Regulation of Dopamine D 1 Receptor Trafficking and Desensitization by Oligomerization with Glutamate N -Methyl- D -aspartate Receptors*

Activation of dopamine D 1 receptors is critical for the generation of glutamate-induced long-term potentiation at corticostriatal synapses. In this study, we report that, in striatal neurons, D 1 receptors are co-localized with N -methyl- D -aspartate (NMDA) receptors in the postsynaptic density and that they co-immunoprecipitate with NMDA receptor subunits from postsynaptic density preparations. Using modified bioluminescence resonance energy transfer, we demonstrate that D 1 and NMDA receptor clustering reflects the existence of direct interactions. The tagged D 1 receptor and NR1 sub- unit cotransfected in COS-7 cells generated a significant bioluminescence resonance energy transfer signal that was insensitive to agonist stimulation and that did not change in the presence of the NR2B subunit, suggesting that the D 1 receptor constitutively and selectively inter- acts with the NR1 subunit of the NMDA channel. Oligomerization with the NR1 subunit substantially modified D 1 receptor trafficking. In individually transfected HEK293 cells, NR1 was localized in the endoplasmic reticulum, whereas the D 1 receptor was targeted to the plasma membrane. In cotransfected micrograms of were incubated overnight at 4 °C with antibodies either the NR1 subunit (1 (cid:1) g/ml) or the D 1 receptor (1:250 mouse monoclonal) in 200 m M NaCl, 10 m M EDTA, 10 m M Na 2 HPO 4 , 0.5% Nonidet P-40, and 0.1% SDS (buffer A-agarose (Santa Cruz Biotechnology) were incubation was extensively washed proteins onto blotted for containing 0.1% Tween 20 and 5% low fat dry h atroom temperature with anti-NR1 (1 (cid:1) g/ml) anti-D 1 receptor (1:250 (ECL, Amersham Biosciences, horseradish peroxidase- conjugated secondary Cloning, Expression, and Purification of GST Fusion Proteins— The C-terminal regions of the D 1 receptor (D 1 -CT-(321–446)) and of the D 5 receptor (D 5 -CT-(373–477)) and two fragments of the NR1 subunit C terminus (NR1-CT-(834–930) and NR1-CT-(834–892)) were generated by PCR amplification, cloned into the pGEX-KG plasmid, and expressed in BL21 competent cells. Synthesis of recombinant proteins was induced by 0.1 m M isopropyl- (cid:2) - D -thiogalactopyranoside (Sigma) for 2–4 h. The bacteria were lysed, and the proteins were purified by incubation with glutathione-agarose beads (50% (v/v) in PBS) for 12

Dopaminergic fibers originating in the substantia nigra and cortical glutamatergic neurons extensively interact in the stri-atum to drive the physiological functions of this structure from motor planning to reward seeking and procedural learning (1,2). The critical importance of dopamine in this system is such that the degeneration of nigral dopaminergic neurons leads to the motor and cognitive deficits of Parkinson's disease (3).
At the cellular level, nigral and cortical fibers converge on the medium spiny projection neurons (4), where dopamine D 1and D 2 -like receptors are coexpressed to high degree with glutamate NMDA 1 and non-NMDA receptor channels (5)(6)(7)(8). From a functional point of view, it is well established that dopamine modulates the firing pattern of these neurons. In particular, there is evidence that dopamine, while attenuating the responses mediated by non-NMDA receptors, potentiates those associated with activation of NMDA receptors (2). The D 1 receptor appears to be involved in these interactions. In fact, activation of D 1 receptors in medium spiny neurons enhances NMDA-induced whole cell currents (2,9) and is a critical requirement for the formation of NMDA-mediated long-term potentiation at corticostriatal synapses (2, 10 -12). Moreover, activation of NMDA receptors in striatal neurons triggers the translocation of cytoplasmic D 1 receptors to the plasma membrane and spines (13). Within neuronal spines, D 1 receptors are mainly localized in the spine shaft and, to a lesser extent, also in the spine head and in the postsynaptic density (PSD) (14 -16). This cell structure is typical of the glutamatergic synapse and consists of a complex network of critical proteins involved in synaptic plasticity, many of which bind directly or indirectly to the NMDA receptor, which is an abundant component of the fraction (17,18). The mechanisms that specifically drive D 1 receptor delivery to different spine domains are still unknown. The partial overlap in the subcellular distribution of NMDA and D 1 receptors and the observation that both D 1 and NMDA receptor delivery to synapses is dependent on glutamate transmission (13,19) suggest that direct protein-protein interactions might direct the trafficking of these receptors to the same subcellular domain.
