There is much evidence that the sedative component of anesthesia is mediated by gamma-aminobutyric acid type A (GABA(A)) receptors on hypothalamic neurons responsible for arousal, notably in the tuberomammillary nucleus. These GABA(A) receptors are targeted by gamma-aminobutyric acid-mediated (GABAergic) neurons in the ventrolateral preoptic area (VLPO): When these neurons become active, they inhibit the arousal-producing nuclei and induce sleep. According to recent studies, propofol induces sedation by enhancing VLPO-induced synaptic inhibition, making the target cells more responsive to GABA(A). The authors explored the possibility that propofol also promotes sedation less directly by facilitating excitatory inputs to the VLPO GABAergic neurons.
Spontaneous excitatory postsynaptic currents were recorded from VLPO cells-principally mechanically isolated, but also in slices from rats.
In isolated VLPO GABAergic neurons, propofol increased the frequency of glutamatergic spontaneous excitatory postsynaptic currents without affecting their mean amplitude. The action of propofol was mimicked by muscimol and prevented by gabazine, respectively a specific agonist and antagonist at GABA(A) receptors. It was also suppressed by bumetanide, a blocker of Na-K-Cl cotransporter-mediated inward Cl transport. In slices, propofol also increased the frequency of spontaneous excitatory postsynaptic currents and, at low doses, accelerated firing of VLPO cells.
Propofol induces sedation, at least in part, by increasing firing of GABAergic neurons in the VLPO, indirectly by activation of GABA(A) receptors on glutamatergic afferents: Because these axons/terminals have a relatively high internal Cl concentration, they are depolarized by GABAergic agents such as propofol, which thus enhance glutamate release.
THE cellular and molecular mechanisms underlying the effects of general anesthetics are not well understood. The sedative component of anesthesia seems to be mediated by γ-aminobutyric acid type A receptors (GABAARs) in an endogenous sleep pathway,1,2the relevant γ-aminobutyric acid–mediated (GABAergic) neurons being located in the ventrolateral preoptic area (VLPO).3–8Recently, Zecharia et al. 2showed that GABAergic agents such as muscimol and propofol potentiate the GABAergic inhibition of orexinergic perifornical9and histaminergic tuberomammillary neurons.6Together with other monoaminergic and cholinergic projections to the cortex, the neurons in tuberomammillary nuclei directly promote cortical arousal,10,11and all are activated by orexinergic neurons.12Therefore, their inhibition by VLPO neurons is pivotal in the initiation of sleep.13,14Although the activity of VLPO neurons is strongly influenced by circadian fluctuations in input from suprachiasmatic cells,15the detailed mechanisms underlying activation of VLPO neurons are largely unexplored.16
General anesthetics are known to have marked effects on synaptic transmission. Most previous studies found that propofol enhances the function of inhibitory GABAARs17–20and may depress the release of glutamate, the major excitatory neurotransmitter.21However, in vitro electrophysiologic studies have shown that, in several brain areas, activation of presynaptic GABAARs have a depolarizing effect that enhances glutamatergic transmission.22–25We hypothesized that by potentiating the depolarizing action of presynaptic GABAARs, propofol may increase glutamatergic transmission in the VLPO and thus increase the activity of VLPO GABAergic neurons. To test this hypothesis, we recorded spontaneous excitatory postsynaptic currents (sEPSCs) in rat VLPO GABAergic neurons, either in slices or isolated by a mechanical, enzyme-free procedure. These neurons preserve some functional glutamate-releasing terminals after isolation.
Materials and Methods
All experiments were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals , and they were approved by the Institutional Animal Care and Use Committee of the University of Medicine and Dentistry of New Jersey, Newark, New Jersey. The experiments were performed on brains from Sprague-Dawley rats.
Slice Preparation and Mechanical Dissociation
The hypothalamic slices were prepared as described previously.26,27Briefly, Sprague-Dawley rats (aged 10–28 postnatal days) were anesthetized and then decapitated. Coronal slices (300 μm thick) were cut using a VF-200 slicer (Precisionary Instruments Inc., Greenville, NC). They were prepared in ice-cold glycerol-based artificial cerebrospinal fluid containing 250 mm glycerol, 1.6 mm KCl, 1.2 mm NaH2PO4, 1.2 mm MgCl2, 2.4 mm CaCl2, 25 mm NaHCO3, and 11 mm glucose and saturated with 95% O2–5% CO2(carbogen).28Slices were allowed to recover for at least 1 h in a holding chamber at 32°C in carbogen-saturated regular artificial cerebrospinal fluid, which has the same composition as glycerol-based artificial cerebrospinal fluid, except that glycerol was replaced by 125 mm NaCl.
