Volatile Anesthetics Depress Glutamate Transmission Via Presynaptic Actions

Experiments were conducted on brain slices isolated from adult male Sprague-Dawley rats (weighing 90–110 g). Animals were housed with a 12-h light cycle with food and water provided ad libitum, and were anesthetized with halothane or diethyl ether before use. While deeply anesthetized, rats were killed by a blow to the back of the thorax to stop the heart and reduce bleeding as the brain was removed. At all stages care was taken to ensure gentle handling of animals; the research protocol was approved by the Stanford University Animal Research Committee, sanctioned by the Council of Laboratory Animal Medicine (USA), and adhered to National Institutes of Health guidelines. For experiments conducted at 35 degrees Celsius an interface brain slice chamber was used, as previously described. [3,30]Room temperature (22 +/- 1.7 degrees Celsius, mean +/- SD, n = 30) experiments were conducted using a submersion chamber. [35].

Coronal brain slices (450 micro meter) were prepared on a vibratome (Vibroslice, Boston, MA), hemisected and placed on filter papers at the interface of a humidified oxygen/carbon dioxide (95%/5%) carbogen gas phase and artificial cerebrospinal fluid (ACSF; see Materials) liquid phase. Slices were allowed at least 1 h for recovery before being submerged in ACSF in a recording chamber. The ACSF was saturated with carbogen gas and perfused at a rate of 2.5 ml/min, at room temperature. Rapid and accurate solution changes were made using an AutoMate (Oakland, CA) computerized perfusion system, and high-quality polytetrafluorethylene was used for reservoirs, valves, and tubing to minimize volatile anesthetic loss and drug binding. Volatile agent concentrations were adjusted using calibrated commercial vaporizers (Fraser Harlake, Orchard Park, NY) and a Puritan-Bennett UV anesthetic agent monitor (Model 222, Wilmington, MA) to measure concentrations. Anesthetic concentrations in the ACSF were measured using calcium-sensitive electrodes* to ensure that steady-state concentrations were achieved.

Bipolar tungsten microelectrodes (0.5 M Omega; Frederick Haer, Brunswick, ME) were placed in stratum radiatum to electrically activate Schaffer-collateral fibers (Figure 1). Electrical stimuli consisted of square wave paired pulses (0.25 ms duration; 5–8 V) delivered every 10 s, to minimize frequency-dependent changes in transmission, from digitally timed and controlled isolation units (Grass Instruments, Quincy, MA). An interpulse interval of 120 ms was used to study paired pulse facilitation and was varied between 10 and 1000 ms to determine the time course of facilitation. Glass microelectrodes filled with ACSF (0.5–1.0 M Omega, Garner Glass Company, Claremont, CA) were placed in stratum radiatum to record EPSPs, or at the stratum pyramidal/oriens border to record evoked population spikes (Figure 1). Recorded signals were amplified (10,000x), filtered (1.0 Hz to 30 kHz, bandpass) and digitized at 50 kHz for computer storage and analysis (Data Wave Technologies, Longmont, CO). Response measures were plotted on-line during experiments to ensure stable baseline activity (< 5% variation for 20 min) was achieved from each preparation before the administration of anesthetics.

Excitatory postsynaptic potential amplitudes were measured as peak negativity from baseline. Population spike amplitudes were measured from peak negativity to peak positivity. All measurements were made using Data Wave Technologies software. Anesthetic effects were expressed as a percent of control (effect measure/control measure x 100) and the mean +/- SD was calculated based on measures from at least eight preparations. Each preparation was exposed to only a single concentration of a given anesthetic, to avoid possible residual drug effects. Statistical significance was determined using analysis of variance (ANOVA; AXUM, TriMetrix, Seattle, WA) to compare time-matched control responses with those produced by the anesthetics.

