Dopamine–glutamate interactions may contribute to persistent electrical activity in the prefrontal cortex (PFC). We tested whether a D1 modulation of NMDA function can result in persistent depolarization in vitro. D1–NMDA co-activation yielded depolarizing plateaus resembling in vivo up states in PFC pyramidal neurons recorded in slices from adult (PD 45–65), but not pre-pubertal (PD 29–38) rats. These plateaus required intracellular Ca2+, activation of L-type Ca2+ channels and protein kinase A. These events were eliminated by intracellular administration of the voltage-gated Na+ channel blocker QX-314 or by interrupting synaptic activity with bath application of tetrodotoxin or the AMPA antagonist CNQX, suggesting that they require both intrinsic and synaptic mechanisms. These recurrent depolarizations could constitute important elements in cortical information processing, allowing synaptic plasticity and memory functions. Acquiring these PFC D1–NMDA interactions after puberty may be a critical element for developing mature cognitive abilities.
It is well known that dopamine (DA) can modulate glutamate function in the brain. The interactions between these neurotransmitters are diverse and depend on the receptor subtypes involved. A consistent finding in this regard has been a D1 enhancement of NMDA-mediated responses (Cepeda and Levine, 1998; Cepeda et al., 1998; Wang and O'Donnell, 2001). It may not be coincidence that cells in which this interaction has been observed (e.g. pyramidal cortical neurons and striatal medium spiny neurons) exhibit spontaneous glutamate-driven plateau depolarizations when recorded in vivo (O'Donnell and Grace, 1995; Wilson and Kawaguchi, 1996; Lewis and O'Donnell, 2000; Goto and O'Donnell, 2001). Indeed, in vivo intracellular recordings from prefrontal cortical (PFC) pyramidal neurons reveal a very negative resting membrane potential (down state) interrupted by plateau depolarizations (up states; Branchereau et al., 1996; Lewis and O'Donnell, 2000). Up states are driven by synchronous excitatory afferents and are periods of active information processing during which synaptic plasticity mechanisms may be enabled (O'Donnell, 2003). Although up-down transitions have been extensively characterized in anesthetized animals and sleep (Mahon et al., 2001; Steriade et al., 2001), a recent study demonstrated their presence in awake animals and a role of sensory inputs in up states (Petersen et al., 2003). There is some indication that DA may control PFC up states. For example, electrical and chemical stimulation of the ventral tegmental area (VTA) evokes plateau depolarizations resembling up states, which are shortened by a D1 antagonist (Lewis and O'Donnell, 2000). Complete DA receptor blockade fails to abolish the onset of the evoked depolarization, suggesting that the transition to the up state is mediated by a non-DA mechanism, with DA acting to sustain the depolarization. Because D1 receptors enhance NMDA function in the PFC (Zheng et al., 1999; Wang and O'Donnell, 2001; Seamans et al., 2001a; Gonzalez-Islas and Hablitz, 2003), it is conceivable that a sufficient level of D1–NMDA co-activation could elicit plateau depolarizations resembling up states in a slice preparation. To address this issue, the effects of a D1 agonist and NMDA on spontaneous activity in PFC pyramidal neurons were tested using whole-cell current clamp recordings.
The age of animals used to prepare slices is an important consideration in whole cell recordings. It is well known that postnatal developmental transformations in the brain, especially in the PFC and limbic system, are prominent during adolescence across a variety of species (for a review, see Spear, 2000). Most rat studies have been conducted to date in slices from pre-pubertal animals. It is rare to find reports of PFC recordings, for example, in rats >40 days old. Because puberty in rats occurs between postnatal day (PD) 40 and PD 45 and cortical circuits may mature at that age (Spear, 2000; Zhu, 2000), it is important to determine DA–glutamate interactions in adult animals. This becomes essential when studying persistent activity that may be relevant to in vivo up states. In vivo intracellular recordings from pre-pubertal animals (PD 28–35) have shown up states of smaller amplitude than similar recordings obtained from adult animals (O'Donnell et al., 2002). Therefore, the experiments aimed at elucidating the cellular signaling pathways involved in D1–NMDA interactions were conducted in slices obtained from animals at least 45 days old. Additional recordings were conducted in slices from younger animals to determine the age at which these interactions emerge.