Neurons with functional presynaptic terminals attached were obtained by mechanical dissociation (fig. 1A), as previously described,29with some modifications.28,30A slice containing the VLPO was transferred to a 35-mm culture dish (Falcon, Rutherford, NJ) and held down with a flat U-shaped wire. The dish was filled with standard external solution containing 140 mm NaCl, 5 mm KCl, 2 mm CaCl2, 1 mm MgCl2, 10 mm HEPES, and 10 mm glucose (320 mOsm, pH adjusted to 7.3 with Tris base). Under an inverted microscope (Nikon, Tokyo, Japan), we identified VLPO by its stereotaxic coordinates,31and located the VLPO cells that project to the noradrenergic locus ceruleus and the histaminergic tuberomammillary nucleus.3,6,32A fire-polished pipette, held by a micromanipulator, lightly touched the surface of the VLPO region and vibrated horizontally at 15–20 Hz for 2–5 min. After 20 min, isolated neurons that adhered to the bottom of the dish were kept for electrophysiologic recordings at room temperature (21°–22°C). These mechanically dissociated neurons often preserve functional nerve terminals, including some that release glutamate.29,30
Electrical signals were obtained in whole cell configurations with an Axon 200B amplifier or a MultiClamp 700A (Molecular Devices Co., Union City, CA), a Digidata 1320A A/D converter (Molecular Devices Co.), and pCLAMP 9.2 software (Molecular Devices Co.). Data were filtered at 1 kHz and sampled at 5 kHz. The patch electrodes had a resistance of 2–5 MΩ when filled with the pipette solution, containing 135 mm CsF, 5 mm KCl, 2 mm MgCl2, 10 mm HEPES, 2 mm Mg adenosine-5′-triphosphate, and 0.2 mm guanosine-5′-triphosphate (for voltage clamp). For current clamp recordings, the CsF in the above pipette solution was replaced by K-gluconate. The pH of the pipette solution was adjusted to 7.2 with Tris base. Neurons were voltage clamped at −60 mV to record α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptor–mediated spontaneous and miniature excitatory postsynaptic currents (sEPSCs and mEPSCs).
In several experiments, a single coronal slice was transferred into a 0.4-ml recording chamber where it was held down by a platinum ring. Warm carbogenated artificial cerebrospinal fluid flowed through the bath (1.5–2.0 ml/min). Under infrared video microscopy (E600FN; Nikon), we recorded from neurons located in the core of VLPO, where a cluster of sleep-active neurons were originally identified, as illustrated in figure 1.5,6Cells were deemed of the sleep promoting type according to previously established criteria3: First, they were triangular in shape (fig. 1A1), in contrast to the fusiform, non–low-threshold spike type (fig. 1A2); second, when recorded in current clamp mode, they generated characteristic low-threshold spikes in response to a depolarizing pulse or after a hyperpolarizing current step (figs. 1B1 and C; fig. 1B2shows the high-threshold spikes of non–low-threshold spike cells); and finally, they were inhibited by noradrenaline (fig. 1D1). Triangular-shaped cells inhibited by noradrenaline always displayed low-threshold spikes. By contrast, noradrenaline excited a non–low-threshold spike neuron (fig. 1D2). The series resistance (15–30 MΩ) or input resistance (300–500 MΩ) was monitored throughout the whole cell recording, and data were discarded if the resistance changed by more than 20%. All these recordings were made at 32°C, maintained by an automatic temperature controller (Warner Instruments, Hamden, CT).
Chemicals and Applications
The chemicals, including 2, 6-diisopropylphenol (propofol), gabazine (SR 95531), 2,3-dihydroxy-6,7-dinitroquinoxaline (DNQX), tetrodotoxin, bicuculline, and bumetanide were obtained from Sigma-Aldrich Chemical Company (St Louis, MO). Drugs were added to the superfusate or applied to a cell by a fast perfusion system (Y tube). Solutions in the vicinity of a neuron could be completely exchanged within 40 ms without damaging the seal.33
sEPSCs and mEPSCs were counted and analyzed with Clampfit 9.2 (Molecular Devices Co.); sEPSCs and mEPSCs were screened automatically (6-pA amplitude threshold), checked visually, and accepted or rejected according to their rise and decay times. The frequency and amplitude of all events, during and after drug application, were compared with the mean of the values observed during the initial control period. Cumulative probability plots of the incidence of various interevent intervals and amplitudes, recorded under different conditions from the same neuron, were subjected to the Kolmogorov–Smirnov test. For other plots, data obtained over 1-min periods at the peak of a drug response were compared with the average values of frequency (or amplitude) of the sEPSC (or mEPSC) recorded during the initial control period (1–4 min). Data are expressed as mean (±SEM). sEPSCs were fitted with a standard single exponential equation (Clampfit 9.2) to determine the time constant of decay from the EPSC peak. The statistical significance of drug effects was assessed by a paired two-tailed t test on normalized data with Sigma Plot (Systat Software Inc., San Jose, CA). Values of P < 0.05 were considered significant.
