Dopamine plays an important role in synaptic plasticity and learning and is involved in the pathogenesis of various neurological and psychiatric disorders. Here, we reveal staining of dopaminergic fibers in stratum oriens of the mouse hippocampal CA1 region, a finding that is consistent with earlier reports. Furthermore, we examined the effect of dopamine agonists on NMDAR-dependent early long-term potentiation (LTP) (40 min) during γ-aminobutyric acid (GABA)A-mediated blockade. LTP of the AMPA component was strongly reduced in stratum oriens but barely affected in stratum radiatum. This layer-specific effect was caused by D4 receptor activation, which augmented the inactivation of synaptic NMDAR-mediated currents (NMDA EPSCs) during LTP induction through a Ca2+-dependent G-protein–independent mechanism. A similar dopaminergic modulation of both NMDA EPSCs and LTP was also observed in mice constitutively lacking NR2A but was absent in mice lacking NR2B in principal forebrain neurons. Together, these experiments strongly indicate that dopaminergic modulation of early LTP in stratum oriens occurs through NMDARs containing NR2B subunits via D4Rs. Thus, a dopamine hyperfunction in stratum oriens may result in NMDAR hypofunction that could affect both normal and pathological conditions.
The dopamine system is thought to play an important role in hippocampus-mediated learning and memory (Lisman and Grace 2005; O'Carroll et al. 2006) and seems also highly relevant to the pathogenesis of several psychiatric disorders including attention-deficit/hyperactivity disorder (ADHD) and schizophrenia (Lewis and Lieberman 2000; Winterer and Weinberger 2004). The hippocampal manifestation of schizophrenia may be characterized by dysfunctions of 2 neurotransmitter pathways: a dopamine (DA) hyperfunction caused by an increased firing rate of dopaminergic neurons localized to the ventral tegmental area (VTA) and an NMDA hypofunction of the glutamatergic system (reviewed in Lisman et al. 2008). The NMDA hypofunction appears to comprise the main features of schizophrenia: positive symptoms like hallucinations, negative symptoms like social withdrawal and cognitive defects including deficits in association and working memory (Greene 2001). The characterization of the functional link between the dopaminergic and the glutamatergic systems in the hippocampus is still incomplete.
The VTA and the substantia nigra supply the hippocampal formation with dopaminergic fibers. These fibers are mainly localized in the subiculum, hilus, and the stratum lacunosum-moleculare (SLM) (Gasbarri et al. 1997). This dopaminergic innervation could modulate glutamatergic synaptic transmission and synaptic plasticity within the hippocampal CA1 region through both D1-like and D2-like receptor families (Tarazi and Baldessarini 2000). Electrophysiological recordings performed in distinct CA1 layers have demonstrated a dopaminergic modulation of the direct cortical pathway to CA1 in SLM with a possible contribution from the inhibitory network (Otmakhova and Lisman 1999; Romo-Parra et al. 2005; Ito and Schuman 2007). At the Schaffer collateral–commissural synapses in stratum radiatum (RAD), the application of DA agonists or antagonists have indicated that D1-like receptors (D1R, D5R) increase the magnitude of early long-term potentiation (LTP) (Otmakhova and Lisman 1996) and stabilize late LTP both in vitro (Frey et al. 1991; Huang and Kandel 1995) and in vivo (Swanson-Park et al. 1999; Lemon and Mnahan-Vaughan 2006). Furthermore, selective genetic inactivation of the D1R suggested that D1Rs but not D5Rs are critical for both LTP at CA1 synapses and spatial learning (Granado et al. 2008; Ortiz et al. 2010). In contrast, the effects of D2-like receptors (D2R, D3R, D4R) are less explored in CA3-to-CA1 synapses and observations are still inconclusive. A D3R agonist has been shown to enhance LTP (Swant and Wagner 2006), whereas D3R mutant mice exhibited normal learning abilities (Xing et al. 2010). Notably, a D4R agonist depotentiated LTP in response to neuregulin-1 and theta-patterned stimuli (Kwon et al. 2008). Most studies have investigated the dopaminergic modulation of synaptic transmission and synaptic plasticity in CA1 with intact inhibition, which precluded an attribution of dopamine-mediated effects to either glutamatergic or γ-aminobutyric acid (GABA)ergic synapses. Moreover, very few studies have examined the dopaminergic modulation in stratum oriens (OR), in spite of the anatomical evidence for a stronger dopaminergic innervation of OR than RAD (Verney et al. 1985; Gasbarri et al. 1994; Kwon et al. 2008).
Here, we compared dopaminergic modulation of NMDAR-dependent early LTP in OR and RAD in acute hippocampal slices using D1R/D5R and D4R agonists. We blocked GABAARs to avoid dopaminergic effects on inhibition and examined the AMPA component of glutamatergic transmission. Even though any NMDAR subtype constellation is capable of inducing NMDAR-dependent LTP in hippocampal RAD synapses (Berberich et al. 2005), dopaminergic modulation may lead to a preferential involvement of certain NMDAR subtypes. NMDARs containing NR2A have been proposed to interact with D1R/D5R (Lee et al. 2002; Varela et al. 2009), whereas NMDARs containing NR2B seem to interact with D2Rs (Liu et al. 2006). Therefore, we explored NMDAR subtype dependency for both NMDA and AMPA components of excitatory postsynaptic currents (EPSCs) in constitutive NR2A knockout (NR2A−/−) and forebrain-specific NR2B knockout mice (NR2BΔFb) in comparison with wild-type mice. We found that during LTP induction DA agonists activating D4Rs reduced the activity of NR2B-containing NMDARs in OR, causing input-specific modulation of LTP.
