The interactions between N-methyl-d-aspartate (NMDA) and D1 dopamine receptors in the rat prefrontal cortex were examined using whole-cell recordings from pyramidal neurons. The effects of NMDA, the D1 agonist SKF38393, or both compounds combined were tested on measures of cell excitability. Both NMDA (10–100 μM) and SKF38393 (5–10 μM) independently increased the number of spikes and decreased the latency of the first spike evoked by intracellular depolarizing current pulses. Combining low doses of NMDA (5 μM) and SKF38393 (2 μM) resulted in a marked increase of cell excitability. This synergism was blocked by SCH23390, protein kinase A (PKA) inhibitors, and the Ca2+ chelator BAPTA, and reduced by nifedipine. These results indicate the presence of a dopamine– glutamate interaction in the prefrontal cortex at the postsynaptic level, by which D1 dopamine receptors may maintain NMDA- mediated responses in prefrontal cortical pyramidal neurons through both a PKA-dependent pathway and Ca2+-dependent mechanisms.
Dopamine (DA) in the prefrontal cortex (PFC) plays an important role in a variety of its functions, including working memory and executive functions (Goldman-Rakic, 1995; Williams and Goldman-Rakic, 1995; Murphy et al., 1996; Seamans et al., 1998; Romanides et al., 1999). Prefrontal glutamatergic transmission, particularly that mediated by N-methyl-d-aspartate (NMDA) receptors, also participates in these functions (Moghaddam et al., 1997; Aura and Riekkinen, 1999; Romanides et al., 1999; Wang, 1999). Thus, it is possible that DA and NMDA receptors interact in the PFC. Indeed, several electrophysiological studies have shown a variety of DA–glutamate interactions in cortical and striatal regions. Mutual control via presynaptic mechanisms has been reported, with glutamate receptors modulating DA terminals (Chéramy et al., 1986; Grace, 1991) and DA exerting a presynaptic control of glutamatergic actions (Brown and Arbuthnott, 1983; Pennartz et al., 1992; O'Donnell and Grace, 1994; Nicola and Malenka, 1997). Postsynaptic interactions have also been described (Levine et al., 1996a,b; Harvey and Lacey, 1997; Umemiya and Raymond, 1997). However, the overall nature of a DA modulation of glutamatergic transmission is still a matter of controversy, with evidence supporting both positive and negative interactions. As an attempt to bring some clarity to this issue, Levine and colleagues have shown that such inter- actions may depend on the receptor subtype involved for each transmitter; for example, D2 receptor typically decreased non-NMDA responses, whereas D1 receptors enhanced NMDA responses in striatal neurons (Levine et al., 1996a,b; Cepeda and Levine, 1998; Cepeda et al., 1998). A similar D1–NMDA interaction has been reported in nucleus accumbens neurons (Chergui and Lacey, 1999). Most studies, however, have been conducted using afferent stimulation to evoke glutamatergic activation. This can involve presynaptic interactions as a source of confounding results. In the striatum, the possibility of interactions occurring at a postsynaptic level was indicated by a D1 facilitation of responses to bath-applied NMDA (Cepeda and Levine, 1998; Cepeda et al., 1998). Although D1–NMDA interactions have been extensively studied in the striatum and have been reported in human epileptic cortical neurons (Cepeda et al., 1999), their presence and mechanisms in the normal rodent PFC remain to be established. In this study, we investigated whether the activation of D1 receptors can enhance NMDA-mediated changes in PFC pyramidal cell excitability using whole-cell recordings from rat brain slices.
Materials and Methods
Rats were obtained from Taconic Farms (Germantown, NY). All experimental procedures were carried out in accordance with the USPHS Guide for Care and Use of Laboratory Animals and were approved by the Albany Medical College Institutional Animal Care and Use Committee. Sprague–Dawley rats (24–28 days old) were anesthetized with chloral hydrate (400 mg/kg, i.p.) before being decapitated. Brains were removed and coronal slices (300 μm thick) containing the medial PFC were cut on a Vibratome (TPI Inc.). Slices were transferred into ice-cold artificial CSF (aCSF) containing (in mM): 125 NaCl, 25 NaHCO3, 10 glucose, 3.5 KCl, 1.25 NaH2PO4, 0.5 CaCl2, 3 MgCl2 (pH 7.4, osmolarity: 300 ± 5mOsm), and incubated in aCSF for at least 1h at room temperature, while constantly oxygenated with 95% O2–5% CO2. After incubation, slices were transferred to a submerged recording chamber attached to the stage of a fixed-stage upright microscope (Olympus BX50WI) and the perfusion solution was delivered with a multistaltic pump (Labconco, Kansas City, MO) in continuously flowing oxygenated aCSF (2–3 ml/min). In the recording aCSF solution, CaCl2 was increased to 2 mM and MgCl2 was decreased to 1mM. All experiments were performed at 32–34°C.