Spontaneous EPSCs Recorded in Neurons Mechanically Dissociated from VLPO
In VLPO, two thirds of the neurons have a characteristic triangular and multipolar shape and low-threshold-spikes; moreover, because these cells contain the GABA-synthesizing enzyme glutamic acid decarboxylase and are inhibited by noradrenaline and carbachol, they correspond to the GABAergic VLPO cells known to be active during sleep.3,8In the current study, we recorded from such multipolar, triangular-shaped neurons within VLPO (fig. 1). Most of the experiments were performed on mechanically dissociated neurons (fig. 1A1), which combine several advantages: good space clamp; preservation of functioning synaptic terminals, including some that release glutamate; and better control of the surrounding solution.29,30,34,35The traces in figures 1B–Dwere recorded under current clamp. The voltage traces in figures 1B1, C, and Dwere from a triangular-shaped, multipolar VLPO neuron. Depolarization from a relatively hyperpolarized level was followed by the characteristic depolarizing hump and burst of firing (inset in figs. 1B1 and C), and noradrenaline-induced membrane hyperpolarization (fig. 1D1). Conversely, in the bipolar, fusiform VLPO neuron (fig. 1A2), in response to a depolarization pulse there was a slow voltage rise toward the first action potential (fig. 1B2, arrow) and noradrenaline-induced membrane depolarization and firing (fig. 1D2). Under voltage clamp, sEPSCs were recorded with the CsF-based internal solution at a holding potential of −60 mV. Under these conditions, postsynaptic responses to GABA or glycine are suppressed,36allowing us to monitor changes in frequency and amplitude of the isolated EPSCs.30,37,38DNQX (10 μm), an antagonist of AMPA/kainate receptor, reversibly blocked the spontaneous synaptic events (fig. 1E), confirming that these VLPO neurons receive glutamatergic excitatory inputs.15
Propofol Increases sEPSC Frequency in Neurons Mechanically Dissociated from VLPO
Propofol (1–100 μm) was applied to isolated VLPO neurons by a fast perfusion system (Y-tube).39Propofol (10 μm) did not significantly change the holding current, whereas it robustly but reversibly increased the frequency of sEPSCs (fig. 2). A second application of propofol had a comparable effect (88 ± 7% of first response; n = 8, P > 0.5; figs. 2A and B). As depicted by averaged traces recorded before and during propofol application (fig. 2C) and the dose–response data in figure 2E, propofol (10 μm) did not significantly alter the mean amplitude of sEPSCs (7.6 ± 1.2 pA in control and 10.3 ± 1.8 pA in presence of propofol; P > 0.05, n = 30 events). The superimposed normalized mean traces at right in figure 2Cillustrate the absence of any change in sEPSC kinetics: The decay time constant of sEPSCs was not significantly altered (control: 3.9 ± 1.3 ms, propofol: 2.3 ± 0.4 ms; P = 0.13, by t test). The effect of propofol on sEPSC frequency depended on its concentration, as shown in figure 2D. The apparent EC50for propofol was 4.2 μm. These results are consistent with a presynaptic mechanism of propofol action, enhancing glutamatergic transmission by increasing glutamate release.
Propofol Acts Only on sEPSCs That Are Suppressed by Tetrodotoxin and Cd2+
We next examined how block of synaptic transmission affects the propofol-induced facilitation of sEPSC frequency. First, we examined the effect of 1 μm tetrodotoxin, a blocker of voltage-dependent Na+channels. Tetrodotoxin (1 μm) sharply decreased sEPSC frequency to 38 ± 10% of control (from 3.4 ± 1.2 Hz in control to 1.1 ± 0.6 Hz in tetrodotoxin; n = 5, P < 0.01), but a moderate reduction in sEPSC amplitude was not significant (to 75 ± 7% of control, from 17 ± 5 pA to 13 ± 4 pA; n = 5, P > 0.05). In the presence of 1 μm tetrodotoxin, propofol (10 μm) did not significantly alter either the frequency or the amplitude of the remaining EPSCs (presumably mEPSCs) (figs. 3A–C). These results indicate that propofol acts only on sEPSCs generated by ongoing activity of tetrodotoxin-sensitive Na+channels.
