Blockade of Glutamate Receptors and Barbiturate Anesthesia : Increased Sensitivity to Pentobarbital-induced Anesthesia Despite Reduced Inhibition of AMPA Receptors in GluR2 Null Mutant Mice

The sensitivity to pentobarbital was investigated in 28 GluR2  null mutant (−/−), 45 wild-type (+/+), and 26 heterozygous (+/−) littermate mice older than 6 weeks of age. Pentobarbital 30 mg/kg (Somnotol, MTC Pharmaceuticals, Cambridge, Ontario, Canada; 65 mg/ml diluted to 1 mg/ml in 0.9% normal saline) was injected intraperitoneally after an aspiration test. The mice were then placed in a plastic container that was warmed by a heat lamp. An investigator who was blinded to the genotype observed the mice and recorded the time interval between the injection of pentobarbital and the loss of righting reflex (LORR). At 2-min intervals, the animals were gently placed on their backs, and if they failed to right themselves, the time was noted. After the LORR, supplemental oxygen (4 l/min) was delivered through a nose cone placed near the snout. Mice that failed to lose their righting reflex were observed for at least 60 min. In addition, the corneal reflex was tested at 2-min intervals by lightly brushing the cornea with a 1.0 proline suture. The pineal reflex was tested using the suture to irritate the lower aspect of the ear helix, and toe-pinch withdrawal was assessed by squeezing a hindlimb toe using a rubber-shod hemostat. Respiratory rate was measured during a 15-s interval every 2 min. Respiratory depression was defined as a ≥ 25% decrease in baseline respiratory rate determined before the injection. Finally, the time to the recovery of the righting reflex, or sleep time, was noted, and mice were observed for an additional 20 min before being returned to their cages. Animals were maintained under the heat lamp during the observation period. Two weeks to 1 month later, 15 (+/+) and 14 (−/−) mice from this initial group were injected with pentobarbital 35 mg/kg intraperitoneally and observed according to the protocol described.

In eight additional animals, the body temperature was measured by inserting a rectal probe (Fisher Scientific, Hampton, NH) immediately after pentobarbital 30 mg/kg was administered intraperitoneally. Temperature was measured at 2-min intervals until the animals regained the righting reflex or until 15 min had elapsed for mice that did not lose this reflex.

All animal study protocols were approved by the animal facility at the Samuel Lunenfeld Research Centre, Mount Sinai Hospital, Toronto, and the Animal Care Committee of the University of Toronto.

The plasma concentration of pentobarbital was measured in six subjects consisting of three (+/+) mice and three (−/−) mice that had not been previously injected with the barbiturate. Animals were killed by decapitation 5 min after the injection of pentobarbital (30 mg/kg intraperitoneally). The sample was obtained at 5 min because this time approximates the latency to LORR in the (−/−) mice. Approximately 1 ml blood was collected from the decapitation site into a 1.5-ml microcentrifuge tube. Samples were centrifuged, and pentobarbital concentrations were assayed from the plasma using high-performance liquid chromatography.

Mice older than 6 weeks of age were anesthetized with halothane and then decapitated. The entire brain was then rapidly removed and rinsed in cold (4°C) extracellular solution consisting of (in mM): NaCl 140, CaCl²1.3, KCl 5.4, N -2-hydroxy-ethylpiperazine-N ′-2-ethanesulphonic acid (HEPES) 25, glucose 33, and tetrodotoxin 0.0003. This solution was buffered to a p  H of 7.4 with NaOH, and osmolality was adjusted to 320 to 325 mOsm. The hippocampi were then isolated and cut by hand with a razor blade into approximately 500-μm-thick slices. The slices were incubated in extracellular solution containing 0.4 mg/ml papain latex (Sigma Chemical Co., St. Louis, MO) for 15–30 min at room temperature (20–22°C), then washed in papain-free solution and allowed to acclimate while submersed in oxygenated extracellular solution at room temperature for 2 h. Each slice was transferred to a plastic 35-mm tissue culture dish (Nalge Nunc International, Roskilde, Denmark) before the isolation of individual pyramidal neurons. Pyramidal neurons from the CA1 region of the hippocampus were isolated by mechanical trituration using three Pasteur pipettes with consecutively smaller tip diameters (700 to 150 μm). The acutely dissociated hippocampal neurons were allowed to settle to the bottom of the dish before the recordings were made.

The genotypes of all mice used in these studies were confirmed by Southern blotting or polymerase chain reaction of tail genomic DNA.