Slices were viewed using infrared-differential interference contrast (IR-DIC) video microscopy with a 40× water-immersion objective. The image was detected with an IR-sensitive CCD camera (Dage-MTI, Michigan City, IN) and displayed on a monitor. Contrast enhancement and gain controls were provided by an Insta-Gater (Dage-MTI). Digital images were stored on a computer for subsequent analysis with the aid of a BUZ video acquisition PC card (Iomega, Brea, CA). Whole-cell recordings were made from visually identified pyramidal cells located in layer V of the medial PFC. Patch pipettes (4–7 MΩ) were filled with (in mM): 115 potassium gluconate, 10 HEPES, 2 MgCl2, 20 KCl, 2 Mg-ATP, 2 Na2-ATP, 0.3 GTP and 1 Lucifer Yellow (LY) (pH 7.4, osmolarity: 285–290 mOsm). In some experiments, 2mM BAPTA (bis-(o-aminophenoxy)-N,N,N′,N′- tetra-acetic acid) isotonically replaced potassium gluconate. Tight seals (2–10 GΩ) were obtained by applying negative pressure. The membrane was disrupted with additional suction and the whole-cell configuration was obtained. Current-clamp recordings were acquired using an Axoclamp 2A amplifier (Axon Instruments, Foster City, CA), digitized with an interface board (DAP3200-415, Microstar; Bellevue, WA) and fed to a computer (Gateway PII 400) for data storage and off-line analyses. Whole-cell pipette series resistance was less than 20 MΩ and was bridge-compensated. Analyses of electrical activity were performed using custom-made software (Neuroscope; Brian Lowry, University of Pittsburgh). Drug treatment was limited to one neuron per slice. Only cells with a stable resting membrane potential (RMP) of at least –65 mV and overshooting action potentials were included in the study. Passive membrane properties (membrane potential and input resistance) and action potential characteristics (amplitude, duration and number evoked) were measured during a baseline recording period. Current–voltage (I/V) relationships were obtained by injection of depolarizing and hyper- polarizing current pulses, and input resistance was determined in the linear portion of the I/V plots in the hyperpolarizing direction. After completion of the recording protocol, the cells were filled with LY, and cell bodies and apical dendrites were seen on the monitor. All cells included in the study were identified as pyramidal neurons with LY. Whenever membrane potential or series resistance changed by >15%, the data were discarded.
All drugs were made fresh daily from stock solutions and mixed into oxygenated aCSF as needed. Drugs applied included NMDA (Sigma Chemical Co., St Louis, MO), the NMDA receptor antagonist d-2- amino-5-phosphonovaleric acid (APV; Sigma), the D1 receptor agonist SKF38393 (RBI, Natick, MA), the D1 antagonist SCH23390 (RBI), the Ca2+ chelator BAPTA (Sigma), the L-type Ca2+ channel blocker nifedipine (Sigma), the protein kinase A (PKA) inhibitors N-[2-((p-bromocinnamyl)- amino)ethyl]-5-isoquinolinesulfonamide HCl (H-89; Biomol Research Lab Inc., PA) and KT5720 (Calbiochem, La Jolla, CA). H-89, KT5720 and nifedipine were dissolved in dimethylsulfoxide (DMSO; 0.05%) as 1000× stocks and diluted in aCSF for bath application. Because of the light- sensitivity of some drugs in the solution, care was taken to protect the solution from the light and the experiments were conducted in a darkened room. Drugs were bath-applied in known concentrations, reaching the recording chamber by switching a tap in the perfusion line. New solutions reached the recording chamber within 20 s, as determined by exchanging water with a dye. Both control and drug-containing aCSF were continuously oxygenated throughout the experiments.
