By regulating the neocortical excitability, nicotinic acetylcholine receptors (nAChRs) control vigilance and cognition and are implicated in epileptogenesis. Modulation of γ-aminobutyric acid (GABA) release often accompanies these processes. We studied how nAChRs regulate GABAergic transmission in the murine neocortex with immunocytochemical and patch-clamp methods. The cholinergic fibers densely innervated the somatosensory, visual, motor, and prefrontal cortices (PFC). Laminar distribution was broadly homogeneous, especially in the PFC. The cholinergic terminals were often adjacent to the soma and dendrites of GABAergic interneurons, but well-differentiated synapses were rare. Tonically applied nicotine (1–100 μM) increased the frequency of spontaneous GABAergic inhibitory postsynaptic currents (IPSCs) on pyramidal neurons in PFC layer V. The contribution of nAChR types was assessed by using 1 μM dihydro-β-erythroidine (DHβE), to block heteromeric nAChRs, and 10 nM methyllycaconitine (MLA), to block homomeric nAChRs. Both inhibitors antagonized the effect of nicotine on IPSCs, suggesting that mixed nAChR types control pyramidal neuron inhibition in layer V. To determine whether nAChRs are expressed on basket cells’ terminals, we studied miniature IPSCs (mIPSCs). These were revealed using 0.5 μM tetrodotoxin and 50 μM Cd2+ to isolate the GABAergic terminals from the action potential drive. The nicotinic stimulation of mIPSCs was antagonized by DHβE, but not MLA, indicating that heteromeric nAChRs prevail in GABAergic terminals. Immunocytochemistry confirmed the expression of nAChRs on basket cells’ somata and terminals. Finally, when the ionotropic glutamatergic transmission was blocked, nicotine partially inhibited the IPSCs, an effect counteracted by both DHβE and MLA. Therefore, a fraction of nAChRs are capable of activating GABAergic interneurons that in turn inhibit other GABAergic interneurons, thereby reducing the IPSCs. We conclude that heteromeric nAChRs control GABA release presynaptically, whereas mixed nAChRs regulate both excitation and inhibition of interneurons, the balance depending on the overall glutamatergic drive.
The mammalian neocortex is diffusely innervated by the cholinergic fibers ascending from basal forebrain (Semba 2004). Acetylcholine (ACh) release increases in aroused cerebral states, including the rapid eye movement (REM) phase of sleep (Steriade and McCarley 2005). On a shorter timescale, cholinergic transmission participates in shaping information processing in the frontoparietal attentional network (Everitt and Robbins 1997; Sarter et al. 2006). In keeping with the widespread functions of ACh in the brain, alterations in the cholinergic system often accompany the development of neurodegenerative diseases, although little conclusive evidence is yet available about the roles of ACh receptors in these disorders (Mesulam 2004).
ACh exerts its effects through both metabotropic (muscarinic) and ionotropic receptors (nicotinic, nAChR), whose function and interplay are only partially understood. Nicotine generally stimulates cognition in mammals and improves vigilance, memory, and learning in smokers and patients suffering from neuropsychiatric disorders (Newhouse et al. 2004; Mansvelder et al. 2006). Direct recent evidence shows that nAChR activation increases the threshold for the potentiation of glutamatergic synapses dependent on the action potential frequency in layer V of the murine prefrontal cortex (PFC; Couey et al. 2007). That nAChRs regulate the neocortical excitability is also suggested by the observation that several mutations linked to autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE) map on genes coding for nAChR subunits (Steinlein et al. 1995; De Fusco et al. 2000; Phillips et al. 2001; Aridon et al. 2006). The focal location of the attacks in the frontal lobe strongly suggests a neocortical origin. Nonetheless, the pathogenetic mechanism remains largely speculative because of the wide gaps in current understanding of the nAChR functions in the brain.
The nAChR is a pentamer of subunits surrounding a pore permeable to cations. In the mammalian brain, at least 9 gene products concur in forming functional receptors: α2–α7 and β2–β4. In rodents, the prevalent neocortical forms are the heteropentamer α4β2 and the homopentamer α7. The contribution of other subunits is still debated (Gotti et al. 2007). Homopentameric nAChRs present higher permeability to Ca2+, higher desensitization rates, and lower sensitivity to agonists such as ACh and nicotine (Dani and Bertrand 2007). Besides subunit composition, the physiological complexity of the cholinergic system depends on nAChR expression at pre-, post-, and nonsynaptic sites. Receptors’ distribution in different subcellular compartments points to the relevance of paracrine transmission. In fact, morphological studies in the rat and macaque neocortices highlight conspicuous cholinergic varicose fibers, with rare synaptic specializations (Umbriaco et al. 1994; Mrzljak et al. 1995). The balance of synaptic and nonsynaptic release is uncertain because obtaining precise counts of the synaptic contacts is difficult (Turrini et al. 2001). Nevertheless, other lines of evidence support the notion that ACh exerts important modulatory roles through nonsynaptic transmission. The distribution of acetylcholinesterase does not fully overlap that of nAChRs, and low tonic ACh concentrations are measured in the cerebrospinal fluid, even in the absence of cholinesterase inhibitors (Dani and Bertrand 2007; Lendvai and Vizi 2008; Pepeu and Giovannini 2008). Hence, understanding the cerebral physiology of nAChRs requires distinction between fast postsynaptic transmission and slower effects produced by sustained agonist concentrations. Consistently, behavioral experiments coupled with real-time detection of cortical choline levels show that ACh release in the rat PFC controls attention tasks on different timescales, ranging from seconds to minutes (Parikh et al. 2007).
In the frontal cortex, nAChRs are implicated in both excitatory and inhibitory transmission. In both rats and mice, they are thought to control glutamate release from thalamocortical fibers (Vidal and Changeux 1993; Gioanni et al. 1999; Lambe et al. 2003). Interspecific differences are however apparent in the nAChR expression on principal cells. Somatic nicotinic currents have been measured in pyramidal neurons from rat layers II/III (Chu et al. 2000), layer V (Zolles et al. 2009), and layer VI (Kassam et al. 2008), but not murine layer V (Couey et al. 2007). In contrast, current evidence suggests that the broad features of the nicotinic regulation of γ-aminobutyric acid (GABA) release are similar in different mammals. Expression of nAChRs on the soma of different interneuronal populations is established in rats (Xiang et al. 1998; Porter et al. 1999; Christophe et al. 2002), humans (Alkondon, Pereira, Eisenberg, et al. 2000), and mice (Couey et al. 2007). In general, nAChR activation tends to stimulate GABA release in different layers.
