Abstract
Dopamine in the nucleus accumbens modulates both motivational and addictive behaviors. Dopamine D1 and D2 receptors are generally considered to exert opposite effects at the cellular level, but many behavioral studies find an apparent cooperative effect of D1 and D2 receptors in the nucleus accumbens. Here, we show that a dopamine-induced enhancement of spike firing in nucleus accumbens neurons in brain slices required both D1 and D2 receptors. One intracellular mechanism that might underlie cooperativity of D1 and D2 receptors is activation of specific subtypes of adenylyl cyclases by G-protein βγ subunits (Gβγ) released from the Gi/o-linked D2 receptor in combination with Gαs-like subunits from the D1 receptor. In this regard, dopaminergic enhancement of spike firing was prevented by inhibitors of protein kinase A or Gβγ. Furthermore, intracellular perfusion with Gβγ enabled D1 receptor activation but not D2 receptor activation to enhance spike firing. Finally, our data suggest that these pathways may increase spike firing by inhibition of a slow A-type potassium current. These results provide evidence for a novel cellular mechanism through which cooperative action of D1 and D2 receptors in the nucleus accumbens could mediate dopamine-dependent behaviors.
Introduction
Dopamine (DA) in the nucleus accumbens (NAcb) has long been considered an important modulator of addiction and goal-directed behaviors (Spanagel and Weiss, 1999). The shell region of the NAcb in particular is implicated in a number of cellular and behavioral phenomena, especially in relation to addictive drugs (Zahm, 1999). Although a diversity of functional effects and ionic targets can be modulated by DA, the exact role of DA receptor activation in the NAcb is only partially understood (Greengard et al., 1999; Nicola et al., 2000). DA receptors are generally grouped into two subfamilies, the D1-like receptors and the D2-like receptors (Missale et al., 1998). Opposing influences of D1 and D2 receptor activation on cAMP-dependent signaling have been reported in many studies (Stoof and Kebabian, 1981; Missale et al., 1998), with D1 receptors acting through the stimulatory Gs-like Golf, and D2 receptors acting through the inhibitory Gi/o proteins.
In contrast, results from a number of behavioral studies suggest a cooperative interaction of D1 and D2 receptors in the NAcb (Plaznik et al., 1989; Chu and Kelley, 1992; Wolterink et al., 1993; Phillips et al., 1994; Hodge et al., 1997; Ikemoto et al., 1997; Gong et al., 1999; Koch et al., 2000; Nowend et al., 2001). For example, animals will self-administer D1 and D2 agonists directly into the NAcb in combination but not alone (Ikemoto et al., 1997). Furthermore, self-administration of amphetamine (Phillips et al., 1994) and ethanol (Hodge et al., 1997), lever pressing for a conditioned reinforcer (Chu and Kelley, 1992; Wolterink et al., 1993), and evaluating the relative cost of obtaining a reward (Koch et al., 2000; Nowend et al., 2001) all may involve coactivation of D1 and D2 receptors in the NAcb. Dose-dependent modulation of firing rate and TTX-independent Fos induction by cooperative D1 and D2 receptor activation have also been reported (Chiodo and Berger, 1986; Wachtel et al., 1989; Williams and Millar, 1990; Hu and White, 1997; LaHoste et al., 2000).
Despite the implication of D1and D2 receptor cooperativity in several behaviors, the specific cellular and biochemical pathways that mediate the interaction between D1 and D2 receptors are uncertain. The present study was designed to examine the effects of DAergic agonists on spike firing in medium spiny neurons (MSNs) in the NAcb shell and the cellular mechanisms that might mediate these effects.
Materials and Methods
Slice preparation and electrophysiology. Coronal slices (300 μm) were prepared from male postnatal day 22–28 Sprague Dawley rats (50–80 gm). After cutting, slices recovered at 32°C in carbogen-bubbled artificial CSF (ACSF (in mm: 126 NaCl, 1.6 KCl, 1.2 NaH2PO4, 1.2 MgCl2, 2.4 CaCl2, 18 NaHCO3, and 11 glucose, pH 7.2–7.4, 301–305 Osm) for 30 min to 5 hr. During experiments, slices were submerged and continuously perfused (using a peristaltic pump, ∼2 ml/min) with carbogen-bubbled ACSF warmed to 31–32°C and supplemented with CNQX (10 μm, to block AMPA-type glutamate miniature EPSPs), picrotoxin (50 μm, to block GABAA receptors), and sodium metabisulfite (50 μm), an antioxidant to preserve DAergic reagents (Nicola and Malenka, 1997). CNQX and picrotoxin were added to isolate the cell from several major sources of neurotransmitter input whose release is known to be inhibited by dopamine (Pennartz et al., 1992a; Nicola and Malenka, 1997). In preliminary experiments, DA-induced increases in spike firing were observed in the absence of these three reagents (data not shown). All reagents were bath-applied.
