Effects of Intravenous Anesthetic Agents on Glutamate Release: A Role for GABAAReceptor–Mediated Inhibition

Female Wistar rats (250–300 g) were killed by cervical dislocation and decapitation. The brain was rapidly removed and placed in ice-cold oxygenated (95% O², 5% CO²) bicarbonate buffer, pH 7.4, of the following composition: 115 mM NaCl; 4.7 mM KCl; 2.0 mM CaCl²; 1.2 mM MgCl²; 25 mM NaHCO³; and 8.8 mM glucose. The cortex was dissected from the other cerebral structures, cut into 350 × 350 μM slices using a MacIlwain Tissue Chopper (Mickle Lab Engineering Co. Ltd., Surrey, UK), and suspended in bicarbonate buffer. After being washed three times in fresh bicarbonate buffer, slices were agitated in a shaking water bath at 37°C for 40 min. Approximately 1 ml gravity-packed slices (protein not determined) were pipetted into a perfusion chamber and held in place by a greased diffuser. The perfusion chamber consisted of a 2-ml syringe barrel cut at approximately 2 cm and packed at the needle end with a 0.5-cm-thick layer of glass wool. The chamber was sealed around the diffuser with two O-rings. The perfusate entered the diffuser via  a peristaltic pump, which provided a continuous, steady flow to the cerebrocortical slices at the diffuser end and exited at the needle end of the syringe, with the eluate being collected by a fraction collector. Slices were perfused at 37°C at 1 ml/min for 60 min to allow for stabilization before collection of 2-min fractions for the estimation of glutamate concentrations. After 6 min of perfusion, 46 mM K⁺(Na⁺adjusted to maintain tonicity) was applied (total K⁺was therefore 46 + 4.7 mM = 50.7 mM) for 2 min (S¹). Slices were perfused for an additional 30 min prior to the second application of a 2-min pulse of 46 mM K⁺(S²). Fractions were collected for 8 min after S².

The fractions collected were analyzed for glutamate using an adaptation of the fluorescence method previously described. 13Glutamate dehydrogenase catalytically reduces any glutamate present to 2-oxoglutarate, accompanied by the reduction of NAD⁺/NADP (nicotinamide adenine nucleotides) to NADH/NADPH. As NADH undergoes reoxidation, NADP⁺is used. 13Assay volume was 500 μl and consisted of 15 μl of glutamate dehydrogenase (final concentration, 30 U), 5 μl of NADP (final concentration, 1 mM), and 480 μl of perfusate. The mixture was incubated at 37°C for 10 min, and fluorescence intensity was determined using a Perkin-Elmer LS50B (Beaconsfield, Bucks, UK) spectrofluorometer with excitation and emission wavelengths set at 366 and 430 nm, respectively. Perfusate sample fluorescence intensity was then compared with a known set (0.5–20 pmol) of glutamate standards.

Thiopental (3–300 μM), propofol (10 μM), and ketamine (3–300 μM) were introduced to the buffer immediately after S¹until the end of the experiment (thus including S²). In some experiments, 100 μM pentobarbital and 300 μM barbituric acid were included as additional active and inactive barbiturates, and 1 μM dizocilpine (MK-801) was included as a noncompetitive NMDA receptor antagonist. In experiments using 1–100 μM bicuculline, it was introduced when S¹response had returned to the basal level by perfusing the slices with a solution of the relevant concentration of bicuculline. After 10 min, coincubation of bicuculline with 100 μM thiopental, 10 μM propofol, or 30 μM ketamine took place until the end of S². Control response with and without a high and low concentration of test agent were obtained from a single rat (i.e. , three conditions per animal), and these data were combined to produce a full concentration response curve.

Time course of glutamate release is presented relative to the mean of the first three basal samples collected during S¹or S². This measurement both confirms slice responsiveness (S¹) and normalizes for differences in protein content among experiments. The areas under the stimulation curves (S¹and S²) were then calculated and data presented as S²/S¹ratio. In some experiments, the inhibition of this ratio is presented. Statistical analysis was performed using Student paired t  test for paired comparisons of S²/S¹ratios and analysis of variance when a series of comparisons was necessary. P < 0.05 was taken as an indication of statistical significance. The analysis of concentration–response curves in release and release-inhibition studies to yield half-maximum inhibition of release (EC⁵⁰;i.e. , an agonist effect), and half-maximum reversal of the agonist effect (IC⁵⁰) was performed using computer-assisted curve fitting (sigmoid concentration response curve-variable slope) on GraphPad Prism 2.0. (GraphPad Software, San Diego, CA). Because each concentration response curve is produced from data collected from multiple experiments, it is not possible to derive error estimates for these values.

