Amnesic Concentrations of the Nonimmobilizer 1,2-Dichlorohexafluorocyclobutane (F6, 2N) and Isoflurane Alter Hippocampal θ Oscillations In Vivo 

All experiments were conducted according to the guidelines laid out in the Guide for the Care and Use of Laboratory Animals  (Institute of Laboratory Animal Resources, National Research Council) and were approved by the University of Wisconsin Animal Care and Use Committee, Madison Wisconsin.

Young adult male Sprague-Dawley rats were housed in the animal care facility of the University of Wisconsin with free access to rat chow and water. Thirteen rats underwent electrode implantation. After the implantation of the electrode, the animals recovered for at least 7 days before being exposed to volatile agents. After each exposure, the animals were observed until they fully recovered and were then transferred back into their home cage. Each animal was exposed to an agent only once per day.

Linear microwire array recording electrodes with four sites spaced 200 μm apart were fabricated from 30-μm Formvar-insulated Nichrome wire14(CFW-188-0012-HFV; California Fine Wire, Grover Beach, CA). Briefly, a piece of polytetrafluoroethylene tubing of 300 μm ID was cut along its long axis, and one half of the tubing was used as a form for the electrode array. Four holes were punched in the tubing at 200-μm intervals, and a single piece of 35-μm Formvar-insulated Nichrome wire was passed through each hole. Each wire was bent at a 90° angle to lie parallel to the inside wall of the form. The hemicylinder of tubing with the microwire bundle was filled with epoxy resin (Epoxylite 6001-M; Epoxylite Corp., St. Louis, MO) and heat cured. After discarding the polytetrafluoroethylene tubing, the hardened resin hemicylinder formed the shank of the electrode, and the wires were then broken off at right angles to the long axis of the array to form the recording sites. Finally, the resin matrix was sharpened at the tip to facilitate penetration of the brain, with the closest electrode approximately 25 μm away from the tip. The bare wires of the electrode array were connected to a Neuralynx EIB-16 interface board (Neuralynx, Tucson, AZ) with an Omnetics nanoconnector suitable for connecting to a small headstage (Neuralynx HS-16).

For the implantation, animals were anesthetized with 1.5–2% isoflurane and placed in a stereotactic apparatus. The skull was depilated, the bone was exposed by an incision, and holes were drilled in the right and left frontal, right parietal, and left occipital bones. Miniature stainless steel machine screws were inserted and advanced until contact was made with the dura. They served as anchors for the electrode assembly and as surface electrodes, with the screw in the occipital bone used as animal ground for reference. Finally, an additional hole was drilled 3.0 mm caudal to the bregma and 2.0 mm lateral to the midline. After incision of the dura, the electrode array was lowered until it contacted the cortex and then advanced 2.6 mm. Postmortem histologic examination showed that electrodes were successfully implanted into the CA1 region of the dorsal hippocampus in 10 of the rats, and only data from these animals were used for field-potential analysis (fig. 1A). After at least 1 week of recovery, electroencephalographic signals from the CA1 electrodes and from surface electrodes were recorded under control conditions and in the presence of isoflurane or F6 (fig. 1B).

After an animal had undergone all planned experiments, it was anesthetized deeply with isoflurane and perfused intracardially with 0.1 m phosphate-buffered saline followed by 4% paraformaldehyde. The brain was removed and placed in 4% paraformaldehyde. After fixation, 100-μm-thick slices were prepared, stained with cresyl violet, and mounted on coverslips to verify the location of the electrode by the tissue disruption it had caused (fig. 1A). The most ventral extent of the tissue damage was taken as the location of the tip of the linear array, and the electrode locations were estimated using the spacing parameters from the fabrication process.