All experiments were performed using whole-cell recording, except where specifically indicated that an amphotericin perforated patch was used. Patch clamping was performed using visualized infrared differential interference contrast with 2.5–3.5 MΩ electrodes. Current pulses were applied using Clampex 8.0 and an Axo-1D patch amplifier in current-clamp mode (Axon Instruments, Foster City, CA). On breaking into neurons, the resting membrane potentials were between -95 and -80 mV. In most experiments, the membrane potential for each neuron was set to approximately -85 mV using the patch amplifier ∼5 min after breaking into a cell, except for experiments involving intracellular perfusion with okadaic acid, norokadaone, or Gβγ subunits, in which the resting membrane potential of cells was maintained at given potential (either -90 or -80 mV) throughout experiments to compensate for small drifts in membrane potential that can occur after break-in. Series resistance correction was 15–20 MΩ. To record EPSPs, a bipolar stimulating electrode was placed ∼100 μm lateral to the recording electrode. Afferents were stimulated with 10 pulses at 20 Hz (20 μsec) every 30 sec using a Master 8 (AMPI, Jerusalem, Israel), and EPSPs were recorded using Clampex in current-clamp mode.
For voltage-clamp experiments, ACSF was modified to contain 3.6 mm MgCl2, 0.2 mm CaCl2, 0.1 mm CdCl2, and 1 μm TTX to block calcium and sodium current. Neurons were held at -80 mV with the patch amplifier. Our voltage protocol for determining effects of DA and 4-aminopyridine (4-AP) on potassium currents involved a series of 11 steps. In each step, a neuron was hyperpolarized to -100 mV for 700 msec and then depolarized for 300 msec. The depolarization ranged from -80 to +20 mV across the 11 steps. This protocol was applied every 30 sec. For analysis, magnitude of the total evoked potassium current was determined 280 msec into the current response after depolarization to 0 mV.
All data are shown as mean ± SEM. Unless otherwise indicated, all statistics were performed using a two-tailed, unpaired t test.
Analysis of spike firing. All such data were analyzed using Axograph (Axon Instruments, Foster City, CA). To calculate percent change in spiking, a current pulse was selected that exhibited approximately seven or eight spikes at baseline. The same current pulse was used for all time points of a given cell. Spike firing rates during the 3 min before addition of the reagent were averaged, and this value was normalized to 100%. Statistical significance was determined for the average spike firing change during the last 2 min of exposure to reagents. In most cases, statistical significance of changes in spike rate for a particular experimental condition was determined in comparison with an appropriate control condition. For experiments in which we systematically varied resting membrane voltage (Vm-rest) and determined the relationship between change in Vm-rest and change in spike firing, we first determined, for each cell, the change in number of spikes per millivolt change in Vm-rest. Then, to make these data comparable with the percent change in spike firing measured with DAergic agonists, we chose a baseline number of spikes of 7.2 (the mean number of spikes before addition of DAergic agonists in the cells shown in Fig. 5C) and determined the percent increase in spike firing per millivolt change in Vm-rest for each cell.
We attempted to estimate the proportion of neurons that responded to a given treatment, and, as is commonly observed (Uchimura et al., 1986; Surmeier et al., 1995), we found that most but not all MSNs responded to DAergic agonists. Approximately 70% of cells responded under whole-cell conditions, and ∼80% of cells responded under perforated patch conditions (with a threshold of 15% increase in spiking to be considered responding), suggesting that a small washout of some signaling molecules might have occurred under whole-cell conditions. However, to avoid the arbitrary designation required to delineate cells as responders or not, we included all neurons exposed to a given condition in all of our analyses.
Reagents. Whole-cell experiments were performed with potassium methanesulfonate- or potassium chloride-based solutions. KMeS/KOH, 0.95% (v/v); methanesulfonic acid, 0.76% (v/v); 20 mm HEPES; 0.2 mm EGTA; 2.8 mm NaCl; 2.5 mg/ml Mg-ATP; 0.25 mg/ml GTP, pH 7.2–7.4, 275–285 Osm; and KCl/KCl, 144 mm replacing KOH and methanesulfonic acid. Amphotericin experiments used the methanesulfonate-based solution, except without ATP and GTP. Amphotericin was made fresh as a 30 mg/ml stock in DMSO, sonicated, added at 0.2% (v/v) to a pipette solution containing 0.25 mg/ml pluronic acid, and sonicated again. Peptide sequences were as follows: FVIII, YEDSYEDISAYLLSKNNAIPR; and SPβγ, DALRIQMEERFMASNPSKVSYEPIT (Ma et al., 1997; synthesized by Synpep, Dublin, CA). Peptides were prepared as a 500× stock in DMSO and kept at -80°C. Peptides were used within 2 weeks of dilution in DMSO. In half of the peptide experiments, the experimenter did not know the identity of the peptide. Purified bovine whole-brain Gβγ subunits (Calbiochem, Lucerne, Switzerland; Huang et al., 1998) were aliquoted and kept at–80°C. Maltose-binding protein (MBP; a generous gift from Dr. Dorit Ron and Alicia Vagts, Ernest Gallo Clinic and Research Center), which is approximately the same molecular weight as Gβγ (∼50 kDa for MBP and ∼46 kDa for Gβγ), was prepared in the same vehicle as Gβγ subunits, with a final concentration of 20 μm DTT, 20 μm EGTA, and 0.002% Lubrol.
