Extracellular Magnesium Ion Modifies the Actions of Volatile Anesthetics in Area CA1 of Rat Hippocampus In Vitro 

After approval had been obtained from the Animal Research Committee of Toyama Medical and Pharmaceutical University, male Wistar rats (weight, 100–200 g) were anesthetized with sevoflurane and decapitated. The technique for preparing rat hippocampal slices was identical to the methods previously described by Hirota and Roth. 12The brain was rapidly removed, and the dissected hippocampus was sliced in cold artificial cerebrospinal fluid (ACSF) transversely to its longitudinal axis (400 μm thick) with a Rotorslicer DTY–7700 (DSK, Osaka, Japan). Slices were placed on a nylon mesh at a liquid–gas interface in a recording chamber at 37°C. A humidified gas mixture (95% O²–5% CO²) was applied to the chamber at a flow rate of 1 l/min. ACSF was continuously perfused at a rate of 90 ml/h. Slices were incubated for 90 min without electrical stimulation.

Bipolar nichrome stimulating electrodes were placed in the region of stratum radiatum to activate Schaffer-collateral inputs to CA1 pyramidal neurons. Glass microelectrodes (3–5 MΩ filled with 2 m NaCl) were placed in the cell body region of CA1 neurons and stratum radiatum to record the evoked field PSs and EPSPs, respectively. Stimuli were applied via  the alveus hippocampi to produce antidromic responses of pyramidal neurons in the absence of synaptic transmission. The minimum stimulus intensity (5–10 V) that elicited a maximum PS amplitude was used to evoke a response. Square-wave stimuli (5–10 V, 0.05 ms, 0.1 Hz) were delivered using a SEN-3301 stimulator (Nihon Kohden, Tokyo, Japan). Field potentials were amplified (Nihon Kohden MEZ-8301), filtered (1 Hz to 10 kHz), and digitally converted (100 kHz) using an iNet system (GWI, Somerville, MA) and stored on a Macintosh computer (Apple, Cupertino, CA) for later analysis.

The PS amplitudes were measured from peak positive to peak negative. EPSP slopes were calculated by fitting digitized data points between onset and peak negativity to a linear function (dV/dt). 13Results were collected in a group of five, averaged, and stored as a single record. All preparations used in this study showed control variability less than 5% during the initial data acquisition period and after washout of anesthetics.

The general anesthetics, sevoflurane and isoflurane, were applied as vapors to the tissue chamber via  the prewarmed and humidified 95% O²–5% CO²gas stream above the slices, using the appropriate vaporizers, Tec 3 (Omeda, Steeton, West Yorkshire, United Kingdom) and Forawick (Muraco, Tokyo, Japan), respectively. Concentrations, expressed as volume percent (vol%), refer to dial settings on the vaporizer. The vaporizers had been previously calibrated with an anesthetic gas analyzer (Capnomac; Datex, Helsinki, Finland). Volatile anesthetics were applied for a minimum of 5 min to reach equilibrium, as the preliminary experiments demonstrated that the onset time for sevoflurane and isoflurane effects were less than 30 s and that maximum effects reached a plateau within 3 min (not shown). Concentrations of sevoflurane and isoflurane in the perfusate of the recording chamber were determined using gas chromatography (Shimazu, Kyoto, Japan). The concentrations of anesthetics in solution were found to be linear (0.64 and 0.55 mm per 1 vol%, respectively) up to 5.0 vol%. Recovery responses were recorded at least 10 min after washout of anesthetic-equilibrated ACSF from the chamber. In this study, the concentrations of sevoflurane and isoflurane that depressed PS amplitudes and EPSP slopes to 50% of control in normal (ACSF) condition were used, except for the antidromic experiments. The NMDA receptor antagonists (DL-2-amino-5-phosphonovaleric acid [AP-5] and (+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine [MK-801]) and the non-NMDA receptor antagonist (6-Cyano-7-nitroquinoxaline-2,3-dione [CNQX]) were dissolved into ACSF and applied as perfusate for 20 min. Stock solutions of CNQX (1 mm) were prepared in dimethyl sulfoxide and diluted in ACSF before it was perfused into the chamber. The concentration of dimethyl sulfoxide used in the experiments alone did not affect the field potentials. In some experiments, concentrations of Mg²⁺and Ca²⁺in the perfusate were altered by changing the concentrations of MgSO⁴and CaCl²in ACSF; the modified perfusate was also applied for 20 min.

