Muscarinic activation of inwardly rectifying K⁺ conductance reduces EPSPs in rat hippocampal CA1 pyramidal cells

Abstract

1. To determine how acetylcholine (ACh) modulates the somatodendritic processing of EPSPs, we performed whole-cell recordings from CA1 pyramidal cells of hippocampal slices and examined the effect of the cholinergic agonist, carbachol (CCh), on α-amino-3-hydroxy-5-methyl isoxazole-4-propionate (AMPA) EPSPs, miniature EPSPs, and EPSP-like waveforms evoked by brief dendritic glutamate pulses (glutamate-evoked postsynaptic potentials, GPSPs).

2. Although CCh is known to enhance the intrinsic excitability of the neuron in several ways, activation of atropine-sensitive (muscarinic) receptors on the apical dendrite or the soma of CA1 pyramidal cells consistently reduced the amplitude of EPSPs and GPSPs.

3. Cholinergic inhibition of evoked and simulated EPSP waveforms displayed considerable voltage dependence, with the amplitude of the postsynaptic potentials progressively declining with membrane hyperpolarization indicating the involvement of an inwardly rectifying current.

4. Extracellular Ba²⁺ (200 μm) and tertiapin (30 nm), a novel and selective blocker of G protein-activated, inwardly rectifying K⁺ (GIRK) channels, completely blocked the effect of CCh on GPSP amplitude.

5. Muscarinic reduction of GPSPs was not sensitive to the M1 receptor-preferring antagonist, pirenzepine, but was suppressed by the M2 receptor-preferring antagonist, methoctramine, and by the allosteric M2 receptor antagonist, gallamine.

6. In voltage-clamp recordings, CCh induced an ion current displaying inward rectification in the hyperpolarizing direction, which was identified as a GIRK current based on its sensitivity to low Ba²⁺ and tertiapin. Its pharmacological profile paralleled that of the cholinergic GPSP reduction.

7. We link the observed reduction of postsynaptic potentials to the cholinergic activation of a GIRK conductance, which serves to partially shunt excitatory synaptic input.

The hippocampus receives a major cholinergic input from the medial septum/diagonal band that has been implicated in the generation of behaviourally relevant, rhythmic network activity, such as theta rhythm and gamma oscillations (Buzsaki et al. 1983; Bland et al. 1988; Fisahn et al. 1998). At the cellular level, cholinergic afferents exert a plethora of actions, comprising both pre- and postsynaptic effects. Activation of presynaptic acetylcholine (ACh) receptors leads to reduced transmitter release at excitatory and inhibitory synapses (Hounsgaard, 1978; Valentino & Dingledine, 1981; Sheridan & Sutor, 1990; Sugita et al. 1991; Behrends & ten Bruggencate, 1993; Bellingham & Berger, 1996; Kimura & Baughman, 1997). On the postsynaptic side, a variety of intrinsic ion conductances are subjected to cholinergic modulation, including voltage-dependent, Ca²⁺-dependent and leak-K⁺ currents (Halliwell & Adams, 1982; McCormick & Prince, 1986a; Madison et al. 1987; Benson et al. 1988; Uchimura & North, 1990; Müller & Connor, 1991; Zhang et al. 1992; Kitai & Surmeier, 1993; Guérineau et al. 1994), voltage-dependent Ca²⁺ currents (Misgeld et al. 1986; Gähwiler & Brown, 1987; Howe & Surmeier, 1995; Toselli & Taglietti, 1995), non-selective cation currents (Colino & Halliwell, 1993; Guérineau et al. 1995; Haj-Dahmane & Andrade, 1996; Klink & Alonso, 1997), and fast and persistent Na⁺ currents (Cantrell et al. 1996; Mittmann & Alzheimer, 1998). With few exceptions, cholinergic stimulation of the postsynaptic neuron increases its intrinsic excitability. In hippocampal CA1 pyramidal neurons, hallmarks of ACh receptor activation are membrane depolarization, impairment of adaptation during repetitive spike firing, suppression of slow after-hyperpolarization (AHP), and appearance of after-depolarization (ADP) (reviewed in Misgeld, 1988; Nicoll et al. 1990).

A postsynaptic action of ACh largely overlooked so far is its effect on the integration of synaptic signals. Acetylcholine causes a slow and long-lasting facilitation of an NMDA receptor-mediated EPSP component (Markram & Segal, 1990), but the question remains, how cholinergic activity influences the processing of AMPA receptor-mediated EPSPs in the postsynaptic neuron. That cholinergic input should have indeed considerable impact on fast synaptic integration is suggested by the following lines of evidence: (i) it is widely accepted that intrinsic, in particular dendritic, ion conductances play a pivotal role in the local integration and propagation of synaptic signals (reviewed in Johnston et al. 1996), and (ii) cholinergic afferents, owing to their diffuse projection, are likely to modulate ion channels in all neuronal compartments, including the apical dendrite (Frotscher & Leranth, 1985). In the present study, we used whole-cell recordings from CA1 pyramidal cells of the hippocampal slice preparation to investigate how the activation of somatodendritic ACh receptors affects electrically evoked EPSPs and EPSP-like waveforms elicited by short glutamate pulses. We report here the unexpected finding, that stimulation of postsynaptic ACh receptors, presumably of the M2 receptor subtype, leads to an appreciable reduction of AMPA EPSPs in CA1 pyramidal cells, resulting from the cholinergic activation of an inwardly rectifying K⁺ conductance.

