Although barbiturates activate alpha-aminobutyric acid type A receptors as part of their hypnotic effect, these drugs also inhibit voltage-gated calcium channels. The authors determined if barbiturates could decrease neuronal intracellular Ca2+ transients and the resulting glutamate release.
Neonatal rat cerebellar granule neurons were isolated and cultured on coverslips and studied at 37 degrees C. Spectrofluorometric assays were used during identical conditions to monitor intracellular Ca2+ with the Ca2+ -sensitive fluorophore fura-2 and glutamate release by a glutamate dehydrogenase-coupled assay, which produced the reduced form of nicotinamide-adenine dinucleotide phosphate in proportion to the amount of glutamate released. Neurons were depolarized by a rapid increase in external [K+] from 5 to 55 mM. Control responses were compared with those in the presence of 10, 30, and 100 microM thiopental; 3, 10, and 30 microM methohexital; decreased external [Ca2+]; or voltage-gated calcium channel blockers.
Thiopental and methohexital depressed the intracellular Ca2+ transient peak and plateau in a dose-dependent manner, as did decreased Ca2+. The intermediate dose of either drug caused approximately 50% decrease in peak intracellular Ca2+ and 60% decrease in glutamate release. In the presence of specific L- and/or N-type voltage-gated calcium channel blockade by nicardipine or omega-conotoxin-GVIA, respectively, 30 microM thiopental further decreased the intracellular Ca2+ transient. Thiopental caused a dose-dependent decrease in glutamate release, which was proportional to the decreased peak intracellular Ca2+.
Thiopental and methohexital depress the depolarization-induced increase in intracellular Ca2+ and the accompanying glutamate release, actions which can contribute to the anesthetic and neuronal protective effects of these drugs.
THE anesthetic potency of the barbiturate class of sedative hypnotic agents has long been recognized. A major site of action of these drugs regarding their sedative hypnotic effects is related to their binding to the [Greek small letter gamma]-aminobutyric acid type A (GABAA) receptor-chloride channel complex. [1–4]A variety of studies have demonstrated, however, that in addition to effects on the GABAAreceptor, barbiturates also modulate several other ion channels, [5,6]including voltage-gated neuronal calcium channels. [4,7–9]Granule neurons are the most common neurons in the central nervous system, and those isolated from the cerebellum have been used widely in isolated culture for toxicologic and pharmacologic investigation as a well-established, uniform model of a neuronal cell. [11,12]In cerebellar granule (CG) neurons, influx of Ca2+through a variety of types of voltage-gated calcium channels (VGCCs)[12–16]leads to an increase in intracellular Ca2+concentration ([Ca2+]i)and release of the excitatory neurotransmitter glutamate. [17–20]Using CG neurons, we examined the effects of clinical concentrations of the rapidly acting barbiturates thiopental and methohexital on the depolarization-induced increase of [Ca2+]iand the resulting glutamate release. Drug actions were compared with those observed with altered extracellular [Ca2+] and with blockade of specific VGCCs.
Cell Isolation and Culture
Cerebellar granule neurons were prepared using a modification of the method of Novelli et al. Using a protocol approved by the University of Virginia Animal Research Committee, cerebella were dissected from 10 - 14 Sprague-Dawley rat pups 5 - 7 days old. The tissues were cross-chopped into 0.3-mm2blocks and suspended for 45 min at 37 [degree sign]C in solution A, which contained 0.25 mg/ml trypsin type III and (in mM) 140 Na+, 5.4 K+, 1.2 Mg2+, 140 Cl-, 1 H2PO4(-), 14.5 glucose, 25 HEPES, and 0.3% bovine serum albumin, adjusted to pH 7.4. After 45 min, DNase I and trypsin inhibitor were added, and the suspension was gently centrifuged for 2 min. The supernatant was discarded, and the pellet was triturated 70 times in solution A without trypsin. After 5 min, MgCl2(to 2.5 mM) and CaCl2(0.1 mM) were added to solution A, and the cell suspension was collected, filtered through 70-[micro sign]m mesh, and recentrifuged for 2 min. Neurons (2 x 106) from the resuspended pellet (2 x 106) were plated onto glass coverslips (11 x 22 mm, coated with poly-L-lysine or lightly etched with 10 M NaOH) in culture dishes. Neurons were cultured in basal Eagle's medium with 10% heat-inactivated fetal calf serum, 2 mM glutamine, 100 [micro sign]g/ml gentamicin, and 25 mM KCl, a partially depolarizing medium which stabilizes growth of CG neurons and prevents apoptosis. Proliferation of glial cells was prevented by addition of 10 [micro sign]M cytosine arabinoside 24 h after plating. Cerebellar granule neurons were maintained in 5% CO2/95% air at 37 [degree sign]C and were used at 4 - 8 days in culture. At the time of study, 7- to 10-[micro sign]m neurons were clustered in 30- to 70-[micro sign]m aggregations of cells, with “cables” forming connections between the scattered cell aggregates, a pattern identical to that demonstrated previously. 
