| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Netherlands Institute for Brain Research, 1105 AZ Amsterdam; 2Vrije Universiteit University Medical Center, Department of Anatomy, 1007 MC Amsterdam; and 3University of Amsterdam, Department of Animal Physiology and Cognitive Neuroscience, 1098 SM Amsterdam, The Netherlands
Submitted 11 September 2003; accepted in final form 20 October 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
; Pennartz et al. 1994). It has been implicated in several behavioral and cognitive functions, such as translation of behaviorally relevant stimuli into behavioral responses (Cardinal et al. 2002; Robbins and Everitt 1996), coding of reward and reward expectancy (Schultz et al. 2000), control over impulsive behavior (Cardinal et al. 2001), instrumental learning, and approach behavior (Baldwin et al. 2002; Balleine and Killcross 1994; Corbit et al. 2001; DiCiano et al. 2001; Hernandez et al. 2002; Parkinson et al. 2002). In general, the Nac can be considered a neuronal interface between sensory and associative brain areas processing and evaluating salient stimuli within their context and the motor and autonomous systems mediating responses to these stimuli (Mogenson et al. 1980; Pennartz et al. 1994).
Anatomical, electrophysiological and behavioral studies have disclosed many aspects of the functional organization of the Nac (Arts and Groenewegen 1992; Cardinal et al. 2002; Chang and Kitai 1986; Groenewegen et al. 1996, 1999; Kelley 1999; Meredith 1999; Pennartz and Kitai 1991; Pennartz et al. 1990, 1991, 1992, 1994; Sarter et al. 1999; Zahm 1999). The Nac forms an integral part of the striatum, and as in the caudate-putamen region (which forms the dorsal part of the striatum; Bolam et al. 2000; Kawaguchi et al. 1995), about 90-95% of the neuronal population of the Nac consists of medium-sized spiny neurons (MSNs), a class of GABAergic cells. In the Nac, MSNs receive extensive glutamatergic input from different limbic areas such as prefrontal cortex, thalamus, hippocampus, and basolateral amygdala, and project their output to downstream areas of the meso-cortico-limbic system like the ventral pallidum, the lateral hypothalamus, and mesencephalic dopaminergic areas (Chang and Kitai 1985; Groenewegen and Russchen 1984; Nauta et al. 1978).
In addition to the large majority of MSNs, interneurons compose the remaining 5-10% of ventral and dorsal striatal neurons, such as fast spiking interneurons (FS), low-threshold spike interneurons, and cholinergic interneurons (Bracci et al. 2003; Cowan et al. 1990; Hidaka and Totterdell 2001; Kawaguchi et al. 1995; Wilson et al. 1990). Despite our knowledge about these different cell types and their characteristics (Meredith et al. 1990; Pickel et al. 1988; Sesack and Pickel 1990), detailed information about intra-accumbens cellular interconnectivity is still lacking. It has long been known that axons of MSNs project not only outside the Nac but also branch profusely within it (Chang and Kitai 1986; Pennartz et al. 1991). However, it is not known whether MSNs in the Nac are in fact functionally connected to each other and to what extent.
Anatomical evidence supporting the presence of a relatively extended axonal collateralization of MSNs has been provided also in the dorsal striatum (Kawaguchi et al. 1990; Preston et al. 1980; Ramon y Cajal 1911; Somogyi et al. 1981; Wilson and Groves 1980; Yung et al. 1996). Together with these anatomical data, early electrophysiological studies suggested the existence of GABAergic, bicuculline-sensitive inhibition in the dorsal striatum in vivo (Bernardi et al. 1975; Katayama et al. 1981; Park et al. 1980) and in vitro (Lighthall and Kitai 1983; Misgeld et al. 1982). However, direct evidence of GABAergic inhibition between MSNs in striatal slices has been provided only recently. Contrary to an earlier study suggesting the absence of lateral inhibition (Jaeger et al. 1994), Tunstall et al. (2002) and Czubayko and Plenz (2002), using dual intracellular and whole cell recordings in rat slices and organotypic cultures, recently demonstrated the existence of bicuculline-sensitive GABAergic connectivity between pairs of MSNs in the dorsal striatum. In addition, FS interneurons also provide GABAergic inhibition onto MSNs in the dorsal striatum (Koos and Tepper 1999).
In the Nac, previous studies have demonstrated GABAergic transmission in MSNs by means of local or afferent electrical stimulation (Chang and Kitai 1986; Pennartz and Kitai 1991; Pennartz et al. 1991), but no demonstration of GABAergic lateral inhibition between pairs of ventral striatal MSNs has been provided so far. We do not consider it warranted to extrapolate the very recent evidence for lateral inhibition from the dorsal striatum to the ventral striatum, because several major physiological differences between these areas have been revealed, such as the presence versus lack of a dopamine receptor-mediated attenuation of fast excitatory postsynaptic potentials (EPSPs) and inhibitory postsynaptic potentials (IPSPs) in the ventral and dorsal striatum, respectively (Nicola and Malenka 1998). Also, dopamine has been reported to gate corticostriatal long-term potentiation (LTP) and long-term depression (LTD) of glutamatergic EPSPs in the dorsal (Calabresi et al. 2000) but not ventral striatum (Pennartz et al. 1993). Thus the existence, extent, and functional properties of lateral inhibition in the ventral striatum are no trivial matters that can be simply extrapolated from findings in the dorsal striatum.
