Modulation of γ-Aminobutyric Acid Type A Receptor–mediated Spontaneous Inhibitory Postsynaptic Currents in Auditory Cortex by Midazolam and Isoflurane

All experimental protocols conformed to American Physiologic Society/National Institutes of Health guidelines and were approved by the University of Wisconsin Research Animal Resources Committee (Madison, Wisconsin). Male and female CBA/J mice (P28-P35) were decapitated during isoflurane anesthesia, and the brain was extracted and immersed in artificial CSF (ACSF; composition: 126 mm NaCl, 26 mm NaHCO³, 1.8 mm KCl, 2.1 mm CaCl², 1.4 mm MgSO⁴, 1.2 mm KH²PO⁴, and 10 mm glucose) at 0°–4°C. Slices (500 μm) were obtained from the left hemisphere in a plane 15° off the horizontal plane, as described for auditory thalamocortical slices.20Slices were maintained in ACSF saturated with 95% O²–5% CO²at 34°C for 1 h followed by 24°C for 1 h before transfer to the recording chamber, which was perfused at 3 ml/min with ACSF saturated with 95% O²–5% CO²at 34°C.

Layer IV and V pyramidal cells in ACx were visualized and identified based on morphology using a video camera (VE-1000; DAGE MTI, Michigan City, IN) connected to an upright microscope (BX-50WI; Olympus America Inc., Melville, NY) equipped with an infrared band-pass filter (775 ± 75 nm), a long working-distance water-immersion objective (Olympus, 40X, N.A. 0.7), and differential interference contrast optics. Pyramidal cells were identified by the triangular shape of their somata, prominent single apical dendrites, and two smaller caliber basal dendrites. The microscope, recording, and stimulating electrodes were under remote control using an integrated motorized control system (Luigs & Neumann, Ratingen, Germany). Most cells were also labeled with 0.3% biocytin, and morphologic identification was confirmed after processing the tissue using standard histologic techniques.21 

Patch pipettes were fabricated from borosilicate glass (KG-33, 1.7-mm OD, 1.1-mm ID; Garner Glass, Claremont, CA) using a Flaming-Brown two-stage puller (P-87; Sutter Instruments, Novato, CA). Patch pipettes had open-tip resistances of 2–4 MΩ. Whole cell voltage clamp recordings were obtained at 34°C with patch pipettes filled with 100 mm KCl, 40 mm K-gluconate, 10 mm NaCl, 10 mm HEPES, 0.1 mm EGTA, 4 mm MgATP, and 5 mm QX-314 (pH 7.3). Access resistances were less than 20 MΩ and compensated by 20–60%. Cells were held at −60 mV. Spontaneous IPSCs (sIPSCs) were isolated by bath application of 4 mm kynurenic acid to block ionotropic glutamate receptor–mediated currents. All remaining currents were blocked by 20 μm bicuculline (n = 5 cells; data not shown). Kynurenic acid, bicuculline, and all components of the pipette solution and control ACSF were obtained from Sigma-Aldrich (St. Louis, MO).

After forming a gigaohm seal and rupturing the membrane to achieve intracellular access, glutamate receptor antagonists were allowed to equilibrate for 5 min before recording a baseline control period of 15 min. Midazolam was then bath applied for 20–25 min (0.1 and 1 μm) or 35 min (0.01 μm). The midazolam data were taken from the final 5 min of drug application. It was not possible to obtain reliable recovery data from slices exposed to midazolam even when slices were washed for more than 30 min. Midazolam hydrochloride (injectable form, pH approximately 3.0) was obtained from Bedford Labs (Bedford, OH) as a 1-mg/ml solution; to obtain a solution of 1 μm, this stock solution was diluted 2,500-fold.