Most reagents were prepared fresh each day, including 4-aminopyridine, DA, sodium metabisulfite (all in Ringer's solution), ω-conotoxin-GVIA and ω-agatoxin (in water), eticlopride, propylnorapomorphine (NPA), okadaic acid, quinpirole, SCH23390, and U0126 (all in DMSO), and nifedipine (in 95% ethanol; Bargas et al., 1994). Dendrotoxin (Alomone Labs, Jerusalem, Israel), CNQX, Rp isomer of cAMPS (Rp-cAMPs) (Biolog, La Jolla, CA), and amphetamine were dissolved in water and kept at -20°C. SKF81297, SKF82957, forskolin, and dideoxy-forskolin were dissolved in DMSO and kept at -20°C. More than 85% of D1 agonist experiments used SKF81297, and, when tested, we saw similar results with both. Picrotoxin was dissolved in water and kept as a room temperature stock. Unless otherwise indicated, all reagents were made at 1:1000 stock and were purchased from Sigma (St. Louis, MO) or Research Biochemicals (Natick, MA).
Results
Coactivation of D1 and D2 receptors increases spike firing
Coherent excitatory synaptic inputs in vivo drive MSNs from a strongly hyperpolarized state, the “down state,” to a depolarized state, the “up state,” which is close to the threshold for action potential generation (Plenz and Kitai, 1998; Wickens and Wilson, 1998; Nicola et al., 2000). Although dopamine can have a number of effects within the basal ganglia (Greengard et al., 1999; Nicola et al., 2000), including modulation of release of several transmitters (McGinty, 1999), we focused on the postsynaptic effects of dopamine receptor activation on spike firing. Understanding how dopamine could modulate spike firing is critical, because spike firing is a major mechanism by which neurons process information. In addition, there is considerable interest in modulation of spike firing of NAcb MSNs in relation to behavioral events (Schultz et al., 1992; Bowman et al., 1996).
We used two criteria to restrict our investigation to MSNs. First, we only recorded from medium-sized cells to exclude the much larger cholinergic interneurons. The majority of neurons showed the slow, repetitive spike-firing pattern typically reported for MSNs (Nisenbaum et al., 1994; Plenz and Kitai, 1998; Wickens and Wilson, 1998; Mahon et al., 2000), and all such neurons were included for study. A small proportion of cells (∼5%) exhibited a clearly different firing pattern, with higher rates of firing, a larger fast afterhyperpolarization, and a more depolarized resting membrane potential. These properties are typical of the fast-spiking GABAergic interneurons (Plenz and Kitai, 1998; Bracci et al., 2002), and these cells were not analyzed further.
To test the firing properties of MSNs during continuous depolarization, a series of 300 msec current pulses was delivered to an MSN every 30 sec. The current pulses ranged from -100 pA (hyperpolarizing) to +350 pA (depolarizing, both subthreshold and suprathreshold for action potentials) in 50 pA steps (Fig. 1A). Bath application of either 75 or 30 μm DA significantly and reversibly elevated spike firing (75 μm, Fig. 1B–E; n = 10 and 27 for perforated patch and whole-cell experiments, respectively; 30 μm, 17.3 ± 6.6%; n = 7; both p < 0.05, paired t test), but 10 μm DA did not (1.9 ± 4.7%; n = 5). Enhancement of spike firing was relatively slow to develop, reaching a peak ∼5–7 min after application of DA. Spike firing was also significantly enhanced by amphetamine (10 μm, 18.5 ± 5.4%; n = 5; p < 0.05, paired t test), which causes release of DA by reversal of the DA transporter (Seiden et al., 1993). However, spike firing was not altered by a selective D1 agonist alone (SKF81297 or SKF82957, 1–10 μm, Fig. 2A1,B,D) or by a selective D2 agonist alone (quinpirole, 1–10 μm, Fig. 2 A1,B,D; NPA, 3 μm, Fig. 2D). Instead, spike firing was significantly increased after exposure to a D1 agonist and the D2 agonist quinpirole in combination with 3 μm of each (Fig. 2D; p < 0.05, paired t test) or 10 μm of each (Fig. 2A2,B,D; p < 0.05 vs D1 or D2 agonist alone) but not with 1 μm of each (Fig. 2D). Enhancement of spike firing was also observed using a D1 agonist in combination with NPA, a D2 receptor agonist structurally unrelated to quinpirole (Fig. 2D). Thus, only D1 and D2 agonists in combination produced an enhancement in spike firing similar to that observed with DA.