METHODS

Using standard procedures, transverse hippocampal slices 300 μm thick, were prepared from the brain of Wistar rats (2-3 weeks old), which were deeply anaesthetized with ether prior to decapitation. All experiments were carried out according to the guidelines and with the approval of the Animal Care Committee at the University of Munich. After dissection, slices were incubated in warmed (30 °C) artificial cerebrospinal fluid (ACSF) for 30 min and then maintained at room temperature (21-24 °C) in the same solution. ACSF was constantly gassed with 95 % O₂-5 % CO₂ and had the following composition (mm): NaCl, 125; KCl, 3; CaCl₂, 2; MgCl₂, 2; NaH₂PO₄, 1.25; NaHCO₃, 25; d-glucose, 10 (pH 7.4). For electrophysiological measurements, individual slices were transferred to the recording chamber that was mounted on the stage of an upright microscope (Zeiss Axioskop). Dodt infrared gradient contrast in conjunction with a contrast-enhanced CCD camera (Hamamatsu) served to identify somata and dendritic processes of pyramidal cells in the hippocampal CA1 region. During experiments, slices were kept submerged in ACSF that was constantly exchanged by means of a gravity-driven superfusion system (flow rate 2-3 ml min⁻¹). All experiments were performed at room temperature. Drugs were either added to the bathing medium, or pressure ejected through the broken tip (diameter 1-3 μm) of a glass micropipette which was positioned under visual control at different locations along the somatodendritic axis of the patch-clamped neuron (Fig. 1). Pressure was set to 1-4 p.s.i. (7-28 kPa) and pressure pulses lasting 7 ms were delivered at 0.2 Hz for periods of 5-10 min. Locally applied drugs were dissolved in modified ACSF in which bicarbonate was replaced with Hepes/NaHepes (25 mm) to avoid pH changes in the pipette solution. Control experiments (n = 3) ruled out that the focal application of drug-free pipette solution altered resting membrane potential or EPSP shape (cf. Fig. 2, inset). EPSPs were evoked by means of concentric bipolar stainless steel electrodes placed in outer stratum radiatum to stimulate afferent fibres of the Schaffer collateral/commissural pathway exciting predominantly the most distal portion of the apical dendrite of CA1 pyramidal neurons. Constant current pulses of 50-500 μA (pulse width 0.1 ms) were delivered at 0.2 Hz. Illustrated EPSPs are means of five consecutive sweeps. The GABA_A receptor antagonist, bicuculline (10 μm), and the NMDA receptor antagonist, d-APV (20 μm), were routinely added to the bathing solution to obtain AMPA receptor-mediated EPSPs. In addition to electrical stimulation, EPSP-like waveforms were evoked using short pulses of glutamate (1 mm), which was focally applied onto the apical dendrite of CA1 pyramidal neurons (∼150 μm from the soma) by means of a micropipette. Pressure was set to 4 p.s.i. (28 kPa) and pulses 7 ms long were delivered at 0.2 Hz. To functionally isolate the recorded neuron from synaptic input arising in neighbouring neurons coactivated by glutamate pulses, these experiments were always conducted in the presence of TTX (0.5-1 μm). Again, the NMDA component was suppressed with d-APV (20 μm), and spontaneous IPSPs were abolished with bicuculline (10 μm). Illustrated EPSP and GPSP traces are averages of five consecutive sweeps, refiltered at 0.5 kHz by means of a Gaussian filter. To evoke miniature EPSPs (mEPSPs), TTX (1 μm) was added to the bathing solution and an external solution of high osmolarity was pressure applied locally onto a visually identified spot of the apical dendrite at a distance of about 150 μm from the soma. The solution of high osmolarity consisted of modified ACSF (25 mm Hepes/NaHepes instead of bicarbonate) with the addition of 300 mm sucrose and d-APV, bicuculline and TTX at the same concentrations as they were in the bath solution. Electrophysiological signals obtained in the whole-cell configuration of the patch-clamp technique were recorded, amplified and analysed with the use of an Axopatch 200A amplifier (Axon Instruments, CA, USA) in conjunction with a Digidata 1200 interface and pCLAMP 6 software (Axon Instruments). Signals were low-pass filtered at 1 kHz and digitized at 3-5 kHz, with the exception of mEPSPs, which were sampled at 10 kHz (low-pass filter 5 kHz). Recording pipettes were filled with (mm): potassium gluconate, 140; MgCl₂, 3; EGTA, 5; Hepes, 5; Na₂-ATP, 2; Na-GTP, 2 (pH 7.25-7.30); and had a resistance of about 5 MΩ. Miniature EPSPs were individually inspected and analysed using Origin 6.0 software. Events were binned at 0.1 mV for amplitude distribution and fitted to a Gaussian curve. Voltage-clamp experiments were performed in the presence of TTX (1 μm). For voltage-clamp recordings in high extracellular K⁺ solution, ACSF was modified as follows (mm): NaCl, 100; KCl, 20; MgCl₂, 3; NaH₂PO₄, 1.25; NaHCO₃, 25; EGTA, 5; TEA-Cl, 5; 4-amino-pyridine, 5; d-glucose, 10; (pH 7.4). In the whole-cell configuration, series resistance was about 12 MΩ, which was, in voltage-clamp experiments, compensated by 75-85 %. Voltage readings were corrected for liquid junction potentials (10 mV). Data are expressed as means ±s.e.m.. Statistical comparison of data was performed using Student's t test.

