Amnestic Concentrations of Sevoflurane Inhibit Synaptic Plasticity of Hippocampal CA1 Neurons through γ-Aminobutyric Acid–mediated Mechanisms

As approved by the Institutional Animal Care and Use Committee (Gunma University Graduate School of Medicine, Maebashi City, Japan), Sprague-Dawley rats at 30–35 days old (mean body weight, 100 g) were decapitated during anesthesia, and the head was immediately immersed in cold (1–4°C) modified Ringer's solution, comprising 234 mm sucrose, 2.5 mm KCl, 1.25 mm NaH²PO⁴, 10 mm MgSO⁴, 0.5 mm CaCl², 26 mm NaHCO³, and 11 mm glucose saturated with 95% O²and 5% CO². A block of tissue containing the hippocampus was dissected out quickly and glued to a DTK-1000 vibratome tray (Dosaka EM, Tokyo, Japan) using oxygenated cold modified Ringer's solution. Hippocampal slices (thickness, 500 μm) were cut from the brain and then kept in the prechamber (Brain Slice Chamber System; Harvard Apparatus, Holliston, MA) filled with artificial cerebrospinal fluid consisting of 125 mm NaCl, 2.5 mm KCl, 2 mm CaCl², 1 mm MgCl², 26 mm NaHCO³, 1.25 mm NaH²PO⁴, and 11 mm glucose, bubbled with 95% O²and 5% CO²at room temperature (22°–24°C). Slices were allowed at least 1 h for recovery in the prechamber, which was designed to keep 8–12 slices viable for several hours.

The methods of brain slice electrophysiology have been described previously.15,18,19Briefly, slices were transferred to a submerged recording chamber (1 ml in volume, Brain Slice Chamber System) and perfused at a rate of approximately 3 ml/min with oxygenated artificial cerebrospinal fluid. PSs and fEPSPs were evoked using bipolar tungsten stimulating electrodes (1–2 MΩ; World Precision Instruments, Sarasota, FL) placed in the stratum radiatum of CA1 regions to activate Schaffer-collateral-commissural fibers under a dissection microscope (SD30 and SZ-STU1 base unit; Olympus, Tokyo, Japan). Electrical stimuli comprised square wave paired pulses (duration, 0.35 ms; interpulse interval, 60 ms) delivered by a SEN 3301 stimulus generator (Nihon Kohden, Tokyo, Japan) connected through an SS-202 J stimulation isolation unit (Nihon Kohden) to electrodes every 15 s, to minimize frequency-dependent changes in synaptic transmission. Stimulus intensity was set to evoke 50–70% of the maximal amplitude of responses. In the LTP study, the intensity of stimulation used was 40–50% of maximum amplitude to avoid a ceiling effect. LTP was induced using a tetanic stimulation (100 Hz, 1 s). In all experiments, baseline recordings were monitored for at least 20 min before drug application or delivery of tetanic stimulation.

All fEPSPs and PSs were recorded using a glass microelectrode filled with artificial cerebrospinal fluid (GDC-1.5; OD, 1.5 mm; ID, 1.0 mm; Narishige, Tokyo, Japan). Recording electrodes were placed directly in line with stimulating electrodes in striatum radiatum using an MO-150 micromanipulator (Narishige), but separated from stimulating electrodes by 1.0–1.5 mm, or placed in stratum pyramidale to record PS field potentials. Signals were recorded using a DAM-80 AC differential amplifier (World Precision Instruments), amplified at least 1,000-fold, filtered at 2–5 kHz, and digitized at 10 kHz using a Digidata 1322A system (Axon Instruments, Union City, CA). All data were stored on a Pentium-based PC for later analysis. All experiments were performed at room temperature (21°–24°C).

Artificial cerebrospinal fluid solution at room temperature was bubbled with a carrier gas (95% O², 5% CO²) passing through a calibrated commercial vaporizer (Sevotec 5; Ohmeda, BOC Health Care, West Yorkshire, United Kingdom) at the designated concentration and was applied to the recording chamber using a gravity-feed and vacuum system. High-quality polytetrafluorethylene was used for tubing and valves to minimize loss of volatile anesthetic and drug binding. Sevoflurane (Maruishi Pharmaceutical, Osaka, Japan) concentrations used for this study were 0.5–5.0%. To determine the actual aqueous concentrations of sevoflurane in the submerged recording chamber for each concentration used, aliquots of the solution were taken from the recording chamber and filled into airtight glass containers for gas chromatographic measurements as described previously.20 Figure 1shows the final aqueous sevoflurane concentrations in our recording condition (n = 5, each). A strong positive correlation (r = 0.99971) was observed; for example, 0.5, 2.8, and 5.0% of sevoflurane were actually 0.04, 0.23, and 0.41 mm, respectively. Therefore, 0.04 mm was used as an amnestic dose, and 0.23–0.41 mm was used as clinically relevant concentrations. Drugs were purchased from Sigma-Aldrich Chemicals (Tokyo, Japan).