If the DA-mediated increase in spike firing required cooperative activation of D1 and D2 receptors, then either a D1 or a D2 antagonist should block this activation. DA-mediated enhancement of spike firing was prevented by preexposure to either the D1 antagonist SCH23990 (1 μm, -0.5 ± 4.6%; n = 5; 10 μm, Fig. 2C; n = 6; both concentrations p < 0.05 vs DA without antagonists) or the D2 antagonist eticlopride (300 nm, 2.5 ± 5.5%; n = 6; 3 μm, Fig. 2C; n = 11; both concentrations p < 0.05 vs DA without antagonists). Therefore, DA-mediated increases in spike firing required activation of both D1 and D2 receptors.
To address whether D1 and D2 receptor signaling might involve a synaptically released factor, slices were preincubated for 15–60 min with irreversible antagonists of the N-type (ω-conotoxin GVIA, 500 nm) and P/Q-type (ω-agatoxin IVA, 250 nm) calcium channels, as well as continuous exposure to the L-type calcium channel antagonist nifedipine (30 μm). This treatment completely inhibited evoked glutamatergic EPSCs even 1 hr after exposure to toxins (data not shown) but did not prevent the enhancement in spike firing by DA (24.0 ± 5.0%; n = 6; p < 0.05, paired t test). These data suggest that the DA-mediated signaling events did not require a synaptically released factor.
Because firing of NAcb neurons in vivo usually requires glutamatergic excitation to elicit action potentials (Plenz and Kitai, 1998; Wickens and Wilson, 1998; Nicola et al., 2000), we determined whether activation of DA receptors would increase the number of spikes evoked during synaptically driven spike firing. Thus, using 10 pulses at 20 Hz (with stimulation current set to evoke four or five spikes in the basal condition), we found that exposure to a combination of 3 μmSKF81297 and quinpirole significantly enhanced the number of spikes elicited by synaptically driven excitation (Fig. 2E; from 4.7 ± 0.3 to 6.8 ± 0.7 spikes; n = 5; p < 0.05). These results suggest that DA receptor activation can enhance spike firing under conditions that more closely mimic the in vivo situation.
DA receptor-mediated increase in spike firing requires cAMP and G-protein βγ subunits
Several studies suggest that protein kinase A (PKA) plays a major role in DA signaling (Greengard et al., 1999). Addition of 100 μm Rp-cAMPS (Rp), an inhibitor of cAMP-dependent processes, during the DA response significantly reduced spike firing to pre-DA levels (Fig. 3A; n = 4 with Rp; n = 5 without Rp; p < 0.01). However, Rp alone did not affect basal firing activity (-3.8 ± 2.6%; n = 4). These data suggest that the cAMP system did not regulate the spike rate under basal conditions but was required for expression of the increased firing rate observed during exposure to DA. In support of this possibility, forskolin (5 μm; n = 12), an activator of adenylyl cyclases, increased spike firing to a similar degree as DAergic agonists, whereas dideoxyforskolin (5 μm; n = 6), an inactive analog of forskolin, had no effect (Fig. 3B).
One intracellular mechanism that might underlie cooperativity of D1 and D2 receptors is activation of specific subtypes of adenylyl cyclases by G-protein β-gamma subunits (Gβγ) released from the Gi/o-linked D2 receptor in combination with Gαs-like subunit signaling from the D1 receptor. This Gαs/Gβγ interaction allows Gi/o-linked receptors to contribute to, rather than oppose, activation of the PKA system (Fig. 3D) (Sunahara et al., 1996; Watts and Neve, 1997). In this regard, dialysis of neurons with 200 μm SPβγ (n = 9), an inhibitory peptide that interferes with binding of Gβγ to several targets (Ma et al., 1997), prevented the increase in spike firing elicited by D1and D2 coactivation, whereas a 200 μm concentration of the inactive peptide FVIII (n = 10) had no effect (Fig. 4A,B; p < 0.05). Activation of spike firing by forskolin was not prevented by either SPβγ or FVIII (Fig. 4C; n = 4 and 6, respectively), indicating that the inhibition mediated by SPβγ was upstream of adenylyl cyclase.
