The Effect of Thiopental and Propofol on NMDA- and AMPA-mediated Glutamate Excitotoxicity 

The experiments were approved by the Institutional Animal Care and Use Committee of the State University of New York Health Science Center at Brooklyn. The methods used have been described in detail previously. [19,23]

Adult male Sprague-Dawley rats (aged 100–120 days) were preoxygenated in a plexiglass chamber for 3 min and then anesthetized with 2% isoflurane in oxygen until they showed no response to tactile stimuli. Then the rats were decapitated, and the brain rapidly removed. The hippocampi were dissected from the brain and cut into 500-micro meter thick transverse slices. During these procedures, the tissue was kept in ice-cold (4 [degree sign] Celsius) artificial cerebrospinal fluid (aCSF). The slices were placed on nylon mesh attached to a plexiglass grid and superfused with aCSF at a rate of 60ml/min. The aCSF was preequilibrated with 95%O²/5%CO², and the temperature was maintained at 37 +/- 0.5 [degree sign] Celsius. The composition of the aCSF was (in mM): NaCl 126, KCl 3.0, KH²PO⁴1.4, NaHCO³26, glucose 4, MgSO⁴1.3, and CaCl²1.4.

After 1 h of incubation, a bipolar metal electrode was used to stimulate the Schaffer collateral pathway supramaximally. A tungsten electrode was placed in the CA1 pyramidal cell layer to record the postsynaptic-evoked population spike. The population spike was monitored for at least 30 min before any treatment was initiated to ensure the stability of the response. The electrophysiologic response was quantified by measuring the amplitude of the first spike of the postsynaptic population spikes.

For all treatments, the drug was added to the aCSF superfusing the slices to achieve a certain final concentration. At the end of the treatment, the drug was washed out. NMDA (25 micro Meter) or AMPA (15 or 10 micro Meter) was applied for 10 min, 2 h after the beginning of superfusion. In drug-treated groups, 600 micro Meter thiopental or 112 micro Meter propofol (equivalent to 20 micro gram/ml) was present 15 min before, during, and 10 min after the NMDA or AMPA treatment. In another thiopental/NMDA group, the thiopental exposure was prolonged to 2 h after the washout of NMDA. Thiopental, 600 micro Meter, is the concentration that demonstrated significant protection against anoxia in our previous study [19] and is approximately equal to a dose that causes barbiturate coma. Propofol, 20 micro gram/ml, protected against anoxia at 39 [degree sign] Celsius but not at 37 [degree sign] Celsius; a higher concentration (40 micro gram/ml) also did not improve the recovery at 37 [degree sign] Celsius. [23] Propofol, 20 micro Meter, is a dose that would cause complete anesthesia. For propofol experiments, either propofol in its clinical formulation (Diprivan injection [Zeneca Pharmaceuticals, Wilmington, DE], with a fat emulsion as solvent, referred to as Diprivan hereafter) or propofol as chemical reagent (dissolved in dimethyl sulfoxide [DMSO] before application, referred to as pure propofol hereafter) was used. The vehicle control experiments were performed by using exactly the same concentrations of a fat emulsion (Intralipid 10%, Pharmacia, Clayton, NC, final concentration 0.02%) or DMSO (0.2%, v/v) as those existing in the aCSF when propofol had been added.

The change in the amplitude of the population spike after a certain treatment was calculated by dividing the posttreatment value by its pretreatment control value and expressed as a percentage. Recovery from NMDA-induced damage was observed for up to 3 h after the washout of NMDA and for up to 5 h after AMPA-induced damage to ensure that the recovery had reached a stable level.

All values are expressed as mean +/- SD; n equals the number of animals, one slice per animal was examined. Sigma Stat (Jandel Corporation, San Rafael, CA) was used for statistical analysis. When two groups of data are compared, the nonparametric Mann-Whitney rank sum test is used (this is indicated in the tables as N). If the normality and equal variance tests are passed for both data groups, the software automatically chooses a parametric t test instead (this is indicated in the tables as P). When three groups of data are compared, the nonparametric Kruskal-Wallis one-way analysis of variance (ANOVA) on rank is used, and a multiple comparison procedure, Dunn's method, is selected as a post test (this is indicated in the tables as N). If the normality and equal variance tests are passed for all of the three data groups, the software automatically chooses a parametric one-way ANOVA; the Student-Newman-Keuls test is then used as a post test (this is indicated in the tables as P).