Isoflurane (Novaplus; Abbott Labs, N. Chicago, IL) was bath applied to slices as follows. Isoflurane was prepared as an aqueous solution from a saturated stock solution (15 mm in ACSF22) and diluted to final concentration in ACSF on the day of the experiment. The isoflurane solution was prepared in 500 ml Teflon gas sampling bags (Fisher Scientific International Inc., Hampton, NH; cat. No. 10-923-5) that contained ACSF bubbled with 95% O²–5% CO². Teflon tubing was used between the ACSF reservoirs and the recording chamber to minimize loss of the volatile agent before it reached the tissue. Preliminary experiments in which isoflurane concentration was measured in the liquid phase using Ca²⁺-sensitive electrodes23indicated that approximately 15% of the isoflurane was lost before it reached the center of the recording chamber, similar to a previous study using identical techniques and equipment.24We calibrated the Ca²⁺-sensitive electrode by determining the relation between electrode potential and isoflurane concentration in solutions bubbled with isoflurane gas, and in parallel with solutions prepared as for our slices from stock concentrations assuming a saturating concentration of 15 mm.23We then measured the electrode potential in the slice chamber while perfusing with different isoflurane concentrations and calculated percent loss according to the measured linear relation between electrode potential and isoflurane concentration. This loss was accounted for in the reported values of the concentration of isoflurane used in these experiments. A drug application protocol identical to that used for the midazolam experiments was used for the isoflurane experiments, i.e. , 5 min equilibration with kynurenic acid, 15-min baseline recording period, 20- to 30-min drug recording period. Recovery data were obtained in the isoflurane experiments by recording for more than 30 min after removal of isoflurane from the bathing medium.

We typically observed a significant decrease in mean interevent interval over time, independent of whether midazolam or isoflurane was applied. This likely reflects changes in network activity under our recording conditions over time, unrelated to the application of midazolam or isoflurane, e.g. , due to application of glutamate receptor antagonists.25Because we were focused on the effects of midazolam and isoflurane on the time course of IPSCs and because similar effects on mean interevent interval were observed independent of drug condition, we do not anticipate that this phenomenon has any effect on our results.

Data were amplified (MultiClamp-700A; Axon Instruments, Union City, CA), low-pass filtered (10 kHz), digitized (20 kHz, DigiData 1322A; Axon Instruments), and recorded using pCLAMP v8.2 (Axon Instruments).

Data were analyzed using ClampFit (Axon Instruments) and Origin (OriginLab Corporation, Northampton, MA). Spontaneous IPSCs were analyzed using an automated event detection algorithm.26In this algorithm, two windows were moved along the data, a “peak” window and a “baseline” window. At each time point, the data within each window was averaged and the baseline point subtracted from the peak. This yielded a pseudodifferentiated form of the data that was characterized by large, rapid peaks at the onset of GABA^AIPSCs. Threshold-level crossings were identified in this pseudodifferentiated data, with threshold set as 3*ς^Noise, where ς^Noisewas measured during periods of no visually detectable events and was typically 3–5 pA. Because the baseline value was constantly updated during the analysis, slow changes in baseline had no effect on the accuracy of the algorithm. This algorithm successfully detected more than 99% of sIPSCs. Individual sIPSCs (n > 100 in each cell for each condition) were selected for averaging when no other detected events occurred within ±100 ms of the peak. Averaged sIPSCs were then used to measure 10–90% rise times (t^Rise). Decay phases were best fit by two exponential components, and decay kinetics were characterized in addition by the weighted sum of the exponential decay components, τ^Dec,Wt= A¹τ¹+ A²τ², where A¹and A²are the amplitudes of the decay components and τ¹and τ²are their time constants. The time courses of individual, uncontaminated events were also characterized by the time to decay from peak to within 37% of the baseline. Statistical comparisons were performed using the nonparametric Wilcoxon signed rank test with a significance level of 0.05, because 12 of 48 variables tested did not pass the Shapiro-Wilk normality test (P < 0.05). Identical results were obtained with the Student t  test. Data are expressed as mean ± SD.

