Disruption of Hippocampal Multisynaptic Networks by General Anesthetics

GENERAL anesthetics affect brain activity and function at different doses.^1,2  The dose required for surgical anesthesia, which suppresses incision pain and movements, is higher than that for loss of consciousness, validated by a lack of response to verbal commands in humans. Lower doses of general anesthetic disrupt cognitive function, with different types of memory showing differential sensitivity.^3,4  The mechanisms whereby general anesthetics affect behavioral and neural functions are not completely known.^1,2 

Many studies of general anesthetics have focused on in vitro brain preparations, and advances have been made on the effect of general anesthetics on ion channels and receptors.⁵  Early work of Richards and White⁶  reported that volatile anesthetic decreased excitatory synaptic transmission in the dentate gyrus (DG) of the hippocampus in vitro, at doses less than 1 minimal alveolar concentration (MAC). Volatile anesthetics less than 1 MAC were also shown to decrease non-N-methyl-D-aspartate receptor–mediated glutamatergic transmission in the hippocampal CA1 area.^7–11  Glutamatergic neurons have been shown to be susceptible to metabolic depletion by a volatile anesthetic.¹²  γ-Aminobutyric acid–mediated (GABAergic) inhibition in vitro was sensitive to low doses of volatile anesthetics.^13–15 

The relation between brain activity and behavioral functions is best studied in freely moving animals in vivo. This approach can elucidate the brain mechanisms underlying cognitive impairment and loss of consciousness after low doses of general anesthetic. The in vivo brain offers intact neural networks that are not altered by in vitro slicing or culturing. To our knowledge, other than a few exceptions,¹⁶  the effects of low-dose general anesthetic on neural activity in vivo have not been studied.

We focused on the hippocampal cortex, which is particularly well structured for network studies of neural interactions.^17,18  The hippocampus is also important for contextual fear conditioning,¹⁹  which was disrupted by isoflurane of 0.25 MAC,³  and MAC for isoflurane was ~1.4% in rats.²⁰  A multisynaptic pathway from the entorhinal cortex (EC) to the DG, CA3, and CA1 areas of the hippocampus (fig. 1A)^17,18  has been implicated as a main neural circuit for declarative memory in humans.²¹  Each of the excitatory synapse at the EC–DG–CA3–CA1 circuit is glutamatergic, but GABAergic interneurons provide inhibition at somatic, dendritic, and presynaptic sites.²²  There is controversy over whether isoflurane affects the excitatory²³  or inhibitory transmission²⁴  in the hippocampus of intact animals, and electrophysiologic responses to low doses of isoflurane (less than 0.35%) have not been studied.

We hypothesize that the multisynaptic networks in the hippocampus in vivo are sensitive to isoflurane, even at low doses (less than 0.5%). The first objective was to compare multisynaptic (EC to CA1) transmission with monosynaptic (CA3 to CA1) excitation at the basal or apical dendrites of the CA1 pyramidal cells. A second objective was to study networks by spontaneous local field potentials (LFPs), and Δ, θ, and γ frequencies in the hippocampal LFPs are known to be generated by extrinsic and intrinsic neural circuitry.^25–30  For comparison to low-dose isoflurane responses, the effects of a single subanesthetic dose of sodium pentobarbital (10 mg/kg intraperitoneal) on hippocampal synaptic transmission and spontaneous LFPs were also studied.

Adult male Long-Evans rats (Charles River, Canada) were kept in a temperature-controlled room under 12:12 h light–dark cycle (lights on/off at 7:00/19:00 h) with food and water available ad libitum. All experiments were conducted between 9:00 and 17:00 h in accordance with the guidelines established by the Canadian Council on Animal Care (Ottawa, Canada) and approved by the local Animal Use Committee at The University of Western Ontario (London, Ontario, Canada).