All electrophysiologic studies were performed at room temperature (22°C). Patch electrodes were pulled from thin-walled borosilicate glass (1.5-mm outer diameter; World Precision Instruments, Sarasota, FL) using a two-stage vertical puller (Narishige PP-83, Tokyo, Japan) to a series resistance of 3–10 MΩ. The electrodes were filled with intracellular solution consisting of (in mM): CsF 140, CsOH 35, HEPES 10, MgCl²2, EGTA 11, tetraethylammonium chloride 2, CaCl²1, magnesium adenosine 5′-triphosphate 4. This solution was buffered to a p  H of 7.4 using CsOH, and the osmolality was adjusted to 300–310 mOsm. For the experiments in which GABA was applied as the agonist, the recording solution contained CsF 70 mM and CsCl 60 mM instead of CsF 140 mM. The CsCl was added to increase intracellular chloride (Cl⁻) concentrations so that a detectable outward current could be recorded after the application of GABA.

Acutely isolated neurons were voltage clamped at a holding potential of −60 mV. Whole-cell currents were recorded using the Axopatch 1D amplifier (Axon Instruments Inc., Foster City, CA) and data were filtered (2 kHz), digitized, and acquired on-line using the pCLAMP5 program (Axon Instruments). Changes in series resistance were monitored during the recordings by measuring the capacitance transient resulting from a hyperpolarizing test pulse of 10 mV. Recordings that demonstrated marked changes in access resistance were not used for data analysis.

A multibarrel perfusion system 21was used to achieve a rapid exchange (30 ms) of the extracellular solution. Three square capillary tubes (400 μM × 400 μM) were glued together and mounted on a Leitz manipulator (Wild Leitz Canada Ltd., Midland, Ontario, Canada). Each barrel was connected to a 7-ml reservoir of perfusion fluid, and the flow rate set to approximately 0.5 ml/min by adjusting the height of the reservoir. A computer-driven motor-based stepper (Vexta motor; Oriental Motor Co., Torrance, CA) was used to laterally move the capillary assembly. After the formation of the whole-cell configuration, neurons were lifted into the outflow of the barrels. Cells were perfused with the extracellular solution or solutions containing the AMPA receptor agonist, kainate (Sigma Chemical Co.), GABA (Sigma Chemical Co.), or barbiturates (pentobarbital, thiopental, and phenobarbital).

Glutamate-evoked currents mediated by AMPA receptors rapidly desensitize to a low steady-state amplitude. Although kainate-evoked currents also desensitize, the extent of desensitization is considerably less, and the steady-state current is greater compared with glutamate-evoked responses. 22Therefore, kainate was used to investigate the inhibition of AMPA receptors by barbiturates as previously described. 19,20,23,24Various concentrations of barbiturates were applied during the steady-state response evoked by kainate. It is unlikely that the activation of kainate receptors, which undergo a fast pronounced desensitization, 25contributes appreciably to the currents recorded in these experiments. Because it is the endogenous ligand of AMPA receptors, glutamate (3 mM) was applied to activate AMPA receptors in another set of experiments. This saturating concentration of glutamate was applied with the N -methyl-D-aspartate receptor antagonist, DL-2-amino-5-phophonovaleric acid (40 μM; Sigma Chemical Co.), and MgCl²2 mM in the absence and presence of pentobarbital 30 μM and 100 μM. For these experiments, bicuculline 20 μM was added to all solutions to inhibit barbiturate activated Cl⁻current mediated by GABA^Areceptors.

The current amplitude was measured using pCLAMP (Axon Instruments), and data were plotted using GraphPad Prism (Graph Pad, San Diego, CA). Concentration–response relationships for kainate-evoked currents and GABA-evoked currents recorded in the presence and absence of pentobarbital 30 μM were fitted using a modified version of the Michaelis-Menten equation by the least squares method. The concentration of kainate that produced 50% of the maximal response (EC⁵⁰) and the Hill coefficient (n^H) were determined according to the equation:

where I^maxis the maximal response observed at a saturating concentration of the agonist. Membrane capacitance, as an approximation of neuron size, was estimated from the area under the capacitance transient evoked by a 10-mV hyperpolarizing pulse applied after the formation of the whole-cell configuration.

For concentration–inhibition analysis, kainate was applied at the EC^55–60concentration in the absence or presence of the barbiturates. The concentrations of pentobarbital, thiopental, or phenobarbital that produced 50% of the maximal inhibition (IC⁵⁰) were determined from the concentration–inhibition curves. Data points were normalized to the maximal inhibition and fit according to the equation:

where I^maxis the maximal inhibition produced by a saturating concentration of the antagonist.