In further experiments, we applied Cd2+, a general blocker of voltage-dependent Ca2+channels, which also suppresses synaptic transmission. Indeed, acting somewhat like tetrodotoxin, 100 μm Cd2+reduced sEPSC frequency by 36 ± 10% (P = 0.03, n = 5) and amplitude by 32 ± 9% (P = 0.02, n = 5), and in its presence, 10 μm propofol did not significantly alter the frequency of the remaining EPSCs (figs. 3D–F). Judging by these results, propofol targets only EPSCs generated by Na+channel–dependent terminal depolarization and the resulting Ca2+influx.
The Propofol-induced Increase in Glutamatergic Activity Is Eliminated by a GABAAR Antagonist and Mimicked by a GABAAR Agonist
It is well known that propofol enhances the function of GABAARs, in particular those containing β2and β3subunits,40and that activation of presynaptic GABAARs enhances glutamate release in several brain areas.22–25To test for possible involvement of GABAARs in the acceleration of sEPSC frequency, we applied gabazine (10 μm), a specific antagonist of GABAARs. As shown in figures 4A–C, in the presence of gabazine, 10 μm propofol caused a mild reduction in sEPSC frequency (−17.9 ± 2.1%, n = 6), presumably mediated by its GABAAR-independent depressant effect on glutamate release.21To further confirm the role of GABAARs in the facilitation of sEPSCs, we applied muscimol, the specific GABAAR agonist. In five cells tested, 10 μm muscimol significantly increased sEPSC frequency, but not amplitude (figs. 4C–E).
The Propofol-induced Increase in Glutamatergic Activity Is Attenuated by an Antagonist of Na+–K+–2Cl−Cotransporters
We have shown that voltage-dependent Na+and Ca2+channels, as well as GABAARs, are involved in the propofol-induced facilitation of sEPSC frequency. These data suggest that the activation of presynaptic GABAARs depolarizes some glutamatergic nerve terminals by generating Cl−efflux. This implies that the intraterminal Cl−concentration ([Cl−]) is relatively high, perhaps owing to inwardly directed Cl−transport systems via Na+–K+–2Cl−cotransporters (NKCCs) and/or the Cl−–HCO3−exchanger.41–43In peripheral sensory axons and immature or injured CNS neurons, NKCC increases intracellular [Cl−]; hence, in these cells, GABA induces depolarization.43Indeed, in some other hypothalamic neurons, NKCC is responsible for a relatively high [Cl−] in glutamatergic nerve terminals.44,45To determine whether a high internal [Cl−], maintained by NKCC, is essential for the propofol-induced facilitation of sEPSC frequency, we tested propofol in the presence of 10 μm bumetanide, which selectively blocks NKCC43,46(but not GABA receptors47). As illustrated in figure 5by original traces in A, the plot in B, and mean data in C, after 5 min of 10 μm bumetanide application, 10 μm propofol did not significantly alter the frequency of sEPSCs. The effect of propofol slowly recovered to control values after washing out bumetanide (data not shown). This result suggests that functional NKCCs are present on the juxtaterminal portion of glutamatergic axons, where they maintain a relatively high internal [Cl−].
Propofol Raises the Frequency Both of sEPSCs and of Ongoing Firing of VLPO Neurons in Brain Slices
We confirmed that propofol modulates glutamate release onto VLPO neurons under more physiologic conditions in slices (fig. 6). Propofol was tested on multipolar VLPO neurons, identified by the standard criteria (fig. 6A). As illustrated by the traces (fig. 6B), plot (fig. 6C), and histograms (fig. 6D), 2 μm propofol robustly increased the frequency (but not amplitude) of glutamatergic sEPSCs in these VLPO neurons. Having established that propofol (2 μm) facilitates glutamatergic transmission in slices, we next assessed the physiologic consequences of the action of propofol by examining the ongoing discharge of VLPO neurons. As shown in figures 6E and F, 0.1 μm propofol significantly increased the frequency of spontaneous firing of VLPO neurons in brain slices. Interestingly, as illustrated in figures 6G and H, at a higher concentration (2 μm) propofol had a sharp depolarizing action, which inactivated firing.
