Etomidate Targets α⁵γ-Aminobutyric Acid Subtype A Receptors to Regulate Synaptic Plasticity and Memory Blockade

All experimental procedures and protocols were approved by the Animal Care Committee of the University of Toronto (Toronto, Ontario, Canada). Mice were obtained from two sources. Male mice (postnatal age [P] P90–P120) were purchased from Taconic Laboratories (Germantown, NY) or were obtained from the authors' breeding colony. The commercially purchased mice had the same hybrid genetic background (50% C57Bl/6 and 50% Sv129Ev) as Gabra5 −/− mice.13WT and Gabra5 −/− mice were bred in the University of Toronto animal care facilities. The generation, genotyping, and characterization of Gabra5 −/− mice have been previously described.13Behavioral studies of Gabra5 −/− mice used aged-matched male WT controls. All mice were handled in 5-min epochs every day, for 1 week before their use in behavioral experiments. The experimenter was blind to the drug treatment and genotype of the mice for all studies.

The LTP of excitatory potentials was studied with hippocampal slices prepared from P90–P120 mice. Mice were decapitated during isoflurane anesthesia, and their brains were quickly removed and placed in ice-cold oxygenated (95% O², 5% CO²) artificial cerebrospinal fluid (composition: 124 mm NaCl, 3 mm KCl, 1.3 mm MgCl², 2.6 mm CaCl², 1.25 mm NaH²PO⁴, 26 mm NaHCO³, and 10 mm d-glucose), with the osmolarity adjusted to 300–310 mOsm. Transverse brain slices (350 μm thick) were prepared with a VT1000E tissue slicer (Leica, Deerfield, IL). After a recovery period of 1 h in the oxygenated artificial cerebrospinal fluid, the slices were transferred to a submersion-type recording chamber. The residual concentration of isoflurane is assumed to be negligible under the above conditions. Field excitatory postsynaptic potentials (fEPSPs) were recorded at room temperature (21°–23°C) as previously described.9Baseline stimulation frequency was 0.05 Hz, and the stimulus intensity was adjusted to evoke a half-maximal fEPSP amplitude. LTP was induced in the slices by stimulating with a theta burst stimulation (TBS) protocol, which consisted of 10 stimulus trains at 5 Hz, with each train including 4 pulses at 100 Hz.

Vehicle (dimethyl sulfoxide), etomidate (1 μm), L-655,708 (20 nm), or both etomidate and L-655,708 were perfused into the recording chamber for 15 min before the induction of LTP. The fEPSPs were monitored before and 60 min after TBS. L-655,708 is an inverse agonist, which is a compound that binds to the same receptor binding site as the agonist but has an opposite pharmacologic effect. It has a 100-fold higher functional affinity for α⁵GABA^ARs than for GABA^ARs that contain the α¹, α², or α³subunits.14,16,17The concentration of L-655,708 selected for use in this study binds preferentially to α⁵GABA^ARs in tissue slices,17whereas the concentration of etomidate was selected because it occurs in the brains of mice injected with an amnestic dose of etomidate.9Others have suggested that the free aqueous concentration of etomidate that corresponds to amnesia in vivo  is 0.25 μm.18This value was based on an estimate of the brain:artificial cerebral spinal fluid partition coefficient (3.35). However, to achieve an appropriate steady state concentration at the depth of the recording electrode, perfusion of the slices for up to several hours may be required. Given the time-dependent decline in the integrity of the hippocampus slices, studies of plasticity were performed approximately 15–20 min after application of the drug, and a higher concentration of etomidate was added to the extracellular solution (1 μm).

The extracellular recording solution contained 6-cyano-7-nitroquinoxaline-2,3-dione (20 μm) and (2R)-amino-5-phosphonovaleric acid (10 μm) to block ionotropic glutamate receptors and tetrodotoxin (0.3 μm) to block voltage-dependent sodium channels. Patch pipettes had open tip resistances of 3–5 MΩ when filled with an intracellular solution that contained mm CsCl (140), 10 mm HEPES, 10 mm EGTA, 2 mm MgATP, and 1 mm CaCl²(pH 7.3 with CsOH, 295–305 mOsm). Currents were sampled at 10 kHz and filtered at 2 kHz by using an eight-pole low-pass Bessel filter. All cells were recorded at a holding potential of −60 mV. A stable baseline current (< 20% change) was confirmed before the application of drugs. The amplitude of the tonic current under control conditions was measured as the difference in the holding current before and during the application of etomidate (1 μm), L-655,708 (20 nm), bicuculline (10 μm), or a combination of these drugs.