We have investigated the steady-state nicotinic modulation of GABA release onto layer V pyramidal neurons in the murine PFC. The inhibitory control is particularly potent in this layer because of profuse interneuronal innervation (Douglas and Martin 2004). Cholinergic transmission in the deep layers of PFC is implicated in modulating the cortical tone because the PFC is the only neocortical region that projects back to subcortical regions such as the basal cholinergic nuclei and the monoaminergic brain stem nuclei (Uylings et al. 2003; Gabbott et al. 2005). In general, layer V is the main subcortical output channel of the neocortex and has a major role in seizure initiation and spread (Richardson et al. 2007). Recent results show that GABA release in the frontal cortex is sensitive to alterations of the nAChR expression, with ensuing epileptogenic effects. Rodents expressing mutant nAChR subunits linked to ADNFLE often present modified GABAergic transmission in the presence of tonically applied nicotine (Klaassen et al. 2006; Zhu et al. 2008).
By combining patch-clamp and immunocytochemical methods, we show that nAChRs can produce tonic control of GABA release at different levels. Heteromeric receptors regulate presynaptic terminals, whereas mixed receptor types control GABA release less directly through complex modulation of interneuronal activity, which appears to depend on the glutamatergic tone and include inhibition between interneurons.
Materials and Methods
Experiments were carried out according to the Principles of Laboratory Animal Care (directive 86/609/EEC). All efforts were made to minimize the number of animals used. Unless otherwise indicated, chemicals and drugs were purchased from Sigma-Aldrich, Milan, Italy.
Identification of Neocortical Areas
The different regions (prefrontal, motor, somatosensory, and visual) were identified by selecting appropriate sections in nonoverlapping ranges of rostrocaudal stereotaxic coordinates. According to Paxinos and Franklin (2001), the cortical sections for motor cortex range between +1.70 and +1.34 mm from bregma, and we examined the area indicated in the atlas as “primary motor cortex” (M1). For somatosensory cortex, sections were chosen between +0.50 and +0.02 mm from bregma (“primary somatosensory cortex,” S1). The visual cortex was analyzed in sections comprised between −2.46 and −2.80 mm from bregma (“primary visual cortex,” V1). As for PFC, no general agreement about nomenclature has yet been reached. For both morphological and electrophysiological experiments, we studied the dorsomedial shoulder region, often called frontal area 2 (Fr2), following the criteria recently applied for the rat PFC (Heidbreder and Groenewegen 2003; Uylings et al. 2003; Palomero-Gallagher and Zilles 2004). These authors adopt the nomenclature of Krettek and Price (1977), except that the term medial precentral area was substituted by Fr2 (Zilles 1985). Our sections were thus prepared between + 2.68 and +2.10 mm from bregma (for both immunocytochemical and electrophysiological experiments). We notice, however, that in the atlas of Paxinos and Franklin (2001), the area named Fr2 covers both the medial part of “frontal association cortex” (rostrally) and the “secondary motor area” (M2, caudally).
Cortical Tissue Preparation for Immunocytochemistry
FVB mice (Harlan, San Pietro al Natisone, Italy), aged 21–40 days, were anesthetized with intraperitoneal 4% chloral hydrate (2 mg/100 g) and sacrificed by intracardiac perfusion of 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.2–7.4). Brains were removed and immersed in the same solution at 4 °C. Coronal sections (50 μm thick) were cut with a VT1000S vibratome (Leica Microsystems, Mannheim, Germany). For each cortical region, 3–4 sections were selected for immunocytochemistry. Cytoarchitecture controls were performed on alternate sections adjacent to those processed for immunocytochemistry and stained with thionin (Paxinos and Franklin 2001).
Immunocytochemistry under Light and Electron Microscopy
After aldehyde quenching with NH4Cl and inactivation of endogenous peroxidases with H2O2, in standard phosphate buffer saline (PBS), sections were permeabilized for 30 min with 0.2% Triton X-100 in 10% normal goat serum (for immunocytochemistry of the vesicular ACh transporter, VAChT) or normal horse serum (for immunocytochemistry of choline acetyltransferase, ChAT). Sections for ultrastructural analysis were mildly pretreated with ethanol (10%, 25%, and 10%, 5 min each), instead of Triton X-100, to increase the immunoreagent penetration. Subsequently, they were incubated overnight at room temperature with one of the following primary antibodies: 1) anti-ChAT (polyclonal, made in goat, diluted at 1/500, Chemicon International, Temecule, CA) and 2) anti-VAChT (polyclonal, made in rabbit, diluted at 1/1000, Sigma-Aldrich). The biotinylated secondary antibodies (b-GAR: biotinylated anti-rabbit immunoglobulin G [IgG] made in goat; b-HAG: biotinylated anti-goat IgG made in horse; Vector Inc., Burlingame, CA, diluted 1/200) were then applied for 75 min. After washing, sections were treated with the avidin-biotinylated complex (ABC kit, Vector, Inc., diluted 1/100) and then with a freshly prepared solution of 0.075% 3-3′-diaminobenzidine tetrahydrochloride supplemented with 0.002% H2O2. For control, some sections were processed without primary antibody. In this case, no specific staining was ever observed. Sections were examined under a light microscope after dehydration and mounting with Permount (BDH, Poole, UK).
Ultrastructural analysis of the immunocytochemical reaction was carried out after fixation with 2.5% glutaraldehyde for 10 min followed by application of 1% OsO4 for 1 h. Sections were next dehydrated and flat embedded in Epon–Spurr between acetate sheets. After polymerization, selected areas were excised and glued to cured resin blocks. Semithin sections (1 μm thick) were cut with a Reichert ultramicrotome and collected on glass slides with or without toluidine blue counterstaining for inspection in light microscopy. Thin sections were counterstained with lead citrate or left unstained and examined with a Jeol T8 electron microscope.
Sections from different cortical areas were permeabilized for 30 min with 0.2% Triton X-100 and diluted in PBS containing 1% bovine serum albumin. They were next incubated overnight in a mixture of 2 primary antibodies or 1 primary antibody and NeuroTrace™ (NT; see Densitometric analysis) for cytoarchitecture analysis. Finally, a 2-h incubation was carried out in a mixture of the appropriate secondary antibodies at room temperature. Images were analyzed with a TCS SP2 AOBS laser scanning confocal microscope (Leica Microsystems). Control sections were processed without primary antibodies to test the specificity of the secondary antibodies.
Anti-VAChT and anti-ChAT were same as for light microscopy. Monoclonal mouse anti-neurofilament H, non phosphorylated (SMI32): monoclonal, made in mouse, diluted at 1/1000 (Sternberger Monoclonals Inc., Lutherville, MD); anti-parvalbumin (PV): polyclonal, made in rabbit, 1/1500 (Swant, Bellinzona, Switzerland); anti-vesicular GABA transporter (VGAT): polyclonal, made in rabbit, diluted at 1/800 (Synaptic Systems, Göttingen, Germany); anti-glutamic acid decarboxylase, 65 kDa form (GAD65): monoclonal, made in mouse, diluted at 1/300 (Chemicon); anti-α4 nAChR subunit: polyclonal, made in guinea pig, diluted at 1/500 (Chemicon); and anti-β2 nAChR subunit: polyclonal, made in rabbit, diluted at 1/200 (Immunological Sciences, Rome, Italy).