Data Quantification and Statistics
In each experiment, measures indicative of cell excitability were recorded; these include the membrane potential, latency of the first spike and number of spikes evoked by a 500 ms duration depolarizing (0.1–0.5 nA) current pulse. Values are expressed as mean ± SD. In most experiments, drug effects were compared with pre-drug baseline measures from the same cell using a paired Student's t-test. When different drugs were administered to a single cell, a repeated measures ANOVA (followed by a Tukey post hoc test) was used. Differences between baseline and experimental conditions were considered statistically significant when P < 0.05.
Visually Identified Cells
Whole-cell recordings were obtained from 116 layer V pyramidal neurons located within the medial PFC. These neurons were identified using IR-DIC video microscopy by their large (≥ 20 μm diameter) pyramidal-shaped cell bodies with long apical dendrites extending toward the pial surface (Fig. 1A,B). Resting membrane potential (RMP) averaged –70.5 ± 1.4mV (n = 116). All neurons recorded were silent at rest and were held for periods ranging between 15 min and 1 h. The voltage response to a series of current pulses (500 ms duration) revealed inward rectification in both the hyperpolarizing and depolarizing directions (Fig. 1C,D). Input resistance was 162.2 ± 32.1 MΩ (n = 116) calculated from the linear portion of the curves in the hyperpolarizing direction (obtained with pulses of up to –0.5 nA). In order to test for changes in cell excitability across different drug treatments, depolarizing current pulses (up to 0.5 nA, 500ms duration) were delivered every 10s. The intensity of these pulses was adjusted so that the response of the cell consisted of only one to three spikes; the average number of spikes evoked was 2.0 ± 0.2 (n = 116). In response to supra- threshold current steps, most pyramidal neurons exhibited modest spike frequency accommodation, often after an initial spike ‘doublet’ (Fig. 1E). Doublets were observed in 61% of recorded cells (n = 68), and followed by an afterhyperpolarization (AHP). With current intensities lowered so only one spike was evoked, the action potential was followed by an afterdepolarization (ADP, Fig. 1E). Other neurons responded to suprathreshold current injection with relatively regular interspike intervals, without doublets (n = 38; Fig. 1F). These 106 neurons were similar to what had been described as ‘regular spiking’ neurons (Connors and Gutnick, 1990). The remaining five neurons responded to current injection with repeated bursts of progressively decreasing amplitude action potentials that were similar to the ‘intrinsic bursting’ cells, as reviewed by Connors and Gutnick (1990).
NMDA Increased Neuron Excitability
The effects of NMDA on PFC pyramidal neuron excitability were studied by measuring changes in the number of spikes evoked with a depolarizing current pulse (0.1–0.5 nA) and in the latency of the first spike. The average number of evoked spikes in this group was 2.0 ± 0.3 (n = 25) before drug application. After ~5 min of baseline recording, NMDA (5–100 μM) was bath- applied for ~10 min. All neurons were responsive to NMDA and the effect could be detected 0.5–1 min after the onset of NMDA administration.
NMDA increased the number of evoked spikes in a dose- dependent manner (Fig. 2A,C; n = 25). In the presence of 25–100 μM NMDA, the numbers of evoked spikes were significantly higher than the values obtained before applying NMDA at P < 0.01 (paired t-test). Applying 10 μM NMDA increased the spike number at P < 0.05 (paired t-test), while 5 μM NMDA did not cause a significant increase (P > 0.05, n = 8). Similarly, a dose-dependent decrease of the first spike latency by NMDA (Fig. 2B,C; n = 25) was revealed at 10–100 μM (paired t-test); 5 μM NMDA did not produce a significant decrease of the first spike latency (P > 0.05, n = 8). Addition of the NMDA receptor antagonist APV (50 μM) to the bath blocked NMDA effects at all concentrations (data not shown).
SKF38393 Increased Neuron Excitability
We examined the effects of the D1 agonist SKF38393 (1–10 μM) on cell excitability in 27 PFC neurons. Only the highest concentrations of SKF38393 increased the number of spikes evoked by a constant-amplitude depolarizing current pulse (Fig. 3A,C). Administration of 10μM SKF38393 significantly increased this measure at P < 0.01 (paired t-test; n = 6); 5 μM SKF38393 increased evoked spikes at P < 0.05 (paired t-test; n = 6). Similarly, 5–10 μM SKF38393 reduced significantly the first spike latency (paired t-test; Fig. 3B,C). Lower concentrations SKF38393 were ineffective.