Male Long-Evans rats (250 to 350 g) were handled and habituated to a recording chamber (30 × 30 × 30 cm Plexiglas cage) for at least 3 days before surgery. Under sodium pentobarbital anesthesia (60 mg/kg intraperitoneal, supplemented as needed), animals were placed in a stereotaxic apparatus. Body temperature was maintained between 35 to 37°C with a rectal thermometer monitoring system. In seven rats, a 16-channel silicon probe (50-μm interelectrode distance; NeuroNexus, USA) was inserted to span all layers of the right dorsal hippocampal CA1 area (P4.6, L3.2, V3.3; atlas of Paxinos and Watson³¹ ). Teflon-coated steel wires of 125-μm diameter were used as stimulating electrodes, placed at (1) right medial perforant path (MPP; P8.0, L4.4, V3.5), (2) stratum radiatum (RAD), and (3) stratum oriens (OR) of CA1 bilaterally (P4.6, L3.0, V3.0; P3.2, L2.0, V3.0). All recordings in CA1 were made with a 16-channel probe, unless otherwise stated. In four additional rats, recordings were done with Teflon-coated stainless steel wires in the dentate gyrus (P3.8, L 2.7, V4), and two of the rats also had stimulus electrodes in OR and recording electrodes in CA1. Optimal placement of the stimulating and recording electrodes were made by electrophysiologic criteria.^32,33  A recording screw was located at the frontal cortex (A1.0, L1.0) and other ground and reference screws were placed in the skull over the cerebellum. All electrodes were fixed by creating a head cap made of dental cement. A stimulus isolation unit (PSIU6 connected to Grass S44 stimulator, Astro-Med, Inc., USA) provided constant current stimulus pulses (0.2-ms duration, less than or equal to 400 µA), delivered cathodally to a stimulus electrode, with a screw over the cerebellum serving as anode.

Recording sessions commenced at least 7 days after surgery, after further habituation in the Plexiglas recording chamber. Signals recorded by a silicon probe were fed through a headstage to a commutator, and then amplified and filtered (1 to 10 kHz) by a Medusa preamplifier and RA16 base station (Tucker-Davis Technologies, USA) before digitization (typically at 6 kHz) by a custom-made program.

A rat underwent two or more recording sessions with isoflurane anesthesia, separated by at least 5 days (fig. 1B). Isoflurane in 100% oxygen from a vaporizer was delivered by tubing through a small outlet to the bottom of the chamber, and allowed to flow out through a small hole at the top of the recording chamber for continuous monitoring by means of an infrared gas analyzer (Ohmeda RGM5250, USA). For the first session, recordings were first made during baseline before anesthetic administration, followed by isoflurane administration at 0.25% and then 1%, before returning to room air (0%). Follow-up recordings were made at 24 h after isoflurane exposure. Recordings started at less than 20 min after each isoflurane concentration change, when the gas analyzer indicated a stable isoflurane concentration. In order to keep the rat near its normal body temperature, a heating pad maintained at 36°C was placed underneath the recording chamber. For the next session 5 or more days later, after baseline recording, isoflurane was delivered at 0.5%, and then at 1.5%. For subsequent sessions, some rats were given 0.375% isoflurane, or a previous isoflurane dose was repeated. No attempts were made to randomize the doses or to blind the experimenter, since the behavioral effects of high-dose isoflurane were clearly discernible. In another session, sodium pentobarbital (10 mg/kgintraperitoneal) was injected, with recordings done during baseline (before pentobarbital) and at 15 min, 1 h, and 24 h after pentobarbital.

LFPs were recorded for minutes during different behavioral states before and after different isoflurane doses. Immobility state was defined when the rat spontaneously showed no gross head and body movements. If not administered with an anesthetic, the rat had to be awake and typically held its head up. Walking state was operationally defined to include horizontal locomotion, rearing, and large head and body movements. Rapid eye movement (REM) sleep was defined when the rat was immobile while a regular hippocampal θ rhythm (3 to 10 Hz) was detected. Hippocampal average evoked potentials (n = 5 sweeps) were recorded during immobility, excluding REM periods, after stimulation of OR and RAD,^32,33  or MPP³⁴  (fig. 1C). Different stimulus intensities were applied to each pathway, including two and four times the stimulus threshold, with threshold defined as the stimulus intensity that evoked a visually detectable response. Behaviors after anesthetic administration were noted, including locomotor activity, ataxia, or extended periods of inactivity. Single- or paired-pulse stimulation used for evoking field potentials did not induce detectable behaviors. At the end of the experiment, under pentobarbital (80 mg/kg intraperitoneal) anesthesia, the rat was perfused with saline followed by 4% formaline through the heart. The brain was retrieved and subsequently, the locations of the electrodes were verified in 40-μm thionin-stained brain sections.