For anti-ChAT: biotinylated horse anti-goat IgG (bHAG; Vector, Inc.), diluted at 1/200, and Alexa-488–labeled streptavidin (1/200, Molecular Probes, Carlsbad, CA). For the monoclonal antibodies SMI32 and anti-GAD65: indocarbocyanine (Cy3)-conjugated donkey anti-mouse (DAM-Cy3; Jackson Immunoresearch Laboratories, West Grove, PA), diluted at 1/200. For VAChT, VGAT, PV, and β2 nAChR subunit: biotinylated goat anti-rabbit (b-GAR; Vector, Inc.), diluted at 1/200, and Alexa-488–labeled streptavidin (1/200, Molecular Probes). For α4 nAChR subunit: biotinylated donkey anti-guinea pig IgG (bDAGp; Vector, Inc.), diluted at 1/200, and Alexa-488–labeled streptavidin or Rhodamine RedX streptavidin (1/200, Molecular Probes).
Choice of nAChR Antibodies
We tested several monoclonal and polyclonal antibodies against the α4 and β2 nAChR subunits. In agreement with the results of Moser et al. (2007), most turned out to be unsuitable for immunocytochemistry in the murine cerebral cortex. The most reproducible results were obtained with AB5590 (Chemicon, anti-α4) and AB11687 (Immunological Sciences, anti-β2). AB5590 was previously used to verify Förster resonance energy transfer experiments on cultured midbrain neurons (Nashmi et al. 2003) and to study the nAChR expression in murine urothelium, in which α4 immunolabeling was independently validated by real-time polymerase chain reaction (Zarghooni et al. 2007). We confirmed that AB5590 binds α4-containing nAChRs by treating human embryonic kidney (HEK 293) cells transiently transfected with different combinations of α4, β2, α2, and β4 subunits, as described previously (De Fusco et al. 2000; Aridon et al. 2006). As expected, signal was only detected in transfected cells, with good specificity for α4-expressing cells (data not shown). Therefore, AB5590 does bind to α4-containing receptors, which is the relevant point for our purposes, although some cross-reaction with other nicotinic subunits or other proteins may also occur (Moser et al. 2007).
AB11687 was raised in rabbit immunized with a peptide analog of the β2 subunit C-terminus (residues 493–502). Again, we tested its specificity in HEK cells. Moreover, preadsorption assays were performed with the corresponding blocking peptide (β2 C-terminus; Immunological Sciences). Brain sections were treated with anti-β2, previously incubated (for 12–24 h at 4 °C, with mild agitation) with the specific peptide it recognizes (3- to 5-fold molar excess of synthetic peptide, relative to IgG concentration at the normal antibody dilution). No signal was detected after preadsorption, as expected.
Before applying the anti-VAChT antibody, histological staining was carried out with NT (Molecular Probes), an alkaline compound that gives results similar to those obtained with the Nissl method. Tissue was permeabilized with 0.2% Triton in PBS for 30 min. NT was applied for 40 min diluted at 1/50 in PBS. Sections were next treated with 0.1% Triton in PBS for 20 min. After this procedure, the usual VAChT immunofluorescence reaction was carried out. The pattern of cholinergic innervation in the adult was quantified on 3 P40 mice. For each animal, the prefrontal, motor, somatosensory, and visual areas were analyzed with Leica Confocal Software (LITE version, Leica Microsystems) on confocal microscope images acquired by using identical parameters. Consecutive nonoverlapping images (350 × 350 μm, ×40 objective) were acquired starting from the pial surface to fully cover each cortical region up to the subcortical white substance. Four images were necessary to cover the thickest cortices (motor and PFC), 3 for the somatosensory, and 2 for the visual cortices. Cortical layers were identified with NT counterstaining. By using the same software, a rectangular field was subsequently drawn in each image. The field area was about 8000 μm2, and care was taken to maintain its dimension below the limits of the smallest layers (I and IV). This field was then used to sample 3 regions of interest per layer, in which immunofluorescence was quantified. The same procedure was repeated for all cortical areas. For each animal, the mean fluorescence values of the different layers in each cortical region were normalized to the value obtained in the layer showing the highest signal. The values thus obtained were averaged between animals, and plotted in Figure 1, to convey an immediate comparison of the cholinergic innervation in different laminas.
Brain Slice Preparation for Electrophysiology
FVB mice (Harlan), aged 17–24 days, were deeply anesthetized with ether and decapitated. Brains were removed and placed in ice-cold solution containing the following (in millimolar): 87 NaCl, 21 NaHCO3, 1.25 NaH2PO4, 7 MgCl2, 0.5 CaCl2, 2.5 KCl, 25 D-glucose, and 75 sucrose, equilibrated with 95% O2 and 5% CO2 (pH 7.4). Coronal slices (300 μm thick) were cut with a VT1000S vibratome (Leica Microsystems) from the medial PFC and in particular from Fr2, as previously discussed. Slices were incubated at room temperature for at least 1 h, in the same solution as above, before being transferred to the recording chamber. During experiments, slices were superfused at 1.8 mL/min with artificial cerebrospinal fluid containing the following (in millimolar): 135 NaCl, 21 NaHCO3, 0.6 CaCl2, 3 KCl, 1.25 NaH2PO4, 1.8 MgSO4, and 10 D-glucose, aerated with 95% O2 and 5% CO2 (pH 7.4). Cells were examined with an Eclipse E600FN (Nikon Instruments, Sesto Fiorentino, Italy) equipped with a water immersion differential interference contrast (DIC) objective and an infrared (IR) CCD 100 camera (DAGE-MTI, Inc., Michigan City, IN).
For postrecording morphological characterization, pipettes contained 0.5% biocytin. Slices were fixed in phosphate buffer with 4% paraformaldehyde at 4 °C for 24–48 h. After embedding in 4% agar, 50-μm sections were cut with a VT1000S vibratome (Leica Microsystems) and collected in PBS. They were next incubated in Alexa-488–labeled streptavidin (1/200), in PBS supplemented with 0.2% Triton X-100, for 75 min. After washing with PBS, sections were mounted on gelatin-coated slides and analyzed with our confocal microscopy apparatus.
