(R,S)-3,5-DHPG – mTOR signal

Functional and ultrastructural analysis of group I mGluR in striatal fast-spiking interneurons

Keywords: GABAergic interneuron, mGluR1, mGluR5, rat, striatum

Abstract

Striatal parvalbumin-containing fast-spiking (FS) interneurons provide a powerful feedforward GABAergic inhibition on spiny projection neurons, through a widespread arborization and electrical coupling. Modulation of FS interneuron activity might therefore strongly affect striatal output. Metabotropic glutamate receptors (mGluRs) exert a modulatory action at various levels in the striatum. We performed electrophysiological recordings from a rat striatal slice preparation to investigate the effects of group I mGluR activation on both the intrinsic and synaptic properties of FS interneurons. Bath-application of the group I mGluR agonist, (S)-3,5- dihydroxyphenylglycine (3,5-DHPG), caused a dose-dependent depolarizing response. Both (S)-(+)-a-amino-4-carboxy-2-methyl- benzeneacetic acid (LY367385) and 7-(hydroxyimino)cyclopropa[b]chromen-1a-carboxylate ethyl ester (CPCCOEt), selective mGluR1 antagonists, significantly reduced the amplitude of the membrane depolarization caused by 3,5-DHPG application. Conversely, mGluR5 antagonists, 2-methyl-6-(phenylethylnyl)pyridine hydrochloride (MPEP) and 6-methyl-2-(phenylazo)-3-pyridinol (SIB1757), were unable to affect the response to 3,5-DHPG, suggesting that only mGluR1 contributes to the 3,5-DHPG-mediated excitatory action on FS interneurons. Furthermore, mGluR1 blockade significantly decreased the amplitude of the glutamatergic postsynaptic potentials, whereas the mGluR5 antagonist application produced a small nonsignificant inhibitory effect. Surprisingly, our electron microscopic data demonstrate that the immunoreactivity for both mGluR1a and mGluR5 is expressed extrasynaptically on the plasma membrane of parvalbumin-immunoreactive dendrites of FS interneurons. Together, these results suggest that despite a common pattern of distribution, mGluR1 and mGluR5 exert distinct functions in the modulation of FS interneuron activity.

Introduction

The basal ganglia circuitry is critically involved in cognition and control of voluntary movement (Graybiel et al., 1994; Graybiel, 1995). This complex system is composed of several nuclei forming a highly interconnected network, within which the striatum represents the main input station. By inhibiting its proximal targets, the striatum controls the basal ganglia output to the thalamus and brainstem. This striatal inhibitory output is provided by a large population (90–95% in rodents) of GABAergic spiny projection neurons. The activity of spiny neurons is regulated by massive excitatory glutamatergic inputs from the cortex or thalamus and a strong intrinsic GABAergic inhibitory transmission. The latter is mediated by two components: a relatively weak lateral inhibition induced by axon collaterals of spiny neurons themselves and a strong feedforward interneuronal inhibition (Koos & Tepper, 1999; Plenz, 2003; Tepper et al., 2004; Mallet et al., 2005; Gustafson et al., 2006). Four populations of interneurons have been described in the striatum; the cholinergic large aspiny neurons and three populations of GABAergic interneurons, that have been classified, according to their electrophysiological and histochemical properties, as persistent and low-threshold spike calretinin-positive interneurons (PLTS), low-threshold spike nitric oxide-containing neurons (LTS) and parvalbumin (PV)-immunoreactive fast-spiking (FS) interneurons (Kita et al., 1990; Kawaguchi, 1993; Kawaguchi et al., 1995). Though each of the GABAergic interneuron subpopu- lations produces an inhibitory postsynaptic potential in striatal projection neurons (Tepper & Bolam, 2004), some peculiar features of FS interneurons make them likely candidates for the feedforward inhibitory control over spiny neurons. Indeed, FS interneurons have a widely divergent output, each innervating over one hundred projection neurons, and every spiny neuron is contacted by more than one PV-containing cell. These properties, in addition to electrotonic coupling (Kita et al., 1990; Koos & Tepper, 1999), render FS interneurons well suited for amplifying and distributing, over a wide area, the effects of striatal neurotransmitters and modulators through a synchronous, powerful and temporally coded inhibitory action on spiny projection neurons (Koos & Tepper, 1999; Gustafson et al., 2006; Mallet et al., 2006). Members of group I, II and III meta- botropic glutamate receptors (mGluRs) exert complex modulatory effects on striatal function by acting either at the presynaptic or postsynaptic level (Lovinger & McCool, 1995; Colwell et al., 1996; reviewed in Pisani et al., 2003; Gubellini et al., 2004). Striatal spiny and cholinergic neurons show a high coexpression of group I mGluRs, mGluR1 and mGluR5, a peculiar feature mostly restricted to basal ganglia neuronal populations (reviewed in Valenti et al., 2002). Functionally, the complex interactions taking place between mGluR1 and mGluR5 in different neuronal subtypes specifically shape cell activity (Poisik et al., 2003; Bonsi et al., 2005). Though the role of group I mGluRs has been investigated both in spiny projection neurons and cholinergic interneurons, their expression and function in the remaining striatal subpopulations has not been characterized yet. To address this issue, we studied the cellular and subcellular expression of group I mGluRs, and characterized the synaptic and postsynaptic responses of FS striatal interneurons to selective activation of mGluR1 and mGluR5 in rats.