The ACSF had the following ionic composition (in mM): Sodium sup +-- 151, Potassium sup +-- 3.5, Calcium sup ++-- 2.0, Magnesium sup ++-- 2.0, Chlorine sup --- 131, HCO³sup -, -- 26, SO⁴sup ---2.0, PO⁴sup --- 2.0, PO⁴sup --- 1.25, glucose -- 10. All chemicals were reagent grade or better, obtained from J.T. Baker Chemical Company (Phillipsburg, NJ). Water for solutions was spectrophotometric and high-perfusion liquid chromatographic grade, obtained from EM Sciences (Gibbstown, NJ). 6-Cyano-7-nitroquinoxaline-2,3-dione and 2-hydroxypropyl-beta-cyclodextrin (CNQX:HBC complex) and APV ((+/-)-2-amino-5-phosphonopentanoic acid) and bicuculline methiodide were obtained from RBI (Natick, MA). Halothane was obtained from Halocarbon Laboratories (North Augusta, SC). Isoflurane was obtained from Abbott Laboratories (North Chicago, IL). All solutions, including concentrated drug stocks, were made fresh before each experiment. Rats were obtained from Simonsen Laboratories (Gilroy, CA) and housed under the continuous care of technologists and veterinarians in the Stanford University Department of Laboratory Animal Medicine facilities.

Synaptically evoked population spike (PS) responses recorded from CA1 neurons were similar in appearance at the two temperatures, although response amplitude was smaller and the time course was prolonged at 22 degrees Celsius. After Schaffer-collateral stimulation, a single negative compound action potential (PS) was superimposed on a slower positive potential (EPSP) seen at both temperatures (Figure 1). Population spike half decay constants were 1.5 times slower at 22 degrees Celsius (3.93 +/- 0.19 ms; mean +/- SD; n = 8) compared with 35 degrees Celsius (2.57 +/- 0.35 ms; n = 5). Excitatory postsynaptic potential decay times were slowed by 1.4, from 9.7 +/- 0.9 ms at 35 degrees Celsius to 13.2 +/- 0.8 ms at 22 degrees Celsius. Decay time constants for both PS and EPSP responses were significantly (P < 0.005, ANOVA) slowed at 22 degrees Celsius.

Population spike responses were nearly abolished in the presence of 1 minimum alveolar concentration (MAC) of halothane (1.2 vol %, Figure 1) or isoflurane (1.4 vol %), regardless of temperature. The profiles of anesthetic effects were also similar at both temperatures, with an increase in PS latency accompanying depression of responses, although discharge thresholds were not altered. Halothane produced a 98.2 +/- 3.1% depression of PS amplitude at 22 degrees Celsius and a 97.4 +/- 4.2% depression at 35 degrees Celsius. Isoflurane produced a 98.7 +/- 3.3% depression at 22 degrees Celsius and a 98.5 +/- 4.2% depression at 35 degrees Celsius. There were no significant differences observed between halothane and isoflurane effects (P > 0.1, ANOVA, n = 11) or anesthetic effects at either temperature (P > 0.1). Anesthetic effects were comparable at both temperatures despite the observed slowing of PS voltages observed at 22 degrees Celsius and the expected increase in agent/ACSF solubility. Concentration-dependent depression of PS amplitudes was also observed at 0.25 and 0.5 MAC concentrations for both anesthetics at 22 degrees Celsius and 35 degrees Celsius. To further characterize temperature effects on CA1 circuitry, latency profiles for EPSP and PS facilitation were studied.

The time course of EPSP facilitation reflects presynaptic, calcium-dependent, vesicle release processes in Schaffer-collateral nerve terminals (see Discussion). When two identical stimuli were used to generate action potentials in these presynaptic fibers, the second EPSP amplitude was larger than the first for interstimulus intervals between 10 and approximately 500 ms. Peak facilitation occurred at 20–30-ms intervals (approximately 40 Hz), and facilitation slowly declined during the next 300–600 ms. Facilitation remained intact at 22 degrees Celsius, and latency profiles were similar to those at 35 degrees Celsius (Figure 1), especially for responses at longer interstimulus intervals (> 100 ms). Lowering the temperature did not appear to slow calcium-dependent release mechanisms in these nerve terminals, and comparable levels of facilitation (120–150%) were observed at both 22 degrees Celsius and 35 degrees Celsius.

Population spike facilitation provided a measure of postsynaptic feedforward and feedback inhibition. [22,33]An early period of inhibition (0–20-ms intervals, Figure 1) results from the activation of voltage-dependent potassium channels (I^KA) and GABA sub Afast -gated chloride channels, which combine to limit CA1 neuron discharge. Maximal spike facilitation occurred with interstimulus intervals of 20–30 ms, similar to EPSP facilitation. With interstimulus intervals > 40 ms, PS facilitation was markedly reduced, reflecting the activation of late inhibitory currents, including: I^KAHP, I^KCA, GABA^Aslow, and GABA^B. These late currents depressed spike discharge even though excitatory synaptic drive remained elevated (compare the time course of EPSP and PS facilitation, Figure 1). Lowering the temperature to 22 degrees Celsius did not significantly (P > 0.05 compared to 35 degrees Celsius) change the interstimulus latency profile for PS facilitation. Early inhibitory responses appeared to decay sooner at 35 degrees Celsius; late inhibition activated somewhat earlier and also appeared to decay sooner. These results, together with a similar EPSP facilitation profile, indicate that both excitatory and inhibitory synaptic circuits remained intact at 22 degrees Celsius. This temperature (22 degrees Celsius) was used for all subsequent experiments.

