Cholinergic neurotransmission in the medial prefrontal cortex (mPFC) is critical for normal processing of cue detection and cognitive performance. However, the mechanism by which cholinergic system modifies mPFC synaptic function remains unclear. Here we show that activation of muscarinic acetylcholine receptors (mAChRs) by carbamoylcholine (CCh) induces long-term depression (CCh-LTD) of excitatory synaptic transmission on mPFC layer V pyramidal neurons. The induction of CCh-LTD is dependent on M1 mAChR activation but does not require N-methyl-D-aspartate receptor activation or coincident synaptic stimulation. Activation of phospholipase C (PLC), protein kinase C (PKC), and postsynaptic Ca2+ release from inositol 1,4,5-triphosphate (IP3) receptor–sensitive internal stores are required for CCh-LTD induction. The expression of CCh-LTD is likely to be presynaptic because it is accompanied by a decrease in 1/(coefficient of variance)2 and an increase in synaptic failure and paired-pulse ratio of synaptic responses. CCh-LTD is blocked by nitric oxide (NO) synthase inhibitors, soluble guanylyl cyclase (sGC) inhibitor, and protein kinase G (PKG) inhibitor. Synaptic stimulation of M1 mAChRs with prolonged paired-pulse low-frequency stimulation also triggers LTD. These results suggest that activation of M1 mAChRs can induce LTD on mPFC layer V pyramidal neurons through the activation of postsynaptic PLC/PKC/IP3 receptor- and subsequently presynaptic NO/sGC/PKG-dependent signaling processes.
The cholinergic system is essential for a wide variety of higher cognitive functions (Everitt and Robbins 1997; Sarter and Bruno 1997) and gates the processing of sensory information (Ma and Suga 2005). The medial prefrontal cortex (mPFC) receives extensive cholinergic innervation originating from the nucleus basalis magnocellularis, the diagonal band, and the mesopontine laterodorsal nucleus (Lehmann et al. 1980; Satoh and Fibiger 1986; Gaykema et al. 1990). In addition to the afferent inputs, acetylcholine (ACh) is also released from local circuit neurons in the mPFC (Houser et al. 1985). In vivo microdialysis studies have indicated that ACh release is increased in the mPFC during the performance of attentional tasks (Himmelheber et al. 2000; Passetti et al. 2000). Lesions of cholinergic inputs to the mPFC impair cue detection (Parikh et al. 2007) and attentional performance (McGaughy et al. 2002). In contrast, increases in mPFC cholinergic transmission enhance behavioral arousal and locomotor activity (Day et al. 1991). However, it is not clearly known how mPFC cholinergic system controls cognitive functions at the synaptic levels.
Long-term depression (LTD) is a persistent activity-dependent decrease of synaptic strength that together with the converse process, long-term potentiation, has been considered to be crucial for information storage and adaptation to external stimuli (Malenka and Bear 2004). Although LTD is a widespread phenomenon expressed in various brain regions, much of our understanding of the properties and functional relevance comes from studies on the hippocampus. In the hippocampal CA1 region, it is generally accepted that the induction of LTD by repetitive synaptic stimulation at 0.5–5 Hz requires the activation of N-methyl-D-aspartate (NMDA) receptors, a rise in postsynaptic intracellular Ca2+, and the consequential activation of serine–threonine protein phosphatase cascades (Dudek and Bear 1992; Mulkey and Malenka 1992). In the mPFC, LTD has been induced by 2 different induction protocols, each of which involves different mechanisms. The first form of mPFC LTD is induced by high-frequency tetanic stimuli, consisting of four 2-s trains of stimuli separated by an intertrain interval of 10 s at 50 Hz, in the presence of dopamine, and that this form of the LTD induction depends on the concurrent synaptic activation of groups I and II metabotropic glutamate receptors (mGluRs; Otani et al. 1998, 1999). The second form of mPFC LTD is induced by pharmacological activation of group II mGluRs with the agonist (2S, 2′R, 3′R)-2-(2′,3′-dicarboxycyclopropyl) glycine, and its induction relies on the postsynaptic protein kinase C (PKC) activation and inositol 1,4,5-triphosphate (IP3) receptor–mediated postsynaptic increase of Ca2+ concentration (Otani et al. 2002; Huang and Hsu 2008). In this study, we have demonstrated a novel form of LTD that is induced by pharmacological activation or synaptic stimulation of the M1 muscarinic ACh receptors (mAChRs) on mPFC layer V pyramidal neurons. Our results show that this novel form of mPFC LTD is induced postsynaptically but expressed presynaptically. Furthermore, we found that postsynaptic release of nitric oxide (NO) may act as a retrograde messenger on presynaptic site to induce persistent synaptic depression via a soluble guanylyl cyclase (sGC)/protein kinase G (PKG)–dependent mechanism.
Materials and Methods
All experimental procedures were conducted in adherence to the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of National Cheng Kung University. The coronal slices containing the prelimbic area (2.5–3.5 mm anterior to bregma) were prepared from 14- to 30-day-old male Sprague–Dawley rats for whole-cell patch-clamp recordings by procedures described previously (Huang and Hsu 2006; Huang et al. 2007). Most experiments (unless otherwise noted) were performed on 21- to 23-day-old rats. Briefly, rats were deeply anesthetized with halothane and decapitated with guillotine, and coronal slices (250 μm) were prepared using a vibrating microtome (Leica VT1000S or VT1200S; Leica, Nussloch, Germany). The slices were placed in a holding chamber of artificial cerebrospinal fluid (aCSF) oxygenated with 95% O2–5% CO2 and kept at room temperature for at least 1 h before recording. The aCSF solution was composed of the following (in mM): 117 NaCl, 4.7 KCl, 2.5 CaCl2, 1.2 MgCl2, 25 NaHCO3, 1.2 NaH2PO4, and 11 glucose at pH 7.3–7.4.