In this study, we report that the dopamine D 1 receptor forms a heteromeric complex with the NR1 subunit of the NMDA receptor in both purified striatal PSDs and cotransfected cells. This interaction is constitutive, occurs in the endoplasmic reticulum (ER), influences D 1 receptor targeting to the cell membrane, and prevents agonist-induced D 1 receptor internalization.

EXPERIMENTAL PROCEDURES
Materials-Human embryonic kidney cells (HEK293) were provided by Deutsche Sammlung von Mikroorganismen und Zellculturen GmbH (Braunschweg, Germany). Tissue culture media and fetal bovine serum were obtained from Euroclone Celbio (Milano, Italy). Dopamine, glutamate, D-butaclamol, SKF-81297, and the rat monoclonal anti-D 1 receptor antibody (clone 1-1-F11-S.E6) were purchased from Sigma. Glycine was obtained from Tocris (Avonmouth, UK). The rabbit anti-PDI antibody was from Stressgen Biotech Corp. (Victoria, British Columbia, Canada). Cy3-labeled anti-rat and anti-rabbit secondary antibodies were from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). The anti-NR1 and anti-NR2A/B and mouse monoclonal anti-D 1 receptor antibodies were from Chemicon International (Temecula, CA). The horseradish peroxidase-conjugated anti-mouse antibody was purchased from DAKO (Milano), and the horseradish peroxidaseconjugated anti-rabbit antibody was from Santa Cruz Biotechnology (Heidelberg, Germany). Dr. Marc Caron (Duke University, Durham, NC) kindly provided D 1 and D 5 receptor cDNAs. Dr. Hannah Monyer (Heidelberg University) kindly provided the NR2B cDNA, and the NR1 cDNA was a gift of Dr. Shigetada Nakanishi (Kyoto University, Kyoto, Japan).
PSD and Triton-insoluble Fraction Preparation-Striatal PSD were isolated according to Carlin et al. (20) with minor modifications as described previously (21). Briefly, the tissue was homogenized in icecold 0.32 M sucrose containing 1 mM Hepes, 1 mM MgCl 2 , 1 mM NaHCO 3 , 0.1 mM phenylmethylsulfonyl fluoride, and a mixture of protease inhibitors (Complete, Roche Diagnostic, Milano) at pH 7.4 (buffer A) and centrifuged at 1000 ϫ g for 10 min. The supernatant was centrifuged at 3000 ϫ g for 15 min. The resulting pellet (containing mitochondria and synaptosomes) was resuspended in ice-cold 0.32 M sucrose containing 1 mM Hepes, 1 mM NaHCO 3 , and 0.1 mM phenylmethylsulfonyl fluoride (buffer B); overlaid on a sucrose gradient (0.85 to 1.0 to 1.2 M); and centrifuged at 82,500 ϫ g for 2 h. The fraction between 1.0 and 1.2 M was diluted with buffer B containing 1% Triton X-100, stirred at 4°C for 15 min, and centrifuged at 82,500 ϫ g for 30 min. The resulting pellet was resuspended, layered on a sucrose gradient (1.0 to 1.5 to 2.1 M), and centrifuged at 100,000 ϫ g for 2 h at 4°C. The fraction between 1.5 and 2.1 M was removed and diluted with 150 mM KCl containing 1% Triton X-100. PSD were collected by centrifugation at 100,000 ϫ g for 30 min at 4°C.
To isolate the Triton-insoluble fraction (TIF), tissue was homogenized in ice-cold buffer A and centrifuged at 1000 ϫ g for 10 min. The resulting supernatant was centrifuged at 3000 ϫ g for 15 min, and the pellet was resuspended in 1 mM Hepes and centrifuged at 100,000 ϫ g for 1 h. The pellet was resuspended in 75 mM KCl containing 1% Triton X-100, and TIF was collected by centrifugation at 100,000 ϫ g for 1 h. TIF was characterized by enrichment in PSD proteins as previously described (22).