Several model studies have indicated that lateral GABAergic inhibition in the striatum might be an important feature in selecting proper output patterns, as part of a coherent response to environmental stimuli encoded and evaluated by the prefrontal cortex and other glutamatergic areas afferent to the striatum. Since different MSNs may receive overlapping inputs from multiple common sources, this selection function has been proposed to be implemented by a "Winner-take-all" (WTA) strategy, which would ensure the detection of salience differences by rapidly solving conflict between competing stimuli, preventing distortion of proper outputs, and avoiding rapid switching between closely matched competitive states. These characteristics may endow the striatum with the attributes of a competitive network for pattern classification, subserving selection of behavioral strategies (Redgrave et al. 1999) and cognitive functions (Beiser and Houk 1998). Moreover, GABAergic inhibitory postsynaptic potentials may significantly delay the occurrence of postsynaptic action potentials, possibly causing disruption of the temporal relationship (coincidence detection) between excitatory synaptic inputs and backpropagating spikes that are thought to lie at the basis of LTD or LTP (Markram et al. 1997; Stuart and Hausser 2001; see also Plenz 2003).
Given these comparative and theoretical considerations, the general aim of this study was to assess whether lateral inhibition does exist in the Nac, and if so, to assess its electrophysiological properties. We show that MSNs in acutely prepared slices of rat Nac have a relatively high probability of being synaptically interconnected. This chemical neurotransmission is mediated by GABAA receptors and shows forms of frequency facilitation and depression. No evidence for electrical coupling between MSNs was found. These findings advance our fundamental understanding of intra-accumbens communication and provide empirical constraints on biologically plausible network models of the basal ganglia.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Wistar rats (23-30 days of age) were anesthetized with an intra-peritoneal injection of sodium pentobarbital sodium (60 mg/kg) and decapitated. Adequate measures were taken to minimize pain or discomfort of the animals. The experiments were carried out in accordance with the European Community Council Directive 86/609/EEC and approved by the National Committee on animal experiments.
Slices containing the Nac were prepared as described previously (Taverna and Pennartz 2003). Briefly, brains were removed from the skull, and 280-µm-thick coronal slices were cut in artificial cerebrospinal fluid (ACSF) at 4°C using a vibratome (VT1000S, Leica). The slices were allowed to recover for at least 1 h in ACSF at room temperature. Individual slices were submerged in a recording chamber in which ACSF was continuously flowing (1 ml/min) at 32°C. The composition of ACSF was as follows (in mM): 124 NaCl, 3.5 KCl, 1 NaH2PO4, 2.5 CaCl2, 26 NaHCO3, 1.3 MgSO4, and 10 D-glucose, saturated with 95% O2-5% O2 (pH 7.3). NBQX (5 µM) was added to ACSF at the beginning of experiments.
Patch-clamp glass pipettes (5-7 M
) contained the following solution (in mM): 60 K-gluconate, 58 KCl, 10 HEPES, 0.5 EGTA, 1 MgCl2, 2 Na2-ATP, 0.3 Na3-GTP, 20 Na2-phosphocreatine, 0.1 leupeptin, and 26 biocytin (pH 7.2, adjusted with KOH). We chose this composition to obtain a relatively depolarized reversal potential for Cl- ions (-20.8 mV at 32°C according to the Nernst equation) compared with the resting membrane potential and thus to enhance the amplitude of Cl--mediated ionic currents evoked at resting level. A liquid junction potential of +3 mV between intra- and extracellular solution was corrected for. At the end of recording sessions, slices were fixed in formalin and stored at +4°C.
Electrophysiology
Voltage- and current-clamp dual-cell recordings were made using a MultiClamp 700A amplifier (Axon Instruments). Signals were filtered at 1-3 kHz, digitized at 10 kHz, and stored on a computer using pClamp 8.0 software (Axon Instruments). Whole cell configurations were established in voltage-clamp (VC) mode simultaneously in two cells, chosen under visual control using an Axioscope upright microscope (Zeiss) equipped with Hoffman modulation contrast. The series resistance was compensated (40-60%), and electrode capacitance was canceled. We subsequently switched to current-clamp (CC) mode, and after bridge-balance compensation, assessed the intrinsic membrane properties of both cells by applying positive and negative rectangular current pulses (range: +100/-400 pA, 2 s) that were also used to calculate the membrane input resistance (Table 1). The time constant was calculated by fitting the initial 150 ms of the membrane potential response to a small (20-50 pA) depolarizing current injection using a single exponential equation. Action potential threshold was measured at the point of clear-cut up rise of membrane potential (i.e., spike outbreak). Action potential amplitude was measured as the difference between peak and threshold values.