Biochemical reagents were obtained from Sigma Chemical Company (St. Louis, MO) unless otherwise indicated. Appropriate aliquots of 100 mM thiopental (Gensia Labs, Ltd., Irvine, CA) or 100 mM methohexital (Eli Lilly, Indianapolis, IN) as the Na+salt were added to the solutions 5 min before depolarization by addition of KCl.
Measurement of Intracellular Calcium Concentration in Cultured Granule Neurons
Cytosolic [Ca2+]iwas measured using fura-2. Neurons on coverslips were incubated at 37 [degree sign]C for 20 min in basal medium containing 3 [micro sign]M fura-2-AM (Molecular Probes, Eugene, OR), 16 [micro sign]M bovine serum albumin, and (inmM) 155 Na+, 5.0 K+, 155 Cl-, 5 HCO3-, 1 H2PO4-, 1.25 Mg2+, 1.2 Ca (2+), 5 glucose, and 20 N-tris(hydroxymethyl)-methyl-2-aminoethanesulfonic acid (TES), adjusted to pH 7.4. In some experiments, HEPES was substituted for TES with no alteration in behavior. After washing the neurons twice in fura-2-free solution, coverslips were inserted into a holder and placed in cuvette containing 2 ml of solution; the neurons were then washed twice more with 2 ml of fresh medium. Measurements of [Ca2+]iwere performed at 37 [degree sign]C in a PTI luminescence spectrofluorometer (Photon Technology International, Monmouth Junction, NJ) equipped with a thermostatically warmed cuvette and magnetic stirrer. Fluorescence at 510 nm was determined for alternating excitation wavelengths of 340 and 380 nm; ratios were collected every 0.33 s for 1.5 min. Influx of Ca2+was initiated by rapid addition of 100 [micro sign]mol of KCl (in 33 [micro sign]l) to achieve a final concentration of 55 mM. Calibration was performed by obtaining minimum and maximum fluorescence values using 10 [micro sign]M ionomycin for maximum (Ca2+-saturated) values and 5 mM EGTA for minimum values, respectively, for each coverslip. After calibration, [Ca2+]iwas calculated according to the standard formula of Grynkiewicz et al. using a Ca2+-fura-2dissociation constant of 224 nM, using PTI software (Felix [TM]; Photon Technology International) configured for the analysis. The computed value of [Ca2+]iwas displayed and stored for subsequent analysis.
Measurement of Glutamate Release
Glutamate release was measured by a glutamate dehydrogenase - coupled assay. A coverslip with adherent CG neurons was rinsed with buffer solution and then placed in a cuvette containing 2 ml of buffer solution, which contained 1 mM of the oxidized form of nicotinamideadenine dinucleotide phosphate (NADP+) and 50 U/ml glutamate dehydrogenase (Boehringer Mannheim, Indianapolis, IN, GmbH, Germany) to catalyze from the released glutamate the formation of [Greek small letter alpha]-ketoglutarate and the fluorescent species of the reduced form of NADP (NADPH), as previously described. [20–23]NADPH fluorescence was excited at 340 nm and measured at 460 nm using the PTI spectrofluorometer. Glutamate release was again activated by rapid addition of 100 [micro sign]mol KCl for a final [K+] of 55 mM, and the change in NADPH fluorescence was monitored for 5 min at a sampling rate of 1 - 2 Hz.