Materials and Methods
Experimental procedures were in accordance with the animal welfare guidelines of the Max Planck Society, the “Norwegian Animal Welfare Act” and the “European Union’s Directive 86/609/EEC.”
Z/EG mice (Novak et al. 2000) were crossbred with transgenic DATCre mice (Parlato et al. 2006) to generate Z/EGDATcre reporter mice. Four-week-old Z/EGDATcre mice were sacrificed with CO2, perfused through heart with 4% paraformaldehyde/phosphate buffered saline (PFA/PBS) and brains were isolated and postfixed with 4% PFA/PBS at 4 °C overnight. Vibratome sections (50 μm thick) were stained for green fluorescent protein (GFP) with anti-GFP antibody (1:500, Invitrogen), biotinylated secondary anti-Rabbit antibody (Vector Laboratories), amplified by VECTASTAIN ABC kit (Vector Laboratories) and developed with 3,3′-diaminobenzidine (Sigma). Sections were mounted with Eukitt (Fluka). Images were taken on an Axioplan microscope (Zeiss) and processed with Adobe Creative Suite.
Extracellular Field Experiments
Wild-type (C57BL/6) mice (P27–P30) were killed with isoflurane. Transverse slices (400 μm) from the dorsal half of each hippocampus were cut with a vibroslicer in artificial cerebrospinal fluid (ACSF) containing (in mM): 123 NaCl, 4 KCl, 4 MgCl2, 1.24 NaH2PO4, 4 CaCl2, 26 NaHCO3, and 12 D-glucose; 4 °C, pH 7.4 with 95% O2/5% CO2. Slices were placed in an interface chamber at 28–32 °C and perfused with ACSF containing 10 μM bicuculline methiodide (BMI) to block GABAA receptors. Orthodromic synaptic stimulation at 0.1 Hz was delivered alternately through 2 tungsten electrodes placed in the CA1 region, 1 in stratum radiatum (RAD), about 200 μm from pyramidal cell layer, and the other one in stratum oriens (OR), close to the alveus. Extracellular field excitatory postsynaptic potentials (fEPSPs) were monitored in the corresponding layers by 2 glass electrodes filled with ACSF and positioned 500–600 μm away from the stimulation electrodes. Synaptic efficacy and tetanization procedures were assessed as described earlier (Köhr et al. 2003). Six consecutive responses (1 min) were averaged and normalized to the mean value recorded 4–7 min before high-frequency stimulation (HFS: 100 Hz, 1 s). Data sets were pooled and presented as mean ± standard error of the mean (SEM). n indicates the number of slices. The subsequent number in brackets indicates the number of mice. Statistical evaluation was performed by using a paired or unpaired 2-tailed Student's t-test.
The brains were removed from deeply anesthetized (isoflurane) P27–P30 wild-type (C57BL/6), constitutive NR2A−/− (Sakimura et al. 1995) and forebrain-specific NR2B knockout mice (NR2BΔFb) (von Engelhardt et al. 2008). Hippocampal coronal slices (250 μm) were prepared, allowed to recover at 35 °C for 1 h, and thereafter maintained at room temperature (23–25 °C) in ACSF containing (in mM): 125 NaCl, 25 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 2 CaCl2, 1 MgCl2, and 25 D-glucose; pH 7.4 with 95% O2/5% CO2 (320 mOsm). A slice was then transferred to a submerged recording chamber and continuously perfused with ACSF (1.5–2 mL/min) containing 10 μM BMI. Whole-cell recordings were performed at room temperature, using patch pipettes pulled from 0.5 mm walled borosilicate glass capillaries (Hilgenberg, Germany). Patch pipettes had resistances of 4–6.5 MΩ when filled with (in mM): 125 Cs-gluconate, 20 CsCl, 10 NaCl, 10 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 0.2 ethyleneglycol-bis(2-aminoethylether)-N,N,N ′,N ′-tetra acetic acid (EGTA), 4 MgATP, 0.3 Na3GTP, 2.5 QX-314Cl (pH 7.3, 270–320 mOsm). In some experiments, EGTA was increased to 10 mM. Liquid junction potentials were not corrected. Series and input resistances were monitored continuously by measuring peak and steady-state currents in response to hyperpolarizing pulses (−5 mV, 20 ms). The Schaffer collateral–commissural fibers were stimulated at 0.2 Hz with 2 glass electrodes filled with 1 M NaCl placed in RAD or in OR, 150 μm lateral from the recorded CA1 cell, and 150 and 50 μm away from the pyramidal cell layer, respectively. EPSCs were recorded at −70 mV. Six consecutive EPSCs were averaged and normalized to the responses obtained 5 min before washing in a DA agonist or before LTP induction and compared with responses obtained about 30 min later. For LTP recordings, the patch pipettes were filled with (in mM): 120 Cs-gluconate, 10 CsCl, 8 NaCl, 10 HEPES, 0.2 EGTA, 2 MgATP, 0.3 Na3GTP, 10 phosphocreatine (pH 7.3, 270–320 