Accumulation of the excitatory neurotransmitter glutamate in ischemic brain tissue contributes to neuronal cell death. Volatile anesthetics at clinically relevant concentrations are neuroprotective in in vivo models of brain ischemia and reduce glutamate release in vivo and in vitro, but they appear to have weaker neuroprotective effects than hypothermia. The purpose of this study was to determine whether isoflurane reduces glutamate release in hypoxic brain slices, how large this effect is compared to that of hypothermia, and if it is diminished by hyperthermia.
Glutamate released from rat cortical brain slices during chemical anoxia (100 microM NaCN) was measured continuously with a fluorescence assay. The release rate was compared at three temperatures (28 degrees C, 37 degrees C, and 39 degrees C) with and without isoflurane at concentrations equipotent to 1 minimum alveolar concentration. At the same three temperatures, glutamate release rates before and after exposure to isoflurane were compared.
Isoflurane reduced glutamate release from brain slices during chemical anoxia at 37 degrees C (19.6%, P < 0.01) and at 39 degrees C (25.4%, P < 0.01), but not at 28 degrees C. The reduction in glutamate release with hypothermia was similar to that with isoflurane. Hyperthermia (39 degrees C) caused greater glutamate release under basal and anoxic conditions than normo- and hypothermia. Isoflurane caused a slight increase in basal glutamate release rates, although this effect was smaller than the increase caused by hyperthermia.
In a brain slice model of cerebral anoxia, 1 minimum alveolar concentration isoflurane decreases glutamate release to a similar extent that hypothermia (28 degrees C) does. The increased glutamate release with hyperthermia (39 degrees C) is not prevented by isoflurane.
Key words: Anesthetics, volatile: isoflurane. Animals: rat. Brain: excitatory neurotransmitters; ischemia. Experimental techniques: brain slices; neurotransmitter release.
A large body of evidence from in vitro experiments suggests that volatile anesthetics have neuroprotective qualities. However, neuroprotective properties of anesthetics have been difficult to demonstrate in in vivo experiments [1-3]and may be weaker than the protective properties of mild hypothermia. [4-6]Neuroprotective effects of hypothermia and anesthetics have been associated with a reduction in cerebral metabolic rates. However, for an equal degree of suppression of metabolic rate, hypothermia is a more potent cerebral protectant than are anesthetic agents. [4,7,8]Mild hypothermia is thought to be protective by reducing the accumulation of excitatory neurotransmitters during ischemia. 
L-glutamate is the major excitatory neurotransmitter in the vertebrate central nervous system. Accumulation of glutamate in brain tissue during hypoxia or ischemia plays an important role in pathophysiologic events. [9-13]In cortical cell cultures, an exposure to 100 micro Meter glutamate (probably 100 times greater than normal extracellular concentrations) for 5 min causes cell death. During ischemia, increases in glutamate concentrations reach the millimolar range. Glutamate at high concentrations leads to an excessive activation of glutamate receptors. Activation of the N-methyl-D-aspartate subtype glutamate receptor then causes uncontrolled influx of calcium ions by increasing the calcium conductance. [11,15]Activation of non-N-methyl-D-aspartate (kainate and AMPA/quisqualate) can alsocontribute to a rise in intracellular calcium either by direct influx of calcium through the activated channels or by calcium influx through voltage gated calcium channels after membrane depolarization. [11,15]High intracellular calcium concentration leads to cell damage and, finally, to cell death. Although the mechanism of hypoxic and ischemic cell damage is not fully understood, the key role of glutamate in this process is well established. Recent evidence suggests that neuroprotective actions of anesthetics may involve, in part, a reduction in glutamate release from ischemic or hypoxic brain tissue. [16,17]Because glutamate accumulation is very sensitive to changes in temperature, [9,18]the interpretation of the results of many previous studies is complicated by reductions in brain temperature during anesthesia. 