Neurons were voltage (or current) clamped with a Multiclamp 700A patch-clamp amplifier (Molecular Devices, Union City, CA) at room temperature. Low-resistance micropipettes (2–3 MΩ) were pulled from borosilicate capillaries with a P-97 Flaming/Brown Micropipette Puller (Sutter Instrument Company, Novato, CA). The cell capacitance and series resistance were always compensated. Experiments in which series resistance did not remain below 10 MΩ (typically 5–8 MΩ) were discarded. Input resistance was generally close to 100 MΩ. Synaptic currents were low-pass filtered at 2 kHz and digitized at 5 kHz with pClamp/Digidata 1322A (Molecular Devices). Pipettes contained (in millimolar) 140 Cs gluconate, 2 MgCl2, 10 4-2-hydroxyethyl-1-piperazine ethanesultanic acid (HEPES), and 2 MgATP (pH 7.2). Cs+ was used to inhibit the K+ currents when recording at +10 mV. To assess the health of our slices, we routinely tested a fraction of pyramidal cells in current-clamp mode by using pipettes containing (in millimolar) 135 K gluconate, 5 KCl, 2 MgCl2, 10 HEPES, and 2 MgATP (pH 7.2). Resting membrane potential was measured in open circuit mode soon after obtaining the whole-cell configuration. The membrane potential (Vm) values given in the text were not corrected for liquid junction potentials. Stock solutions of (−)-nicotine hydrogen tartrate salt, dihydro-β-erythroidine hydrobromide (DHβE), methyllycaconitine (MLA) citrate hydrate, (−)-bicuculline methiodide, D(−)-2-amino-5-phosphono-pentanoic acid (AP5), and tetrodotoxin (TTX; Tocris Bioscience, UK) were prepared in distilled water. Stock solutions of 20 mM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) were prepared in dimethyl sulfoxide. Agonists and antagonists were perfused in the bath and their effect calculated at the steady state, which was usually reached within 2 min. Only one cell per slice was treated with nicotine, as it is known that, because of nAChR desensitization, prolonged application of this compound exerts long-term effects on the nicotine sensitivity of central neurons (Mansvelder et al. 2002).
Analysis of Data
IPSCs were analyzed offline with Clampfit 9 (Molecular Devices) and OriginPro 8 (OriginLab, Northhampton, MA) software. The IPSCs included both smoothly shaped isolated signals with a peak amplitude larger than 40 pA and composite signals. Events were inspected one by one, and those not presenting the typical shape of synaptic currents were rejected. For analysis of the miniature IPSCs (mIPSCs), only events with a single peak amplitude higher than 10 pA were considered. The baseline noise (peak to peak) was about 5 pA. Statistical comparison of the cumulative distributions of IPSC amplitudes and interevent intervals in the different experimental conditions was carried out, for each cell, with the Kolmogorov–Smirnov test. For comparison of the results obtained in different cells, we also give the average frequency of events calculated for 2 min, at the steady state, in each experimental condition (e.g., Fig. 4D). In this case, data are given as mean values ± standard errors of the mean, with the number of experiments referring to neurons sampled in different brain slices. Statistical significance was determined with paired Student’s t-test. The level of significance was generally set to P < 0.05.
Cholinergic Innervation in the Murine Neocortex
By applying an immunoperoxidase method, the VAChT+ and ChAT+ fibers were found to be widely distributed throughout the neocortex. The laminar pattern showed minor differences between regions. In the S1, the supragranular layers were intensely stained. The label decreased slightly in layer IV and increased again in layers V and VI (Fig. 1A,B; the color version of Fig. 1 is found in Supplementary Material). Similar results were obtained in the visual cortex, whereas immunoreactivity was distributed more homogeneously in the motor cortex and PFC (data not shown).
A better quantification was achieved by studying the VAChT+ fibers with immunofluorescence in the PFC, motor, somatosensory, and visual cortices. The different layers were identified by staining the neuronal bodies with NT (Fig. 1C). Results are summarized in Figure 1D. Bars give the average VAChT immunofluorescence values in the indicated regions and layers. Values obtained in each animal were normalized to the highest signal for easier comparison. In general, innervation was denser in layers I/II. In the somatosensory cortex, immunoreactivity decreased in the deeper layers. A similar tendency, though less marked, was observed in the visual and motor cortices. The lowest signal was usually detected in layer IV. In the PFC, the cholinergic innervation density in layers V–VI was similar to the one observed in the upper layers, indicating that the cholinergic modulation presents similar levels of efficacy in input and projective layers in the associative areas.
Morphological and Ultrastructural Analysis
Irrespective of labeling method, the cholinergic terminals appeared like an intricate network of varicose, branching fibers across all the regions and layers we have examined (Fig. 1A,B). As occurs in other species (Avendaňo et al. 1996; Descarries et al. 2004), labeled cholinergic fibers resembled pearl necklaces in which the lace is the axon and pearls are axonal enlargements (varicosities) where the neurotransmitter is stored. The immunoreactive puncta, more often observed in the deep layers, are instead believed to identify classical synapses (Umbriaco et al. 1994; Turrini et al. 2001). Because bona fide synapses are below the resolution power of optical and confocal microscopy, we have further characterized the cholinergic terminals by ultrastructural analysis.
Sections were treated with antibodies against VAChT or ChAT and labeled with an immunoperoxidase method. In all layers, the reaction product was stored in numerous neuronal processes of different sizes, filled with vesicles, and rich in mitochondria (Fig. 2A,C,G). Synaptic contacts were more frequent in deep layers but were overall very rare (Fig. 2B,E). In layer V, in particular, small and medium caliber axonal processes were usually adjacent to unlabeled boutons forming symmetric (i.e., inhibitory) synapses on neuronal cell bodies (Fig. 2C,D) or distal dendrites (Fig. 2F,G). These results support the notion that the cholinergic control of neocortical GABAergic synapses includes a significant paracrine component.
GABAergic Postsynaptic Currents in Layer V Pyramidal Neurons
Following the lead of the morphological data, we studied whether and how tonic stimulation of nAChRs modulates GABA release onto pyramidal neurons. Pyramidal neurons were inspected with IR/DIC microscopy and distinguished on the basis of their laminar location and shape. Postrecording morphological reconstruction of cells loaded with biocytin usually showed the typical apical and basal dendrites with numerous spines and a pyramid-like cell soma (Fig. 3A).
A fraction of these neurons were routinely tested in current-clamp mode. In response to depolarizing current pulses (Fig. 3B), most of them exhibited the classic low-frequency action potential firing with adaptation (Porter et al. 1999; Chang and Luebke 2007). They also presented a slight time-dependent inward rectification, which is usually attributed to activation of IH-type currents at hyperpolarized membrane potential (Vm; e.g., Yang et al. 1996). In the series of experiments illustrated in Figure 3, the average resting Vm was −70.9 ± 0.59 mV (n = 11). Similar results were obtained in different mice litters.