If D1and D2 receptor cooperativity occurred via the Gβγ-dependent mechanism described above, we would predict that, after intracellular perfusion with Gβγ subunits, D1 receptor agonists should enhance spike firing, whereas D2 receptor agonists should not. Intracellular perfusion with purified bovine brain Gβγ subunits (20 nm) had no effect alone (3.7 ± 3.6% change in spiking 10 min after break-in; n = 9). However, a D1 agonist (SKF81297, 10 μm; n = 5) but not a D2 agonist (quinpirole, 10 μm; n = 4) significantly enhanced spike firing in cells perfused with Gβγ (Fig. 4D; p < 0.05). The D1 agonist did not enhance spike firing in neurons perfused with the Gβγ vehicle plus maltose-binding protein, which is similar in size to Gβγ, (-2.5 ± 2.5%; n = 5). Although we cannot completely rule out effector sites for Gβγ other than adenylyl cyclases, taken together, these data strongly suggest that Gβγ and the cAMP system are required for the D1- and D2-mediated enhancement of spike firing.
PKA signaling can involve other downstream signaling molecules such as MAP kinase (Impey et al., 1998) and protein phosphatase 1 (PP1) (Surmeier et al., 1995; Schiffmann et al., 1998; Greengard et al., 1999). The MAP kinase inhibitor U0126 (10 μm), which blocks long-term potentiation in cortical neurons in a slice (Di Cristo et al., 2001), did not prevent DA-induced enhancement of spike firing (22.9 ± 5.8%; n = 4; p < 0.05, paired t test). However, spike firing was significantly enhanced by intracellular perfusion of okadaic acid (1 μm; Fig. 3C; n = 5; p < 0.01, paired t test), an inhibitor of PP1 and PP2A, and this enhancement occluded DA-mediated changes in spike firing (p > 0.1, paired t test). Norokadaone (1 μm), an inactive analog of okadaic acid, had no effect by itself and did not prevent the effects of DA (Fig. 3C; n = 5). In several studies, okadaic acid mimics the inhibition of PP1 by DA and cAMP-regulated phosphoprotein of Mr 32,000 (DARPP-32) (Surmeier et al., 1995; Schiffmann et al., 1998), which is normally PKA-dependent (Greengard et al., 1999), suggesting that the DA- and PKA-mediated enhancement of spike firing observed here may involve signaling through DARPP-32.
Role of slow A-type potassium current in spike firing enhancement
A number of ionic mechanisms have been reported to be modulated by dopamine (for review, see Greengard et al., 1999; Nicola et al., 2000). Therefore, we first analyzed several baseline electrophysiological parameters that might be altered on exposure to DA or the combination of a D1 and a D2 receptor agonist. However, we observed no significant changes in resting membrane voltage (Fig. 5A1, Vm-rest) or input resistance (Fig. 5A2, Rinput; both p > 0.25, each parameter tested by one-way ANOVA across all groups shown in Fig. 5A1). We also performed a within-cell comparison of the spike firing change and the change in Rinput or Vm-rest. For this analysis, we grouped cells exposed to 75 μm DA and a 10 μm concentration of each of the D1 and D2 receptor agonists (n = 51 cells). Although change in Rinput did not significantly correlate with change in spike firing (p > 0.2; data not shown; Spearman rank order correlation), the change in Vm-rest nearly did (p > 0.07; Fig. 5B). To further address this issue, we systematically varied Vm-rest and determined the relationship between change in Vm-rest and change in spike firing (3.5 ± 0.4% increase in spike firing per millivolt depolarization; n = 5; Fig. 5C; for details of analysis, see Materials and Methods). Thus, for the majority of neurons shown in Figure 5B, the contribution of the change in Vm-rest to change in spike firing is negligible. Finally, in several experiments (intracellular perfusion with Gβγ, okadaic acid, or norokadaone), we maintained the Vm-rest of a neuron at a given potential (either -80 or -90 mV) throughout the experiment by injecting current through the patch amplifier while in current-clamp mode. Of course, in current-clamp mode, the membrane potential would be free to deviate from this Vm-rest in response to hyperpolarizing or depolarizing current pulses. Under these conditions, we still could observe an enhancement of spike firing with DAergic agonists, further supporting the possibility that DAergic enhancement of spike firing did not require, and could occur independently from, changes in Vm-rest. Thus, a mechanism other than changes in Vm-rest primarily accounts for the elevation in spike firing observed here, although DA-dependent depolarization of Vm-rest can occur in a minority of cells and can account for the observed changes in spike firing in these cells.
We also measured a number of features of the action potential and found that DAergic activation did not significantly change action potential threshold (APT), action potential width (APW), action potential peak (APP), fast afterhyperpolarization (fAHP), and magnitude of depolarization by subthreshold current pulses (data not shown; all p > 0.25, each parameter tested by one way ANOVA across all groups shown in Fig. 5A1). Figure 5D shows an example cell with an overlay of the action potential before and after exposure to D1 and D2 agonists in combination, whereas averaged data for APP, APT, and fAHP are shown in Figure 5E. In particular, several changes that would be predicted by alterations in the function of sodium channels (changes in APT, APW, and APP; Calabresi et al., 1987; Schiffmann et al., 1998) and the delayed rectifier potassium channel (changes in APW and fAHP; Rudy and McBain, 2001) were not observed (Fig. 5D,E).