Materials and methods Electrophysiology experiments Tissue preparation

Male Wistar rats (3–4-weeks old) were used for the experiments. All efforts were made to reduce the number of animals used. The animal experiments were carried out in accordance with the guidelines of the European Communities Council Directive (86 ⁄ 609 ⁄ EEC). As previ- ously described (Pisani et al., 1997; Bonsi et al., 2005), rats were killed by cervical dislocation under deep anaesthesia and the brain rapidly removed from the skull. Coronal corticostriatal slices (180– 200 lm) were cut in oxygenated (95% O2, 5% CO2) Krebs’ solution by using a vibratome. The composition of the solution was (in mM) 126 NaCl, 2.5 KCl, 1.3 MgCl2, 1.2 NaH2PO4, 2.4 CaCl2, 10 glucose,
18 NaHCO3. After recovering for 30 min in oxygenated Krebs’ solution (32 °C), slices were transferred in a recording chamber (0.5– 1 mL volume), placed on the stage of an upright microscope (BX51WI, Olympus, Milan, Italy) equipped with a 20·, 0.95 n.a. water immersion objective (XLUMPlan Fl, Olympus). During the electrophysiological recording, slices were submerged by temperature- controlled (32–33 °C) oxygenated Krebs’ solution, flowing at 2.5–3 mL ⁄ min.

Electrophysiological recordings

Conventional intracellular recordings in the current-clamp mode were performed using sharp microelectrodes filled with 2 M KCl (40– 60 MW). Signal acquisition and off line analysis were performed using an Axoclamp 2B amplifier and pClamp9 software (Axon Instruments, Foster City, CA, USA). For synaptic activity recordings, a bipolar stimulating electrode was placed, approximately 0.3–0.5 mm apart from the recorded neuron. Electrical stimulation (intensity 3–10 V, duration 10–20 ls) was delivered at low frequency (1 pulse ⁄ 10 s).

Whole-cell patch-clamp recordings were made with borosilicate glass pipettes (3–5 MW) containing (in mM) 125 K+-gluconate, 10 NaCl, 1 CaCl2, 2 MgCl2, 0.5 1,2-bis (2-aminophenoxy) ethane- N,N,N,N-tetraacetic acid (BAPTA), 19 HEPES, 0.3 guanosine tri- phosphate (GTP), 1 Mg-adenosine triphosphate (Mg-ATP), adjusted to pH 7.3 with KOH. Membrane currents were monitored using a MultiClamp 700B amplifier (Axon Instruments, Foster City, CA, USA). Interneurons were voltage-clamped at –80 mV. Neurons in which series resistance (8–25 MW) changed by more than 10% during drug application were discarded from the statistics. Excitatory postsynaptic currents (EPSCs) were evoked by 0.1 Hz electrical stimulation delivered with a bipolar electrode, in the presence of 10 lM bicuculline to block GABAA receptors. Traces were stored and analysed off line with pClamp9 software (Axon Instruments). For paired-pulse experiments, two stimuli were delivered with an inter- stimulus interval of 50 ms.

Values given in the text and in the figures are mean ± SEM of changes in the respective groups. Student’s t-tests and nonparametric Wilcoxon tests for paired observations were used to compare the means.

Drug source and handling

(S)-a-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid (AMPA), N-methyl-D-aspartate (NMDA), (S)-3,5-dihydroxyphenylglycine (3,5-DHPG), 6-methyl-2-(phenylazo)-3-pyridinol (SIB1757), (S)-(+)- a-amino-4-carboxy-2-methylbenzeneacetic acid (LY367385), 2- methyl-6-(phenylethylnyl)pyridine hydrochloride (MPEP), 7-(hydrox- yimino)cyclopropa[b]chromen-1a-carboxylate ethyl ester (CPCCOEt), 6-cyano-7-nitroquinoxaline-2,3-dione disodium (CNQX), dizocil- pine ⁄ (5S,10R)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohep- ten-5,10-imine maleate (MK801), a-amino-5-carboxy-3-methyl-2- thiopheneacetic acid (3-MATIDA) were from Tocris Cookson (Avon- mouth, UK); tetrodotoxin (TTX) from Alomone Laboratories (Jeru- salem, Israel); (–)-bicuculline methiodide from Sigma–Aldrich (Milan, Italy). Drugs were dissolved to the final concentration in Krebs’ solution and applied by turning on a three-ways tap and switching the perfusion from control to drug-containing solution. Complete replacement of the medium in the chamber took 40–60 s.

Immunocytochemistry experiments

Animals and tissue preparation

Four male Sprague–Dawley rats were used for this study. All procedures were approved by the animal care and use committee of Emory University and conform to the US National Institutes of Health guidelines. All animals were deeply anaesthetized with a cocktail of ketamine (60–100 mg ⁄ kg, i.p.) and dormitor (0.1 mg ⁄ kg, i.p.). The animals were then transcardially perfused with cold oxygenated Ringer’s solution followed by a fixative containing 4% paraformal- dehyde and 0.1% glutaraldehyde in phosphate buffer (0.1 M; pH 7.4). Following perfusions, brains were removed from the skull, postfixed in 4% paraformaldehyde for 24 h, cut into 60-lm-thick sections using a vibrating microtome and stored in PBS at 4° C until processed for immunocytochemistry. Prior to immunocytochemical processing, all sections were put into a 1% sodium borohydride solution for 20 min and then washed with PBS.