Materials and Methods
Slices containing the medial PFC were obtained from adult (PD 45–65) or pre-pubertal (PD 29–38) rats. Animals were anesthetized with chloral hydrate (400 mg/kg, i.p.) before being decapitated. Brains were rapidly removed into ice-cold artificial CSF [aCSF (in mM): 125 NaCl, 25 NaHCO3, 10 glucose, 3.5 KCl, 1.25 NaH2PO4, 0.5 CaCl2, 3 MgCl2; pH 7.45, osmolarity 295 ± 5 mOsm]. Coronal slices (300 μm) were cut on a Vibratome in ice-cold aCSF, transferred and incubated in warm (33–35°C) aCSF solution constantly oxygenated with 95% O2–5% CO2 for at least 90 min before recording.
Whole-cell current clamp recordings were performed in pyramidal neurons located in deep layers of the medial PFC, identified under visual guidance 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 and displayed on a monitor. All experiments were conducted at 33–35°C. In the recording aCSF (perfused at 2 ml/min), CaCl2 was increased to 2 mM and MgCl2 was decreased to 1 mM. Patch pipettes (5–8 MΩ) were filled with (in mM): 115 K-gluconate, 10 HEPES, 2 MgCl2, 20 KCl, 2 MgATP, 2 Na2-ATP and 0.3 GTP (pH = 7.3, 280 ± 5 mOsm). Signals were acquired using a Multiclamp 700A amplifier, digitized with a Digidata 1322A (Axon) and fed to a computer for off-line analysis. In each cell, membrane potential and input resistance were analyzed before (baseline) and after drug treatment. Only cells with a stable resting membrane potential of at least −65 mV (baseline) 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 period and during drug delivery. Whole-cell pipette series resistance was <20 MΩ, and was bridge compensated by ∼80% (with the computer-controlled amplifier). Series resistance was monitored throughout the experiments and if it changed by >15%, the data were excluded for analysis. The D1 agonist SKF38393 (Sigma, St Louis, MO), NMDA (RBI) and the D1 antagonist SCH23390 (RBI) were dissolved in the external solution. For intracellular Ca2+ chelation, BAPTA (Sigma) was included in the recording micropipette at a concentration of 2 mM. L-type Ca2+ channels were blocked by bath application of nifedipine (Sigma, 10 μM). In some experiments, excitatory inputs arriving from other activated neurons contained in the slice were blocked by bath application of tetrodotoxin (TTX; Sigma, 0.05 μM) or CNQX (Sigma, 10 μM). To block protein kinase A activity, KT-5720 (Calbiochem, La Jolla, CA) was included in the bath solution. Both control and drug-containing aCSF were continuously oxygenated throughout the experiments. Typically, baseline recordings were conducted for ∼15 min before perfusing with a solution containing SKF38393 (2 μM) or NMDA (8 μM) for 5 min. When both agonists were combined, baseline recordings were followed by SKF (2 μM) for 2 min, then followed by the D1 + NMDA cocktail for 5 min. The effects of D1 + NMDA were evident during the first minute of bath application and become stable during the subsequent 2–3 min. All neurons recorded in the present study showed a similar pattern of responses to D1 + NMDA and required 10–20 min to completely wash out the effects. All data shown were obtained from the stable 2–3 min following drug administration. The doses were selected based on pilot experiments determining dose–response curves for these agents, which revealed that 8 μM NMDA is an effective concentration yielding constant depolarization and action potential firing and 2 μM SKF38393 is a marginally effective concentration that does not have a significant effect on its own, but potentiates NMDA responses (P. O'Donnell and K.Y. Tseng, unpublished observations).
The Fisher exact probability test (FET) was used to evaluate the relationship between two dichotomous variables. Student's t-test was used for two-group comparisons involving a single continuous variable. Drug effects along two or more variables were compared using repeated measures ANOVA. If data were not normally distributed or had unequal variances, Kruskal–Wallis ANOVA by ranks was preferred for multiple comparisons involving interrelated proportions. Normality was assessed with the Kolmogorov–Smirnov test and homogeneity of variances was assessed with Levene's test. Differences between experimental conditions were considered statistically significant when P < 0.05.
Whole-cell Recordings from Adult PFC Neurons
Whole-cell current clamp recordings of PFC pyramidal neurons were performed in slices obtained from adult (PD 45–65) rats. These neurons (n = 83) displayed a negative resting membrane potential (−69.8 ± 3.3 mV, mean ± SD) and had an input resistance of 120.1 ± 31.1 MΩ. Action potentials could be observed in response to depolarizing current (Fig. 1a) or occasionally during spontaneous depolarizations (Fig. 1b).
