Although propofol is known to produce amnesia when used for anesthesia, mechanisms underlying its effects on memory are poorly understood. The current study was designed to examine the effects of propofol on forms of synaptic plasticity thought to contribute to memory processing.
Extracellular excitatory postsynaptic potentials were recorded from the CA1 region of rat hippocampal slices. Long-term potentiation (LTP) was induced using theta-burst stimulation (10 bursts of 4 pulses at 100 Hz, applied at 5 Hz) of the Schaffer-collateral pathway, while low-frequency stimulation (1 Hz x 900 pulses) was delivered to induce long-term depression. The authors also used higher-frequency stimulation (10 bursts of 4 pulses at 200 Hz, applied at 5 Hz) in the presence of MK-801 to examine the effects of propofol on an N-methyl-D-aspartate receptor-independent form of LTP.
At 30 microM, propofol inhibited LTP induction produced by theta-burst stimulation but had less effect on LTP maintenance. Similarly, when LTP was induced by 200-Hz stimulation in the presence of MK-801, propofol also blocked LTP induction. Propofol did not block LTP induction in the presence of picrotoxin, a specific antagonist of gamma-aminobutyric acid type A receptors, suggesting that modulation of gamma-aminobutyric acid type A receptors participates in propofol-mediated LTP inhibition. Propofol did not inhibit long-term depression.
Propofol inhibits LTP induction through modulation of gamma-aminobutyric acid type A receptors but not via inhibition of N-methyl-D-aspartate receptors. However, other factors also possibly contribute to propofol-mediated LTP inhibition.
AMNESIA, analgesia, and hypnosis are important properties of agents used for general anesthesia. The goal of anesthetic amnesia is to prevent painful memories from developing during medical or surgical procedures. Under certain circumstances, however, patients may retain memories of events occurring during invasive procedures, a phenomenon known as awareness during anesthesia . Patients who experience awareness during anesthesia may have recurring nightmares of events experienced during anesthesia, resulting in insomnia, anxiety, and even suicidal thoughts.1Therefore, amnesia for the procedure is a key feature of effective anesthesia.
Propofol (2,6-diisopropylphenol) has proven to be a highly effective intravenous anesthetic with rapid onset and short recovery time after injection. Because of these advantages, propofol is now widely used both for general anesthesia and for sedation with local anesthesia.2,3Propofol is also useful for sedation in intensive care units.4
Like other anesthetics, propofol has amnesic effects.5Although suppression of memory function can persist for several hours after propofol administration,6the amnesic effects of propofol do not seem to be particularly strong with inhibition of explicit memory being weaker than that observed with midazolam.7In animal studies using a Morris water maze, rats did not show inhibition of memory tasks immediately after recovery from propofol administration.8
To improve anesthetic amnesia in clinical situations, it is important to understand how various anesthetics alter memory processing. Although the mechanisms involved in memory are complex, it is believed that persistent changes in synaptic function contribute to memory formation. Long-term potentiation (LTP) of synaptic transmission is one model for cellular processes thought to contribute to memory and learning, and understanding how pharmacologic treatments affect LTP can be useful for assessing mechanisms by which these agents disrupt memory.9It has been reported previously that propofol inhibits the maintenance but not the induction of LTP in vivo .10Effects on LTP in ex vivo preparations, however, have not yet been studied. LTP studies in hippocampal slices offer the opportunity to examine potential mechanisms contributing to propofol-induced amnesia under a high degree of experimental control. Using rat hippocampal slices, we studied the effects of propofol on several forms of long-term synaptic plasticity, including N -methyl-d-aspartate (NMDA) receptor–dependent forms of LTP and long-term depression (LTD) as well as a form of LTP not involving NMDA receptors.
Materials and Methods
Hippocampal Slice Preparation
Hippocampal slices were prepared from 30-day-old male Sprague-Dawley rats. Rats were decapitated under deep halothane anesthesia. The hippocampus was quickly dissected into gassed (95% oxygen–5% carbon dioxide) artificial cerebrospinal fluid containing 124 mm NaCl, 5 mm KCl, 1.25 mm NaH2PO4, 2 mm MgSO4, 22 mm NaHCO3, 10 mm glucose, and 2 mm CaCl2at 4°–6°C. Transverse slices (500 μm thick) were cut with a vibrotome (WPI, Sarasota, FL). After recovering for at least 1 h at 30°C in an incubation chamber, each slice was submerged in a constant-flow (2 ml/min) recording chamber and maintained at 30°C during the course of an experiment. Procedures for animal studies were approved by the Animal Care and Use Committee at Washington University School of Medicine (St. Louis, Missouri).