On the day of an experiment, the animal was placed into a custom-made 10-l polymethyl methacrylate (acrylic glass) chamber equipped with ports for drug injection, gas sampling, and connection to the recording equipment. Adhesive tape sealed the lid of the chamber and all ports. Five minutes of acclimation were followed by a 15-min control period, after which a loading dose of the drug to be tested was injected. Distilled water was injected for control experiments. A fan accelerated the evaporation and distribution of volatile agents that were deposited into a glass Petri dish via  a polytetrafluoroethylene tube. Soda lime scavenged carbon dioxide. We took gas samples from the chamber with airtight glass syringes for gas chromatography at 5, 15, 22.5, and 30 min after the initial drug injection. The last three measurements were averaged and considered to represent the tested concentration. The concentration of the drug was kept approximately constant by injecting additional small boluses 10 and 20 min after the initial injection. In preliminary experiments, we determined that, with this protocol, it was possible to maintain stable concentrations (within 20% of the reported concentrations) of the agent throughout the 30 min of drug exposure. Drug washout was achieved by suctioning the chamber and allowing fresh air to enter the chamber. The oxygen concentration in the chamber was constantly monitored with a gas analyzer (POET II; Criticare Systems Inc, Waukesha WI) and maintained above 20%. In most experiments, a dedicated but electroencephalogram-blinded observer scored the animal's behavior. In addition, we videotaped the animal for post hoc  analysis. The 15 min before drug injection (0–15 min), before drug removal (30–45 min), and before the end of the experiment (60–75 min after beginning of the recording) were defined as the control, test, and recovery time periods for analysis.

The observer classified the behavior as immobile, exploring, grooming, or undefined. Exploring was considered any behavior that involved movement of the animal's head or body that was not grooming related, and included walking, sniffing, or manipulating any of the objects present in the tray. Grooming included scratching, face and paw washing. Immobile did not include assumption of the sleep posture (curling up), which was infrequent and which we classified as undefined. In practice, only exploring and immobile behaviors, corresponding to type I and type II behaviors, respectively,15were observed with sufficient duration and frequency for subsequent data analysis.

Electroencephalographic signals were amplified by a unity-gain HS-16 headstage preamplifier to reduce movement-related artifacts, a Lynx-8 second-stage amplifier (both from Neuralynx, Tucson AZ), band-passed between 1 and 325 Hz, and digitized at 1,000 Hz using a DigiData 1200 A/D converter (Axon Instruments, Foster City, CA). Data acquisition and processing were controlled with the pClamp software suite (Axon Instruments) and stored for analysis on a Pentium-based (Intel Corporation, Santa Clara, CA) personal computer.

Only the signal from the electrode closest to the hippocampal fissure, as revealed by postmortem histologic examination, was used for analysis. Data analysis was performed primarily with custom-written routines in Matlab (MathWorks, Natick, MA). Origin (MicroCal, Northampton, MA) and Instat (GraphPad Software, San Diego, CA) were used for graphical presentation and statistical analysis. In a preprocessing step, the raw data were passed through a digital band-pass filter (attenuation 40 dB/octave) designed for the extraction of signals in the θ frequency band. The −3 dB (corner) frequencies were 4 and 12 Hz. Artifacts occurred primarily from mechanical causes and were excluded from analysis together with adjacent data points (± 0.5–1 s, as necessary). Subsequently, data were sorted by behavior and the raw and filtered data were subdivided into segments of 4,096 points each (approximately 4 s) with 30% overlap (1,365 points). Shorter data segments were not analyzed. All parameters described below were computed for each segment and then averaged with other segments obtained for the same drug condition and behavior for a given animal.

For spectral analysis, we computed the fast Fourier transform using a Hamming window to obtain the power spectral density. From the power spectral density, we extracted the power in distinct frequency bands (the integral of the power spectral density within the frequency ranges detailed above) as well as amplitude and frequency of the θ peak (fig. 1C).

Unless indicated otherwise, all numerical results are expressed as changes relative to time- and behavior-matched control experiments. For individual data points, linear regression lines were fitted according to the model Yⁱ= A + BXⁱ, with the parameters A (intercept) and B (slope) estimated by the method of least squares minimization. The correlation between drug concentration and change in the measured parameter was considered significant if the null hypothesis of zero slope could be rejected at P < 0.05.