Clinically effective concentrations (0.5–2.0 MAC) of both isoflurane and halothane produced a marked and consistent depression of Schaffer-collateral evoked EPSPs. The time course of anesthetic effects was slow at 22 degrees Celsius; 10–20 min was required to reach steady-state levels of depression and recovery required a similar amount of time (Figure 2). Onset and recovery kinetics followed a first-order exponential for both anesthetics, but isoflurane had a faster equilibration time (16.2 +/- 0.6 min vs. 18.7 +/- 1.4 min for halothane, mean +/- SD, P < 0.01, n = 8 for each agent). Decreasing extracellular calcium concentrations from 2.0 to 1.5 mM produced a 70.4 +/- 13.6% depression of EPSP amplitude with a faster equilibration time of 11 +/- 0.7 min (n = 6).

Isoflurane produced a 75.8 +/- 5.2% depression of EPSP amplitudes at a concentration of 1.7 vol %(1.2 rat MAC; mean +/- SD, n = 11). Halothane was less potent and produced a 46 +/- 9.4% depression at a concentration of 1.2 vol %(1.0 rat MAC; n = 8). Isoflurane was significantly more potent than halothane at low concentrations (0.5 MAC; P < 0.001, ANOVA, Figure 3), but the anesthetics were equipotent at higher concentrations (1.5–2.0 MAC). The concentration for 50% depression (EC⁵⁰) of EPSP responses was 0.71 MAC for isoflurane and 1.1 MAC for halothane, determined from a least-squares fit of the concentration-response data (Figure 3).

It was unlikely that enhanced recurrent GABA-mediated inhibition contributed to anesthetic-induced EPSP depression, because stimulus intensities were below threshold for producing CA1 discharge. Anesthetic-enhanced feedforward inhibition, however, could have contributed by shunting EPSP dendritic currents. To test whether feedforward inhibition was involved, the GABA^Areceptor antagonist bicuculline was studied. Bicuculline (10 micro Meter) did not significantly alter EPSP amplitudes when tested alone (105 +/- 3.7% of control; n = 5; P > 0.01, ANOVA) and produced only a small (<5%) reversal of responses to 1 MAC halothane (42.7 +/- 9.9% depression, n = 5; P > 0.5 compared with halothane alone). A concentration of 10 micro Meter bicuculline is known to depress electrically evoked monosynaptic and spontaneous GABA-mediated inhibitory postsynaptic currents by > 90% at both 22 degrees Celsius and 35 degrees Celsius. [36,37]The small effect of bicuculline to reverse anesthetic-induced EPSP depression was not accompanied by a change in facilitation, indicating that enhanced feedforward inhibition did not contribute to this presynaptic measure.

An interstimulus interval of 120 ms was used to study paired pulse facilitation of Schaffer-collateral EPSP responses, because this interval produced a comparable degree of facilitation at both 22 degrees Celsius and 35 degrees Celsius (Figure 1). In control conditions using a stimulus intensity of 75% of maximal response, EPSP facilitation was 1.22 +/- 0.07 (n = 15, approximately 20% increase in amplitude of the second EPSP compared to the first). When stimulus intensity was decreased to produce smaller EPSP responses, facilitation did not change even for responses that were < 10% of control amplitudes (Figure 4). A change in facilitation should not have occurred because decreasing the stimulus intensity reduced the number of Schaffer-collateral fibers activated, but would not change the characteristics of synaptic transmission. Each nerve ending should still be depolarized by the same amount because the action potential in activated fibers would not change. Similarly, reducing EPSP amplitudes by blocking postsynaptic glutamate receptors with a competitive antagonist (CNQX 10 micro Meter) did not change paired pulse facilitation (Figure 5(A)). CNQX produced a 73.4 +/- 1.9% depression of EPSP amplitudes (n = 6), with no apparent change in presynaptic action potential amplitude or conduction velocity, since Schaffer-collateral fiber volley responses superimposed before and after CNQX application (Figure 5(A)).

