Gamma-aminobutyric acid type A (GABA(A)) receptor-mediated inhibition in the central nervous system exists in two forms: phasic (inhibitory postsynaptic currents, IPSCs) and tonic (nonsynaptic). Phasic inhibition is further subdivided into fast (GABA(A,fast)) and slow (GABA(A,slow)) IPSCs. By virtue of its dendritic location and kinetics, GABA(A,slow) has been proposed to control synaptic plasticity and memory. Etomidate is a nonbarbiturate, intravenous anesthetic that selectively modulates GABA(A) receptors and produces amnesia at low doses in vivo. This study tested whether correspondingly low concentrations of etomidate in vitro alter GABA(A,fast) and GABA(A,slow) phasic inhibition.
Electrophysiological recordings were obtained from hippocampal slices prepared from postnatal day 3-8 mice and maintained in organotypic culture for 10-14 days. Etomidate was applied at concentrations corresponding to one-half to four times the half maximal effective concentration that impairs hippocampus-dependent learning and memory--i.e., 0.125-1.0 microm.
Etomidate 0.25 microm (the half maximal effective concentration) doubled the time constant of decay of GABA(A,slow) IPSCs, but it had no detectable effect on GABA(A,fast) IPSCs. Higher concentrations of etomidate had stronger effects on both types of phasic inhibition: 0.5 and 1 microm etomidate prolonged the time constant of decay by 310% and 410% for GABA(A,slow) and by 25% and 78% for GABA(A,fast). Concentrations of etomidate up to 1 microm had no significant effects on the amplitudes of either GABA(A,fast) or GABA(A,slow) IPSCs.
At concentrations that impair hippocampus-dependent memory, etomidate modulates GABA(A,slow) more strongly than GABA(A,fast) IPSCs. Effects of etomidate on GABA(A,slow) IPSCs may contribute to etomidate-induced amnesia.
MODULATION of γ-aminobutyric acid type A (GABAA) receptor-mediated inhibition is thought to play a key role in the mechanism of action of general anesthetics.1GABAAreceptors exist in a variety of subunit combinations that confer distinct physiologic and pharmacological properties to the individual receptors.2Moreover, it has been recently recognized that GABAAreceptor-mediated inhibition presents in three distinct forms: two transient (phasic) types mediated by synaptic receptors, and one continuous (tonic) type mediated by receptors that are largely extrasynaptic.3Therefore, even anesthetics with a relatively narrow and well-defined receptor-level activity profile (such as etomidate and propofol)1may differentially engage different types of GABAA-ergic inhibition to achieve the desirable components of the anesthetic state (e.g. , sedation, amnesia, hypnosis, and immobility).
Significant progress has been made recently in understanding the chain of events linking molecular to behavioral actions of etomidate. Evidence obtained from knockout animals has suggested a role for α5 subunit-containing GABAAreceptors in the control of learning and memory.4,5The α5 subunit is expressed most strongly in the hippocampus, a structure that plays a central role in the formation of declarative memories. Located at extrasynaptic sites on pyramidal neurons, receptors containing α5 subunits have been shown to mediate tonic inhibition.6The finding that tonic inhibition is enhanced at low anesthetic concentrations, together with the demonstration that mice lacking the α5 subunit are resistant to the amnestic action of etomidate, thus pointed to tonic inhibition in the hippocampus as the principal mechanism for this behavioral effect.7
However, recent findings indicate that receptors containing α5 subunits are also located at dendritic synapses8and that they underlie a slow form of synaptic (phasic) inhibition in hippocampal CA1 pyramidal neurons.9,10By virtue of its dendritic location and slow kinetics, which match those of N -methyl-d-aspartate receptor–mediated excitation, this inhibitory current (which has been termed GABAA,slow) is well suited to control synaptic plasticity and memory formation.11,12It represents an additional or alternative effector mechanism by which etomidate and other drugs that enhance GABAAreceptor function may produce amnesia.