To calibrate the fluorescent response to glutamate release, studies were performed with the direct addition of NADPH or glutamate during identical conditions. Addition of NADPH in the cuvette solution to obtain concentrations of 0.2, 0.5, and 1.0 [micro sign]M resulted in abrupt increases in the fluorescence signal of 0.92 +/- 0.21, 1.94 +/- 0.86, and 3.1 +/- 1.31 x 105counts/s, respectively (mean +/- SD, n = 5). When glutamate was added (as aliquots of 0.5 mM solution) to solutions containing 50 mg/ml glutamate dehydrogenase enzyme solution, the fluorescence signal increased with an exponential time course time constant of [almost equal to] 60 s at 37 [degree sign]C. When the added [glutamate] was 0.2, 0.5, and 1.0 [micro sign]M (equimolar to the increases in [NADPH]), the respective steady-state increases in fluorescence signals were 0.89 +/- 0.19, 1.83 +/- 0.29, and 2.89 +/- 0.30 x 105counts/s, or [almost equal to] 95% of the NADPH values. The close agreement of the fluorescence signal between the same quantity of NADPH and glutamate suggests that the glutamate reaction producting NADPH proceeded largely to completion.
With the addition of KCl and the depolarization of neurons grown on coverslips, there was a sudden increase in the fluorescence signal, often followed by a much smaller and slower increase, which typically stabilized by 5 - 15 s (Figure 1B). The increase in the fluorescence signal was typically on the order of 0.4 - 1.5 x 105counts/s, suggesting an immediate glutamate release of 0.1 - 0.2 nmol. The value varied with the degree of confluence and coverage of neurons on the coverslips. Compared with the addition of glutamate in solution, the stabilization of the fluorescence signal in the presence of depolarization-induced glutamate release from neurons was more rapid. Such rapidity suggests that there must be rapid release of a high concentration of glutamate, followed by the rapid arrest (<or= to 1 s) of glutamate release and local uptake of glutamate into neurons to cause cessation of production of NADPH in the first few seconds. A high concentration of glutamate (>100 [micro sign]M) has been predicted in synaptic clefts, whereas a high-capacity system for uptake of glutamate present in neurons could account for the rapid stabilization of the signal. 
Analysis and Statistics
To reduce artifacts attributable to inherent noise in the fluorescence signal, [Ca2+]iwas determined by averaging the individual values obtained during a 10-s period before depolarization (basal), during the 1 - 3 s immediately after depolarization (maximum), and during the final 5 s of the sample (plateau). The peak [Ca2]itransient is reported as the maximum value achieved minus the initial baseline value, although results did not differ if instead the actual maximum [Ca2+]iwas used for the calculation. NADPH fluorescence was calculated similarly except that a 1- to 5-s sample after depolarization was used and the plateau value was not determined. For comparison among various concentrations, results were expressed as the fraction of same-day control to allow for variations that occasionally occurred in cell density and behavior during the period of culture or among different preparations. The dose-dependent drug effects were fit to a standard logistic Equation ofthe form: Drug effect (as fraction of control)= 1/(1 +([drug]/IC50)n), whereas IC50represents the concentration for the half-maximal inhibitory effect and n is the slope factor. Unless otherwise indicated, results are expressed as sample mean +/- sample SD. Statistical test of the drug effect was performed using one-group, two-tailed t test versus unity (or 100%). Differences between groups were tested by analysis of variance and protected least-significant difference test.