GABA and glutamate responses recorded in the presence of pentobarbital were examined for changes in amplitude of the peak and steady-state currents, respectively.

All results were reported as mean ± SEM unless otherwise indicated. For the behavioral study, the latency to LORR and sleep time were compared between groups using a one-way analysis of variance (SigmaStat 2.0 software; SPSS Inc., San Rafael, CA). The chi-square and Fisher exact tests were used to compare the number of mice in each group that showed the loss of the righting, corneal, pineal, and toe-pinch withdrawal reflexes. Where appropriate, the EC⁵⁰values for kainate and GABA-evoked currents, as well as the IC⁵⁰values for pentobarbital, thiopental, and phenobarbital for (+/+) and (−/−) neurons were compared using the Student t  test. Data sets for which tests of normality failed were analyzed with corresponding nonparametric tests.

The study population of (+/+) and (−/−) mice were initially organized into three groups according to postnatal age: group 1 (5–10 weeks), group 2 (23 weeks), and group 3 (50 weeks). No significant differences in weight were observed between the age-matched groups (table 1). Similarly, no age-dependent differences were demonstrated for the latency to LORR, sleep time, or loss of corneal or pineal reflexes within each genotype. Therefore, the results were pooled according to the genotype for all ages.

After intraperitoneal injection of pentobarbital 30 mg/kg, a greater proportion of the (−/−) mice demonstrated LORR (table 2) and a shorter latency to LORR (5.1 ± 0.5 min, n = 26; 6.5 ± 0.4 min, n = 17; 7.8 ± 0.9, n = 25;P < 0.05) compared with (+/−) or (+/+) littermates, respectively (fig. 1A). GluR2  null mutant mice also demonstrated a longer sleep time (18.1 ± 1.5 vs.  10.3 ± 1.3 and 10.8 ± 1.5 min;P < 0.05;fig. 1B). A greater proportion of (−/−) mice lost their corneal and pineal reflexes compared with the (+/+) and (+/−) mice (P < 0.05;table 2). Loss of the toe-pinch withdrawal reflex was not demonstrated in any of the mice after injection of pentobarbital. The enhanced sensitivity to pentobarbital in (−/−) mice was not likely attributed to pharmacokinetic factors because the plasma concentration of pentobarbital was similar, at least at the 5-min interval, in (−/−) mice and (+/+) littermates (128.8 ± 15.9 μM, n = 3 vs.  113.3 ± 12.8 μM, n = 3, respectively;P > 0.05).

Some of the mice that were not killed for the electrophysiologic studies were allowed to recover for a minimum 2-week period. Pentobarbital 35 mg/kg was administered intraperitoneally as previously described. Three mice, two (−/−) and one (+/+), died from respiratory failure, whereas 14 (+/+) and 12 (−/−) mice were successfully anesthetized, and most of the subjects demonstrated LORR (table 3). Consistent with our previous findings, the (−/−) mice showed a decreased time to LORR (4.1 ± 0.4 min, n = 12 vs.  5.7 ± 0.3 min, n = 13;P < 0.05) and increased sleep time (24.5 ± 3.5 min, n = 12 vs.  12.6 ± 2.8 min, n = 12;P < 0.05). A greater proportion of (−/−) mice lost their pineal and corneal reflexes compared with (+/+) littermates; however, these differences were not statistically significant (table 3). The incidence of respiratory depression was not different between the (+/+) and (−/−) mice (tables 2 and 3).

Because the sensitivity to anesthetics can be influenced by core temperature, eight additional mice were observed for changes in rectal temperature after pentobarbital 30 mg/kg was administered intraperitoneally. No significant differences in temperature was detected at any time after the pentobarbital injection, including at 4 min [37.0 ± 0.4°C (+/+), n = 4 vs.  36.5 ± 0.4°C (−/−), n = 4] and 10 min [37.3 ± 0.2°C (+/+), n = 4 vs.  36.8 ± 0.3°C (−/−), n = 4].

In summary, these results demonstrate an increased anesthetic sensitivity to pentobarbital in the (−/−) mice compared with (+/−) and (+/+) littermates as evidence by a shorter latency to LORR, longer sleep time, and the greater proportion of mice that lost the corneal and pineal reflexes. This enhanced sensitivity could not be attributed to differences in blood concentrations or changes in rectal temperature.