Spontaneous Plateau Depolarizations in Mature Prefrontal Cortical Slices
Spontaneous plateau depolarizations were observed sporadically during baseline recordings. These events lasted 452.9 ± 122.2 ms and occurred about once every 12 min (0.0014 ± 0.0004 Hz, n = 16). Spontaneous depolarizations were not observed in slices from immature animals (PD < 40, n = 26). Because neocortical pyramidal neurons undergo several changes in both morphological and physiological properties during postnatal development until PD 42 (Zhu, 2000), we interpreted the depolarizing events as resulting from critical glutamate activation in a mature cortical network. Bath-application of the D1 agonist SKF38393 (8 μM) significantly increased the duration of plateau depolarizations in all cells tested (n = 4, Fig. 1b), supporting the idea that D1 receptors may contribute to sustaining up states in PFC pyramidal neurons (Lewis and O'Donnell, 2000). The frequency of these events, however, remained unchanged in presence of the D1 agonist (from 0.0013 ± 0.0004 to 0.0012 ± 0.0003; P = 0.7).
Given that a D1 enhancement of NMDA-mediated responses has been observed in PFC pyramidal neurons (Wang and O'Donnell, 2001), we investigated whether combined administration of NMDA with a marginally effective dose of the D1 agonist enhanced the occurrence of spontaneous plateau depolarizations. Bath-application of SKF38393 (2 μM) did not induce significant changes in resting membrane potential (from −69.2 ± 2.5 mV to −68.8 ± 2.9 mV; n = 8; Fig. 2a) or in cell excitability measured as the response to intracellular current injection (data not shown; higher agonist doses were required to increase excitability). In contrast, 12 of 14 cells treated with 8 μM NMDA (85%) exhibited a marked membrane depolarization (from −69.4 ± 3.1 mV to −62.3 ± 4.3 mV) yielding spontaneous cell firing (Fig. 2b). Combining SKF38393 (2 μM) and NMDA (8 μM) resulted in spontaneous depolarizing plateaus resembling the membrane potential fluctuations typically observed in vivo in 10 of 11 cells tested (90%; Fig. 2c). This proportion is significantly higher than what was observed in the presence of the D1 agonist alone (n = 0/8, P = 0.00012, FET) or NMDA alone (n = 2/12, P = 0.0002, FET). The spontaneous plateaus recorded in the presence of SKF38393 + NMDA lasted 380.0 ± 169.3 ms (P < 0.0003 compared with NMDA: 102.2 ± 87.9 ms, Kruskall–Wallis ANOVA) and exhibited a frequency of 0.92 ± 0.58 Hz (P < 0.00001 compared with NMDA: 0.03 ± 0.08 Hz, Kruskall–Wallis ANOVA). The duration of these plateaus was not significantly different from those occasionally observed in control conditions. Pre-treatment with the D1 antagonist SCH23390 (10 μM) prevented the membrane potential oscillations when SKF38393 + NMDA were applied in six of eight cells tested (P = 0.0044 compared with NMDA + SKF38393, FET). What was left were the shorter (120.8 ± 118.5 ms; P < 0.005 compared with NMDA + SKF38393, Kruskall–Wallis ANOVA) sporadic (0.13 ± 0.21 Hz; P < 0.002 compared with NMDA + SKF38393, Kruskall–Wallis ANOVA) depolarizations resembling those recorded in control conditions. To address whether the increase in plateau depolarizations was dependent on NMDA receptor activation and not simply due to the depolarization elicited by NMDA, SKF38393 (2 μM) was applied while the neuron was depolarized with somatic current injection instead of NMDA administration. Combining the D1 agonist with positive current injection did not result in any depolarizing plateau, but elicited an increase in cell firing (Fig. 2e, n = 6/6). On the other hand, co-activation of D1 and AMPA receptors (SKF38393 2 μM + AMPA 0.4 μM, n = 7) depolarized PFC pyramidal neurons causing cell firing, but not membrane potential oscillations (data not shown). These results indicate that D1–NMDA receptor co-activation can induce spontaneous membrane potential fluctuations resembling up–down transitions in PFC pyramidal neurons.