To test whether GABAA,slowshows the requisite sensitivity to etomidate to be considered a viable candidate for contributing to etomidate-induced amnesia, we examined the effects of a range of concentrations of etomidate bracketing the half maximal effective concentration that impairs hippocampus-dependent learning and memory (EC50-amnesia) on GABAA,fastand GABAA,slowinhibitory postsynaptic currents (IPSCs). We found that in contrast to GABAA,fastIPSCs, which were unaffected or only weakly modulated by etomidate at amnestic concentrations, GABAA,slowIPSCs were strongly enhanced. Therefore, along with tonic GABAergic inhibition, GABAA,slowrepresents a potentially important effector mechanism for suppression of hippocampal learning and memory.
Materials and Methods
Organotypic Hippocampal Slice Cultures (OTChip)
All experiments conformed to the guidelines laid out in the Guide for the Care and Use of Laboratory Animals, and were conducted with the approval of the University of Wisconsin-Madison (Madison, Wisconsin) Animal Care and Use Committee.
OTChip were prepared from 3- to 8-day-old B6/129 hybrid mice as described by Stoppini et al. 13In brief, mice were anesthetized with isoflurane and decapitated. The brain was quickly removed and immersed in ice-cold dissecting medium containing 50% Gey’s balanced salt solution, 10 mm MgCl2, and 60 mm glucose. The brain was then transferred to a dissecting chamber, blocked, and mounted on the stage of a vibrating microtome (Vibratome series 1000; Ted Pella, Inc., Redding, CA). Brain slices were cut 350 μm thick and placed on MilliCell-CM membrane inserts (Millipore, Billerica, MA) in a 6-well culture plate. Each well contained one membrane insert and 1 ml of culture medium. The culture medium was composed of 25% Hank’s salt solution, 25% horse serum with 30 mm D-glucose, 50% minimum essential medium with Earl’s salts, L-glutamine, and sodium bicarbonate. After 1 day in culture, cytostatics (10 μm Uridine, 10 μm ARA-C, and 10 μm 5-Fluoro-2′deoxyuridine; Sigma-Aldrich, St. Louis, MO) were added to the culture medium. The slice cultures were maintained in an incubator at 36°C in 5% carbon dioxide. The culture medium was exchanged (90% of volume) twice per week, with cytostatics added to the medium once weekly. Slices were used between 10 and 14 days in culture, at which time hippocampal structures remained easily identifiable (fig. 1A).
A brain slice recording chamber was mounted on the stage of an upright microscope (Olympus BX50WI; Olympus America Inc., Center Valley, PA) with a long working distance water immersion objective and differential interference contrast optics, which was connected to a video camera (Hamamatsu C2400; Hamamatsu, Tokyo, Japan). A single OTChip was removed from the six-well culture plate by cutting the membrane from the well with a scalpel blade, and it was then transferred to the recording chamber by using a Pasteur glass pipette. For experiments, slices were continuously superfused with artificial cerebrospinal fluid at 34 ± 1°C bubbled with carbogen (95% O2+ 5% CO2).
Whole cell pipettes were fabricated from borosilicate glass (KG33; Garner Glass, Claremont, CA). Pipettes were filled with a solution containing (in mm): CsCl 135 (for evoked IPSCs recording, CsCl was partially replaced by 40 mm K-gluconate), Na-HEPES 10, EGTA 10, MgATP 3, GTP 0.5, lidocaine N -ethyl bromide (QX-314) 5, pH = 7.25. Whole-cell patch clamp recordings were obtained from neurons within the pyramidal cell layer of the CA1 area that had pyramidal cell-like morphology. Recordings generally remained stable for up to 1.5 h at a holding potential of –50 to –60 mV. Signals were amplified using a MultiClamp 700A amplifier (Axon Instruments, Foster City, CA) and ClampEx software (Axon Instruments), filtered at 5 kHz and sampled at 10 kHz by using a Digidata 1322A (Axon Instruments). Open tip resistance ranged from 1.5–4 MΩ. Whole-cell access resistance was less than 15 MΩ before compensation by 50–80%.