mOsm), and the ACSF contained (in mM): 124 NaCl, 26 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 4 CaCl2, 4 MgSO4, 10 D-glucose, and the NMDAR coagonist glycine (10 μM); pH 7.4 with 95% O2/5% CO2 (320 mOsm). LTP was induced by pairing postsynaptic depolarization (0 mV membrane potential for 3 min) with low-frequency synaptic stimulation (120 pulses, 0.67 Hz) of either the RAD or OR pathways (LFS pairing) (Chen et al. 1999), unpaired pathway served as a control. The average charge transfer per EPSC evoked during LFS pairing was determined by analyzing the 120 EPSCs individually. NMDAR-mediated EPSCs (NMDA EPSCs) were recorded in the presence of the AMPAR blocker NBQX (5 μM) at a membrane potential of +20 mV. NMDA EPSCs were elicited at 0.05 Hz, and if stable, they were normalized to 1 before applying the LFS pairing protocol at 0.67 Hz. The resulting NMDA EPSC inactivation was estimated either by averaging the peak amplitude or the charge transfer of 10 consecutive NMDA EPSCs during LFS pairing. To compare the decay times for NMDA EPSCs across different conditions, a weighted time constant (τweighted) was calculated using the formula: (Ifast/(Ifast + Islow)) × τfast + (Islow/(Islow + Ifast)) × τslow, where I is the amplitude of the fast or slow component. Data were pooled across animals of the same genotype and are presented as mean ± SEM. In addition to the number of cells (n), the number of slices (x) and mice (y) are presented in brackets (x, y). Statistical significance was evaluated by a 2-tailed Student's t-test (paired for washing in experiments and unpaired for all other experiments) or ANOVA with Fisher's post hoc analysis when comparing more than 2 pharmacological conditions or more than 2 genotypes at the same time.
NBQX and R-(+)-SCH23390 (both Biotrend), BMI and glycine (both Sigma), and SKF38393 and L-745,870 (both Tocris) were dissolved in water to stock solutions and further dissolved in the ACSF on the experimental day. PD168077 and S-(−)-sulpiride (both Biotrend), R(+)-6-bromo-APB (b-APB, Sigma), and clozapine (Tocris) were dissolved in dimethyl sulfoxide (DMSO) to a final DMSO concentration not exceeding 0.1%. The drugs were applied at least 15 min before whole-cell recordings or during washing in. G-proteins were blocked by adding GDPßS (1 mM, Sigma) to the aliquots of the electrode solutions, while holding pH and osmolarity constant. GDPßS was allowed to diffuse into the cell for at least 25 min prior to baseline recordings.
Distribution of Dopamine Active Transporter within Hippocampal Area CA1
Neuronal tracer techniques combined with immunocytochemistry have revealed a differential dopaminergic innervation of the hippocampal formation in rats (Gasbarri et al. 1994; Gasbarri et al. 1997). Therefore, the rat dopamine active transporter (DAT), which is enriched in dopaminergic terminals (Cragg and Rice 2004) could also be differentially distributed. To visualize DAT-enriched fibers in mice, we employed a reporter mouse (Z/EGDATcre) which expresses GFP in cells expressing DAT. Thus, via projections of dopaminergic neurons residing in the pars compacta of substantia nigra, the dorsal striatum was intensively stained in sections from 6 Z/EGDATcre mice (Fig. 1A). Staining was more pronounced in stratum oriens (OR) than in stratum radiatum (RAD) within the dorsal hippocampal CA1 region. Higher resolution images showed DAT-stained fibers passing along the alveus in sagittal sections and ending radially in coronal sections in OR (Fig. 1A; see also dopaminergic innervation in OR obtained by immunofluorescence in Kwon et al. 2008). Although we could not observe DAT-stained fibers in RAD, we cannot rule out that dopaminergic input is also conveyed via fibers in RAD.
In order to unmask whether the skewed distribution of dopaminergic fibers modulated the glutamatergic transmission in the mouse hippocampal slice preparation, we evoked EPSCs by LFS (0.2 Hz) of the RAD and the OR pathways. We blocked the GABAA-mediated inhibition by bicuculline (BMI, 10 μM) to avoid effects of dopamine agonists on inhibitory transmission. Forty minutes of washing in the D1R/D5R agonist SKF38393 (50 μM) failed to affect EPSCs elicited in the 2 pathways (OR: 0.93 ± 0.16, n = 6, P = 0.64; RAD: 0.96 ± 0.09, n = 6 (6, 6), P = 0.67) (Fig. 1B). The present results extend earlier reports showing that the AMPA component of EPSCs in RAD was not modulated by DA agonists during single-pulse stimulation (Otmakhova and Lisman 1996; Mockett et al. 2004).