The subunit composition of ligand-gated receptors influences the EC⁵⁰value of the receptor for agonist, as well as their sensitivity to pharmacologic agents. Furthermore, the extent of inhibition produced by an antagonist can vary with the concentration of agonist. Therefore, to ensure that equi-effective concentrations of agonist were used to elicit control currents in (−/−) and (+/+) neurons, the concentration–response relationships for kainate-evoked currents were first examined.

Applications of kainate (> 10 μM) activated an inward current in all hippocampal neurons tested. The maximum current evoked by a saturating concentration of kainate (I^max) was not influenced by the presence of the GluR2  subunit:(−/−)I^max= 1,341 ± 217 pA, n = 15 versus  (+/+)I^max= 1,746 ± 233 pA, n = 12 (P > 0.05). Similarly, no differences were observed in the membrane capacitance in the dissociated cells:(−/−) 14.2 ± 3.9 pF versus  (+/+) 14.9 ± 3.8 pF. The concentration–response relationship indicated the EC⁵⁰value for kainate-evoked current was significantly lower for (−/−) neurons compared with (+/+) neurons (136 ± 7 μM, n = 17 vs.  226 ± 23 μM, n = 15;P < 0.05), respectively. Our results are consistent with previous reports of a higher potency for kainate in recombinant AMPA receptors that lack the GluR2  subunit. 26The EC⁵⁰values for receptors containing the GluR2  ^flipsubunit in combination with GluR1  (EC⁵⁰= 110 μM) or GluR1 /GluR3  ^flip(EC⁵⁰, 100 μM) were greater than those values for homomeric GluR1  (EC⁵⁰, 32 μM) or dimeric GluR1 /GluR3  ^flip(EC⁵⁰, 55 μM) receptors. The EC⁵⁰value we report for (+/+) neurons (EC⁵⁰, 226 μM) approximate those values reported for pyramidal neurons isolated from the CA3 region (EC⁵⁰, 344 μM) or CA1 region (EC⁵⁰, 474 μM) of the rat hippocampus. 27In addition, the Hill coefficient (1.69 ± 0.04 vs.  2.06 ± 0.13;P < 0.05) was lower for (−/−) neurons compared with (+/+) neurons, respectively (fig. 2), suggesting a decreased cooperativity for kainate activation of GluR2 -deficient AMPA receptors. 28 

To investigate the effects of barbiturates on AMPA receptors, equi-effective concentrations of kainate were applied to (−/−) neurons (EC⁵⁵, 150 μM) or (+/+) neurons (EC⁶⁰, 300 μM). Kainate was applied for 1.5 s before the application of the various concentrations of barbiturates. Only recordings that demonstrated a stable level of inhibition and full recovery after washout of the barbiturate were used for the data analysis.

The concentration–inhibition relationship for pentobarbital blockade of kainate-evoked currents is illustrated in figures 3 and 4A. The IC⁵⁰value for pentobarbital inhibition was approximately sixfold greater for (−/−) neurons compared with (+/+) neurons (301 ± 52 μM vs.  51 ± 10 μM;P < 0.05). Similarly, the IC⁵⁰values for thiopental (153 ± 29 μM vs.  34 ± 6 μM;P < 0.05) and phenobarbital (930 ± 344 μM vs.  205 ± 55 μM;P < 0.05) were fourfold to fivefold greater in (−/−) compared with (+/+) neurons, respectively (figs. 4B and 4C). These results demonstrate a reduced barbiturate sensitivity of AMPA receptors in GluR2 -deficient neurons. No significant differences in the Hill coefficient (n^H) were observed for pentobarbital [(+/+) n^H= 1.00 ± 0.03 and (−/−) n^H= 1.31 ± 0.27] and thiopental [(+/+) n^H= 1.08 ± 0.05 and (−/−) n^H= 0.98 ± 0.16]. However, the Hill coefficient for phenobarbital was increased in the (−/−) neurons:(+/+) n^H= 1.12 ± 0.07 and (−/−) n^H= 1.63 ± 0.18 (P < 0.05). The reduction in the slope factor for barbiturate concentration–inhibition curves reported for recombinant GluR2 -deficient receptors 19was not demonstrated in (−/−) neurons.