In the absence of any excitatory or inhibitory antagonists, recordings revealed frequent large-amplitude spontaneous inward currents (fig. 1B). In chloride-loaded neurons, these events include both GABAergic IPSCs and glutamatergic excitatory postsynaptic currents. Even in the presence of 20 μm 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 40 μm (2R)-amino-5-phosphonovaleric acid (APV), which together block all ionotropic glutamatergic synaptic input, there remained sufficiently frequent and large spontaneous fast IPSCs that individual events were difficult to separate for kinetic analysis (fig. 1C). Therefore, to characterize the effects of etomidate on GABAA,fast, we examined miniature IPSCs (mIPSCs) by adding tetrodotoxin to the superfusate (fig. 1D). To characterize the effects of etomidate on GABAA,slowIPSCs, we placed a glass electrode filled with artificial cerebrospinal fluid in the stratum lacunosum-moleculare (SL-M, fig. 1A) and stimulated using current pulses of 40–60 μA, 0.1 ms in duration, at a frequency of 0.05 Hz.
All drugs were purchased from Sigma-Aldrich. Etomidate was dissolved in dimethyl sulfoxide, frozen, and freshly dissolved in artificial cerebrospinal fluid before adding it to the bath perfusion by using a syringe pump with the flow rate adjusted to yield the desired concentration.
Data were analyzed offline on a personal computer by using Mini Analysis (Synaptosoft, Decatur, GA) and ClampFit (Molecular Devices, Sunnyvale, CA). The threshold for event detection was set at three times the root mean square noise level. Events were analyzed by automated event detection that acquires amplitude, 10–90% rise times, and the time to 63% decay.
The decay phase of averaged mIPSCs was fit with a double exponential function using Microcal Origin (version 6.0, 7.0; OriginLab Corporation, Northampton, MA). The weighted time constant (τwt) was calculated as [A1τ1+ (1 − A1)τ2], where A1is the amplitude of the fast component of the biexponential equation and where τ1and τ2are the time constants of the fast and slow components, respectively. The decay phase of averaged GABAA,slowIPSCs was fit with a single exponential function.
Concentration-response data were fit by using microcal origin to the logistic function:
i =i max + (i min −i max)/(1 + ([etomidate]/EC50)n )
where i is the current amplitude normalized to the control response, i max is the maximal current amplitude, i min is the minimum current amplitude, EC50 is the concentration of etomidate eliciting half maximal current amplitude, and n is the slope factor. During the fitting procedure, all values were allowed to float freely, with the exception of i min, which was fixed to a value of “1.”
Statistical analysis was performed by using Microcal Origin (version 7; OriginLab Corporation) or GraphPad Prism (version 7; GraphPad Software, San Diego, CA). Data are presented as mean ± SD. Statistical comparisons were made by using a one-tailed Student t test to examine effects of etomidate on decay, and a two-tailed t test to examine effects of etomidate on amplitude. Effects were considered significant at P < 0.05. Fits to the logistic equation were compared by examining 95% confidence intervals; nonoverlapping confidence intervals indicated significant differences.
Etomidate Slows the Decay of Evoked GABAA,slowIPSCs
We examined the effects of etomidate on GABAA,slowIPSCs at concentrations ranging from 0.125–1 μm (i.e. , 0.5–4 times EC50 amnesia) by using electrically evoked, action potential–dependent responses. At its EC50–amnesia concentration, etomidate markedly prolonged the decay time constant (τdecay) of GABAA,slowIPSCs: 0.25 μm slowed τdecayapproximately two-fold (fig. 2A). The effect was rapid in onset, reversible (fig. 2B), and concentration-dependent (fig. 2C). Figure 3summarizes the effects of etomidate on the amplitude and τdecayof GABAA,slowobtained from 6 cells. Over the range of concentrations tested, etomidate slowed IPSC decay in a concentration-dependent manner (fig. 3, left axis; P < 0.05 for all concentrations); 0.5 and 1 μm etomidate prolonged τdecayby 310 and 410%. By contrast, etomidate did not alter the amplitude of evoked GABAA,slowIPSCs at any concentration (fig. 3, right axis; P > 0.05).