Extracellular recordings were obtained from the apical dendritic layer of the CA1 region using 5- to 10-MΩ glass electrodes filled with 1 m NaCl. Evoked synaptic responses were elicited with 0.1- to 0.2-ms constant current pulses through a bipolar electrode placed in the Schaffer collateral pathway. Stimuli, 50 μs in duration, were applied every minute. In all experiments, baseline synaptic transmission was monitored for 20 min before drug administration or delivery of tetanic stimulation. The stimulus intensity was set to evoke 50–60% of the maximal amplitude of field excitatory postsynaptic potentials (EPSPs). In another set of experiments, extracellular recordings were obtained from the pyramidal cell layer of the CA1 region to investigate the effects of propofol and pentobarbital on paired pulse inhibition of population spikes. For these studies, two stimuli of maximal intensity were delivered through a bipolar electrode placed in the Schaffer collateral pathway at an interval of 21 ms. Paired pulse inhibition was estimated as a ratio of the amplitude of the second population spike to the first population spike.
To induce LTP, θ-burst–patterned stimulation (TBS) was used. For most experiments, TBS consisted of 10 bursts (unless otherwise indicated) of 4 pulses at 100 Hz, applied at 5 Hz. In another set of experiments, LTP was induced by TBS consisting of 200-Hz stimulation in the presence of 1 μm MK-801, an NMDA receptor antagonist. LTD was induced using 900 pulses at 1 Hz. We also examined effects of 900-pulse stimulation at 10 Hz to examine effects on synaptic plasticity threshold.
For recording NMDA receptor–mediated synaptic responses, artificial cerebrospinal fluid containing low Mg2+(0.1 mm) and high Ca2+(2.5 mm) was used to facilitate activation of NMDA receptor–gated channels. Non-NMDA receptor–mediated components of synaptic responses were blocked using 6-cyano-7-nitro-quinoxaline-2, 3-dione (CNQX; 30 μm).
Field EPSPs were monitored and analyzed using the pCLAMP software data acquisition system (version 9.0; Axon Instruments, Foster, CA). EPSP slopes were measured as the maximum slope of the negative rising phase of the waveform. Changes in EPSP slopes were compared at the 50% point on baseline input–output curves. All responses were normalized as a percentage of control.
To evaluate effects of anesthetic agents on synaptic plasticity, each chemical was administered starting 20 min before conditioning stimulation. Picrotoxin (1 μm) was dissolved in ethanol. The final concentration of ethanol in artificial cerebrospinal fluid was 0.1%. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO).
Propofol was made up as a 300-mm stock solution in dimethyl sulfoxide. The final concentration of dimethyl sulfoxide in artificial cerebrospinal fluid was 0.1%.
Data are expressed as mean ± SE, and levels of significance were set at P < 0.05. Statistical significance was determined using the Student t test or Mann–Whitney U test to compare values between groups.
We initially evaluated the effects of propofol on basal synaptic transmission at Schaffer collateral synapses in the CA1 region. At concentrations of 30 μm or less, 20-min administration of propofol had little effect on baseline EPSPs. Administration of 100 μm propofol depressed EPSPs by 81 ± 5% (n = 4). We also examined the ability of propofol to modulate paired pulse plasticity of CA1 population spikes (PSs) using an interpulse interval of 21 ms and a stimulus intensity sufficient to elicit a maximal second PS. Under these conditions, there is little depression of the second PS, but treatments that augment γ-aminobutyric acid–mediated (GABAergic) tone promote paired pulse inhibition of the second spike. Propofol suppressed the second PS in a concentration-dependent manner with an IC50of 19 ± 5 μm (fig. 1). For comparison, we also examined the effects of pentobarbital, an anesthetic known to augment GABAergic function. Pentobarbital enhanced paired pulse inhibition with an IC50of 29 ± 3 μm (fig. 1). At a fixed concentration of 30 μm, propofol and pentobarbital changed the ratio of second to first PS by −80 ± 9% (n = 6) and −53 ± 9% (n = 6), respectively. As reported by other investigators,11the enhancement of paired pulse inhibition by propofol was blocked by 1 μm picrotoxin, a γ-aminobutyric acid type A (GABAA) receptor antagonist (change in the ratio of second to first PS by 30 μm propofol was −1 ± 3%, n = 5; data not shown). Similarly, enhancement of paired pulse inhibition by pentobarbital was also masked by picrotoxin (change in the ratio of second to first PS by 30 μm pentobarbital was −6 ± 2%, n = 5).