Primary antibodies

To localize mGluR1a, an affinity-purified rabbit polyclonal antibody against the C-terminus of rat mGluR1a (PNVTYASVILR- DYKQSSSTL) conjugated to KLH with glutaraldehyde was used at a concentration of 1 : 1000 (Chemicon, Temecula, CA; Cat# AB1551,Lot# 21100471). In Western blot analysis by the manufacturer, this antibody labels a single band of ~140 kDa. In addition, recent studies from our laboratory have used a combination of knock-out mice, transfected HEK-293 cells, and preadsorption to determine the specificity of the mGluR1a antibody. These studies demonstrated that mGluR1a knockout mice do not show any labelling for mGluR1a vs. wild-type, while only cells transfected for mGluR1a label a band of 140 kDa (Kuwajima et al., 2004) and preadsorption studies in rat retina cells abolish mGluR1a labelling (Koulen et al., 1997). An affinity-purified synthetic rabbit polyclonal antibody against the C-terminus of mGluR5 with a lysine added to the N-terminus (KSSPKYDTLIIRDYTNSSSSL) in a concentration of 1 : 5000 (Upstate Biotechnology, Lake Placid, NY; Cat# 06–451, Lot# 27884) was used to localize mGluR5. According to the manufacturer’s immunoblot analysis, the mGluR5 antibody labels a band of ~30 kDa. Specificity studies from our laboratory have shown that mGluR5 knockout mice do not stain for mGluR5 using this antiserum; immunoblotting of HEK-293 cells transfected with mGluR5 or various brain regions label a single band corresponding to the mGluR5 molecular weight (Marino et al., 2001; Kuwajima et al., 2004). A monoclonal antibody against the calcium-binding protein parvalbumin (PV) was used in a concentration of 1 : 5000 [Swant, Bellinzona, Switzerland, Cat# 235, Lot# 10–11(F)]. This antibody was produced by hybridization of mouse myeloma cells with spleen cells from mice immunized with PV purified from carp muscles and creates a 12-kDa band in immunoblots (Manufacturer’s information). The specificity of the PV antibody was demonstrated through the use of immunoblots and radioimmunoassay (RIA; Celio, 1986; Celio et al., 1988). RIA showed inhibition of binding of immunoreactive PV by unlabelled rat- muscle PV with no cross-reaction to other calcium-binding proteins (Celio et al., 1988). Nitrocellulose paper treated with numerous amino acids and calcium binding proteins showed that PV only reacted with PV and immunoreactivity was displaced by a PV conjugate (Celio, 1986).

Double preembedding immunoperoxidase and immunogold labelling for electron microscopy colocalization of mGluR1a or mGluR5 with PV
Following sodium borohydride treatment, sections were placed in a cryoprotectant solution for 20 min (PB 0.05 M, pH 7.4, 25% sucrose, and 10% glycerol), frozen at –80 °C for 20 min, returned to a decreasing gradient of cryoprotectant solutions, and rinsed in PBS. Sections were incubated for 30 min in PBS containing 5% dry milk at room temperature and then rinsed in TBS-gelatin buffer. Sections were then transferred to primary antibody solutions that contained a mixture of the primary antibody for either receptor subtype and PV with 1% dry milk in TBS-gelatin buffer for 24 h at room temperature and then rinsed again in TBS-gelatin. After rinses, sections were treated for 2 h at room temperature with a mixture of secondary goat anti-rabbit IgGs for mGluR1a or mGluR5 conjugated with 1.4 nm gold particles at a concentration of 1 : 100 (Nanoprobes, Yaphank, NY) and horse anti- mouse biotinylated IgGs for PV at a concentration of 1 : 200 (Vector Laboratories, Burlingame, CA) with 1% dry milk in TBS-gelatin. Sections were rinsed in TBS-gelatin and 2% sodium acetate buffer before gold particles were silver intensified to 30–50 nm (upon electron microscopic examination) using the HQ silver kit (Nano- probes) for approximately 10 min. Following silver intensification, the sections were rinsed again in PBS and then incubated for another 90 min with the avidin-biotin peroxidase complex (ABC) at a dilution of 1 : 100 (Vector Laboratories). Finally, the sections were washed in PBS and Tris buffer (50 mM; pH 7.6) and transferred to a solution containing 0.025% 3,3¢-diaminobenzidine tetrahydrochloride (DAB; Sigma, St Louis, MO), 10 mM imidazole, and 0.005% hydrogen peroxide in Tris buffer for 10 min. Immediately following the DAB reaction, the tissue was rinsed in PB (0.1 M, pH 7.4) and treated with 1% OsO4 for 20 min. It was then returned to PB and dehydrated with increasing concentrations of ethanol. When exposed to 70% ETOH, 1% uranyl acetate was added to the solution for 35 min to increase the contrast of the tissue at the electron microscope. Following dehydra- tion, sections were treated with propylene oxide and embedded in epoxy resin for 12 h (Durcupan ACM, Fluka, Buchs, Switzerland), mounted onto slides and placed in a 60 °C oven for 48 h. Samples of the dorsal striatum were cut out from larger sections, mounted onto resin blocks and cut into 60-nm sections using an ultramicrotome (Leica Ultracut T2). The 60-nm sections were collected on Pioloform-

coated copper grids, stained with lead citrate for 5 min to enhance tissue contrast and examined on the Zeiss EM-10C electron micro- scope. Electron micrographs were taken with a CCD camera (DualView 300 W; Gatan, Inc., Pleasanton, CA) controlled by DigitalMicrograph software (Gatan, Inc.).