In untreated normoxic slices, the postsynaptic population spike remained stable for at least 6 h after the start of superfusion (Figure 1). In normoxic slices, 35 min of thiopental exposure reduced the population spike to 78% of its predrug level. Fifty min after the washout of thiopental, the population spike recovered to 92% of its predrug value (Figure 1). The data from all thiopental-treated slices demonstrated that 15 min of thiopental exposure increased the latency of the population spike to 132 +/- 3%(n = 29) and reduced the amplitude to 74 +/- 7%(n = 30) of their predrug value.

The mean time until complete block of the population spike after the addition of 25 micro Meter NMDA was 61 +/- 5 s, and there was invariably no response at the end of the 10-min NMDA treatment. The evoked population spikes 1 and 3 h after the washout of NMDA were 49% and 44% of their pre-NMDA amplitude. In tissue treated with thiopental, the mean amplitude of the response after 10 min of NMDA treatment was 33 +/- 10% of its pre-NMDA level (P < 0.05). After the washout of NMDA and thiopental, the 1- and 3-h recoveries were, respectively, 91% and 79% of their pre-NMDA amplitude (Table 1). The recovery of the population spike in the thiopental group was significantly better than that with NMDA alone throughout the 3-h period. An example of traces from single experiments are shown in Figure 2. If the thiopental exposure was extended to 2 h after the NMDA washout, the 3-h recovery was 72%(Table 1). This is significantly better than with NMDA alone, but it is not different from the group in which thiopental was discontinued 10 min after NMDA.

The mean time until complete block of the population spike after 15-micro Meter AMPA application was 44 +/- 1 s. The 1-h recovery of the population spike after the washout of AMPA was 3% of the pretreatment amplitude; slices recovered to 15% after 5 h. Thiopental significantly prolonged the mean time until complete block of the population spike with 15 micro Meter AMPA to 94 +/- 8 s. The recovery of the population spike amplitude with thiopental 1 and 5 h after the washout of AMPA was 20% and 50%, respectively. This recovery was significantly better than that with AMPA alone between 1 and 5 h (Table 1).

Diprivan (112 micro Meter) reduced the amplitude (42 +/- 7% vs. 97 +/- 2%) and prolonged the latency (123 +/- 18% vs. 98 +/- 4%) of the population spike when compared with its vehicle, Intralipid. Pure propofol had a similar effect compared with its solvent, DMSO. The amplitude was decreased (79 +/- 18 vs. 111 +/- 12), and the latency was increased (106 +/- 2% vs. 95 +/- 2%)

Propofol in Clinical Formulation (Diprivan). The population spike returned to 36% and 40% of its pretreatment amplitude 1 and 3 h after 10 min of 25 micro Meter NMDA. In slices treated with Diprivan, the clinical formulation of propofol, which contains Intralipid as a solvent, the population spike returned to 8% of its pretreatment amplitude 1 and 3 h after NMDA; with intralipid alone, the recovery was 14% 1 and 3 h after NMDA. The recovery of the population spike was significantly worse in the Diprivan-treated tissue than in tissue treated with NMDA alone (Table 2). The changes with Intralipid were not significantly different from untreated tissue.

The evoked population spike recovered to 3% and 15% 1 and 5 h after 10 min of 15 micro Meter AMPA. Diprivan did not significantly alter the response; the population spike recovered to 9% and 8%, 1 and 5 h after treatment of 15 micro Meter AMPA with Diprivan (Table 2).

A lower concentration of AMPA (10 micro Meter) was examined; the evoked response recovered to 64% and 69% of its pre-AMPA amplitude after 3 and 4 h. When Diprivan was present, the evoked response recovered to 41% and 46% of its pretreatment value after 3 and 4 h. When Intralipid was present, the recovery of the population spike after AMPA treatment was 42% and 45% at 3 and 4 h (Table 2). Although Diprivan did not significantly worsen the response after AMPA, it is clear that Diprivan did not improve recovery from AMPA; this is in contrast to thiopental.