When EPSP amplitudes were reduced by decreasing extracellular calcium concentrations (control = 2.0 mM), facilitation was increased to 131.2 +/- 7.6%(1.5 mM, n = 8) or 152.1 +/- 18.3%(1.0 mM, n = 5). Decreasing extracellular calcium produced a depression of EPSP amplitudes with no apparent change in Schaffer-collateral fiber volley amplitudes or conduction velocity (Figure 5(B)). Halothane, at a concentration of 1.0 MAC, produced a significant increase in facilitation to 143.2 +/- 13.8%(n = 8, P < 0.005 compared to control, ANOVA). This halothane-induced increase in facilitation was accompanied by a small (< 20%) depression of Schaffer-collateral fiber volley amplitudes (Figure 5(C)). Isoflurane (1.2 MAC) also significantly increased facilitation to 160 +/- 11.9%(n = 11; P < 0.001 compared with control, ANOVA), but this occurred with no apparent depression in fiber volley amplitude (Figure 5(D)).

A comparison of anesthetic effects on EPSP amplitude and facilitation are shown in Figure 6, together with a comparison of low calcium, decreased stimulus intensity, and postsynaptic receptor block. The anesthetic-induced increase in facilitation was consistent with a presynaptic action to reduce glutamate release, comparable to that produced by lowering calcium levels. Increases in facilitation were not produced by reducing the number of active fibers (decreased stimulus intensity;Figure 4) or by blocking postsynaptic glutamate receptors with CNQX.

Isoflurane and halothane depressed glutamate-mediated excitatory synaptic transmission in a concentration-dependent manner. Significant depression of EPSP amplitudes occurred at 0.5 MAC for both anesthetics; half maximal depression occurred at 0.71 MAC for isoflurane and 1.1 MAC for halothane. These results confirm previous findings that volatile anesthetics [24,30,38]and especially halogenated ethers [31]depress glutamate-mediated transmission at clinically relevant concentrations. The increase in facilitation that accompanied EPSP depression is consistent with a presynaptic anesthetic action to reduce glutamate release from Schaffer-collateral terminals. These findings confirm earlier electrophysiologic results from brain slices at physiologic temperatures [30,31]and agree with recent findings from biochemical studies of anesthetic-induced depression of glutamate release. [29,39].

Decreasing the temperature had little effect on Schaffer-collateral evoked synaptic responses (Figure 1) or on second-order physiologic responses such as EPSP and PS facilitation. A 13 degrees Celsius temperature decrease (from 35 degrees Celsius to 22 degrees Celsius) slowed synaptic and spike discharge responses by 1.5- to 2.0-fold, but a comparable level of EPSP facilitation was observed (approximately 120% for interstimulus intervals of 100 to 200 ms). Similarly, the time course of PS facilitation did not change significantly, nor did the early and late stages of inhibition, which limit spike discharge. These results agree with previous findings from hippocampal brain slices [40,41]and from rats with chronically implanted hippocampal electrodes. [42,43]Hippocampal synaptic currents appear to slow with a Q¹⁰of approximately 1.5. Excitable membranes and CNS function adapt well to decreasing temperatures [44,45]and rats remain capable of learning motor tasks at brain temperatures of 28–30 degrees Celsius. [42]Long-term potentiation of hippocampal synaptic responses remains functional at temperatures above 20 degrees Celsius. [46]Although cooling can produce anesthesia, rodents remain conscious at brain temperatures from approximately 20–40 degrees Celsius. [47].

Volatile anesthetic-induced depression of population spike responses were not significantly different at 22 degrees Celsius compared with effects at 35 degrees Celsius. The increase in volatile agent solubility that occurs at 22 degrees Celsius was offset by evaporative loss in the submersion chamber, whereas experiments conducted at 35 degrees Celsius were performed in a closed chamber where the ACSF concentrations were in equilibrium with gas phase concentrations. Preliminary experiments using an ion sensitive electrode to measure aqueous halothane concentrations in both chambers (22 degrees Celsius and 35 degrees Celsius), indicated that similar concentrations of anesthetic were present (220–250 micro Meter at 1.0 MAC, P > 0.05). Thus, comparable concentrations of halothane appeared to produce a similar degree of synaptic depression at 22 degrees Celsius compared with that at physiologic temperatures. This would argue against changes in bulk membrane fluidity contributing to the anesthetic effect, because fluidity would be reduced over two orders of magnitude by a 13 degrees Celsius temperature change compared to increases in fluidity produced by the anesthetics. [45].