In wild-type mice, HFS (100 Hz, 1 s) of either the RAD or the OR pathway in the presence of BMI (10 μM) produced a persisting homosynaptic potentiation of extracellular fEPSPs 40–45 min after the tetanization, characteristic of early LTP of synaptic transmission. The magnitude of LTP was comparable in the 2 CA1 pathways (OR: 1.26 ± 0.04, n = 16 (9); RAD: 1.39 ± 0.06, n = 21 (9), P = 0.09) (Fig. 2A,B), whereas the untetanized control pathway remained unchanged. This indicates that in our slice preparation branches from CA3 axons, targeting both OR and RAD, hardly contribute to evoked responses. In presence of 50 μM SKF38393, the tetanization elicited a less pronounced posttetanic potentiation/short-term potentiation phase (P < 0.05 for the initial 8 min in OR and the initial 3 min in RAD after tetanization) (Fig. 2A,B). In contrast, 40–45 min after the tetanization LTP was strongly reduced in OR (1.07 ± 0.05, n = 14 (8), P = 0.008) but only marginally in RAD (1.21 ± 0.06, n = 15 (9), P = 0.04) (Fig. 2A,B). These results contrast with very recent observations, which showed enhanced field LTP both in OR and in RAD when using 20 μM SKF38393 in rat hippocampal slices with intact GABAA receptor mediated inhibition (Stramiello and Wagner 2008, 2010). Therefore, our results in presence of BMI unmask a pathway-specific dopaminergic modulation on glutamatergic transmission.
LFS Pairing at Membrane Potentials of 0 and +20 mV
Next, we used whole-cell recordings in the presence of BMI (10 μM) to examine in the 2 dendritic CA1 subfields the effect of SKF38393 on AMPAR- as well as NMDAR-mediated currents. Following a period with stable synaptic responses (EPSCs) elicited at 0.2 Hz, LFS at 0.67 Hz was paired with postsynaptic depolarization of 0 mV membrane potential for 3 min (Chen et al. 1999). In presence of SKF38393 (50 μM), the amount of LTP obtained 30 min after LFS pairing was similar to LTP obtained in the control solution in the RAD pathway (control solution: 2.52 ± 0.17, n = 6 (4, 3); with SKF: 2.35 ± 0.38, n = 8 (7, 6); P = 0.68), However, in the OR pathway, LTP was significantly reduced (control solution: 1.99 ± 0.16, n = 9 (7, 7); with SKF: 1.34 ± 0.14, n = 7 (4, 4) P < 0.05) (Fig. 3A,B). Since GABAA receptors were blocked, the results indicate a direct modulatory effect of SKF38393 on excitatory synaptic transmission in OR. In addition, analysis of the EPSC charge transfer during the LFS pairing paradigm at 0 mV revealed a selective statistically significant decreased charge transfer in presence of SKF38393 in the OR pathway (OR, control solution: −2.73 ± 0.39 pC, n = 9 (7, 7); OR with SKF: −0.89 ± 0.16 pC, n = 7 (4, 4), P < 0.01; RAD, control solution: −4.44 ± 0.5 pC, n = 6 (4, 3); RAD with SKF: −3.62 ± 1.15 pC, n = 8 (7, 6), P = 0.53) (Fig. 3C). Thus, the reduced LTP in OR in the presence of SKF38393 could be due to a reduced Ca2+ influx through NMDARs during the LFS pairing paradigm.
To further elucidate whether NMDARs are modulated by D1R/D5R agonists during LTP induction, we recorded at +20 mV membrane potential synaptic NMDAR-mediated currents (NMDA EPSCs) while blocking AMPARs (5 μM NBQX). Following baseline responses elicited at 0.05 Hz, we applied the LFS pairing protocol (Fig. 4). During the LFS pairing protocol, peak amplitudes and charge transfers of NMDA EPSCs gradually decreased in both OR and RAD (Fig. 4B,C). These decreases in NMDAR-mediated responses were significantly and selectively accentuated in the presence of SKF38393 in the OR pathway. On average, the NMDAR-mediated charge transfer, which is correlated to whether LTP develops (Berberich et al. 2007), was reduced by ∼35% in OR (OR relative to baseline with control solution: 0.73 ± 0.03, n = 22 (18, 14); OR with SKF: 0.46 ± 0.04, n = 7 (5, 5), P < 0.0001), whereas no effect was observed at RAD synapses (RAD relative to baseline with control solution: 0.59 ± 0.03, n = 6 (6, 6); RAD with SKF: 0.63 ± 0.06, n = 7 (2, 2), P = 0.44) (Fig. 4D). Therefore, the D1R/D5R agonist SKF38393 seems to interfere with LTP induction in the OR pathway through an augmented inactivation of NMDA EPSCs.
We also found a similar pathway-specific dopaminergic modulation of the NMDA EPSC charge transfer in presence of R(+)-6-bromo-APB (b-APB, 50 μM), another D1R/D5R agonist (Otmakhova and Lisman 1996; Navakkode et al. 2007) (OR, control solution: 0.73 ± 0.03, n = 22 (18, 14); OR with b-APB: 0.56 ± 0.04, n = 7 (4, 3), P < 0.01; RAD, control solution: 0.59 ± 0.05, n = 6 (6, 6); RAD with b-APB: 0.64 ± 0.06, n = 9 (6, 3), P = 0.43) (Fig. 4E).
The Involvement of NR2A and NR2B Subunits
D1Rs have been reported to interact directly with NR1 and NR2A subunits in heterologous cells and cultured hippocampal neurons (Lee et al. 2002), whereas D2Rs are known to interact directly with NR2B subunits in striatal neurons (Liu et al. 2006). Based on our observed pathway-specific dopaminergic modulation of NMDAR activity during the LFS pairing paradigm and the subsequent reduction in the amount of LTP, we examined NR1/NR2B receptors in NR2A−/− mice and NR1/NR2A in NR2BΔFb mice to determine if either NR2A- or NR2B-containing NMDARs might be a preferential target for DA agonist modulation. In the hippocampus of adult (>P60) NR2BΔFb mice, NR2B protein expression is absent in most forebrain neurons, and in these neurons, the decay time of NMDA EPSCs is significantly faster than in wild type (von Engelhardt et al. 2008; Rauner and Köhr 2011). In slices from NR2BΔFb mice at P27–P29, we restricted our analysis to cells that had NMDA EPSCs with fast decay.