Control immunohistochemistry experiments

To control for nonspecific immunoreactivity, each of the primary antibodies was removed in turn from the primary antibody solutions while the remainder of the procedure was the same. These sections contained only the gold deposits when the PV antibodies were omitted, while sections were solely labelled with DAB following omission of mGluRs antibodies. The overall pattern of labelling for either mGluRs or PV in these sections was the same as in sections immunostained with both antibodies, suggesting that the double-labelling procedure had no negative effect on the quality of labelling. A further control experiment was to reverse the order of antigen localization (i.e. revealing PV with gold and mGluRs with peroxidase). The relative proportion of double-labelled PV dendrites in this material did not significantly differ from the original experiments (data not shown).

Analysis of material for double immunoperoxidase and immunogold labelling of Group I mGluRs and PV

Data for double immunoperoxidase and immunogold labelling were collected from seven blocks of mGluR1a ⁄ PV- or mGluR5 ⁄ PV- immunostained rat tissue as described above. Serial ultrathin sections were collected and 40–45 electron micrographs were taken at 25 000· from each block, for a total tissue surface area of 1956 lm2 for mGluR1a ⁄ PV and 1514 lm2 for mGluR5 ⁄ PV. The tissue was scanned for PV-containing elements (i.e. immunoperoxidase labelled elements) in fields where both PV-containing and mGluR1a- or mGluR5- immunoreactive elements (i.e. immunogold particle labelling) could be seen. Therefore, in the electron microscope, analysis was restricted to the most superficial sections of the block where the penetration of both antibodies and reaction products was optimal. Electron micro- graphs of dendrites containing immunoperoxidase labelling were taken (i.e. labelled for PV) and the percentages of these that also contained immunogold labelling (for mGluR1a or mGluR5) was calculated and averaged (± the SEM) across three or four animals. To avoid false positives double labelling due to light gold background labelling, a PV-containing structure had to contain at least two gold particles to be considered double labelled. The double-labelled dendrites were then subdivided into a large and a small subgroup based on their cross-sectional diameter (> 0.75 lm, large dendrite; £ 0.75 lm, small dendrite). The subcellular distribution of immuno- gold labelling in these double-labelled elements was categorized as intracellular or plasma membrane-bound depending on their localiza- tion relative to the plasma membrane. Plasma membrane-bound gold particles were in direct contact with the membrane while all other particles were considered intracellular. The per cent of plasma membrane-bound and intracellular gold particles was calculated from the total number of gold particles counted and then averaged across three or four animals. Data were analysed for significant differences between large and small dendrites’ subsynaptic localization of receptors using one-way repeated measures anovas and Tukey’s posthoc tests with the Sigma Stat software program. The plasma membrane-bound gold particles were further classified into three categories; perisynaptic (touching or within a 20-nm range of the edges of postsynaptic specializations); synaptic (in contact with the main body of postsynaptic specializations); or extrasynaptic (on the plasma membrane but not associated with synapses).

The percentage for each of these categories was determined from the total number of plasma membrane-bound gold particles and averaged across three or four animals.