Data acquisition and analysis were performed with pCLAMP version 8.1 software (Axon Instruments) and IGOR Pro version 5.0 software (WaveMetrics, Lake Oswego, OR). Amplitudes of PSs were measured from peak positive to peak negative voltages. Fiber volley amplitudes were measured from baseline to peak negative voltage. Rise times and decay times of fEPSPs were measured as time intervals between 10% and 90% of peak amplitude and as the decay time from peak amplitude to half amplitude, respectively. In PS and fEPSP recordings, responses were normalized as a percent of control, based on mean amplitude during a 10-min recording immediately before the perfusion of anesthetic. In the LTP study, after 10 min of control recordings, sevoflurane was administered for 20 min before providing an LTP-inducing stimulus, in the continued presence of the anesthetic.

The statistical significance of data were determined using a Student t  test or one-way analysis of variance with a Tukey post hoc  test to compare the differences from three or more groups and to determine anesthetic effects with time-matched control responses. Data are expressed as mean ± SD. P  values less than 0.05 were considered statistically significant.

Sevoflurane (0.04–0.23 mm) depressed synaptically evoked discharge of CA1 neurons, measured as a block of PS responses (figs. 2A and B). The following study focused on two concentrations: amnestic dose (0.04 mm) and clinically relevant concentrations of sevoflurane (0.23 mm). Sevoflurane produced a dose-dependent depression of PS amplitude (57.3 ± 15.0% of control at 0.04 mm and 26.9 ± 6.19% of control at 0.23 mm, respectively; P < 0.001, n = 5 slices from individual rats, compared with preanesthetic control responses, using analysis of variance–Tukey). These effects were reversible on washout of the agent with drug-free artificial cerebrospinal fluid. PS responses completely disappeared when a higher dose (0.41 mm) of sevoflurane was applied (n = 5).

To determine whether the effects of sevoflurane on PSs involve GABA^Areceptors, we examined the effects of sevoflurane on PS responses in the presence of bicuculline, a GABA^Areceptor antagonist. The effects of bicuculline alone on PS responses were initially tested in the absence of sevoflurane. Although bicuculline promoted epileptoid activity in PS responses and made it difficult to measure the amplitude correctly, bicuculline (10 μm) increased PS amplitude to 127.2 ± 8.1% of control (n = 5, P < 0.01). Next, bicuculline (10 μm) was applied in the continued presence of sevoflurane to attempt reverse depression of PS amplitude. Bicuculline (10 μm) reinstated 0.04 mm sevoflurane–induced PS depression to 94.3 ± 9.2% of control (n = 5; fig. 2A). However, bicuculline (10 μm) only partially reinstated 0.23 mm sevoflurane–induced depression to 51.8 ± 16.0% of control (n = 5; fig. 2B). Taken together, γ-aminobutyric acid–mediated (GABAergic) inhibition seemed to play an important role, at least in part, in PS depression with amnestic doses of sevoflurane.

To determine whether sevoflurane-induced PS depression resulted from depressed glutamate-mediated excitatory synaptic inputs to CA1 neurons, fEPSPs were recorded from dendritic regions in the stratum radiatum. Sevoflurane depressed the amplitude of fEPSP responses in a dose-dependent manner (figs. 3A and B), with 0.23 mm sevoflurane to 70.7 ± 11.7% of control (P < 0.05, n = 10) and 0.41 mm sevoflurane to 41.9 ± 19.2% (P < 0.001, n = 10). Bicuculline (10 μm) did not reverse fEPSP depression produced by sevoflurane at any concentrations tested (up to 0.41 mm; fig. 3C). Depressed glutamate-mediated synaptic excitation thus seemed to play an important role in PS depression produced by clinical concentrations of sevoflurane. To determine the effects of sevoflurane on nerve conduction velocity, the duration from electrical stimulation to fiber volley response was measured. Sevoflurane up to 0.41 mm did not change the duration from electrical stimulus to fiber volley amplitude throughout the experiments (figs. 3A and B). Bicuculline alone had no effect on the amplitude of fiber volley or fEPSP.