Figure 1. Recording situation in the CA1 region of the hippocampal slice.

Schematic drawing illustrates positioning of pipettes for somatic whole-cell recording and for local dendritic substance application in relation to afferent fibres that were electrically stimulated to elicit EPSPs. Neurons were visualized using Dodt infrared gradient contrast in conjunction with a contrast-enhanced CCD camera. In the bottom right-hand corner of the photograph, a patch pipette for whole-cell recording is attached to the soma of a pyramidal neuron. The apical dendrite of this cell extends towards the upper left-hand corner where a micropipette for local substance application is positioned in close proximity. If EPSPs were evoked by electrical stimulation of remote synaptic input, micropipettes were filled with CCh to focally activate postsynaptic ACh receptors. To elicit EPSP-like waveforms in TTX, micropipettes were filled with glutamate.

Figure 2. Activation of somatodendritic ACh receptors reduces AMPA EPSPs.

In the presence of bicuculline (10 μm) and d-APV (20 μm), electrical stimulation in outer stratum radiatum evoked remote AMPA EPSPs in CA1 pyramidal cells. Arrows indicate time of stimulation (stim.) as represented by stimulus artifacts. CCh (20 μm) was focally applied at different sites along the somatodendritic axis as indicated above traces. Recordings are from different cells. The resting membrane potential was manually held at -70 mV by current injection through the recording pipette as necessary. The inset demonstrates that focal application of CCh-free solution did not affect EPSP waveform.

Tetrodotoxin (TTX) and tertiapin were purchased from Alomone Labs (Jerusalem, Israel), pirenzepine and methoctramine from Tocris (Köln, Germany), and all other substances from Sigma (Deisenhofen, Germany).

RESULTS

Postsynaptic cholinergic inhibition of electrically evoked EPSPs

To examine how the activation of somatodendritic ACh receptors affected AMPA receptor-mediated EPSPs in the postsynaptic neuron, we combined whole-cell recordings from visually identified CA1 pyramidal neurons with local application of carbachol (CCh, 20 μm) and electrical stimulation of afferent fibres in outer stratum radiatum to evoke remote excitatory input (Fig. 1). Stimulation strength was adjusted to obtain subthreshold EPSPs at the soma of approximately 10 mV peak amplitude (9.8 ± 0.6 mV, n = 17) at a resting membrane potential of -70 mV. Under visual guidance, CCh was focally administered onto the soma and different regions of the apical dendrite. As shown in Fig. 2, CCh acting on somatic or dendritic ACh receptors produced an appreciable inhibition of EPSPs. The mean reduction of EPSP peak amplitude was 19.1 ± 2.0 % (n = 8) during somatic CCh application, and 19.8 ± 0.9 %, 20.9 ± 0.8 % and 17.8 ± 1.1 % during dendritic application, at approximately 150 μm (n = 7), 300 μm (n = 3), and 400 μm (n = 4), respectively, from the soma.

In all experiments, somatic resting membrane potential was manually held at -70 mV by means of current injection through the recording pipette. If the membrane potential was allowed to change freely, focal CCh application onto the soma produced a membrane depolarization of 6.5 ± 1.4 mV (n = 4, resting membrane potential in these cells was -65.0 ± 3.3 mV), whereas activation of dendritic ACh receptors at ≥ 150 μm from the soma caused no or only a negligible (≤ 1 mV) deviation of the membrane potential. We next examined a possible voltage dependence of the cholinergic inhibition of EPSPs. The somatic resting membrane potential was varied between -60 and -90 mV using appropriate current injection and remote EPSPs were evoked at each potential. As shown in the sequence of Fig. 3, the efficacy of dendritically applied CCh (20 μm) to inhibit EPSPs increased substantially with membrane hyperpolarization. To determine whether CCh also affected EPSP time course, EPSP waveforms recorded in the absence and presence of CCh were scaled to their peak amplitude and superimposed. As demonstrated by the insets of Fig. 3, CCh did not alter the rising phase, but it modestly accelerated EPSP decay at all potentials.

Figure 3. Cholinergic inhibition of EPSPs in the postsynaptic neuron is voltage dependent.

EPSPs were electrically evoked in outer stratum radiatum. Arrow indicates stimulus artifact. CCh (20 μm) was focally applied to the apical dendrite (∼150 μm from the soma). Resting membrane potential was varied between -60 and -90 mV by means of current injection through the recording pipette. All recordings are from the same neuron. Note increase in cholinergic inhibition with membrane hyperpolarization. To determine whether CCh also affects their kinetics, EPSPs recorded before and during drug application were scaled to peak and superimposed. Insets show that CCh caused a modest acceleration of EPSP decay at all potentials tested.