Analogous to wild-type mice, the NMDA EPSC charge transfer was examined in the mutants at +20 mV membrane potential during the LFS pairing paradigm in absence and presence of SKF38393 (Fig. 5). In slices from NR2A−/− mice (Fig. 5A), as observed in wild type (Fig. 4C), SKF38393 significantly reduced the NMDA EPSC charge transfer in OR but not in RAD (OR, control solution: 0.73 ± 0.06, n = 7 (4, 3); OR with SKF: 0.48 ± 0.04, n = 8 (6, 4), P < 0.05; RAD; control solution: 0.68 ± 0.04, n = 7 (6, 2); RAD with SKF: 0.67 ± 0.07, n = 7 (6, 3), P = 0.96). In slices from NR2BΔFb mice, however, SKF38393 failed to affect the inactivation of NMDA EPSCs in both layers (OR, control solution: 0.51 ± 0.11, n = 4 (3, 2); OR with SKF: 0.4 ± 0.05, n = 4 (4, 3), P = 0.51; RAD, control solution: 0.66 ± 0.04, n = 5 (3, 2); RAD with SKF: 0.56 ± 0.05, n = 6 (3, 2), P = 0.21) (Fig. 5B). These results suggest that the applied DA agonist selectively modulated NMDA EPSC charge transfer in OR via NMDARs containing NR2B subunits (Fig. 5C).
LFS Pairing at 0 mV in NR2BΔFb and NR2A−/− Mice
In accordance with von Engelhardt et al. (2008), LTP was significantly reduced in the RAD pathway of NR2BΔFb mice following the LFS pairing paradigm (RAD: 1.54 ± 0.1, n = 9 (5, 5), P < 0.01) (Fig. 6A,B). We also found a similar reduction of LTP in the OR pathway (OR: 1.33 ± 0.15, n = 8 (6, 4), P < 0.05) (Fig. 6A,B). As expected from the analysis of the charge transfer mediated by NMDA EPSCs, which was not affected by SKF38393 (Fig. 5B), the presence of SKF38393 failed to modulate the magnitude of LTP both in RAD and in OR (RAD: 1.49 ± 0.15, n = 9 (6, 5), P = 0.82; OR: 1.38 ± 0.14, n = 8 (4, 3), P = 0.83). Analysis of the charge transfer during LFS pairing at 0 mV membrane potential substantiated this finding (Fig. 6C) because SKF38393 failed to alter the EPSC charge in both pathways (RAD, control solution: −5.36 ± 1.71 pC, n = 9 (5, 5); RAD with SKF: −4.91 ± 1.7 pC, n = 9 (6, 5), P = 0.86; OR, control solution: −1.28 ± 0.27 pC, n = 8 (6, 4), OR with SKF: −0.83 ± 0.09 pC, n = 8 (4, 3), P = 0.20). It is therefore unlikely that NR1/NR2A receptors are responsible for the SKF38393-mediated reduction of LTP in OR in wild-type mice. Therefore, the dopaminergic modulation of LTP in OR in wild-type mice might possibly be ascribed to NR2B-containing NMDARs.
To support this conclusion, we employed NR2A−/− mice (Sakimura et al. 1995) and induced LTP in the 2 pathways by LFS pairing (Fig. 7). LTP in OR was comparable to that obtained in wild-type mice in OR (1.94 ± 0.21, n = 8 (5, 4), P < 0.83), whereas LTP in RAD was significantly reduced (1.78 ± 0.09, n = 8 (6, 4), P < 0.01). In the presence of SKF38393, LTP in OR was reduced (1.38 ± 0.14, n = 10 (5, 3), P < 0.05), whereas SKF failed to modulate LTP in RAD (1.62 ± 0.09, n = 10 (7, 5), P = 0.23). Analysis of the charge transfer at 0 mV during the LFS pairing supported this finding (Fig. 7C), revealing a significant charge reduction in OR but not in RAD (OR, control solution: −4.85 ± 0.29 pC, n = 8 (5, 4); OR with SKF: −0.85 ± 0.13 pC, n = 10 (5, 3), P < 0.0001; RAD; control solution: −4.01 ± 0.62 pC, n = 8 (6, 4); RAD with SKF: −3.44 ± 0.43 pC, n = 10 (7, 5), P = 0.48). These results indicate that the dopaminergic modulation of LTP induction in OR is closely linked to NR2B-containing NMDARs.