For whole-cell patch-clamp recording, a single slice was transferred to a submerged recording chamber and fixed at the glass bottom of the chamber with a nylon grid on a platinum wire frame. The chamber consisted of a circular well of low volume (1–2 mL) and was perfused constantly at 32.0 ± 0.5 °C at a rate of 2–3 mL/min. Visualized whole-cell patch-clamp recording of synaptically evoked excitatory postsynaptic currents (EPSCs) was conducted using standard methods as described previously (Huang and Hsu 2006; Huang et al. 2007). The layer V pyramidal neurons were identified by their pyramidal shape, presence of a prominent apical dendrites, and distance from the pial surface with an upright microscope (Olympus BX50WI; Olympus, Tokyo, Japan) equipped with a water immersion ×40 objective and a Nomarski condenser combined with infrared videomicroscopy. Patch pipettes were pulled from borosilicate capillary tubing and heat polished. The electrode resistance was typically 3–6 MΩ. The composition of intracellular solution was (mM) 115 K-gluconate, 20 KCl, 10 N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, 2 MgCl2, 0.5 ethyleneglycol-bis(aminoethylether)-tetraacetic acid, 3 Na2ATP, 0.3 Na3GTP, 5 QX-314, and sucrose to bring osmolarity to 290–295 mOsm and pH to 7.3. After a high-resistance seal (>2 GΩ before breaking into whole-cell mode) was obtained, suction was applied lightly through the pipette to break through the membrane. The cell was then maintained at −70 mV for several minutes to allow diffusion of the internal solution into the cell body and dendrites. Recordings were made using a patch-clamp amplifier (Axopatch 200B; Axon Instruments, Union City, CA). Electrical signals were low-pass filtered at 2 kHz and digitized at 10 kHz using a 12-bit analog-to-digital converter (Digidata 1320; Axon Instruments). An Intel Pentium–based computer with pCLAMP software (Version 8.0; Axon Instruments) was used for online acquisition and off-line analysis of the data. For measurement of synaptically evoked EPSCs, a bipolar stainless steel stimulating electrode was placed in layer V, 40–60 μm away the apical dendrites of the recorded neurons to stimulate excitatory afferents at 0.05 Hz and the superfusate routinely contained bicuculline methiodide (1 μM) to block inhibitory synaptic responses. Because EPSCs were almost completely blocked by α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainate receptor antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 20 μM) plus NMDA receptor antagonist D-2-amino-5-phosphonovalerate (D-APV, 50 μM; Supplementary Fig. S1), they were predominantly mediated by ionotropic glutamate receptors (Huang and Hsu 2006). The strength of synaptic transmission was mostly quantified by measuring the initial rising slope of EPSCs (2-ms period from its onset; pA/ms), which contains only a monosynaptic component. For recording inhibitory postsynaptic currents (IPSCs), potassium gluconate was replaced with KCl in the intracellular solution and CNQX (20 μM) and D-APV (50 μM) were added to the bath. In some experiments, LTD was induced by 1200 paired pulses of low-frequency stimulation (LFS) at 1 Hz with a 50-ms interstimulus interval in the presence of D-APV (50 μM). Series resistance and input resistance were monitored online throughout the whole-cell recording with a 5-mV depolarizing step given after every afferent stimulus and data were discarded if resistance changed by more than 20%. Only cells with a stable resting membrane potential at more negative than −60 mV were used for experiments. Data were not corrected for junction potential.
For extracellular recordings, a single slice was placed in a submersion recording chamber, maintained at 32.0 ± 0.5 °C, and continually perfused with aCSF solution at a rate of 2–3 mL/min. Extracellular electrodes (filled with 1 M NaCl, 2–3 MΩ resistance) were placed on layer V of the mPFC to monitor field excitatory postsynaptic potentials (fEPSPs) evoked with a bipolar stainless steel stimulating electrode placed on layers II–III, where the input fibers are located. The stimulation strength was set to elicit a response having amplitude that was 50–60% of the maximum spike-free response. The superfusate routinely contained bicuculline methiodide (10 μM) to block inhibitory synaptic responses and D-APV (50 μM) to block NMDA receptor–mediated synaptic responses. An Axoclamp 2B amplifier (Axon Instruments) was used for recordings. The responses were low-pass filtered at 2 kHz, digitally sampled at 5–10 kHz, and analyzed using pCLAMP software (Version 7.0; Axon Instruments). The strength of synaptic transmission was quantified by measuring the initial slope of fEPSPs, which contains only a monosynaptic component (Huang et al. 2004). LTD was induced by 1200 paired pulses of LFS at 1 Hz with a 50-ms interstimulus interval.
Drugs were applied by manually switching the superfusate and were diluted from stock solutions just before application. Drug concentrations were selected on the basis of previously published studies or our preliminary results. U73122, U73343, bisindolylmaleimide I (Bis-I), bisindolylmaleimide V (Bis-V), thapsigargin, ryanodine, nimodipine, U0126, N-piperidino-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methylpyrazole-3-carboxamide (SR141716A), 2-(4-Carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (Carboxy-PTIO), 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ), and KT5823 were dissolved in dimethyl sulfoxide (DMSO) stock solutions and stored at −20 °C until the day of experiment. Other drugs used in this study were dissolved in the distilled water. The concentration of DMSO in the perfusion medium was 0.1%, which alone had no effect on the excitatory glutamatergic transmission (Huang and Hsu 2006). Methyllycaconitine (MLA), D-APV, Bis-I, (S)-α-methyl-4-carboxyphenylglycine (MCPG), U73122, thapsigargin, ryanodine, NG-nitro-L-arginine methyl ester (L-NAME), Nω-propyl-L-arginine (NPA), ODQ, Carboxy-PITO, S-nitroso-N-acetyloenicillamine (SNAP), 8-bromoguanosine cyclic 3′,5′-monophosphate (8-Br-cGMP), D15, CNQX, bicuculline methiodide, and U0126 were purchased from Tocris Cookson (Bristol, UK); carbamoylcholine (CCh), eserine, pirenzepine, nimodipine, GDPβS, heparin, kynurenic acid, atropine, and bicuculline methiodide were obtained from Sigma (St Louis, MO); and U73343, Bis-V, 1,2-bis(o-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid (BAPTA), 4-amino-5-(4-cholophenyl)-7-(t-butyl)pyrazolo-[3,4-d]pyrimidine (PP2), PKG inhibitor (PKGI), pep2-SVKE, and pep2-SVKI were purchased from Calbiochem (La Jolla, CA). SR141716A was gift from Sanofi Recherche (Montpellier, France).