In the pavlovian fear conditioning tasks, mice were exposed to a tone, which was subsequently paired with a foot shock in a novel conditioning context, with either no time delay (0 s for cued conditioning) or an interval of 20 s between the tone and the foot shock (trace conditioning).19,20Several different associative memory tests were conducted to determine the contribution of certain brain regions for which the extent of expression of α⁵GABA^ARs differs. More specifically, the hippocampus, which has a high expression of α⁵GABA^ARs, is known to play a key role in contextual and trace fear conditioning.19,21In contrast, the expression of α⁵GABA^ARs is relatively low in the amygdala.22Cued fear conditioning, which requires the basal lateral nucleus of the amygdala, served as a control.23 

Thirty minutes before being placed in the fear conditioning chamber, mice were randomly assigned to receive an intraperitoneal injection (2 ml/kg) of the vehicle (35% propylene glycol, 10% dimethyl sulfoxide), etomidate (4 mg/kg), L-655,708 (0.7 mg/kg), or the combination of etomidate and L-655,708 (administered together). For these experiments, the dose of etomidate9was carefully selected to cause conscious amnesia, a state characterized by minimal sedation (which confounds the study of learning and memory) combined with loss of explicit or episodic memory.24In addition, the dose of L-655,708 was selected to modify learning behaviors via  preferential modulation of α⁵GABA^ARs, as previously determined.25On day 1, single animals were allowed to explore the chamber for 180 s. An 800-Hz tone, created by a frequency generator, amplified to 70 dB, and lasting 20 s, was then presented. For cued fear conditioning, the last 2 s of each auditory tone was paired with an electric foot shock (2 s, 1 mA); for trace fear conditioning, the auditory stimulus and foot shock (2 s, 0.5 mA) were separated by 20 s. Each of these sequences was presented three times, separated by 60 s (for cued fear conditioning) or 240 s (for trace fear conditioning). For contextual fear conditioning, either a strong (2 s, 1 mA) or weak (2 s, 0.5 mA) foot shock was applied, depending on the protocol. On day 2, 24 h after the conditioning session, each mouse was assessed for a freezing response by placing it in the original context and scoring every 8 s for a total of 8 min to determine contextual fear. On day 3, the conditioning chamber was modified to measure the freezing response to the tone to study either cued or trace fear conditioning. Mice were monitored for 180 s for freezing to the modified context, to rule out contextual influences. After the monitoring period, the auditory tone was presented continuously for 300 s, and the freezing response was measured every 8 s.

The water maze is a hippocampus-dependent spatial navigation task that requires the mouse to use visual cues positioned around the room to locate a hidden platform in a circular tub of opaque water and has been previously described.26Briefly, a circular pool of diameter 1.2 m was filled with tap water (25°± 2°C), which was made opaque by the addition of a white nontoxic paint. Mice were pretrained for 10 days, with four trials on each day to locate a hidden platform. In the match-to-place paradigm, the location of the platform was changed daily, and the first trial of each day was used as a comparator or reference trial to determine learning on trials 2, 3, and 4. During the acquisition phase of the probe trial, each mouse was randomly assigned to receive an intraperitoneal injection of vehicle, etomidate (4 mg/kg), L-655,708 (0.7 mg/kg), or both etomidate and L-655,708 (administered together) 30 min before the experiment. The next day, a probe trial was performed to test the ability of the mice to recall the correct spatial location that previously contained the hidden platform. Data records were stored with HVS Water 2020 software (VHS Image, Hampton, United Kingdom) for off-line analysis. The time, swim path, and latency of each mouse were recorded during each trial, and the percentage of time spent in the correct region was calculated by the software during analysis.