Dopaminergic Modulation of NMDAR Activity in Stratum Oriens is Ca2+ dependent and G-protein independent
Changes in intracellular Ca2+ through NMDARs are necessary for long-term changes in synaptic AMPAR function (reviewed in Malenka and Bear 2004). Intracellular Ca2+ is also known to rise following the activation of DA receptors (Hasbi et al. 2010) and may contribute to the Ca2+-dependent inactivation of NMDARs (Zhang et al. 1998; Krupp et al. 1999; Kotecha et al. 2002). To test Ca2+ dependencies regarding the dopaminergic modulation of NMDA EPSCs in OR, we used 10 mM EGTA in the patch pipette (Fig. 8A). In the absence of SKF, the charge transfer of NMDA EPSCs evoked at +20 mV in OR was reduced to a similar degree in 0.2 and 10 mM EGTA (0.2 mM, 0.73 ± 0.03, n = 22 (18, 14), Fig. 4C; 10 mM, 0.69 ± 0.07, n = 7 (6, 4), P = 0.65; Fig. 8A). In contrast, the SKF38393-augmented inactivation of NMDA EPSCs in OR was no longer observed in the presence of 10 mM EGTA (with SKF: 0.61 ± 0.1, n = 7 (6, 4), P = 0.51; Fig. 8A). This indicates a Ca2+ dependency of the observed dopaminergic NMDA EPSCs modulation in OR.
Different intracellular signaling systems have been linked to DA receptors. Receptor coupling to Gs- and Gi-proteins regulate the adenylate cyclase catalysis of cAMP formation, whereas coupling to Gq-proteins works through the PLC/IP3 pathway (Hasbi et al. 2010). We inhibited G-proteins by allowing the stable GDP analog guanosine 5′-O-2-thiodiphosphate (GDPßS, 1 mM) to diffuse into the CA1 neuron via the patch pipette 25 min prior to baseline recordings. The presence of GDPßS did not affect the reduction of NMDA EPSC charge transfer in OR, neither in absence nor in presence of SKF38393 (control solution, P = 0.29; with SKF, P = 0.14). Consequently, the SKF effect remained significant in OR (control solution, 0.68 ± 0.04, n = 9 (2, 1), with SKF, 0.50 ± 0.03, n = 9 (4, 4), P < 0.001; Fig. 8B and for absence of GDPßS, see Fig. 4C). Thus, DA receptor–mediated activation of G-proteins does not appear to contribute to the modulation of NMDA EPSCs in OR.
Effects of Dopamine Receptor Agonists and Antagonists in Stratum Oriens
To elucidate whether the modulation of NMDA EPSCs in OR by SKF38393 and b-APB during the LFS pairing protocol at +20 mV (see Fig. 4) were mediated via D1-like receptors, we examined the NMDA EPSC charge transfer in the presence of both the SKF38393 (50 μM) and the D1R/D5R antagonist SCH23390 (1 or 10 μM). Surprisingly, the SKF38393-induced augmentation of NMDA EPSC inactivation in OR was not antagonized by SCH23390 (1 μM, 0.38 ± 0.05, n = 4 (1, 1), P < 0.01; 10 μM, 0.45 ± 0.04, n = 11 (5, 5), P < 0.01) (Fig. 8C). The D2R antagonist sulpiride behaved in a similar way (15 μM, 0.33 ± 0.04, n = 4 (2, 1), P < 0.01) (Fig. 8C). The reported existence of D1/D2 heterooligomers (Hasbi et al. 2010) led us coperfuse SCH23390 (1 μM) and sulpiride (10 μM), which has a 100-fold higher affinity to D2Rs and D3Rs than to D4Rs (Seeman and Van Tol 1994). Although the SKF38393-induced reduction in NMDA EPSC charge transfer was synergistically diminished (P < 0.05), SCH23390 and sulpiride together failed to fully antagonize the SKF38393-mediated modulation (0.56 ± 0.07, n = 5 (2, 1), P < 0.05) (Fig. 8C). These experiments indicated that D1R/D5Rs and D2R/D3Rs with high affinity to sulpiride are not involved in the modulation of NMDA EPSCs in stratum oriens.
Next, we considered the contribution of D4Rs, which negatively regulate NMDAR transmission and LTP in rat hippocampal slices (Kotecha et al. 2002; Stramiello and Wagner 2008). First, we tested clozapine (20 μM), an atypical neuroleptic with a relatively high affinity for D4Rs compared with other DA receptors. This compound blocked the SKF-induced decrease in charge transfer (0.66 ± 0.09, n = 6 (5, 3), P = 0.32) (Fig. 8C). Because clozapine also has affinity to alpha-adrenergic and serotonergic receptors (reviewed in Meltzer 1995; Svensson 2003), we subsequently applied the potent and selective D4R antagonist L-745,870 (100 nM) which also fully blocked the SKF-induced decrease in charge transfer (0.67 ± 0.07, n = 5 (2, 2), P = 0.92; Fig. 8C). These experiments indicated that SKF38393 at 50 μM is nonselective, since SKF38393 not only binds to D1Rs but also to D4Rs, which in turn modulated NMDA EPSCs in OR.
To explore whether a selective activation of D4Rs was sufficient to reduce the NMDA EPSC charge transfer, we applied the LFS pairing protocol at +20 mV in the presence of the selective D4R agonist PD168077 (100 nM). PD168077 augmented the reduction in the OR but not in the RAD pathway (OR, control solution: 0.73 ± 0.03, n = 22 (18, 14); OR with PD168077: 0.53 ± 0.04, n = 8 (5, 3), P < 0.01; RAD, control solution: 0.72 ± 0.08, n = 4 (2, 2); RAD with PD168077: 0.82 ± 0.05, n = 7 (2, 2), P = 0.25) (Fig. 8D). This input-specific effect by PD168077 was completely blocked by the selective D4R antagonist L-745,870 (50 nM; 0.68 ± 0.05, n = 6 (2, 2), P = 0.81) (Fig. 8D). Thus, the dopaminergic modulation of NMDA EPSCs in OR appears to be exclusively mediated via D4Rs.