Population postsynaptic potential (pEPSP) and population spike (PS) were distinguished in the average evoked potentials. PS was generated by synchronously firing single neurons.^17,32  Monosynaptic excitation after stimulation of OR (or RAD) was measured by the rising slope of the negative pEPSP at OR (or RAD). The maximal slope over 1-ms interval was evaluated during the negative going phase of the pEPSP (fig. 1B), at 0.5 ms before the onset of a PS. The latter slope measure was measured within 1 to 4 ms of onset, during which evoked inhibitory postsynaptic potential would be expected to be minimal. The multisynaptic excitation in CA1 was evaluated by the maximal negative slope of the apical dendritic pEPSP at 10 to 15 ms after MPP stimulation (fig. 1B). Paired-pulse facilitation (at 50-ms interpulse interval) was evaluated by the ratio of the pEPSP slope evoked by the second pulse (E2) to that evoked by the first pulse (E1).

For silicon probe recordings, a one-dimensional current source density (CSD) was calculated from the average evoked potentials. CSD shows the local current source and sink in a laminated structure, without the effects of volume conduction.³⁵  CSD(z, t), a function of depth z and time t, was calculated by a second-order difference formula:³⁵ 

where Φ(z, t) is the field potential at depth z and time t, and Δz is the spacing (50 µm) between adjacent electrodes on the 16-channel probe. The conductivity σ was assumed to be constant and the CSDs were reported in units of V/mm².

CSD analysis was used to give the input–output relation between PS and excitation after RAD stimulation. For a CSD trace near the CA1 pyramidal cell layer, the amplitude of the PS sink was assessed by the vertical distance of the negative sink peak to the tangent line linking the positive peaks surrounding the negative peak (fig. 1B). The strength of synaptic excitation was evaluated by the maximal slope (1 ms interval) of the apical dendritic excitatory sink. PS (y) values were plotted against excitatory sink (x) values for each experiment, before and after isoflurane. A second-order polynomial (y = ax² + bx + c, where a, b, c are fitted numbers) was drawn to fit all the points during baseline. For a given x range, a cluster of isoflurane points consistently above the baseline curve was considered to be a significant deviation from the baseline. Wilcoxon test between baseline and isoflurane groups (minimum four per group over a fixed x range) will determine the P value.

Spontaneous LFPs were sampled at 6.103 kHz, and power spectral analysis was done by custom software, as described previously.³⁶  Artifact-free activity was selected manually. Each LFP segment consisted of 8,192 points (1.34 s), and one tenth of the total segment length was tapered by a cosine function. (1 − cos(10 πt/T)) where t is time from each end and T is the segment length. After fast Fourier transform, the spectral values were smoothed by an elliptical window decreasing at [1 − (fi/fm)²]^0.5, where fi = deviation from the center frequency and fm = half the smoothing bandwidth = 2. Effectively, the frequency resolution is 0.745 Hz (reciprocal of 1.34 s), and after smoothing over five frequency bins, statistically independent values were 3.73 Hz apart. Power from 57 to 63 Hz (line power) was excluded. Each power spectral value derived from one segment is estimated to have 10 degrees of freedom (df), and the average spectrum of each condition consisted of 6 to 12 segments, that is, 60 to 120 df.

Based on previous studies, the SD of the pEPSP slope (normalized to mean =1) was less than or equal to 0.122.³³  For a mean pEPSP decrease of 0.2, and a two-tailed t-distribution (α = 0.05), a sample size n ≥ 5 predicts power greater than 75%. GB Stat (Dynamic Microsystems, Inc., USA) and SAS 9.1.4 (SAS Institute, Inc., USA) were used for statistical evaluation. The mean ± SD are shown. The change of a neural measure (pEPSP or excitatory sink slope, LFP power at a certain frequency slope, and peak θ frequency) with isoflurane dose was analyzed by one-way repeated measures ANOVA. Unless otherwise noted, there were no missing values in the ANOVA. If the overall repeated measures ANOVA was significant (P < α, α = 0.05; two-tailed), post hoc Newman–Keuls test was applied to compare values at different isoflurane doses to the baseline; significance of each post hoc test was kept at the P < 0.05 level.