Addition of 46 mM K⁺produced monophasic releases of glutamate during both S¹and S²from perfused rat cerebrocortical slices. Over the entire study, the control (n = 96) S²/S¹ratio was 1.07 ± 0.33 (mean ± SD;fig. 1). There was considerable variation in the S²/S¹ratio (95% confidence interval, 1.00–1.13; range, 0.47–1.88) between animals, but each animal acted as its own control. Glutamate release from rat cerebrocortical slices was dependent on extracellular Ca²⁺(data not shown), indicating exocytotic release of vesicular glutamate and not reversal of the glutamate uptake transporter.

Thiopental (fig. 2) and ketamine (fig. 3) produced a concentration-dependent inhibition of 46 mM K⁺-evoked glutamate release from rat cerebrocortical slices with maximum inhibition of 55.4 ± 16.8% and 65.1 ± 8.1%, respectively, produced at 100 μM. The EC⁵⁰for this inhibition was 18.2 and 10.9 μM, respectively (figs. 2A and 3). In addition, 100 μM pentobarbital also significantly inhibited glutamate release by 60.4 ± 14.4% (P < 0.05). In contrast, the inactive barbiturate barbituric acid was ineffective. MK-801, a noncompetitive NMDA antagonist, inhibited release by 81.2 ± 7.5% compared with control (P < 0.05). At 10 μM, propofol inhibited glutamate release by 74.8 ± 8.2% (P < 0.05).

In a separate series of pilot studies, we showed that bicuculline 100 μM alone had no effect on 46 mM K⁺-evoked glutamate release (S²/S¹ratio control, 1.01 ± 0.10; bicuculline, 0.96 ± 0.04;P > 0.05; n = 5) and that this concentration also fully reversed 100-μM thiopental–induced inhibition of release. We further probed this inhibitory effect of bicuculline by constructing a full concentration–response curve. Bicuculline produced a complete reversal of 100 μM thiopental inhibition of release with an estimated IC⁵⁰of 10.9 μM (fig. 2B). Bicuculline (100 μM) also fully reversed the inhibition produced by propofol (fig. 4A). In contrast, the inhibition produced by 30 μM ketamine was completely unaffected by bicuculline (fig. 4B).

We report a reliable monophasic release of glutamate from perfused oxygenated rat cerebrocortical slices in response to 46 mM K⁺depolarization. We used this model system to probe the effects of a range of intravenous anesthetic agents on glutamate release. We have clearly showed that ketamine, thiopental, and propofol markedly reduce glutamate release. In addition, MK-801 (a noncompetitive NMDA antagonist) and pentobarbital, but not barbituric acid, also significantly inhibit glutamate release. These data confirm our initial hypothesis that glutamate release is a target for intravenous anesthetic agents. Bicuculline, a GABA^Aantagonist, reversed the inhibition of 46 mM K⁺-evoked glutamate release produced by thiopental and propofol, suggesting that this inhibition of glutamate release is mediated indirectly via  an agonist action at the GABA^Areceptor. As a negative control, the inhibition of release produced by the NMDA antagonist ketamine was unaffected by bicuculline. Collectively, our data indicate that there is a subtle interaction between glutamatergic and GABA-releasing (GABAergic) transmission in the production of the anesthetized state (fig. 5). In this model, it is assumed that GABA^Areceptor activation of glutamatergic neurons by thiopental and propofol results in hyperpolarization, which reduces the amount of glutamate released. In vivo , we envisage an interplay between GABAergic and glutamatergic transmission in the control of cortical activity, with release of transmitter from the former affecting glutamate release from the latter.

Enhancement of GABA^Areceptor activity has been championed by many authors as the principal target site for anesthetic agents. 6,7,14–17Certainly, barbiturates have been shown to enhance the affinity of GABA for the GABA^Areceptor and to slow the dissociation of GABA from its receptor. 14,15Propofol has also been shown to enhance the inhibitory actions of GABA. 14,16Recent work using chimeric receptor constructs has identified a 45–amino acid sequence on the GABA^Areceptor that is both necessary and sufficient to enhance receptor function, supporting the notion that ethanol and the volatile anesthetic agent enflurane act at this site to exert a specific effect on this important ion-channel protein. 17 

Our finding of reversal of an anesthetic effect by bicuculline (and hence an action at GABA^Areceptors) is supported by two electrophysiologic studies. In the first, using rat hippocampal CA1 slices, inhibition of electrically evoked population spikes by pentobarbital and propofol were completely reversed by bicuculline, but ketamine had no such effect 18; however, glutamate release was not measured. Another study, also using rat hippocampal CA1 slices, found that halothane and isoflurane reduced glutamate transmission (excitatory postsynaptic potentials), but the reduction was completely insensitive to bicuculline 10 μM. 19The underlying mechanism of the discrepancy between intravenous agents and volatile agents in their susceptibility to reversal by bicuculline is unclear but clearly warrants further investigation.