The time it took to completely block the population spike during NMDA or AMPA treatment with Diprivan or Intralipid is not statistically different from untreated slices. The population spike was blocked after 61 +/- 6 s of NMDA alone and after 65 +/- 13 s of NMDA in slices treated with Diprivan. The response was blocked after 44 +/- 1 s of 15 micro Meter AMPA; slices treated with Diprivan were blocked after 43 +/- 3 s of AMPA.

Propofol in DMSO (Pure Propofol). The population spike recovered to 41% and 45% 1 and 3 h after 10 min of 25 micro Meter NMDA. If pure propofol dissolved in DMSO was present, there was significantly less recovery of the population spike to 6% and 15% 1 and 3 h after NMDA treatment. There was no difference between the recovery of the DMSO group and NMDA control group (Table 3).

The recovery after 10 micro Meter AMPA treatment was 64% and 69% after 3 and 4 h. Pure propofol did not significantly alter recovery; slices recovered to 44% and 51% of their pretreatment amplitude 3 and 4 h after AMPA. DMSO, the solvent for pure propofol, also did not significantly alter recovery (Table 3).

The time it took for AMPA or NMDA to completely block the population spike in either the pure propofol or DMSO groups was not significantly different from untreated slices. The response was blocked after 54 +/- 3 s of NMDA alone, 67 +/- 4 s after NMDA in slices treated with propofol, and 60 +/- 11 s after NMDA in slices treated with DMSO. The response was blocked after 47 +/- 3 s of AMPA (10 micro Meter) alone, 42 +/- 2 s of AMPA in slices treated with propofol, and 39 +/- 1 s of AMPA in slices treated with DMSO.

Glutamate excitotoxicity plays a key role in ischemic and anoxic neuronal damage. [1–4] It is also involved in the pathology of epilepsy [5] and brain trauma. [6] Glutamate receptor antagonists reduce anoxic and ischemic damage [1,2,7,9–11] and demonstrate anticonvulsive action in experimental epilepsy. [24] NMDA receptor antagonists provide protection against seizure-related neuronal damage. [5,25]

The hippocampal slice used in this study provides the capability of examining the direct effect of anesthetics on neuronal damage while excluding indirect effects such as cerebrovascular actions. This is ideal for studying the mechanisms of drugs with complicated actions. [19,23] Anoxic damage measured electrophysiologically in this model correlates with histologic and biochemical damage. [10,11,26] Results found with this model have been confirmed by in vivo studies. For example, barbiturates have demonstrated neuroprotection in vivo and in vitro. [12,14,18,19] Nimodipine fails to protect neurons against anoxia directly in this in vitro model, [26] which is in agreement with the results of clinical trials that could demonstrate no benefit after cerebral ischemia. [27] This model is also useful for examining the effects of excitotoxicity and the importance of excitotoxicity as a factor leading to anoxic damage. Aminophosphonovaleric acid, an NMDA receptor blocker, protected against ischemia and anoxia in an in vivo and in a hippocampal slice model. [1,9,10] Exposing the slice to brief periods of NMDA or AMPA leads to physiologic, protein synthetic, and histologic alterations that persist after the washout of the excitotoxic agent. [28–31]

The concentration of thiopental (600 micro Meter) used in the study is the one that showed protection against anoxia in previous experiments. [19] This is comparable with a high barbiturate coma dose of thiopental, which is similar to that used by Nussmeier in a clinical study. [18] At this concentration, thiopental reversibly reduces the amplitude of the population spike.

Thiopental significantly prolongs the time required to block the population spike after the application of NMDA and AMPA. The block of the population spike is a result of the loss of ion and electrical gradients across the neuronal membrane, which makes the neurons unexcitable. [32,33] This effect of thiopental reflects its ability to delay depolarization and reduce ion flux through voltage-dependent ion channels and possibly glutamate receptor channels, so that the transmembrane electrical gradient is better maintained.