Carbachol reduces glutamate-evoked EPSP-like waveforms (GPSPs)

Although the voltage dependence of the CCh effect strongly argued against a presynaptic site of action, we repeated these experiments in the presence of TTX using brief focal glutamate pulses delivered to the apical dendrite at a distance of approximately 150 μm from the soma to simulate EPSP-like waveforms. Since presynaptic actions of CCh were irrelevant under these experimental conditions, CCh (20 μm) was added to the bathing solution. Confirming the findings obtained with electrically evoked EPSPs, CCh produced a very similar decrease of glutamate-evoked postsynaptic potentials (GPSPs, Fig. 4A-C). Again, cholinergic inhibition of GPSPs increased as the membrane potential was made more negative. After scaling and superimposition, the decay of GPSPs appeared moderately faster in the presence of CCh (Fig. 4A-C, insets), consistent with its action on EPSP kinetics. To determine the significance of this observation, the decay phase of GPSPs was fitted to a single exponential. A comparison of decay time constants before and during CCh superfusion showed that CCh moderately, but significantly, accelerated EPSP decay at all potentials (Fig. 4D). Bath-applied CCh caused a slow membrane depolarization of 10.6 ± 1.0 mV (n = 18, resting membrane potential in these cells was -66.7 ± 1.4 mV), which was compensated for by hyperpolarizing current injection to maintain equal driving forces for GPSPs. To discriminate between the effects of muscarinic vs. nicotinic receptor activation, we examined cholinergic inhibition of GPSPs in the absence and presence of either the muscarinic receptor antagonist, atropine (1 μm), or the nicotinic receptor antagonist, mecamylamine (10 μm). Only atropine (n = 6) reversed the decrease of GPSPs during CCh superfusion (Fig. 4A-C and Fig. 5C) as well as the slow membrane depolarization (not shown), whereas mecamylamine did not influence either effect of CCh (n = 4, data not shown). The inhibitory, voltage-dependent action of CCh on EPSPs (n = 17) and GPSPs (n = 18) is summarized in Fig. 5.

Figure 4. Cholinergic inhibition of glutamate-evoked postsynaptic potentials (GPSPs).

A-C, in the presence of TTX (1 μm), bicuculline (10 μm) and d-APV (20 μm), short glutamate pulses (arrows) locally delivered onto the apical dendrite (∼150 μm from the soma) served to mimic EPSP-like waveforms at RMPs of -70 mV (A), -80 mV (B) and -100 mV (C). CCh (20 μm) was added to the bathing solution. Closely resembling its action on electrically evoked EPSPs, CCh reduced GPSPs in a voltage-dependent fashion and accelerated their decay (see insets, in which scaled GPSPs were superimposed). The inhibitory action of CCh was reversed by atropine (1 μm). All recordings were obtained from the same neuron. D, decay time constants (τ_d, mean ±s.e.m.) were plotted as a function of the membrane potential at which GPSPs were evoked. CCh produced a significant decrease of decay time constants at all potentials investigated (*P < 0.005, **P < 0.001). Numbers above asterisks indicate number of observations.

Figure 5. Synopsis of CCh effects on EPSP and GPSP amplitudes.

The columns (means ±s.e.m.) of the histograms summarize the voltage-dependent reduction of electrically evoked EPSPs by focally applied CCh (A), the voltage-dependent reduction of GPSPs by bath- applied CCh (B) and the antagonistic action of atropine (C). Asterisks indicate level of statistical significance (*P < 0.05, **P < 0.005, ***P < 0.001).

Pharmacology of muscarinic inhibition of GPSPs

We next used the M1 receptor-preferring antagonist, pirenzepine, and the M2 receptor-preferring antagonist, methoctramine, to discriminate between the muscarinic receptor subtypes mediating the various electro- physiological effects of CCh in CA1 pyramidal cells. In this set of experiments, CCh and the two antagonists were all bath applied at 2 μm to obtain a maximum of antagonist selectivity in the hippocampal slice preparation (Müller & Misgeld, 1989). At this concentration, CCh depolarized the neurons by 5.1 ± 1.4 mV (n = 8), an effect significantly smaller (P < 0.005) than that produced by 20 μm CCh. In contrast, the lower concentration of CCh was as effective as the higher one in decreasing GPSP amplitude (see below). Given the much higher affinity of CCh for M2 than for M1 receptors (Müller & Misgeld, 1989), this was a first hint that the two actions of CCh are mediated by different muscarinic receptor subtypes. In fact, only methoctramine (n = 7), but not pirenzepine (n = 7), was capable of suppressing cholinergic inhibition of GPSPs. In contrast, pirenzepine, but not methoctramine, antagonized the depolarizing action of CCh. Resting membrane potential before CCh application was -66.1 ± 1.6 mV and it returned to -65.4 ± 1.4 mV when pirenzepine was added to the CCh-containing bath solution (n = 7). This is in agreement with previous studies ascribing muscarinic depolarization to M1 receptor activation (Dutar & Nicoll, 1988). This is shown in the experiment of Fig. 6A, in which the membrane potential was free to change. Carbachol (2 μm) depolarized the neuron by 5 mV (red trace). Exactly as if we had injected hyperpolarizing DC, this depolarizing action of CCh was completely blocked by pirenzepine (2 μm, green trace). It is also obvious from this figure that pirenzepine did not abrogate the GPSP decrease. It was only after further addition of methoctramine that the GPSP regained its control amplitude (blue trace). That the cholinergic reduction of GPSPs was sensitive to the M2 receptor antagonist, but not to the M1 receptor antagonist, is also evident from the experiment illustrated in Fig. 6B. Here, DC injection was used to hold the membrane potential at -80 mV. Again, pirenzepine (n = 5) failed to abolish the CCh-induced GPSP decrease, whereas methoctramine (n = 6) produced an almost complete reversion. Given the relatively small discriminative potential of methoctramine between M1 receptors where it has a pA₂ value of 7.1, and M2 receptors, where it has a pA₂ value of 7.8 (Hume et al. 1990), it might be argued that, in the above experiments, the antagonist would abolish responses regardless of whether they are mediated by M1 or M2 receptors. To further substantiate the involvement of M2 receptors, we used gallamine, which acts as a selective allosteric antagonist at M2 receptors (Caulfield & Birdsall, 1998) and has been employed to identify M2 receptor-mediated electrophysiological actions in hippocampal pyramidal cells (Dutar & Nicoll, 1988). As shown in Fig. 6C, gallamine (20 μm) almost completely reversed the effect of CCh (20 μm) on the amplitude of GPSPs (n = 7). In contrast, gallamine did not abrogate the depolarization of the membrane potential associated with CCh application (9.42 ± 1.1 mV, n = 7), suggesting that the antagonist did not suppress M1 receptor-coupled responses. The histogram of Fig. 6D, which summarises the pharmacological profile of the effect of CCh on GPSP amplitude, shows that the pirenzepine insensitivity and the methoctramine/ gallamine sensitivity was a consistent and statistically significant finding.