In such cells, we studied the spontaneous GABAergic currents (IPSCs) in voltage-clamp mode. The common term IPSC is used for reader's convenience, although we are aware that, even in the adult, GABA release does not necessarily have inhibitory effects (Marty and Llano 2005). The current–voltage relation was determined between −110 and +10 mV (Fig. 3C), in the presence of AP5 (50 μM) and CNQX (10 μM), to inhibit N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptors, respectively. The remaining currents were sensitive to the GABAA receptor inhibitor bicuculline (10 μM). Moreover, their reversal Vm was close to −60 mV (Fig. 3D), in agreement with a permeability ratio around 10 for Cl− and gluconate and 2 for Cl− and HCO−3, in GABAA receptors (Fatima-Shad and Barry 1993). IPSCs such as these were observed in all the pyramidal neurons we have tested (n = 177). In the following experiments, we illustrate IPSCs recorded at +10 mV to maximize their amplitude and minimize the contribution of excitatory postsynaptic currents. Patch pipettes contained cesium to block the voltage-gated K+ currents. The synaptic currents recorded at +10 mV were completely inhibited by 10 μM bicuculline (Fig. 4A) and were insensitive to AP5 and CNQX (Fig. 8). These experimental conditions allowed us to avoid systematic use of ionotropic glutamate receptor blockers, which were only applied during the experiments shown in Figure 8.
Nicotine Increased the Frequency of IPSCs
To activate nAChRs, nicotine was preferred to ACh. Using the latter requires simultaneous application of muscarinic receptor inhibitors, whose effects on nAChRs are not fully defined. Atropine, for example, affects different nAChR types at concentrations as low as 100 nM (Zwart and Vijverberg 1997). Figure 4A,B illustrates, respectively, the IPSC traces and the time course of the events’ frequency in a typical experiment in which 5 μM nicotine was applied through the bath. Nicotine almost doubled the IPSC frequency, in a reversible way, as is also clear from the corresponding distribution of the interevent intervals (Fig. 4D).
Stimulation usually peaked within 2 min and, on average, brought the events’ frequency from 1.55 ± 0.47 to 2.92 ± 0.97 Hz (n = 8; P < 0.05 with t-test; Fig. 4E). On the other hand, nicotine scarcely affected the IPSC amplitude, as is apparent from the cumulative distribution plot shown in Figure 4C. On average, the mean amplitude of IPSCs was 47.86 ± 8.82 pA before nicotine application and 45.17 ± 7.21 pA in the presence of nicotine (n = 8; not significantly different with t-test). In the long run, a progressive reduction of the IPSC amplitude was often observed, which was presumably caused by a partial depletion of the GABA-containing vesicles (Bekkers 2005). The threshold for the effect of nicotine was between 0.1 and 1 μM (Fig. 4F), in agreement with the typical ACh and nicotine levels that have been measured in the cerebrospinal fluid, respectively, during cholinergic fiber stimulation and smoking (Alkondon, Pereira, Almeida, et al. 2000; Pepeu and Giovannini 2008). The maximal effect was observed at 5 μM nicotine (full statistics are given in the legend to Fig. 4), which is close to the concentration that produces the peak of the “window current” for both α4β2 and α7 receptors (Fenster et al. 1997). Around this ligand concentration, the steady-state current carried by nAChRs is maximal. Hence, the fact that higher doses produced no further stimulation of GABA release likely depends on a higher contribution of nAChR desensitization. Below, we mostly illustrate experiments carried out by applying 5 μM nicotine in order to elicit the maximal steady-state response from both heteromeric and homomeric nAChRs.
Presynaptic Nicotinic Modulation of GABA Release onto Pyramidal Neurons
The nicotinic regulation of GABA release is potentially complex because of nAChR expression on interneuronal somata and terminals. Moreover, ACh could exert indirect control by modulating the excitatory and inhibitory fibers that form synapses on interneurons. As mentioned earlier, nAChRs are known to be activated by somatic application of nicotine on several populations of interneurons in layer V of PFC. On the other hand, whether presynaptic nAChRs modulate the GABAergic terminals has not been previously investigated in this region. Because it is presently impossible to obtain direct recordings from these synaptic boutons, we have tested the effect of nicotine indirectly by measuring mIPSCs on pyramidal neurons. Miniature events were isolated by blocking action potentials with 0.5 μM TTX and 50 μM Cd2+, which inhibit, respectively, voltage-gated fast Na+ channels and high-threshold voltage-gated Ca2+ channels (e.g., Klaassen et al. 2006). As expected, perfusion with TTX and Cd2+ strongly decreased both the frequency and the amplitude of synaptic events. Figure 5A shows current traces from a typical experiment. Current traces from a representative neuron are illustrated in Figure 5A. Once again, all events were inhibited by bicuculline.
The typical time course of our experiments is illustrated in Figure 5B. The corresponding cumulative distributions for current amplitudes and interevent intervals are plotted, respectively, in Figure 5C,D. These panels show that 5 μM nicotine significantly increased the mIPSC frequency, with scarce effect on their amplitudes. On average, nicotine increased the frequency of mIPSCs by about 40% (Fig. 5E). Statistics are given in the legend to Figure 5. The relation between nicotine concentration and the percent increase in the mIPSC frequency (Fig. 5F) was broadly similar to the one observed for the spontaneous IPSCs, although the overall effect was weaker. We conclude that the nicotinic control of GABA release includes a presynaptic component.
The Nicotinic Stimulation of mIPSCs Was Mediated by Heteromeric nAChRs
The heteromeric α4β2 and the homomeric α7 nAChRs can be distinguished by application of 1 μM DHβE (selective antagonist for the former) or 10 nM MLA (selective antagonist for the latter; Alkondon et al. 1999; Maggi et al. 2001). As shown in Figure 6A,D, after TTX and Cd2+ had revealed the mIPSCs, the antagonists were added for 3–4 min to obtain full equilibration before nicotine addition.
When 5 μM nicotine was then added to DHβE, the usual stimulatory effect was prevented (Fig. 6A; the corresponding distribution of amplitudes and interevent intervals are shown in Fig. 6B,C). In contrast, MLA did not impede the effect of nicotine on the events’ frequency, as illustrated in Figure 6D,E,F. The results of these experiments are summarized in Figure 6G (for DHβE) and Figure 6H (for MLA). On average, in the presence of MLA, nicotine increased the mIPSC frequency by about 50% (Fig. 6H; statistics are given in the legend to Fig. 6), whereas DHβE effectively antagonized the action of nicotine (Fig. 6G). No different results were obtained by applying the same inhibitors in the presence of higher concentrations of nicotine (up to 100 μM; data not shown). These data indicate that the steady-state effect of nicotine on GABAergic presynaptic terminals is mostly mediated by heteromeric nAChRs.
