Opposing Actions of Etomidate on Cortical Theta Oscillations Are Mediated by Different γ-Aminobutyric Acid Type A Receptor Subtypes

Mice of both sexes homozygous for an asparagine to methionine point mutation at position 265 of the GABA^Areceptor β³subunit (N265M) and homozygous wild type controls on the same genetic background as described previously (statistically 87.5% 129/SvJ; 12.5% 129/Sv) were used for this study.14All procedures were approved by the animal care committee (Eberhard-Karls-University, Tuebingen, Germany) and were in accordance with the German law on animal experimentation.

Neocortical slice cultures were prepared from 2- to 5-day-old mice as described by Gähwiler et al.  15,16In brief, for the preparation of somatosensory cortex, animals were deeply anesthetized with halothane and decapitated. Cortical hemispheres were aseptically removed and stored in ice-cold Gey’s solution. After removal of the meninges, 300-μm thick coronal slices were cut. Slices were transferred onto clean glass coverslips and embedded in a plasma clot. The coverslips were transferred into plastic tubes (Nunc) containing 750 μl of nutrition medium and incubated in a roller drum at 37°C. After 1 day in culture, antimitotics were added. The suspension and the antimitotics were renewed twice a week. Cultures were used after 2 weeks in vitro .

Extracellular recordings were performed in a recording chamber mounted on an inverted microscope. Slices were perfused with an artificial cerebrospinal fluid consisting of (in mM) NaCl 120, KCl 3.3, NaH²PO⁴1.13, NaHCO³26, CaCl²1.8, and glucose 11. Artificial cerebrospinal fluid was bubbled with 95% oxygen and 5% carbon dioxide. Artificial cerebrospinal fluid-filled glass electrodes with a resistance of approximately 3 to 5 MΩ were positioned on the surface of the slices and advanced into the tissue until extracellular spikes exceeding 100 μV in amplitude were visible. All experiments were conducted at 34°C.

Test solutions were prepared by dissolving etomidate (Janssen-Cilag, Neuss, Germany) in the artificial cerebrospinal fluid to yield the desired concentration. A closed, air-free system was used to prevent evaporation of the anesthetics.

Etomidate was applied via  bath perfusion using syringe pumps (ZAK, Marktheidenfeld, Germany), connected to the experimental chamber via  Teflon tubing (Lee, Frankfurt, Germany). The flow rate was approximately 1 ml/min. When switching from artificial cerebrospinal fluid to drug-containing solutions, the medium in the experimental chamber was replaced by at least 95% within 2 min. Effects on the spike patterns were stable approximately 5 min later. To ensure steady state conditions, recordings during anesthetic treatment were carried out 10 min after commencing the change of the perfusate.

Data were acquired on a personal computer with the digidata 1200 AD/DA interface and Axoscope 9 software (Axon Instruments, Foster City, CA). The recorded signal was composed of fast action potentials and slow local field potentials (LFP)17,18separated from each other by digital band-pass filtering (200–2000 Hz for action potentials and 1–10 Hz for LFP). Extracellularly recorded spikes were counted offline using self-written programs in OriginPro7 (OriginLab Corporation, Northampton, MA). The mean of spikes occurring during a recording period of 180 s was used as the average spike rate. As firing occurred typically in bursts (episodes of ongoing activity, EOAs), we furthermore calculated the spike rate within a time window 200–1000 ms after the onset of the EOA.

Computation of power-density spectra from the LFP is schematized in figure 1. First, EOAs were cut from the continuous LFP recording, starting where the first peak had decayed to zero. The appropriate window length (1–6 s) was chosen for the control recording of each culture to contain as much as possible of and only the EOA (fig. 1A). Power density between zero and 30 Hz in the EOAs was then determined by fast Fourier transform using the Pwelch method under Matlab 6.5 (MathWorks Inc., Natick, MA). The spectrum was first computed for individual Hanning windows at width 1 fitted to each sweep, with zero padding. The average over these windows was then taken as the power-density spectrum for each EOA. Subsequently an average spectrum for each 180 s recording was obtained by averaging the spectra of the individual EOAs (fig. 1B).