The composition of the ACSF was 124 mm NaCl, 5 mm KCl, 2 mm CaCl², 1.25 mm NaH²PO⁴, 2 mm MgSO⁴, 26 mm NaHCO³, and 10 mm glucose, prepared in purity water. The Mg²⁺and Ca²⁺concentrations in cerebrospinal fluid previously reported in rats are 0.53–0.63 mm 14and 0.75–0.98 mm, 15respectively. It is necessary to use higher concentrations of [Mg²⁺]^o(2.0 mm) in normal ACSF for slice preparations because many of the afferents and efferents are disrupted during the slicing procedure. This results in preparations that produce spontaneous spiking. The ACSF was precooled (8–10°C) and saturated with a 95% O²–5% CO²gas mixture before use (pH 7.3–7.5). Sevoflurane and isoflurane were donated by Dinabot (Osaka, Japan) and Maruishi Pharmaceutical Co. (Osaka, Japan), respectively. Most of the chemicals were obtained from Sigma (St. Louis, MO), except for MK-801 and CNQX, which were obtained from Tocris Cookson (Bristol, United Kingdom).

Experiments were conducted on 61 hippocampal slices prepared from 48 rats. Anesthetic effects are expressed as percent of control (normalized to the value in the absence of anesthetic), and the mean ± SD was calculated based on measures from at least four slices. Statistical differences among groups of different [Mg²⁺]^oor extracellular Ca²⁺concentration ([Ca²⁺]^o) were determined by one-way analysis of variance with Bonferroni-corrected post hoc  analyses. Differences between the effects of anesthetics or antagonists on the field potential parameters in various [Mg²⁺]^owere determined by two-way factorial analysis of variance (Statview 4.0; Abacus Concepts, Berkeley, CA). A P  value < 0.05 was considered significant. Linear and quadratic functions were fitted to the data using the SigmaPlot program (Jandel Scientific, San Rafael, CA).

In normal [Mg²⁺]^o(2.0 mm), the magnitudes of EPSP slopes and PS amplitudes of CA1 pyramidal neurons in response to stimulation of the Schaffer collateral fibers are 24 ± 11 v/s and 7.4 ± 1.5 mV (n = 5), respectively. Changing the perfusate to low (0.5 mm) enhanced the EPSP slopes and the PS amplitudes to 114 ± 10% and 119 ± 11% of those in each normal [Mg²⁺]^o, respectively (n = 4). Higher [Mg²⁺]^o(4.0 mm) reduced the signals to 66 ± 3% and 62 ± 11%, respectively (n = 5). Table 1summarizes the alterations of PS amplitudes in various [Mg²⁺]^o/[Ca²⁺]^oconditions.

Figure 1shows representative recordings of the effects of sevoflurane (3.0 and 5.0 vol%) on the evoked PSs (fig. 1A) and EPSPs (fig. 1B), respectively, in various [Mg²⁺]^oconcentrations. In normal [Mg²⁺]^o, sevoflurane reduced the PS amplitudes and the EPSP slopes in a reversible manner. The inhibitory actions of sevoflurane on PSs and EPSPs were attenuated in the [Mg²⁺]^o-free condition but were enhanced in high [Mg²⁺]^o. Figure 2summarizes the relations between [Mg²⁺]^oand the effects of sevoflurane and isoflurane on the PS amplitude (fig. 2A) and the EPSP slopes (fig. 2B). The inhibitory actions on PS amplitudes in the presence of sevoflurane and isoflurane were largely dependent on [Mg²⁺]^o(r²= 0.92 and 0.90;P < 0.0001 and P < 0.0001, respectively), whereas the EPSP slopes in the presence of sevoflurane and isoflurane were loosely correlated with [Mg²⁺]^o(r²= 0.79 and 0.39;P < 0.0001 and 0.05, respectively). The data points for PSs and EPSPs of sevoflurane were not significantly different from those of isoflurane.