Spontaneous IPSCs were recorded from layer IV and V pyramidal cells in ACx in the presence of ionotropic glutamate receptor antagonists at −60 mV and 34°C and served as an assay of IPSC properties under physiologic conditions (i.e. , release from small numbers of release sites due to action potential activity or spontaneous vesicle fusion). Recordings were obtained from 73 pyramidal cells. Most sIPSCs had rapid rise kinetics (t^Rise< 0.6 ms) and biexponential decay kinetics (see below). A small fraction (typically < 1%) of the sIPSCs in these cells had slower rise times (> 2 ms) and slower decay times (> 15 ms) than the majority of the recorded sIPSCs, presumably reflecting a second population of distal dendritic inhibitory inputs, as observed in the hippocampus.26Because of the small number of these events, it was not possible to obtain reliable information on pharmacologic modulation, and we excluded these events from our analyses by restricting individual sIPSC rise times to be less than 1 ms. The remaining sIPSCs on which these data are based occurred at rates of 7.8 ± 6.6 Hz and had rapid rise and decay kinetics (t^Rise= 0.33 ± 0.087 ms; τ¹= 3.25 ± 1.19 ms, A¹= 0.61 ± 0.12, τ²= 13.4 ± 4.08 ms).

Previous studies have shown that the modulation of binding and gating kinetics of GABA^Areceptors by benzodiazepines is manifested at GABAergic synapses by an increase in the amplitude or the duration of GABA^Areceptor–mediated synaptic currents or both.6,27,28We observed no consistent effects of midazolam on the mean peak amplitudes of sIPSCs (table 1and fig. 1). In contrast, bath application of 1 μm midazolam prolonged the decay of spontaneous IPSCs by approximately 70% (table 1and fig. 1). These effects were observed in every cell tested, indicating that in layer IV and V pyramidal cells in ACx, most GABA^Areceptors are sensitive to benzodiazepine agonists. (Distributions of 63% decay times within cells were tested with the Kolmogorov-Smirnoff test [P < 0.05].) Application of 0.1 μm midazolam caused a smaller (approximately 24%) increase in decay time course of sIPSCs, whereas 0.01 μm midazolam produced no significant effects on sIPSCs (table 1and figs. 2 and 3). The latter observation should be interpreted with caution, however, because the value of SD for τ^Dec,Wt(table 1) precludes detecting changes in this parameter smaller than approximately 23% (power analysis with α= 0.05, β= 0.8). However, plotting the decay time as a function of time during the experiment (fig. 2A) strongly suggests that this concentration of midazolam produced no change in decay time course of sIPSCs. Changes in decay time course were not observed in the absence of drug application or when the benzodiazepine antagonist flumazenil (5 μm) was coapplied with 1 μm midazolam (table 1and fig. 3). Significant decreases in mean interevent interval were observed during bath application of 0.01 and 1 μm midazolam, but these effects were likely secondary to nonspecific changes in receptor and network properties over time, because they were also observed under control conditions and in the presence of 5 μm flumazenil (data not shown).

Recent reports suggest that changes in tonic inhibition secondary to either direct activation of GABA^Areceptors or increases in the level of or sensitivity to ambient γ-aminobutyric acid may mediate many of the suppressive effects on cortical circuit activity of benzodiazepines and general anesthetics.29However, we observed no drug-specific effects on holding current in these experiments. The magnitude of the holding current typically increased by approximately 50 pA over the hour-long course of a recording (e.g. , fig. 1), and no significant differences were observed between the magnitude of this change in holding current for control versus  drug conditions (one-way analysis of variance, F^4,47= 2.523, P > 0.05).

Isoflurane is an inhalation anesthetic that in addition to GABA^Areceptors is reported to target other protein targets in cortical cells.3We tested the effects of isoflurane on sIPSCs in pyramidal cells in ACx to compare with the GABA^Areceptor–specific drug midazolam. We chose three concentrations of isoflurane, 0.l mm (approximately 0.3 minimum alveolar concentration [MAC]), 0.25 mm (approximately 0.85 MAC), and 0.5 mm (approximately 1.7 MAC) to span the range of doses evaluated for their effects on MLAERs in previous studies. At 0.1 mm, isoflurane reversibly prolonged sIPSC decay by approximately 45% (table 1and fig. 4; data after wash of isoflurane not shown), an effect intermediate between that observed in response to 0.1 and 1 μm midazolam. The decay of sIPSCs was reversibly prolonged by approximately 140 and 230% by 0.25 and 0.5 mm, respectively, effects far greater than those achieved by even the highest concentration of midazolam tested (table 1and fig. 4; data after wash of isoflurane not shown). No significant changes in holding current were detected (isoflurane vs.  control, P > 0.05, Student t  test; not shown). Similar to previous reports in the hippocampus and neocortex,24the highest dose of isoflurane caused a significant decrease in the mean amplitude of sIPSCs (table 1).