Thus, DA or the combination of D1 and D2 receptor agonists enhanced spike firing without altering baseline parameters or several features of the action potential. Changes in calcium channel function (Surmeier et al., 1995; Cepeda et al., 1998) might underlie the observed pattern, but the experiments described above using calcium channel antagonists suggested that L-, N-, and P/Q-type calcium channels were not required for the DA-related enhancement of spike firing. Instead, our data indicate that the DAergic enhancement of spike firing was mediated by inhibition of the slow A-type potassium channel (IAs; Surmeier et al., 1991; Surmeier and Kitai, 1993; Nisenbaum et al., 1994; Gabel and Nisenbaum, 1998; Mahon et al., 2000).
We tested pharmacological inhibitors of IAs, including α-dendrotoxin (α-dtx), which is highly selective for IAs, and 4-AP, which is relatively selective for IAs at a concentration range of 5–60 μm (Surmeier et al., 1991; Nisenbaum et al., 1994). All these compounds significantly enhanced spike firing (α-dtx, 0.5 μm, Fig. 6A,B; n = 4; 4-AP, 5–60 μm, Fig. 6C,D; all p < 0.01, paired t test). The enhancement of firing observed with 10 μm 4-AP persisted in the presence of the combination of calcium channel inhibitors described above (25.7 ± 4.2%; n = 4), suggesting an action via a postsynaptic mechanism. Furthermore, α-dtx and 4-AP significantly occluded the effects of DA (α-dtx, Fig. 6B; n = 4; 10 μm 4-AP, 11.4 ± 5.1%; n = 5; 60 μm 4-AP, Fig. 6D; n = 6; all p > 0.05, paired t test testing the effect of DA). Occlusion was not simply attributable to a limitation on the number of spikes a neuron could fire after exposure to 60 μm 4-AP, because application of glutamate (200 μm) further increased spike rate (Fig. 6D; p < 0.01, paired t test). In addition, the enhancement of spike firing by okadaic acid was occluded by ∼13 min preexposure to 60 μm 4-AP (16 ± 4.3% firing change with okadaic acid in the presence of 4-AP; n = 4; p < 0.05 vs okadaic acid without preexposure to 4-AP). Taken together, these occlusion experiments suggest that DA, Gβγ, okadaic acid, α-dtx, and 4-AP all enhance spike firing by a common mechanism, inhibition of IAs.
We also examined the effects of DA and 4-AP on total potassium currents, comprising IAs, delayed rectifier, and noninactivating potassium currents (Surmeier et al., 1991; Surmeier and Kitai, 1993; Nisenbaum et al., 1994; Gabel and Nisenbaum, 1998; Mahon et al., 2000; Rudy and McBain, 2001), by performing voltage-clamp experiments in which sodium and calcium currents were blocked (for details, see Materials and Methods). DA (75 μm; n = 8) and 4-AP (10 μm; n = 6) produced similar small but significant inhibition in the potassium current evoked by a 300 msec pulse to 0 mV (Fig. 6E,F; DA and 4-AP, both p < 0.05 change in current, paired t test), consistent with previous reports showing that IAs contributes a minor amount to the total evoked potassium current (Surmeier and Kitai, 1993; Bekkers and Delaney, 2001). However, Bekkers and Delaney (2001) showed that, despite the modest contribution of IAs to the total potassium current, inhibition of IAs produces a significant enhancement in spike firing, emphasizing the critical role that IAs plays in regulation of action potential firing (Nisenbaum et al., 1994; Wickens and Wilson, 1998; Mahon et al., 2000).
Current-clamp experiments of IAs in striatal MSNs also suggest that IAs is a key regulator of the latency to firing the first action potential during prolonged depolarization (see Fig. 1A) (Nisenbaum et al., 1994; Mahon et al., 2000); thus any condition that inhibits IAs should shorten the latency to firing. As shown in Table 1, a significant reduction in latency to fire was observed after exposure to DAergic agonists, forskolin, or antagonists of IAs. These data further support the contention that DA, okadaic acid, α-dendrotoxin, and 4-AP enhanced spike firing through inhibition of IAs.
Discussion
The present study shows that DA increases spike firing in MSNs from the NAcb shell. This enhancement of spike firing requires coactivation of D1 and D2 receptors, because neither agonist alone modifies spike firing, and the effect of DA is inhibited by either a D1 oraD2 receptor antagonist. The increased spike firing after coactivation of D1 and D2 receptors is mediated intracellularly by Gβγ and cAMP-dependent processes. Finally, our biophysical and pharmacological studies suggest that enhancement of spike firing occurs primarily through inhibition of a slow A-type potassium current (IAs).