Thiopental reduces NMDA- and AMPA-induced neuronal damage. This may result from its effect on depolarization or blockade of voltage-dependent ion channels. [34,35] It may also be a result of its inhibition on NMDA and AMPA/kainate receptors. Barbiturates have been shown to reduce glutamate-mediated postsynaptic excitation;[34] noncompetitively inhibit the responses evoked by kainate, quisqualate, and NMDA;[36] and suppress NMDA-, kainate-, and quisqualate-induced increase of intracellular Ca. [35] Blockade of NMDA and AMPA/kainate receptors by thiopental contributes to the attenuated neuronal damage induced by NMDA and AMPA. The ability of thiopental to reduce Na sup + and Ca²+ influx [19] may also be related to its inhibition of NMDA and AMPA/kainate receptors. NMDA and AMPA receptor-mediated excitotoxicities are involved in the pathogenesis of cerebral ischemia. [1,4,7] Barbiturates have been shown to be effective against cerebral ischemia in animals and humans. [13,14,18,19] Our results indicate that barbiturate protection against cerebral ischemia or anoxia, which we demonstrated previously, [18] may be due in part to the blockade of NMDA- and AMPA-mediated glutamate excitotoxicity. The attenuation of NMDA-mediated glutamate toxicity may be one mechanism by which barbiturates exert their antiepileptic action and might possibly reduce seizure-induced neuronal damage.

Glutamate excitotoxicity causes lipolysis and a subsequent increase in free fatty acids, including arachidonic acid. [4,37–39] Free fatty acids are harmful to cell membranes, and arachidonic acid generates free radicals after undergoing further metabolism. [37,38] Free radicals damage lipid and proteins. [39] Because thiopental can scavenge free radicals, [40] this may be one mechanism by which it improves the recovery.

Prolonged thiopental exposure after the NMDA-induced damage did not further increase the recovery. In general, thiopental given before and during ischemia or anoxia provides much better protection than given after the insult. [14] It seems that the protective mechanism of thiopental is effective during the period of insult.

The concentration of propofol used in our study (112 micro Meter, which is equivalent to 20 micro gram/ml) is within clinically applicable range for anesthesia. Burst suppression can be achieved at this concentration, [41] and it exerts reversible inhibition of the population spike. [23] Unlike thiopental, propofol does not alter the time required by NMDA or AMPA to block the population spike. This suggests that it might not sufficiently reduce ion flux and prolong the maintenance of the electrical gradient during this period.

We found that propofol increases the neuronal damage induced by NMDA. Some investigators have shown that propofol attenuates NMDA-induced toxicity and inhibits NMDA receptors in cultured hippocampal neurons. [42,43] There is no previous report suggesting that propofol worsens NMDA-mediated excitotoxicity. However, there exist many conflicting phenomena regarding the effect of propofol on cerebral ischemia and anoxia. Propofol reduces anoxic damage during hyperthermia but not normothermia in hippocampal slices. Despite the attenuation of anoxic changes in Ca²+, Na sup +, and ATP at 37 [degree sign] Celsius, there is no corresponding improvement in electrophysiologic recovery. [23] Propofol demonstrates protection against temporary focal ischemia in animals in some studies but fails to show any effect in the others. [20–23] These results indicate that propofol has complicated actions, as least with respect to its effects on cerebral ischemia. The beneficial effects of propofol on Ca²+, Na sup +, and ATP during anoxia may be offset by the damage due to increased NMDA-mediated excitotoxicity and thereby explain its lack of protection against anoxic damage at 37 [degree sign] Celsius. Facilitation of NMDA-mediated glutamate excitotoxicity by propofol may help explain the reported central excitatory side effects of propofol such as postanesthesia dreams, hallucinations, paroxysmal movements, and even convulsions. [41,44,45] The effect of propofol on epilepsy is still controversial because the nature of what are apparently epileptic seizures has not been identified, and propofol also shows antiseizure activity in situations of refractory status epilepticus. [44,45]

Diprivan did not alter the damage after 15 micro Meter AMPA. To increase the sensitivity of the study, the concentration of AMPA was decreased to 10 micro Meter. Diprivan only slightly and nonsignificantly decreased the recovery from the AMPA insult. We do not think this small difference, even if additional studies allowed it to achieve statistical significance is of biological importance. Pure propofol also does not significantly affect the recovery from AMPA damage.

We conclude that thiopental reduces NMDA- and AMPA-induced neurotoxicity, whereas propofol enhances the neuronal damage induced by NMDA. Thiopental and propofol have different properties with respect to glutamate excitotoxicity, which may partially explain the differential effect of these agents during anoxia and ischemia.