As shown by the control response in Figure 1A, a sudden increase in [K+] to 55 mM caused an increase in [Ca2+]ifrom [almost equal to] 40 to >800 nM, which subsequently declined to less than one-half the peak value in <15 s. [Ca2+]itypically reached final plateau levels of 160 - 210 nM by 60 - 90 s. To verify the relation between peak [Ca (2+)]itransient and the fluorescence attributed to glutamate release, a series of experiments was performed in which extracellular [Ca2+]([Ca (2+)]e) was decreased to <1.2 mM. Omission of added external Ca2+in the absence of added EGTA produced a solution in which residual trace [Ca (2+)]ewas typically 0.005 - 0.010 mM (measured by ion-selective electrode). With [Ca2+]eof 0.60, 0.40, 0.20, and [almost equal to] 0.01 mM, the [Ca2+]itransient was reduced to values that were 56 +/- 11, 37 +/- 6, 20 +/- 3, and 8.1 +/- 5.8% of the same day control, respectively (n = 5, 3, 6, and 3, respectively). The plateau of the [Ca2+]itreatment showed slightly more modest decreases to 78 +/- 33, 64 +/- 18, 33 +/- 8, and 32 +/- 19% of control, respectively. For identical decreases in [Ca2+]e, glutamate release decreased to 46 +/- 14, 41 +/- 12, 21 +/- 7, and 9.6 +/- 6.0%(n = 7,7,7, and 4, respectively) of same day control. The decrease was very similar to that for the peak [Ca2+](i) transient. The Ca2+transient and the NADPH fluorescence attributed to glutamate release are shown in Figure 1for 1.2 and 0.4 mM Ca (2+) and for no added Ca2+. Figure 1C shows the relation between the peak Ca2+transient and glutamate release for the reduction in [Ca2+]e. With 1 mM EGTA and no Ca2+added to the extracellular solution, [Ca2+]ewas reduced to <100 nM. The [Ca2+]itransient and glutamate release activated by KCl were then completely abolished, verifying the requirement for external Ca2+to elicit glutamate release.
In CG neurons, the contribution to the Ca2+transient of [Ca (2+)]icurrent through L-type, N-type, or P/Q-type VGCC was determined, respectively, by use of 1 [micro sign]M nicardipine, 100 nM [Greek small letter omega]-conotoxin-GVIA ([Greek small letter omega]-CgTx), or [Greek small letter omega]-Agatoxin-IVA (Aga-IVA)(Alexis Biochemical, San Diego, CA). [26–28]Neurons were exposed to nicardipine, [Greek small letter omega]-CgTx, or Aga-IVA for 20 min before experimental measurement. The alterations in the peak of the Ca2+and the NADPH signal attributed to glutamate release in the presence of these specific VGCC blockers are listed Table 1. Each blocker decreased the Ca2+transient by [almost equal to] 45% with less prominent and more variable effects on glutamate release. The more modest effects of [Greek small letter omega]-CgTx on glutamate release compared with Aga-IVA or nicardipine agree with previous reports suggesting that N-type channels contribute modestly to glutamate release. [18,20]When combined, the nicardipine and [Greek small letter omega]-CgTx showed an effect greater than either drug alone, but the increase was nonadditive. Such overlapping drug sensitivity also has been previously reported. In two experiments, all three agents were combined, and the residual Ca2+transient was reduced to 16 and 21% of control, suggesting that the bulk of the Ca2+transient was VGCC-mediated Ca2+entry.
In theory, the KCl-induced depolarization should inactivate sodium channels after an initial opening (<10 ms) so that the Ca2+transient should reflect Ca2+entry mediated by opened VGCCs. To determine if any of the observed effects were mediated by actions on sodium channels, depolarization also was performed in the presence of 10 [micro sign]M tetrodotoxin to block sodium channels. A small ([almost equal to] 10%) fraction of the Ca2+transient was depressed by the presence of tetrodotoxin, and its action of similar magnitude inhibiting glutamate release did not achieve significance. To verify that the observed events were activated by depolarization, and not the 28% increase (+100 mOsm) is osmotic strength, 100 [micro sign]mol of NaCl was added instead of KCl, with no effect on the [Ca2+]itransients or on glutamate release (n = 3).
Additional control experiments were performed to exclude possible effects of barbiturates on other processes in CG neurons known to influence glutamate release and VGCC function, such as GABAA, GABA type B (GABA (B)), and N-methyl-D-aspartate (NMDA) receptor activation. [30,31]GABA (A) receptors (chloride ion channels) were blocked using 100 [micro sign]M bicuculline; GABABreceptors were activated by 10 [micro sign]M baclofen; and NMDA glutamate receptors were inhibited by D-(-)-2-amino-5-phosphonovaleric acid. None of these interventions had any significant action on the depolarization-evoked Ca2+transient or glutamate release (Table 1). To determine if Ca2+release from intracellular stores (endoplasmic reticulum) contributed to the Ca2+transient or glutamate release, intracellular Ca2+release was manipulated by modifying the function of the ryanodine receptor (Ca2+release channel). Neurons were incubated with 10 [micro sign]M ryanodine or 5 mM caffeine for 20 min before depolarization. During these conditions, ryanodine caused a small decrease (14%) in glutamate release (Table 1), suggesting that internal Ca2+release could contribute modestly to glutamate release in this setting.