Figure 6. Pharmacology of muscarinic depression of GPSPs suggests involvement of M2 receptors.

A, in this recording, where the membrane potential was free to change, CCh (2 μm) depolarized the neuron by 5 mV. Addition of the M1 receptor antagonist, pirenzepine (2 μm), reversed the depolarizing action of CCh, but did not affect the CCh-induced GPSP decrease (green trace). Only after further addition of the M2 receptor antagonist, methoctramine (2 μm), the GPSP waveform (blue trace) matched that of control. B, same experiment as in A, except that the membrane potential was manually voltage clamped at -80 mV. Again, the cholinergic GPSP inhibition was pirenzepine insensitive, and methoctramine sensitive. C, same experiment as in B, except that the M2 receptor antagonist gallamine (20 μm) was added to the CCh (20 μm)-containing bath solution. D, the columns of the histogram summarize the effects of pirenzepine, methoctramine and gallamine on the reduction of GPSPs by CCh. Colours of columns correspond to colours of traces in A, B and C. Asterisks indicate level of statistical significance (**P < 0.001).

Suppression of inwardly rectifying K⁺ channels abrogates cholinergic inhibition of GPSPs

Considering the characteristic voltage dependence, we speculated that ACh inhibits EPSPs by augmenting hyperpolarizing inward rectification in the postsynaptic neuron. Two inwardly rectifying currents are present in CA1 pyramidal neurons, the cholinergic activation of which would progressively shunt excitatory postsynaptic currents with membrane hyperpolarization, namely a G protein-activated, inwardly rectifying K⁺ (GIRK) current, and a hyperpolarization-activated cation current (I_h). Since inwardly rectifying K⁺ currents, but not I_h, are selectively blocked by Ba²⁺ at low concentrations, we first examined whether Ba²⁺ (200 μm) would suppress the CCh-induced inhibition of GPSPs. If examined in Ba²⁺-containing solution, CCh (20 μm) consistently failed to reduce the amplitude of GPSPs (relative change to the amplitude in Ba²⁺ alone: -1.3 ± 2.3 %, n = 5, Fig. 7A). Demonstrating the specificity of the blocker, Ba²⁺ did not affect the concomitant depolarization by CCh (data not shown). Whereas low concentrations of Ba²⁺ suppress various types of inwardly rectifying K⁺ channels, tertiapin, a novel peptide inhibitor isolated from the venom of the honey bee, blocks GIRK1 and GIRK4 channels with nanomolar affinity while having no effect on IRK1 inwardly rectifying K⁺ channels, ATP sensitive inwardly rectifying K⁺ channels or voltage-dependent K⁺ channels (Jin & Lu, 1998; Kitamura et al. 2000). As shown in Fig. 7B, tertiapin (30 nm) fully reversed the CCh-induced reduction of GPSP amplitude (to 97.8 ± 1.7 % of control, n = 5), thus strongly implicating muscarinic GIRK channel activation as the underlying mechanism.

Figure 7. Suppression of inwardly rectifying K⁺ channels abrogates the action of CCh on GPSPs.

A, Ba²⁺ (200 μm) prevented the CCh (20 μm)-induced decrease of GPSP amplitude. B, the selective GIRK channel inhibitor, tertiapin (30 nm), reversed the effect of CCh on GPSP amplitude.

Effects of carbachol on miniature EPSPs (mEPSPs)

While the measurement of glutamate-evoked postsynaptic potentials (GPSPs) allowed us to isolate postsynaptic mechanisms contributing to muscarinic depression of excitatory synaptic input, it would be still more convincing to demonstrate directly the postsynaptic effect using synaptically elicited EPSPs. In order to address this issue, we evoked mEPSPs by means of local dendritic application of a solution of high osmolarity in the presence of TTX (cf. Bekkers & Stevens, 1995; Magee & Cook, 2000). Whereas it was virtually impossible to identify mEPSPs in normal bath solution, focal application of a high concentration of sucrose solution to the apical dendrite (see inset in Fig. 8B) resulted in the occurrence of spontaneous mEPSPs (Fig. 8A). If evoked in the presence of bath-applied CCh (20 μm), the mean amplitude of mEPSPs was reduced significantly from 0.87 ± 0.07 to 0.47 ± 0.04 mV (n = 6, P < 0.001, Fig. 8B and C) and the frequency of mEPSPs was reduced to 80.2 ± 4.1 % of control (P < 0.05, Fig. 8E). As illustrated in Fig. 8A-C, the depressant effect of CCh on the amplitude of mEPSPs was reversed by the allosteric M2 receptor antagonist, gallamine (20 μm, mEPSP amplitude 0.80 ± 0.06 mV, n = 6), consistent with the results obtained with GPSP recordings. Gallamine also reversed the decrease in the frequency of mEPSPs (data not shown). That the observed inhibition of mEPSPs was indeed a postsynaptic phenomenon was confirmed in experiments where we recorded mEPSPs under conditions that selectively eliminated the postsynaptic effect of CCh in the recorded cell. As shown in Fig. 8D, CCh failed to affect the mean amplitude of mEPSPs in cells dialysed with Cs⁺ and GDPβ S (0.5 mm) instead of K⁺ and GTP, respectively. In contrast, CCh was still capable of reducing the frequency of mEPSPs at the presynaptic side (Fig. 8E).