Miniature GABAA,fastIPSCs: Relative Resistance
We examined the effects of etomidate on GABAA,fastmIPSCs at concentrations ranging from 0.25–1 μm (i.e. , 1–4 times EC50 amnesia). Examples of action potential-independent mIPSCs under control conditions and in the presence of 0.5 and 1 μm etomidate are shown in figure 4A. Individual mIPSCs were characterized by their amplitude and interevent interval and plotted as cumulative distributions (fig. 4, B and C). In addition, several hundred events under each condition were averaged to evaluate their kinetics (fig. 4D). Just as we found for GABAA,slowIPSCs, the amplitude of GABAA,fastmIPSCs was not changed by etomidate (fig. 4B). The aggregate data confirmed that mIPSC amplitude was unchanged by 1 μm etomidate, the highest concentration tested (33.6 ± 14.1 pA and 32.2 ± 14.3 pA, for control and drug, respectively, n = 4, P > 0.05). Similarly, there was no significant effect of etomidate on mIPSC frequency (interevent interval 47.3 ± 10.3 ms and 58.9 ± 10.1 ms for control and 1 μm etomidate, respectively, P > 0.05). However, 0.5 and 1 μm etomidate did appear to slow GABAA,fastmIPSC decay modestly (fig. 4D).
To characterize the effect of etomidate on decay kinetics of GABAA,fastin detail, we fit the averaged currents to biexponential functions (table 1). This analysis showed that the effect of etomidate was restricted to a slowing of the slow component (τ2) of the biexponential decay, with no changes in either the relative amplitude (A1) or the decay rate (τ1) of the fast component. The value of τ2was nearly doubled (from 12.3 to 21.1 ms) at 1 μm, but there was no appreciable effect of etomidate on τwtat EC50 for amnesia (0.25 μm; table 1, P > 0.05).
Differential Effects of Etomidate on GABAA,fastversus GABAA,slow
Our experiments showed that etomidate did exert pronounced effects on the kinetics of GABAAreceptor-mediated synaptic currents but not on peak amplitude (or frequency, for GABAA,fastmIPSCs). However, the degree to which the decay time constant (and hence the charge transfer) was altered differed substantially between the two forms of phasic inhibition. The difference is illustrated in figure 5, which compares τdecayof GABAA,slowwith the weighted time constant of decay of GABAA,fast(τwt).
To compare the overall effects of etomidate on the two different types of phasic inhibition, we compared the parameters of the fits of the logistic equation to these two data sets. Both i max (GABAA,slow4.3 ± 0.9 vs. GABAA,fast1.9 ± 0.1) and EC50 values (GABAA,slow0.37 ± 0.17 μm vs. GABAA,fast0.61 ± 0.07 μm) favored greater modulation of GABAA,slowcompared to GABAA,fast, with nonoverlapping 95% confidence intervals supporting a statistical difference in these parameters. The steepness of the two curves did not differ (GABAA,slow2.5 ± 1.0 vs. GABAA,fast3.8 ± 1.3) These results suggest that, to the extent that etomidate exerts its amnesic action via synaptic GABAA-ergic inhibition in the hippocampus, it involves primarily GABAA,slow.
Our principal finding is that etomidate, at concentrations associated with amnesia in vivo , enhances GABAA,slowinhibition to a substantially greater extent than GABAA,fast. Therefore, we conclude that GABAA,slowdoes have the requisite sensitivity to serve as an effector mechanism for etomidate-induced impairment of hippocampal-dependent learning and memory.