The purpose of the current study was to determine whether isoflurane, a commonly used anesthetic in neurosurgical procedures, reduces glutamate release in anoxic brain slices and whether the magnitude of this effect is greater or lesser than that of hypothermia. In addition, because hyperthermia is detrimental during anoxia, we determined whether hyperthermia causes glutamate release and whether it decreases the effect of isoflurane on glutamate release.
Materials and Methods
The following methods were approved by the University of California San Francisco Committee on Animal Research, and they conform to relevant National Institutes of Health guidelines.
Preparation of Brain Slices
The methods for brain slice preparation used in this study are commonly used and are well described in the literature. [16,20-22]Brain slices were prepared from 18 10- to 35-day-old Sprague-Dawley rats anesthetized with 1.5-2% halothane in oxygen. After decapitation, the brain hemispheres were rapidly dissected. The skull was opened, ice cold (1-3 degrees Celsius) artificial cerebrospinal fluid (aCSF) (Earle's balanced salts, composition in mM: NaCl 116, NaHCO sub 3 25, KCl 5.4, CaCl21.8, MgCl20.9, NaH2PO40.9, glucose 10, pH 7.40, bubbled with 5% CO2-95% Oxygen2) was poured over the brain, and the brain hemispheres were removed. All following steps of the preparation were done in an ice-cold environment. The brain hemispheres were glued with cyanoacrylate to a holder and immersed in 1-3 degrees Celsius aCSF. Slices (300-350 micro meter thick) were then prepared with a vibrating tissue slicer (Campden Instruments, Cambridge, UK). Slices were transferred to vials of gassed aCSF and stored at room temperature (approximately 25 degrees Celsius), completely submersed in aCSF, bubbled with 95% Oxygen2/5% CO2. After 30 min, the solution in the vials was exchanged. Slices were allowed to recover from slicing trauma for at least 45 min before starting the experiments.
Glutamate Release Assay
Glutamate released from brain slices during experiments was detected with a fluorescence assay in a Hitachi F-2000 fluorometer (Hitachi, Tokyo, Japan). Slices were mounted on a mesh screen and placed in a fluorometer cuvette that contained aCSF prewarmed to 28 degrees Celsius, 37 degrees Celsius, and 39 degrees Celsius, respectively, 1 mM nicotinamide adenine dinucleotide (NAD) (Sigma Chemical), and 6 IU/ml glutamate dehydrogenase (Sigma Chemical). The total volume in the cuvette was 1.7 ml. Glutamate released from the slice was converted to alpha -ketoglutarate with the reduction of NAD sup + to NADH. The reaction catalyzed by glutamate dehydrogenase is shown below. Equation 1
The fluorescence from NADH was measured with an excitation wavelength of 340 nm and an emission wavelength of 460 nm. A magnetic stirbar ensured rapid mixing of the solution. The measured fluorescence was plotted against the time, and the glutamate release rate was calculated from the slope of the curves after calibration. Calibration of the assay was performed by injecting 2-20 nmol L-glutamate into the cuvette and measuring the increase of fluorescence after completion of the reaction. Calibration curves were run for all three different temperatures.
Interference of isoflurane and NaCN with the assay was excluded by control experiments. Time control studies showed stable recordings for at least 15 min for the slice without any insult as well as for the slice after exposure to isoflurane.
A heating element attached to the cuvette holder was used to maintain the temperature throughout the experiment. During mock experiments with continuous measurement, the temperature of the heating element was adjusted so that changes in temperature for the duration of the experiments did not exceed plus/minus 0.2 degree Celsius. Once adjusted, the temperature was measured at the beginning and end of each experiment.