A series of experiments was then conducted to study the mechanism(s) of the [Mg²⁺]^o-dependent depression of PS amplitudes by volatile anesthetics. First, the role of presynaptic Ca²⁺channels in the [Mg²⁺]^o-dependent events was examined. Because Mg²⁺competes with Ca²⁺at calcium channels, the ratio of [Mg²⁺]^oto [Ca²⁺]^oin the perfusate was maintained at a constant, i.e. , [Ca²⁺]^owas adjusted in proportion to [Mg²⁺]^o. Figure 3demonstrates the poor correlation between [Mg²⁺]^oand anesthetic-induced depression, indicating that the [Mg²⁺]^o-dependent effects could be suppressed during conditions in which a ratio of [Mg²⁺]^oto [Ca²⁺]^owas fixed.

To examine the involvement of the NMDA receptor in the [Mg²⁺]^o-dependent anesthetic depression, the effects of NMDA receptor antagonists, AP-5 (10⁻⁴m) and MK-801 (4 × 10⁻⁵m), were studied on the [Mg²⁺]^o-dependent depression produced by isoflurane (0.7 vol%). The concentrations of the antagonists have been reported to completely block the NMDA receptor in hippocampal preparations in vitro  16,17and have been used as pharmacologic tools in electrophysiologic experiments. 18AP-5 or MK-801 alone had no consistent effect on the shape of EPSP and PS. The anesthetic-induced depression of PS amplitudes was well correlated with [Mg²⁺]^oin the presence of AP-5 (r²= 0.83;P < 0.0001;fig. 4) and MK-801 (r²= 0.99;P < 0.0003), suggesting that blockade of NMDA receptors had no effect on the [Mg²⁺]^o-dependent effects. In addition, the participation of the non-NMDA receptor antagonist was examined. The complete blockade of the non-NMDA receptor diminishes field potentials, since the non-NMDA receptor is primarily responsible for the orthodromic activation of pyramidal neurons. A relatively low concentration of CNQX (5 × 10⁻⁷m) was used in the current study: CNQX alone reduced the PS amplitudes from 9.1 ± 4.4 mV to 7.1 ± 2.8 mV (n = 5). CNQX failed to modify the [Mg²⁺]^o-dependent inhibition (r²= 0.99;P < 0.0003) in the presence of isoflurane (0.7 vol%).

In subsequent experiments, two independent recording electrodes were placed in the dendrite and soma regions of CA1 pyramidal neurons to simultaneously record EPSPs and PSs, respectively. The relations between the EPSP slopes and the PS amplitudes describe an input–output function for pyramidal neurons. The input–output relations at different stimulus intensities were then studied to address whether the postsynaptic events were involved in the [Mg²⁺]^o-dependent volatile anesthetic depression. Figure 5shows representatives of the effects of volatile anesthetics on input–output curves in low, normal, and high [Mg²⁺]^o. In normal [Mg²⁺]^o, PS amplitudes steeply increased as a function of EPSP slopes. Neither sevoflurane nor isoflurane produced consistent effects on the configuration of the input–output curves and on the EPSP-axis intersections (threshold for PS generation) in the range of 0.5 to 4.0 mm of [Mg²⁺]^o. The threshold did not exhibit the [Mg²⁺]^o-dependent modification in the absence and presence of volatile anesthetics (fig. 6).

The [Mg²⁺]^o-dependent anesthetic inhibition was also tested on antidromically elicited PSs in response to the activation of nonsynaptic pathways. The baseline amplitudes of antidromic PSs at 0.5, 2.0, and 4.0 mm of [Mg²⁺]^owere 2.7 ± 1.2, 2.3 ± 1.3, and 2.6 ± 1.5 mV (n = 4–7), respectively. Figure 7shows that the effects of sevoflurane and isoflurane on antidromic PSs were not dependent on [Mg²⁺]^o, suggesting that the soma or axons of pyramidal neurons are not responsible for the [Mg²⁺]^o-dependent anesthetic inhibition.

We have demonstrated that [Mg²⁺]^ocan modify the inhibitory actions of volatile anesthetics on field potentials elicited by stimulation of monosynaptic pathways in mammalian CNS preparations. Because the input–output analysis showed that neuronal excitation in PS steeply increases as a function of the EPSP amplitude (fig. 5), the [Mg²⁺]^o-dependent effects were more evident on PS amplitudes than on EPSP slopes in the current study.