In a last set of experiments, we tested the possibility that the D4R-mediated modulation of NMDAR function affects LTP in wild-type mice. First, we evoked stable EPSCs at −70 mV at 0.2 Hz before washing in PD168077 at 100 nM, which did not affect the peak amplitude of EPSCs in wild-type mice (OR: 0.95 ± 0.06, n = 11 (11, 4), P = 0.59; RAD: 1.09 ± 0.1, n = 11 (11, 4), P = 0.42) (Fig. 8E). In separate experiments and in continuous presence of 100 nM PD168077, we applied the LFS pairing protocol in OR, which significantly reduced LTP compared with control experiments lacking the D4R antagonist (control solution: 1.99 ± 0.16, n = 9 (7, 7); with PD168077: 1.23 ± 0.08, n = 9 (7, 5), P < 0.01). Thus, PD168077 reflects all functional aspects that we had observed with SKF38393 at 50 μM (Figs 1B, 3B, and 4D), indicating that the suppression of LTP in OR by PD168077 or SKF38393 was caused through modulation of NMDAR function via D4Rs.
Here, we compared the effects of synthetic DA receptor ligands in stratum oriens (OR) and stratum radiatum (RAD) in the CA1 region of the dorsal hippocampus of 4-week-old mice. D4R activation selectively reduced early LTP in OR by reducing NMDAR-mediated charge transfer via an NR2B Ca2+-dependent G-protein–independent mechanism.
Dopaminergic Input to the Hippocampal CA1 Region Differentially Modulates LTP
We observed intensively stained DAT-positive fibers in OR, a finding that is supported by the identification of dopaminergic fibers in rat and mouse hippocampus (Gasbarri et al. 1994; Gasbarri et al. 1997; Kwon et al. 2008). These findings may be consistent with a more pronounced dopaminergic innervation in OR than in RAD, which requires further investigation in the future. In the presence of the D1R/D5R agonist SKF38393 at a high (μM) concentration or in the presence of the D4R agonist PD168077 at a low (nM) concentration, we observed a pathway-specific reduction in the amount of LTP in OR. In field recordings, the reduction of the AMPA component developed gradually following the induction by HFS, whereas in whole-cell configuration, the synaptic efficacy was immediately reduced following LFS pairing (low-frequency synaptic stimulation paired with postsynaptic depolarization). This effect was in contrast to the lack of modulatory effects by SKF38393 in RAD.
Similar to the lateral amygdala (Bissiere et al. 2003), the dopaminergic gating of LTP in the temporoammonic (TA)–CA1 pathway in SLM has been ascribed to a diminished feed-forward inhibition (Ito and Schuman 2007), thus facilitating the LTP induction (Wigstrom and Gustafsson 1983). The reduced LTP we observed in OR in presence of SKF38393 was obtained while GABAA receptors were blocked with bicuculline, which suggests a different gating mechanism than the one mediated through inhibitory interneurons. Distinct mechanisms underlying the dopaminergic modulation of LTP induction and expression in SLM versus RAD/OR are also indicated by the fact that SKF38393 and dopamine itself reduced low-frequency activated synaptic responses in SLM (Otmakhova and Lisman 1999; Ito and Schuman 2007) but neither in RAD (Otmakhova and Lisman 1996; Mockett et al. 2004) nor in OR (our results).
Mechanism for Negative Dopaminergic LTP Modulation in OR
The major DA receptor signaling cascades are G-protein linked and activate or inhibit adenylate cyclase, thereby affecting cAMP production, the activity of PKA, cAMP-regulated DARPP-32, and voltage-gated Ca2+ channels (Seamans and Yang 2004). Alternative signaling cascades may involve interactions between D2-like DA receptors and the platelet-derived growth factor receptor beta (PDGFRbeta) (Kotecha et al. 2002) or direct protein–protein interactions between the intracellular C-termini of DA receptors and NMDARs. Interactions exist either between D1 and NR1 or NR2A (Lee et al. 2002) or between D2 and NR2B (Liu et al. 2006). These interactions most likely take place in postsynaptic membranes, where PDGFRbeta is expressed (Beazely et al. 2009) and DA receptors and NMDARs are known to colocalize (Hara and Pickel 2005). Thus, postsynaptic membranes provide the basis for functional interactions between dopaminergic and glutamatergic systems.
Our pharmacological analysis of NMDA EPSCs indicated that the augmented inactivation of NMDARs in OR is mediated through D4Rs, since the D4R antagonist L-745,870 prevented the SKF38393-induced NMDAR inactivation—a finding also supported by the experiments using clozapine. This possibility was strengthened by our demonstration that a D4R agonist (PD168077) also induced the pathway-specific NMDAR modulation in OR, an effect which was antagonized by the D4R antagonist L-745,870. Therefore, the D1R/D5R agonist SKF38393 at 50 μM is not D1 selective. Additionally, SKF38393 activated D4Rs, which modulated NMDA EPSCs in OR, whereas the binding of SKF38393 to D1Rs did not affect the inactivation of NMDA EPSCs for 2 reasons. 1) The rather selective D1R/D5R antagonist SCH23390 did not prevent the SKF38393 (50 μM)-induced NMDAR inactivation. 2) Lower concentrations of SKF38393 (10 and 20 μM) were insufficient to reduce the NMDA EPSC charge transfer in OR (data not shown). In contrast, a contribution of D1/D2 heterooligomers cannot be excluded (Hasbi et al. 2010) because the SKF38393-induced NMDAR inactivation was reduced by simultaneously antagonizing D1R/D5Rs (SCH23390) and D2-like DA receptors (sulpiride).