Immunoprecipitation and Western Blotting-Ten micrograms of PSD were incubated overnight at 4°C with antibodies against either the NR1 subunit (1 g/ml) or the D 1 receptor (1:250 dilution; mouse monoclonal) in 200 mM NaCl, 10 mM EDTA, 10 mM Na 2 HPO 4 , 0.5% Nonidet P-40, and 0.1% SDS (buffer C). Protein A-agarose beads (Santa Cruz Biotechnology) were added, and incubation was continued for 2 h at room temperature. The beads were collected and extensively washed with buffer C. The resulting proteins were resolved by SDS-PAGE, transferred onto polyvinylidene difluoride membranes, and blotted for 1 h at room temperature in Tris-buffered saline containing 0.1% Tween 20 and 5% low fat dry milk. Membranes were incubated for 2 h at room temperature with the anti-NR1 (1 g/ml) or anti-D 1 receptor (1:250 dilution) antibodies. Detection was performed by chemiluminescence (ECL, Amersham Biosciences, Milano) with horseradish peroxidaseconjugated secondary antibodies (1:1500 dilution).
Affinity Purification ("Pull-out")-TIF proteins (35 g) were diluted with PBS containing 0.1% SDS and incubated for 1 h at room temperature with glutathione-agarose beads saturated with GST fusion proteins. Beads were washed with PBS containing 0.1% Triton X-100, and bound proteins were resolved by SDS-PAGE and immunoblotted with anti-NR1 and anti-NR2A/B antibodies. Generation of Bioluminescence Resonance Energy Transfer (BRET 2 ) Fusion Constructs-The D 1 receptor and NR1a subunit coding sequences were amplified out of their original vectors using sense and antisense primers containing unique XhoI and BamHI sites and Hin-dIII and BamHI sites, respectively, and the native Pfu DNA polymerase (Stratagene, Milano) to generate stop codon-free fragments. The D 1 receptor fragment was cloned in-frame into the Renilla luciferase-containing vector pRluc-N2(h) (PerkinElmer Life Sciences, Milano) to generate the plasmid D 1 -Rluc. The NR1a fragment was cloned in-frame into the pGFP 2 -N2(h) vector containing the green fluorescent protein (GFP 2 ) (PerkinElmer Life Sciences) to generate the plasmid NR1-GFP 2 . The D 1 -Rluc receptor was tested for its efficiency in activating adenylyl cyclase in transfected COS-7 cells as previously described (24). The influence of GFP 2 on glutamate-mediated 45 Ca 2ϩ influx in COS-7 cells cotransfected with NR1-GFP 2 and NR2B was assessed by standard methods.
Cell Culture, Transfection, and BRET 2 Assay-COS-7 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Semiconfluent cells were cotransfected for 3 h with D 1 -Rluc and NR1-GFP 2 at a 1:4 DNA ratio, which was shown to give the best BRET 2 signal, in the absence or presence of NR2B using the LipofectAMINE technique (Invitrogen, Milano). The total amount of DNA was kept at 10 g. Forty-eight hours post-transfection, cells were harvested, centrifuged, and resuspended in PBS containing 0.1 mg/ml CaCl 2 , 0.1 mg/ml MgCl 2 , and 1 mg/ml D-glucose. Approximately 50,000 cells/well were distributed in a 96-well microplate (white Optiplate, PerkinElmer Life Sciences) and incubated in the absence or presence of 50 M dopamine, 100 M glutamate, and 10 M glycine for 10 min at 37°C. DeepBlueC TM coelenterazine (PerkinElmer Life Sciences) was added at a final concentration of 5 M, and BRET 2 signals were determined using a Fusion TM universal microplate analyzer (PerkinElmer Life Sciences), which allows sequential integration of signals detected at 390/400 and 505/510 nm. Untransfected cells and cells transfected with D 1 -Rluc alone were used to define the nonspecific signals, and cells transfected with the pRluc-GFP 2 control vector (PerkinElmer Life Sciences) were used as positive controls. The BRET signal was calculated as the difference in the ratio between emission at 510 and 395 nm of cotransfected Rluc and GFP 2 fusion proteins and the ratio between emission at 510 and 395 nm of the Rluc fusion protein alone.