These effects were blocked by a D1 antagonist. SCH23390 (10 μM; n = 6) was applied to the bath before adding SKF38393 (10 μM), and prevented changes in the number of spikes and in the latency of the first evoked spike by SKF38393 (Fig. 3D). SCH23390 (10 μM) did not by itself change the number of spikes (from 2.2 ± 0.5 to 1.9 ± 0.5 spikes) or the first spike latency (from 79.2 ± 6.5ms to 80.2 ± 7.2ms). Subsequent administration of SKF38393 (10 μM) failed to change these values significantly; the number of evoked spikes remained at 2.4 ± 0.3 and the first spike latency was 68.4 ± 7.8 ms. A repeated measures ANOVA indicated an absence of overall treatment effect on the number of evoked spikes [F(2,15) = 0.1827, P > 0.05, n = 6] and on the latency of the first spikes [F(2,18) = 1.7725, P > 0.05, n = 6].
Combining low doses of the D1 agonist SKF38393 (2 μM) with NMDA (5 μM) increased cell excitability in all neurons tested in the medial PFC (n = 8). The number of spikes evoked by a 0.1–0.5 nA depolarizing current pulse was not affected by 5 μM NMDA alone; the addition of 2 μM SKF38393 significantly increased this value (Fig. 4A,C). A repeated measures ANOVA revealed an overall treatment effect [F(2,21) = 23.3365, P < 0.01, n = 8]. A post hoc analysis indicated a significant difference between NMDA alone and NMDA plus SKF38393 (Tukey test, P < 0.01, n = 8). Similarly, the latency of the first spike was not affected in the presence of 5 μM NMDA; however, after adding 2 μM SKF38393 into the bath, this measure was decreased (Fig. 4A,B,D). A repeated measures ANOVA revealed an overall treatment effect [F(2,21) = 5.2692, P < 0.01, n = 8]. Post hoc analysis revealed that the effect of the combined agonists was significantly different from the effect of NMDA before applying the D1 agonist (Tukey post hoc test, P < 0.01, n = 8). Preliminary data indicated that this increase in cell excitability by SKF38393 occurred when the D1 agonist was administered 5–30 min following the onset of NMDA administration.
These changes in cell excitability by combined administration of a D1 agonist and NMDA were much more pronounced than the sum of their independent actions (Fig. 4C,D), indicating a synergistic effect. Indeed, after 5 min of 5 μM NMDA or 2 μM SKF38393 treatment, the number of evoked spikes increased to 133.8 ± 7.5% (n = 8) and 127.9 ± 8.6% (n = 8) of baseline values, respectively (Fig. 4C). The sum of these independent effects (161.7 ± 11.4%) was much smaller than the increase in the spike number observed after the two agonist combined (365.6 ± 17.6%). An ANOVA analysis revealed an overall difference between these groups [F(4,35) = 45.6290, P < 0.01]. Post hoc analysis indicated a significant difference between the sum of NMDA and SKF38393 independent effects and the effects of simultaneous administration of both components (Tukey test, P < 0.01, n = 8). Similarly, treatment with either NMDA (5 μM) or SKF38393 (2 μM) decreased the latency of first spike to 89.5 ± 10.7% (n = 8) and 91.4 ± 9.5% (n = 8) of their respective controls (Student's paired t-test, P > 0.05; Fig. 4D). The sum of NMDA and SKF38393 independent effects on first spike latency (to 80.9 ± 15.6% of controls) was also much smaller than the reduction observed with combined application of both agonists (to 38.4 ± 9.2% of control). An ANOVA analysis revealed an overall difference between groups [F(4,35) = 38.6512, P < 0.01]. Post hoc analysis indicated a significant difference between the sum of independent effects of 5 μM NMDA and 2 μM SKF38393, and the effect of simultaneous administration of both compounds (Tukey test, P < 0.01, n = 8). Input resistance, measured with negative current pulses, did not increase following NMDA and SKF38393 administration. Baseline values were 160 ± 13 MΩ; following NMDA (5 μM) they were 161 ± 13 MΩ; and after adding SKF38393 (2 μM) they were 163 ± 14 MΩ (repeated measures ANOVA; P = 0.98; n = 5).