Hetero- and Homomeric nAChRs Cooperate in Modulating the IPSCs
We next measured the effect of nAChR inhibitors on the IPSCs, that is, in the absence of TTX and Cd2+. Application of either DHβE (1 μM) or MLA (10 nM) in the absence of nicotine produced no effect on the IPSC frequency (not shown). This agrees with the data shown in Figure 6 for the miniature events and rules out spurious effects of our inhibitors. In the presence of 5 μM nicotine, both DHβE (Fig. 7A) and MLA (Fig. 7B) counteracted the normal stimulation of the IPSCs. Hence, differently from the modulation of miniature events, mixed nAChR types participate in the overall control of GABA release. Moreover, in the presence of nicotine, DHβE produced significant inhibition of the IPSC frequency (Fig. 7C). A tendency to inhibition of GABA release may also occur with MLA, although the effect was not statistically significant (Fig. 7D). This is an indirect evidence that, in certain experimental conditions, nicotine can produce inhibition of GABA release onto pyramidal neurons. An interpretation of these results is suggested below.
Nicotine Inhibited the Spontaneous IPSCs when Ionotropic Glutamate Receptors Were Blocked
In layer V of the murine PFC, nAChRs are known to modulate glutamate release from thalamocortical terminals (Lambe et al. 2003). Interneurons can thus be activated by the nAChR-dependent increase of glutamatergic transmission, which mostly targets fast-spiking (FS) PV+ cells in the neocortex of rats and mice (Gibson et al. 1999; Couey et al. 2007; Kruglikov and Rudy 2008). To exclude this component of the effect of nicotine, we monitored the IPSC frequency in the presence of CNQX and AP5, which inhibit the ionotropic glutamate receptors. These inhibitors did not alter significantly the basal IPSC frequency, showing that the spontaneous interneuronal activity was scarcely affected by background glutamate release in our slices. However, when 5 μM nicotine was added to CNQX and AP5, the frequency of IPSCs turned out to decrease in a reversible way (Fig. 8A). The corresponding cumulative distributions of amplitudes and interevent intervals are given, respectively, in Figure 8B,C. The overall results of these experiments are summarized in Figure 8D, which shows the average steady-state IPSC frequencies, in the indicated conditions. Thus, nAChR activation tends to inhibit GABA release when the AMPA and NMDA receptors are blocked. This effect was prevented by either DHβE (Fig. 8E) or MLA (Fig. 8F). Detailed statistics for panels 8D, E, and F are given in the legend to Figure 8.
These results indicate that nAChR activation can produce stimulatory or inhibitory effects in layer V depending on the level of glutamatergic activation. Our interpretation is as follows (see the schematic diagram in Fig. 10). In basal conditions, nicotine stimulates glutamate release onto basket (FS) cells. This effect tends to dominate and produces overall inhibition of pyramidal neurons, which may be reinforced by direct activation of nAChR expressed in GABAergic terminals (Fig. 5). When the glutamatergic transmission is blocked, the remaining effect of nicotine on interneurons depends on 2 main components. First, activation of somatic nAChRs (both hetero- and homomeric), which are mainly expressed by non-FS interneurons, in layer V (Couey et al. 2007). Because frequent chemical transmission occurs between non-FS and FS interneurons in the rodents’ neocortex (Gibson et al. 1999), the most likely effect of nicotinic stimulation of these neurons appears to be inhibition of other interneurons, in agreement with the decrease of GABA release onto pyramidal cells (Fig. 8). A second possible way through which nicotine can produce inhibition of GABA release is direct activation of reciprocal GABAergic terminals between interneurons. The efficacy of this mechanism depends on the density of reciprocal interneuronal innervation and on the level of expression of nAChR therein. In general, to clarify the relative importance of the physiological mechanisms discussed above, it is necessary to determine more precisely the innervation pattern and the nAChR distribution in relation with different cell types.
Cholinergic Innervation of Pyramidal Cells and Interneurons in Layer V
Pyramidal neurons were identified with SMI32, which labels the nonphosphorylated epitope of the neurofilament heavy subunit, mainly expressed in the soma and proximal processes of principal cells (Kirkcaldie et al. 2002). The GABAergic interneurons were distinguished with antibodies against PV, a typical marker of basket cells, and GAD65, an enzyme expressed in most of the GABA-synthesizing terminals (Fig. 9D) and occasionally also in the neuronal somata (Soghomonian and Martin 1998). We focused on PV because current evidence suggests that, in layer V, PV+ interneurons largely prevail (Hof et al. 1999; Galarreta and Hestrin 2002; Gonchar et al. 2008; Kruglikov and Rudy 2008; Uematsu et al. 2008). VAChT+ (or ChAT+) varicosities rarely appeared in proximity of the SMI32+ somata of pyramidal neurons (Fig. 9A,B). On the other hand, cholinergic fibers were often observed to follow the apical dendrites of pyramidal cells (Fig. 9A), especially in the supragranular layers. Double immunostaining for VAChT (or ChAT) and PV (or GAD65) showed that the cholinergic fibers frequently lay near both interneuronal cell bodies (see the arrowhead in Fig. 9B) and their neuropilar elements (Fig. 9C), which typically surrounded as a basket the pyramidal cell bodies (Fig. 9B,C). Thus, in the deep cortical layers, the cholinergic fibers profusely innervate both the soma and the synaptic terminals of GABAergic interneurons. This agrees with the dual nature (pre- and postsynaptic) of the nicotinic effects, indicated by the patch-clamp experiments. On the other hand, in pyramidal neurons, ACh release appears to mainly target the dendritic arborization rather than the somata.
Distribution of nAChRs in Layer V
Altogether, our results indicate that nAChRs are expressed by GABAergic interneurons at both the somatic and terminal levels in murine PFC layer V. As is well known, nAChR localization with antibodies in brain slices is not completely reliable because of cross-reactivity with different subunits and other undefined proteins (Jones and Wonnacott 2005; Moser et al. 2007). Nonetheless, we deemed it useful to study the nAChR distribution with antibodies against the α4 and the β2 subunit for comparison with the other methods we have applied. We used AB5590 (anti-α4) and AB11687 (anti-β2), which we characterized as discussed earlier. By double immunofluorescence experiments (not shown), we observed that both these antibodies produced a labeling pattern quite similar to the previous immunocytochemical localization on rat cerebral cortex (Hill et al. 1993; Nakayama et al. 1995). Signal was observed on both neuronal somata and processes throughout the neocortex, with a laminar distribution that paralleled the one displayed by ChAT and VAChT. For both principal cells and interneurons, immunolabeling of cell somata prevailed in layers II/III and V, whereas the neuropilar signal was denser in the upper layers.
Our patch-clamp experiments indicate that presynaptic heteromeric nAChRs control GABA release onto pyramidal cells. We have thus analyzed the coimmunofluorescence signals produced by the above antibodies and 1) VGAT (to mark the GABAergic terminals) and 2) GAD65 plus PV (to mark both GABAergic cell bodies and terminals). Numerous VGAT+ fibers were also labeled by the anti-nAChR antibodies (Fig. 9E). The GAD65+ (or PV+) terminals often surrounded pyramidal neurons (Fig. 9D,F) and contained the 2 nicotinic subunits, which were also found in the GAD65+ and PV+ interneuronal somata (Fig. 9D,F). Hence, in keeping with the results obtained with VGAT, the PV+ terminals surrounding the pyramidal neuron somata (Fig. 9F) were often immunolabeled for the nAChR subunits, in agreement with the electrophysiological data.