Consistent with these results, the steady-state current evoked by glutamate (3 mM) was also sensitive to inhibition by pentobarbital. Pentobarbital 30 μM and 100 μM caused a significant inhibition of the current recorded from (+/+) neurons (30%± 7% and 49%± 6%, respectively; n = 11;P < 0.05) but not from (−/−) neurons (8%± 11% and 24%± 12%; n = 4;P = 0.40 and 0.26, respectively). Therefore, a decrease in barbiturate inhibition of AMPA receptors in (−/−) neurons was also demonstrated when the endogenous ligand, glutamate, was used to activate AMPA receptors.

The GABA^Areceptor is thought to be a primary site of action of barbiturates. 29Therefore, we also investigated the effects of pentobarbital (30 μM) on GABA^Areceptor–mediated current activated by a subsaturating (30 μM) and a saturating concentration of GABA (600 μM). No differences were observed in the maximal amplitude of currents evoked by GABA 30 μM recorded from (−/−) neurons (1,489 ± 345 pA, n = 10) or (+/+) neurons (1,160 ± 235 pA, n = 13), respectively. Similarly, no difference in the maximal current activated by GABA 600 μM was apparent for responses from (−/−) neurons (4,654 ± 612 pA, n = 12) and (+/+) neurons (4,205 ± 695 pA, n = 12). Pentobarbital (30 μM) applied in the absence of GABA did not activate inward current in any of the neurons. However, pentobarbital (30 μM) potentiated the peak currents evoked by 30 μM GABA but not 600 μM GABA (fig. 5). The effects of pentobarbital on the GABA concentration–response relationship were also investigated. The calculated EC⁵⁰values for GABA were:(−/−), 51 ± 6 μM (n = 17); and (+/+), 46 ± 6 μM (n = 13). Pentobarbital caused a shift to the left of the GABA concentration–response curve and significantly reduced the GABA EC⁵⁰value to 43 ± 6 μM (n = 11) and 34 ± 5 μM (n = 11) for (−/−) and (+/+) neurons, respectively (fig. 6). These results indicate that GABA^Areceptors present in (−/−) and (+/+) neurons are similarly influenced by pentobarbital.

Our results show that GluR2  null mutant mice are more sensitive to the anesthetic effects of pentobarbital compared with wild-type littermates. In contrast, AMPA receptors in GluR2 -deficient neurons were resistant to inhibition by barbiturates as indicated by a fourfold to sixfold increase in the IC⁵⁰values for pentobarbital, thiopental, and phenobarbital. Taken together, these data demonstrate that the behavioral effects of barbiturates do not correlate with the inhibition of AMPA receptors in this mouse model. Although the rank order of potencies of the barbiturates for AMPA receptor inhibition (thiopental > pentobarbital > phenobarbital) is consistent with the behavioral potencies in animals and humans, 30this rank order also correlates with the modulation of GABA^Areceptors, a primary target site for anesthetic drugs.

Previous behavioral studies suggest that inhibition of AMPA receptors contributes to the neurodepressive effects of barbiturates. The selective non–N -methyl-D-aspartate receptor antagonist, NBQX, administered intravenously in rats caused a dose-dependent increase in the duration of LORR caused by hexobarbital. 31Potentiation of the anesthetic effect of hexobarbital by NBQX, together with evidence that barbiturates inhibit the AMPA receptor in vitro , led to the suggestion that AMPA receptors contribute to the clinical properties of barbiturates. Alternatively, these behavioral data could also be interpreted as indicating that NBQX reduced the baseline level of excitatory neurotransmission, rendering the central nervous system more sensitive to the inhibitory GABAergic effects of barbiturates.

Electrophysiologic studies support a reduction in excitatory synaptic signaling in GluR2  null mutant mice as the amplitude of excitatory postsynaptic potentials and currents are reduced in hippocampal slices from these mice. 13This suggests that the absence of the GluR2  subunit is associated with a generalized reduction in excitatory neurotransmission, a state analogous to that induced by low concentrations of NBQX. Consequently, barbiturates may unmask an inherent susceptibility to neurodepressive drugs.

An alternative mechanism to account for the enhanced sensitivity to pentobarbital in GluR2  null mutant mice is attributed to the modulation of polyneuronal networks. For example, in hippocampal circuits, inhibitory GABAergic interneurons are activated by glutamate. AMPA receptors present in these interneurons can contain GluR2  subunits as indicated by in situ  hybridization and immunofluorescence labeling studies. 32,33In wild-type mice, pentobarbital blockade of AMPA receptors would reduce the activation of inhibitory interneurons, thereby reducing GABAergic transmission in these mice. In contrast, in GluR2  null mutant mice, barbiturate-resistant AMPA receptors would permit the persistent activation of inhibitory interneurons.