After being placed into the experimental chamber and connected to the recording equipment, each rat spontaneously explored its environment, groomed, or remained immobile. The fraction of time each individual spent on these behaviors differed, but a certain pattern prevailed: Over the first 45 min in the drug-free experiments (water injections and sham sampling, see time course of experiments in fig. 2), as the rats became familiar with the chamber, they typically spent an increasing fraction of time immobile (fig. 2A). This pattern of progressive immobility/diminishing exploration was interrupted by opening the chamber to rapidly evacuate volatile agents 45 min after the start of the experiment. This intervention always led to a burst of exploratory activity that quickly abated as the rats returned to their baseline activity level, and this remained approximately constant for the remaining 30 min of the experiment.

F6 and isoflurane both affected the behavioral pattern, but in opposite directions. Increasing concentrations of F6 progressively reduced the amount of time the rats spent immobile (figs. 2B and C), so that at 3.5% (0.85 MAC^pred), they explored the cage almost constantly. After evacuating the drug, the animals gradually reverted to a behavioral pattern that resembled the drug-free experiments. Isoflurane had the opposite effect. With increasing isoflurane concentrations, the rats spent progressively more time immobile. The effect of 0.35% isoflurane is illustrated in figure 2D. After suctioning, the rats gradually reverted to an activity pattern similar to that under drug-free conditions. We conclude that F6 has no sedative properties at concentrations that were shown previously to produce amnesia.7,8 

In the rodent hippocampus, θ oscillations are prominent during two behavioral states: rapid eye movement sleep,11and waking behaviors associated with locomotor activity that have been described by the terms preparatory , voluntary , orienting , or exploring .16To validate our behavioral scoring with respect to its ability to separate low-θ from high-θ states under control conditions and during drug exposure, we compared the characteristics of the power spectrum of the electroencephalogram in the θ band during behaviors we characterized as exploring, versus  immobile. The averaged results from three animals tested drug free (i.e. , water vapor), in F6 (2.2 ± 0.08%), and in isoflurane (0.35 ± 0.04%) are presented in figure 3. Under drug-free conditions (i.e. , during water vapor exposure), the animals spent an average of 3 ± 1, 17 ± 16, and 78 ± 16% of the time grooming, exploring, and immobile, respectively (fig. 3A, pie chart). Comparison of the aggregate electroencephalographic activity in the θ band confirmed the expected difference between immobile and exploring states (fig. 3A): θ oscillations were more prominent during exploratory behavior. Note, however, that a clear peak in the spectral density within the θ frequency range was also present during the immobile state, albeit with a slower peak frequency (7.7 ± 0.6 Hz exploring vs.  6.6 ± 0.8 Hz immobile; P < 0.05). In the presence of 2.2% F6, the difference in θ power between exploring and immobile states was maintained, again with θ-peak frequency slower during immobile than during exploring (6.8 ± 0.1 vs.  7.7 ± 0.1 Hz; P = 0.02; fig. 3B). In the presence of 0.35% isoflurane, the animals spent most of the time immobile (90 ± 6% vs.  6 ± 4%). Still, the average power of θ oscillations during the brief periods of exploration was higher than during immobility (fig. 3C). There was, however, no difference in the peak frequency (5.9 ± 0.3 vs.  5.8 ± 0.3 Hz). We therefore conclude that our behavioral assessment was able to separate high-θ from low-θ states under control conditions and during the application of volatile agents. The observed difference in θ power between behavioral states supports the separation of drug effects on θ oscillations by making behavioral state-specific comparisons.

Hippocampal θ oscillations facilitate mnemonic processes, and interventions that impair the generation or the expression of θ oscillations in the hippocampus impair memory (for review, see Vertes et al.  17,18). To test the hypothesis that inhalational agents with different behavioral profiles (anesthetic vs.  nonimmobilizer) modulate θ rhythms at concentrations that are amnesic, we measured the effects of isoflurane and F6 on θ oscillations in the hippocampus in vivo . Because the behavioral state affects the characteristics of this rhythm, we analyzed the effects separately for the exploring and the immobile states.