We employed NMDAR mutant mice in which either the NR2A or the NR2B alleles are inactive and demonstrated that NR2B- but not NR2A-containing NMDARs allowed the dopaminergic modulation in OR, that is, the reduced NMDAR-mediated postsynaptic charge transfer during LFS pairing and the consequent reduction in LTP magnitude. Because NR2A expression is substantial in 4 week-old wild-type mice, DA modulation in hippocampal synapses may embrace triheteromeric NR1/NR2A/NR2B receptors (Köhr 2006). These NMDAR subtypes comprise a notable fraction of the total NMDAR population based on biochemical analyses of whole hippocampus (Al-Hallaq et al. 2007) and on electrophysiological analyses of hippocampal CA1 synapses (Rauner and Köhr 2011). Our experiments, in which we applied a Ca2+ chelator or a G-protein inhibitor via the patch pipette intracellularly, indicated that the dopaminergic modulation of NMDA EPSCs in OR involves a postsynaptic mechanism that is Ca2+ dependent but G-protein independent. PDGFRbeta signaling may be one potential postsynaptic mechanism. Activation of PDGFRbeta in hippocampal slices by the D2-like agonist quinpirole, known as transactivation, was reported to enhance calmodulin- and Ca2+-dependent inactivation of NMDARs (Kotecha et al. 2002). Interestingly, PDGFRbeta activation by growth factor preferentially inhibited NR2B-containing NMDARs in isolated CA1 neurons (Beazely et al. 2009). Since quinpirole did not augment the NMDAR inactivation in OR (data not shown), we favor the possibility that a direct protein–protein interaction between D2R and NR2B subunits might underlie the dopaminergic modulation of LTP in OR. Direct D2R–NR2B interaction has been reported to be enhanced in response to cocaine, disrupting the association between CaMKII and NR2B, and reducing NMDA currents in dissociated striatal neurons (Liu et al. 2006). Consistently, transgenic mice with a disrupted CaMKII–NR2B interaction are impaired in hippocampal LTP and spatial learning (Zhou et al. 2007). A similar CaMKII–NR2B disruption may occur during the dopaminergic modulation of LTP that we describe in OR, which will be of interest for future studies.
Dopaminergic Modulation of Hippocampal Learning and Memory
The DA contribution to hippocampus-dependent learning and memory is incompletely understood. Recent studies have focused on the D1R/D5R-mediated modulation of hippocampus-dependent learning (Tran et al. 2008; Bethus et al. 2010; Nai et al. 2010; Ortiz et al. 2010; Xing et al. 2010). Our results add new dimensions, including D4R contribution in stratum oriens. Beside OR, SLM could be another distinct CA1 subregion for dopaminergic modulation of learning and memory, as previously implied in plasticity studies (Otmakhova and Lisman 1999; Ito and Schuman 2007). Thus, the future challenge will be to understand the function of the differential DA modulation in OR and SLM during hippocampal learning and memory. These hippocampal considerations will also be interesting because of the recently reported dopaminergic modulation of synapses at the cortical input and the subicular output (Roggenhofer et al. 2010; Mu et al. 2011).
Stratum Oriens as Functional Link in Cognitive Disorders?
Both dopaminergic hyperfunction in VTA and NMDAR hypofunction in the hippocampus have been implicated in the pathophysiology of schizophrenia (reviewed in Lisman et al. 2008). Whether one neurotransmitter system causes the alterations in the other or whether they coexist without causal relation is unknown. Aberrant hippocampal activity can lead to abnormal firing of DA neurons in the VTA (Floresco et al. 2003; Lodge and Grace 2007). However, local effects may play a role, as, for example, stimulation of dopamine release within the hippocampus by neuregulin-1 is sufficient to suppress LTP without VTA involvement (Kwon et al. 2008).
Here, we suggest that stratum oriens could be an overlooked link between dopaminergic and glutamatergic systems underlying cognition. Malfunctions in either system or in their interactions may contribute to the deficits observed in a variety of mental disorders (Heckers et al. 1998). For instance, DA neurons in the VTA are highly active in schizophrenia, leading to increased DA release in target regions. In OR, the subsequent activation of DA receptors on basal dendrites of CA1 neurons inactivates NR2B-containing NMDARs in a Ca2+-dependent manner, impairing LTP as a consequence. This mechanism links DA hyperfunction in VTA and NMDAR hypofunction in the hippocampus. Thus, the severity of symptoms of mental disorders may depend on dysfunctions in stratum oriens of the dopaminergic and glutamatergic neurotransmitter pathways, and consideration could be beneficial therapeutically.
Letten Foundation (to V.J. and O.H.).
We thank Dr M. Mishina (University of Tokyo) for providing NR2A−/− mice, Dr H. Monyer (University of Heidelberg) for NR2BΔFb mice, Dr S. Berberich and A. K. Lee for valuable comments, and Dr P. H. Seeburg for generous support. Conflict of Interest : None declared.