Results

Electrophysiological identification of striatal FS interneurons

Data included in the study were obtained from intracellular recordings of 98 FS interneurons, identified by their specific electrophysiological features (Fig. 1) (Kawaguchi, 1993; Kawaguchi et al., 1995; Koos & Tepper, 1999; Bracci et al., 2002; Bracci et al., 2003). Recorded cells had a mean resting membrane potential (RMP) value of –76.8 ± 3.3 mV. Input resistance was 86.6 ± 32.4 MW. FS neurons were silent at rest; upon depolarization, they showed short duration action potentials (spike width at half amplitude < 1 ms) followed by short duration afterhyperpolarizations (AHP; amplitude –10.7 ± 1.9 mV; time to peak 4.9 ± 2.1 ms). In response to depolarizing current injection, FS interneurons showed either a brief burst of action potentials or an early spike followed by tonic firing with no accommodation (Fig. 1A and B). A brief, small amplitude AHP followed the end of the depolarizing pulse (Fig. 1B). At rest, FS interneurons showed spontaneous postsynaptic potentials of variable frequency and amplitude (Fig. 1C). At membrane potentials more depolarized than –40 mV, intermittent bursts of action potentials and subthreshold membrane potential fluctuations during quiescent periods were recorded (Fig. 1C). FIG. 1. Electrophysiological features of striatal fast-spiking interneurons. (A) Current–voltage relationship was measured by injection of steps of negative and positive current (from –0.8 nA to 0.8 nA, 600-ms duration) through the recording electrode. (B) In the same neuron, injection of a prolonged, high intensity pulse of depolarizing current (1 nA, 1 s) induced the typical high frequency firing showing little accommodation. The maximal firing frequency recorded in this cell was 160 Hz. Termination of firing was followed by a short-duration AHP (arrowhead). RMP, –80 mV. (C) A representative recording from another fast-spiking interneuron shows the absence of firing activity at resting membrane potential (RMP, –77 mV). Spontaneous synaptic activity is indicated by the empty arrow. When the cell is depolarized to its firing threshold the typical tonic firing interrupted by abrupt pauses is observed. The black arrow indicates the characteristic subthreshold membrane potential fluctuations. Effect of group I mGluR blockade on synaptic activity Synaptic potentials were elicited at low frequency (0.1 Hz) to address the effect of group I mGluRs blockade on FS interneuron basal synaptic activity. The glutamatergic component of the excitatory postsynaptic currents (EPSCs) was pharmacologically isolated by adding the GABAA receptor selective inhibitor bicuculline to the perfusion solution. Pretreatment with bicuculline (10 lM, 10 min) caused a reduction in the amplitude of the postsynaptic potential (Fig. 4A 79.8 ± 4.4% of control; n ¼ 8, P < 0.05). To identify the ionotropic glutamate receptors mediating the glutamatergic component of the EPSP, the selective NMDA and AMPA receptor antagonists, MK801 and CNQX,respectively, were used. Addition of MK801 (30 lM, 10 min) to the perfusion solution did not significantly modify the EPSP amplitude (Fig 4A; 92.5 ± 3.4% of control; n ¼ 5, P > 0.05), whereas bath application of CNQX (10 lM, 10 min) largely reduced the EPSP amplitude (Fig 4A; 17.2 ± 1.0% of control; n ¼ 5, P < 0.05). These observations indicate that the major component of the EPSP in FS interneurons is mediated by the activation of AMPA ⁄ kainate receptors, in agreement with previous data obtained on spontaneous excitatory glutamatergic potentials (Plotkin et al., 2005). Then, we addressed the possible contribution of group I mGluRs to pharmacologically isolated glutamate-mediated EPSP. In the presence of bicuculline (10 lM, 10 min), bath application of the mGluR1 antagonist LY367385 (30–50 lM, 15 min) caused a signi- ficant reduction in the amplitude of the EPSP (Fig. 4, B1 and C; 58.4 ± 4.0% of control; n ¼ 7, P < 0.05). The noncompetitive antagonist, CPCCOEt (100 lM, 15 min), induced a similar effect (Fig 4C; 56.5 ± 3.2% of control; n ¼ 3; P < 0.05). In addition, a novel mGluR1 antagonist with different pharmacological properties, 3-MATIDA (Costantino et al., 2004), was utilized. Bath application of 100 lM 3-MATIDA (10–15 min) induced a similar reduction in the amplitude of the EPSP (Fig. 4, B2 and C; 58.5 ± 6.6% of control; n ¼ 3, P < 0.05). Conversely, pretreatment with SIB1757, a selective mGluR5 antagonist (2–5 lM, 15 min) caused only a slight, nonsignificant reduction of the EPSP amplitude (Fig. 4, B3 and C; 86.9 ± 0.3% of control; n ¼ 5, P > 0.05). Coapplication of both group I mGluR antagonists, LY367385 and SIB1757, depressed to
57.5 ± 3.2% the EPSP amplitude (Fig. 4, B3 and C; n ¼ 3,P < 0.05). It could be argued that the effect observed with LY367385 on the glutamatergic EPSP might be ascribed to a modulatory action by mGluR1 on either the NMDA- or AMPA- mediated ionotropic components. To address this issue, 3,5-DHPG was applied during whole-cell voltage-clamp experiments. FS interneurons recorded in the whole-cell patch-clamp configuration (n ¼ 21) had mean RMP of –81.9 ± 1.7 mV and input resistance of 86.8 ± 8.6 MW. Cells were clamped at –80 mV for voltage-clamp recordings. NMDA and AMPA-evoked inward currents were recor- ded in FS interneurons challenged with 3,5-DHPG. Bath application of 10–50 lM 3,5-DHPG induced an inward current (25 lM, 25.7 ± 2.6 pA amplitude; not shown; n ¼ 5). In the control condi- tion, brief bath applications of NMDA (30 lM, 30 s) elicited a transient inward current (Fig. 5, A1; 90.0 ± 8.7 pA; n ¼ 3). NMDA was then applied after preincubation in 3,5-DHPG, when the group I mGluR agonist-induced current had reached the steady state (about 3 min). In this condition, the NMDA-induced current was not significantly different from control currents (Fig. 5, A1; 87.3 ± 6.9 pA; n ¼ 3, P > 0.05). Perfusion with AMPA (1 lM, 25 s) induced a transient inward current (Fig. 5, A2; 76.3 ± 5.6 pA, n ¼ 4). In the presence of 3,5-DHPG (25 lM), the AMPA-evoked current was unaffected (Fig. 5, A2 and 77.7 ± 7.4 pA; n ¼ 4, P > 0.05), suggesting that neither the NMDA- nor the AMPA-receptor mediated responses are modulated by group I mGluRs in FS interneurons. However, as synaptic and extrasynaptic ionotropic glutamate receptors could be differentially modulated, the effect of 3,5-DHPG was also investigated on the synaptic response of FS neurons. Excitatory postsynaptic currents (EPSCs) were evoked by 0.1 Hz electrical stimulation in the presence of bicuculline (10 lM). Bath application of the group I mGluR agonist 3,5-DHPG (50 lM) did not cause a significant modification in the amplitude of the recorded EPSCs (Fig. 5B; 99.55 ± 5.0% of control; n ¼ 3, P > 0.05). As changes in paired-pulse ratio (PPR) have been considered as an indicator of modified presynaptic release (Schulz et al., 1994), synaptic responses to paired stimulation (50 ms interstimulus interval) were recorded under control conditions or in the presence of the mGluR1 antagonist LY367385. The inhibitory effect of LY367385 on synaptic currents was not associated with a significant change in PPR (control 1.19 ± 0.12, LY367385 1.24 ± 0.10; n ¼ 6, P > 0.05), ruling out a presynaptic site of action of mGluR1 antagonists in the effects observed on FS interneurons (Fig. 5C).