Visible platform trials were also performed to test for possible differences in motivational factors, perceptual and motor abilities, and any possible nonspecific effects of etomidate and L-655,708 as previously described.9The mice were injected 30 min before the visible platform trial, similar to the treatment during the learning acquisition phase before the probe trial.

The elevated plus maze is designed to measure the anxiety levels of the mice.27The test hinges on the natural tendency of rodents to explore a novel environment and their aversion to open, elevated, and brightly lit areas. The elevated plus maze consisted of four arms (5 cm × 27.5 cm) that were joined by a central area (5 cm × 5 cm). Two opposite arms were enclosed by 30-cm-high walls, and the other two arms were open.

Mice were injected intraperitoneally with vehicle, etomidate (4 mg/kg), L-655,708 (0.7 mg/kg), or both etomidate and L-655,708 (administered together) 30 min before the experiment. Mice were placed in the central area of the maze facing an open arm and were scored for the amount of time they spent in the central area, the open arms, or the closed arms. The number of entries into the open and closed arms was also monitored. All mice were allowed to explore the maze for 5 min.

Statistical analyses for the electrophysiologic and behavioral data were completed with GraphPad Prism Version 4.0c (San Diego, CA). All pooled data are presented as mean ± SEM. Electrophysiologic and behavioral statistical comparisons were completed using a one-way (i.e. , drug treatment only) or two-way (drug treatment vs.  genotype) analysis of variance with two-tailed inference testing. Post hoc  analyses were conducted using the Tukey–Kramer method, which accounted for both equal and unequal sample size comparisons. For the plots of LTP, the data points (slope of the fEPSP measured between 25% and 70% of the rising phase) were binned in 1-min increments to facilitate readability. The extent of LTP was quantified for statistical comparisons by averaging the slope of the fEPSPs during the final 5 min of each experiment and normalizing to baseline values. P < 0.05 was considered statistically significant.

First, to determine whether a reduction in α⁵GABA^AR activity modifies synaptic plasticity induced by TBS, L-655,708 (20 nm) was applied at a concentration that selectively blocks the tonic inhibitory conductance in CA1 pyramidal neurons without substantially altering inhibitory synaptic transmission.15In vehicle-treated slices, robust LTP was induced following a 1-s presentation of TBS, such that the slope of the fEPSP was significantly increased (P = 0.03 vs.  baseline; n = 8; fig. 1A). The application of L-655,708 did not modify the strength of LTP (P = 0.02 vs.  control slices; n = 8; fig. 1A). Next, the effect of L-655,708 on synaptic excitability was studied by comparing the slope of the input–output relation, where current intensity was plotted against the slope of the fEPSP (fig. 1B). The application of L-655,708 after TBS did not further enhance synaptic excitability. Therefore, a decrease in α⁵GABA^AR activity, similar to a reduction in the expression of α⁵GABA^ARs, did not modify synaptic plasticity or neuronal excitability under baseline conditions.

Next, slices from WT mice were cotreated with the same concentration of L-655,708 and etomidate to determine whether inhibiting α⁵GABA^ARs can reverse etomidate-induced blockade of LTP. Etomidate inhibited LTP in WT slices stimulated with the TBS protocol (P = 0.01 vs.  control LTP; n = 8; fig. 1C).9This effect of etomidate was reversed by the coapplication of L-655,708 (P = 0.02 vs.  etomidate-treated slices; n = 8; fig. 1C). Therefore, inhibition of α⁵GABA^ARs reverses the LTP-blocking property of etomidate. In addition, etomidate blocked an increase in synaptic excitability after TBS; this effect was not observed in slices treated with both etomidate and L-655,708 (fig. 1D).