Group differences in LFP power were assessed in two ways. First, for each condition (baseline or anesthesia), a mean ensemble power spectrum was constructed by averaging the logarithmic power at each digitized frequency for all rats. At each digitized frequency bin, statistical difference in power between two conditions was evaluated by a paired Wilcoxon test. To compensate for multiple comparisons, only significant power differences that extend over five or more frequency bins (estimated P < 0.001) were shown in the plots. Second, for dose response, LFP power at selected frequency bands was averaged for each rat for all experiments of a given isoflurane dose (baseline = zero dose). For each frequency band, if a one-way repeated measures ANOVA was significant (P < 0.05), post hoc Newman–Keuls test was used to reveal differences at a particular dose as compared to the baseline. A two-way randomized block repeated measures ANOVA was applied when the same two factors were studied in the same rats (apical dendritic responses to RAD/MPP stimulation and isoflurane dose; peak θ frequency in relation to behavioral state and isoflurane dose).

Two different synapses from CA3 to CA1 pyramidal cells were assessed: basal dendritic excitation after stimulation of OR and apical dendritic excitation after stimulation of RAD (fig. 1). Basal dendritic pEPSPs were negative at the basal dendrites and positive at the cell layer and apical dendritic layer. A representative example showed that the slope (E1) and peak of the negative wave (pEPSP) recorded in CA1 OR evoked by first pulse of a pair of OR stimulus pulses were strongly decreased by 0.5% isoflurane (fig. 2A1). The slope of the pEPSP (E2) evoked by the second pulse was less strongly decreased by 0.5% isoflurane (fig. 2A2). However, the population spikes evoked by the first (PS1) and second pulse (PS2) were both sensitive to 0.5% isoflurane (fig. 2A). Increase of isoflurane to 1.5% only attenuated E1 slightly more than 0.5% (fig. 2A1). CSD analysis revealed an excitatory sink (negative CSD) at the CA1 OR, accompanied by a source at CA1 cell layer and RAD (fig. 2B). Isoflurane (0.25%) decreased the amplitude of the whole sink-source pattern (fig. 2C), also found for other isoflurane doses (fig. 2B).

The group data showed similar results. One-way repeated measures ANOVA showed that the normalized first-pulse pEPSP slope (E1) evoked by 2 x threshold stimulus intensity was significantly different for six conditions (baseline, four isoflurane doses, and 24 h after isoflurane), as indicated by a main group effect (F[5,15] = 10.3, P < 0.001, n = 4 rats; fig. 3). Post hoc Newman–Keuls test showed that each of the four isoflurane doses (0.25, 0.5, 1.0 and 1.5%) was significantly different from the baseline, or from 24 h postisoflurane, but not from each other (fig. 3). Similar results were found for E1 evoked by 4 x threshold stimulus intensity (fig. 3). The low-dose (less than 0.5%) isoflurane response was also confirmed in two other rats recorded by single-wire electrodes in CA1 after OR stimulation (data not shown).

The first pulse apical dendritic pEPSP (apical E1) was slightly decreased by 0.5% isoflurane and strongly attenuated by 1.5% isoflurane (fig. 4A). CSD analysis showed that the apical dendritic pEPSP was generated by a dominant sink at RAD, with source at the cell layer, and the whole source-sink pattern was suppressed by isoflurane in a dose-dependent manner (fig. 4B). Repeated measures ANOVA of the group apical E1 data at 2 x threshold stimulus intensity confirmed an isoflurane treatment effect (F[5,25] = 11.7, P < 0.0001, n = 6), with significant (P < 0.05) post hoc differences (Newman–Keuls test) only found at greater than or equal to 1% isoflurane (fig. 3). Group data of apical E1 of 4 x threshold stimulus intensity also showed a significant isoflurane treatment effect with post hoc differences at greater than or equal to 1% isoflurane (fig. 3).