Glutamate release rate from the brain slices was measured at three temperatures (28 degrees Celsius, 37 degrees Celsius, and 39 degrees Celsius). The cortex portion of each brain slice was divided in half; one half was exposed to chemical anoxia, the other to chemical anoxia plus isoflurane. This was done to decrease the influence of variability due to slight differences between brain slices. For both treatments, glutamate release was followed until a stable release rate was achieved. For isoflurane-treated slices, a saturated isoflurane solution was injected into the cuvette and the headspace filled with 1.5% (approximately 1 minimum alveolar concentration [MAC] for adult rats ) isoflurane in Oxygen2/5% CO2. Saturated isoflurane solution (pure isoflurane in aCSF, concentration of isoflurane in aCSF was 12.4 mmol/l) was prepared and added to the cuvette to achieve 1 MAC or EC50(350 micro Meter for adult rats ) concentrations. Minimum alveolar concentration was corrected for the effect of age and temperature on anesthetic potency. Aqueous concentrations for different temperatures were calculated using Franks and Lieb's equations (310 micro Meter for 28 degrees Celsius, 350 micro Meter for 37 degrees Celsius, and 359 micro Meter for 39 degrees Celsius). The isoflurane EC50for rats given in that article was corrected for the age effect using the values for halothane given by Cook et al. (values corrected for age effect: 400 micro Meter for 28 degrees Celsius, 450 micro Meter for 37 degrees Celsius, and 463 micro Meter for 39 degrees Celsius). We assumed that age affects MAC of halothane and isoflurane similarly. The final amount of saturated isoflurane solution that was added to give a 1.7 ml total cuvette volume was 66, 76, and 78 micro liter for 28 degrees Celsius, 37 degrees Celsius, and 39 degrees Celsius, respectively (corrected for age effect and estimated loss).
In all of the experiments, pH-adjusted NaCN solution (final concentration 100 micro Meter) was added to the cuvettes as the anoxic stimulus. This concentration of NaCN causes loss of more than 90% of ATP in brain slices within 10 min after exposure. 
The numbers of slices for each temperature were as follows: 30 slices at 28 degrees Celsius, 40 slices at 37 degrees Celsius, and 26 slices at 39 degrees Celsius.
To estimate the protein content, brain slices were transferred to 50 mM Hepes buffer (pH 7.8) at 100 degrees Celsius, sonicated, and stored at -80 degrees Celsius for later assay (Enhanced Protein Assay, Pierce, Rockford, IL).
Glutamate release rates after exposure to isoflurane (1 MAC) were compared with basal glutamate release rates at all three temperatures (28 degrees Celsius [n = 25], 37 degrees Celsius [n = 25], and 39 degrees Celsius [n = 26]) without an additional insult.
Measured glutamate release rates were normalized with respect to the protein content of the brain slices.
The data for the assessment of a possible dose-dependent effect of isoflurane was obtained as a separate set. Three different anesthetic concentrations (0.5 MAC, 1 MAC, 2 MAC, (225, 450, 900 micro Meter, respectively, corrected for age)) were used, and the experiments were performed at 37 degrees Celsius.
Measurement of Anesthetic Concentrations
Mock experiments were performed to determine isoflurane concentrations during the experiments. Samples of the solution were taken from the cuvette at the end of the experiments to measure partial pressure of isoflurane. In a constant temperature waterbath, isoflurane was extracted into air and injected into a Gow Mac Gas Chromatograph (Bridgewater, NJ) calibrated with the appropriate standards. Isoflurane concentration was calculated from the partial pressure at the end of the experiments, and a loss of 24 plus/minus 9% (n = 6) over the timecourse of the experiments was determined. The amount of isoflurane added to the cuvettes was corrected for the mean loss.
Results are reported as mean plus/minus SD. N values correspond to the number of brain slices and not to the number of animals. The influence of isoflurane and temperature on NaCN-evoked release was tested by one-way analysis of variance (ANOVA) with Student-Newman-Keuls test as post hoc test for relevant pairwise comparisons. Analysis of variance with Student-Newman-Keuls test was used to assess changes in basal glutamate release caused by isoflurane. Data for the analysis of the isoflurane effect on basal release were taken from the same experiments, but the numbers of samples are smaller because artifacts in the recording made it impossible to use all experiments for this analysis. The dose response effect of isoflurane on cyanide-evoked glutamate release was evaluated in a separate experiment with a separate set of data, which was analyzed using linear regression analysis.