|
|
; Galarreta and Hestrin 2001; Gibson et al. 1999; Koos and Tepper 1999; Tamas et al. 2000). However, we were unable to block the current-induced depolarization in the paired cell with the putative inhibitor of gap junctions octanol (1 mM). Furthermore, similar electrical coupling was found when two model cells were mounted in cell-mode configuration on the two headstage inputs of the amplifier. This latter effect was the ultimate hint for an electrical artifact due to cross-talk between input channels. Such artifact was detected in all recorded pairs, but did not interfere with the detection of synaptic transmission between cells. Morphology
Following fixation, the 280-µm slices were cryoprotected by storage for 1 h at room temperature in a mixture of 20% glycerin and 2% dimethyl sulfoxide (DMSO) in aqua bi-dest or 0.1 M Na2HPO4. 2H2O/KH2PO4 [phosphate buffer (PB)], pH 7.4. Unless stated otherwise, all reagents used for morphological analysis were obtained from Merck (Darmstadt, Germany). The slices were rapidly frozen in 30% sucrose in aqua bi-dest or PB onto the stage of a sliding microtome. The slices were cut into thinner sections of 30 µm to facilitate penetration of reagents during the staining procedures. Sections were collected in a 24-wells plate containing PB for direct processing. The sections were stained for biocytin as follows. After three rinses with PB followed by three rinses with 0.05 M Tris HCl supplemented with 0.15 M NaCl, pH 7.6 (TBS) and 0.5% Triton X-100 (TBS-Tx) for 10 min each, the sections were incubated in avidin-biotin-peroxidase complex (Vector, Burlingame, CA) in TBS-Tx for 4 successive days at 4°C. The incubation was stopped by three rinses with Tris HCl (pH 8.0) for 10 min each. The sections were stained with nickel-enhanced diaminobenzidine (DAB-Ni) substrate: 7.5 mg 3,3'-diaminobenzidine-tetrahydrochloride (DAB; Sigma, St. Louis, MO), 0.225 g nickel-ammonium sulfate (Boom Meppel), and 10 µl of 30% H2O2 in 50 ml Tris HCl, pH 8.0, for 5-20 min at room temperature. Progress of staining was frequently monitored microscopically. As soon as nonspecific background staining became visible, the reaction was terminated by several rinses in Tris HCl (pH 8.0). The sections were mounted on glass slides from a Tris HCl (pH 8.0) solution, containing 0.2% gelatin (Oxoid, Basingstoke, UK), and air dried. Mounted sections were dehydrated through an ascending series of alcohol (50%, 70%, 80%, 2 times at 96%, and 2 times at 100%) and coverslipped from xylene using Entellan.
Data analysis and statistics
Analysis of pre- and postsynaptic signals recorded in CC and VC mode was accomplished using the statistical and fitting analysis tools of Clampfit 8.0 (Axon Instruments) and Origin 5.0 (Microcal Software).
Inhibitory postsynaptic current (IPSC) onset latency was measured as the time interval between the peak of the first derivative of the presynaptic action potential and the onset of the postsynaptic response. Peak amplitudes of postsynaptic events were measured by averaging values within a time window of 0.5-2 ms for IPSCs and 3-5 ms for dPSPs. Baseline references were taken by averaging values within a 2- to 5-ms time window. The decay phase of IPSCs could be fitted well with a single exponential time constant. A failure was defined as the lack of a postsynaptic event in response to a presynaptic action potential. In particular, a postsynaptic current or potential was considered absent when its value, measured within a 20-ms time window starting 0.5 ms after the peak of the presynaptic action potential, failed crossing an empirically determined threshold value that was set as twice the standard deviation (SD) of the baseline mean value, measured within a 20-ms time window preceding the presynaptic action potential.
To quantify frequency potentiation and depression, postsynaptic CC traces were integrated across the first 300 ms departing from the onset of the 2-s lasting presynaptic pulse and across the final 300 ms before the offset of the presynaptic pulse. Averaged areas under the curve within the two time windows were plotted in histograms.
For paired pulse recordings, peak amplitude values of the first IPSC (IPSC1) and second IPSC (IPSC2) were measured and averaged across four different pairs. The ratio between averages for each interspike interval was calculated and plotted (Kim and Alger 2001). This method yields one single ratio value per interspike interval; thus no standard error (SE) was calculated. Failures were included in the analysis.
We also sorted IPSC1s of each cell pair according to their amplitude and normalized them relative to the largest value (maximal normalized value was set to 1). Data were grouped into 0.1-sized bins, and across-pair averages (±SE) of normalized values were calculated within each bin and plotted against paired-pulse ratios (these were calculated by averaging IPSC2 amplitudes within the same bin interval as for the normalized IPSCs1 and computing the ratio between mean IPSC2 and mean IPSC1). Note that, in Fig. 6, some bins remain empty, because no averaged value fell within those bins. The relationship between the normalized amplitude of IPSC1 and paired-pulse ratios was fit with a hyperbolic function (see Debanne et al. 1996).
|
Drugs
Bicuculline methochloride and 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo-[f]quinoxaline-7-sulfonamide disodium salt (NBQX) were obtained from Tocris Cookson. Biocytin was obtained from Sigma.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Seventeen slices with electrophysiologically identified pairs of neurons were available. After immunohistochemical processing, 4 of these 17 slices failed to show any stained neuron, whereas in 9 slices, only one of the neurons of a pair could be anatomically visualized. In six of these nine slices, this neuron could be unequivocally identified as a medium-sized spiny cell. In the remaining four slices, both cells of a pair could be identified and characterized as MSNs. Two of these cell pairs exhibited extensive filling of both axons and dendrites (Fig. 1A). A few close appositions between axons of the stimulated neuron and dendrites or cell body of the responsive neuron were identified.