Because most forms of long-term plasticity require activation of NMDA receptors, we examined the effects of propofol on isolated NMDA receptor–mediated EPSPs (NMDA EPSPs). At concentrations of 300 μm or less, propofol exhibited only partial inhibition of NMDA EPSPs (fig. 2A), with 30 μm propofol producing 36 ± 3% inhibition (n = 9). This partial depression of NMDA EPSPs by 30 μm propofol was not altered when propofol was coapplied with 1 μm picrotoxin (EPSP change: 38 ± 6% inhibition, n = 4). To determine whether the partial inhibitory effects of propofol reflect NMDA receptor subtype selectivity, we examined interactions between propofol and ifenprodil on NMDA EPSPs. Ifenprodil inhibits NMDA receptors expressing NR2B subunits fairly selectively at low micromolar concentrations. NMDA EPSPs were depressed by approximately 40% by 10 μm ifenprodil. In the presence of ifenprodil, 30 μm propofol still partially depressed EPSPs, indicating that its effects were not completely occluded by ifenprodil (21 ± 2% depression, n = 5; fig. 2B). Based on these observations, it does not seem that propofol is selective for NMDA receptors containing NR2B subunits.
We next determined the effects of propofol on NMDA receptor–dependent forms of synaptic plasticity. In control slices, TBS consistently induced LTP (EPSPs 60 min after TBS: 163 ± 10% of control, n = 6) and this degree of LTP was not altered by 0.1% dimethyl sulfoxide, the agent used as a solvent for propofol (159 ± 8% of control, n = 5; fig. 3A). Continuous administration of 3 μm propofol starting 20 min before the delivery of TBS and maintained throughout the experiment did not alter LTP induction (166 ± 21% of control, n = 5; fig. 3B), whereas 10 μm propofol partially inhibited LTP induction (121 ± 10% of control, n = 5, P < 0.05 vs. control; fig. 3C). In the presence of 30 μm propofol administered continuously throughout the experiment, TBS did not induce LTP (99 ± 1%, n = 6; fig. 3D).
Although these results suggest that low micromolar concentrations of propofol do not alter LTP, it is possible that 20-min administration before TBS is insufficient for propofol to exhibit full activity. Therefore, we preincubated slices for 4 h in moderate concentrations of propofol and delivered TBS during continuous administration of propofol at the same concentration. Even after preincubation with 3 μm propofol, TBS consistently induced LTP (172 ± 6%, n = 5). However, preincubation with 10 μm propofol completely inhibited LTP (100 ± 4%, n = 5; data not shown).
To determine whether propofol blocks the induction or maintenance of LTP, propofol was administered either during or after TBS. Administration of 30 μm propofol for 20 min before and during the delivery of TBS blocked LTP induction (EPSP slope: 109 ± 1%, n = 6, figs. 4B and 5; P < 0.05 vs. control, fig. 4A). Administration of propofol starting 5 min after TBS, however, did not block LTP completely (139 ± 9%, n = 6; fig. 4C), although the degree of LTP was statistically less than control (P < 0.05). These findings indicate that propofol inhibits the induction but has less effect on the maintenance of NMDA receptor–dependent LTP.
The induction of LTP using 100-Hz TBS requires activation of NMDA receptors. In contrast, LTP induced by very-high-frequency stimulation (200 Hz) is independent of NMDA receptor activation. Delivery of a 200-Hz TBS induced a persistent increase in EPSPs in the presence of 1 μm MK-801, a noncompetitive NMDA receptor antagonist (EPSP slope: 190 ± 5%, n = 5; figs. 5 and 6). This NMDA receptor-independent LTP was inhibited by 30 μm propofol administered before and during the 200-Hz stimulation (EPSP slope: 115 ± 2%, n = 5, P < 0.05 vs. control; figs. 5 and 6).