Plateau depolarizations observed during D1 + NMDA activation are typically accompanied by rapid action potential firing. However, the threshold for action potential firing shifted to a more depolarized value (from −54.6 ± 2.5 mV to −51.5 ± 1.9 mV; P < 0.005, t-test; Fig. 3). This shift could be explained by an inhibitory action of D1 receptors on Na+ currents via PKA mediated phosphorylation of voltage-gated Na+ channels (Cantrell and Catterall, 2001) or by an excitatory effect of D1 receptors on interneurons yielding inhibitory postsynaptic currents that can in turn reduce pyramidal neuronal excitability (Zhou and Hablitz, 1999; Seamans et al., 2001b; Gorelova et al., 2002). Indeed, the membrane potential between plateaus in D1 + NMDA activation is similar to that of NMDA alone (see Fig. 2b,c). Yet, at that membrane potential value NMDA causes rapid firing, but adding the D1 agonist blocks firing. It is conceivable that the D1 enhancement of postsynaptic NMDA actions is sufficient to periodically overcome this inhibition.
Plateau Depolarizations Induced by D1–NMDA Require Local Network Activity
D1 + NMDA-induced membrane potential oscillations could reflect activity of a local neural network impinging on the recorded neuron. Holding the membrane potential to its baseline value failed to block plateau depolarizations induced by D1–NMDA in six of six cells tested (Fig. 4a), suggesting that voltage-dependent intrinsic currents may not be initiating the depolarizations (although they could still contribute to sustaining then). To investigate whether excitatory inputs arriving from other activated neurons contained in the slice were involved, subsets of experiments were conducted in the presence of the Na+ channel blocker TTX or the AMPA receptor antagonist CNQX. Bath-application of TTX (0.05 μM; Fig. 4b) or CNQX (10 μM; Fig. 4c) prevented the D1 + NMDA-induced membrane potential oscillations. Occasional plateau depolarizations were observed in only two of nine cells treated with TTX (P = 0.0045 compared with D1 + NMDA, FET) and one of seven cells in presence of CNQX (P = 0.0025 compared with D1 + NMDA, FET) every several minutes (∼0.01 Hz). These results indicate that membrane potential oscillations induced by co-activation of D1 and NMDA receptors in PFC pyramidal neurons may reflect the activity of a local neuronal network.
D1–NMDA-mediated Plateau Depolarizations Involve Intracellular Ca2+, L-type Ca2+ Channels, Protein Kinase A and Voltage-gated Na+ Channels
Several cellular mechanisms may also contribute to the depolarizing plateaus induced by D1–NMDA co-activation. For example, it has been shown that D1–NMDA synergism involves intracellular Ca2+ and activation of L-type Ca2+ channels (Cepeda and Levine, 1998; Cepeda et al., 1998; Wang and O'Donnell, 2001). To test whether this applies to plateau depolarizations, a subset of recordings was conducted with electrodes containing the Ca2+ chelator BAPTA (2 mM) or in presence of nifedipine in the bath. Intracellular application of BAPTA (n = 8) reduced significantly the duration of plateau depolarizations (146.1 ± 109.6 ms; P < 0.005, compared with NMDA + SKF38393, Kruskall–Wallis ANOVA; Fig. 5a) to a value within the range of those observed in control conditions. The frequency of these depolarizations was also reduced to 0.07 ± 0.14 Hz (P < 0.0006 compared with NMDA + SKF38393, Kruskall–Wallis ANOVA). Similarly, bath application of nifedipine (10 μM, n = 7) shortened the duration (184.5 ± 104.1 ms; P < 0.02 compared with NMDA + SKF38393, Kruskall–Wallis ANOVA) and decreased the frequency (0.19 ± 0.19 Hz, P < 0.005 compared with NMDA + SKF38393, Kruskall–Wallis ANOVA) of D1 + NMDA-mediated plateau depolarizations (Fig. 5b). However, plateaus observed in the presence of nifedipine were more frequent than with normal aCSF. These results suggest that both intracellular Ca2+ and L-type Ca2+ channels may contribute to plateau depolarizations in PFC pyramidal neurons.