Effects of Barbiturates
Administration of thiopental or methohexital did not alter the basal [Ca2+]ibut did markedly depress the depolarization-induced influx of [Ca2+]i. In the presence of thiopental or methohexital (Figure 2A and Figure 2B) the peak [Ca2+]itransient phase was decreased in a concentration-dependent manner. When analyzed as the fraction of same-day controls, the decrease in the peak [Ca2+]itransients by the varying concentrations of barbiturate were fit by a logistic equation (Figure 2C). The calculated IC50(mean +/- SEM) for thiopental was 34.5 +/- 4.5 [micro sign]M, 3.7 times greater than the value for methohexital of 9.6 +/- 2.0 [micro sign]M. The plateau phase of the Ca2+transient was also depressed by the barbiturates but typically not to the same extent as the peak [Ca2+]i.
Detailed studies of the effects of thiopental were performed in the presence of specific calcium channel blockade. In the presence of a near IC50concentration of thiopental of 30 [micro sign]M, the [Ca2+](i) transient peak was depressed to 56 +/- 9% of control (n = 8). A similar decrease to [almost equal to] 55% of control was observed with 1 [micro sign]M nicardipine, 100 nM [Greek small letter omega]-CgTx (n = 14), or 100 mM Aga-IVA (n = 4). Nevertheless, when 30 [micro sign]M thiopental was present with 1 [micro sign]M nicardipine or [Greek small letter omega]-CgTx, a greater decrease in the Ca2+transient was observed than seen with complete blockade of either the L- or N-type channels (Figure 3). These results suggest that thiopental was depressing multiple types of calcium channels. Conversely, a combination of either 1 [micro sign]M nicardipine or 100 nM [Greek small letter omega]-CgTx with 30 [micro sign]M thiopental caused a significantly greater decrease in the [Ca2+]itransient peak compared with thiopental alone, suggesting that this near IC50concentration of thiopental had not completely blocked either N- or L-type channels. When L- and N-type channels were blocked by the combination of nicardipine and [Greek small letter omega]-CgTx, the peak Ca2+transient was decreased (40 +/- 12% of control) more than with either agent alone (P < 0.01). When 30 [micro sign]M thiopental was also present, there was further marked decrease in the transient (Figure 3). This result suggests that 30 [micro sign]M thiopental is able to depress Ca2+entry profoundly through the P/Q-type and R-type (residual or blockade-resistant) VGCCs that remain functional in the presence of nicardipine and [Greek small letter omega]-CgTx.
The inset in Figure 4A shows the increase in NADPH fluorescence observed with KCl depolarization, which represents formation of NADPH from glutamate and NADP+. Thiopental caused a dose-dependent reduction in glutamate release in these neurons (Figure 4A), which, by logistic analysis, was found to have an IC50value of 12.5 +/- 1.4 [micro sign]M. In the presence of 10 [micro sign]M methohexital, glutamate release was decreased to 41 +/- 4% of same-day control, significantly less than the decrease to 54 +/- 7% of control observed with 10 [micro sign]M thiopental. To delineate more clearly the relation between effects on the peak [Ca2+]itransient and on glutamate release, the average fraction decrease in glutamate release is plotted versus the average decrease in the peak [Ca (2+)]itransient for the various barbiturate concentrations studied (Figure 4B). Clearly, there is a parallel decrease in the transient peak and glutamate release with either barbiturate, similar to the action observed in decreasing Ca2+entry by reductions in [Ca2+]e.
The current results have implications regarding mechanisms of anesthetic action and possibly the neuroprotection afforded by barbiturates during certain circumstances. Although it is clear that barbiturate activation of GABAAreceptors/chloride channels mediates their hypnotic effect and may contribute to their anticonvulsant actions, [1–3,32]concentrations that produce anesthesia and lack of response to pain are typically higher. Previous studies have suggested the ability of barbiturates to inhibit Ca2+currents [8,9,34,35]and the depolarization-induced increase in [Ca2+]iat higher concentrations. Macdonald et al. have suggested that for pentobarbital, although low concentrations can provide sedation, the higher concentrations required for complete anesthesia or coma are those that block calcium channels in neurons. 