FIG. 3. The 3,5-DHPG-induced membrane depolarization response is mediated by mGluR1 activation. (A) In the presence of TTX, 50 lM 3,5-DHPG (45 s) induced a membrane depolarization in the recorded neuron. (A1) Upon incubation with the mGluR1 antagonist, LY367385 (50 lM, 15 min), a significant reduction in the amplitude of the 3,5-DHPG-induced membrane depolarization was observed. RMP, –74 mV. (B1) Another representative recording showing that the amplitude of the 3,5-DHPG-induced depolarization (B) was unaffected by preincubation with the mGluR5 antagonist, SIB1757 (2 lM, 15 min). (B2) Further addition of LY367385 (30 lM, 15 min) to SIB1757 caused a significant reduction in the amplitude of the 3,5-DHPG-induced membrane depolarization. RMP, –74 mV. (C1) Representative trace showing the inhibitory effect of the mGluR1 antagonist CPCCOEt (100 lM, 15 min) on the 3,5-DHPG-induced membrane depolarization (C). RMP, –80 mV. (D) Summary plot of the pharmacological data.

Finally, we determined the involvement of mGluR1 in the EPSP recorded from FS interneurons directly, in conditions of pharmaco- logic isolation obtained by blocking both glutamate ionotropic components. In the presence of 10 lM bicuculline, plus 30 lM MK801 and 10 lM CNQX, a reliable slow EPSP was evoked by increasing the stimulus intensity (Fig. 6, B1; n ¼ 15). Perfusion with either LY367385 or CPCCOEt (30–50 and 100 lM, respectively).

FIG. 4. Effect of group I mGluR blockade on the glutamatergic EPSP. (A) Pharmacological isolation of the GABAergic and glutamatergic components of the postsynaptic potential in a representative striatal FS interneuron, by incubation with selective GABAA, NMDA and AMPA antagonists (10 lM bicuculline, 30 lM MK801, and 10 lM CNQX, respectively). RMP, –80 mV. (B1) In the presence of bicuculline, blockade of mGluR1 by 50 lM LY367385 caused a significant reduction in the amplitude of the glutamatergic EPSP in another FS interneuron. RMP, –83 mV. (B2) Representative recording showing a similar inhibitory effect obtained with the mGluR1 antagonist, 3-MATIDA (100 lM). RMP, –85 mV. (B3) Representative traces from a different recording showing that SIB1757 (mGluR5 antagonist, 2 lM) slightly reduced the amplitude of the glutamatergic EPSP. Addition of LY367385 strongly inhibited the EPSP. RMP, –84 mV.(C) Plot summarizing the pharmacological data.