The data for each experiment were normalized relative to the baseline and are presented as mean ± standard error of the mean. Numbers of experiments are indicated by n. The statistic significance was evaluated by the 2-tailed unpaired Student's t-test or the 1-way analysis of variance followed by the Newman–Keuls multiple-comparison post hoc test. Probability values (P) of less than 0.05 were considered to represent significant differences.
Activation of M1 mAChRs Induces LTD of Excitatory Synaptic Transmission on mPFC Layer V Pyramidal Neurons
To examine the role of cholinergic activation on glutamatergic transmission in the mPFC, we performed whole-cell recordings from layer V pyramidal neurons in acutely prepared mPFC slices. Bath application of a broad-spectrum cholinergic agonist CCh for 10 min induced a dose-dependent reduction of EPSCs (Fig. 1A,B). At a concentration of 50 μM, the slope of EPSCs did not fully recover after washout of CCh. Instead, the synaptic response remained at a depressed level (64.5 ± 5.2% of pre-CCh baseline, n = 12; measured 30 min after CCh washout; Fig. 1B). In all subsequently studies, CCh (50 μM, 10 min) was used to induce what we will refer to as CCh-LTD. This LTD was completely blocked by a nonselective mAChR antagonist atropine (1 μM) and a selective M1 mAChR antagonist pirenzepine (100 nM) but not by nicotinic AChR antagonist MLA (100 nM), suggesting that it is specifically mediated by M1 mAChRs (Fig. 1C,D). To determine whether the magnitude of CCh-LTD changes with age, we applied CCh to mPFC slices from rats at different postnatal ages. Our results show no significant age differences in the magnitude of CCh-LTD among groups (Fig. 1E). On average, the magnitude of CCh-LTD was 42.6 ± 5.8% (n = 8) at postnatal day (P) 14–16, 35.5 ± 5.2% (n = 12) at P21–23, and 32.8 ± 4.9% (n = 7) at P28–30, respectively, measured 30 min after CCh washout. We also examined the effect of CCh on GABAergic transmission on mPFC layer V pyramidal neurons. Monosynaptic IPSCs were evoked while holding neurons in voltage clamp at −70 mV in the presence of CNQX (20 μM) and D-APV (50 μM). Bath application of CCh (50 μM) for 10 min induced a transient depression of IPSCs (28.5 ± 5.3%, n = 4; P < 0.05), which completely recovered to baseline value following washout of CCh (96.5 ± 3.6% of baseline, n = 4; Supplementary Fig. S2). These results suggest that CCh-LTD selectively expresses on excitatory synaptic transmission in mPFC layer V pyramidal neurons.
Role of Ionotropic Glutamate Receptor and mGluR in the Induction of CCh-LTD
We next determined whether the coincident activation of ionotropic glutamate receptor and mGluR is required for the induction of CCh-LTD in mPFC layer V pyramidal neurons. To assess the involvement of NMDA receptor activation in the CCh-LTD induction, CCh-LTD was attempted in the presence of NMDA receptor antagonist D-APV (50 μM). Normal levels of CCh-LTD were observed in the presence of D-APV (34.7 ± 4.5%, n = 6; P > 0.05 when compared with CCh alone group; Fig. 2A). We next examined whether CCh-LTD occurs in the absence of synaptic stimulation during CCh application. In these experiments, after a stable baseline EPSC was achieved, synaptic stimulation was turned off during the 10-min CCh application and was resumed 10 min after the wash was begun with drug-free aCSF. As shown in Figure 2B, normal CCh-LTD (36.3 ± 5.7%, n = 6) was induced in slices that did not receive synaptic stimulation during CCh application. The synaptic stimulation interruption alone did not significantly alter the basal synaptic transmission (97.5 ± 2.9% of baseline, n = 4; P > 0.05). These results suggest that pharmacological activation of mAChRs alone, in the absence of synaptically evoked glutamate release, is sufficient to elicit LTD in the mPFC. To further test this conclusion, we applied CCh in the absence of synaptic transmission by using kynurenic acid, which blocked both NMDA and non-NMDA receptors (Kemp et al. 1988) together with a broad-spectrum mGluR antagonist MCPG (Hayashi et al. 1994). In control experiments, coapplication of kynurenic acid (20 mM) and MCPG (1 mM) quickly and completely blocked synaptic transmission, and the synaptic responses recovered quickly after washout of these antagonists (98.5 ± 4.5% of baseline, n = 4; P > 0.05; Fig. 2C). After complete blockade of synaptic responses with kynurenic acid and MCPG, CCh (50 μM) was applied. After washout of kynurenic acid, MCPG, and CCh, the EPSCs did not completely recover to the baseline levels. On average, the slope of EPSCs measured 30 min after washout of CCh was 63.9 ± 5.4% of baseline (n = 6), which was not significantly different from that of CCh-LTD recorded under control condition (Fig. 2C), indicating that activation of ionotropic glutamate receptor and mGluR is not necessary for the induction of CCh-LTD.
CCh-LTD Requires Activation of Postsynaptic PLC, PKC, and IP3 Receptors
We next explored the cellular mechanism underlying the induction of CCh-LTD. To assess the possible contribution of postsynaptic M1 mAChR to CCh-LTD, we inhibited activation of G protein–coupled signaling with the postsynaptic application of GDPβS (300 μM). Cells were dialyzed for at least 20 min before CCh (50 μM) application to ensure completely dialysis of GDPβS into the cells. Although neurons loaded with GDPβS showed normal levels of acute depression of synaptic responses by CCh, CCh-LTD was blocked (6.5 ± 4.1%, n = 6; P < 0.05 when compared with GTP-loaded group; Fig. 3A). Furthermore, control GDPβS-loaded cells did not show a significant change of synaptic responses when held for the identical period of time (n = 4; data not shown). Therefore, postsynaptic M1 mAChR is required for the CCh-LTD induction.