Figure 8. Effect of CCh on miniature EPSPs.

A, miniature EPSPs (mEPSPs) were virtually absent in normal bath solution containing TTX (1 μm, top trace). Local dendritic application of a solution of high osmolarity gave rise to spontaneous mEPSPs (control trace), the amplitudes of which were reduced in the presence of CCh (20 μm), but recovered after further addition of the M2 receptor antagonist gallamine (20 μm, bottom trace). Resting membrane potential was held at -70 mV. B, histograms show amplitude distribution of mEPSPs under the different recording conditions as indicated. Schematic drawing (inset) illustrates recording situation. C, histogram summarizes effects of CCh and gallamine (applied in the presence of CCh) on mean amplitude of mEPSPs. D, CCh (20 μm) failed to reduce mEPSPs when neurons were dialysed with Cs⁺ and GDPβS instead of K⁺ and GTP, respectively (n = 5). E, the relative reduction of the frequency of mEPSPs by CCh was independent of the pipette solution used (n = 6 for normal pipette solution, n = 5 for Cs⁺/GDPβS-containing pipette solution).

Carbachol activates a GIRK conductance, presumably through M2 receptors

To demonstrate cholinergic activation of a GIRK conductance in CA1 pyramidal neurons directly, pyramidal cells were voltage clamped and current responses to hyperpolarizing voltage ramps were recorded in the absence and presence of CCh. Extracellular K⁺ ([K⁺]_o) was elevated to 20 mm to enhance current flow through GIRK channels. Na⁺ and Ca²⁺ channels were pharmacologically eliminated and TEA and 4-AP were added (see Methods) to reduce cholinergic effects on other ion channels. Before CCh application, the current-voltage (I-V) relationship of CA1 pyramidal neurons typically displayed constitutive inward rectification in the hyperpolarizing direction. Addition of CCh (20 μm) to the bathing solution substantially enhanced inward rectification (Fig. 9A). Subtraction of the two current traces yielded the I-V relationship of the CCh-induced current (Fig. 9A, inset), which displayed strong inward rectification and reversed polarity near the calculated equilibrium potential for potassium (E_K) (-49 mV) as expected for a GIRK conductance. To quantify the action of CCh, we calculated the slope conductance of the I-V relationship of the neurons between -170 and -130 mV. In 20 mm[K⁺]_o, control slope conductance was 18.9 ± 1.4 nS, whereas CCh increased slope conductance to 29.4 ± 2.3 nS (P < 0.001, n = 8). In 3 mm[K⁺]_o, slope conductance was 9.8 ± 2.1 nS in control solution and 14.6 ± 3.0 nS in the presence of CCh (P < 0.005, n = 5). The relative increase in slope conductance by CCh was 49 % in physiological K⁺, and 56 % in high K⁺ solution.

Figure 9. CCh induced GIRK current in voltage-clamped CA1 pyramidal neurons.

A, in 20 mm[K⁺]_o, hyperpolarizing voltage ramps (0.1 V s⁻¹) evoked inwardly rectifying current responses that were reversibly augmented by CCh (20 μm). Subtraction of the control trace from the trace recorded in the presence of CCh yielded the I-V relationship of the CCh-induced current (I_CCh, inset). B, cholinergic potentiation of hyperpolarizing inward rectification was reversed after addition of tertiapin (30 nm) to a CCh-containing solution. Green area indicates tertiapin-sensitive shift of the I-V relationship. [K⁺]_o was 3 mm. Inset demonstrates the lack of effect of CCh (20 μm) on the I-V relationship, if recorded in neurons dialysed with high Cs⁺ and GDPβS. C, to detect a modulatory action of CCh (20 μm) on I_h, [K⁺]_o was elevated to 20 mm, and square-voltage commands to test potentials from -60 to -160 mV were applied. Note slow activation of I_h in the absence (thin traces, ▪) and presence of CCh (thick traces, •; Ca). The I-V curves shown below were obtained by plotting the current amplitude at the end of the pulse as a function of the test potential (Cb). Subtraction of the two curves showed that CCh induced a very small current with inwardly rectifying properties, compatible with a subtle upregulation of I_h. In normal ACSF, however, no significant action of CCh on I_h was observed (see text). D, pharmacology of the effect of CCh on the I-V relationship of the neuron. The M2 receptor antagonist gallamine (20 μm) reversed the increase in hyperpolarizing inward rectification resulting from GIRK current activation by CCh (20 μm). Green area indicates gallamine-sensitive shift of the I-V curve. In contrast, the M1 receptor-preferring antagonist, pirenzepine (2 μm, red trace of inset), blocked the parallel, rightward shift of the I-V curve by CCh (2 μm) between -70 and -30 mV, but did not affect the tertiapin- and gallamine-sensitive muscarinic increase in hyperpolarizing inward rectification.