Voltage clamp experiments were performed to determine whether L-655,708 reversed the enhancement of the tonic conductance by etomidate. Etomidate caused a significant increase in the tonic current, as evidenced by an inward shift in the holding current (I^Hold; n = 5; figs. 2A–C), as previously reported in studies of pyramidal neurons grown in dissociated cell cultures.9To determine the proportion of the etomidate-potentiated tonic current that was attributed to α⁵GABA^ΑRs, L-655,708 (20 nm) was applied. L-655,708 caused a 73 ± 9.04% reduction in the I^Hold(n = 5; figs. 2A–C), suggesting that a large proportion of the etomidate-enhanced tonic current is mediated by α⁵GABA^ΑRs. The L-655,708-treated tonic current was not significantly different from control (P = 0.93; n = 5). In addition, the coapplication of bicuculline, L-655,708, and etomidate caused a reduction in I^Holdby 120 ± 11.98% (fig. 2D). This reduction in I^Holdbeyond the baseline indicates that a tonic conductance is present in the absence of etomidate or interventions intended to increase the extracellular concentration of γ-aminobutyric acid.28Interestingly, etomidate and L-655,708, at the concentrations tested, did not influence the kinetics of miniature inhibitory postsynaptic currents (table 1), suggesting that these drugs mediate their effects predominantly by modifying extrasynaptic GABA^AR activity.

The general procedure and the training protocols used to assess fear conditioning are shown in figures 3A and B, respectively. In contextual fear conditioning, mice pretreated with L-655,708 exhibited robust freezing that was similar to that of vehicle-injected control mice (P < 0.01; n = 8/group; fig. 3C). These results suggest that either α⁵GABA^ARs are not important for contextual fear memory or the fear conditioning protocol produced a saturating response under the experimental conditions, such that L-655,708 could produce no further enhancement. The strong contextual fear response was considerably reduced in mice that had been injected with etomidate (P < 0.01 vs.  control; n = 8; fig. 3C). Mice treated with both etomidate and L-655,708 displayed freezing levels comparable to those injected with the vehicle control (P = 0.62; n = 8; fig. 2C). This latter result indicates that decreasing α⁵GABA^AR activity completely reversed the impairment of memory by etomidate.

To address the concern that the initial experimental conditions used to study contextual fear memory produced a saturated freezing response (i.e. , a ceiling effect), the level of the foot shock was reduced (2 s, 0.5 mA). Under these new conditions, the effects of L-655,708 and etomidate were studied in WT and Gabra5 −/− mice. The presentation of the weaker foot shock significantly reduced the baseline freezing in control mice when compared with mice trained with the stronger (2 s, 1 mA) foot shock (freezing with strong foot shock, fig. 3C,vs.  weak foot shock, fig. 3D; P < 0.01). Despite a weak contextual fear conditioning protocol, there were no differences between WT and Gabra5 −/− mice (n = 10 and n = 9, respectively; P = 0.53; fig. 3D). Notably, L-655,708 did not further enhance freezing in WT mice or Gabra5 −/− mice (n = 10 and n = 9, respectively; P = 0.58; fig. 3D). Injections of etomidate reduced the freezing scores for WT but not Gabra5 −/− mice (n = 11 and n = 10, respectively; P = 0.04). The ability of etomidate to reduce freezing scores was reversed by L-655,708 (P = 0.03 compared with control mice; fig. 3D). Consistent with the changes in synaptic plasticity, these findings indicate that α⁵GABA^AR activity is not important for baseline contextual fear conditioning, but these receptors can be activated by etomidate to impair memory performance in this task. Finally, the effects of etomidate and L-655,708 were studied on basolateral amygdala-dependent cued fear conditioning. Neither drug had a significant effect (one-way analysis of variance, P = 0.52; fig. 3E).

In trace fear conditioning, the strength of classic conditioning can be reduced by introducing a time interval (or “trace”) between the tone and the foot shock. For trace fear conditioning, Gabra5 −/− mice treated with vehicle outperformed their WT littermates, as indicated by significantly higher freezing scores (n = 10 and n = 11, respectively; P = 0.03; fig. 3F). To confirm that the difference between the genotypes was attributable to a reduction in α⁵GABA^AR activity, mice were injected with L-655,708, which considerably improved freezing scores for WT mice but had no effect on Gabra5 −/− mice (n = 8/group; P = 0.34; fig. 3F). Interestingly, etomidate did not significantly decrease freezing in WT and Gabra5 −/− mice (n = 10/group) because the freezing scores were similar to those of vehicle-injected mice (P = 0.26; fig. 3F). However, WT mice that received both etomidate and L-655,708 displayed a high level of freezing, which was no different from Gabra5 −/− mice (n = 10 and n = 9, respectively; P = 0.23; fig. 3F).