P values less than 0.05 were considered statistically significant.
Effect of Isoflurane and Temperature on Anoxia-evoked Glutamate Release
Exposure of brain slices to chemical anoxia (100 micro Meter NaCN) caused rapid release of glutamate (Figure 1). Cyanide anoxia-evoked glutamate release increased with temperature. The release rate for 37 degrees Celsius was 13.6 plus/minus 4.1 pmol *symbol* mg sup -1 *symbol* sec sup -1 (n = 40), release rate at 39 degrees Celsius (17.9 plus/minus 4.2 pmol *symbol* mg sup -1 *symbol* sec sup -1 [n = 26]) was 31.6% greater, and the rate at 28 degrees Celsius (9.9 plus/minus 2.4 pmol *symbol* mg sup -1 *symbol* sec sup -1 [n = 30]) was 27.4% lower, respectively (ANOVA, P < 0.01; Figure 2). Isoflurane at equipotent concentrations of approximately 1 MAC reduced the amount of glutamate released at 39 degrees Celsius by 21.3% (from 17.9 plus/minus 4.2 to 14.1 plus/minus 2.7 pmol *symbol* mg sup -1 *symbol* sec sup -1 [n = 26], P < 0.01) and at 37 degrees Celsius by 19.2% (from 13.6 plus/minus 4.1 to 11.0 plus/minus 3.7 pmol *symbol* mg sup -1 *symbol* sec sup -1 [n = 40], ANOVA, P < 0.01). The difference between the means of the control (9.9 plus/minus 2.4 pmol *symbol* mg sup -1 *symbol* sec sup -1 and isoflurane group (8.2 plus/minus 3.0 pmol *symbol* mg sup -1 *symbol* sec sup -1) at 28 degrees Celsius (mean value of treatment group was 17.2% less than the controls) is not statistically significant (ANOVA, P > 0.05, n = 30). There was no difference in the degree of inhibition of glutamate release with 1 MAC isoflurane (37 degrees Celsius) and hypothermia (28 degrees Celsius).
The reduction in glutamate release caused by isoflurane was greater at 39 degrees Celsius than at 37 degrees Celsius. These data are graphically displayed in Figure 2.
Effect of Temperature on Basal Glutamate Release
Basal glutamate release from brain slices increased with temperature. A decrease in temperature from 37 degrees Celsius to 28 degrees Celsius blocked the basal glutamate release (from 1.0 plus/minus 1.1 to -0.1 plus/minus 0.5 pmol *symbol* mg sup -1 *symbol* sec sup -1, ANOVA, P < 0.01), whereas a 2 degrees Celsius temperature rise from 37 degrees Celsius to 39 degrees Celsius caused an enormous increase in basal glutamate release (from 1.0 plus/minus 1.1 to 5.7 plus/minus 2.2 pmol *symbol* mg sup -1 *symbol* sec sup -1, ANOVA, P < 0.01; Figure 3).
Effect of Isoflurane on Basal Glutamate Release
Isoflurane at a concentration of approximately 1 MAC caused a statistically significant increase in glutamate release in otherwise untreated slices at 28 degrees Celsius (from -0.1 plus/minus 0.5 to 0.7 plus/minus 0.9 pmol *symbol* mg sup -1 *symbol* sec sup -1, ANOVA, P <0.05) and 37 degrees Celsius (from 1.0 plus/minus 1.1 to 3.1 plus/minus 1.5 pmol *symbol* mg sup -1 *symbol* sec sup -1 ANOVA, P < 0.01). Figure 3shows the amount of glutamate released in slices before and after adding approximately 1 MAC isoflurane. The temperature change from 37 degrees Celsius to 39 degrees Celsius causes a greater increase than does 1 MAC of isoflurane (ANOVA, P < 0.01), and isoflurane plus the temperature change causes the same increase than does the temperature change alone (5.6 plus/minus 1.3 and 5.7 plus/minus 2.2 pmol *symbol* mg sup -1 *symbol* sec sup -1, ANOVA, P > 0.05).