The ascending cholinergic system diffusely releases ACh throughout the neocortex and thalamus. ACh modulates the cortical tone on a tenths of minutes to hours scale, thus regulating vigilance and the sleep-waking cycle. More localized and briefer (down to seconds) effects are exerted on the efficiency of neocortical information processing (Mansvelder et al. 2006; Sarter et al. 2006; Parikh et al. 2007). This functional complexity is brought about by the combination of a wide array of receptor subtypes (both ionotropic and metabotropic) localized at both synaptic and nonsynaptic sites. Despite a vast literature about the pharmacological and behavioral aspects, physiological insight on cerebral cholinergic transmission is still fragmentary. For the murine neocortex, even the broad features of cholinergic innervation have received scarce attention, which is unfortunate considering the increasing use of murine models of human cognition and neuropathology.
Our immunocytochemical results show a diffuse cholinergic innervation in different regions and layers. Electron microscopy indicates that the cholinergic varicosities are frequently adjacent to the GABAergic neurons, but their location is usually extrasynaptic. These observations prompted us to investigate the tonic effects of nicotinic stimulation on GABA release by applying patch-clamp methods to layer V pyramidal neurons. Nicotine turned out to stimulate the spontaneous GABA release on pyramidal neurons, with a threshold around 1 μM. The global effect appears to depend on both heteromeric and homomeric nAChRs, but analysis of the mIPSCs revealed that direct effect on GABAergic boutons is mostly exerted by heteromeric receptors. The expression of nAChRs on GABAergic cell bodies and terminals was confirmed by immunocytochemistry. In contrast, in the presence of inhibitors of the ionotropic glutamate receptors, activation of nAChRs tended to decrease the IPSCs measured in pyramidal cells, suggesting that nAChRs can regulate both excitation and inhibition of interneurons depending on the level of glutamatergic input.
Cholinergic Innervation in the Murine Neocortex
In rodents and cats, diffuse cortical cholinergic innervation was originally observed by labeling ChAT (Eckenstein et al. 1988; Kitt et al. 1994; Avendaňo et al. 1996). Subsequent studies have mostly focused on VAChT, which marks more reliably the cholinergic terminal fields (Arvidsson et al. 1997; Ichikawa et al. 1997). In the mouse, however, only the parietal and motor cortices have been studied in this way (Wong et al. 1999). By comparing the distribution of ChAT and VAChT, we have observed a profuse cholinergic innervation across the somatosensory, visual, motor, and prefrontal murine cortices. The signal tended to decrease in deeper layers, but not in the PFC, which underlines the relevance of ACh-dependent modulatory processes in both input and output channels of the associative areas. The pattern we have observed with the immunoperoxidase method is in broad agreement with the results obtained in the rat (Eckenstein et al. 1988; Avendaňo et al. 1996; Mechawar et al. 2000; Henny and Jones 2008). Irrespective of labeling method, the cholinergic terminals appeared like an intricate network of varicose, branching fibers. In all cortical regions, a significant incidence of immunoreactive puncta, likely to be classical synapses, was only detected in the deeper layers. Such a pattern was confirmed by electron microscopy, which highlighted only rare classical synaptic contacts, usually in layers V–VI. These results are consistent with recent observations in the rat, in which most of the neocortical VAChT+ terminals are not associated with typical postsynaptic proteins (Henny and Jones 2008). They support the notion that paracrine ACh release is a prominent feature of cholinergic transmission in the cerebral cortex.
Association between Cholinergic Fibers and Neuronal Types in Layer V
Immunofluorescent staining of VAChT and ChAT was also used to study the relationship between cholinergic fibers and the 2 main neuronal populations in layer V: pyramidal neurons and basket cells. VAChT+ and ChAT+ varicosities rarely appeared in proximity of the pyramidal neuron somata (stained with SMI32), whereas they often accompanied the course of apical dendrites in the supragranular layers. In contrast, proximity between cholinergic fibers and the cell bodies of pyramidal neurons is frequent in the rat (Henny and Jones 2008). Such a difference agrees with the electrophysiological results that we have previously discussed. Scarce success is generally encountered in activating nicotinic currents from pyramidal neurons in the murine PFC (Couey et al. 2007), whereas in rats, clear nicotinic currents are recorded from these cells in frontal and somatosensory cortical layer V (Zolles et al. 2009) and PFC layer VI (Kassam et al. 2008). All these studies were performed during the first postnatal month and suggest that species-specific differences in the nAChR expression by pyramidal neurons occur during neocortical development. Differences between rats and mice have also been observed in the maturation of intrinsic neocortical cholinergic cells, which is slower in mice (Mechawar and Descarries 2001; Consonni et al. 2009). The precise function of these cells is unknown, but they are thought to be implicated in the regulation of pyramidal neuron excitability (von Engelhardt et al. 2007). Hence, the maturation of the nAChR-dependent control of principal cells may be delayed in the murine neocortex.
On the other hand, the association between the cholinergic fibers and the GABAergic system in the mouse showed a pattern similar to the one observed in the rat. We found in particular that the cholinergic varicosities often followed the PV+ terminals, suggesting a strict relationship between the cholinergic and the GABAergic systems. Likewise, in the rat PFC, the VAChT+ axonal varicosities in the fibers arising from basal forebrain are often placed in close proximity to PV+ cell bodies and processes (Henny and Jones 2008). These results are consistent with the electrophysiological similarities in the nicotinic control of GABA release in different mammals.
Nicotinic Control of the GABAergic Terminals
To the best of our knowledge, the presynaptic nAChR expression in GABAergic terminals was not previously studied in PFC layer V. Our conclusion that part of the nicotinic regulation of GABA release occurs at the presynaptic level is based on the following evidence: 1) nicotine increased the frequency of mIPSCs; 2) stimulation was prevented by a specific inhibitor of neuronal nAChRs; 3) the cholinergic fibers innervated both the soma and the terminals of basket cells; 4) electron microscopy revealed cholinergic varicosities in the vicinity of symmetric synapses; and 5) antibodies against nAChR subunits stained the GABAergic terminals labeled with different markers, such as VGAT, PV, and GAD65. Because the effect of nicotine on mIPSCs was prevented by DHβE, but not MLA, we conclude that heteromeric nAChRs prevail in GABAergic terminals. The lack of very specific antibodies for nAChR subunits prevents us from definitely exclude that α7 is expressed in basket cells’ terminals. Nevertheless, its functional contribution at the steady state should be minor. As we mentioned earlier, the large majority of interneurons in PFC layer V is constituted by PV+ FS cells. The other interneuron types have different discharge patterns and tend to express colecystokinin (CCK), although at different levels. Recent work (Couey et al. 2007) shows that nicotinic currents can be measured in CCK+ cells, which express α4, β2, and α7 nAChR subunit mRNAs. FS interneurons present instead negligible somatic nicotinic currents and express much lower levels of messenger mainly for β2. Expression of the latter agrees with our conclusion that these cells express presynaptic heteromeric nAChRs. If this also applies to other interneuronal types remains to be determined. A diagram of these results is given in Figure 10.