Immunofluorescence and Confocal Microscopy-HEK293 cells were maintained in high glucose Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 2 mM glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 units/ml penicillin, and 100 g/ml streptomycin. Semiconfluent cells were transfected with different combinations of D 1 receptor, NR1-GFP 2 , and NR2B cDNAs using LipofectAMINE 2000 reagent (Invitrogen). Twenty-four hours after transfection, cells were plated onto poly-L-lysine-coated coverslips, fixed in 4% paraformaldehyde for 20 min at room temperature, and permeabilized with 0.1% Triton X-100 in PBS containing 5% bovine serum albumin and 5% normal goat serum for 10 min at room temperature. Cells were incubated overnight at 4°C with either the rat monoclonal anti-D 1 receptor antibody (1:600 dilution in PBS containing 1% normal goat serum) or the anti-PDI antibody (1:400 dilution in PBS containing 1% normal goat serum) and then for 45 min at room temperature with the Cy3-conjugated anti-goat secondary antibody (1:1000 dilution). The immunolabeled cells were recorded with a Bio-Rad laser scanning confocal microscope. Untransfected cells and omission of the primary antibodies were used as negative controls.
Sequestration Assay-HEK293 cells, which spontaneously express different G protein-coupled receptor kinases and arrestin (25), were transfected with the D 1 receptor in the absence or presence of NR1 and NR2B subunits using the LipofectAMINE 2000 method, plated onto poly-L-lysine-coated glass coverslips, and allowed to recover for 1 day. Cells were incubated for 1 h at 37°C in the absence or presence of 10 M SKF-81297 and processed as described above for confocal microscopy detection of the D 1 receptor.

Membrane Preparation and [ 3 H]SCH23390
Binding-Cells were rinsed, harvested, and centrifuged at 100 ϫ g for 10 min. Cells were homogenized with a Polytron homogenizer in 5 mM Tris-HCl containing 2 mM EDTA and a mixture of protease inhibitors (pH 7.8) and centrifuged at 80 ϫ g for 10 min to pellet unbroken cells and nuclei. The supernatant was centrifuged at 30,000 ϫ g for 20 min at 4°C. The resulting pellet was resuspended in 50 mM Tris-HCl containing 5 mM MgCl 2 , 1 mM EGTA, and the protease inhibitors (pH 7.8), layered on a 35% sucrose cushion, and centrifuged at 150,000 ϫ g for 90 min to separate the light vesicular and heavy membrane fractions as described by Lamey et al. (26). The heavy fraction, at the bottom of the sucrose cushion, was resuspended in 50 mM Tris-HCl containing 5 mM EDTA, 1.5 mM CaCl 2 , 5 mM MgCl 2 , 5 mM KCl, and 120 mM NaCl (pH 7.4) and used for binding assay. Protein concentration was determined according to Lowry et al. (27) using DC protein assay reagent (Bio-Rad, Milano). Aliquots of membrane suspension (50 g of protein/sample) were incubated at room temperature for 90 min with a saturating concentration (4 nM) of [ 3 H]SCH-23390 (86 Ci/mmol; Perkin-Elmer Life Sciences). Nonspecific binding was defined with 1 M D-butaclamol. The reaction was stopped by rapid filtration under reduced pressure through Whatman GF/C filters.

Dopamine D 1 and Glutamate NMDA Receptors
Are Co-clustered in Striatal PSD-Striatal PSD were isolated and analyzed for the presence of D 1 receptors and other PSD-associated proteins. Fig. 1A shows the results from Western blot analysis performed with different tissue fractions with antibodies recognizing the D 1 receptor, the NR1 subunit, ␣-Ca 2ϩ /calmodulindependent protein kinase II, and protein kinase C⑀. As previously reported (28), immunoreaction with the anti-D 1 receptor antibody revealed a major specific band of ϳ50 kDa. This species was detectable in all tissue fractions, but was enriched in the PSD fraction. This pattern of D 1 receptor distribution paralleled that of ␣-Ca 2ϩ /calmodulin-dependent protein kinase II and the NR1 subunit of the NMDA receptor, two known PSD constituents. The purification yield and the purity of our PSD fraction were confirmed by the absence of immunoreactivity for protein kinase C⑀, a protein present exclusively in the presynaptic compartment. Co-immunoprecipitation studies were performed to evaluate whether D 1 and NMDA receptors might interact in striatal PSD. As shown in Fig. 1B, a 50-kDa band, which was detected by the anti-D1 receptor antibody, was present in PSD proteins immunoprecipitated with the antibody raised against the NR1 subunit (lane 2). Similarly, the anti-D 1 receptor antibody immunoprecipitated a 116-kDa band corresponding to NR1 (lane 3), indicating relevant complex formation between these two holoreceptor moieties in vivo. These bands did not appear when either an irrelevant antibody, such as that directed against protein kinase C⑀ (lane 4), was used or the precipitating antibody was omitted (lane 5).