To confirm that the increase in hyperpolarizing inward rectification was due to the activation of a GIRK conductance, we again employed the selective GIRK channel inhibitor, tertiapin (30 nm). As shown in Fig. 9B, tertiapin reversed the non-linear increase in inward rectification by CCh (green area, n = 6). Consistent with the lack of effect of CCh on mEPSPs after intracellular suppression of postsynaptic muscarinic responses (Fig. 8D), CCh did not affect hyperpolarizing inward rectification in voltage-clamp experiments performed with pipettes containing Cs⁺ and GDPβ S instead of K⁺ and GTP, respectively (n = 4, Fig. 9B, inset). Although these findings indicated that GIRK channels are the predominant target of cholinergic modulation, they do not entirely rule out a small contribution of I_h. Thus we performed additional experiments to detect a possible cholinergic modulation of I_h under conditions where CCh did not produce concomitant activation of GIRK channels. For this purpose, we added Ba²⁺ (200 μm) to the bathing solution, which suppressed the cholinergic activation of GIRK current, as demonstrated in independent experiments (n = 3, data not shown). Again, [K⁺]_o was enhanced to 20 mm to facilitate inward current flow through I_h channels (Pape, 1996). As shown in Fig. 9C, CCh (20 μm) produced only a very small effect on hyperpolarizing inward rectification. If repeated with 3 mm[K⁺]_o, the CCh-induced increase in slope conductance was 1.9 ± 0.3 % (n = 6, P > 0.1) on average, indicating that CCh exerted a negligible action on I_h.

To link muscarinic GIRK current activation to the observed reduction of postsynaptic potentials, it is essential to demonstrate that the two effects have the same pharmacological profile. The I-V curves of Fig. 9D show that this is indeed the case. Mimicking the action of tertiapin on the I-V relationship of the neuron, gallamine (20 μm) reversed the CCh-induced increase in hyperpolarizing inward rectification (n = 5). The same antagonistic effect was observed when we applied methoctramine instead of gallamine (n = 7, data not shown). In contrast, when pirenzepine was added to the CCh-containing bath solution, a complementary antagonistic effect was observed: the CCh-induced parallel rightward shift of the I-V curve between -30 mV and approximately -70 mV was abrogated, whereas, with more negative potentials, the CCh effect proved increasingly insensitive to pirenzepine (Fig. 9D, inset). From these experiments, it appears safe to conclude that CCh produces a pirenzepine-insensitive, gallamine-sensitive enhancement of hyperpolarizing inward rectification. Based on the inhibitory effect of tertiapin and its pharmacology, we attribute this cholinergic effect to the activation of GIRK channels, presumably mediated by M2 receptor activation.

DISCUSSION

This study demonstrates a novel inhibitory action of ACh in hippocampal CA1 pyramidal cells. We found that, in addition to its multiple excitatory actions, ACh also potentiates hyperpolarizing inward rectification via activation of a GIRK current. While this cholinergic action is not evident from somatic membrane potential recordings, in which the depolarizing influence of ACh prevails, experiments using electrically evoked or simulated EPSPs suggest that muscarinic GIRK current activation bears particular significance on the processing of excitatory synaptic input. Since EPSP-like waveforms evoked by focal dendritic glutamate application (GPSPs) were inhibited by CCh in a virtually identical fashion, it is safe to conclude that this action of CCh occurs on the postsynaptic side. Furthermore, analysis of mEPSPs in the absence and presence of CCh allowed us to identify a clear postsynaptic contribution to the depression of synaptically evoked EPSPs. Our evidence that muscarinic upregulation of GIRK current is responsible for the postsynaptic inhibition of EPSPs is the following. (1) The decrease of EPSPs and GPSPs during application of the cholinergic agonist CCh was voltage dependent, with the amplitude of the postsynaptic potentials progressively declining with membrane hyperpolarization. (2) The CCh-evoked GPSP reduction was reversed by Ba²⁺ at a concentration which selectively suppresses inwardly rectifying K⁺ channels, and more specifically, by the selective GIRK-channel inhibitor tertiapin at nanomolar concentrations. (3) Voltage-clamp recordings directly demonstrated the existence of a CCh-activated, tertiapin-sensitive GIRK current in CA1 neurons. (4) Both cholinergic activation of GIRK conductance and inhibition of excitatory synaptic input displayed the same pharmacological features.

Activation of GIRK current is a hallmark of muscarinic receptor activation in cardiac pacemaker cells leading to a slowing of heart beat (Yamada et al. 1998). In central nervous system neurons, evidence for a similar muscarinic action has been reported from thalamic reticular neurons and from interneurons of striatum, neocortex and hippocampus, with some studies suggesting the involvement of M2 receptors (Egan & North, 1986; McCormick & Prince 1986b; Calabresi et al. 1998; Xiang et al. 1998; Yamada et al. 1998; McQuiston & Madison, 1999). Owing to the lack of very high subtype selectivity of any single muscarinic receptor antagonist, one cannot unambiguously attribute an electrophysiological action of ACh to a particular receptor subtype, unless the experimental design involves determination of (apparent) affinity constants (Caulfield & Birdsall, 1998). Although such data are not available for the postsynaptic reduction of EPSPs reported here, preliminary evidence favours the participation of M2 receptors. Both the decrease of EPSPs and the augmentation of inward rectification were insensitive to pirenzepine at a concentration which suppressed membrane depolarization, an effect generally ascribed to M1 receptor activation. Vice versa, the selective allosteric M2 receptor antagonist gallamine abrogated the former effects of carbachol at a concentration which did not affect the latter one. Finally, it is widely accepted that muscarinic activation of GIRK conductance involves binding of agonists to M2 receptors, or to M4 receptors, which share similar G protein-coupling properties, but are difficult to discriminate by classical pharmacological techniques (Wess, 1996; Yamada et al. 1998).