Our results may provide a cellular mechanism to explain observations from behavioral studies suggesting a cooperative action of D1 and D2 receptors in the NAcb. For example, rats will self-administer D1 and D2 agonists into the NAcb in combination but will not self-administer either alone (Ikemoto et al., 1997). Both synergistic and additive effects of D1 and D2 receptor activation in the NAcb have been reported by studies of conditioned reinforcement (Chu and Kelley, 1992; Wolterink et al., 1993). Also, results from studies of amphetamine self-administration (Phillips et al., 1994) and evaluating the relative cost of obtaining a reward (Koch et al., 2000; Nowend et al., 2001) are suggestive of a cooperative role for D1 and D2 receptors in the NAcb in behavioral expression. Although D1–D2 interaction is not observed for all behaviors involving NAcb DA (Coccurello et al., 2000), these studies suggest that D1 and D2 receptors interact cooperatively in the expression of a number of reward- and motivation-related behaviors mediated by the NAcb.
These behavioral observations are quite intriguing, given that D1 and D2 receptors are traditionally thought to oppositely couple to the G-protein–PKA system (Missale et al., 1998). Gβγ provides a mechanism by which Gs- and Gi/o-coupled receptors, such as D1 and D2, respectively, can act cooperatively to activate PKA (Sunahara et al., 1996; Watts and Neve, 1997), especially perhaps for behaviors involving NAcb PKA signaling (Self et al., 1998; Sutton et al., 2000). Interestingly, a recent paper demonstrated that self-administration of ethanol is significantly reduced after inhibition of Gβγ function in the NAcb (Yao et al., 2002), in agreement with decreased ethanol consumption observed after block of D1 or D2 receptors within the NAcb (Hodge et al., 1997). Here, enhancement of spike firing after coactivation of D1 and D2 receptors required both Gβγ- and cAMP-dependent processes. In particular, intracellular perfusion with Gβγ enabled D1 but not D2 enhancement of spike firing, indicating that Gβγ derived from D2 was required for a spike firing increase. These data raise the interesting possibility that the DAergic signaling pathway we have identified mediates self-administration of ethanol and perhaps other behaviors.
Our results are also consistent with studies suggesting that PKA plays a major role in DA signaling in MSNs (Greengard et al., 1999). We should note that there are likely to be multiple forms of DA receptor and Gβγ signaling, including presynaptic modulation (McGinty, 1999) and interaction with signaling pathways other than PKA (Seiden et al., 1993; Sunahara et al., 1996; Missale et al., 1998; Hernandez-Lopez et al., 2000). Also, although we used relatively high concentrations of DA (Pennartz et al., 1992a; Nicola and Malenka, 1997), it is likely that the high density of dopamine transporters and strong dopamine transporter activity around MSNs (Uchimura and North, 1990; Jones et al., 1995; Hersch et al., 1997) greatly reduce the extracellular concentration of DA. Finally, the relatively slow onset of response after application of agonists, which has been observed in other studies (Nicola et al., 1996), likely reflects both the time required to superfuse the slice with a reagent and the time for intracellular signaling events to occur. For example, enhancement of spike firing after bath application of direct IAs antagonists (e.g., 4-AP) still takes several minutes, suggesting that some delay in onset of drug action is to be expected in brain slice experiments.
Our data also indicate that DA might increase the firing rate of MSNs in the NAcb in vivo. However, in vivo studies of DAergic modulation of MSN spike firing have produced mixed results, with observations of both excitation and inhibition (for review, see Siggins, 1978; Nicola et al., 2000). Several factors might contribute to these discrepancies. First, DAergic reduction of firing might be attributable to inhibition of glutamate release (Pennartz et al., 1992a; Nicola et al., 1996; Nicola and Malenka, 1997). Second, several studies have observed dose-dependent effects of DA, with lower doses activating and higher doses inhibiting firing (Chiodo and Berger, 1986; Wachtel et al., 1989; Williams and Millar, 1990; Hu and White, 1997). In this regard, DA release after stimulation of the ventral tegmental area or the median forebrain bundle, which might result in more moderate DA levels compared with direct application, can strongly enhance spike firing in MSNs (Chiodo and Berger, 1986; Gonon and Sundstrom, 1996). Thus, DA likely has multiple effects, perhaps depending on dose or signaling context, but there is a strong precedent for DAergic activation of MSNs. Of particular interest are recent studies of NAcb firing in response to cues that indicate food reward (S. M. Nicola, I. A. Yun, K. T. Wakabayashi, and H. L. Fields, unpublished observations). In some NAcb cells, firing rates increase during presentation of the cue, and this enhancement of firing is greatly reduced by VTA inactivation. VTA inactivation or infusion of dopamine receptor antagonists into the NAcb also greatly inhibits behavioral responding to the cue. Taken together, these data suggest that DA enhances firing in a set of NAcb neurons, and that this change in firing may be important for proper task performance after the cue is observed.