Pull-out experiments were then performed with GST fusion proteins containing the C-terminal domains of both the D 1 receptor and NR1 subunit. Striatal TIF proteins were incubated with GST fusion proteins containing the D 1 receptor C-terminal tail or, as a control, the D 5 receptor C terminus. D 1 and D 5 receptors display, in fact, particular sequence divergence within the C-terminal domain, a region that might confer subtype-selective properties (29). As shown in Fig. 2A, a 116-  4), was able to bind the NR1 subunit, but not the NR2A/B subunits. B, fusion proteins of GST with two different fragments of the NR1 subunit C terminus (GST-NR1-CT-(834 -930) and GST-NR1-CT-(834 -930)) were incubated with membranes obtained from HEK293 cells expressing the D 1 receptor fused to luciferase (D 1 -Rluc). After extensive washing, the glutathione-agarose beads containing the pulled out proteins were assayed for luciferase activity using DeepBlueC coelenterazine as a substrate. Both NR1 C-terminal fragments were able to bind the D 1 receptor. Bars represent the means Ϯ S.E. of three experiments. *, p Ͻ 0.001 versus GST (Student' t test). WB, Western blot; RLU, relative light units. aration (lane 1) did not interact with GST-D 1 -CT-(321-446), indicating that, in striatal PSD, the D 1 receptor selectively complexes with the NR1 subunit of the NMDA channel through its C-terminal tail. To identify the NR1 region responsible for this interaction, GST fusion proteins containing two different domains of the NR1 C-terminal tail were constructed. As shown in Fig. 2B, GST-NR1-CT-(834 -930), encoding the entire C-terminal region of the NR1a/b isoforms, was able to pull-out luciferase-tagged D 1 receptors from solubilized membrane preparations obtained from transfected HEK293 cells. This activity was still present when the region downstream of the alternatively spliced C1 domain in the NR1 subunit C terminus was deleted, suggesting that both NR1a/b and NR1e/f isoforms may potentially interact with the D 1 receptor. D 1 and NMDA Receptors Constitutively Interact in Living Cells-BRET is a newly developed biophysical approach that detects energy transfer between a luminescent donor and a fluorescent acceptor when they are Ͻ50 -80 Å apart. To evaluate whether D 1 and NMDA receptors could exist as oligomers in living cells, we used an improved BRET technology (BRET 2 , PerkinElmer Life Sciences) that takes advantage of the properties of a particular luciferase substrate, DeepBlueC coelenterazine, which allows a spectral resolution between the Rluc and GFP 2 emissions at ϳ105 nm, a characteristic that confers high sensitivity to the assay. For this purpose, the D 1 receptor was fused to Renilla luciferase, and the NR1 subunit of the NMDA receptor was fused to GFP 2 . The kinetic and transduction properties of these fusion receptors were superimposable with those of their wild-type counterparts (data not shown). BRET 2 signals were determined in COS-7 cells simultaneously or individually expressing the D 1 -Rluc and NR1-GFP 2 constructs. As shown in Fig. 3A, no BRET 2 was observed in cells expressing only NR1-GFP 2 , and a negligible nonspecific signal was detected in cells expressing only the D 1 -Rluc construct. A significant BRET 2 signal was observed in cells expressing a fusion construct covalently linking Rluc to GFP 2 (pRluc-GFP 2 ), confirming the importance of molecular proximity between the BRET partners for signal detection. Coexpression of the tagged D 1 receptor and NR1 subunit yielded a BRET 2 ratio that was significantly higher than that observed with cells expressing D 1 -Rluc alone or with cells individually expressing D 1 -Rluc and NR1-GFP 2 and mixed before analysis. The specificity of this interaction is illustrated by the absence of significant energy transfer between the D 1 -Rluc construct and the pGFP 2 -N2(h) vector (Fig. 3A). This BRET 2 ratio was unchanged when the NR2B subunit of the NMDA receptor was also expressed, suggesting that there is no competition between NR1 and NR2B for interaction with the D 1 receptor. Moreover as shown in Fig.  3B, the BRET 2 signal recorded in cells cotransfected with D 1 -Rluc, NR1-GFP 2 , and NR2B was insensitive to stimulation by 50 M dopamine with or without 100 M glutamate and 10 M glycine. These data demonstrate a physical proximity between D 1 -Rluc and NR1-GFP 2 that can be explained best by the formation of constitutive protein dimers.