Because M1 mAChR is coupled to Gq/11 protein, which activates phospholipase C (PLC) and therefore can stimulate phosphoinositide hydrolysis, leading to the formation of inositol 1,4,5-trisphosphate (IP3) and diacylglycerol to stimulate the release of Ca2+ from intracellular stores and the activation of PKC (Hulme et al. 1990), we explored the possibility that M1 mAChR–mediated activation of PLC and PKC is involved in the induction of CCh-LTD. Toward this end, a broad-spectrum PLC blocker U73122 and a selective PKC inhibitor Bis-I were used. As shown in Figure 3B,C, U73122 (10 μM) or Bis-I (2 μM) did not affect the acute EPSC depression induced by CCh but significantly reduced CCh-LTD (U73122: 4.5 ± 3.2%, n = 6; Bis-I: 12.1 ± 4.8%, n = 6; P < 0.05 when compared with CCh alone group), demonstrating an involvement of PLC/PKC-coupled signaling processes in the CCh-LTD induction. In contrast, pretreatment of the slices with the inactive structural analog of U73122 and Bis-I, respectively, U73343 (10 μM) or Bis-V (2 μM), failed to affect the induction of CCh-LTD (U73343: 31.5 ± 4.9%, n = 5; Bis-V: 26.5 ± 5.6%, n = 5; P > 0.05 when compared with CCh alone group).
Recent work has shown that activation of both Src family of tyrosine kinases and extracellular signal-regulating kinase1/2 (ERK1/2) is involved in the induction of mAChR-dependent LTD in the visual cortex (McCoy and McMahon 2007) and the hippocampal CA1 region (Scheiderer et al. 2008). To test whether either of these kinases is required for the induction of CCh-LTD in the mPFC, we examined the effects of Src kinase inhibitor PP2 and mitogen-activated ERK kinase inhibitor U0126. We found that neither PP2 (10 μM) nor U0126 (20 μM) pretreatment significantly affects CCh-LTD induction (PP2: 34.5 ± 5.3%, n = 4; U0126: 29.5 ± 5.8%, n = 6; P > 0.05 when compared with CCh alone group; Fig. 3D). Therefore, CCh-LTD on mPFC layer V neurons does not rely on the activation of Src family of kinases or ERK1/2.
To address the potential role of postsynaptic intracellular Ca2+ concentration ([Ca2+]i) in the CCh-LTD induction, we first examined the effect of thapsigargin, which has already been shown to deplete intracellular Ca2+ stores by preventing their refilling (Goeger et al. 1988). As shown in Figure 3E, thapsigargin (10 μM) did not affect the acute EPSC depression induced by CCh but blocked CCh-LTD (6.9 ± 3.4%, n = 7; P < 0.05 when compared with CCh alone group). We next examined the effect of intracellular Ca2+ chelation on CCh-LTD by postsynaptic application of membrane-impermeable Ca2+ chelator BAPTA (10 mM). Although the acute EPSC depression by CCh was unaffected in BAPTA-loaded cells, CCh-LTD was blocked (10.3 ± 4.2%, n = 6; P < 0.05 when compared with CCh alone group; Fig. 3F). Control BAPTA-loaded cells did not show a significant change of synaptic responses when held for the identical period of time (n = 3; data not shown).
Experiments were also carried out to determine whether Ca2+ release triggered by IP3 receptors or ryanodine receptors from internal stores is involved in the CCh-LTD induction. As shown in Figure 3G, postsynaptic application of IP3 receptor blocker heparin (5 mg/mL) did not significantly affect the acute EPSC depression by CCh but blocked CCh-LTD (11.5 ± 5.1%, n = 7; P < 0.05 when compared with CCh alone group). In contrast, normal levels of CCh-LTD (30.5 ± 5.3%, n = 5; P > 0.05 when compared with vehicle control [1% DMSO]–loaded group: 32.6 ± 4.2%, n = 5) were observed in cells loaded with ryanodine (100 μM), which blocks ryanodine receptors in the internal stores (Fig. 3H). We also examined the role of extracellular Ca2+ influx in the CCh-LTD induction with the use of L-type voltage-gated Ca2+ channel blocker nimodipine. Normal levels of CCh-LTD (38.6 ± 4.8%, n = 5) were observed in the presence of nimodipine (10 μM) compared with control slices (Fig. 3D). Together, these results indicate that a rise in [Ca2+]i from IP3 receptor–sensitive internal stores is specifically required for the induction of CCh-LTD.
It has been reported that de novo protein synthesis is required for the induction of mAChR-dependent LTD in the perirhinal cortex (Massey et al. 2001) and the hippocampal CA1 region (Volk et al. 2007). To investigate the possible requirement for protein synthesis in the establishment of CCh-LTD, we preincubated slices in the protein synthesis inhibitors, cycloheximide (60 μM), or anisomycin (25 μM), for at least 1 h before CCh (50 μM) application. We found that neither cycloheximide nor anisomycin pretreatment has effect the acute EPSC depression by CCh or the induction of CCh-LTD (cycloheximide: 28.5 ± 4.8%, n = 4; anisomycin: 25.5 ± 4.6%, n = 6; P > 0.05 when compared with CCh alone group), suggesting that the induction of CCh-LTD in the mPFC is independent of protein synthesis processes (Supplementary Fig. S3).
Presynaptic Expression of CCh-LTD
To explore whether the expression of CCh-LTD in the mPFC was presynaptic or postsynaptic in origin, 4 different approaches were used. We first examined the effect of CCh on pharmacologically isolated NMDA receptor–mediated EPSC (EPSCNMDA) in Mg2+-free aCSF solution containing AMPA/kainate receptor antagonist CNQX (20 μM). If CCh-LTD were expressed presynaptically, changes in both the AMPA and NMDA receptor–mediated components of EPSCs by CCh would be expected. Consistent with this view, we found that application of CCh (50 μM) for 10 min induced a nearly identical LTD of AMPA and NMDA receptor–mediated EPSCs. On average, the magnitude of CCh-LTD of EPSCNMDA was 31.5 ± 5.5% (n = 5; measured 30 min after CCh washout; Fig. 4A). Comparable results were obtained with EPSCAMPA (34.7 ± 4.5%, n = 6; Fig. 2A).