|
; Pennartz et al. 1991) and their morphological properties as described above. Other cell types encountered during the recording sessions will be presented elsewhere. We recorded a total of 38 pairs of MSNs, of which 13 (34.2%) were connected unidirectionally and 1 (2.6%) was connected bidirectionally (Table 1). We did not find significant differences between the intrinsic properties of connected versus unconnected MSNs (Table 2).
|
Synaptic communication between MSNs: basic properties and GABAergic nature
Figure 2 shows an example of a dual-cell recording from two synaptically connected neurons. AMPA receptor-mediated synaptic transmission was blocked by 5 µM NBQX throughout all experiments to prevent spontaneous, AMPA receptor-mediated synaptic potentials that may interfere with evoked GABAergic synaptic potentials. In VC mode, unitary inward currents (IPSCs) followed presynaptic spikes closely in time, although not every spike elicited a response. The average IPSC peak amplitude was -31 ± 11 pA (range, -8 to -75 pA). Table 3 summarizes the properties of IPSCs evoked by a single presynaptic action potential. A failure was defined as a lack of a postsynaptic IPSC in response to a presynaptic action potential and was not included in the computation of other parameters.
|
|
The GABAergic nature of the postsynaptic response was further confirmed by dual recordings in which a single action potential was evoked in the presynaptic cell in CC mode, while the postsynaptic cell was voltage-clamped at different levels with steps of 10 mV, starting from -73 mV (n = 4 pairs, Fig. 4). A linear fit of the current-voltage relationship reversed at -22 mV, which is close to the expected value of the reversal potential for Cl- ions in our experimental conditions. These findings are consistent with the MSN-to-MSN synaptic response being an inhibitory, GABAA receptor-mediated postsynaptic current (IPSC) (cf. Chang and Kitai 1986; Czubayko and Plenz 2002; Misgeld et al. 1982; Pennartz and Kitai 1991; Tunstall et al. 2002).
|
We wondered whether intrastriatal GABAergic synapses show properties of short-term, activity-dependent plasticity, which could be revealed by differential patterns of presynaptic stimulation. Figure 5 illustrates short-term plastic properties of evoked dPSPs, studied with dual CC recordings. When the presynaptic cell was stimulated with relatively low amounts of current (+80 to +150 pA), its firing rate ranged between 8 and 15 Hz; at these rates, frequency facilitation (FF) of the postsynaptic response occurred (n = 9 of 11 pairs), as the amplitude of evoked dPSPs progressively augmented during the presynaptic spike train despite fluctuations in individual dPSP amplitude (Fig. 5A, left). This augmentation could not be attributed to frequency adaptation of the presynaptic spike trains, since in all connected pairs tested presynaptic cells fired trains of action potentials in a regular spiking or nonadapting fashion. Conversely, when presynaptic firing was evoked at higher rates (
20 Hz) by means of larger current injections, the train of postsynaptic responses was markedly switched into frequency depression (FD; Fig. 5A, right). Intermediate firing rates (16-20 Hz) evoked trains of dPSPs having largely stable amplitudes. In 9 of 11 pairs tested with high presynaptic firing rates, the first dPSP was the largest in amplitude, followed by depression of dPSPs amplitude down to a relatively stable value within the first 200-300 ms (Fig. 5A). The amplitude of the first dPSP in 20-Hz trains was significantly larger than the one in 8- to 15-Hz and 16- to 20-Hz trains (6 ± 1, 4 ± 1, and 4 ± 1 mV, respectively; P < 0.05, Wilcoxon's matched-pairs signed-rank test).
|
FF and FD were quantified by computing the integral values for the first 300 ms and last 300 ms of the dPSP train response. These were 192 ± 66 and 631 ± 292 ms · mV, respectively, for presynaptic trains of 8-15 Hz (Fig. 5B, left: n = 9 pairs; FF was significant at P < 0.02). For trains at 20 Hz, values were 549 ± 221 and 384 ± 168 ms · mV, respectively (Fig. 5B, right: n = 9 pairs; FD was significant at P < 0.02).
When low-rate and high-rate stimulations were alternately repeated, postsynaptic responses of FF and FD, respectively, were also alternately evoked, showing that these phenomena are specific to presynaptic firing rate and readily reversible.
Paired-pulse recordings
Responses to paired pulse stimuli are a useful tool to study the dynamic properties of short-term plasticity. Characteristics of paired-pulse response patterns are thought to reflect presynaptic dynamics in particular (Katz and Miledi 1968; Zucker 1989). We further characterized intra-accumbens GABAergic inhibition by determining whether an IPSC evoked by a presynaptic spike was able to influence the amplitude of a subsequent IPSC induced by a second presynaptic spike, with time intervals ranging between 120 and 380 ms from the first spike in 20-ms steps. The ratio of the mean amplitude of IPSC2 relative to the mean amplitude of IPSC1 was plotted against the interspike interval (see METHODS and Fig. 6A). The time course did not reveal a clear dependence of paired-pulse facilitation or depression on the interspike interval; no significant difference was found when averages of IPSC1 and IPSC2 amplitudes were compared for each time point (P > 0.05, paired t-test).