Oligomerization with the NMDA Receptor Regulates D 1 Receptor Targeting to the Plasma Membrane-To identify the cellular compartment in which the D 1 receptor and NR1 subunit are assembled, HEK293 cells transfected with the D 1 receptor and NR1-GFP 2 construct, either individually or simultaneously, were analyzed for receptor distribution by confocal microscopy. As shown in Fig. 4a, the D 1 receptor expressed in HEK293 cells was completely targeted to the plasma membrane. By contrast, as previously reported (19,30), when expressed alone, the NR1 subunit accumulated in the perinuclear region and in cytoplasmic compartments with a reticular staining pattern (Fig. 4d) that was identified as the ER using an antibody to PDI, a specific marker for this structure (Fig. 4e). Virtually all the intracellular NR1 staining was in fact colocalized with PDI (Fig. 4f). When the D 1 receptor and NR1 subunit were coexpressed in the same cells, the D 1 receptor was only partially targeted to the cell membrane (Fig. 4, h and i), with the majority of D 1 receptor staining retained in cytoplasmic structures (Fig. 4h), where it was co-localized with NR1 (Fig. 4, g and i). Coexpression of the D 1 receptor with both the NR1 and NR2B subunits relieved the cytoplasmic retention of the complex, allowing insertion of both the NR1 subunit (Fig.  4l) and D 1 receptor (Fig. 4m) at the plasma membrane, where they were completely co-localized (Fig. 4n). These data suggest that D 1 and NMDA receptors are assembled as oligomeric units in the ER and transported to the cell surface as a preformed complex.
Oligomerization with the NMDA Receptor Abolishes Agonistmediated D 1 Receptor Sequestration-A common adaptive response of G protein-coupled receptors to agonist stimulation is redistribution from the plasma membrane to cytosolic compartments. Using confocal microscopy and receptor binding in transfected HEK293 cells, which spontaneously express different G protein-coupled receptor kinases and ␤-arrestin (25), we investigated whether interaction with NMDA receptors alters D 1 receptor sequestration induced by agonist administration (26,31). As shown in Fig. 5A, in unstimulated cells, the fluorescence distribution of the D 1 receptor was exclusively localized at the plasma membrane (panel a). Exposure to 10 M SKF-81297 for 1 h resulted in D 1 receptor sequestration into cytosolic compartments, as shown by the D 1 receptor fluorescence that was detectable also in the cytoplasm with a punctate appearance (panel b). In contrast, when the D 1 receptor was coexpressed with NR1 and NR2B subunits, SKF-81297 failed to induce D 1 receptor internalization. Under these conditions, D 1 receptor immunofluorescence was in fact retained at the plasma membrane (panel c). Similar results were obtained by [ 3 H]SCH23390 binding in the purified heavy membrane fraction. As shown in Fig. 5 (B and C), pretreatment with 10 M SKF-81297 resulted in 20 Ϯ 2.8% reduction of cell-surface [ 3 H]SCH23390 binding in HEK293 cells expressing only the D 1 receptor. On the other hand, exposure to 10 M SKF-81297 did not modify cell-surface [ 3 H]SCH-23390 binding in cells expressing both the D 1 receptor and NR1 and NR2B subunits (Fig. 5, B and C). The dose-response curve and the time course of SKF-81297-induced D 1 receptor sequestration in HEK293 cells expressing the D 1 receptor either alone or in combination with NR1 and NR2B subunits are shown in Fig. 6. The SKF-81297-induced decrease in membrane [ 3 H]SCH-23390 binding was dose-dependent, with an EC 50 of 80 Ϯ 2 nM in cells expressing the D 1 receptor, but not in those coexpressing also the NR1 and NR2B subunits (Fig. 6A). Moreover, in cells expressing only the D 1 receptor, SKF-81297-induced receptor internalization was detectable after 10 min of incubation and reached a maximum within 30 min (Fig. 6B). By contrast, in cells coexpressing the D 1 receptor and the NR1 and NR2B subunits, no decrease in membrane [ 3 H]SCH-23390 binding was detectable at any time tested. Increasing SKF-81297 incubation to 2 h did not modify [ 3 H]SCH-23390 binding as well (data not shown). Taken together, these data suggest that interaction with the NMDA receptor immobilizes the D 1 receptor at the plasma membrane, impairing the mechanisms of the receptor plasticity that normally occurs as an adaptive response to agonist stimulation.