MGluR1a and mGluR5 subcellular localization

To characterize further the potential substrate that underlies the physiological roles of group I mGluRs in FS neurons, we performed a double immunocytochemical electron microscopic study to determine the degree of colocalization and subcellular expression of mGluR1a and mGluR5 in parvalbumin-containing striatal interneurons in the rat striatum. In striatal tissue double labelled for mGluR1a or mGluR5 and PV, 80.04 ± 3.37% (mean ± SEM, n ¼ 3 animals) of PV-labelled dendrites also expressed mGluR1a (n ¼ 191 dendrites), while 79.56 ± 1.83% (mean ± SEM, n ¼ 4 animals) coexpressed mGluR5 (n ¼ 178 dendrites). Figure 7 shows representative examples of double-labelled dendrites. One-way RM anovas and Tukey’s posthoc tests showed that the relative proportion of plasma membrane-bound gold particles for both group I mGluRs was significantly higher in small dendrites than in large dendritic shafts (62.1 ± 2.9% vs. 41.9 ± 4.4% for mGluR1a, n ¼ 152 dendrites, 888 gold particles, P < 0.01; 69.2 ± 1.7% vs. 40.4 ± 3.0% for mGluR5, n ¼ 142 dendrites, 851 gold particles, P < 0.01; Fig. 8A and C). Overall, the pattern of subsynaptic labelling for mGluR1a and mGluR5 was quite similar in PV-containing dendrites (Fig. 8B and D). For both receptor subtypes, more than 95% of labelling was extrasynaptic, while less than 5% of labelling was either perisynaptic or synaptic to asymmetric or symmetric synapses. There was no difference in the overall pattern of labelling on the plasma membrane between large and small dendrites (Fig. 8B and D). None of the PV-containing terminals examined displayed group I mGluRs immunoreactivity (Fig. 7D). FIG. 5. Whole-cell patch-clamp analysis of the effect of mGluR1 on NMDA and AMPA-mediated currents and synaptic responses. (A1 and A2) Representative traces of ionotropic glutamate receptor-mediated inward currents recorded in the whole-cell patch-clamp configuration from FS interneurons clamped at –80 mV. (A1) Preincubation in 25 lM 3,5-DHPG did not significantly affect the amplitude of the inward current induced by a brief application (30 s) of 30 lM NMDA (black trace, control; grey trace, in 3,5-DHPG). (A2) Similarly, 3,5-DHPG did not affect the AMPA-mediated inward current (1 lM AMPA, 25 s; black trace, control; grey trace, in 3,5-DHPG). (B) Traces from a representative recording, showing that preincubation in 50 lM 3,5-DHPG did not modify the amplitude of the EPSC. (C1) Whole-cell patch-clamp recordings show synaptic responses to paired stimulation (50-ms interstimulus interval) in FS interneurons, in the control condition and in the presence of LY367385 (50 lM) in the perfusing solution. Note the decrease in EPSC amplitude without any significant change in paired-pulse ratio (PPR ¼ EPSC2 ⁄ EPSC1) induced by LY367385 (grey trace). (C2) The graph shows that PPR was not significantly modified in the presence of LY367385, as compared to controls. Discussion Growing evidence supports the central role of striatal FS interneurons in regulating striatal activity. Although these cells represent less than 2% of the entire striatal neuronal population, it has become evident that they exert a powerful GABAergic inhibitory control over spiny projection neurons (Koos & Tepper, 1999). The data presented in this study show that both mGluR1 and mGluR5 are coexpressed and show a similar subcellular localization in a large population of striatal FS interneurons in the rat striatum. In spite of their coexpression, only mGluR1, but not mGluR5, modulates the excitability of FS interneurons. Two lines of evidence support this conclusion. First, group I mGluRs activation by 3,5-DHPG induced a direct membrane depolarization of FS neurons that was significantly reduced by mGluR1 antagonists, but not affected by blockade of mGluR5. Second, mGluR1 antagonists significantly depressed the glutamatergic EPSP, while mGluR5 blockade had a negligible impact on EPSP amplitude. FIG. 6. Effect of group I mGluR blockade on the slow EPSP evoked in the presence of GABAA, NMDA and AMPA receptor antagonists. (A) Represen- tative traces showing the effect of coapplication of the GABAA antagonist bicuculline (10 lM) plus the ionotropic glutamate receptor antagonists MK801 (30 lM), and CNQX (10 lM) on the EPSP recorded from a FS interneuron. RMP, –76 mV. (B1). In the same recording, after blockade of GABAA, NMDA and AMPA receptors, by progressively increasing the stimulus intensity, a slow EPSP was evoked. (B2) Bath application of the mGluR1 antagonist LY367385 (50 lM) caused a significant reduction in the amplitude of the slow EPSP, while further application of 2 lM SIB1757 (mGluR5 antagonist) did not cause any additional modification. MGluR1 activation mediates membrane depolarization in FS interneurons Although our electron microscopy data revealed the coexpression on both mGluR1 and mGluR5 on striatal FS interneurons, the 3,5-DHPG- induced membrane depolarization was blocked only by mGluR1 antagonists, whereas mGluR5 blockers were ineffective. The apparent mismatch between receptor localization and electrophysiological effects of group I mGluRs in basal ganglia neurons is not surprising and has been described in other nuclei. MGluR1 and mGluR5 are, indeed, coexpressed throughout the basal ganglia, where they exert distinct actions on the same neuron subtype. For instance, in subthalamic neurons, both the membrane depolarization and the potentiation of NMDA inward current by 3,5-DHPG are mediated by mGluR5, despite strong plasma membrane expression of both mGluR1 and mGluR5 (Awad et al., 2000). Conversely, activation of mGluR1, but not mGluR5, exerts direct depolarizing effects on GABAergic substantia nigra pars reticulata (SNr) neurons (Marino et al., 2001, 2002). In type II neurons of the globus pallidus the mGluR1-mediated depolarization is enhanced by blockade of mGluR5 (Poisik et al., 2003). Distinct functional roles for group I mGluR subtypes have also been demonstrated in striatal spiny projection neurons. Selective activation of mGluR5 is responsible for the enhancement of the NMDA-induced membrane depolarization, while mGluR1 is required for induction of striatal long-term depression (Gubellini et al., 2001; Pisani et al., 2001a). It is noteworthy that, within the same brain region, striatal cholinergic interneurons express either mGluR1 or mGluR5, which are both involved in the 3,5-DHPG-induced membrane depolarization (Pisani et