To determine whether presynaptic actions also contribute to anesthetic effects at glutamatergic synapses, paired-pulse facilitation (PPF) of Schaffer-collateral-commissural fiber evoked fEPSPs (60-ms intervals) were studied. In the presence of 0.41 mm sevoflurane, no apparent change in fEPSP rise time or decay kinetics was observed (fig. 4A), but PPF was transiently increased to approximately 130–150% of control during sevoflurane application and returned to baseline after washout (fig. 4B). This effect was observed in a dose-dependent manner (fig. 4C) and was independent of GABA^Areceptor–mediated actions because bicuculline (10 μm) did not affect sevoflurane-induced changes in PPF. This increase in facilitation is consistent with a presynaptic depression of glutamate release from nerve terminals, perhaps via  depressant actions on voltage-activated calcium or sodium channels that couple axon spike depolarization to transmitter release.

Tetanic stimulation (100 Hz, 1 s) induced a stable LTP (fEPSP slope after tetanization, 172.9 ± 15.1% of baseline, n = 9; fig. 5A), which was maintained for at least 1 h (fEPSP slope 30 min after tetanization, 154.8 ± 7.4% of baseline). First, we tested the effects of sevoflurane on the induction of LTP. Sevoflurane was administered 20 min before tetanic stimulation and delivered to the chamber throughout the experiment. In the presence of 0.04 mm sevoflurane, tetanic stimulation induced only posttetanic potentiation (141.8 ± 10.7% of baseline, n = 6; fig. 5A) and then failure of LTP. Posttetanic potentiation lasted only 5 min or less in all experiments (n = 6). To determine whether amnestic sevoflurane blocks the maintenance of LTP, 0.04 mm sevoflurane was applied after the LTP induction. As shown in figure 5B, administration of 0.04 mm sevoflurane 10 min after the LTP induction, however, had no influence on the maintenance of LTP (n = 7). The degree of LTP maintenance was similar to that of control LTP (fig. 5A). These findings indicate that amnestic sevoflurane inhibits the induction of LTP but has less effect on the maintenance. In the continued presence of higher concentrations of sevoflurane (0.41 mm), tetanic stimulation (100 Hz, 1 s) completely failed to induce LTP (fig. 6). Because 0.41 mm sevoflurane had inhibitory actions on fEPSP amplitude (fig. 3), baseline in the 0.41 mm sevoflurane group (n = 4) was decreased during anesthetic application (fig. 6).

To clarify concentration-dependent differential actions of sevoflurane on LTP, experiments were performed in the presence of bicuculline (10 μm). In this condition, tetanic stimulation could induce a stable LTP (fEPSP slope after tetanization 172.1 ± 10.5% of baseline; 30 min after tetanic stimulation, 134.2 ± 11.2%, n = 7; fig. 7A), indicating that bicuculline reinstates 0.04 mm sevoflurane–induced depression of LTP. The degree of LTP induction was similar to that of control LTP (fig. 5A), although the degree of maintenance was statistically less than control (P < 0.05). In the presence of 0.41 mm sevoflurane, bicuculline reinstated only partially sevoflurane-induced depression of LTP (fig. 7B). Only posttetanic potentiation was observed after tetanic stimulation (151.1 ± 21.7% of baseline, n = 9). The effects of bicuculline alone on LTP were also examined in the absence of sevoflurane. Bicuculline (10 μm) alone had no effect on the induction and maintenance of LTP (fEPSP slope after tetanization 185.0 ± 19.2% of baseline; 30 min after tetanic stimulation, 150.2 ± 7.5%, n = 6; fig. 8). The degree of LTP was statistically similar to that of control LTP (fig. 5A) in the absence of bicuculline. Although bicuculline (10 μm) promoted epileptiform activity in PS responses, such a waveform was not significant in fEPSP recordings. This difference is probably due to relatively smaller stimulus intensity to evoke fEPSP.

Major findings of the current study are summarized as follows. First, amnestic concentrations of sevoflurane inhibited PS amplitude and the induction of LTP with less effect on the maintenance through a GABAergic mechanism. Second, clinically relevant concentrations of sevoflurane dose-dependently depressed fEPSP amplitude and completely blocked LTP, in part by decreasing glutamate release as well as by enhancing GABAergic inhibitory tone. These data suggest that amnestic concentrations of sevoflurane inhibit synaptic plasticity of hippocampal CA1 neurons through GABAergic mechanisms and may thereby impair memory function.