Second messenger cascades could also be involved in this interaction. D1 receptors are coupled to the G protein Gs and adenylyl cyclase and their activation increases cyclic adenosine monophosphate (cAMP) formation and protein kinase A (PKA) activity (Surmeier et al., 1995), stimulating the phosphoprotein DARPP-32, allowing phosphorylation of NMDA receptors (Snyder et al., 1998). Indeed, synergistic interactions between NMDA and D1 receptors in the PFC involve PKA (Wang and O'Donnell, 2001). Therefore, we tested whether D1 + NMDA-dependent membrane potential oscillations in the PFC require a PKA-dependent cascade. Oscillations induced by co-activation of D1 and NMDA receptors were prevented by application of a PKA inhibitor (KT-5720, 1 μM) in the bath in 9 of 10 cells tested (Fig. 5c). Plateau potentials had a duration of 107.6 ± 165.5 ms (P < 0.005, compared with NMDA + SKF38393, Kruskall–Wallis ANOVA) and their frequency was 0.03 ± 0.08 Hz (P < 0.0005 compared with NMDA + SKF38393, Kruskall–Wallis ANOVA). These results indicate that PKA mediates the D1 + NMDA-induced membrane potential oscillations in PFC pyramidal neurons.
The blockade of plateau depolarizations with TTX does suggest that afferent activity may be needed, but it could also be explained by the participation of persistent Na+ currents. These currents are important for synaptic integration in pyramidal neurons (Crill, 1996; Yang et al., 1996). To test whether slowly inactivating voltage-gated Na+ currents are required for plateau depolarizations, a subset of recordings was conducted with electrodes containing the Na+ channel blocker QX-314 (2.5 mM). Depolarizing plateaus were observed in only one of seven cells tested (P = 0.0012, compared with SKF38393 + NMDA, FET). Short spontaneous depolarizations resembling Ca2+ spikes (Stafstrom et al., 1985) were observed instead of plateau depolarizations in the remaining six neurons. These depolarizations lasted 73.3 ± 82.1 ms (P < 0.002, compared with SKF38393 + NMDA, Kruskall–Wallis ANOVA; Fig. 5d) and exhibited a frequency of 0.61 ± 0.82 Hz. Thus, although slow Ca2+ events can be evoked by D1 + NMDA when Na+ channels are blocked in the recorded neuron, plateau depolarizations were not observed. This suggests that slowly inactivating voltage-gated Na+ channels could contribute to sustaining synaptically induced depolarizations, although a role of backpropagating dendritic Na+ spikes cannot be ruled out. Figure 6 illustrates the frequency of plateau depolarizations in all treatment groups.
D1–NMDA-mediated Plateau Depolarizations Require a Mature PFC
To determine the age at which plateau depolarizations elicited by D1–NMDA interactions emerge, an additional set of experiments were conducted in slices from pre-pubertal rats. Eighteen PFC pyramidal neurons were recorded and displayed a negative resting membrane potential (PD 29–31, −65.5 ± 2.7 mV, n = 9; PD 36–38, −67.9 ± 2.4 mV, n = 9). None of them exhibited spontaneous action potential firing, but action potentials could be evoked in response to somatic depolarizing current injection (not shown). The sporadic plateau depolarizations observed during baseline recordings in brain slices obtained from adult animals could not be detected in these younger slices. Bath application of the D1 agonist SKF38393 (2 μM) and NMDA (8 μM) resulted in membrane depolarization (PD 29–31, −57.9 ± 3.9 mV; PD 36–38, −59.3 ± 4.0 mV) with spontaneous action potential firing (Fig. 7), but failed to induce plateau depolarizations. Instead, short depolarizations were observed (PD 29–31, 41.6 ± 50.3 ms; PD 36–38, 51.4 ± 66.2 ms; P < 0.0001 compared with PD >45, Kruskall–Wallis ANOVA; Fig. 7). A regression of all data points at all ages indicates that adult levels of plateau duration were achieved at around PD 45 (Fig. 7a). These findings indicate that the D1 + NMDA-mediated plateau depolarizations occur in a mature, post-pubertal PFC.
D1–NMDA co-activation elicited plateau depolarizations resembling in vivo up states in slices obtained from adult animals. PFC pyramidal neurons exhibited a very stable and negative membrane potential with very sporadic plateau depolarizations that were increased in duration by a D1 agonist. Combining the D1 agonist with NMDA resulted in repeated plateau depolarizations in the PFC of post-pubertal, but not younger animals, that resembled in vivo up states, which were observed. These recurrent depolarizations required intracellular Ca2+, L-type Ca2+ channels and PKA signalling. Interrupting synaptic activity with bath-application of TTX or CNQX blocked these plateaus. These events were also eliminated by intracellular administration of the Na+ channel blocker QX-314. Thus, co-activation of D1 and NMDA receptors could enhance the activity of a mature PFC to the point of causing spontaneous depolarizing plateaus in individual neurons, an effect that may be mediated by both intrinsic and synaptic mechanisms.