The mean spectrum was then calculated from all the spectra obtained under the same experimental condition (fig. 1C). We observed that small variations in electrode positioning resulted in pronounced differences in the amplitudes of the recorded signal, making it difficult to compare the absolute amplitudes between different recordings. Therefore, all spectra were normalized to the total power density in their spectrum under control conditions. Data acquired in the presence of etomidate were processed in the same way.

Difference spectra were finally obtained by subtracting the mean control spectrum from the mean drug spectrum. We used analysis of variance and post hoc  Student t  test for statistical testing. All results are given as mean ± SEM.

Actions of etomidate were investigated in cultured brain slices derived from the somatosensory cortex of postnatal wild type and β³(N265M) mutant mice. As in our previous studies, spontaneous neuronal activity was induced by removing Mg²⁺ions from the extracellular solution.19,20In figure 2, data from a typical recording are presented: EOAs lasting approximately 3 to 7 s were separated by periods of neuronal silence (fig. 2A). During EOAs spontaneous action potential firing was accompanied by changes in the LFP (fig. 2B). EOAs exhibited a characteristic time structure: at the onset, neurons discharged at a high rate. Simultaneously a biphasic signal occurred in the LFP, exhibiting a large positive and a smaller negative peak (fig. 2C). In the following seconds neurons discharged at lower rates while the LFP was close to baseline, indicating the presence of uncorrelated action potential activity. Towards the end of EOAs an oscillatory component in the LFP developed, arising from progressive synchronization of neuronal activity. During this phase, action potentials appeared time locked to the rising phase of the oscillations in the LFP (fig. 2D).

In slices derived from wild type mice, EOAs occurred at a frequency of 0.128 ± 0.018 Hz (n = 51). In cultures from mutant mice the frequency of EOAs was 0.145 ± 0.017 Hz (n = 66). These mean values were not statistically different (P > 0.5). Similarly, the corresponding discharge rates, calculated from action potential activity during a time period of 180 s in slices from wild type and mutant mice, were not significantly different (7.88 ± 1.06 Hz and 10.29 ± 0.85 Hz, n = 49/67, P > 0.05). In summary, activity patterns monitored in slices from wild type and β³(N265M) mutant mice did not differ under control conditions.

In a recent study we have reported that etomidate depresses spontaneous action potential activity in slices prepared from wild type mice to a significantly larger extent than in slices derived from β³(N265M) mutant mice.14Here, a more detailed analysis is provided on specifically how the anesthetic affects the discharge patterns. Throughout these studies, etomidate was applied at 0.2 μm. At this concentration neuronal activity was depressed in slices from wild type animals by 65% compared with 31% in slices derived from mutant mice.14Example recordings of the effect of etomidate are displayed in figure 3. Etomidate did not change the rate of EOAs (wild type: −11.92 ± 11.36%; mutant: −0.25 ± 6.45%; P > 0.05) but decreased action potential activity by reducing the discharge rate within EOAs.

We further analyzed the time course of depressant action of etomidate within EOAs to find out whether synaptic inhibition mediated by β³-containing GABA^Areceptors is homogeneous within EOAs. As described above the firing pattern is characterized by high firing rates at the immediate onset of an EOA. Therefore we quantified the etomidate-induced depression within the time window 200–1000 ms after the onset of EOAs. Similar values to the average spike rate were obtained when restricting the analysis to this time window (wild type 71.24 ± 5.54% versus  mutant 27.25 ± 19.6%; P < 0.05; n = 9; fig. 4).