The D1 antagonist SCH23390 prevented the D1–NMDA synergism. Delivering SCH23390 (10 μM) prior to the addition of 2 μM SKF38393 effectively blocked the agonist potentiation of NMDA responses. Unlike in slices without the D1 antagonist, a combination of 2 μM SKF38393 and 5 μM NMDA failed to affect the number of evoked spikes [repeated measures ANOVA: F(3,24) = 3.2804, P > 0.05, n = 7; Fig. 4E]. Similarly, the presence of the D1 antagonist prevented the SKF38393- and NMDA-evoked decrease in latency of the first spike [repeated measures ANOVA: F(4,24) = 4.9514, P > 0.05, n = 7; Fig. 4E,F]. The D1 agonist was administered 25 min after the onset of NMDA in these experiments. As illustrated in Figure 4A, this is a time point at which SKF38393 was still effective in increasing cell excitability.
The potential role of membrane depolarization in these facilitatory effects of SKF38393 on NMDA responses was considered, as bath-applied NMDA induced a dose-dependent membrane depolarization (Fig. 2A). In the presence of 100 μM NMDA, PFC neurons depolarized 17.6 ± 1.5 mV (Student's t-test; P < 0.01, n = 4); 50 μM resulted in a 12 ± 1.2 mV depolarization (P < 0.01, n = 4); 25 μM resulted in a 10.5 ± 1.0 mV depolarization (P < 0.05, n = 4); 10 μM depolarized PFC neurons by 4.5 ± 0.7 mV (P < 0.05, n = 5). A complete recovery of these effects was observed 5–10 min after washing-out the drug. Perfusion with 5 μM NMDA (n = 8), on the other hand, did not affect the membrane potential significantly, depolarizing the neurons by only 1–3 mV. Bath-application of SKF38393 did not produce a membrane depolarization at most concentrations tested (n = 27). Only the highest dose (50 μM) resulted in a significant membrane depolarization (5.5 ± 1.2 mV; P < 0.05, n = 6). In the experiments combining 5 μM NMDA with 2 μM SKF38393, NMDA depolarized the membrane by only 2.2 ± 0.3mV, a value increased to 2.7 ± 0.3 mV after adding 2 μM SKF38393 into the bath. To control for a possible role of such depolarization in the increase in excitability observed, some experiments were performed in the presence of SKF38393 and NMDA after the membrane potential was adjusted to control levels (69.8 ± 3.4 mV, n = 6) by intracellular constant DC current injection. When NMDA and SKF38393-elicited depolarizations were corrected in this way, the number of evoked spikes and first spike latency were not affected in the presence of 5 μM NMDA; however, after adding 2 μM SKF38393 into the bath and adjusting the membrane potential again, the latency of the first spike was decreased and the number of spikes was increased. A repeated measures ANOVA indicated an overall treatment effect on the number of spikes [F(2,15) = 8.3751, P < 0.01, n = 5, Fig. 5A,B] and on the latency of the first spike [F(2,15) = 6.4522, P < 0.01, n = 6, Fig. 5A,C,D]. Post hoc analyses revealed that these values were significantly different from the effects of NMDA before applying the D1 agonist (Tukey test; P < 0.01, n = 6). To further test whether NMDA-induced depolarization was required for the enhancement of NMDA responses by SKF38393, 2 μM SKF38393 was applied to cells that instead of receiving NMDA had been depolarized by 3 mV. This was the maximal membrane depolarization observed with 5 μM NMDA. In these conditions, 2 μM SKF38393 was still ineffective in increasing the number of evoked spikes and decreasing first-spike latency (Fig. 5E). An intracellular current pulse evoked 1.8 ± 0.4 spikes before (baseline) and 2.1 ± 0.5 spikes during SKF38393 (2 μM) with a 3 mV depolarization (P > 0.05, paired t-test, n = 5). In addition, the first spike latency was 83.5 ± 7.8 ms before (baseline), and 77.3 ± 8.9ms during SKF38393 administration with a 3 mV depolarization (P > 0.05, paired t-test, n = 5). Thus, the increase in PFC neuron excitability observed is not caused by drug- induced depolarization.