Thus, two mechanisms are postulated to account for our experimental results. The first attributes the enhanced sensitivity in GluR2  null mutant mice to a global reduction of excitatory neurotransmission and suggests that barbiturates act at non-AMPA receptors (such as GABA^Areceptors). The second mechanism attributes the increased sensitivity in GluR2  null mutant mice to barbiturate-resistant AMPA receptors that activate inhibitory interneurons. Anesthetics that cause minimal inhibition of AMPA receptors and cause no differential modulation of wild-type and GluR2 -deficient receptors could be used to distinguish between these two possibilities. We predict that if excitatory neurotransmission is globally reduced in GluR2  null mutant mice, then AMPA receptor–independent anesthetics 34would also be more potent in these mice. In this regard, the volatile anesthetic halothane causes minimal inhibition of AMPA receptors at clinically relevant concentrations. 35Preliminary studies indicate that halothane is more potent in GluR2  null mutant compared with wild-type mice for LORR, 36supporting the model in which the increased anesthetic sensitivity is attributed to a reduction in glutamate-mediated excitatory neurotransmission.

Subtle but important differences in the number or function of synaptic AMPA receptors present in GluR2  null mutant and wild-type mice may not be revealed by analyzing kainate-evoked currents recorded from the soma of acutely dissociated neurons. Although we demonstrated that the potency of AMPA receptors for kainate is lower in wild-type (EC⁵⁰, 226 μM) compared with GluR2 -deficient receptors (EC⁵⁰, 136 μM), we observed no difference in the maximal amplitude of kainate-evoked currents in GluR2 -deficient and wild-type neurons. This result is consistent with the findings of Feldmeyer et al. , 14who reported that the maximal macroscopic conductance to kainate was similar for wild-type and GluR2 -deficient (GluR^neo/neo) neurons. However, the GluR2  subunit plays a critical role in the surface expression and trafficking of glutamate receptors to the postsynaptic compartment. 37A reduction in the number or function of synaptic AMPA receptors in GluR2 -deficient neurons could contribute to the reduction in excitatory field potentials and increased sensitivity to anesthetics.

Resistance of GluR2 -deficient receptors to barbiturate inhibition was first demonstrated in recombinant AMPA receptor subunits (GluR1  to GluR4 ) expressed in Xenopus  oocytes. Pentobarbital inhibited GluR2 -containing receptors (GluR1/2  or GluR2/3 ) with an IC⁵⁰of approximately 180–200 μM, whereas GluR2 -deficient receptors (GluR1 , GluR1/3 , GluR3 ) had IC⁵⁰values of 1.1–2 mM. 19The absence of the GluR2  subunit was also associated with a decrease in the slope of the concentration–inhibition curve and a Hill coefficient less than 1. These results indicate that the absence of the GluR2  subunit may cause a decrease in the number of drug binding sites or a change in the cooperativity or access to the binding site. 28Our results in native neurons from GluR2  mutant mice demonstrate a similar decrease in potency for barbiturate inhibition of GluR2 -deficient receptors and a low value for the Hill coefficient. However, deciphering the molecular mechanism(s) of inhibition by analyzing the concentration–response relation in mutant receptors is problematic because differences in agonist binding or gating of AMPA receptors may not be revealed by the shape of the concentration–response curve. Further studies are necessary to clarify the mechanisms of barbiturate inhibition of AMPA receptors.

In summary, our behavioral study indicates that GluR2  null mutant mice are more sensitive to barbiturates despite a reduced inhibition of GluR2 -deficient AMPA receptors by barbiturates. Our behavioral experiments were primarily designed to measure LORR as a surrogate end point for the loss of consciousness. However, the state of general anesthesia is comprised of multiple components, including hypnosis, amnesia, analgesia, and autonomic stability. Our results cannot be extrapolated to include the other components of general anesthesia such as analgesia or immobility in response to pain, because the absence or presence of the GluR2  subunit might differentially influence these end points. Nevertheless, from a clinical perspective, our results are compelling because a variety of neurologic insults, including ischemia and epilepsy, are associated with the downregulation of the GluR2  subunit. 38Our data predict that disorders associated with a relative reduction in GluR2  subunit expression, including stroke, 39,40schizophrenia, 41,42Alzheimer's disease, 43and amyotrophic lateral sclerosis, 44,45would be associated with a decreased dose requirement for anesthetic drugs.