Next we investigated the effect of etomidate on local field potentials separately for the onset (early LFP) and towards the end of EOAs. Etomidate decreased the peak amplitude of the early LFP in slices from wild type mice by 30% and in slices from mutant mice by approximately 10%. However, this difference did not reach statistical significance (P > 0.05). Furthermore, the anesthetic shortened the duration of the early LFP. To quantify this effect, we computed the area under the early LFP curve where it deviated from the baseline. This area was reduced by etomidate in slices from wild type mice to a larger extent than in slices from β³(N265M) mutant mice (P < 0.05, n = 9/11; fig. 5).

Etomidate also had a prominent effect on the oscillatory activity developing towards the end of EOAs. To quantify these actions, power-density spectra of the LFP were computed as described above. In figure 6Athe power spectra obtained from LFP recordings in slices from wild type (n = 9) and mutant mice (n = 8) in the absence of etomidate are displayed. A peak close to 6 Hz is observable in both types of preparations. Across all computed frequencies the averaged power densities were not significantly different (P > 0.05). This result provides further evidence that neuronal activity patterns monitored in slices derived from wild type and mutant mice did not differ under control conditions.

In preparations from wild type mice, etomidate is predicted to modulate GABA^Areceptors containing β²or β³subunits. The effect of etomidate on oscillatory activity in the LFP of the wild type is shown in figure 6B: the anesthetic enhanced power densities between 3 and 8 Hz. However, drug-induced amplification of oscillations did not reach statistical significance (P > 0.05). It is remarkable that in the wild type, but not in the β³(N265M) mutant, effects of etomidate displayed a large variability, as indicated by the error bars in figure 6B. Possible variations in the expression of β³subunits, making up only 15–20% of all GABA^Areceptors could serve as an explanation.

In slices from β³(N265M) mutant mice the effects of etomidate are largely restricted to GABA^Areceptors containing β²subunits, assuming that other potential targets are not relevant in this respect. The corresponding spectra are displayed in figure 6C: here, etomidate exhibited the opposite effect and significantly decreased power densities to between 3 and 8 Hz less than control values (P < 0.05).

To provide a direct comparison of the effects of etomidate observed in slices from wild type and mutant mice, difference spectra are displayed in figure 6D. The difference between the computed traces, most prominent between 3 and 8 Hz, is apparently produced exclusively by GABA^Areceptors containing β³subunits.

In the current study we characterized subunit specific actions of 0.2 μm etomidate on neocortical neurons. To judge the relevance of our results for clinical anesthesia, it is necessary to answer the question of whether the anesthetic concentration tested here falls within the clinically relevant range.21Dickinson et al.  22proposed that an aqueous concentration of 1.5 μm etomidate should be close to the EC⁵⁰value for lack of responses to painful stimuli. This concentration represents an uppermost limit because amnesia, sedation, and hypnosis—concordant with a significant depression of cortical neurons in vivo —are achieved with considerably smaller concentrations. Unfortunately, quantitative estimates on etomidate blood concentrations producing amnesia, sedation, and hypnosis are lacking. However, we hypothesize that at 0.2 μm, which is approximately 15% of the concentration causing immobility, the drug might produce amnesia and sedation in vivo .

Dickinson et al.  22quantified the effects of etomidate on gamma oscillation in hippocampal brain slices. These investigators report a 30% decrease in oscillation frequency at 2 μm, which is 10-fold higher than the concentration used in the current work. At first glance, this difference in the effective concentrations is surprising. However, the effects of etomidate on different forms of network activity, namely γ-and θ-range oscillations, have been investigated in the study of Dickinson et al.  22and the current work. The frequency of hippocampal gamma oscillations is largely determined by the decay time of GABA^Areceptor mediated inhibitory postsynaptic currents.23,24Interestingly, Lukatch and MacIver have demonstrated that this is also true for slower neocortical theta oscillations.18An important difference between these rhythms is defined by the kinetic properties of GABA^Areceptor-mediated inhibitory postsynaptic currents: inhibitory postsynaptic currents involved in gamma oscillations have a fast decay time of 10–15 ms compared with inhibitory postsynaptic currents involved in theta oscillations (approximately 100–150 ms).18,23Obviously, different subtypes of the GABA^Areceptor participate in γ- and theta oscillations. It seems possible that subtypes of the GABA^Areceptor, highly sensitive to etomidate, are part of the mechanism producing theta oscillations but of minor importance for gamma oscillations.