Large amplitude variability in the evoked responses may have dampened clear paired-pulse effects as a function of interspike interval. Thus other independent variables affecting paired-pulse dynamics were examined. Paired-pulse depression (PPD) and facilitation (PPF) were induced depending on whether the amplitude of IPSC1 was small or large, respectively. This observation was quantified by plotting paired-pulse ratios against mean normalized values of IPSC1, as described in METHODS (Fig. 6B). Paired-pulse ratios relative to mean normalized values of IPSC1 of 0.12 ± 0.04 and 0.9 ± 0.1 were 3.3 and 0.71, respectively. Averaged amplitudes of IPSC1s and IPSC2s from different cell pairs (-6 ± 2 and -21 ± 5 pA, respectively) were significantly different at the lowest mean normalized IPSC1 tested (0.12 ± 0.04; P < 0.02). These data suggest a major dependence of paired-pulse dynamics on the amplitude of the first IPSC.
Lack of a consistent paired-pulse effect, other than due to the amplitude of IPSC1, reflected the apparent randomness of the relationship between pairs of dPSPs evoked by trains of presynaptic action potentials (Fig. 5). To quantify this latter observation, the ratio between the second and the first dPSP in a train was calculated at two different presynaptic firing rates (8-15 Hz and 20 Hz); as described above, these frequency ranges induced FF and FD, respectively. Average paired-pulse ratios were 0.94 ± 0.10 and 0.91 ± 0.10 at 8-15 Hz and 20 Hz, respectively (n = 9, P = 0.8, paired t-test). No significant correlation between this measure of PPF and FF or between PPD and FD was found (r = -0.22, P = 0.27, n = 26 trains); for example, the second dPSP in a facilitating train at 8-15 Hz could be smaller in amplitude than the first one, but facilitation would still occur later during the train (Fig. 5A).
In summary, these results indicate that trains of GABAergic dPSPs in the Nac show marked frequency-dependent facilitation and depression and that paired-pulse effects are primarily governed by the relative amplitude of IPSC1 rather than by interspike intervals.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Depending on the strength of presynaptic current injection setting the presynaptic firing rate, GABAergic synaptic potentials showed different types of short-term plasticity: frequency facilitation (FF) at 8-15 Hz and frequency depression (FD) at 20 Hz. Paired-pulse facilitation or depression was detected when IPSC1 was relatively small or large, respectively. Dependence of paired-pulse dynamics on presynaptic spike interval was not apparent.
Besides being the first demonstration, to our knowledge, of GABAergic lateral inhibition between MSNs in the Nac in situ, the data presented in this study extend similar results recently obtained in dorsal striatal neurons by means of double recordings (Czubayko and Plenz 2002; Tunstall et al. 2002; but see Jaeger et al. 1994). Tunstall et al. (2002) reported for the first time unitary postsynaptic GABAergic IPSPs induced by stimulation of single presynaptic striatal cells, with a 20% probability of finding a connected pair. This relatively low probability may be explained by a difference in postsynaptic signal-to-noise ratio relative to our study. Czubayko and Plenz (2002) showed GABAergic transmission between MSNs in cultured slices of the rat neostriatum, with an approximately 38% probability of finding connections. In the same study, GABAergic connections between pairs of MSNs in acutely prepared slices from rats aged 11 days were also reported, but with substantially lower probability (
13%). In culture, GABAergic transmission showed short-term plasticity properties, i.e., FF or FD at different presynaptic firing rates and paired pulse depression at interpulse intervals <100 ms.
In our results, we found some similarities as well as differences with the results of Czubayko and Plenz (2002). First, the approximately 34% probability of finding a connected pair in our study is comparable to the probability reported for cultured slices but higher than that found in acutely prepared slices of dorsal striatum in their study. It is possible that different experimental procedures account for this discrepancy between their study and ours, e.g., the difference in rat age (p11 vs. p23-28, respectively) and the recording temperature (22°C vs. 31-33°C, respectively). In cultured slices of Czubayko and Plenz, in which cells were recorded at 35 ± 0.5°C, the probability of finding connected pairs was similar to ours, but it should be noted that the authors found a relatively high fraction of mutually interconnected MSNs (8/26, 30.7%), while we found only 1 reciprocally connected pair out of 13 (7.7%). Besides a possible difference between dorsal and ventral striatum that may explain this discrepancy, organotypic culturing may lead to re-patterning of axonal connections between neurons.
Furthermore, Czubayko and Plenz reported some pairs to be connected by electrical synapses, contrary to our study in which no electrical synapses between MSNs were detected. Importantly, we found that cross-talk between the two input channels of our analog/digital interface could be mistaken for electrical coupling among recorded cells (see METHODS for details). Therefore our results suggest that gap junctions between MSNs are not present or occur with very low probability in the Nac in situ.
Altogether, these observations suggest that GABAergic transmission between MSNs of the ventral striatum shares certain properties with that in the dorsal striatum. Nonetheless, some differences with recent studies in the dorsal striatum are noted, which may reflect different experimental approaches or a genuine structural and physiological divergence between the two areas.