To determine whether the effects of propofol on LTP involve GABAAreceptors, we examined the effects of propofol on LTP induction in the presence of picrotoxin. Because high concentrations of picrotoxin promote epileptiform activity in the CA1 region and make it difficult to measure EPSP slopes, we used 1 μm picrotoxin in the current study. Previously, we found that this concentration of picrotoxin diminishes GABAergic inhibition but does not alter baseline synaptic potentials.12In control slices treated with picrotoxin, we observed robust TBS-induced LTP (EPSP slope: 180 ± 4%, n = 6; fig. 7A). Propofol did not alter the induction of LTP in the presence of picrotoxin (176 ± 3%, n = 6; fig. 7A). Similarly, propofol did not inhibit NMDA receptor–independent LTP in the presence of picrotoxin. Delivery of 200-Hz very-high-frequency stimulation induced LTP in the presence of 1 μm MK-801, 1 μm picrotoxin, and 30 μm propofol (EPSP slope: 146 ± 10% n = 5, P = 0.34 vs. control; fig. 7B).
The finding that picrotoxin overcomes the effects of propofol on LTP induction suggests that modulation of GABAAreceptors contributes to LTP inhibition. We examined this further using pentobarbital. At 30 μm, pentobarbital did not alter NMDA EPSPs (EPSP slopes in the presence of CNQX and low magnesium were 99 ± 5% of control, n = 5; fig. 8C) but did enhance PS paired pulse inhibition via a GABAergic effect (fig. 1). Administration of pentobarbital before and during 100-Hz TBS did not completely suppress LTP induction, although the amplitude of LTP was less than control (EPSP slope: 135 ± 7%, n = 6, P < 0.05 vs. control; figs. 5 and 8A).
Because the effects of pentobarbital on TBS-LTP differed from those of propofol, we also examined whether pentobarbital inhibits NMDA receptor–independent LTP in the presence of 1 μm MK-801. Similar to effects on NMDA receptor–dependent LTP induced by TBS, administration of 30 μm pentobarbital for 20 min before and during the delivery of very-high-frequency stimulation diminished but did not completely block LTP induction (EPSP slope: 137 ± 10%, n = 5, P = 0.15 vs. control; fig. 8B). Given that 10–30 μm propofol completely blocks LTP, these results suggest that the effects of pentobarbital on LTP differ from those of propofol and that propofol has actions in addition to modulation of GABAAreceptors that contribute to its LTP inhibition.
We also examined the effects of propofol on the induction of homosynaptic LTD, a form of plasticity that requires activation of NMDA receptors in the CA1 region. In control slices, delivery of low-frequency stimulation consisting of 900 pulses delivered at 1 Hz (900-s stimulation) successfully induced LTD in control slices (EPSP slopes: 69 ± 3%, n = 7; fig. 9A). LTD induction was not blocked by administration of 30 μm propofol for 20 min before and during low frequency stimulation (66 ± 5%, n = 6; fig. 9A). Because propofol inhibits the induction of LTP but not LTD, it is possible that propofol shifts the frequency dependence of synaptic plasticity and thus would allow 10-Hz intermediate-frequency stimulation to induce LTD. In control slices, delivery of 900 pulses at 10 Hz (90-s stimulation) did not induce LTD (EPSP slopes: 92 ± 2%, n = 5; fig. 9B). Administration of 30 μm propofol for 20 min before and during delivery of 10 Hz stimulation had no significant effect (87 ± 3%, n = 5; fig. 9B), indicating that propofol did not shift the overall frequency dependence of synaptic plasticity towards LTD.