How can we reconcile the hypothesis that muscarinic GIRK current activation partially shunts excitatory synaptic current leading to a postsynaptic reduction of EPSPs with the fact that the ACh-induced GIRK conductance fails to counterbalance cholinergic depolarization of the somatic membrane potential? First of all, a somatic recording pipette is unlikely to reflect accurately the voltage changes occurring at, or near, the site of synaptic stimulation on the apical dendrite (Spruston et al. 1994). It is not known whether the depolarizing response to ACh is uniformly distributed or whether it displays some degree of compartmentalization. In patch recordings from apical dendrites of CA1 pyramidal neurons, bath-applied CCh depolarized the dendritic membrane potential by a few millivolts (Tsubokawa & Ross, 1997), but it is not clear whether this depolarization was genuinely produced in the dendrite, or whether it rather reflected the electrotonic spread of a somatically generated depolarization. In our hands, local CCh application to dendritic sites approximately 150 μm away from the soma failed to influence the somatically recorded resting potential. Although the directional asymmetry of steady-state voltage attenuation in a dendritic tree (somatopetal > somatofugal) has to be taken into account (Rall et al. 1992), the absence of a somatic voltage deviation during focal dendritic CCh administration points to a relatively weak (or maybe absent) depolarizing response in the dendrite, possibly reflecting a decline in M1 receptor density along the somatodendritic axis. While such a compartmentalization of cholinergic actions has to remain speculative for the moment, several lines of evidence argue in favour of a highly localized, and thus highly efficient, interaction between ACh-induced GIRK current and excitatory synaptic currents in the dendrite. (i) In electron microscopic studies of cholinergic projections to hippocampal pyramidal neurons, ChAT-immunoreactive terminals were found to form synaptic contacts with dendritic shafts and spines (Frotscher & Leranth, 1985; Heimrich & Frotscher, 1993). (ii) Electron microscopy revealed prominent GIRK1 channel subunit immunoreactivity immediately adjacent to asymmetric (excitatory) postsynaptic densities in spines of apical dendrites of CA1 pyramidal neurons (Drake et al. 1997). (iii) The functional significance of the latter finding is underscored by patch-clamp recordings demonstrating stronger GIRK current responses to several agonists in isolated apical dendritic segments compared with isolated somata of pyramidal neurons (Takigawa & Alzheimer, 1999). We thus propose that, owing to their presumed location on dendritic spines and shafts, ACh-induced GIRK currents will dampen EPSPs near the site of their generation as well as during their propagation to the soma.

Although GIRK current activation is certainly the primary mechanism underlying muscarinic inhibition of EPSPs, the subtle, but consistent, acceleration of EPSP decay during CCh application would also be compatible with a small muscarinic enhancement of I_h. Synaptic depolarization deactivates a standing I_h, giving rise to an effectively outward current that causes faster EPSP decay and produces a transient after-hyperpolarization (Pape, 1996; Magee, 1998). Vice versa, pharmacological suppression of I_h slows EPSP decay (Fig. 7B; Stuart & Spruston, 1998; Magee, 1999). While a previous study reported cholinergic potentiation of I_h (Colino & Halliwell, 1993), we did not obtain evidence for a robust modulatory action of ACh under our recording conditions, suggesting that the cholinergic effects on EPSP amplitude and kinetics are almost exclusively mediated by GIRK current activation.

A salient feature of cholinergic activity in the hippocampal formation, which has been implicated in the formation of heteroassociative memory (Hasselmo & Schnell, 1994), is the inhibition of excitatory transmission at the Schaffer afferent-CA1 synapse. This was generally believed to represent a purely presynaptic phenomenon (see Introduction). However, after isolation of the postsynaptic contribution to the depression of EPSPs by ACh, it becomes evident that cholinergic effects generated in somatodendritic compartments of the postsynaptic neuron do play a significant role. We thus propose that the postsynaptic mechanism identified in this study augments the presynaptic action of ACh and assures that it is not counteracted by the multiple excitatory effects on the postsynaptic side. It is noteworthy that, because of its intrinsic voltage dependence, the postsynaptic component of cholinergic EPSP inhibition is unlikely to operate in a purely parallel fashion with the presynaptic side, but might also serve a function of its own. The biophysical features of GIRK current predict that the extent of postsynaptic EPSP reduction critically depends on the membrane potential when synaptic input arrives. EPSP reduction should be most effective for small excitatory input reaching a resting or hyperpolarized cellular compartment. When strong and repetitive excitatory synaptic input moves the neuron towards more depolarized voltages, the influence of GIRK current will rapidly disappear. Even more, deactivation of GIRK current with strong depolarizing potentials might even allow supralinear EPSP summation, as recently demonstrated in an invertebrate preparation (Wessel et al. 1999). In functional terms, muscarinic activation of GIRK currents might be thus seen as a mechanism to filter small, less relevant input.