The effects of both drugs on power spectra acquired from one animal during exploration are illustrated in figure 4A. This animal was tested with two concentrations of isoflurane (0.35% and 0.77%) and one concentration of F6 (2.2%). The spectrograms show that both drugs altered θ oscillations, but also illustrate differences between the two agents in their effect on all three analyzed parameters: isoflurane (top panel) progressively reduced oscillation frequency and markedly increased the amplitude of the θ peak (from 7.57 Hz to 4.88 Hz and from 0.0434 to 0.0963 mV2/Hz, respectively at 0.77%) while slightly reducing the total power in the θ band (from 0.079 mV2to 0.071 mV2). By contrast, 2.2% F6 (shown inverted in the lower panel), i.e. , the MAC^predequivalent of 0.77% isoflurane, reduced the amplitude of the θ peak and, rather than slowing the oscillation, slightly accelerated it (from 0.0419 mV2/Hz to 0.0249 mV2/Hz and from 7.57 to 7.81 Hz, respectively). F6 also markedly reduced the total power between 4 and 12 Hz (from 0.081 mV2to 0.051 mV2).

The results from all experiments are summarized in figure 4B. Theta-peak amplitude was decreased by F6 (slope =−0.52, P < 0.0001), whereas it was increased by isoflurane (slope = 1.47, P = 0.0005; fig. 4B, top). Power in the θ band was reduced by F6 but not significantly affected by isoflurane (slope =−0.38, P < 0.0001 and slope = 0.43, P = 0.18, respectively; fig. 4B, middle). Isoflurane slowed θ-peak frequency in a dose-dependent manner (slope =−0.61, P < 0.0001). By contrast, F6 did not alter θ-peak frequency (slope = 0.06, P = 0.09; fig. 4B, bottom).

We also analyzed the effects of isoflurane and F6 on the electroencephalogram during immobility. Theta oscillations occur during immobility in the awake rat, under certain behavioral conditions.19Immobility represented a major fraction of the experimental time under most conditions and is likely to include multiple behavioral states that we did not further separate. The results are summarized in figure 5. In the immobile state, both drugs affected θ oscillations in several ways qualitatively similar to effects during the exploring state, with slowing of the peak frequency by isoflurane and suppression of power by F6 (fig. 5A). The results from all experiments are presented in figure 5B. F6 suppressed power in the θ band (slope =−0.22, P = 0.02; fig. 5B, middle), and isoflurane slowed the θ peak (slope =−0.36, P < 0.0001). The other effects were not significant.

We investigated the effect of a nonimmobilizer and an anesthetic on hippocampal network activity in freely behaving rats. We found that the nonimmobilizer F6, at concentrations that inhibit hippocampal-dependent learning,8suppressed the power of oscillations in the θ band (4–12 Hz) without sedating the animals. Isoflurane also affected θ rhythm, but in a contrary way: It slowed the oscillations and increased power in the θ band. At amnesic concentrations,20isoflurane caused pronounced sedation.

A striking behavioral observation is the increased locomotor activity exhibited by the animals at concentrations of F6 above 0.5 MAC^pred. This finding is in stark contrast to the reduced locomotion observed with isoflurane at or above 0.25 MAC (fig. 2). Motor activity is a measure of sedation and, in the case of benzodiazepines, sedation is mediated selectively by enhancement of α¹subunit–containing γ-aminobutyric acid type A (GABA^A) receptors.21In expressed receptor studies, isoflurane was also shown to enhance currents in response to low and high concentrations of GABA mediated by α¹subunit–containing receptors.22,23F6, by contrast, did not enhance responses to high GABA concentrations.24Similarly, in studies of synaptic receptors in situ , isoflurane was shown to enhance charge transfer by GABA^Aergic synaptic currents, whereas F6 did not.25A differential effect of the sedating anesthetic isoflurane versus  the nonsedating nonimmobilizer F6 at synaptic GABA^Areceptors is thus consistent with (but does not prove) a role for these receptors in mediating the sedative effects of isoflurane and other potent inhalational agents. We suggest that F6 could be useful in probing the role of other receptors, e.g. , extrasynaptic GABA^Areceptors, which have been shown to be sensitive to very low concentrations of anesthetics26,27in mediating volatile anesthetic–induced sedation.