It is known that increasing total divalent cation decreases cell excitability, 5and the alteration of [Mg²⁺]^oitself modified the parameters of field potentials in the current study. Previous work reported that an intravenous infusion of Mg²⁺produced neither anesthesia nor analgesia in two human volunteers, even when their serum Mg²⁺concentration reached 7.3–7.7 mm (11 times the normal value). 19Aldrete et al.  20observed a sleep-like state in mongrel dogs after intravenous administration of Mg²⁺. However, they concluded that the cardiovascular depression and respiratory muscle paralysis by high [Mg²⁺]^oimpaired the level of consciousness. Taken together with the current results, Mg²⁺per se  could not produce anesthesia or analgesia, but could probably modify the actions of volatile anesthetics in CNS. Our study was designed to test whether the [Mg²⁺]^o-dependent modification of anesthetic actions could arise from (1) presynaptic Ca²⁺channels, (2) NMDA receptors, (3) postsynaptic mechanisms, or (4) neural excitability of pyramidal cell soma.

Presynaptic Ca²⁺channels regulate neurotransmitter (glutamate) release in the CNS. Isoflurane has been reported to block Ca²⁺channels in hippocampal pyramidal neurons, 21and these could be a major target site for volatile anesthetics in the CNS. The current results showed that the [Mg²⁺]^odependence was attenuated under the fixed [Mg²⁺]^o/[Ca²⁺]^oratio, suggesting that the competition between Mg²⁺and Ca²⁺at the presynaptic Ca²⁺channels modify the volatile anesthetic depression of the Ca²⁺currents. Presynaptic Ca²⁺channels could be, at least in part, responsible for the [Mg²⁺]^o-dependent anesthetic inhibition.

Mg²⁺is known to regulate activation of NMDA receptors in a voltage-dependent manner. 6,7Tramer et al.  22showed in clinical trials that intravenous administration of magnesium sulfate reduced the postoperative pain score, and suggested that high concentrations of [Mg²⁺]^ocan modify analgesic effects caused by NMDA receptor–mediated mechanisms. However, the results of the current study demonstrate that neither NMDA nor non-NMDA receptors are involved in the [Mg²⁺]^o-dependent modification of anesthetic effects.

Input–output (EPSP–PS) relations were examined to investigate whether the postsynaptic events, such as inhibitory postsynaptic potentials, were involved in the [Mg²⁺]^o-dependent modification of anesthetic actions. Because the inhibitory postsynaptic potential has a latency to onset of 3 ms or less in CA1 of rat hippocampus, 23the feedforward inhibitory postsynaptic potential could modulate PS generation. If the inhibitory postsynaptic potential is modified in a [Mg²⁺]^o-dependent manner, the alteration of application of [Mg²⁺]^owould affect the input–output curve and shift the threshold for PS generation in the presence of volatile anesthetics. However, the current results did not support this concept. Sevoflurane and isoflurane had minimal effects on the curves and thresholds, and the effects were not modified by changing [Mg²⁺]^o. In addition, the actions of volatile anesthetics on antidromically elicited responses were not [Mg²⁺]^o-dependent. Taken together, it is unlikely that the postsynaptic events or the excitability of cell soma are responsible for the [Mg²⁺]^o-dependent anesthetic actions.

It has been reported that serum Mg²⁺concentration can be altered during general anesthesia. 8,9Considering that the cerebrospinal fluid concentration of Mg²⁺is regulated via  active transport mechanisms, a small amount of Mg²⁺could cross the blood–brain barrier. 24The alteration of cerebrospinal fluid Mg²⁺concentration is found to be highly correlated with serum Mg²⁺concentration, 25and systemically administered Mg²⁺reduces minimum alveolar concentration of volatile anesthetic in rats in vivo . 26Hence, it might be possible that [Mg²⁺]^o-dependent modification of volatile anesthetic actions can take place during clinical anesthesia.

In conclusion, the current study provides evidence that alteration of [Mg²⁺]^ocan modify the actions of volatile anesthetics on stratum radiatum synaptic input to CA1 pathway in rat hippocampus in vitro . The results suggest that [Mg²⁺]^o-dependent anesthetic inhibition could be, in part, a result of the modification of presynaptic Ca²⁺channels.

The authors thank Toshio Narahashi, Ph.D. (Professor, Department of Molecular Pharmacological and Biological Chemistry, Northwestern University Medical School, Chicago, IL), Dinabot Co. Ltd., Osaka, Japan, and Maruishi Pharmaceutical Co., Osaka, Japan.