combination therapy protects ischemic brain in rats: a glutamate antagonist plus a -aminobutyric acid agonist. Stroke 25: 189−195. |
| 26. | Mayer ML, Westbrook GL & Guthrie PB. (1984) Voltage-dependent block by Mg++ of NMDA responses in spinal cord neurones. Nature 309: 261−263. |
| 27. | MacDermott AB & Dale N. (1987) Receptors, ion channels and synaptic potentials underlying the integrative actions of excitatory amino acids. TINS 10: 280−284. |
| 28. | Meldrum B. (1985) Possible therapeutic applications of antagonists of excitatory amino acid neurotransmitters. Clin Sci 68: 113−122. |
| 29. | Michelson HB & Wong RKS. (1991) Excitatory synaptic responses mediated by GABAA receptors in the hippocampus. Science 253: 1,420−1,423. |
| 30. | Monyer H, Hartley DM, Ehsani H, Seeberg PH & Choi DW. (1990) Muscimol attenuates slow excitatory amino acid-induced injury of cultured cortical neurons. Soc Neurosci Abstr 16: 288. |
| 31. | Muir JK & Choi DW. (1995) Modulation of NMDA-mediated excitotoxicity by GABA uptake inhibitors or benzodiazepines in cortical cell culture. Soc Neurosci Abstr 21: 994. |
| 32. | Muir JK, Lobner D & Choi DW. (1994) Muscimol attenuates excitotoxicity, but exacerbates oxygen-glucose deprivation neuronal injury in cortical cell cultures. Soc Neurosci Abstr 20: 182. |
| 33. | Nowak L, Bregestovski P & Ascher P. (1984) Magnesium gates glutamate-activated channels in mouse central neurons. Nature 307: 462−465. |
| 34. | Ohkuma S, Chen SH, Katsura M, Chen DZ & Kuriyama K. (1994) Muscimol prevents neuronal injury induced by NMDA. Jpn J Pharmacol 64: 125−128. |
| 35. | Rose K, Goldberg MP & Choi DW. 1993 Cytotoxicity in murine neocortical cell culture. (In) In Vitro Biological Methods in Toxicology (Tyson CA, Frazier JM, eds) San Diego, Academic Press (pp) 46−60. |
| 36. | Schwartz RD, Huff RA, Yu X, Carter ML & Bishop M. (1994) Postischemic diazepam is neuroprotective in the gerbil hippocampus. Brain Res 647: 153−160. |
| 37. | Schwartz RD, Yu X, Katzman MR, Hayden-Hixson DM & Perry JM. (1995) Diazepam, given postischemia, protects selectively vulnerable neurons in the rat hippocampus and striatum. J Neurosci 15: 529−539. |
| 38. | Shuaib A, Ijaz S, Hasan S & Kalra J. (1992) Gamma-vinyl GABA prevents hippocampal and substantia nigra reticulata damage in repetitive transient forebrain ischemia. Brain Res 590: 13−17. |
| 39. | Shuaib A, Mazagri R & Ijaz S. (1993) GABA agonist "muscimol" is neuroprotective in repetitive transient forebrain ischemia in gerbils. Exp Neurol 123: 284−288. |
| 40. | Simon RP, Swan JH, Griffith T & Meldrum BS. (1984) Blockade of N-methyl-D-aspartate receptors may protect against ischemic damage in the brain. Science 226: 850−852. |
| 41. | Spetzler RF & Hadley MN. (1989) Protection against cerebral ischemia: the role of barbiturates. Cereb Brain Metab Rev 1: 212−229. |
| 42. | Staley KJ, Soldo BL & Proctor WR. (1995) Ionic mechanisms of neuronal excitation by inhibitory GABAA receptors. Science 269: 977−981. |
| 43. | Sternau LL, Lust WD, Ricci AJ & Ratcheson R. (1989) Role for -aminobutyric acid in selective vulnerability in gerbils. Stroke 20: 281−287. |
| 44. | Sydserff SG, Cross AJ, West KJ & Green AR. (1995) The effect of chlormethiazole on neuronal damage in a model of transient focal ischaemia. Br J Pharmacol 114: 1631−1635. |
Acknowledgements
We thank Susan Adams and Aghdas Jafari for technical support and Dr. Sandra Hewett for helpful discussions. This work was supported by NIH research grant NS-32636 (D.W.C.).