Functional implications and mechanistic aspects
GABAergic postsynaptic potentials shown in our results were depolarizing because we used a high intracellular Cl- concentration, which was useful to increase the amplitude of synaptic responses. Considering the average conductance of 0.6 nS obtained here, it is possible to estimate the amplitude of GABAergic IPSCs at a more physiological reversal potential for Cl- ions (ECl-). An estimate of ECl- (-71 mV) in Nac was obtained in a study using recording pipettes containing potassium methylsulphate (Pennartz and Kitai 1991).
Considering this value for ECl- and applying Ohm's law ICl- = GCl- x (E - ECl-) (where ICl- is the chloride current, GCl- is the GABAA receptor conductance, and E is the membrane potential), GABAergic IPSCs occurring at a membrane potential of -80 mV would be -5.4 pA. Assuming an input resistance of 100-200 M, this yields a dPSP value of approximately 0.5-1.1 mV. At membrane potentials less negative than ECl-, such as those resembling the "up states" of MSNs (from - 60 to - 40 mV), GABAergic IPSCs will be outward, and corresponding PSPs will be hyperpolarizing, indicating that GABAergic IPSPs arising from lateral contacts may significantly counteract firing of action potentials. This is consistent with the common view that GABAA receptors act to stabilize the membrane potential around the Cl- reversal potential and to lower the membrane resistance, yielding an overall inhibitory effect (Shepherd 1994). Nevertheless, our data do not rule out additional GABAA effects that may be excitatory (cf. Plenz 2003).
What is the physiological role of frequency-dependent plasticity of intrastriatal inhibitory transmission, and which mechanisms of action underlie it? Answers to these questions are still to be given, but the presently available findings permit us to advance several ideas that can be tested by future experiments.
FF may be suitable for efficacious silencing of the postsynaptic cell, probably accompanied by a capacity to suppress simultaneous potentiation of glutamatergic synapses onto it. Indeed, GABAA receptors have been shown to inhibit LTP induction of limbic inputs to Nac (Pennartz et al. 1993).
FD occurring at relatively high presynaptic rates (20 Hz) reflected a negative feedback on dPSPs amplitude to a stable level after the first dPSP in the train had reached a large amplitude. Figure 5 illustrates that the dPSP evoked by the first action potential had a large amplitude and was followed by smaller dPSPs elicited by subsequent action potentials. Notably, the first two or three dPSPs of a train evoked by high presynaptic rates were significantly larger than the first dPSP of a train evoked by low or intermediate rates. We denote this phenomenon as the "first dPSP effect." The mechanism underlying this phenomenon—which is logically independent from sequence-associated dPSP changes because it is already observed in the first dPSP—is presently unknown. However, an explanation must be sought in those parameters in the dual-cell configuration that differ between high- and low-frequency trains and precede the first dPSP. Thus it is reasonable to hypothesize that the rate of current-induced depolarization in the presynaptic cell and the first spike latency are involved, such that short-latency firing during presynaptic depolarization results in a larger dPSP. Tentatively, one may speculate that an inhibitory (K+ or Cl-) current, which slowly activates on depolarization and affects vesicle release probability, is at play here. A possible candidate may be the persistent potassium current (IKrp), characterized by Nisenbaum et al. (1996) in neostriatal slices and dissociated neurons. If, indeed, the "first dPSP" effect described here would be mediated at the level of the presynaptic terminal, it may well affect the probability and amplitudes of subsequent dPSPs in the train, thus contributing to FD. As such, the effect may lead to the depletion of the releasable pools of synaptic vesicles (Betz 1970; Dobrunz and Stevens 1997; Stevens and Tsujimoto 1995). Other factors regulating FD in the Nac may comprise the action of GABAB receptors, which are known to inhibit GABAA receptor-mediated currents in Nac (Uchimura and North 1991) and dorsal striatum (Calabresi et al. 1991; Nisenbaum et al. 1993). Alternatively, desensitization of postsynaptic GABAA receptors may occur (Jones and Westbrook 1995, 1996; Numann and Wong 1984).
Regardless of the precise mechanism of action involved in FD, we noted that FF could be restored by subsequent return to lower presynaptic rates, indicating that the induction of depression is reversible and not maintained by long-lasting mechanisms.
Synaptic dynamics and Winner-take-all model
Ventral striatal MSNs project their axons to extrastriatal structures like the ventral pallidum, substantia nigra pars reticulata and compacta, lateral hypothalamus, and ventral tegmental area, where they are thought to exert GABAergic synaptic inhibition (Groenewegen et al. 1993; Mogenson et al. 1983; Nauta et al. 1978; Walaas and Fonnum 1980; Yim and Mogenson 1980). According to anatomical results that revealed specific convergence of prefrontal cortex, hippocampus, amygdala, and thalamic projections to ventral striatal compartments and the maintenance of such topographic specificity in striatofugal projections (Groenewegen et al. 1999), it was postulated that the ventral striatum is organized into a mosaic-like pattern of neurochemically defined compartments or "ensembles" of cells processing functionally distinct patterns of information in a modular fashion (Pennartz et al. 1994). An immediate question arising from this hypothesis concerns the functional relationship between ensembles, as well as the role they may play in mediating aspects of cognition and behavior such as motivated approach behavior, instrumental conditioning, and responding for delayed rewards (Apicella et al. 1991; Cardinal et al. 2002; Hernandez et al. 2002; Kelley et al. 1997). In addition, dysfunctional communication between cells and cell groups in Nac has been proposed to contribute to psychopathological states such as schizophrenia (Swerdlow and Koob 1987).