Dose-response Effects of Isoflurane on Cyanide-evoked Glutamate Release
The data for the dose-response relation were normalized with respect to the controls. A regression analysis failed to show a significant effect (P = 0.079). Although differences in mean glutamate release were calculated between the control group and the groups with 0.5-2 MAC isoflurane, these differences were too small to reach statistical significance. (Figure 4).
The results show that (1) isoflurane at clinical concentrations significantly reduces cyanide-evoked glutamate release from brain slices and (2) hypothermia reduces cyanide-evoked glutamate release to the same extent as 1 MAC isoflurane. These findings provide an important comparison of the potency of isoflurane and hypothermia in reducing glutamate release during cerebral ischemia, although an extrapolation from the brain slice model to intact brain is difficult.
Suppression of Glutamate Release by Isoflurane or Hypothermia
Although certain volatile anesthetics (halothane, sevoflurane, [5,6]and isoflurane ) have been shown to be histologically or neurologically protective during focal cerebral ischemia in rats, their protective potency relative to hypothermia appears to be low. In our study, we found that hypothermia (28 degrees Celsius) produced a similar reduction in glutamate release as did 0.5 - 2.0 MAC isoflurane. Patel et al. also found that isoflurane and hypothermia reduced glutamate accumulation to a similar extent in the brain of rats, although only mild hypothermia (33-34 degrees Celsius) was used. In some studies, even mild hypothermia completely eliminated glutamate accumulation during ischemia. The difference in the potency of hypothermia to reduce glutamate release/accumulation in an in vitro and in vivo animal model may be due to factors intrinsic to these very different model systems. These differences include lack of circulation in brain slices, differences in the effects of simulated and actual ischemia, differences in diffusion and accumulation of extracellular neurotransmitters, and injured neurons/glia in the superficial layer of the slices.
Although the effects of isoflurane and hypothermia on cyanide-evoked glutamate release appear small in brain slices, we do not yet know how large a reduction in glutamate release is necessary to be protective in vivo brain ischemia. Although an extrapolation to intact brain is difficult, the findings suggest that isoflurane is not a potent drug for reduction of glutamate release during ischemia or prevention of neurologic injury from ischemia. This agrees with previous in vivo studies. [2-4,17,28]The potency of isoflurane in reducing histologic injury from focal ischemia is uncertain, because none of the recent, temperature-controlled studies [3,4]included a control group that received no anesthetic.
Effect of Hyperthermia on Glutamate Release
Mild hyperthermia is deleterious in brain anoxia or ischemia. [18,29]We found that mild hyperthermia increased both basal and cyanide-evoked glutamate release. In fact, the increase in glutamate release caused by a 2 degrees Celsius increase in temperature (approximately 32%) was as great as the decrease caused by a 9 degrees Celsius drop in temperature (approximately 31%). This comparison shows how even mild hyperthermia greatly increases the deleterious effects of anoxia. Warner et al. also reported significant worsening of histopathologic outcome when pericranial temperature was risen from 38 degrees Celsius to 39.2 degrees Celsius. A protective effect described for halothane in a following study was not large enough to persist during mild hyperthermia.