To further test the possibility that CCh-LTD expresses presynaptically, we examined the effect of CCh (50 μM) on the failure rate of single-fiber EPSCs evoked by minimal stimulation, which reflects changes in presynaptic transmitter release (Stevens and Wang 1994). As a typical example shown in Figure 4B, the expression of CCh-LTD was accompanied by an increase in the synaptic failure rate. On average, the failure rate was increased from 0.52 ± 0.05 to 0.78 ± 0.06 measured 25–30 min after CCh washout (n = 7; P < 0.05, paired Student's t-test; Fig. 4C). We also addressed the synaptic locus of CCh-LTD expression by examining the trial-to-trial amplitude fluctuation in EPSCs with the variance analysis. The coefficient of variance (CV) value varies with quantal content but is independent of changes postsynaptic response to a fixed amount of transmitter and is a useful measure of changes in presynaptic function (Bekkers and Stevens 1990). Because variance analysis is best done on unitary synaptic responses, we carried out a variance analysis of unitary single-fiber EPSCs evoked by minimal stimulation before and after CCh-LTD induction. We found a decrease in 1/CV2 from 34.5 ± 2.8 to 20.8 ± 2.3 measured 25–30 min after CCh washout (n = 10; P < 0.05, paired Student's t-test; Fig. 4D).
We next calculated the paired-pulse ratio (PPR) before and after the CCh-LTD induction. If the expression of CCh-LTD involves a presynaptic mechanism of action, it would be associated with an increase in the PPR. Two consecutive stimuli with a 50-ms interstimulus interval elicited a pair of EPSCs with the second EPSC significantly larger than the first EPSC (Fig. 4E). Under control conditions, the ratio of the slope of second EPSC divided by the first was 1.42 ± 0.05 (n = 8). The expression of CCh-LTD was accompanied by a significant increase in the PPR to 1.72 ± 0.06 measured 25–30 min after CCh washout (n = 8; P < 0.05, paired Student's t-test; Fig. 4C). Taken together, these findings clearly indicate that the expression locus of CCh-LTD on mPFC layer V pyramidal neurons is presynaptic.
There is considerable evidence that AMPA receptor endocytosis represents an important mechanism contributing to the expression of LTD (Carroll et al. 2001). To assess a role for AMPA receptor internalization in CCh-LTD, we performed experiments in which postsynaptic cells were loaded with D15 (1 mM), a peptide that interferes with clathrin-mediated endocytosis (Lüscher et al. 1999), before CCh application. We found that postsynaptic loading with D15 did not affect the acute EPSC depression by CCh or the induction of CCh-LTD (24.7 ± 5.2%, n = 4; P > 0.05 when compared with vehicle control-loading group; Supplementary Fig. S4A). The interaction of protein interacting with C-kinase 1 (PICK1) with the C-terminal regions GluR2/3 subunit has been implicated in the endocytosis and reinsertion of actively internalized AMPA receptors at hippocampal synapses (Daw et al. 2000; Lu and Ziff 2005). We also tested whether the GluR2/3–PICK1 interaction is important for the expression of CCh-LTD by introducing a peptide corresponding to the last 11 amino acids of GluR2 (YNVYGIESVKI; pep2-SVKI) into neurons to disrupt this interaction (Dev et al. 1999). As shown in Supplementary Figure S4B, postsynaptic application of pep2-SVKI (100 μM) did not affect the acute EPSC depression by CCh or the induction of CCh-LTD (31.1 ± 5.6%, n = 5; P > 0.05 when compared with the control peptide pep-SVKE-loading group: 32.6 ± 4.2%, n = 4; Supplementary Fig. S4B). These results rule out a role of clathrin-mediated endocytotic removal of postsynaptic AMPA receptors in governing the expression of CCh-LTD.
A Role of the NO/cGMP/PKG Signaling Pathway in CCh-LTD
Our results so far show that activation of postsynaptic M1 mAChR signaling with CCh persistently suppresses glutamate synaptic transmission through presynaptic effect on neurotransmitter release. As such, it is possible that a retrograde messenger signal is required for the expression of CCh-LTD. Furthermore, the retrograde messenger that mediates the effect of CCh should be capable of producing inhibition of glutamate release. Because activation of M1 mAChRs can stimulate NO formation and NO has been shown to be involved in LTD in the dentate gyrus (Wu et al. 1997), the neostriatum (Calabresi et al. 1999) and the cerebellum parallel fibers (Shibuki and Okada 1991; Lev-Ram et al. 1995), we therefore determined whether NO could be the retrograde messenger signal that mediates the development of CCh-LTD. We first used the broad-spectrum NO synthase (NOS) inhibitor L-NAME, which blocks both neuronal NOS and endothelial NOS activity (Southan and Szabó 1996). We found that L-NAME (50 μM) treatment did not affect the acute EPSC depression by CCh but blocked CCh-LTD induction (7.5 ± 4.6%, n = 5; P < 0.05 when compared with CCh alone group; Fig. 5A). Similarly, pretreatment of slices with a highly selective neuronal NOS inhibitor NPA (10 μM) also blocked the induction of CCh-LTD (8.3 ± 3.9%, n = 6; P < 0.05 when compared with CCh alone group; Fig. 5A). These results show that NO production is required for M1 mAChR to develop LTD. To test whether postsynaptic NO production is important in CCh-LTD, we postsynaptically injected NPA (10 μM) into recorded cells. As shown in Figure 5B, postsynaptic application of NPA almost completely blocked CCh-LTD (7.5 ± 5.9%, n = 5; P < 0.05 when compared with CCh alone group) without significantly affecting acute EPSC depression by CCh. We also found that CCh-LTD induction was blocked by bath application of NO scavenger carboxy-PTIO (30 μM; 4.7 ± 3.1%, n = 6; P < 0.05 when compared with CCh alone group; Fig. 5C).