al., 2001b; Bonsi et al., 2005). Synaptic activity With respect to synaptic activity, our results demonstrate the involvement of mGluR1, but not mGluR5 in synaptic activity recorded from FS interneurons. Either LY367385, CPCCOEt or 3-MATIDA were indeed able to decrease the EPSP amplitude, whereas selective mGluR5 antagonists failed to modify synaptic responses. Interestingly, each mGluR1 antagonist inhibited the EPSP to a similar extent, nearly 40%. Furthermore, this inhibitory effect was observed either when a test synaptic stimulation was delivered, but also at higher intensity stimulation in experimental conditions of pharmaco- logical isolation, in which the glutamate ionotropic component was fully blocked. The apparent discrepancy of these data might argue for a modulatory action of mGluR1 on the ionotropic, NMDA- or AMPA- mediated components in the former experimental condition. However, the observation that 3,5-DHPG did not modify either NMDA- and AMPA-evoked currents or the glutamatergic EPSC ruled out this possibility, suggesting that in both conditions synaptic stimulation might recruit extrasynaptic mGluR1 receptors. Another possibility was that mGluR1 antagonists action was mediated through a presynaptic mechanism. As changes in PPR are considered as an indicator of modified presynaptic release (Schulz et al., 1994), the effect of LY367385 on PPR was analysed. The absence of significant changes in PPR suggests a postsynaptic action of mGluR1 antagonists. Our electrophysiological data are not supportive of an active involvement of mGluR5 in synaptic activity recorded from FS interneurons. Consistent with our results, in the SNr both mGluR1 and mGluR5 have been found at postsynaptic sites, though only mGluR1 actively participates to synaptic responses (Marino et al., 2001). It might be speculated that mGluR5 exerts a modulatory role in intracellular signalling pathways independent of ion channel regula- tion. For instance, as mGluR5 has been demonstrated to regulate intracellular calcium mobilization, its activation might determine the amount of calcium available for synaptic regulation. Future studies on the role of mGluR5 in these cells may provide important insight into the distinct functional roles of closely related receptor subtypes within a single neuronal population. Although a differential subtype-specific trafficking and subcellular localization of mGluR1 and mGluR5 has been suggested to underlie their distinct electrophysiological effects in the SNr (Hubert et al., 2001), such is unlikely to be the case in FS interneurons, because our electron microscopic data show a similar subcellular and subsynaptic distribution of mGluR1a and mGluR5 in these neurons. Receptor phosphorylation and protein–protein interaction processes are regula- tory mechanisms that affect the activity of mGluR1 or mGluR5 in different neuronal populations and might in part explain the divergent roles of these closely related receptors in FS neurons (Valenti et al., 2002). In particular, scaffolding proteins like Homer, that specifically associate with group I mGluRs, might have a prominent role in differentially regulating mGluR1 and mGluR5 activity (Tu et al., 1998; Kammermeier et al., 2000; Xiao et al., 2000; Thomas, 2002; Fagni et al., 2004). FIG. 7. Double immunoperoxidase and immunogold labelling for mGluR1a ⁄ PV (A–D) or mGluR5 ⁄ PV (E–F) in the rat striatum. Single arrows indicate extrasynaptic gold particles, arrowhead indicates intracellular labelling, double arrowheads point at perisynaptic labelling at asymmetric synapses, while asterisks indicate perisynaptic labelling at symmetric synapses. den, dendrite; sp, spine; te, terminal. Scale bars, 0.25 lm. FIG. 8. Histogram showing the subcellular and subsynaptic localization of mGluR1a and mGluR5 in large and small PV-immunoreactive dendrites. A and C show the percentages (± SEM) of plasma membrane-bound and intracellular gold particles for mGluR1a and mGluR5. Asterisks indicate significantly higher proportions of plasma membrane-bound receptors in small dendrites compared to large dendrites. B and D show the localization of mGluR1a and mGluR5 on the plasma membrane of large and small dendrites. In conclusion, even if the exact mechanisms underlying this functional segregation of the two group I mGluRs in specific populations of striatal neurons remain to be established, these findings indicate that the electrophysiological properties of mGluR1 and mGluR5 are highly specific and dependent on the basal ganglia region and neuronal phenotypes they are expressed in. Potential role of mGluR1 in the regulation of FS excitability Colocalization between mGluR1 and neuronal glutamate transporters has been demonstrated in different brain regions (Brasnjo & Otis, 2001). Such overlapping postsynaptic distribution seems to suggest that either mGluR1 or glutamate transporter can compete for synaptically released glutamate. Thus, one possibility would be that perisynaptic mGluR1 might serve as sensors of glutamate levels and in turn activate glutamate transporters to buffer glutamate itself. Our observation that mGluR1 antagonists reduce synaptic responses independently of the ionotropic- or metabotropic component, is consistent with this hypothesis, in that mGluR1 blockade would redirect synaptic glutamate towards the transporter, which would in turn clear glutamate from the synaptic cleft. An alternative interpretation is that mGluR1 activation might be actively involved in the generation of complex activity patterns, thereby exerting a profound influence on striatal output neurons. Consistent with previous ultrastructural data, there is compelling evidence of electrical coupling in a high percentage of electrophysiologically tested pairs of striatal FS interneurons (Koos & Tepper, 1999). The biophysical properties of the electrotonic interactions between interneurons are critical determinants of their population activity, resulting in complex oscillatory and rhythmic activity patterns (Berke et al., 2004). In distinct brain regions, such as the thalamic reticular nuclei and cerebral cortex, it has been shown that group I mGluRs can trigger such synchronized activity (Long et al., 2004). Moreover, both in cortical and hippocampal slices, mGluR activation resulted in the appearance of an oscillatory activity in networks of inhibitory interneurons connected (R,S)-3,5-DHPG by synapses using GABAA (Whittington et al., 1995).