The NMDA–SKF38393 Synergism Was Blocked by PKA Inhibitors
The D1 receptor is coupled to G proteins, activating adenylyl cyclase, increasing the level of cAMP and phosphorylating PKA. To determine whether the cAMP–PKA signal pathway plays a role in the observed D1–NMDA interactions, we studied the effects of two PKA inhibitors: H-89, which also blocks PKC (Chijiwa et al., 1990), and KT5720, an inhibitor selective for PKA (Kase et al., 1987). Adding H-89 (1 μM) by itself had no effect, but blocked the potentiating effect of 2 μM SKF38393 on NMDA responses. The number of spikes evoked by a 500 ms duration depolarizing (0.1–0.5 nA) current pulse was not increased by NMDA (5 μM) and SKF38393 (2 μM) in the presence of H-89 [repeated measures ANOVA: F(3,24) = 2.0329, P > 0.05, n = 7; Fig. 6A]. Similarly, the latency of the first spike was not decreased when 5 μM NMDA and 2 μM SKF38393 were applied in the presence of H-89 [repeated measures ANOVA: F(3,24) = 1.3340, P > 0.05, n = 7; Fig. 6A,B]. Furthermore, the more selective PKA inhibitor KT5720 (10 μM) blocked the potentiating effect of SKF38393 on NMDA responses (Fig. 6C,D). The combination of 5 μM NMDA and 2 μM SKF38393 failed to affect the number of evoked spikes in the presence of 10 μM KT5720 [repeated measures ANOVA: F(3,24) = 2.5242, P > 0.05, n = 7; Fig. 6C]. Similarly, the addition of 10 μM KT5720 prevented the SKF38393- and NMDA-evoked decrease in the latency of the first spike [repeated measures ANOVA: F(3,24) = 1.1542, P > 0.05, n = 7; Fig. 6C,D]. In these experiments, SKF38393 was applied after NMDA has been present in the bath for 25 min. To test for potential time-dependent changes in these responses, in a subset of neurons SKF38393 was delivered 25 min after NMDA application onset. The D1 potentiation of NMDA responses was still present (P < 0.01 for both first spike latency and number of spikes; repeated measures ANOVA; n = 5; Fig. 6E,F). Taken together, these results lead to the conclusion that PKA activation is necessary for D1–NMDA interactions controlling cell excitability in PFC pyramidal neurons.
The NMDA–SKF38393 Synergism Involves Activation of Ca2+ Conductances
D1 receptor activation may affect L-type Ca2+ channels (Higashi et al., 1989; Yang and Seamans, 1996; Hernández-López et al., 1997). Therefore, it is possible that D1–NMDA interactions also involve these channels. To assess this possibility, experiments were performed in the presence of the Ca2+ channel blocker nifedipine (Fig. 7A,B, n = 6). In these experiments, NMDA (5 μM) and SKF38393 (2 μM) still increased the number of evoked spikes in the presence of nifedipine. A repeated measures ANOVA indicated a significant treatment effect [F(3,20) = 5.3537, P < 0.05, n = 6]. Post hoc comparisons revealed that the number of evoked spikes (4.7 ± 0.8, n = 6) in the presence of NMDA, SKF38393 and nifedipine combined was significantly higher than baseline (1.9 ± 0.3, P < 0.05, Tukey test; n = 6; Fig. 7A), but not than NMDA alone (2.6 ± 0.4, n = 6). However, this value was significantly lower than that measured in neurons treated with only NMDA and SKF38393 (7.1 ± 0.9, P < 0.05, Student's t-test; n = 8, see Fig. 4A). The latency of the first spike (56.4 ± 9.2 ms, n = 6; Fig. 7A,B) was reduced by NMDA (5 μM) and SKF38393 (2 μM) in the presence of 10 μM nifedipine. A repeated measures ANOVA indicated a treatment effect [F(3,20) = 6.7194, P < 0.05]. Post hoc comparisons revealed that the latency of the first evoked spike in the presence of NMDA, SKF38393 and nifedipine was lower than in baseline (82.5 ± 8.7 ms, P < 0.05, Tukey test; n = 6) and with NMDA alone (71.5 ± 7.5ms, P < 0.05, Tukey test; n = 6). Again, this effect of NMDA, SKF38393 and nifedipine combined was markedly different from the effect of just NMDA and SKF38393 (38.6 ± 8.2 ms, P < 0.05, Student's t-test; n = 8; see Fig. 4A,B). These results indicate that although the L-type Ca2+ channel blocker nifedipine did not block the D1–NMDA synergism, it reduced the efficacy of combining both compounds on cell excitability.