Recent evidence indicates that propofol-induced anesthesia is not accompanied by recall of events occurring during the hypnotic state.13This suggests that propofol disrupts memory formation during anesthesia and raises the possibility that anesthetically relevant concentrations of propofol also disrupt forms of synaptic plasticity thought to contribute to memory processing. Although we observed that propofol inhibits LTP in hippocampal slices, it is difficult to determine how the concentrations required for modulation of synaptic function and plasticity correlate with concentrations required for effective anesthesia. In humans, blood concentrations of propofol required for sedation range from 0.5 to 1 μg/ml (2.8–5.6 μm),14whereas those for general anesthesia range from 2 to 6 μg/ml (11.2–33.6 μm).15In rats, the blood concentrations obtained upon awakening are less than 3.5 μg/ml (19.6 μm), and additional doses are required to maintain hypnosis.16These blood concentrations, however, may not exactly mirror the concentrations in extraneuronal spaces. It has been shown that concentrations of propofol in rat brain are greater than those in blood after intravenous infusion.17Investigators have often used relatively high concentrations of propofol to demonstrate inhibitory effects on population spikes,11,18presumably because of poor penetration of propofol into brain slices when administered by bath perfusion.19Although the concentrations of propofol that inhibited LTP are greater than the concentrations in the cerebrospinal fluid calculated from doses used in clinical practice,20the concentrations for LTP inhibition are similar to those required for paired pulse inhibition of population spikes. Therefore, the concentrations of propofol used in the current study seem to correlate with concentrations achieved during anesthesia, given that paired pulse inhibition in slices and clinical anesthesia are both thought to involve GABAergic actions. Considering that the IC50for propofol-mediated enhancement of paired pulse inhibition was 19 μm in hippocampal slices, it seems that the 10- to 30-μm concentrations associated with LTP inhibition are within a range that could be relevant for anesthesia.
To study possible mechanisms underlying memory disruption by propofol, we focused on examining 30 μm propofol, a concentration that completely inhibited LTP. We found that propofol inhibits LTP induction but has less effect on LTP maintenance in the CA1 region. A previous study found that intraperitoneal injections of propofol inhibit the maintenance but not the induction of LTP in vivo .10Reasons for this discrepancy are not clear but may involve differences in experimental conditions including drug administration methods.
Propofol is a partial inhibitor of NMDA receptors, and this partial inhibition in combination with GABAergic modulation may be critical for explaining effects on synaptic plasticity. When the effects of propofol are blocked by picrotoxin, the partial inhibitory effects of propofol on NMDA receptors are insufficient to block NMDA receptor–dependent LTP. Furthermore, propofol also blocks a form of LTP not dependent on NMDA receptors. These observations strongly suggest that there are several mechanisms contributing to the effects of propofol on synaptic plasticity.
Because activation of NMDA receptors plays a key role in LTP induction, we initially hypothesized that inhibition of NMDA receptors would account for propofol-mediated LTP inhibition. Previous studies have shown that propofol partially inhibits NMDA receptor–mediated responses in dissociated cultured neurons and in Xenopus oocytes.21,22Partial inhibition of NMDA receptor–mediated synaptic transmission, however, is typically not sufficient to inhibit LTP induction,23unless there is relatively complete block of the component of transmission mediated by NR2A containing NMDA receptors.24Our results indicate that propofol is not selective for specific subtypes of NMDA receptors, and thus, partial inhibition of NMDA receptors by propofol seems unlikely to account for the block of LTP.
Consistent with the idea that partial block of NMDA receptors does not account for the effects of propofol on LTP, we found that propofol also inhibits LTP induced by 200-Hz TBS that is independent of NMDA receptor activation.25Previous studies have shown that LTP induced by 200-Hz stimulation involves activation of L-type calcium channels but not NMDA receptors.25Therefore, 200-Hz stimulation induces LTP even in the presence of NMDA receptor block with MK-801 (fig. 6). Our observation that propofol inhibited the induction of NMDA receptor–independent LTP further supports the idea that effects on NMDA receptors cannot fully account for propofol-mediated inhibition of LTP.