The amnesic effects of anesthetics in animal models have been most thoroughly characterized using fear conditioning as an experimental paradigm for learning and memory. The experimental conditions can be modified to preferentially involve different brain structures: fear conditioning to tone and fear conditioning to context. The former depends on the amygdala but not the hippocampus, whereas the latter requires processing by the amygdala and the hippocampus. The hippocampus-dependent learning paradigm (fear conditioning to context) is significantly more sensitive to impairment by isoflurane, with an EC⁵⁰of 0.19%, than fear conditioning to tone, which requires approximately 0.38% isoflurane to achieve a similar degree of suppression (0.13 and 0.26 MAC, respectively).20Intriguingly, in MAC equivalents, F6 has a similar differential potency to prevent learning and memory as isoflurane in these two paradigms, suggesting that it does so by similar mechanisms, on the molecular, the cellular, or the network level. However, recent studies demonstrating that F6 and isoflurane display behavioral antagonism in their amnesic effects over low concentration ranges caused Eger et al.  28to question this conclusion. Our finding that the two drugs alter θ oscillations differently, i.e. , isoflurane increasing and F6 reducing peak amplitude (fig. 4), may help to account for this antagonism.

In humans as well as rodents, the hippocampus plays an important role in spatial and episodic memory.29The most conspicuous electrical activity in the hippocampus is the θ rhythm (θ oscillations), which is among the largest extracellular synchronous signals that can be recorded in the mammalian brain. In the behaving rat, the oscillations are nearly sinusoidal, and their frequency ranges from 5 to 12 Hz.9The θ rhythm is generally thought to consist of two components that can be separated using pharmacologic15and, more recently, genetic30tools. It has been implicated in multiple tasks, including arousal and sensorimotor integration.31In addition, the θ rhythm seems also to serve a critical role in mnemonic functions of the hippocampus.17 

We found that F6 suppressed the peak amplitude of the θ oscillation and the total power in the θ band (fig. 4B, top and middle traces). Surprisingly, the frequency profile was not changed: θ-peak frequency remained primarily dependent on the behavioral state (exploring faster than immobile; fig. 3B) with only minimal acceleration by F6 (fig. 4B, bottom trace). Isoflurane, by contrast, did not reduce power in the θ band but rather slowed the frequency of the θ peak in both behavioral states. It seems, therefore, that F6 and isoflurane have almost opposite effects on the hippocampal θ rhythm. Therefore, the question arises whether these are in any way linked to the memory impairment that both drugs exhibit, and what the mechanism for this differential modulation might be.

Previous work using selective perfusion of various brain structures with local anesthetics involved in the generation and modulation of θ oscillations indicated that frequency and power of evoked θ were encoded in different brain structures and could be modulated independently. The effect of procaine infusion into the medial septum (MS) or the medial forebrain bundle was similar to the effect of systemically delivered F6 in our experiments, in that it reduced the amplitude without changing the oscillation frequency.32Although many of these experiments were conducted in urethane-anesthetized animals, where only the atropine-sensitive type of θ rhythm is present, similar results have been obtained from awake animals.33,34Because θ suppression induced by tetracaine injection into or electrolytic lesions of the MS was associated with learning impairment,35,36it is plausible that depression of θ oscillations by F6 underlies its impairment of learning and memory.