The IC⁵⁰of 10.9 μM for the reversal of thiopental inhibition of K⁺-evoked glutamate release by bicuculline is consistent with other studies. 20–22In these studies, GABA-evoked responses in the vagus nerve in vitro  were inhibited with IC⁵⁰values of 6.1 and 1.2 μM, respectively. Although it has been suggested that bicuculline acts on more than the GABA^Areceptor within the central nervous system, perhaps on dopaminergic midbrain systems (IC⁵⁰of 26 μM), 22we feel it is unlikely that the effects we observed in this study are explained by non-GABA^Aactions of bicuculline.

Total peak serum concentrations of anesthetic agents vary considerably between studies and among different species. Peak serum concentrations for ketamine, propofol, thiopental, and pentobarbital of 9–94, 35, 380, and 200 μM, respectively, have been reported. 23,24Protein binding variably reduces these concentrations. In addition, these may not represent target site (brain) concentrations because propofol has been shown to be concentrated approximately eightfold in the brain. 25The clinical relevance of in vitro  studies can always be questioned; however, the concentrations used in our study represent those that are clinically achievable. EC⁵⁰values for ketamine and thiopental of 18.2 and 10.9 μM fall well within the range of values noted previously, even when corrected for protein binding.

Not all barbiturates inhibit glutamate release. Wei et al.  26reported that [5-(2-cyclohexylidene-ethyl barbituric acid] (CHEB), a convulsant barbiturate, stimulates glutamate release from rat cerebrocortical synaptosomes with an EC⁵⁰of 14.2 μM. In addition, pentobarbital and phenobarbital inhibited this CHEB-evoked release. The inhibitory effect of ketamine on glutamate release is well recognized as an NMDA receptor–mediated event. In a model of neonatal rat spinal cord, Brockmeyer and Kendig 27showed a dose-dependent inhibitory effect of ketamine on glutamate release and ventral root potential, which constitute an NMDA receptor–mediated phenomenon. Importantly, dorsal root potential, a GABA receptor–mediated phenomenon, was unaffected by ketamine. However, the action of ketamine on glutamatergic transmission is more complex than simple inhibition. A study of glutamatergic neurotransmission in the prefrontal cortex of conscious rats indicated that low doses of ketamine increased glutamatergic transmission by an action of non-NMDA postsynaptic receptors, but that higher anesthetic doses similar to those used in our study decreased glutamate release. 28More recently, it has been shown that thiopental and methohexital reduce glutamate release from and Ca²⁺entry into cultured neurons. 29 

We have shown that 10 μM propofol produces a substantial inhibition of 46 mM K⁺-evoked glutamate release. However, in a study of hypoxemia-induced glutamate release from rat brain slices. Bickler et al.  11failed to detect an inhibition of glutamate release with propofol. The variability of glutamate release in response to propofol is further illustrated in the findings of Ratnakumari and Hemmings, 30who reported that propofol inhibited veratridine- and 4-aminopyridine–evoked glutamate release (both Na⁺channel–dependent processes). However, propofol did not inhibit K⁺-evoked glutamate release (a Na⁺channel–independent process) up to 100 μM. The reason for the difference between these results and our findings of inhibition of K⁺-evoked glutamate release by propofol is unclear but may be related to the different experimental models. These investigations used a rat cerebrocortical synaptosome preparation, which may not have contained the complete circuitry of excitatory and inhibitory interneurons, in contrast to our model using rat cerebrocortical slices.

The precise role of glutamate release in the mechanism of general anesthesia remains to be fully elucidated. Although most studies support a role for glutamate release in the mechanism of volatile agent anesthesia, 31–33a recent study using cyclobutane derivatives with anesthetic and nonanesthetic properties found that both anesthetic and nonanesthetic compounds strongly inhibited glutamate release, but only the nonanesthetic compound inhibited K⁺-evoked GABA release, suggesting that inhibition of excitatory neurotransmission may be less important than activation of inhibitory neurotransmission in the primary action of general anesthetics. 34This is consistent with our observation in the current study that inhibition of K⁺-evoked glutamate release by thiopental and propofol is reversed by bicuculline, a GABA^Aantagonist. In humans, clinical anesthesia induced by NMDA antagonists such as ketamine differs in some qualitative respects from that induced by thiopental and propofol. Ketamine anesthesia is characterized by “dissociative anesthesia,” with analgesia, sedation, and detachment. Thiopental and propofol are noted for profound depression of consciousness and lack of analgesic effects. 2,3This is consistent with our laboratory finding that ketamine inhibits glutamate release in a manner unrelated to that of thiopental and propofol. Further studies are warranted to determine the neuronal circuitry involved in GABA-mediated inhibition of glutamate release. Using this model, it will be interesting to determine whether volatile agent inhibition of glutamate release is bicuculline sensitive.

A variety of commonly used intravenous anesthetic agents have been shown to depress the K⁺-evoked release of the excitatory neurotransmitter glutamate. The inhibition produced by thiopental and propofol is probably secondary to activation of GABA^Areceptors.