Role of Glutamate in Brain Ischemia
Inhibition of glutamate release and the accumulation of extracellular glutamate in hypoxic or ischemic brain tissue by anesthetics is significant because glutamate triggers cell damage and cell death. [14,30]The source of glutamate release during anoxia is, for the following reasons, believed to be predominantly extrasynaptic. Synaptic function is rapidly lost after onset of hypoxia or ischemia, most likely because of failure of synaptic vesicle transport and fusion after ATP loss. [12,31]The rapid loss of ATP after onset of anoxia in our brain slice model was demonstrated previously. Second, ATP loss leads to membrane depolarization due to failure of the ATP-dependent Sodium sup + /Potassium sup + transport mechanism. The ATP threshold for ischemic cortical depolarization was reported as 13-18% of normal ATP concentrations in rat brain cortex. Neither the presence of an anesthetic nor temperature variation changed the observed threshold. With depolarization, the transmembrane gradient for potassium can no longer be maintained, and the extracellular potassium concentration rises to approximately 60 mM after a few minutes. Under normoxic conditions, the low extracellular glutamate concentration is maintained by an uptake carrier. This uptake carrier transports two sodium ions into the cell with each glutamate anion, while transporting one potassium ion and one hydroxide ion out of the cell. Because of the cotransport of ions and the net transport of one positive charge into the cell, the carrier is dependent on membrane potential and ion gradients. Because of membrane depolarization and rundown of ion gradients during anoxia, the function of the carrier is reversed, transporting glutamate out of the cell. This reversed uptake mechanism seems to be the major source of glutamate accumulation during anoxia instead of synaptic release. [12,13,16,31]Synaptic glutamate release is a calcium-dependent process, but Calcium sup ++ influx through voltage gated Calcium sup ++ channels after membrane depolarization seems not to be a major contributor to glutamate accumulation, because blockade of the calcium channels has no significant effect on glutamate release during anoxia. 
That glutamate accumulation has been shown to cause neuronal cell death suggests two possibilities, or a combination of both, for brain protection during cerebral ischemia or hypoxia: (1) reduction of the amount of glutamate being released and (2) antagonism of the postsynaptic effects of glutamate. Glutamate receptor antagonists have been shown to reduce neuronal death in vivo and in vitro, [30,35]although their protective effect remains controversial. A reduction of ischemia-induced glutamate release by isoflurane in an in vivo model was reported recently. A similar effect was reported for halothane, enflurane, and sodium thiopental, but not propofol, in vitro. 
Effect of Isoflurane on Basal Glutamate Release
We found that isoflurane itself caused a significant increase in glutamate release without any additional stimulus. Hirose et al. associated an increase in glutamate release from rat brain synaptosomes caused by enflurane with the convulsive properties of enflurane. In the same study, an increase in glutamate release produced by halothane was significantly smaller. An elevation of cytosolic free calcium in brain slices caused by isoflurane could be the cause of the glutamate release we found. The effect could be seen in the normothermic and hypothermic studies. A possible explanation for these observations could be an excitatory effect with onset of anesthetic action. An analog to this effect might be excitation observed in patients during onset of anesthesia with volatile anesthetics.
Limitations of the Brain Slice
Brain slices have limitations that make an extrapolation to intact brain difficult. First, there is an injury layer in each slice that may behave different from normal brain and from the rest of the slice. There is no remaining circulation, and gas exchange, nutrition, secretion, and drug action are dependent on diffusion. Therefore, the results might only show the reaction of an unknown number of cells rather than the entire slice. The health of the slices in the used model was verified earlier by measurement of ATP content, which revealed values in the normal range. Our brain slices maintained low intracellular calcium, an indicator of cellular viability.
Although the glutamate release assay is commonly used, the continuous conversion of glutamate to alpha-ketoglutarate may prevent an accumulation of glutamate within the slice, which is different from the in vivo situation.
We conclude that (1) isoflurane reduces anoxia-evoked glutamate release, (2) hypothermia reduces anoxia-evoked glutamate release to the same extent as isoflurane, and both effects appear to be small, and (3) the increase in glutamate release with hyperthermia (39 degrees Celsius) is not prevented by isoflurane.
The authors thank Bonnie Hansen for technical support and Pompi Ionescu and Edmond I. Eger for measuring anesthetic concentrations. They also thank Leslie T. Buck for important discussion and review of the manuscript.