What might the molecular and cellular targets of F6 be that underlie its selective effects on amplitude but not frequency of the θ oscillation? Because in our experiments F6 was present throughout the brain, its singular effect on θ oscillations could have been mediated in any of the brain structures that shape θ rhythms. However, sites rostral to the medial supramammillary nucleus seem to be likely candidates because injections of procaine into the medial supramammillary nucleus and caudal to it were shown previously to affect either only θ frequency or both frequency and amplitude.32,34 

Procaine and other local anesthetics are sodium channel blockers that indiscriminately silence all neurons in the vicinity of their injection site, so selectivity of their effect is determined by the extent the anesthetic spreads in the tissue around the injection site. With respect to the involvement of specific transmitters in these actions, available experimental data suggest that suppression of θ power without changes in frequency, accompanied by impairment of spatial memory, can be achieved via  activation of GABA^Areceptors in the MS.37,38Because fast GABA^Aergic synaptic currents remain typically unchanged by F6 but are enhanced by isoflurane,25,39it is unlikely that effects of F6 on receptors at these types of synapses account for our observations. However, for some subunit combinations, responses generated by low GABA concentrations are enhanced by F6.40Therefore, effects on tonic GABA currents or on slow synaptic currents produced by subsaturating concentrations of neurotransmitter at perisynaptic sites41may contribute to suppression of θ oscillations by F6.

Activation of opioid receptors in the MS can also lead to θ suppression accompanied by memory impairment.42Although the effects of F6 on opioid receptors have not been studied directly, it is unlikely that F6 is an effective agonist at these receptors as it has no analgesic properties in vivo .43 

Hippocampal θ oscillations are also under strong cholinergic modulation. Selective elimination of septal cholinergic neurons with 193-immunoglobulin G-saporin leads to almost complete suppression of θ in the awake animal, without a change in the peak frequency of the residual rhythm.44Similarly, selective pharmacologic blockade of muscarinic m1-type receptors reduced θ power (as opposed to frequency) in the behaving cat.45Remarkably, F6 but not isoflurane inhibited expressed muscarinic m1 receptors (by 30% at 1 MAC^pred).46,47Another possibility, therefore, is that F6 reduces θ oscillations by inhibition of muscarinic receptors in the MS, and that this underlies its amnesic effects.

If reduction of power in the θ band is the signature of the effect of F6 on synchronized activity in the hippocampus, slowing of θ rhythm is the equivalent for isoflurane. Although isoflurane can affect multiple receptor types, at amnesic concentrations modulation of GABA^Areceptors is most prominent. Evidence from experiments using benzodiazepines, which are selective GABA^Areceptor modulators, supports an association between slowing of θ and amnesia. The benzodiazepine chlordiazepoxide, administered systemically at a dose sufficient to slow θ rhythm by 1 Hz, was accompanied by impaired spatial learning in rats.48Slowing of θ to a similar degree by nonpharmacologic means (systemic cooling) led to a comparable degree of learning impairment.48The conclusion drawn by the authors from these experiments was that θ frequency per se  may be important for some forms of learning. Our observation that 0.25 MAC isoflurane, a concentration that causes significant impairment of hippocampus-dependent learning,20slowed θ-peak frequency from 7.5 ± 0.3 Hz to 6.0 ± 0.3 Hz (P = 0.001, n = 4) is thus consistent with the hypothesis that isoflurane-induced amnesia results from altered θ oscillations.

In summary, we have shown that F6 and isoflurane both affect synchronized neuronal activity in the rodent hippocampus but in qualitatively different ways. F6 caused reduced amplitude but not frequency of θ rhythms at amnesic concentrations without sedation. Isoflurane slowed θ and caused marked sedation. Either effect on θ rhythm is compatible with impairment of learning and memory.

The authors thank Michael Laster, D.V.M. (Assistant Researcher), and Edmond Eger II, M.D. (Professor, Department of Anesthesia, University of California, San Francisco, California), for providing gas phase calibration standards for isoflurane and 1,2-dichlorohexafluorocyclobutane.