Having confirmed a role of NO production in CCh-LTD, we next identify the potential downstream signaling messengers involved in NO-initiated cascade underlying CCh-LTD. The most commonly defined molecular target of NO is sGC, which is stimulated to generate cGMP and subsequently PKG activation (Wang and Robinson 1997). To determine whether CCh-LTD requires activation of sGC and PKG, we used selective sGC inhibitor, ODQ (Garthwaite et al. 1995), and PKG inhibitor, KT5823 (Grider 1993). As shown in Figure 5D, bath application of ODQ (1 μM) blocked CCh-LTD (8.3 ± 5.1%, n = 5; P < 0.05 when compared with CCh alone group) without significantly affecting acute EPSC depression by CCh. Likewise, inhibition of PKG with bath application of KT5823 (1 μM) also blocked the induction of CCh-LTD (6.9 ± 4.2%, n = 6; P < 0.05 when compared with CCh alone group; Fig. 5E). In contrast, postsynaptic injection of the specific PKG inhibitory peptide (RKRARKE; PKGI; 1 mM) had no effect on the induction of CCh-LTD (34.7 ± 4.8%, n = 5; P > 0.05 when compared with CCh alone group; Fig. 5F). We next examined whether persistent NO/cGMP signaling is required for the CCh-LTD expression and maintenance. Toward this goal, we tested the effect of KT5823 on the depression of synaptic transmission induced by CCh application. As shown in Figure 5G, KT5823 (1 μM) had no significant effect on the preestablished CCh-LTD (32.5 ± 5.2%, n = 4; P > 0.05 when compared with CCh alone group) when it was applied for 20 min beginning 20 min after washout of CCh. These results suggest that CCh-LTD is dependent on activation of presynaptic sGC/PKG signaling pathway initiated by postsynaptic NO production, and persistent activation of NO/cGMP signaling is not required for the expression and maintenance of CCh-LTD.
To further confirm the role of NO/cGMP signaling in the induction of LTD in the mPFC, we investigated whether NO donor could mimic the effect of CCh to induce LTD by using a specific NO donor SNAP, which has been reported to effectively induce LTD in the striatum at a concentration of 100 μM (Calabresi et al. 1999). Unexpectedly, bath application of SNAP (100 or 300 μM) for 10 min exhibited no long-lasting effects on EPSCs, although it caused a slight enhancement of EPSCs during and shortly after drug application (Fig. 6A). On average, the slope of EPSCs measured 30 min after washout of SNAP was 104.3 ± 3.6% (n = 6) and 96.5 ± 3.3% of baseline (n = 6), respectively. Similar results were also obtained by using bath application of a selective, membrane-permeable cGMP analog 8-Br-cGMP, which stimulates PKG and depresses glutamatergic responses in isolated hippocampal neurons (Lei et al. 2000). When 8-Br-cGMP (0.2 or 1 mM) was bath applied to slices for 10 min, it produced only a transient depression of EPSCs (5.5 ± 3.4%, n = 4; P > 0.05, and 26.8 ± 5.3%, n = 5; P < 0.05, respectively, at 10 min), which completely recovered to baseline value upon washout (96.5 ± 3.5% of baseline, n = 4, and 98.4 ± 4.2% of baseline, n = 5, respectively, at 30 min; P > 0.05; Fig. 6B). These results suggest the possibility that simply stimulating NO/cGMP signaling alone is not sufficient to elicit LTD. Therefore, we hypothesized that simultaneous activation of NO/cGMP and other forms of signaling stimulated by M1 mAChR activation are required for the induction of CCh-LTD. This hypothesis would predict that strong activation of either types signaling system might induce greater magnitude of LTD. As anticipated, coapplication of 10 μM CCh and 100 μM SNAP or 0.2 mM 8-Br-cGMP induced a greater LTD (CCh + SNAP: 36.4 ± 6.2%, n = 6; CCh + 8-Br-cGMP: 32.7 ± 5.8%, n = 4; P < 0.05) than that elicited by CCh (10 μM) with vehicle control (0.1% DMSO; 10.4 ± 3.9%, n = 5; Fig. 6C,D). We also found that the enhancement effects of SNAP and 8-Br-cGMP on the induction of CCh-LTD were blocked by ODQ (1 μM) treatment (CCh + SNAP: 6.9 ± 3.2%, n = 5; CCh + 8-Br-cGMP: 6.5 ± 4.2%, n = 4; P < 0.05; Fig. 6E).
In addition to NO, previous studies have shown that endocannabinoids can act as retrograde messengers traveling across synapses to inhibit neurotransmitter release (Kreitzer and Regehr 2001; Wilson and Nicoll 2001). We thus examined whether endocannabinoids could mediate CCh-LTD by using a CB1 cannabinoid receptor antagonist SR141716A. However, we found that pretreatment of slices with SR141716A (5 μM) did not significantly affect the induction of CCh-LTD (32.6 ± 4.3%, n = 5; P > 0.05 when compared with CCh alone group; Fig. 3H), indicating that endocannabinoids are not necessary for CCh-LTD. To confirm that SR141716A indeed block CB1 cannabinoid receptors, we found in separate experiments that SR141716A (5 μM) almost completely blocked the inhibition of EPSCs by CB1 receptor agonist WIN 55,212-2 (1 μM; n = 3; data not shown).
Synaptic Activation of M1 mAChR Induces LTD
The results thus far were obtained with exogenous application of CCh, and it is unclear whether the release of ACh within the mPFC can effectively activate M1 mAChRs to elicit LTD. In the final series of experiments, we adapted a recently reported protocol to test this hypothesis (Volk et al. 2007). As shown in Figure 7A, delivery of paired-pulse LFS (PP-LFS; 50-ms interstimulus interval) at 1 Hz for 20 min in the presence of D-APV (50 μM) produced a stable form of LTD (PP-LFS-LTD; Fig. 7A). On average, the slope of EPSCs measured 30 min after the end of PP-LFS was 67.5 ± 5.3% (n = 8; Fig. 7B). The magnitude of PP-LFS-LTD was significantly reduced by the atropine (1 μM; 15.3 ± 4.8%, n = 4) and pirenzepine (100 nM; 14.5 ± 4.6%, n = 5) but not by MLA (100 nM; 34.3 ± 4.7%, n = 4), suggesting that it is mediated, at least partly, by M1 mAChRs (Fig. 7B,E). In addition, to further ensure the contribution of endogenous ACh release in the induction of PP-LFS-LTD, we performed additional experiments in which LTD was induced in the presence of cholinesterase inhibitor, eserine, to enhance ACh concentrations in the slice. As shown in Figure 7C, eserine (2 μM) significantly enhanced the PP-LFS-LTD (45.3 ± 5.4%, n = 5; P < 0.05 when compared with PP-LFS alone group). Because antagonism of M1 mAChRs does not completely abolish PP-LFS-LTD, we speculated that other signaling cascades are also involved in the induction of LTD by PP-LFS. Volk et al. (2007) reported that PP-LFS–induced LTD in hippocampal CA1 area shares a common expression with LTD induced by pharmacological activation of group I mGluRs. This result led us to hypothesize a role of group I mGluRs for the induction of PP-LFS-LTD. To test this hypothesis, LTD was induced by PP-LFS in the presence of pirenzepine (100 nM) and broad-spectrum mGluR antagonist MCPG (1 mM). In all 6 slices tested, PP-LFS-LTD induction was completely blocked under such experimental condition (6.7 ± 5.2%, n = 6; P < 0.05 when compared with PP-LFS alone group; Fig. 7D). Similar results were also obtained by using LY367385, a highly selective group I mGluR antagonist. As shown in Figure 7E, coapplication of pirenzepine (100 nM) and LY367385 (100 μM) almost completely blocked the induction of PP-LFS-LTD (7.3 ± 4.5%, n = 4; P < 0.05 when compared with PP-LFS alone group). These results suggest that the residual PP-LFS-LTD during antagonism of M1 mAChRs was mediated primarily by group I mGluRs.