Barbiturate-induced decrease in Ca2+currents has been previously described in neurons [4,34,35]and cardiac myocytes and for Ca2+currents in Xenopus oocytes resulting from injection of brain-derived messenger ribonucleic acid. The observed reduction in Ca (2+) transients by barbiturates is consistent with a decrease in the influx of Ca2+through VGCCs. Because KCl depolarization was unable to increase [Ca2+]iin solutions that lacked Ca2+, it is likely that KCl-induced increase of [Ca2+]iresulted from the influx of Ca2+through VGCCs. The action of barbiturates observed in this study is similar to the effect of methohexital on cultured embryonic rat hippocampal neurons reported by Bleakman et al., in which 5 - 20 [micro sign]M methohexital caused a 40 - 50% reduction in the peak of the KCl-induced Ca (2+) transient. 
Cultured granule neurons have become a well-established model for the study of neuronal function, because their responses to various stimulatory paradigms are well described and their influx of Ca2+has been studied by radioisotopic tracer, optical, and electrophysiologic methods. Using a series of specific VGCC inhibitors, five distinct channel types (L, N, P, Q, and R) have been identified in cultured rat CG neurons. [15,16,37]In electrophysiologic studies of CG neurons, Aga-IVA results in a decrease of Ca2+current by 50 - 70%, blocking P-type VGCCs at lower concentrations ([almost equal to] 1 nM) and Q-type channels at higher concentrations ([almost equal to] 100 nM). Blockade of Ca2+current using [Greek small letter omega]-CgTx has shown variable effects, causing decreases of 30 - 63%; dihydropyridines show decreases ranging from 20 - 70%. [15,16,29]In this study, nicardipine, [Greek small letter omega]-CgTx, and Aga-IVA all produced decreases in the peak [Ca2+]itransient to [almost equal to] 55% of control, compatible with the electrophysiologic studies. [16,29]When L- and N-type blockade were combined, the effect on the [Ca2+]itransient peak was not strictly additive in that 40% of the control peak [Ca2+]iwas still present. Such nonadditive decreases have been previously reported in an electrophysiologic investigation and may represent overlapping sensitivity between the blockade by these two agents. Nevertheless, the effect of the combined agents was greater than either agent alone, suggesting that overlap was incomplete. When L-, N-, and P/Q-type VGCCs are inhibited by combined application drug and toxins, the 19% of total Ca2+current that remains has been attributed to the R-type channels. Although some discrepancy between voltage clamp studies and the current measures of [Ca2+]iwould be anticipated based on the finding that the fluorometrically measured [Ca2+]imay not be a linear function of Ca2+currents, a surprisingly similar fraction of the [Ca2+] transient ([almost equal to] 19%) remained when combined L-, N-, and P/Q-type blockade was used in this study.
To assess whether thiopental was decreasing Ca2+influx by inhibiting specific VGCCs in these CG neurons, nicardipine (1 [micro sign]M) and [Greek small letter omega]-CgTx (100 nM) were used at concentrations sufficient to provide complete blockade of L- and N-type channels, respectively. [20,29]These studies of nicardipine or [Greek small letter omega]-CgTx with 30 [micro sign]M thiopental, which caused a similar decrease of peak [Ca2+]i, permit certain qualitative conclusions. Because either L- or N-type blockade did cause an additional decrease in [Ca2+](i) when combined with 30 [micro sign]M thiopental, thiopental could not have been completely blocking either L- or N-type VGCCs. Conversely, the finding that thiopental by itself decreased the peak [Ca2+]ito the same extent as nicardipine or [Greek small letter omega]-CgTx current suggests it must have been blocking more than one type of channel. After blockade of both L- and N-type VGCCs, the transient is reduced by 60%, yet addition of 30 [micro sign]M thiopental causes an additional 25% inhibition of the Ca2+transient, suggesting that thiopental is blocking Ca2+current through the P/Q- and R-type calcium channels, which conduct the complement of Ca2+current in these neurons. Consequently, the inhibition of VGCCs by thiopental seems to be relatively nonspecific with regard to VGCC type.