To further test for the role of Ca2+ in this synergism, the Ca2+ chelator BAPTA was included in the recording electrodes in some experiments. BAPTA blocked the D1 potentiation of NMDA responses (Fig. 7C,D). The number of spikes evoked by a 500 ms duration depolarizing (0.1–0.5 nA) current pulse was not increased by 5 μM NMDA and 2 μM SKF38393 if 2 mM BAPTA was loaded into the neurons [repeated measures ANOVA: F(2,21) = 2.8565, P > 0.05, n = 7]. Similarly, the latency of the first spike was not decreased by bath-application of 5 μM NMDA and 2 μM SKF38393 if recording electrodes contained 2 mM BAPTA [repeated measures ANOVA: F(2,21) = 1.9892, P > 0.05, n = 7]. These results indicate that Ca2+ is necessary for a D1–NMDA synergistic increase in neuronal excitability.
The primary finding in this study was that NMDA and the D1 agonist SKF38393 increased PFC neuron excitability in a synergistic manner. Neuron excitability was determined by measuring the number of evoked spikes and the latency of the first spike evoked by an intracellular depolarizing current pulse. Combining marginally effective doses of NMDA (5 μM) and the D1 agonist SKF38393 (2 μM) markedly increased pyramidal cell excitability. This effect was blocked by PKA inhibitors and the Ca2+ chelator BAPTA, and was decreased but not completely eliminated by an L-type Ca2+ channel blocker.
The involvement of D1 receptors in this interaction was confirmed by its recovery to baseline value after washout of the agonist and by its blockade with a selective antagonist. High doses of NMDA or SKF38393 resulted in increased excitability along with a marked depolarization. However, in the experiments combining NMDA and SKF38393, lower doses were used (5 μM and 2 μM, respectively) and only a slight depolarization was observed (1–3 mV). We are confident that the interaction between SKF38393 and NMDA does not arise from a membrane depolarization at the cell body because this synergistic effect was also observed when the membrane potential was adjusted to pre-drug values and was not observed by direct depolarization to the membrane potential typically observed in the presence of NMDA. A role of dendritic depolarization cannot be ruled out, however, given the space-clamp limitations of our preparation.
A number of mechanisms may be involved in a D1–NMDA potentiation. Presynaptic mechanisms and the participation of interneurons are potential sites for DA–glutamate interactions. However, our study assessed cell excitability as the response to intracellular current injection; no synaptic activation was involved in the measures taken. Therefore, it is likely that these actions involve second messenger pathways linking both D1 and NMDA receptors within the recorded neurons. The latency and frequency of action potential firing evoked by intracellular current injection depend on the injected current propagating to the initial segment, where action potentials are generated; however, NMDA may act on glutamate receptors located on dendritic spines (Aoki et al., 1994) and therefore enhancing the effect of the injected current on voltage-gated Na+ channels (Fig. 8A). Another possibility is that NMDA acts at sites that are more proximal. It has been shown that the soma and proximal dendrites are highly sensitive to NMDA, whereas distal dendrites are more sensitive to AMPA (Dodt et al., 1998). But even in that case, NMDA effects on cell excitability may involve distal dendritic locations since current injected at the cell body has been shown to enhance detection of incoming dendritic signals in pyramidal neurons (Stuart et al., 1997). D1 receptors, on the other hand, are coupled to adenylyl cyclase; their activation will increase cAMP formation and PKA activity (Johansen et al., 1991; Surmeier et al., 1995; Price et al., 1999). A D1-activated phosphoprotein (DARPP32) has been shown to phosphorylate voltage-gated Na+ channels, NMDA receptors and other ion channels (Snyder et al., 1998, 2000; Yan et al., 1999), providing a link between D1 receptor activation and increases in cell excitability. Furthermore, it has been reported that a D1–NMDA interaction in the nucleus accumbens involves a D1–PKA pathway (Konradi et al., 1996; Snyder et al., 1998). It is conceivable that a similar mechanism exists in PFC pyramidal neurons. To test for that possibility, we performed some experiments in the presence of PKA inhibitors. These experiments resulted in the blockade of the D1–NMDA synergism, indicating that these interactions may involve PKA. Thus, it is possible that activation of the PKA–cAMP–DARPP32 pathway by the D1 agonist resulted in an increased efficacy of NMDA receptors. Indeed, D1 receptors have been shown to increase NMDA currents in striatal neurons (Cepeda et al., 1998). Another potential site for a D1–NMDA synergism is the phosphorylation of AMPA receptors by DARPP32 (Snyder et al., 1998, 2000); the consequent depolarization would enhance NMDA responses. This is not likely the case in our experiments, since a role of depolarization was ruled out. AMPA receptors, however, could be indirectly activated by glutamate release secondary to NMDA activation. This could contribute to the responses observed. An alternate mechanism for a D1 enhancement of NMDA responses could involve L-type calcium channels as described in striatal neurons (Galarraga et al., 1997; Hernández-López et al., 1997; Cepeda et al., 1998) or voltage-gated Na+ channels located in dendrites. The PKA–cAMP–DARPP32 pathway can phosphorylate these channels (Snyder et al., 1998), enhancing Ca2+ currents in striatal neurons (Surmeier et al., 1995). Although DARPP32 is strongly expressed in the striatum, it was also found in pyramidal cortical neurons (Berger et al., 1990; Ouimet et al., 1992). Therefore, D1 receptor activation could enhance the impact of NMDA receptors at this level (Fig. 8C).