Increased facilitation produced by halothane and isoflurane was consistent with a presynaptic site of action (Figure 6). Perouansky and coworkers [4]also suggest a presynaptic action to account for the equieffective depression of alpha -amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) and N-methyl-D-aspartate (NMDA)-mediated EPSCs, and lack of halothane effect for exogenously applied glutamate responses, on Schaffer-collateral to CA1 neuron synapses in mouse hippocampal slices at 22 degrees Celsius. Similar conclusions were reached using rat hippocampal slices maintained at physiologic temperatures (35 degrees Celsius). [24,31]These electrophysiologic results support the direct measures from biochemical studies showing volatile anesthetic-induced depression of glutamate release from brain slices and synaptosomes. [25,29]This presynaptic action may not account for all of the anesthetic-induced depression of excitatory synaptic responses. There is growing evidence that alcohol, volatile agents, and barbiturates can interfere with postsynaptic glutamate receptor activation, [27,48]and depression of N-methyl-D-aspartate receptors appears to be the major effect produced by ketamine. [49]Similar depressant actions on glutamate receptors also have been reported for propofol. [10]Taken together, there is now good evidence for anesthetic effects at both presynaptic and postsynaptic sites of glutamate synapses. It remains to be determined which sites of action contribute most to depression of glutamate-mediated transmission. These effects at excitatory synapses, in combination with enhanced inhibition at GABA synapses, [1,2]could account for most of the CNS depression produced by some anesthetics. It remains unknown the extent to which these individual presynaptic and postsynaptic effects contribute to the overall CNS depression, but depression of glutamate transmission accounts for approximately 50% of hippocampal PS response depression. [50].

Although our understanding of the physiology underlying glutamate release from nerve terminals is incomplete, it is apparent that volatile anesthetics can act at several sites to depress transmitter release. Halothane has been shown to depress sodium channels [51]and reduce action potential conduction in Schaffer-collateral fibers. [30]In the current study, halothane depressed fiber volley amplitudes, indicative of a conduction failure in these axons, which could account for some of the EPSP depression observed. An anesthetic-induced depression of fiber conduction, however, would not contribute to the observed increase in EPSP facilitation (Figure 3, Figure 5, and Figure 6). Depression of calcium entry into nerve terminals would account for both the depression of EPSP amplitude and increase in facilitation. Volatile anesthetics have been shown to depress calcium channels in hippocampal neurons [19]and this would provide a likely mechanism contributing to EPSP depression. Coupling among terminal depolarization, calcium entry, and glutamate release involves several proteins associated with both vesicle and nerve terminal membranes (Figure 7). [52]These proteins come together via a calcium/actin-driven process to form a docking complex with a vesicle positioned next to a release site and presynaptic calcium channel. Only calcium channels with docked vesicles appear capable of opening to allow calcium entry during terminal depolarization. [53,54]Anesthetics could depress glutamate release and alter facilitation by disrupting protein-protein interactions at vesicle docking sites (Figure 7), thus uncoupling calcium entry and vesicle release from nerve terminal depolarization. Such a mechanism, involving a site of action proximal to calcium influx has been proposed to account for volatile anesthetic-induced depression of glutamate release from synaptosomes. [29]The presynaptic and postsynaptic depressant actions produced at glutamate synapses and opposite enhanced postsynaptic actions at GABA synapses provide further evidence that anesthetics act at multiple sites and via unique mechanisms. [30]Differences in effect produced by halothane and isoflurane (e.g., on facilitation and fiber volley conduction) suggest that different mechanisms can contribute to CNS depression produced by halogenated alkanes versus ethers, in accordance with a multisite agent-specific hypothesis of anesthetic action. [5,30].

The authors thank their Stanford colleagues Dr. Joan J. Kendig, Heath S. Lukatch, and Dr. Donald R. Stanski for helpful comments.

*C. Hagan, JR Trudell, RA Pearce & MB MacIver An aqueous phase volatile anesthetic monitor using Calcium sup ++ Sensitive electrodes, submitted to ANESTHESIOLOGY May 1996.