DISCUSSION
In this study, we have shown that, in striatal neurons and in transfected cells, the dopamine D 1 receptor directly and selectively interacts with the NR1 subunit of the NMDA receptor to form a constitutive oligomeric complex that is recruited to the plasma membrane by the NR2B subunit. In medium spiny neurons, a direct protein-protein interaction with the NMDA receptor is thus one of the mechanisms directing the trafficking of D 1 receptors to specific subcellular compartments. Furthermore, we have shown that this interaction abolishes D 1 receptor internalization, a crucial adaptive response that normally occurs upon agonist stimulation (26,31,32).
Using classical biochemical approaches, we have clearly shown that the D 1 receptor is concentrated in purified striatal PSD, displaying a subcellular distribution that is consistent with the reported localization of NMDA receptors (17,18). In addition, the D 1 receptor was co-immunoprecipitated from striatal PSD with NMDA receptor subunits, suggesting that these proteins are co-clustered in this structure. Recently, energy transfer approaches such as fluorescence resonance energy transfer and BRET have been developed as the systems of choice to study protein-protein interactions (33). These techniques have the advantage of monitoring protein oligomerization in living cells without disrupting the natural environment where they are clustered, thus eliminating the possibility of artifactual aggregation that could happen during the solubilization and concentration of membrane proteins. Using BRET 2 , we have demonstrated that D 1 and NMDA receptor clustering reflects the existence of direct protein-protein binding. In fact, the tagged D 1 receptor and NR1 subunit generated a significant and specific BRET 2 signal for energy transfer when cotransfected in COS-7 cells. This signal did not change when the NR2B subunit was also expressed in the same cells, suggesting that this subunit does not compete with the NR1 subunit for binding to the D 1 receptor. In addition, the association of the NR1 subunit with the D 1 receptor was insensitive to agonist stimulation. Taken together, these observations point to a constitutive, direct, and selective interaction of the D 1 receptor with the NR1 subunit of the NMDA channel. Using specific GST fusion proteins, we have also shown that the interaction between D 1 and NMDA receptors involves the binding of the D 1 receptor C-terminal tail to the C-terminal sequence of the NR1a/b and NR1e/f isoforms, with no contribution from NR2 subunits. The NR1 subunit, the essential component of the NMDA receptor, gives rise to eight splice variants, with four possible C termini (34,35). These isoforms differ in their physiological and pharmacological properties and show different regional and cellular distribution (34,36). Our present data point to the capability of interacting with the D 1 receptor as a further difference among these isoforms and suggest that the interaction between D 1 and NMDA receptors might be a specific feature of certain neuronal populations. In line with our findings, it was reported, while this manuscript was in preparation, that D 1 and NMDA receptors directly interact in the hippocampus (37). In particular, in this brain area, the D 1 receptor apparently associates with both the NR1 and NR2A subunits, but not with the NR2B subunit. Our observation that the D 1 receptor does not interact with NR2 subunits in striatal PSD may reflect the fact that NR2B is the prevalent species in this structure (36).