Although these findings address neither the definition nor the nature of ensembles, they are relevant to the question as to how ensembles or modules within the Nac interact, to ensure that behaviorally relevant stimuli exert an appropriate, well-timed effect on behavior.
The ensemble hypothesis of Nac function, which attributes a major role to lateral inhibition between cell groups, can be considered in the light of more generally applicable WTA models. The WTA models are designed to solve the problem of how a given neuron or population of neurons is selected among others for engagement in information processing tasks and mediating behavioral output (Fukai and Tanaka 1997; Redgrave et al. 1999). If different MSNs subserve integration of separate pieces of information supplied by the neocortex and other glutamatergic afferent areas, some form of selection is required to avoid incoherent or unstable processing and generation of an excess of potentially conflicting output patterns.
Lateral inhibition between MSNs provided by GABAergic synapses, as we have described in this paper, is a good candidate for a mechanism achieving such output selectivity (Groves 1983). Axonal collaterals of MSNs that spread within the Nac ensure fast GABAergic inhibition of nearby cells. Thus action potential firing of a given MSN engaged by particular cues or environmental conditions may be hypothesized to dampen concurrent, glutamate-induced depolarization of connected "competitor" cells or ensembles, ensuring sharp selection of output channels and restricted targeting of extra-accumbens areas by the axonal projecting branches.
Despite the clear demonstration of lateral inhibition in the Nac, a number of theoretically relevant questions remain to be answered. First, some of the original papers speculating on striatal lateral inhibition departed from the assumption of reciprocal GABAergic connections between MSNs (e.g., Groves 1983), and it is far from clear what consequences primarily unidirectional lateral inhibition would have for selectional (WTA type) processes operating at the level of single neurons. If single MSNs are indeed organized in functionally distinct ensembles, it is less problematic to see how aggregates of individual GABAergic outputs from a given ensemble may translate into an overall "weighting factor" of lateral inhibition of that ensemble onto a competing ensemble characterized by its own reciprocal weighting factor.
Second, because the current study was limited to cells lying in each other's immediate vicinity (up to approximately 50 µm), the spatial extent of lateral inhibition remains to be examined. Preliminary anatomical tracing findings indicate that recurrent axon collaterals of MSNs may extend up to several millimeters in the Nac (Y. Van Dongen and H. Groenewegen, unpublished observations).
GABAergic lateral inhibition between MSNs within the Nac is only one part of the complex synaptic organization of this brain area, since other neurotransmitters like glutamate, acetylcholine, and numerous peptides also exert important functions. However, due to the fact that these cells contribute a large majority in Nac, GABAergic lateral inhibition is likely to have a crucial impact on striatal signal integration, and consequently, on behavioral functioning.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
These data support the hypothesis that lateral inhibition between MSNs may provide a fast and efficacious mechanism for selecting neuronal ensembles competitively involved in processing the heterogeneous flows of information that converge to the Nac from different areas of the meso-cortico-limbic system.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
GRANTS
This work was supported by Nederlandse Organisatie voor Wetenschappelijk Onderzoek Grant 903-47-092.
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: C.M.A. Pennartz, Netherlands Institute for Brain Research, Meibergdreef 33, 1105 AZ Amsterdam, The Netherlands (E-mail: c.pennartz{at}nih.knaw.nl).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Arts MP and Groenewegen HJ. Relationships of the dendritic arborizations of ventral striatomesencephalic projection neurons with boundaries of striatal compartments. An in vitro intracellular labelling study in the rat. Eur J Neurosci 4: 574-588, 1992.[CrossRef][ISI][Medline]
Baldwin AE, Sadeghian K, Holahan MR, and Kelley AE. Appetitive instrumental learning is impaired by inhibition of cAMP-dependent protein kinase within the nucleus accumbens. Neurobiol Learn Mem 77: 44-62, 2002.[CrossRef][ISI][Medline]
Balleine B and Killcross S. Effects of ibotenic acid lesions of the nucleus accumbens on instrumental action. Behav Brain Res 65: 181-193, 1994.[CrossRef][ISI][Medline]
Beiser DG and Houk JC. Model of cortical-basal ganglionic processing: encoding the serial order of sensory events. J Neurophysiol 79: 3168-3188, 1998.
Bernardi G, Marciani MG, Morocutti C, and Giacomini P. The action of GABA on rat caudate neurones recorded intracellularly. Brain Res 92: 511-515, 1975.[CrossRef][ISI][Medline]
Betz WJ. Depression of transmitter release at the neuromuscular junction of the frog. J Physiol 206: 629-644, 1970.
Bolam JP, Hanley JJ, Booth PA, and Bevan MD. Synaptic organization of the basal ganglia. J Anat 196: 527-542, 2000.
Bracci E, Centonze D, Bernardi G, and Calabresi P. Voltage-dependent membrane potential oscillations of rat striatal fast-spiking interneurons. J Physiol 549: 121-130, 2003.