Volatile anesthetics are administered to mammals in the gas phase at body temperature (approximately 37°C), whereas most in vitro  physiologic experiments are performed at room temperature. However, gas-phase potencies are basically temperature dependent, increasing markedly with decreasing temperatures.21Therefore, the common procedure of using gas-phase EC⁵⁰concentrations for room temperature experiments can result in overdosing the in vitro  preparations. In this context, Franks and Lieb22have recommended mammalian MAC values for volatile anesthetics expressed as free aqueous concentration in saline, e.g. , sevoflurane was 0.33 mm. As shown in figure 1, we used 0.04 mm for amnestic concentrations and 0.23–0.41 mm for clinical doses. We believe that these values are within reasonable range as clinical relevant concentrations.

Our LTP experiments using 100 Hz stimulus were performed at room temperature. These techniques are widely used, but room temperature is at least 10°C below physiologic conditions for mammalian neurons. It has been reported that temperature has large effects on synaptic transmission, plasma membrane state, and enzyme kinetics.23,24For example, a form of presynaptic plasticity is more prominent at room temperature.23It should be noted that temperature is a crucial experimental factor in studies of mammalian synaptic function.

We found that bicuculline reinstated 0.04 mm sevoflurane–induced depression of LTP and that bicuculline alone had no effect on control LTP in the absence of sevoflurane. Bicuculline is one of widely used GABA^Areceptor antagonists. Several recent studies have reported that bicuculline also blocks small-conductance Ca²⁺activated K⁺(SK) channels.25,26SK channels regulate membrane excitability in CA1 neurons and modulate hippocampal synaptic plasticity and learning. Stackman et al.  27reported that blockade of SK channels by apamin, a selective SK channel antagonist, decreased the threshold for the induction of hippocampal synaptic plasticity. After 50-Hz stimulus, significantly more LTP was induced in SK channel blocked slice, although with 100-Hz stimulus, equal extents of LTP were observed in control and SK channel blocked slice. Moreover, Hammond et al.  28showed that SK2 channels selectively regulated the hippocampal synaptic plasticity, and SK2 overexpression impaired the induction of hippocampal LTP in a frequency-dependent manner. Induction of LTP with 50-Hz stimulus was significantly reduced in SK2 overexpression, although with 100-Hz stimulus, induction of LTP was equivalent. In the current study, we used the stimulus of 100 Hz for the induction of LTP. Taken together, we think that the influence of the blockade of SK channels by bicuculline on LTP is small in our recording conditions, and that GABA^Areceptors are crucially involved in amnestic concentrations of sevoflurane induced inhibition of LTP.

We found that sevoflurane reversibly inhibited PS amplitude in a dose-dependent manner and that the inhibition produced by 0.04 mm sevoflurane was antagonized by bicuculline, but profound inhibition by 0.23 mm sevoflurane was only partially antagonized by bicuculline. However, we found that bicuculline alone slightly but significantly increased PS amplitude. Taking these factors into consideration, enhanced GABAergic inhibition plays a role for amnestic doses of sevoflurane-induced depression. In this regard, Pittson et al.  29reported that halothane and isoflurane at 1.0 MAC produced nearly complete depression of PS responses, but reversal of anesthetic-induced depression after bicuculline administration was only 20% in each anesthetic. Based on these data, amnestic sevoflurane seemed to have similar properties in terms of anesthetic-induced depression of evoked discharge of CA1 neurons.

Sevoflurane may have multiple sites of action to produce overall depression of the central nervous system.30,31Although there are many possible molecular targets, the sensitivity of each component to volatile anesthetic may be different. In fact, we demonstrated that the effects on GABA-mediated components contributed significantly to synaptically evoked discharge of CA1 neurons (fig. 2). We have previously studied the actions of sevoflurane on GABA^Areceptors expressed in human embryonic kidney 293 cells, particularly focusing on the effects of transmembrane 2 mutations in α and β subunits.11We found that low concentrations of sevoflurane (0.1–0.2 mm) significantly potentiated submaximal GABA-activated Cl⁻currents in wild-type GABA^Areceptors, and that positions serine 270 of both the GABA^A1 and 2 subunits were identified as critical for regulation of the GABA^Areceptor by sevoflurane. Recently, Sebel et al.  12also reported that subanesthetic sevoflurane and propofol modulated GABA receptor function in an additive manner and that these two drugs had separate binding sites and converging pathways of action on the GABA^Areceptor. Together, available data suggest that modulation of CA1 neurons by amnestic sevoflurane may be primarily due to activation of GABAergic inhibitory neurons and that clinical concentrations of sevoflurane depress evoked discharge of CA1 neurons by acting not only on GABAergic neurons, but also on other components such as excitatory synapses.