The Organotypic Culture (OTC)
Since its introduction two decades ago,14different types of OTCs have been studied from a variety of brain areas. They have become accepted as useful preparations, complementary to acute brain slices and profitably employed in the study of synaptic transmission, plasticity, and long-term drug exposure.15In the context of anesthetic mechanisms, hippocampal OTCs have been used primarily to study the effects of anesthetics on spontaneous network activity.16–18
The OTChip offered some advantages as a preparation for this project. The hippocampus plays a central role for the formation of explicit memories and offers important targets for amnesia-inducing drugs acting via inhibitory and excitatory receptors. Inhibition in the OTChip resembles that in age-matched acute hippocampal slices.19However, due to the thinness of the preparation (just several cell layers, or 100-μm-thick, by 1–2 weeks in culture), even lipophilic drugs such as etomidate equilibrate quickly enough to allow experiments to be performed at known tissue concentrations within the time limitations imposed by whole-cell recording techniques. Indeed, we found that etomidate’s effect on the most sensitive parameter, the time constant of decay of GABAA,slowIPSCs, stabilized within 10 min of drug application (fig. 2). This is substantially faster than the 1–2 h required in experiments with thicker acute brain slices.20
One important difference between the OTC and the more commonly used acute brain slice preparation is that the rate of ongoing spontaneous inhibitory currents is so high in the OTC that essentially all spontaneous GABAA,fastISPCs overlap temporally (fig. 1). This high frequency of spontaneous synaptic currents complicated kinetic analysis of spontaneous GABAA,fastIPSCs and completely obscured spontaneous GABAA,slowIPSCs. Thus, it precluded a direct comparison of these types of IPSCs under identical conditions. However, we were able to characterize the effect of etomidate on GABAA,fastcurrents by measuring miniature IPSCs because the addition of tetrodotoxin resulted in a lower frequency that permitted kinetic analysis. Since miniature GABAA,slowIPSCs occur too infrequently to use for this purpose,21instead we measured evoked GABAA,slowIPSCs. Although these conditions of activation did differ for the two forms of phasic inhibition, raising the possibility that the observed difference in their modulation reflected the methods used to activate them rather than their intrinsic synaptic responsiveness, we feel that this explanation is unlikely. Our previous studies have shown that the kinetic characteristics of each of the two types of IPSCs are independent of whether they are expressed as miniature, spontaneous, or evoked responses. Rather, their differences reflect intrinsic properties of the two types of synapses, which arise from different classes of presynaptic neurons, utilize different postsynaptic receptors, and produce different neurotransmitter profiles. Which of these factors accounts for the differences in modulation of GABAA,fastversus GABAA,slowby etomidate remains unknown.
The Physiology and Pharmacology of Slow Phasic Inhibition
Synaptic currents of the GABAA,slowtype were discovered in hippocampal neurons, where they were found to underlie the “early IPSP” and to control the excitability of pyramidal cells.11Subsequently, similar synaptic currents were found in other brain areas as well.22–24To date, their properties are best characterized in the hippocampus.25–27
The slow time course of decay of GABAA,slow(30–70 ms as opposed to 3–8 ms for GABAA,fastat 36°C) is its most striking characteristic. A number of explanations for the slow decay, as well as for its functional implications, have been proposed. The time course could be due to the intrinsic kinetic properties imparted by the subunit combination of the underlying receptors. This includes α5subunits, probably in combination with β3and γ2subunits.9,10Their high affinity, which is associated with a slow unbinding rate, would be expected to impart slow deactivation, even after a brief stimulus. However, additional subunits, primarily α1and α2, also have been shown to contribute to GABAA,slowIPSCs,9so other factors evidently also contribute to their slow decay. One such factor is likely to be the prolonged presence of transmitter at these synapses. This could be produced by either of two mechanisms. First, there may be spillover of transmitter onto high-affinity receptors located at perisynaptic or extrasynaptic sites. This would be analogous to the situation in granule cells of the dentate gyrus, where it has been shown that the ‘slow’ time course of IPSCs in the molecular layer arises from spillover onto perisynaptically located δ-subunit containing receptors.28In the hippocampal CA1 area, α5-containing receptors have been found perisynaptically,9lending support to this possibility. Alternatively, the release of GABA into a large-volume synaptic cleft, possibly surrounded by a glial wrap that limits diffusion out of the cleft, may lead to the prolonged presence of a low concentration of transmitter. This mechanism has been proposed for the GABAA,slowIPSC produced by neurogliaform cells on neocortical pyramidal cells.23Multiple types of interneurons can produce slow IPSCs, including neurogliaform,23,27oriens lacunosum-moleculare,29and ivy cells30; therefore, it is possible that different types of slow synapses employ each of these two proposed mechanisms.