Release of neurotransmitters from neurons is mediated by entry of Ca2+into nerve terminals, activating a complex of proteins that cause fusion of the membrane of the transmitter-containing synaptic vesicle with the cell membrane, resulting in exocytosis. In the current study, the actions of the barbiturates in decreasing glutamate release can be attributed to their depression of VGCC-mediated entry of Ca2+. In these CG neurons, as with other central synapses, glutamate release does not appear to be associated with any specific calcium channel, as blockade of various VGCCs is capable of decreasing glutamate exocytosis from granule cells. Voltage-gated calcium channel blockade of N-type channels using [Greek small letter omega]-CgTx, in some instances, has been found not to inhibit glutamate release [18,20]; however, the brief preincubation periods (5 and 8 min) used in these studies may not have been sufficient to observe the decrease in neurotransmitter release reported for other neurons. [39,40]In preliminary experiments, we found that [Greek small letter omega]-CgTx applied for 5 min provided only modest blockade of the [Ca2+] transient, whereas 20 min caused the greater blockade reported in Table 1. The activation of release of internal Ca2+stores by metabotropic glutamate receptors does not appear to occur in these neurons grown during these mildly depolarizing conditions. Although certain components of entry of Ca (2+) have been attributed to entry of Ca2+via other pathways, such as NMDA receptors, block of NMDA receptors also did not have a prominent effect on the depolarization-evoked response in these cells in this study.
In the physiologic setting, when exocytosis is activated by action potentials, the transmitter release is reduced in proportion to some power function of the reduction in [Ca2+], suggesting there is cooperativity in Ca2+dependence and that two or more Ca2+must bind to specific sites. [39,41–43]In this study, the prolonged depolarization induced by KCl produced sustained rather than intermittent entry of Ca2+, which may result in saturation of activating Ca2+sites, resulting in the more linear relation between the peak Ca2+and glutamate release (Figure 1C). In addition, the finding that the measured [Ca2+]imay reflect [Ca2+] at other than intrasynaptic sites means that such quantitative correlations should be interpreted cautiously. Although the VGCC inactivates during this sustained depolarization, electrophysiologic studies of Ca2+currents in CG neurons suggest that inactivation is incomplete at 1 - 3 s (38% inactivation at 0.72 s), which was the period in which sampling of the peak Ca2+transient and glutamate release were performed. Nevertheless, barbiturate actions during physiologic conditions in which action potentials activate VGCC may differ markedly from those described here, because of the intermittent, brief periods of VGCC activity generated by action potentials and the effects of barbiturates on other ion channels, which could alter action potential conduction and configuration.
The [almost equal to] 80% decrease in the Ca2+transient by blockade of drug- and toxin-sensitive VGCCs (not including R-type VGCC) suggests the Ca2+transient in this setting of KCl depolarization is derived largely from entry of Ca2+through VGCC. Because the IC50value for the effects of thiopental on glutamate release was approximately one third that for decrease in concentration, however, thiopental might alter other cellular pathways contributing to glutamate release. Exocytosis also may involve release of Ca2+from internal stores in some instances. Although in cultured CG neurons G protein-linked receptor activation of intracellular Ca2+can be altered by caffeine or ryanodine, manipulation of internal Ca2+release channels did not alter the depolarization-induced Ca2+transient in this study. The modest action of ryanodine in modestly decreasing glutamate release suggests, however, that thiopental actions on internal Ca2+release could contribute to its actions. A thiopental-induced decrease in release of [Ca2+]ihas been described in myocardium for the concentrations used in this study and might explain the slightly greater decrease in glutamate release compared with the Ca2+transient. Although glutamate is the primary excitatory transmitter in the nervous system, a previous study using striatal synaptosomes from rat found that thiopental also is able to depress a depolarization-induced increase in dopamine. 