Our finding of a D1–NMDA synergism is in line with recent data obtained from striatal slices indicating that D1 receptors may enhance NMDA responses (Levine et al., 1996a,b). Previous studies addressing DA actions on PFC neurons, however, have resulted in conflicting data. Bath-application of DA has resulted in a decrease of PFC neuron excitability in some studies (Geijo-Barrientos and Pastore, 1995; Zhou and Hablitz, 1999) and in an increase in others (Penit-Soria et al., 1987; Yang and Seamans, 1996; Shi et al., 1997; Ceci et al., 1999). It is now apparent that the decrease in excitability is mediated by D2, rather than D1, DA receptors (Gulledge and Jaffe, 1998). In the Gulledge and Jaffe (1998) study, however, a D1-mediated increase in excitability was not observed. This could be the result of the experiments being conducted at room temperature. Ours and others (Penit-Soria et al., 1987; Yang and Seamans, 1996; Shi et al., 1997; Ceci et al., 1999) were conducted at 31–35°C and all reported that either DA or D1 agonists increase excitability. In their subset of experiments at 31°C, Gulledge and Jaffe (Gulledge and Jaffe, 1998) still observed a DA-mediated decrease in excitability; however, the effects of D1 agonists were only tested at room temperature in that study.
A D1 agonist-enhancement of NMDA responses in the PFC may have significant implications to our understanding of functional aspects of this brain region. This may explain our recent finding that ventral tegmental area (VTA) stimulation maintains the depolarized membrane potential that defines the ‘up’ state in PFC neurons recorded in vivo (Lewis and O'Donnell, 2000). This action of VTA stimulation was blocked by a D1 antagonist, indicating a role of D1 receptors in maintaining the ‘up’ state of PFC pyramidal neurons. Furthermore, data from behaving animals also indicate a D1 modulation of glutamatergic responses in the nucleus accumbens and striatum (Pierce and Rebec, 1995; Kiyatkin and Rebec, 1996).
It is apparent that an interaction between D1 and NMDA receptors is critical for cognitive functions, including working memory (Goldman-Rakic, 1995; Moghaddam et al., 1997). It is possible that when DA systems become active, typically in response to salient external stimuli that demand the animal's attention (Schultz, 1997), the ongoing activity in the PFC is reinforced by this D1-mediated enhancement of glutamatergic transmission (Jay et al., 1996; Seamans et al., 1998; O'Donnell, 1999). This may result in an ensemble of neurons maintained in their depolarized ‘up’ state (Lewis and O'Donnell, 2000), facilitating plasticity mechanisms. Indeed, it has been shown that hippocampal stimulation-induced LTP in the PFC is facilitated immediately following VTA stimulation (Jay et al., 1995) and blocked by impairing DA transmission (Gurden et al. 1999). Thus, the D1 potentiation of NMDA actions that we have observed in the PFC may be important to memory and cognitive functions.
We thank Dr Mark W. Fleck for his collaboration in setting up the experiments and helpful comments. We also thank Ms Barbara L. Lewis for her technical assistance, Mr Brian Lowry for his data-acquisition software (Neuroscope), Dr James Tepper, Dr Guillermo González-Burgos and Dr Kang Jian for their help regarding the whole-cell technique. This work was supported by USPHS grant MH57683.
Address correspondence to Patricio O'Donnell, Albany Medical College (MC-136), Center for Neuropharmacology & Neuroscience, Albany, NY 12208. Email: firstname.lastname@example.org.