Similar to the effect of CCh, the induction of PP-LFS-LTD was mediated by an NO/sGC/PKG signaling process. In fact, PP-LFS-LTD was significantly blocked by bath application of NPA (10 μM; 8.3 ± 3.2%, n = 4), ODQ (1 μM; 10.2 ± 4.3%, n = 4), or KT5823 (1 μM; 13.3 ± 5.2%, n = 4; Fig. 7E).
To strengthen the tie between pharmacological activation of M1 mAChRs and synaptic stimulation of M1 mAChRs in the induction of LTD, we performed an occlusion experiment. If these 2 forms of LTD utilize similar mechanisms, the saturation of one form of LTD should occlude induction of the other form of LTD at the same synapses. Because maintaining an extremely prolonged period of whole-cell recording is extremely difficult, we therefore attempted to use extracellular field potential recordings to perform occlusion experiments. In each experiment, we applied the induction protocol for the first form of LTD 3 times to saturate expression before applying the second induction protocol. In line with whole-cell recordings, either bath application of CCh (50 μM; 10 min) or PP-LFS consistently elicited stable LTD of the slope of fEPSPs. On average, the magnitude of PP-LFS-LTD and CCh-LTD of fEPSP was 19.8 ± 3.2% (n = 6; P < 0.05) and 25.5 ± 4.3% (n = 6; P < 0.05), respectively (Fig. 8B,D). After CCh-LTD was fully established in the presence of D-APV (50 μM) and MCPG (1 mM), application of PP-LFS failed to induce any further LTD (3.7 ± 2.4%, n = 5; P > 0.05 when compared with pre-PP-LFS baseline; Fig. 8A). Likewise, saturation of PP-LFS-LTD also occluded the induction of CCh-LTD (5.2 ± 3.7%, n = 6; P > 0.05 when compared with pre-CCh baseline; Fig. 8C). Such an occlusion experiment demonstrates that CCh-LTD and PP-LFS-LTD share a common expression mechanism.
The present study shows the first evidence for a role of cholinergic system in the long-term regulation of synaptic transmission on layer V pyramidal neurons of the mPFC and supports a molecular model of cholinergic action (Fig. 9). Our results indicate that pharmacological activation or synaptic stimulation of M1 mAChRs results in the induction of LTD of synaptic transmission. This novel form of LTD is independent of NMDA receptors or coincident synaptic activity but requires a postsynaptic Gq/11 protein–coupled activation of PLC, PKC, and the release of Ca2+ from IP3 receptor–sensitive intracellular stores. In addition, our results demonstrate that the activation of postsynaptic neuronal NOS and subsequent diffusion of NO to the presynaptic terminal to stimulate sGC/PKG signaling pathway is necessary for the induction of LTD.
The mPFC receives rich cholinergic innervation from the basal forebrain nuclei (Lehmann et al. 1980; Satoh and Fibiger 1986; Gaykema et al. 1990). Previous studies addressing the role of cholinergic system in the mPFC focused mainly on behavioral arousal and attentional performance (Day et al. 1991; Himmelheber et al. 2000; Passetti et al. 2000; McGaughy et al. 2002; Parikh et al. 2007). Here we have extended these observations by showing that pharmacological activation of M1 mAChRs can induce an LTD of synaptic transmission on mPFC layer V pyramidal neurons, as it does in the visual cortex (Kirkwood et al. 1999), the perirhinal cortex (Massey et al. 2001), and the hippocampal CA1 region (Volk et al. 2007). Our results also show that the induction of LTD by CCh was not affected by NMDA receptor antagonist D-APV or interruption of synaptic stimulation during drug application, suggesting that the synaptic activation of ionotropic glutamate receptor and mGluR is not likely to be involved in the induction of LTD in the mPFC. Thus, acute activation of M1 mAChRs alone is sufficient to induce LTD.
How might M1 mAChR activation lead to LTD? We found that CCh-LTD is prevented by postsynaptic inhibition of G-protein activity with GDPβS and by bath application of PLC or PKC inhibitors, suggesting that stimulation of postsynaptic Gq/11 protein–coupled M1 mAChRs induces LTD through PLC and PKC activation. The findings that postsynaptic application of Ca2+ chelator BAPTA or IP3 receptor blocker heparin reduces the magnitude of CCh-LTD indicate that IP3 receptor–mediated postsynaptic [Ca2+]i elevation is an additional factor required for the induction of CCh-LTD. In contrast, inhibition of Ca2+ release from ryanodine-sensitive internal stores or extracellular Ca2+ influx through L-type voltage-gated Ca2+ channels does not interfere with CCh-LTD. Interestingly, signaling through PLC activation now appears to be an important mechanism underlying the induction of LTD. For example, Reyes-Harde and Stanton (1998) showed that postsynaptic application of PLC inhibitor prevents the induction of LTD by 1 Hz LFS in the hippocampal CA1 region. In the mPFC, activation of postsynaptic group II mGluRs has also been shown to induce LTD through PLC activation, IP3 receptor–mediated postsynaptic increases of [Ca2+]i, and PKC activation (Otani et al. 2002). In addition, application of PLC inhibitor significantly inhibits CCh-LTD induction in rat visual cortex (McCoy and McMahon 2007). It is therefore possible that the requirement of PLC activation for LTD induction can be fulfilled through different G protein–coupled receptors. It was recently shown that activation of M1 mAChRs can stimulate the Src kinase family, which subsequently leads to increase ERK1/2 activation to induce LTD in the visual cortex (McCoy and McMahon 2007) and the hippocampal CA1 region (Scheiderer et al. 2008). In the present study, however, we found that inhibition of Src family of kinases and ERK1/2 did not significantly alter the induction of CCh-LTD. These results would then suggest that M1 mAChR–mediated CCh-LTD in mPFC layer V neurons is not attributed to the activation of Src kinase family and ERK1/2. The reason for this discrepancy is not clear but could be attributed to the neurobiological differences between the visual cortex and the mPFC, resulting in stimulating different cellular signal transduction pathways.