Consistent with previous reports of other volatile anesthetics,15,32sevoflurane reversibly depressed fEPSP amplitude in a concentration-dependent manner. Differing from PS responses, 0.04 mm sevoflurane had no effect on fEPSPs, but higher concentrations of sevoflurane considerably inhibited the amplitude of fEPSPs. Bicuculline did not reverse sevoflurane (0.41 mm)-induced fEPSP inhibition. These findings support the idea that clinical concentrations of sevoflurane directly depress CA1 excitatory synapses.

Paired-pulse facilitation was used as a measure of presynaptic anesthetic actions at Schaffer-collateral-commissural fiber synapses on CA1 pyramidal neurons.13,14,33PPF has been shown to increase after manipulations that reduce calcium-mediated glutamate release from the Schaffer-collateral-commissural pathway.33In contrast, manipulations that depress CA1 neuron fEPSPs via  postsynaptic actions do not change PPF.33,34We found that sevoflurane at 0.23–0.41 mm but not 0.04 mm significantly increased PPF, suggesting that anesthetic sevoflurane has presynaptic actions to reduce glutamate release from Schaffer-collateral terminals. These data confirm previous findings that clinical concentrations of halothane and isoflurane increased PPF,15suggesting that presynaptic sites are involved in fEPSP depression by volatile anesthetics.

In the presence of 0.04 mm sevoflurane, tetanic stimulation induced only posttetanic potentiation, which lasted for 5 min or less and was believed to involve only modifications at the presynaptic terminal.35Subsequent LTP is thought to involve an increase in the number of quanta released, or in the size of the response each quantum produces in the postsynaptic cells.36Therefore, amnestic sevoflurane appeared to have depressive synaptic actions on either presynaptic or postsynaptic mechanisms to induce LTP through GABAergic mechanisms.

There are several possible cellular targets in amnesic effects produced by general anesthetics; the GABAergic system may be one of the most important mechanisms.16,17Simon et al.  16investigated the effects of clinically relevant concentrations of isoflurane (0.2–0.3 mm) on the induction of LTP and long-term depression in slices from the juvenile (14 days) and adult (2 months) mouse hippocampus. They found that isoflurane (0.2–0.3 mm) blocked the induction of LTP and that picrotoxin (50 μm) partially reversed isoflurane-induced LTP inhibition. Blocking effects of isoflurane on LTP and long-term depression were reversible, suggesting that this agent does not induce persistent change in neural excitability. Similarly, Nagashima et al.  17reported that relatively higher concentrations of propofol (30 μm) inhibited LTP and that this inhibition was partially antagonized by picrotoxin (1 μm). These studies suggest that not only GABA mechanism but other factors may contribute to clinically relevant concentrations of anesthetic-induced LTP inhibition. Therefore, in the current study, we focused on amnestic concentrations of sevoflurane (0.04 mm), which produces amnesia but does not affect sedative/hypnotic actions. Our results offer new evidence that amnestic sevoflurane inhibits the induction of LTP with less effect on the maintenance of LTP, and that these actions alone seem to account fully for the effects of amnestic sevoflurane.

γ-Aminobutyric acid receptors have been previously reported to mediate sedative components of anesthesia37and amnesic properties of benzodiazepines.38Recently, Cheng et al.  39studied the effects of etomidate on GABA α⁵subunit knockout mice. In the wild type, low-dose etomidate (0.1 m), which had no effect on the kinetics of GABA-mediated synaptic currents, significantly increased GABA tonic currents, impaired LTP, and impaired memory performance. However, α⁵null mutant mice showed similar hypnotic dose-responses to the wild type but had little etomidate-induced amnesia, LTP, and tonic current potentiation. These data support the idea that GABAergic mechanisms contribute significantly to anesthetic-induced amnesia.

In conclusion, the current study provides evidence for a GABA^Areceptor involvement contributing to effects of amnestic concentrations of sevoflurane on synaptic plasticity of hippocampal CA1 neurons.