A long-latency CA1 multisynaptic response (triE1), manifested as a negative wave at 10 to 15 ms in CA1 RAD after MPP first-pulse stimulation (fig. 4C1), was significantly attenuated to ~50% of the baseline by isoflurane dose of greater than or equal to 0.25% (figs. 3 and 4C1). After paired-pulse MPP stimulation, the second pulse multisynaptic response (triE2) was less sensitive to isoflurane than triE1, while the PS after the second pulse (triPS in fig. 4C2) was strongly attenuated by 0.5% isoflurane. One-way repeated measures ANOVA of the normalized triE1 group data revealed a significant isoflurane effect (F[5,20] = 11.7, P < 0.0001, n = 5) at 2 x threshold stimulus intensity. Significant isoflurane effect was also found at 4 x threshold stimulus intensity, with less robust post hoc differences (fig. 3).

The effect of isoflurane on the mulitsynaptic response (triE1) was different from that on the apical dendritic response (apical E1) evoked by RAD stimulation. In four rats that had both triE1 and apical E1 recorded, randomized block ANOVA showed a significant group x dose effect (F[5,15] = 13.3, P < 0.0001) without a main group effect (F[1,3] = 2.55, P > 0.2), and post hoc comparisons indicated significant differences only at isoflurane doses of 0.25 to 0.5%.

The multisynaptic CA1 response resulted from activation of the DG, which then activated CA3. The PS evoked by MPP in or near the DG was not significantly changed by isoflurane dose of 0.25 to 0.5% (n = 5; data not shown), suggesting that transmission in CA3 or CA1, and not from MPP to the DG, was the cause of the depression at low-dose isoflurane. Synaptic responses in CA3 were not recorded.

Strong stimulation of CA3 was shown to induce recurrent excitation of the DG through a CA3–CA1–EC–DG pathway.³⁷  A recurrent DG excitation was observed in CA1 after stimulation of OR or RAD, typically after paired pulses (# in fig. 2A2). The late negative wave at ~20 ms latency after strong OR stimulation was abolished by greater than or equal to 0.25% isoflurane (fig. 2A2), which was confirmed in five rats.

During baseline, paired-pulse facilitation (PPF) of the pEPSPs at 50-ms interpulse interval was large for stimulus intensities that did not evoke a first-pulse PS, as shown for 2 x threshold basal (OR stimulation), apical-monosynaptic (RAD stimulation), and apical-trisynaptic (MPP stimulation) excitation (fig. 5A). For 2 x threshold RAD stimulation, a significant increase in PPF was found at greater than or equal to 1% but not at less than 1% isoflurane (fig. 5B; repeated measures ANOVA, n = 7 pathways in five rats). At 2 x threshold stimulus intensity, PPF of the pEPSPs after OR stimulation was typically higher than that after RAD stimulation, and greater than or equal to 0.25% isoflurane further increased the PPF (fig. 5, B and C). Similarly, the late, trisynaptic response in CA1 after MPP stimulation showed a high PPF that greatly increased after greater than or equal to 0.25% isoflurane (fig. 5, B and C).

PS in relation to synaptic excitation was analyzed for RAD stimulation only. At a fixed stimulus intensity, the PS amplitude usually decreased after isoflurane, but the change was not statistically significant at any isoflurane dose tested. The change in spiking with synaptic excitation was shown by plotting the PS1 sink, derived from CSD analysis, with the excitatory apical sink slope E1; both PS1 and E1 were evoked by the first pulse at RAD. For most (four of five) rats, there was no change in the PS1 versus E1 plot after isoflurane, as shown by the scatter plot of three of them before and after 0.25% isoflurane (fig. 6A). However, one of five rats did show an increase in PS1 amplitude at intermediate E1 values after 0.25% isoflurane (arrow at fig. 6B1). A general lack of PS1 versus E1 was also found at 0.5 to 1.5% isoflurane doses.

Peak latency of the PS evoked by a constant stimulus intensity showed a consistent increase after greater than or equal to 0.25% isoflurane dose (data not shown). There was no significant change of PS1 latency in relation to E1 at any dose, as shown by a representative example (fig. 6B2).