In addition to its slow kinetics and dendritic location, another important feature of GABAA,slowthat supports its role in controlling synaptic plasticity is its susceptibility to modulation by presynaptic GABABreceptors. Unlike GABAA,fastIPSCs, GABAA,slowIPSCs are strongly suppressed by repeated stimuli at frequencies that are effective in producing long-term potentiation.25This presynaptic effect, which is mediated by GABABautoreceptors, leads to unopposed excitation. The resulting dendritic depolarization relieves a voltage-dependent block of N -methyl-d-aspartate receptors by magnesium, permitting calcium entry and activating an ensuing cascade of calcium-dependent processes that ultimately leads to a persistent increase in synaptic strength.31In the presence of the GABABantagonist CGP35348, GABAA,slowinhibition is maintained, and this prevents the depolarization required for induction of long-term potentiation.32The ability of GABABantagonists to impair long-term potentiation demonstrates the effectiveness of GABAA,slowin controlling dendritic depolarization and synaptic plasticity. The enhancement of GABAA,slowby amnestic concentrations of etomidate and other drugs may similarly prevent memory formation by enhancing the hyperpolarizing influence of dendritic inhibitory inputs, thereby maintaining a hyperpolarized dendritic membrane potential and preventing sufficient depolarization to initiate synaptic plasticity. However, even if these drugs do modulate the same receptors or synapses and even if these actions do contribute to their amnestic properties, different classes of drugs might achieve amnesia by way of quantitatively different balances of effects on IPSCs versus other targets – or even via qualitatively different types of effects on IPSCs. For example, whereas we found that the major effect of etomidate was to slow GABAA,slowIPSC decay, benzodiazepines were shown recently to increase GABAA,slowamplitude rather than decay,9and halothane-modulated GABAA,fastand GABAA,slowcurrents similarly, altering decay but not amplitude.33
GABAAReceptor α5Subunits, Tonic and Phasic Inhibition, and Amnesia
The existence of a tonic GABAAreceptor-mediated conductance was initially documented in cerebellar granule cells.34,35In those neurons, tonic current arises from receptors located within a glomerulus that are exposed to relatively high and sustained levels of GABA. Similar currents have been identified in a number of other neuron types, including thalamic,36cortical,37dentate gyrus,38,39and hippocampal40neurons. Though these cells lack similar specialized structures, they do contain receptors that are sufficiently sensitive to GABA that ambient levels of neurotransmitter in the extracellular space can activate them.
In hippocampal CA1 pyramidal neurons, tonic current is mediated by GABAAreceptors that contain α5subunits.6,9This current is enhanced by a variety of anesthetics, including amnestic concentrations of isoflurane and etomidate.7,41Behavioral experiments, paired with genetic4,42,43and pharmacologic44manipulations, strongly support a memory-modulating role of α5-containing GABAAreceptors in hippocampal learning paradigms. Furthermore, mice lacking α5subunits are resistant to the effects of etomidate on memory in behavioral experiments.7Taken together, this evidence has pointed to tonic current carried by α5subunit-containing GABAAreceptors as playing a causal role in anesthetic-induced amnesia. However, since the α5subunit-containing receptors that contribute to GABAA,slowIPSCs in CA1 pyramidal cells may very well be the same high-affinity, slowly decaying receptors that produce tonic current in these cells,6slow phasic inhibition provides an additional, or alternative, explanation for these findings.
Our results do not provide experimental support for a prominent role of fast phasic inhibition in etomidate-induced amnesia. In this respect, they are in agreement with previous work that used a genetic approach to conclude that GABAA,fastIPSCs do not play an important role for this anesthetic endpoint.7However, the finding that the decay of fast IPSCs is slowed by etomidate at higher concentrations (fig. 4Dand fig. 5) is compatible with a role for modulation of GABAA,fastIPSCs in other anesthetic end points, such as impaired perception or unconsciousness. Indeed, the decay of fast IPSCs is an important determinant of oscillation frequency in the γ range (40–90 Hz),45a rhythm that has been tied to cognitive processes.46Thus, anesthetic modulation of distinct forms of GABAAreceptor-mediated inhibition may contribute to specific components of the anesthetic state.
The authors thank Anna Bassuener, B.S. (Research Specialist, Department of Anesthesiology, University of Wisconsin, Madison, Wisconsin), Pee Yang (Undergraduate Student, University of Wisconsin), and Mark Perkins, B.S. (Senior Research Specialist, Department of Anesthesiology, University of Wisconsin), for excellent technical assistance.