Although the induction of CCh-LTD occurs in the postsynaptic pyramidal neurons, presynaptic changes in transmitter release are primarily responsible for the expression of this form of LTD. Two lines of evidence support this conclusion. First, CCh equally induces a persistent depression of AMPA and NMDA receptor–mediated components of EPSCs. Second, the expression of CCh-LTD is accompanied by a decrease in 1/CV2 and an increase in synaptic failure rate and PPR, which are generally considered to indicate a reduction of neurotransmitter release probability (Zucker 1989; Bekkers and Stevens 1990; Stevens and Wang 1994). There is compelling evidence that the expression of LTD could result from a reduction in the surface expression of postsynaptic AMPA receptors by facilitating clathrin-mediated endocytosis processes (Hayashi et al. 2000). Although we cannot completely exclude this possibility, evidence against this is provided by the lack of effect of postsynaptic application of D15 and pep2-SVKI on CCh-LTD, manipulations with respect to block the clathrin-mediated endocytosis (Lüscher et al. 1999) and the GluR2/3–PICK1 interactions (Daw et al. 2000; Lu and Ziff 2005), respectively. The postsynaptic induction of CCh-LTD and its presynaptic expression implies the existence of cross talk between the post- and presynaptic neurons and the involvement of a diffusable retrograde messenger. One potential candidate retrograde messenger is NO, which is thought to be involved in the induction of LTD in certain brain areas under certain conditions (Shibuki and Okada 1991; Lev-Ram et al. 1995; Wu et al. 1997; Calabresi et al. 1999). This is supported by the findings that the induction of CCh-LTD and PP-LFS-LTD was prevented by postsynaptic application of NOS inhibitors or bath application of NO scavenger. The findings that sGC and PKG inhibitors blocked LTD induced by CCh and PP-LFS indicate a role of NO sGC/PKG-dependent pathway in the induction of M1 mAChR–mediated LTD. However, the precise mechanism by which PKG inhibits presynaptic glutamate release remains to be determined. Given that PKG can directly phosphorylate and open K+ channels in pituitary tumor cells (White et al. 1993), it will be important for future studies to explore whether such a mechanism might be responsible for the reduction of glutamate release during expression of CCh-LTD. We are surprised to find that although the NO donor SNAP and cGMP analog 8-Br-cGMP can enhance the induction of CCh-LTD, application of SNAP or 8-Br-cGMP alone fails to alter the synaptic transmission. The simplest interpretation of this observation is that simply activating NO/cGMP signaling alone is not sufficient to elicit LTD. Therefore, the simultaneous elevations of NO and other forms of signaling stimulated by M1 mAChR activation are required for the induction of CCh-LTD. In this context, Reyes-Harde et al. (1999) reported that the induction of hippocampal CA1 LTD by a submaximal LFS requires concomitant NO-stimulated PKG activity and Ca2+ release from cyclic ADP-ribose–sensitive stores. It is worth noting, however, that bath application of SNAP or cGMP can effectively elicit a stable corticostriatal LTD via the activation of postsynaptic sGC/PKG signaling cascade (Calabresi et al. 1999).
Recently, Volk et al. (2007) also reported that the application of CCh or PP-LFS induces an LTD at the Schaffer collateral-CA1 synapses. Although their results appear similar to our findings, there are significant differences between these 2 sets of results. First, Volk et al. (2007) showed that the newly protein synthesized proteins and ERK1/2 activation are required for CCh-LTD in the hippocampus. Second, a persistent decrease in surface AMPA receptors is involved in the postsynaptic expression of hippocampal CCh-LTD. It is therefore possible that activation of M1 mAChRs in different brain regions may result in stimulation of different cellular events that may vary in their mode of action.
In conclusion, the results of the present study have provided the first description of M1 mAChR–mediated LTD in the mPFC. This LTD is mechanistically distinct from previously reported forms of mPFC LTD induced by high-frequency tetanic stimuli (Otani et al. 1998, 1999; Takita et al. 1999) or by group II mGluR agonists (Otani et al. 2002; Huang and Hsu 2008) and depends critically on both pre- and postsynaptic processes. We demonstrate that the induction of M1 mAChR–mediated LTD is associated with the activation of postsynaptic PLC, PKC, and the elevation of Ca2+ release from IP3 receptor–sensitive stores. In addition, we show that postsynaptic release of NO may act as a retrograde messenger on presynaptic site to induce persistent synaptic depression via an sGC-/PKG-dependent pathway. These findings may provide a major advance in our understanding of the interaction between cholinergic and glutamatergic neurotransmission systems in the mPFC. Possible functional consequence of M1 mAChR–mediated LTD in the mPFC remains to be determined. Given that LTD, in the mammalian brain, is generally assumed as a synaptic mechanism underlying the learning during novel experiences (Bear and Abraham 1996; Manahan-Vaughan and Braunewell 1999), this LTD may therefore serve as an important role in mPFC-mediated cue detection and sensory information encoding processes in novel contexts.
National Science Council, Taiwan (NSC97-2321-B-006-002-MY2 to C.C.H. and NSC97-2321-B-006-008 to K.S.H.); National Cheng Kung University (A002 and R026 to K.S.H.).
Conflict of Interest: None declared.
- nitric oxide
- protein kinase c
- signal transduction
- cholinergic agents
- cyclic gmp-dependent protein kinases
- depressive disorders
- phospholipase c
- precipitating factors
- prefrontal cortex
- muscarinic acetylcholine receptor
- synaptic transmission
- soluble guanylyl cyclase
- cholinergic synaptic transmission
- cognitive ability
- long-term depression