Oligomerization may play important roles in receptor trafficking and/or signaling. In several cases, receptors appear to fold as constitutive dimers early after biosynthesis, whereas ligand-promoted dimerization at the cell surface has been proposed for others (33). Our data obtained by BRET showing that the D 1 receptor and NR1 subunit interact in the absence of the NR2B subunit and in an agonist-independent way suggest that this interaction is constitutive. The results obtained by confocal microscopy give support to this concept and indicate that the trafficking properties of the D 1 receptor are substantially modified by heteromerization with the NMDA receptor. When the NR1 subunit and D 1 receptor were individually transfected in HEK293 cells, NR1 was retained in the ER, whereas the D 1 receptor was targeted to the plasma membrane. In cotransfected cells, both the D 1 receptor and NR1 subunit were colocalized in cytoplasmic compartments, suggesting that interaction with NR1 blocks D 1 receptor delivery to the plasma membrane. In the presence of the NR2B subunit, however, the NR1-D 1 receptor complex was completely translocated to the plasma membrane. These observations are consistent with previous data showing that, when expressed alone in both heterologous cells and cultured hippocampal neurons, the NR1 subunit accumulates in the ER (19,30) due to the presence of an ER retention motif in the alternatively spliced C1 domain in its C terminus (38) and that coexpression of NR2 subunits is necessary to drive the complex to the cell membrane (19,30). Taken together, these data suggest that, in striatal medium spiny neurons, D 1 and NMDA receptors are assembled within intracellular compartments as constitutive heteromeric complexes that are delivered to functional sites. Interaction with the NMDA receptor thus represents a critical mechanism to recruit the D 1 receptor to the PSD. The postsynaptic specialization of corticostriatal glutamatergic synapses finely regulates the strength of synaptic transmission, thus determining the activity of medium spiny neurons. Several lines of evidence suggest that the efficacy of corticostriatal transmission is highly dependent on the concurrent activation of D 1 and NMDA receptors. In particular, it has been shown that NMDA currents are potentiated by activation of D 1 receptors, which is also an essential requirement for long-term potentiation generation (2, 9 -12). In this context, the direct interaction between D 1 and NMDA receptors may be crucial to recruit the D 1 receptor in the place of synaptic plasticity and to keep it in close proximity with the NMDA receptor to allow rapid cAMP/protein kinase A/DARPP32-mediated potentiation of NMDA transmission (39 -42).
The interaction of the D 1 receptor with the NR1 subunit does not reflect simply a chaperon-like strategy to deliver the D 1 receptor to the PSD, but also implies regulation of D 1 receptor function by interfering with the mechanisms of receptor plasticity. A common adaptive response of G protein-coupled receptors to agonist stimulation is desensitization involving both G protein-coupled receptor kinase-mediated phosphorylation and arrestin binding and internalization (43). In line with this paradigm and with in vitro studies (26,31), there is morphological evidence that, in striatal medium spiny neurons, extrasynaptic D 1 receptors, localized in cell bodies and dendrites, respond to agonist administration by massive internalization (32). We have shown here that association with the NMDA receptor abolishes agonist-induced D 1 receptor cytoplasmic sequestration, indicating that oligomerization with NMDA receptors could represent a novel regulatory mechanism modulating D 1 receptor function. Taken together, these observations suggest that, within a single neuron, D 1 receptor plasticity may be subjected to different regulatory mechanisms in different neuronal microdomains. In particular, agonist stimulation would induce D 1 receptor sequestration in all neuronal compartments except the PSD, where this receptor is immobilized at the plasma membrane by association with the NMDA receptor. Along this line, Dumartin et al. (32) have reported that, in striatal medium spiny neurons, the localization of the perisynaptic D 1 receptor in dendritic spines is apparently unmodified by agonist treatment. It is well known that agonist-induced internalization dynamically calibrates receptor availability for extracellular ligands. Disruption of D 1 receptor cytoplasmic sequestration in response to agonist stimulation due to heteromerization with the NMDA receptor might represent a neuronal mechanism to preserve the optimal synaptic strength at corticostriatal synapses in the presence of alterations in the dopamine environment as occurs, for instance, during drug administration. In conclusion, our present data suggesting that, in striatal medium spiny neurons, D 1 and NMDA receptors are assembled within intracellular compartments and re-cruited to the PSD as a constitutive heteromeric complex may provide a new rationale for a better understanding of the mechanisms that control corticostriatal synaptic transmission under both physiological and pathological conditions.