Isoflurane at 0.25% suppressed power of the LFPs recorded in CA1 RAD during immobility as compared to baseline immobility (fig. 7A), as shown by group-average power spectra (fig. 7B). Selected frequency ranges suppressed by 0.25% isoflurane were 4 to 30 Hz, 50 to 70 Hz, and 100 to 150 Hz (fig. 7, B and C). Isoflurane dose at greater than or equal to 0.5% suppressed power within most of the 0.8 to 300 Hz frequency band, except 30 to 50 Hz power was not suppressed at 0.5% (fig. 7, B and C). LFPs recorded in OR showed a similar dose dependence on isoflurane, except the power decrease was slightly smaller (fig. 7C).

During baseline, a hippocampal θ rhythm was found to accompany walking and REM, with a similar average peak frequency of ~6.8 Hz. The hippocampal θ rhythm found during immobility under isoflurane was considered to be associated with REM. During low isoflurane doses (0.25 to 0.5%), a hippocampal θ rhythm was still found during walking and immobility, but its peak frequency was reduced as compared to baseline (table 1). Two-way (behavioral state x isoflurane dose up to 0.5%) randomized block ANOVA showed that the peak hippocampal θ frequency did not differ between walking and REM states (F[1,4] = 1.11, P > 0.3, n = 5 rats), but progressively decreased with isoflurane dose (F[2,8] = 35.1, P < 0.001; table 1). At greater than or equal to 1% isoflurane, the rat was not capable of spontaneous movements, and the hippocampal LFPs showed irregular slow activity with occasional periods of 3 to 4 Hz θ rhythm.

Evoked pEPSPs and LFPs were recorded in seven rats during no-drug baseline and after a single subanesthetic (10 mg/kg intraperitoneal) dose of pentobarbital. After pentobarbital, the pEPSP after OR stimulation (not shown) and RAD stimulation (fig. 8A) were not significantly changed from baseline, but the trisynaptic CA1 pEPSP at 4 x threshold intensity was significantly reduced (fig. 8B). Group average spectra of LFP power at RAD shows that, as compared to baseline, the power at Δ (1 to 4 Hz) and high γ frequency (78 to 125 Hz) were significantly decreased after pentobarbital injection (fig. 8C). After pentobarbital, apical pEPSPs showed paired-pulse depression (decreased E2/E1 ratio as shown in fig. 8D), as compared to increased PPF after isoflurane.

We reported that hippocampal synaptic transmission in freely moving rats was dose-dependently suppressed by isoflurane. The excitatory synapse on the basal dendrites of CA1 pyramidal cells, as compared to that on the apical dendrites, was suppressed by the lowest dose of isoflurane used (0.25%). The multisynaptic pathway from the EC to CA1 was also suppressed by 0.25% isoflurane. Spontaneous LFP power across a wide band of frequency (0.8 to 300 Hz) also showed a dose-dependent suppression by isoflurane.

The monosynaptic apical dendritic pEPSP (evoked by RAD stimulation) decreased in a graded manner with isoflurane dose and showed a statistically significant decline at greater than or equal to 1% isoflurane (fig. 3B). This isoflurane dose response was similar to the pEPSP¹⁰  or the intracellularly recorded EPSPs in CA1 in vitro.^7,8  The CA1 basal dendritic pEPSP (evoked by OR stimulation) in vivo demonstrated an exquisite sensitivity to isoflurane, being suppressed by 0.25% isoflurane (fig. 3A). Sensitivity of the OR-evoked pEPSP to 0.5% isoflurane was reported in vitro.⁹ 