These electrophysiological studies, taken together with the behavioral results described above, suggest that the D1 and D2 cooperative interaction we have characterized in brain slices may have particular relevance in vivo. However, we should note that, as in vivo, in vitro studies of DAergic effects on MSN function have produced mixed results (Nicola et al., 2000). Some studies from NAcb slices found no DA-related changes in postsynaptic properties (Pennartz et al., 1992a; Nicola et al., 1996; Nicola and Malenka, 1997), whereas others observed significant DA-dependent changes in input resistance or resting membrane potential (Uchimura et al., 1986; Uchimura and North, 1990; O'Donnell and Grace, 1996). The latter may reflect action of DA on cell types other than MSNs (Yasumoto et al., 2002), and such influences were negated here by studying DAergic signaling in relative pharmacological isolation. Some discrepancies may also relate to differences among MSNs from the dorsal striatum, NAcb core, and NAcb shell (Kelly and Nahorski, 1987; Calabresi et al., 1992; Pennartz et al., 1992b; O'Donnell and Grace, 1993; Paxinos, 1995; Thomas et al., 2000). In particular, D1 and D2 receptors in the dorsal striatum are more clearly segregated in the so-called “patch” and “matrix” compartments, whereas such distinction is much less clear in the NAcb shell (Paxinos, 1995). Because slice studies from the dorsal striatum have generally not addressed the compartmental localization of the neurons under investigation, differential signaling among compartments could contribute to the variety of DAergic effects observed in vitro (for review, see Nicola et al., 2000).
Pharmacological and biophysical analyses suggest that the DA-mediated enhancement of spike firing observed here was mediated by inhibition of IAs and was not associated with a change in function of several other channel types, including sodium and L-, N-, and P/Q-type calcium channels, although we should note that other groups have observed modulation of sodium channels by DA in the NAcb (Zhang et al., 1998). Pharmacological inhibitors of IAs, such as α-dendrotoxin and 4-AP (5–60 μm), enhanced spike firing and occluded the effects of DA on spike firing. Also, using voltage-clamp methods, we observed a small but significant inhibition of potassium currents by 4-AP and DA (Surmeier and Kitai, 1993). In this regard, Bekkers and Delaney (2001) found that inhibition of IAs produced only a small decrease in total potassium current but led to a significant enhancement in spike firing, consistent with the critical role IAs contributes to action potential firing (Nisenbaum et al., 1994; Wickens and Wilson, 1998; Mahon et al., 2000). Although these results, particularly occlusion of DAergic effects after inhibition of IAs function, suggest that DA enhances spike firing by inhibiting IAs, we should acknowledge that additional effects beyond those on IAs cannot be definitively excluded, and that changes in Vm-rest in particular can account for the changes in spike firing in a small percentage of cells.
On the basis of analyses of the up and down state transitions and biophysical properties, it has been suggested that potassium channels are key regulators of excitability of MSNs (Wickens and Wilson, 1998). IAs activates at voltages around spike threshold, and inhibition of IAs may allow previously subthreshold synaptic input to elicit action potential firing (Nisenbaum et al., 1994; Wickens and Wilson, 1998; Mahon et al., 2000). In agreement, we found that DAergic activation increased the number of spikes fired during synaptic stimulation. Our data predict that DA will enhance the number of spikes fired in vivo during an up-state transition or during any other coherent glutamate excitation, with little effect on the hyperpolarized down state. This is consistent with the idea that DA is modulatory and normally requires glutamate receptor activation for DAergic effects to be observed (Chiodo and Berger, 1986; Nicola et al., 2000). Thus, by inhibition of IAs, coactivation of D1 and D2 receptors in the NAcb shell could enhance glutamate-mediated cellular excitation and thereby contribute to the expression of goal-directed behaviors.
Footnotes
This work was supported by funds provided by the State of California for medical research on alcohol and substance abuse through the University of California, San Francisco (F.W.H. and A.B.), by the Alcoholic Beverage Medical Research Foundation (M.G.C.), and by Department of the Army Grant DAMD17-0110803 (F.W.H. and I.D.). The United States Army Medical Research Acquisition Activity (Fort Detrick, MD) was the awarding and administering acquisition office. We thank H. Fields, M. Melis, S. Nicola, J. Pierce-Shimomura, B. Schilstrom, and M. Ungless for assistance during preparation of this manuscript.
Correspondence should be addressed to Antonello Bonci, Ernest Gallo Clinic and Research Center, Department of Neurology, University of California, San Francisco, 5858 Horton Street, Suite200, Emeryville, CA94608. E-mail: bonci{at}itsa.ucsf.edu.
Copyright © 2003 Society for Neuroscience 0270-6474/03/235079-09$15.00/0