Next, the Morris water maze was used as an independent measure of hippocampus-dependent learning.26Notably, regardless of drug treatment, the performance of the mice was similar for the acquisition trials of the water maze task (fig. 4A). The mean time savings (time to locate platform during trial 4 minus time required during trial 1) was used to quantify immediate memory. There were no significant differences in the mean time savings to locate the hidden platform between treatment groups (7.2 ± 2.0 s for vehicle-injected mice, 5.7 ± 2.6 s for L-655,708-treated mice, 6.1 ± 3.1 s for etomidate-treated mice, 5.4 ± 3.5 s for etomidate- and L-655,708-treated mice; n = 28/group; P = 0.01; fig. 4A). The results suggest that neither the up- nor down-regulation of α⁵GABA^AR activity influenced the ability of the mice to initially learn and complete the task.

To study long-term memory performance, mice underwent a probe trial 24 h after the acquisition of the task. L-655,708 did not enhance the recall of the hidden platform location, as shown by the percentage of time spent swimming in the correct quadrant of the water maze (n = 28/group; P = 0.35; fig. 4B). In contrast, etomidate decreased the total amount of time spent swimming in the correct quadrant of the pool during the probe trial (n = 28; P = 0.02; fig. 4B). This reduction in performance was not exhibited by the mice that were injected with both etomidate and L-655,708 (n = 28; P = 0.29). Therefore, the mice demonstrated equal learning during the acquisition phase of the water maze task, but etomidate impaired recall of the task 24 h later, and this effect that could be reversed by L-655,708.

The visible platform studies revealed that there were no differences among treatment groups in the latency to locate the platform in the presence of any drug combination (n = 28; P = 0.4; fig. 4C). There were also no differences among the groups in terms of mean swimming speed during the acquisition trial (n = 28; P = 0.87; fig. 4D). The lack of an effect on swimming speed confirmed the procedural ability of the mice to perform the tasks.

The inhibition of α⁵GABA^ARs with L-655,708 has been shown to increase anxiety-like behaviors in the elevated plus maze.17,29However, this anxiogenic effect has been attributed to inhibition of GABA^AR subtypes other than α⁵GABA^AR because nonselective doses were used.29Nevertheless, to determine whether the activity of α⁵GABA^ARs contributed to anxiety-like behaviors that would confound studies of performance in the fear conditioning and Morris water maze tasks, etomidate and L-655,708 were tested in the elevated plus maze at the same doses as used for the memory assays. There were no significant differences in the time spent in the open arms (n = 8/group; P = 0.47; fig. 5A) and the closed arms (n = 8/group; P = 0.31; fig. 5A) of the elevated plus maze. Similarly, there was no difference in the frequency of entries into the open arms (P = 0.21; fig. 5B) and closed arms (P = 0.53; fig. 5C) and closed arms (P = 0.27; fig. 5C), and the frequency of entries into the open and closed arms (data not presented; P = 0.58) were not significantly different. Furthermore, etomidate did not alter the time spent in the open arms (n = 8/group; P = 0.56; fig. 5D) or the closed arms (P > 0.36; fig. 5D), nor did it affect the frequency of entry into either type of arm (data not shown; P = 0.62).

This study supports a pharmacogenetic mechanism to account for resistance to the memory-blocking properties of etomidate. The results show that α⁵GABA^AR activity does not regulate baseline synaptic plasticity evoked by high-frequency stimulation in an in vitro  mouse hippocampus slice model or behavioral performance for contextual fear and spatial navigation memory; nevertheless, etomidate increases α⁵GABA^AR activity and thereby impairs plasticity and memory. These memory-blocking effects of etomidate can be completely reversed by pretreatment with L-655,708.