Cortical GABAergic interneurons do not exhibit unique firing properties. Although low-threshold-spiking neurons are activated by moderate depolarizing inputs, fast-spiking neurons require a considerable stronger excitation to generate action potentials.5,25Therefore, the latter cells are expected to be active predominantly during the early phase of EOAs when cortical neurons discharge at high rates. Because GABA^Areceptor subtypes show specific patterns of subcellular distribution, activation of specific GABA^Areceptor subtypes should be coupled to the firing of specific types of GABAergic interneurons, which selectively project onto the dendrite, the soma, or axon initial segment of pyramidal cells.26Thus it seemed possible that β²-containing and β³-containing GABA^Areceptors are predominantly activated within specific time windows during EOAs. Therefore we quantified the effect of the drug on the discharge rate during the early phase of EOAs separately. However, we found that the fraction of inhibition mediated via β³subunits in the early phases of EOAs compared to the average depression did not differ (fig. 4). Thus from our current results no conclusions can be drawn regarding what type of interneurons activate β²- or β³-containing GABA^Areceptors.

The data summarized in figures 4 and 5clearly indicate that approximately 50% of the overall depressant effect of etomidate on action potential firing and the early peak in LFP are mediated by β³-containing GABA^Areceptors. This result is somewhat surprising because β³subunit-containing receptors clearly constitute a minor fraction of all GABA^Areceptors. However, their high impact on neuronal activity can be explained by the slower decay kinetics of functional GABA^Areceptors containing β³subunits compared with receptors possessing β²subunits.27,28Furthermore, there is evidence that β³receptors are predominantly located on pyramidal cells whereas β²receptors are also found on GABAergic interneurons.26,29Inhibition of GABAergic neurons via β²-containing receptors should decrease GABA release on pyramidal cells and therefore increase their excitability. Finally, the location at strategically important sites such as the axonal hillock should make a considerable difference with regard to the impact of specific GABA^Areceptor subtypes. In conclusion, physiologic impact and relative abundance of diverse GABA^Areceptor subtypes may largely differ.

A large body of evidence indicates that theta activity has a central role in hippocampal learning.12,30More recently, it has been shown that neocortical neurons display theta activity during working memory tasks.8–10However, although data suggesting that cortical theta activity is an important neuronal correlate of memory and learning processes are continuously accumulating, causal relationships are far from being understood. This is also true for the effects of anesthetic agents on learning and memory, although interest in this area of research is increasing. Surprisingly, anesthetic drugs diminish working memory performance when administered during a learning task, whereas post-training exposure to anesthetic agents may improve learning.31,32Obviously, the effect of a single anesthetic strongly depends on the particular neuronal process that is affected. It is therefore important to investigate how specific types of network activity are affected by anesthetic agents.

In the current study we have provided evidence that β²-containing and β³-containing GABA^Areceptors exhibit opposing actions on theta activity in neocortical slice cultures. Overall enhancement of this kind of network activity results from balanced activation of these GABA^Areceptor subtypes, as indicated by the results displayed in figure 6. The involvement of different GABA^Areceptor subtypes in mediating specific actions on cortical neurons might also help to explain the finding of Veselis et al.  3that at concentrations producing a similar degree of sedation, anesthetic drugs acting predominantly via  GABA^Areceptors display different potencies in affecting working memory performance of human subjects. Taken together, these results raise the possibility that different subtypes of cortical GABA^Areceptors contribute in different ways to the sedative and amnestic properties of intravenous anesthetics.

The authors thank Ina Pappe and Claudia Holt (Technical Assistants, Eberhard-Karls-University, Tuebingen, Germany) for excellent technical assistance.