We also found that propofol has GABAergic effects at concentrations that inhibit LTP. Several lines of evidence suggest that inhibition of GABAAreceptors enhances NMDA receptor activation and facilitates LTP.26Loss of GABAergic inhibition may also act as a priming stimulus for the subsequent induction of LTP by increasing calcium influx through L-type calcium channels.27In contrast, strong GABAA-mediated input suppresses calcium ion influx through calcium channels and NMDA receptor channels.28,29Propofol is known to enhance the function of GABAAreceptors expressing a variety of GABAAreceptor subunits.30,31Because picrotoxin overcomes the inhibitory effects of propofol on LTP induction, it seems that enhancement of GABAAreceptor function contributes significantly to the LTP inhibition mediated by propofol. Similarly, other anesthetics like midazolam and isoflurane also inhibit LTP.32,33Because the inhibition of LTP by these agents is overcome by picrotoxin, it is likely that these agents share a common mechanism. Similar to propofol, ethanol partially inhibits NMDA receptors and depresses LTP induction. Schummers and Browning34speculate that ethanol-mediated LTP inhibition results from the synergistic effects of partial NMDA receptor block and potentiation of GABAAreceptors. They showed that GABAAreceptor block with picrotoxin diminished the effects of ethanol on NMDA receptor–mediated EPSPs. In our experiments, the inhibitory effects of propofol on NMDA receptor–mediated EPSPs were not altered by picrotoxin.
To test whether modulation of GABAergic function is sufficient to suppress LTP induction, we examined pentobarbital, a barbiturate known to potentiate GABAAreceptors. Although pentobarbital does not inhibit LTP induction in vivo ,35effects of pentobarbital on LTP induction in vitro have not been studied in detail. Because 100 μm pentobarbital depressed baseline synaptic responses, we focused our studies on 30 μm pentobarbital. Thirty micromolars is a concentration within the range found in rat CSF immediately after intravenous injection of pentobarbital (20–40 mg/kg).36We chose to examine 30 μm pentobarbital based on the observation that pentobarbital enhanced paired pulse inhibition in a picrotoxin-sensitive fashion with an IC50of 29 μm. At 30 μm, pentobarbital had no effect on spike generation elicited by single stimuli and only partially suppressed LTP generation. Previous work has shown that LTP is difficult to induce under conditions where pentobarbital diminishes spike firing in slices.37Because 30 μm pentobarbital did not alter NMDA EPSPs and 10–30 μm propofol produces a greater block of LTP with comparable GABAergic effects, it seems that additional factors are likely to contribute to propofol-mediated LTP inhibition. It is also possible that the modulation of GABAAreceptors by propofol differs from other modulators, resulting in unique effects on LTP induction. Pentobarbital prolongs the open time of GABA channels without affecting closed times,38whereas propofol increases the channel open probability via a decrease in closed times.39Furthermore, pentobarbital does not block NMDA receptor–independent LTP, whereas propofol blocks this form of LTP. This result further suggests that the effects of propofol involve more than GABAergic enhancement. It is possible that effects on calcium channels, intracellular calcium homeostasis, or second messenger systems contribute to effects on synaptic plasticity. In particular, inhibition of L-type calcium channels by propofol40,41coupled with GABAergic effects could account for the block of NMDA receptor–independent forms of LTP.
Long-term depression represents another form of synaptic plasticity in the hippocampus. Recent studies indicate that LTD induction may require a specific NMDA receptor subtype with selective inhibition of NMDA receptors containing NR2B subunits preventing the induction of LTD but not LTP in the CA1 region.24,42We examined whether propofol mimics the effects of the NR2B antagonist ifenprodil on NMDA receptors. Because both ifenprodil-sensitive NMDA EPSPs and ifenprodil-insensitive NMDA EPSPs were inhibited by propofol, it seems that propofol does not block a specific subtype of NMDA receptors. The failure of propofol to inhibit LTD induction may reflect the only partial inhibitory effects of the drug on NR2B-containing NMDA receptors. However, Hendricson et al. 43have shown that ifenprodil can facilitate the induction of LTD under some conditions, leaving some uncertainty about the role of NR2B receptors in LTD induction.
Taken together, our results indicate that propofol inhibits the induction of LTP with less effect on the maintenance of LTP and no effect on LTD. Although propofol is a partial inhibitor of NMDA receptors with actions that are distinct from subtype specific antagonists, the inhibition of LTP cannot be explained simply by partial inhibition of NMDA receptors. In addition, our results indicate that effects on GABAAreceptors contribute significantly to LTP inhibition, but these actions alone seem unlikely to account fully for the effects of propofol on synaptic plasticity.