γ-Aminobutyric acid type A receptors play a critical role in orchestrating neuronal activities by altering spike timing in neurons and synchronized rhythms in neuronal circuits. Etomidate, as well as most inhaled anesthetics, increases the activity of GABA^ARs.7,30This facilitation typically causes membrane hyperpolarization and the shunting of excitatory currents, which reduce neuronal excitability.31The propensity for general anesthetics to block the induction of LTP by increasing GABA^AR activity has been widely reported.32,33Etomidate, at the concentration used for this study blocked LTP through selective potentiation of α⁵GABA^ARs rather than through nonselective enhancement of γ-aminobutyric acid–mediated neurotransmission. Furthermore, nonselectively inhibiting all GABA^ARs with antagonists such as picrotoxin and bicuculline is known to enhance plasticity evoked by high-frequency stimulation34and reverse anesthetic blockade of LTP.32,33This study shows that pretreatment with the α⁵GABA^AR-preferring agent L-655,708 is sufficient to reverse etomidate impairment of LTP and memory blockade.

Etomidate preferentially enhances the tonic rather than synaptic inhibitory conductance in CA1 pyramidal neurons,9and this action can be attenuated by L-655,708. Higher concentrations of anesthetics, which are less selective, may increase both tonic and synaptic inhibition; however, the increase in inhibitory charge mediated by the tonic current is typically many times greater than that mediated by synaptic inhibition.28,35Consequently, we attribute etomidate effects on plasticity and behavior primarily to an increase in the tonic inhibitory conductance, but recognize that inhibitory postsynaptic currents might also contribute. Immunocytoimaging36and electronmicroscopy37studies have shown that α⁵subunits are also expressed in the synaptic regions of hippocampal pyramidal neurons. These synaptic receptors do not appreciably contribute to fast inhibitory postsynaptic currents,38but they may generate a subset of slow synaptic currents that are termed slow inhibitory postsynaptic currents .38The slow inhibitory postsynaptic current contributes to less than 1% of synaptic γ-aminobutyric acid–mediated inhibition but may powerfully regulate plasticity due to its position in the neuronal circuitry and its temporal association with N -methyl-d-aspartate receptor activation.39The influence of low concentrations of etomidate on this slow inhibitory postsynaptic current remains to be studied.

The activity or expression of α⁵GABA^ARs did not seem to influence baseline synaptic plasticity, as was previously reported.9,13Consistent with this result, others have shown that the strength of LTP induced by high-frequency stimulation was similar in hippocampal slices from α⁵(H105R) point mutant and WT mice.40However, another benzodiazepine inverse agonist selective for α⁵GABA^ARs enhanced synaptic plastic in hippocampal slices.41Several factors could account for such discordant results. Notably, although many populations of GABA^ARs are expressed in the neuronal network and within the same cells, only some types may be active during any given experimental condition. With changes in experimental conditions, such as the intensity of the network stimulation, different subpopulations of GABA^ARs may be recruited. Therefore, the contribution of particular GABA^AR populations to synaptic plasticity is critically dependent on the in vitro  experimental protocol. For example, we found that L-655,708 did not alter the baseline synaptic plasticity induced by high-frequency stimulation, whereas others have shown that L-655,708 enhanced plasticity induced by TBS. In the latter study, plasticity was induced by a brief priming stimulus (10 stimuli at 100 Hz) followed 30 min later by TBS.41Notably, after the priming stimulus, synaptic strength was increased to 200% in both the L-655,708-treated and control slices, suggesting that α⁵GABA^ARs do not play a critical role. However, after the second phase of stimulation, L-655,708-treated slices showed greater plasticity. Together, the studies indicate that the stimulation protocol dramatically influences the subpopulation of GABA^ARs that modify plasticity.

The above results also suggest that caution must be exercised when making direct comparisons between studies aimed at understanding the role of GABA^AR subtypes that used different animal species, behavioral protocols, and electrophysiologic measurements. We observed that baseline freezing for fear-associated learning and the Morris water maze task was similar in WT and Gabra5 −/− mice. This result is seemingly at odds with studies showing that a reduction in α⁵GABA^AR activity enhances learning performance.42Others have shown that L-655,708 increased the performance of rats in a water maze probe trial; however, for these experiments, the probe trial was performed 15 min after completion of a series of rigorous training trials.17We studied the probe trial 24 h after training.