The low and intermediate concentrations of barbiturates used represent concentrations of free (unbound) drug that may occur clinically. A 6-mg/kg intravenous bolus dose of thiopental results in a peak total serum concentration of 93 [micro sign]g/ml, whereas the average concentration of anesthetic agent in plasma is 12.6 [micro sign]g/ml. Assuming 83 - 86% plasma protein binding, the resulting free thiopental concentrations should be 7.2 - 60.0 [micro sign]M (1.9 - 16.0 [micro sign]g/ml). A concentration of thiopental of 7.3 - 7.4 [micro sign]g/ml (31 [micro sign]M total, 5 [micro sign]M free drug) represents the concentration in plasma required for “sleep” in 50% of patients, and it is the concentration that induces a 50% halothane minimum alveolar concentration reduction in rats. A more than fourfold greater concentration of 32 [micro sign]g/ml (133 [micro sign]M total, 21 [micro sign]M free drug), however, was required to achieve a 90% reduction in minimum alveolar concentration. Regarding methohexital, a 3-mg/kg induction dose and a 0.05-mg [middle dot] kg-1[middle dot] min-1infusion yielded concentrations in plasma of 2.8 [micro sign]g/ml (10 [micro sign]M) in human volunteers, which, considering the 73% plasma protein binding, represents a free concentration of [almost equal to] 2.7 [micro sign]M. Therefore, the lower concentrations of thiopental (10 [micro sign]M) and methohexital (3 [micro sign]M) used in this study, which represent routine clinical concentrations of free drug, were able to produce a decrease of 30% in the peak Ca2+transient and a >40% reduction in glutamate release for thiopental. The greater potency of methohexital compared with thiopental in inhibiting the [Ca2+]itransient peak and glutamate release appears to reflect the greater anesthetic potency of the former compound.
The ability of thiopental and methohexital to decrease intracellular Ca2+transients and to decrease glutamate release provides a mechanism, through blocking excitatory synaptic transmission, by which they may induce a component of the anesthetic state. Such an action may be additive or synergistic, with the sedative/hypnotic actions mediated by the hyperpolarizing and inhibitory effects of GABAAreceptor activation. In addition, decreased excitatory synaptic transmission and inhibition of entry of Ca2+would be expected to decrease metabolic neuronal activity, which might explain the decreased oxygen consumption associated with administration of barbiturates. Although these inhibitory actions on VGCC and glutamate release may contribute to the anesthetic state, they also may play a beneficial role in neurons subject to ischemic or hypoxic injury. Because glutamate release activates NMDA receptors and Ca2+entry, a barbiturate-mediated decrease in its release, which also has been demonstrated in brain slices, could be protective by preventing excitatory and potentially neurotoxic Ca2+entry. In the presence of depletion of adenosine triphosphate, however, glutamate release from neurons may not be mediated by synaptic processes but may instead represent failure or reversal of processes of glutamate uptake. Although inhibition of reversal of glutamate uptake has been demonstrated, an additional important protective aspect of barbiturates may be the blockade of entry of Ca2+through VGCC. Accumulation of Ca2+through VGCCs, by itself, also is protective by decreasing the entry of Ca2+, which ultimately mediates neurotoxicity. Blockade of N-type VGCC by [Greek small letter omega]-conotoxin MVIIA (SNX-111) has been shown to be effective in decreasing neuronal loss and infarct size in models of global or focal cerebral ischemia, an effect also demonstrable for thiopental. It is noteworthy that the protection afforded by SNX-111 was effective even when administered 24 h after the insult, suggesting that entry of Ca2+via specific VGCCs may be important in mediating later events in the cascade that mediate neuronal death. In addition, blockade of P-type VGCC by [Greek small letter omega]-conotoxin MVIIC (SNX-230) decreased glutamate release but did not provide protection against the ischemic injury. In models examining neural protection from ischemia, doses of thiopental and methohexital are used that are far higher than those used for anesthetic induction. Such doses are likely to achieve sustained concentrations in the intermediate range studied and thereby cause more profound reductions in Ca (2+) entry.
Although barbiturate anesthetic agents have been found to have effects on a variety of ion channels, the barbiturate induction agents thiopental and methohexital at equipotent concentrations depressed VGCC-mediated Ca2+entry into cultured CG neurons and the associated glutamate release. The direct depression of Ca2+entry may contribute to their anesthetic action and their neuronal protective action.
The authors thank John Althaus of Pharmacia and Upjohn Pharmaceuticals for demonstrating cell isolation; Martha J. Frazer for help in initial experiments; and Marcel E. Durieux, M.D., for review of the manuscript.