Neocortical pyramidal neurons can exhibit persistent depolarizing activity in vitro. Recordings from visual and prefrontal cortices in mice and ferrets revealed plateau depolarizations when a bath solution that reflects more closely the K+ concentration of brain interstitial fluid in situ was used (Sanchez-Vives and McCormick, 2000; Cossart et al., 2003). This persistent activity seems to be driven by local synaptic activity and intracellular Ca2+ increase, as indicated by Ca2+ transients observed during the depolarizations (Cossart et al., 2003), probably reflecting activation of somatic voltage-sensitive Ca2+ channels (Smetters et al., 1999; Mao et al., 2001; Cossart et al., 2003). The D1 + NMDA-mediated plateau depolarizations reported here were prevented by interrupting local synaptic activity and shortened by adding the Ca2+ chelator BAPTA or the Na+ channel blocker QX-314 in the recording electrodes. These events were also shortened with bath application of the L-type Ca2+ channel blocker nifedipine. Thus, it seems likely that the cellular and synaptic mechanisms involved in these recurrent depolarizations induced by D1 + NMDA interactions in the PFC are similar to those observed in the visual cortex (Mao et al., 2001; Cossart et al., 2003), involving a combination of increased network activity driving synaptic responses and slow intrinsic currents in pyramidal neurons sustaining the depolarizations.
The precise cellular or network mechanisms involved in generating and sustaining these in vitro depolarizations or the in vivo up states are not clear. Computational models have shown that increasing NMDA and GABA conductances by D1 receptors could lead to persistent activity (Durstewitz et al., 2000; Brunel and Wang, 2001) and that these interactions may play an important role in cognitive functions (for a review, see Cohen et al., 2002). Even in control conditions, plateau depolarizations were sporadically observed in mature brain slices. In presence of the D1 agonist, these events lasted longer. This suggests that there could be some active neurons in the slice in baseline conditions that may occasionally reach a critical level of firing, yielding spontaneous plateaus. If D1 receptors are activated, these plateaus may increase their duration, probably via an enhancement of NMDA responses (Wang and O'Donnell, 2001). It has been suggested that plateau potentials may reflect both action potential backpropagation and NMDA receptor activity (Kerr and Plenz, 2002). Thus, it is possible that Ca2+ transients driven by action potential backpropagation mediate recurrent depolarizations induced by D1 + NMDA interactions in the mature PFC. Although action potential backpropagation in an immature PFC (PD 21–35) is not affected by DA (Gulledge and Stuart, 2003), the possibility of a DA modulation of backpropagation in the PFC remains to be explored in mature animals. A D1 + NMDA synergism would cause an enhanced level of activation in many pyramidal neurons in the recorded slice. Their activation would in turn increase synaptic drive of neighboring neurons (Beggs and Plenz, 2003), with the potential consequence of synaptic currents and slowly inactivating conductances sustaining the depolarization. The repeated depolarizations observed could therefore be a consequence of combination of network and cellular events.
The appearance of D1 + NMDA-dependent membrane potential oscillations was prevented by impairing Ca2+ and PKA signalling as well as by blocking L-type Ca2+ channels. This is in line with recent data obtained from striatal and cortical slices indicating that D1 receptors enhance NMDA responses (Cepeda and Levine, 1998; Cepeda et al., 1998; Wang and O'Donnell, 2001) through a signalling cascade involving PKA activity (Surmeier et al., 1995) and phosphorylation of DARPP-32 (Snyder et al., 1998; Flores-Hernandez et al., 2002). The observed increase in NMDA currents could be due to phosphorylation of NMDA receptors or events downstream to these receptors. Activation of the cAMP-PKA-DARPP32 pathway can also enhance Ca2+ currents (Surmeier et al., 1995). Striatal neurons recorded from slices can exhibit sustained depolarization in response to intracellular current injection via a D1 enhancement of L-type Ca2+ currents (Hernandez-Lopez et al., 1997). It has been recently shown that D1 receptors can enhance synaptically evoked NMDA currents in PFC pyramidal neurons from pre-pubertal rats (Seamans et al., 2001a). In addition, D1 receptors facilitate trafficking of striatal NMDA receptors to the postsynaptic membrane (Dunah and Standaert, 2001). Although there is not evidence linking D1 receptors and trafficking of NMDA receptor subunits in neocortical pyramidal neurons, it is possible that a similar mechanism also plays a role in enhancing NMDA responses in the PFC. Thus, a number of intracellular events can contribute to a D1-mediated enhancement of NMDA function.