Because the strength of associated learning also depends on the strength of the aversive stimulus and the number of presentations,43we sought to determine whether the experimental conditions contributed to the inability to demonstrate improved contextual learning (i.e. , a ceiling effect), and a weaker foot shock was used in some studies. Even under these modified conditions where baseline freezing scores were reduced, L-655,708 did not strengthen contextual learning. Therefore, α⁵GABA^ARs play a minimal role in processes that elicit strong and even moderate contextual memory. In contrast, the performance in trace fear conditioning was greater in Gabra5−/−  mice and WT mice treated with L-655,708 than in WT controls. This result is consistent with studies of α⁵(H105R) point mutant mice, which have a partial deficit of α⁵GABA^ARs.40The α⁵H105R mice exhibit higher baseline freezing scores for trace fear conditioning compared with WT littermates. Trace fear conditioning adds complexity to the delay conditions, because the time interval requires the formation of a temporal relation between the two stimuli. The reasons for differences in trace fear learning but not contextual fear in Gabra5−/−  versus  WT mice remains to be determined because the hippocampus is required for tone-shock association in rodents and humans during contextual learning44and spatial navigation.26 

The concentration of an anesthetic that is required to disrupt behavior is critically dependent on the specific behavioral endpoint under consideration and the methods used to probe behavior. For example, the concentration of inhaled anesthetic that disrupts learning and memory depends on the specific memory tasks used, with hippocampus-dependent learning being particularly vulnerable to disruption by anesthetic drugs.45We showed that memory was impaired during the probe trial; however, during the acquisition tasks, working memory seemed to be intact. Interestingly, working memory is thought to involve the prefrontal cortex. Although early studies suggested that α⁵subunit levels are low in the cortex, a tonic inhibitory conductance is generated by α⁵GABA^ARs in layer 5 of the neocortex.46Also, α⁵GABA^ARs may contribute to inhibitory synaptic transmission in the neocortex.46It is possible that memory tests of higher difficulty or those designed to specifically probe neocortical function may reveal that etomidate modulates these populations of α⁵GABA^ARs.

To reconcile our in vitro  and behavioral data, we developed a schematic model (fig. 6). LTP is mediated, in part, by an increase in the surface expression and function of glutamate receptors in the postsynaptic neuron, and high-intensity stimulation stimulates GABA^ARs, to regulate the induction of LTP. The model proposes the following: (1) α⁵GABA^ARs are expressed in neuronal circuits that normally regulate memory behavior. However, (2) α⁵GABA^ARs regulate network activity in a constrained manner. High-frequency stimulation recruits both high- and low-affinity GABA^ARs, such that the contribution of the α⁵GABA^ARs is overshadowed or obscured by other GABA^AR subtypes. (3) Etomidate preferentially targets α⁵GABA^ARs and “supraactivates” these receptors, causing them to function beyond their normal physiologic limits. Under such conditions, α⁵GABA^ARs exert a dominant role in attenuating plasticity. (4) L-655,708 reverses the effects of etomidate on α⁵GABA^ARs.

The above results provide a compelling foundation for further work, but the studies have important limitations that deserve mention. First, the dose of etomidate was selected to cause amnesia and not general anesthesia. At higher doses, etomidate may modify the activity of other GABA^AR subtypes and other neurotransmitter systems.47,48Second, high-efficacy and affinity-selective α⁵GABA^ARs compounds such as L-655,708 or similar compounds such as α⁵IA, at higher doses, may reduce the activity of α¹, α², and α³subunit-containing GABA^AR subtypes causing anxiogenic and proconvulsant effects that limit their clinical utility.17,29Third, the behavioral paradigms used to study hippocampus-dependent memory do not mimic the clinical scenarios involving etomidate anesthesia. Fourth, α⁵GABA^ARs may play a predominant role under different conditions that evoke plasticity. Finally, it remains to be determined whether α⁵GABA^AR-preferring agents reverse memory impairment by inhaled anesthetics.7 

Preclinical studies show that pathologic conditions, including epilepsy and chronic alcohol abuse, alter expression of the α⁵subunit.49,50Also, polymorphisms of the human Gabra5  gene occur, although their functional significance is still unknown. Mouse models might be useful in developing strategies to treat persistent postanesthetic memory impairment and predict awareness. Finally, the implications of the study extend well beyond the purview of anesthesiology, because the results further implicate α⁵GABA^ARs as targets for the development of memory-modifying drugs.