The time course and magnitude of GABA^Areceptor–mediated inhibition are important for determining the ability of interneurons to regulate integration and spike initiation in pyramidal cells. There is an ongoing debate in the anesthesia research community about the relative importance of the various protein targets of general anesthetics such as isoflurane for the effects of these agents in vivo . We have taken the approach of comparing quantitatively the effects of isoflurane with a receptor-specific agent, midazolam. We have shown that sIPSCs in the primary ACx of mice were prolonged by bath application of both drugs in a dose-dependent manner.

Cell type–specific expression of GABA^Areceptors in neocortical circuits may contribute to circuit-specific modulation by anesthetic agents and thus to distinct sites of action and dose dependencies for different behavioral endpoints.30We focused on layer IV and V pyramidal cells of ACx because these cells are likely to participate in the MLAER, as they are activated by thalamic excitation both directly and via  local excitatory interneurons and are inhibited by feedback and feedforward GABAergic inhibition.31–33However, the cellular origin of the MLAER is likely to extend to other cell types as well, because this multiphasic evoked response contains components with latencies ranging from 8 to 80 ms and spans multiple cortical areas.10In relating the experiments described here to modulation of the MLAER in vivo , we focused on effects of the drugs on sIPSCs in concentration ranges spanning those known to modulate the MLAER in vivo , the assumption being that if modulation of sIPSCs was observed, this modulation likely contributed to the effects observed on the MLAER, whereas if no modulation of sIPSCs was observed, effects on the MLAER likely were mediated by actions on other cell types in ACx. One challenge in this approach is relating the pharmacology of evoked responses recorded in humans to those recorded in rodents. The MLAER recorded from the scalp in humans consists of a series of peaks and troughs with latencies ranging from approximately 8 ms to approximately 80 ms. In rodents, the latencies of the first two pairs of peaks and troughs, spanning latencies of approximately 6–30 ms,34are nearly identical to those recorded in humans,7and there is substantial evidence for a cortical basis of these components in both species.35Effects of anesthetic agents on the MLAER are also similar in humans and rodents,7,34,36,37although the effects of midazolam have not been reported. This similarity in time course, cellular generators, and pharmacology of the MLAER in humans and rodents supports the validity of the correlative approach we have used.

We found that under control conditions, sIPSCs recorded in layer IV and V pyramidal cells in ACx had kinetics comparable to those reported for other primary sensory areas in rat neocortex, with rise times of approximately 0.3–0.4 ms and decay time constants of approximately 5–10 ms.38,39Surprisingly, the kinetics of IPSCs recorded in our experiments were substantially slower than those reported for similar experiments performed in rats,40presumably reflecting a species difference restricted to ACx.

We observed that midazolam at either 0.1 or 1 μm but not 0.01 μm prolongs the duration of IPSCs in pyramidal cells in ACx, consistent with its effects in other cortical areas. The effects of midazolam on the decay time of IPSCs reported here are similar to those observed in hippocampal pyramidal cells, in which midazolam at 1 μm prolonged τ^Decby 50–75%,29,41but no effects were observed in response to 0.01 μm.29Much larger effects were reported for concentrations greater than 1 μm in dentate granule cells.6,42The effects of midazolam on neocortical pyramidal neurons have not been reported previously, but the benzodiazepine receptor agonist zolpidem at 1 μm produced effects in layer V pyramidal cells of visual cortex similar to those reported here.28,43 