The multisynaptic MPP–DG–CA3–CA1 response (triE) and an OR-evoked recurrent excitation of CA1 (fig. 2A) were suppressed by 0.25 to 0.5% isoflurane. The latter isoflurane dose did not significantly alter the MPP-evoked PS in the DG or the RAD-evoked apical dendritic excitation in CA1, but transmission in CA3 had not been studied. Isoflurane may suppress multisynaptic transmission readily because small decrease in excitatory transmission and spike excitability at one synapse may accumulate across multiple synapses. Low doses of isoflurane were shown to enhance GABAergic inhibition^13–15  and suppress neuronal firing of both pyramidal and inhibitory neurons.³⁸  However, γ-aminobutyric acid type A (GABA_A) receptor antagonist bicuculline, which blocked tonic and evoked GABA_A receptor-mediated inhibition, had no consistent effect on the rising phase of monosynaptic EPSPs in CA1 neurons in vitro.³⁹  An effect of tonic¹⁵  or late¹³  GABAergic inhibition on triE in vivo remains to be shown. Still, it may be fair to assume isoflurane’s suppression of triE results from cumulative changes in excitation and inhibition along the multisynaptic pathway.

PPF was found for basal or apical dendritic pEPSPs evoked at low-stimulus intensities. An isoflurane dose that decreased the first-pulse pEPSP also significantly increased PPF. Thus, the isoflurane sensitivity of the PPF of each pathway paralleled that of the first-pulse pEPSP. PPF was more robust for pEPSPs after OR than RAD stimulation (with the same x threshold intensity), and large for the triE after MPP stimulation (fig. 5). PPF at a single synapse is associated with release probability, and a high release probability (strong synapse) is associated with small PPF.⁴⁰  The change in pEPSPs after isoflurane can be explained by a decrease in neurotransmitter release,¹⁰  possibly selective at glutamatergic terminals.¹²  Isoflurane (~0.5 MAC) was shown to decrease exocytosis of synaptic vesicles at glutamatergic terminals in vitro.⁴¹ 

We did not find a consistent isoflurane-induced change in the relation between single-pulse PS amplitude and excitatory strength E1, measured by slope of the pEPSP or excitatory sink in vivo (fig. 6). Spiking in CA1 neurons after intracellular current injection was not sensitive to 0.5 to 3% isoflurane,⁸  suggesting that spike excitability had low sensitivity to isoflurane. The lack of a consistent increase in single-pulse PS1 versus E1 after 0.25% isoflurane (fig. 6A) appears to be different from the conclusion of increased spike excitability induced by paired-pulse stimulation after 0.5% isoflurane.⁹ 

Our results of a significant reduction of glutamatergic excitation in vivo differ from a previous study,²³  which concluded that isoflurane had no clear effects on hippocampal pEPSP and PS after commissural stimulation. However, the latter study²³  did not include a preanesthetic baseline, and effects after isoflurane were compared with those after urethane anesthesia. The current study also differs from another in vivo study that reported 0.38 to 1.4% volume isoflurane caused a leftward shift of the Schaffer-collaterals evoked PS versus pEPSP curve,²⁴  with both PS and pEPSP recorded from the CA1 pyramidal cell layer. Since the excitatory synapses are on the dendrites, optimal recording of pEPSP, or excitatory sink, would be at the apical dendritic layer after RAD stimulation, or the basal dendritic layer after OR stimulation. In our multichannel recordings, we observed that 1.5% isoflurane decreased the positive pEPSP near the soma 25% more than the negative pEPSP at the apical dendrites, possibly because of overlap with spike prepotential and inhibitory conductance near the cell body.

We showed that the spontaneous hippocampal LFP power was decreased by isoflurane dose of greater than or equal to 0.25%. The most sensitive frequency bands appear to be 4 to 15 Hz and 50 to 150 Hz (fig. 7), consistent with our previous studies.^42–44  Similarly, Hudetz et al.⁴⁵  showed that 70 to 140 Hz power of spontaneous LFPs in the hippocampus, visual and frontal cortices were decreased by greater than or equal to 0.4% isoflurane. The current study decreased the sensitivity of hippocampal LFPs to 0.25% isoflurane, and similar sensitivity was found for LFPs recorded in OR and RAD, except the drop in LFP power with isoflurane dose was somewhat smaller in OR than RAD (fig. 7C).