Calabresi P, Mercuri NB, De Murtas M, and Bernardi G. Involvement of GABA systems in feedback regulation of glutamate-and GABA-mediated synaptic potentials in rat neostriatum. J Physiol 440: 581-599, 1991.
Calabresi P, Gubellini P, Centonze D, Picconi B, Bernardi G, Chergui K, Svenningsson P, Fienberg AA, and Greengard P. Dopamine and cAMP-regulated phosphoprotein 32 kDa control both striatal long-term depression and long-term potentiation, opposing forms of synaptic plasticity. J Neurosci 20: 8443-8451, 2000.
Cardinal RN, Parkinson JA, Hall J, and Everitt BJ. Emotion and motivation: the role of the amygdala, ventral striatum, and prefrontal cortex. Neurosci Biobehav Rev 26: 321-352, 2002.[CrossRef][ISI][Medline]
Cardinal RN, Pennicott DR, Sugathapala CL, Robbins TW, and Everitt BJ. Impulsive choice induced in rats by lesions of the nucleus accumbens core. Science 292: 2499-2501, 2001.
Chang HT and Kitai ST. Projection neurons of the nucleus accumbens: an intracellular labeling study. Brain Res 347: 112-116, 1985.[CrossRef][ISI][Medline]
Chang HT and Kitai ST. Intracellular recordings from rat nucleus accumbens neurons in vitro. Brain Res 366: 392-396, 1986.[CrossRef][ISI][Medline]
Corbit LH, Muir JL, and Balleine BW. The role of the nucleus accumbens in instrumental conditioning: evidence of a functional dissociation between accumbens core and shell. J Neurosci 21: 3251-3260, 2001.
Cowan RL, Wilson CJ, Emson PC, and Heizmann CW. Parvalbumin-containing GABAergic interneurons in the rat neostriatum. J Comp Neurol 302: 197-205, 1990.[CrossRef][ISI][Medline]
Czubayko U and Plenz D. Fast synaptic transmission between striatal spiny projection neurons. Proc Natl Acad Sci USA 99: 15764-15769, 2002.
Debanne D, Guerineau NC, Gahwiler BH, and Thompson SM. Physiology and pharmacology of unitary synaptic connections between pairs of cells in areas CA3 and CA1 of rat hippocampal slice cultures. J Neurophysiol 73: 1282-1294, 1995.
Debanne D, Guerineau NC, Gahwiler BH, and Thompson SM. Paired-pulse facilitation and depression at unitary synapses in rat hippocampus: quantal fluctuation affects subsequent release. J Physiol 491: 163-176, 1996.[ISI][Medline]
DiCiano P, Cardinal RN, Cowell RA, Little SJ, and Everitt BJ. Differential involvement of NMDA, AMPA/kainate, and dopamine receptors in the nucleus accumbens core in the acquisition and performance of pavlovian approach behavior. J Neurosci 21: 9471-9477, 2001.
Dittman JS and Regehr WG. Calcium dependence and recovery kinetics of presynaptic depression at the climbing fiber to Purkinje cell synapse. J Neurosci 18: 6147-6162, 1998.
Dittman JS and Regehr WG. Interplay between facilitation, depression, and residual calcium at three presynaptic terminals. J Neurosci 20: 1374-1385, 2000.
Dobrunz LE and Stevens CF. Heterogeneity of release probability, facilitation, and depletion at central synapses. Neuron 18: 995-1008, 1997.[CrossRef][ISI][Medline]
Fukai T and Tanaka S. A simple neural network exhibiting selective activation of neuronal ensembles: from winner-take-all to winners-share-all. Neural Comput 9: 77-97, 1997.[Abstract]
Galarreta M and Hestrin S. Spike transmission and synchrony detection in networks of GABAergic interneurons. Science 292: 2295-2299, 2001.
Gibson JR, Beierlein M, and Connors BW. Two networks of electrically coupled inhibitory neurons in neocortex. Nature 402: 75-79, 1999.[CrossRef][Medline]
Groenewegen HJ, Berendse HW, and Haber SN. Organization of the output of the ventral striatopallidal system in the rat: ventral pallidal efferents. Neuroscience 57: 113-142, 1993.[CrossRef][ISI][Medline]
Groenewegen HJ and Russchen FT. Organization of the efferent projections of the nucleus accumbens to pallidal, hypothalamic, and mesencephalic structures: a tracing and immunohistochemical study in the cat. J Comp Neurol 223: 347-367, 1984.[CrossRef][ISI][Medline]
Groenewegen HJ, Wright CI, and Beijer AV. The nucleus accumbens: gateway for limbic structures to reach the motor system? Prog Brain Res 107: 485-511, 1996.[ISI][Medline]
Groenewegen HJ, Wright CI, Beijer AV, and Voorn P. Convergence and segregation of ventral striatal inputs and outputs. Ann NY Acad Sci 877: 49-63, 1999.
Groves PM. A theory of the functional organization of the neostriatum and the neostriatal control of voluntary movement. Brain Res 286: 109-132, 1983.[Medline]
Hernandez PJ, Sadeghian K, and Kelley AE. Early consolidation of instrumental learning requires protein synthesis in the nucleus accumbens. Nat Neurosci 5: 1327-1331, 2002.[CrossRef]