The central finding of this study is that propofol—at concentrations (1–10 μm) similar to those found effective in potentiating GABAARs in previous studies (e.g. ,48–51)—raises the frequency of spontaneous glutamatergic transmission to VLPO neurons, both isolated and in slices. This is the first electrophysiologic demonstration of the potentiating effect of propofol on glutamatergic transmission at the single cell level. An interesting parallel observation is that propofol, over a range of moderate concentrations (10–13 μm), selectively increases Ca-dependent glutamate release from cortical synaptosomes.52The VLPO neurons under study were multipolar in shape, generated low-threshold spikes, and were inhibited by noradrenaline, and therefore belonged to the GABAergic population3that projects to and inhibits the principal deep nuclei responsible for cortical and behavioral arousal.8According to previous studies,2,20the sedative action of propofol is mediated especially by enhanced GABAergic inhibition of the cells of the tuberomammillary nuclei. Specifically, Zecharia et al. 2recently found that propofol enhances GABAergic inhibition of these cells by increasing the duration of inhibitory synaptic responses. This did not exclude the possibility that propofol also increases the excitability of VLPO neurons in some indirect manner, e.g. , by disinhibition.1The current findings show that propofol increases ongoing excitation of the GABAergic VLPO neurons by enhancing glutamate release. Clearly, this mechanism could be a significant component of the sedative action of propofol.
Our recordings of AMPA receptor–mediated glutamatergic sEPSCs in the VLPO neurons are in keeping with a previous report.15A presynaptic target for the action of propofol was indicated by the increase in sEPSC frequency and insignificant changes in the mean amplitude of sEPSCs and mEPSCs. The presence of GABAARs on or near the glutamatergic terminals was supported by the demonstration that the effect of propofol was mimicked by muscimol, a GABAAR agonist, and blocked by gabazine, a GABAAR antagonist. The fact that activation of these GABAARs enhances glutamate release is readily explained if the Cl−concentration in these glutamatergic axons/terminals is relatively high; in which case, opening the presynaptic Cl−channels would induce Cl−efflux and thus membrane depolarization. The higher internal Cl−concentration seems to be the result of the activity of axonal NKCC-mediated inward Cl−transport, because bumetanide—a selective inhibitor of Cl−transport that does not block GABAAreceptors47—prevented the action of propofol. On the other hand, in view of the block of propofol’s effect by tetrodotoxin, the depolarization directly caused by presynaptic GABAAR–induced Cl−efflux was unable to induce glutamate release, either because it was of insufficient magnitude to activate Ca2+influx into the terminal or because the operative GABAARs are situated at some distance, requiring the conduction of tetrodotoxin-sensitive action potentials. In either case, the participation of voltage-dependent Na+channels was crucial.
Previous studies on spinal cord, hippocampus, thalamus, and neocortex have not reported an excitatory effect of propofol on glutamatergic EPSCs. The simplest explanation could be that in contrast to previous authors, we recorded EPSCs in the absence of GABAAR antagonists, which eliminate this propofol action (as shown by our observations in the presence of gabazine).
Previous studies suggest that some GABAAR agonists promote sleep by inhibiting the histaminergic cells in the tuberomammillary nucleus and weakly activating the VLPO via agonist binding to the α1subunit of GABAARs, whereas gaboxadol (4,5,6,7-tetrahydroisoxazolo[5,4-c]pyridin-3-ol [THIP]) binds to the α4δ subunits, potentially promoting sleep by activation of the VLPO.16However, how activation of GABAARs could excite VLPO neurons was not determined. Our study may provide an answer to this question. Future research will define the relevant subunits of GABAARs and whether the α4δ subunits in particular mediate the propofol-induced potentiation of sEPSCs of VLPO neurons.
In the current study, we tested 2 μm propofol on both mechanically dissociated neurons and slices. Its effects on sEPSCs (in whole cell voltage clamp recordings) were similar in both preparations, but the facilitation of sEPSCs occurred more quickly in mechanically dissociated neurons than in slices, as might be expected in view of the slow diffusion of propofol in slices.53On the other hand, in current clamp recordings of ongoing firing in slices, we found that even 0.1 μm propofol significantly increased the firing rate of some VLPO neurons. Interestingly, at 2 μm, propofol induced a sufficiently large depolarization that firing was inactivated. A possible explanation is that the indirect potentiation of glutamate release produced by propofol is especially pronounced in slices, where many more glutamatergic inputs to VLPO neurons are likely to be preserved.
In conclusion, our results show that chloride channel and GABAAreceptors exist on the glutamatergic axons/terminals that make synapses on the VLPO neurons. By enhancing Cl−efflux from these axons, propofol depolarizes these terminals and stimulates the release of glutamate, which increases the activity of VLPO neurons and thus potentiates GABAergic inhibition of arousal systems. This indirect mode of propofol action can be expected to contribute to its known effectiveness as a general anesthetic.