We found that the peak frequency of the walking hippocampal θ rhythm decreased with isoflurane dose (0.25 to 0.5%), as was reported previously.¹⁶  In addition, under low-dose isoflurane, a θ rhythm that occurred during immobility, tentatively identified as REM, showed a similar peak frequency as that during walking. Thus, isoflurane’s effect in slowing θ frequency was independent of behavioral states. This effect was likely exerted outside of the hippocampus, possibly on the medial septum and supramammillary area.^25,26 

Sensitivity of γ waves to low isoflurane doses is consistent with anesthestic sensitivity of a multisynaptic network. γ activity in the hippocampal and neocortical cortex in vivo is generated by perisomatic parvalbumin-positive interneurons interacting with principal cells (fig. 9A),^46–48  although purely GABAergic networks of interneurons could generate γ oscillations as well.⁴⁹  LFPs recorded in CA1, including γ activity, may be intrinsically generated, or driven by similar generating networks in CA3 and EC.^27,28  The power of spontaneous hippocampal LFPs (0 to 100 Hz) was strongly attenuated by non-N-methyl-D-aspartate receptor antagonists,⁵⁰  supporting the concept that hippocampal LFPs are driven by non-N-methyl-D-aspartate glutamatergic inputs. Using 70 to 100 Hz (RAD-recorded) LFP power to represent γ, the decrease of γ power with 0.5 to 1.5% isoflurane dose was found to be similar to the decrease in power (square amplitude) of the trisynaptic apical pEPSPs (fig. 9B). This suggests that the main cause of LFP γ power decrease with high isoflurane dose is a reduction in glutamatergic synaptic transmission.

Isoflurane of 0.28% induced deficits in a context-dependent fear conditioning in rats,³  a task that depends on the hippocampus.^19,20  The results here showed that 0.25% isoflurane disrupted hippocampal θ and γ activity. More direct evidence of interference of hippocampal function is shown by a strong impairment of the multisynaptic transmission from the EC to CA1, which has been suggested as a main neural circuit responsible for hippocampal memory.²¹ 

Sodium pentobarbital, at a subanesthetic dose (10 mg/kg intraperitoneal) that did not induce loss of righting reflex, disrupted multisynaptic transmission without significantly changing monosynaptic apical or basal dendritic responses. Pentobarbital is known to augment GABA_A receptor-mediated inhibition,^14,51  and triE transmission may decline because of difficulty to generate action potentials at multiple areas along the trisynaptic pathway. Pentobarbital’s effect in decreasing PPF of the pEPSPs (fig. 8D) is consistent with an enhanced inhibition that suppressed the second pulse pEPSP.⁵²  In the presence of a GABA_A receptor antagonist, pentobarbital increased PPF of the pEPSPs.¹⁴  Spontaneous LFPs in the high γ frequency (78 to 125 Hz) range were also suppressed by low-dose pentobarbital.

The current study on field potentials emphasized glutamatergic transmission in vivo and did not fully reveal inhibitory or other ionic events that may be affected by isoflurane. Results in vitro showed that amnestic doses of isoflurane (~0.25 MAC) significantly increased tonic GABAergic currents¹⁵  or prolonged evoked inhibitory postsynaptic currents in hippocampal pyramidal cells.¹³  The current data do not fully reveal the inhibitory changes induced by isoflurane. In addition, population events such as PS do not reveal excitability of the different types of cells, and isoflurane’s action on specific types of interneurons or on CA3 cells is not known. Cellular mechanisms of isoflurane’s action, such as a decrease in glutamatergic release, could only be inferred in the current study, and await direct verification with other techniques.

Isoflurane dose response may differ in different areas of the brain,^1,53  and the current data on the hippocampus do not explain the loss of righting or pain responses. We propose a working hypothesis that isoflurane’s reduction of glutamatergic transmission in the hippocampus accounts for some of the amnestic effects of low-dose isoflurane. While the hypothesis is intuitive, it is tentative in view of the multiple effects of isoflurane on the brain. Furthermore, memory function was not studied in the animals recorded in the current study. In order to validate the working hypothesis, future experiments will need to directly reverse isoflurane’s glutamatergic effects selectively in the hippocampus.

This research was supported by the Canadian Institutes of Health Research (grant No. MOP-15685), Ottawa, Ontario, Canada, and the Natural Sciences and Engineering Research Council (grant No. 1037–2008), Ottawa, Ontario, Canada.

The authors declare no competing interests.