These effects on IPSCs can be compared with isoflurane, an anesthetic agent that has multiple protein targets in the brain in addition to GABA^Areceptors. Here, we found that 0.1 mm isoflurane, corresponding to approximately 0.3 MAC,44prolonged IPSCs by approximately 50%, intermediate to the effects of 0.1 and 1 μm midazolam, whereas 0.25 mm and 0.5 mm isoflurane (approximately 0.85 and 1.7 MAC, respectively), prolonged IPSCs to a far greater degree than 1 μm midazolam. These results are similar to a previous study in hippocampus, in which isoflurane prolonged IPSC decays by 75% at approximately 0.15 mm and 150% at approximately 0.3 mm,24although larger effects were observed in a separate study in cerebellum.45 

The appropriate concentration to use in vitro  is difficult to determine for a drug such as midazolam, because it is typically given as a bolus injection and complicated kinetic modeling is required to determine approximate CSF concentrations during the period after the injection. In one study, steady plasma concentrations of approximately 200 ng/ml midazolam were obtained by combining a bolus injection of 0.2 mg/kg midazolam and a continuous perfusion of 0.1 mg · kg⁻¹· h⁻¹for 2 h.15At steady state, the plasma concentration of free midazolam is the same as the CSF concentration, and therefore, these data can be used to approximate the CSF concentration of midazolam given an estimate of the free unbound fraction of midazolam in plasma, measured at approximately 2% in humans,46and the molecular weight of midazolam (362.25 g/m). These data yield an estimated CSF concentration of approximately 0.01 μm midazolam, consistent with what is currently considered the clinically relevant concentration to use in vitro  29but which produces negligible effects on IPSCs and holding current in cells in ACx (table 1and figs. 2 and 3). This is in contrast to pyramidal cells in hippocampus, in which 0.01 μm midazolam produced a significant increase in holding current although no effects on the time course of IPSCs.29Direct measurements of the concentration of midazolam in the brain after bolus or continuous perfusion of benzodiazepines in vivo  or both will be a significant step toward understanding the mechanism and locus of benzodiazepine effects in cortex.

If these estimates of the concentration of midazolam in CSF are accurate, it has implications for determining the cellular basis for modulation by midazolam of activity in ACx in vivo . Investigations of the effects of benzodiazepines on ACx in vivo  are primarily limited to studies of the modulation of auditory evoked responses in humans and other animals. The effects of midazolam on the MLAER depend on the method of drug application: Bolus injections at doses sufficient for surgical procedures (0.2–0.3 mg/kg) produce only modest effects on latencies and amplitudes of the components of the MLAER.16,47However, in both of these studies, recordings of the MLAER were taken only at one or two time points within 10 min of the bolus injection, and therefore, it is possible that delayed effects of midazolam were not detected. In contrast, bolus injections of 0.2 mg/kg followed by continuous perfusions of midazolam (0.1 mg · kg⁻¹· h⁻¹for 2 h) produced decreases in component amplitudes of 35–40%,15nearly as large as those reported for isoflurane at approximately 0.3 MAC in humans and rats (45–50%),14,34although increases in latencies due to midazolam (5–10%) were smaller than those reported for low doses of isoflurane (20–40%). We and others have observed that this concentration of isoflurane is sufficient to prolong IPSCs in cortical cells by approximately 50–200% (fig. 4).5,24If midazolam can produce effects on the MLAER nearly comparable to those of isoflurane without measurable effects on IPSCs in layer IV and V pyramidal cells in ACx, it is likely that the targets of midazolam underlying the observed effects on the MLAER are other cell types in ACx contributing to the MLAER (e.g. , supragranular pyramidal cells, spiny stellate cells, GABAergic interneurons), and presumably these cellular targets have GABA^Areceptors that are more sensitive to midazolam. Identification of these targets awaits further studies on modulation of IPSCs by midazolam in ACx.

The authors thank Caitlin Cotter (Lab Assistant), Michael Rummel (Lab Assistant), and Abigail Schuh (Lab Assistant) for technical support and Robert A. Pearce, M.D., Ph.D. (Professor), for comments on the manuscript (all from the Department of Anesthesiology, University of Wisconsin, Madison, Wisconsin).