Global Nicotinic Control of Interneuronal Activity
In agreement with literature, we observed that nicotine stimulates the IPSCs registered on pyramidal neurons. A major component of this effect is thought to depend on nAChR-dependent stimulation of glutamate release from thalamocortical terminals (Vidal and Changeux 1993; Gioanni et al. 1999; Lambe et al. 2003), which drives GABA release mostly from PV+ FS cells (Gibson et al. 1999; Couey et al. 2007; Kruglikov and Rudy 2008). A supplementary component depends on direct excitation of the non-FS interneurons that innervate pyramidal cells, although the balance of the action of these cells on pyramidal and FS neurons is presently unclear. The fact that both DHβE and MLA prevented the effect of nicotine on the spontaneous IPSCs agrees with previous work (Couey et al. 2007) and with the conclusion that both the above mechanisms contribute to the overall effect. Nevertheless, DHβE was generally more effective and actually produced significant inhibition of GABA release on pyramidal cells in the presence of nicotine. We attribute the higher efficacy of DHβE to 2 reasons. First, the contribution of heteromeric nAChRs to the glutamate release from thalamocortical fibers is thought to be higher than that of homomeric receptors (Lambe et al. 2003). Second, in the steady-state conditions that we have adopted, the fraction of active heteromeric receptors is likely to be higher because of their slower rate of desensitization.
Finally, in the presence of CNQX and AP5, nicotine turned out to partially inhibit GABA release onto pyramidal cells. This suggests that, in these conditions, reciprocal inhibition between interneurons tends to prevail on stimulation of the GABAergic terminals that innervate principal cells. Recent work has clearly shown that inhibitory synapses between interneurons are very widespread in the neocortex of both rats and mice (Galarreta and Hestrin 1999, 2002; Gibson et al. 1999). Our results suggest that this process is regulated by nAChRs, and we propose 2 possible mechanisms (not mutually exclusive). When glutamatergic transmission is abolished, inhibition between interneurons may prevail because of nicotinic excitation of the cell bodies of the non-FS cells that innervate the FS (PV+) cells. We found this hypothesis on 2 lines of evidence. First, both low-threshold and regular-spiking interneurons often form GABAergic synapses onto FS–PV cells (Gibson et al. 1999). Second, both low-threshold and regular-spiking cells express somatic nicotinic currents caused by expression of α4, α7, and β2 nAChR subunits (Couey et al. 2007). Another possible mechanism to explain our results is suggested by the observation that most of the FS interneurons present reciprocal inhibitory innervation (Galarreta and Hestrin 1999, 2002; Gibson et al. 1999). Therefore, when basket cells are stimulated, a delicate balance must occur between stimulation of GABA release on pyramidal cells and reciprocal inhibition between interneurons. In normal conditions, the former tends to prevail, as indicated by current literature in brain slice preparations. However, our results suggest that when the glutamatergic input decreases, the action of nAChRs on presynaptic terminals may shift the overall balance toward interneuronal inhibition. A scheme of these mechanisms is illustrated in Figure 10. Interestingly, in layers II/III of the murine PFC, the action of nicotine on GABA release onto pyramidal cells remains stimulatory, even when the ionotropic glutamate receptors are inhibited (Klaassen et al. 2006). At the present stage, we attribute this difference to the considerably higher density of reciprocal inhibitory connections in layer V (Douglas and Martin 2004). Regardless, our observations indicate that the equilibrium between the stimulation and inhibition of GABA release caused by nAChR activation is delicate, in layer V, and depends on the general level of glutamate release.
Implications for Epileptogenesis
We have focused our electrophysiological experiments on layer V of the Fr2 region of the medial PFC. This region receives dense innervation from both thalamic and neocortical regions. In particular, it receives afferents from primary and secondary motor cortices and primary and secondary somatosensory, insular, auditory, and occipital areas. Because of these strong intracortical connections, Fr2 is thought to be particularly involved in motor control and spatial orientation (Gabbott et al. 2005). This region may thus be a good model for understanding the frontal sleep-related epileptic forms because hyperkinetic seizures during sleep are a distinctive feature of ADNFLE (Picard and Brodtkorb 2007). It is often observed that mutant nAChR subunits linked to ADNFLE tend to increase the receptor's function in expression systems (De Fusco et al. 2000; Phillips et al. 2001; Aridon et al. 2006; Hoda et al. 2008). Because, however, nAChR activation in the frontal cortex stimulates GABAergic transmission, it is necessary to explain how hyperfunctional nAChRs could lead to network hypersensitivity instead of depression. In mice and rats, expressing some of these mutant subunits did not produce consistent results on GABA release in the PFC (Klaassen et al. 2006; Zhu et al. 2008). At the present stage, it would be premature to decide whether these differences depend on the region- and species-specific differences in neocortical cholinergic control that we have discussed earlier or on the special properties of different mutant channels. Nonetheless, available evidence suggests at least 2 possible mechanisms through which hyperfunctional nAChRs could stimulate excitability. Basing on work in mice, it has been proposed that such mutant nAChRs could abnormally raise GABA release when ACh increases, particularly during the transition between non-REM sleep and wakefulness or REM sleep. The consequent hyperpolarization would partially remove inactivation in low-threshold voltage-gated Ca2+ channels and activate IH currents, thus making pyramidal neurons more sensitive to postinhibitory rebound (Klaassen et al. 2006). A second possibility is suggested by our results. During slow-wave sleep, thalamocortical neurons tend to be inhibited (Steriade and McCarley 2005). In this state, an upsurge of ACh in the neocortex may shift the balance of the basket cells’ network toward reciprocal inhibition, thus facilitating seizure development, before the parallel ACh release in the thalamus stimulates thalamocortical activity. The subtle equilibrium between these neocortical and thalamic actions of ACh may account for the partial nature of ADNFLE crises and for the fragmentation and instability of sleep typically observed in sleep-related frontal epilepsy.
Italian Telethon Foundation (GP0023Y01); Italian Ministry for University and Scientific Research (Programmi di Ricerca di Rilevante Interesse Nazionale 2005); Fondazione Banca del Monte di Lombardia; Ingenio Program of Lombardy Region (fellowship to R.M.).
We thank Prof. Enzo Wanke for advice and support. Thanks are also due to Dr Roberto Bonardi for help in densitometric analysis. Conflict of Interest: None declared.