The PFC plateau depolarizations reported here are not cortical paroxysmal discharges. Paroxysmal activity can be induced in neocortical brain slices from immature (PD 9–30) animals with bath application of high concentration of the GABAA antagonists bicuculline or picrotoxin (Hablitz, 1987; Sutor et al., 1994). D1 + NMDA-mediated plateau depolarizations, on the other hand, require a mature cortical network. Moreover, paroxysmal cortical activity consists of depolarizing events of much longer duration with action potentials of progressively smaller amplitude (Rubinstein et al., 2001). Thus, persistent activity elicited with D1 and NMDA in mature PFC slices is likely to derive from activation of an assembly of cortical neurons (Beggs and Plenz, 2003), but not from exaggerated hyperexcitability.
It is well known that activation of the mesolimbic/mesocortical dopaminergic system is context-dependent and related to attention and salient stimuli (Horvitz, 2000; Cohen et al., 2002; Schultz, 2002). A D1-sustained plateau depolarization would therefore provide a temporal window during which context-relevant inputs can drive the system (O'Donnell, 2003) and NMDA-dependent synaptic plasticity would be enabled. Indeed, hippocampal-PFC long-term potentiation (LTP) is enhanced by activation of D1, but not D2, receptors (Gurden et al., 2000). The dependence of LTP on an intact mesocortical projection is further evidenced by its absence in animals with DA depletion (Gurden et al., 1999). Thus, cognitive functions such as working memory, learning and attention would be supported by this D1–NMDA interaction. This concept is further supported by D1 and NMDA receptors in the PFC being necessary for appetitive instrumental learning (Baldwin et al., 2002) and by PFC D1 receptors improving memory retrieval (Floresco and Phillips, 2001). It is possible that when mesocortical DA systems become active in response to salient stimuli, the relevant ongoing activity in the PFC (i.e. that mediated by NMDA receptors) becomes enhanced and reinforced by setting and maintaining a population of neurons into the up state via activation of local D1 receptors. Up states therefore are active periods of information processing that contribute to learning and cognitive functions (Cossart et al., 2003; O'Donnell, 2003) and they may depend on potentiation of NMDA function by a number of mechanisms (including, but probably not limited to D1 receptor activation) that ultimately enhance PKA activity and Ca2+ signalling.
The plateau depolarizations elicited by D1–NMDA co-activation were observed only in slices from adult rats in our experimental conditions. It is possible that higher D1 + NMDA concentrations could yield up states in PFC slices from younger animals, because it is well known that both D1 and NMDA receptors change during postnatal development. But even if this were the case, it would still suggest that D1–NMDA interactions impact differently on PFC activity depending on the level of maturation of the system, with prepubertal PFC neurons being less likely to exhibit strong plateaus. Indeed, in vivo intracellular recordings from pre-pubertal animals (PD 28–35) have shown PFC up states of smaller amplitude than similar recordings obtained from adult animals (O'Donnell et al., 2002). Although neocortical neurons in pre-pubertal animals have many morphological and physiological properties of mature cells and most cortical synapses have been formed at pre-pubertal (PD 28–40) ages (Petit et al., 1988; Zhu, 2000), there is also data suggesting that some elements in pyramidal cell physiology mature later. For example, dendritic Na+ and Ca2+ regenerative potentials (which may be important for persistent activity) become effective in coupling distal apical dendrites with somata at PD 42 (Zhu, 2000). Therefore, a post-pubertal appearance of persistent activity in the PFC may derive from a late maturation of intrinsic physiological properties in pyramidal neurons. Other factors that may play a critical role in converting pre-pubertal cortical synaptic organization into a mature network include the delayed acquisition of adult levels of D1 dopamine receptors (Leslie et al., 1991; Tarazi et al., 1999) and NMDA receptor subunits (Williams et al., 1993; Monyer et al., 1994). The emergence of D1 + NMDA-mediated persistent activity after puberty may contribute to developing and acquiring mature cognitive abilities in the PFC.
Supported by MH57683 and a NARSAD Independent Investigator Award (P.O'D.).