Differential regulation of the dopamine D2 and D3 receptors by G protein-coupled receptor kinases and beta-arrestins.

The D(2) and D(3) receptors (D(2)R and D(3)R), which are potential targets for antipsychotic drugs, have a similar structural architecture and signaling pathway. Furthermore, in some brain regions they are expressed in the same cells, suggesting that differences between the two receptors might lie in other properties such as their regulation. In this study we investigated, using COS-7 and HEK-293 cells, the mechanism underlying the intracellular trafficking of the D(2)R and D(3)R. Activation of D(2)R caused G protein-coupled receptor kinase-dependent receptor phosphorylation, a robust translocation of beta-arrestin to the cell membrane, and profound receptor internalization. The internalization of the D(2)R was dynamin-dependent, suggesting that a clathrin-coated endocytic pathway is involved. In addition, the D(2)R, upon agonist-mediated internalization, localized to intracellular compartments distinct from those utilized by the beta(2)-adrenergic receptor. However, in the case of the D(3)R, only subtle agonist-mediated receptor phosphorylation, beta-arrestin translocation to the plasma membrane, and receptor internalization were observed. Interchange of the second and third intracellular loops of the D(2)R and D(3)R reversed their phenotypes, implicating these regions in the regulatory properties of the two receptors. Our studies thus indicate that functional distinctions between the D(2)R and D(3)R may be found in their desensitization and cellular trafficking properties. The differences in their regulatory properties suggest that they have distinct physiological roles in the brain.

The neurotransmitter/hormone dopamine modulates the activities of many neuronal pathways and peripheral organ systems (for review, see Ref. 1). The three major central dopaminergic pathways include the nigrostriatal, mesocorticolimbic, and tuberoinfundibular pathways that coordinate movement, affect cognition/emotion, and regulate secretion of pro-lactin from the pituitary. The importance of proper dopaminergic function is evident when any of these systems become compromised such as in Parkinson's disease (2), schizophrenia (3), or hyperprolactinemia (4).
Molecular cloning of the dopamine receptor family has revealed five receptor subtypes, D 1 through D 5 . They have been classically divided into two groups based on ligand specificity and effector coupling. D 1 receptor (D 1 R) 1 and D 5 R are positively coupled to adenylyl cyclase by the G protein G s , whereas the D 2 R, D 3 R, and D 4 R inhibit this enzyme (5,6). The D 2 R and D 3 R, which are major potential targets for antipsychotic drugs, differ in pharmacological profiles and in brain distribution. D 3 R displays much higher affinity for endogenous dopamine, and its distribution in the brain is predominantly localized to the limbic area (7).
The architectures of the D 2 R and D 3 R are similar, with the two sharing 46% overall amino acid homology and 78% identity in the transmembrane domains (8). Similarly D 2 R and D 3 R share many signaling properties when they are expressed in mammalian cells. For example, both regulate adenylyl cyclase (9 -11), extracellular acidification (Na ϩ /H ϩ exchange) (9,12,13), mitogenesis (9,14), mitogen-activated protein kinase activation (15,16), dopamine release (17), and ion channel function (18,19). Although it was reported that D 3 R is exclusively expressed in specific limbic areas such as the islands of Calleja and nucleus accumbens (7), more recent studies in monkey and human brain have shown that D 3 R is also expressed in mesencephalic dopaminergic neurons (20) and seems to be coexpressed with D 2 R in the same cells (21). The question arises why two structurally and functionally similar receptor proteins are expressed in both the same and distinct regions of the brain. Unless one of these receptors utilizes an uncharacterized and unshared signaling pathway, one speculation would be that different brain regions require different regulatory properties of the receptors.
A common paradigm of G protein-coupled receptor desensitization is that agonist-induced receptor signaling is rapidly attenuated via G protein-coupled receptor kinase (GRK)-mediated receptor phosphorylation and arrestin binding (22). The arrestin family of proteins to which the ␤-arrestins belong initiates receptor internalization through clathrin-coated pits (sequestration) (23,24), and in this process additional components of the endocytic machinery like dynamin and ␤ 2 -adaptin are known to be involved (25)(26)(27)(28). The sequestration of the D 2 R involves GRKs and dynamin molecules and seems to be a slow cellular process taking 2 h to plateau (29 -31). In contrast to the D 2 R, the sequestration of the D 3 R has not yet been described.
Considering that the D 2 R and D 3 R have similar structural features and signaling pathways but incompletely colocalize in the brain, it became of interest to determine whether properties of desensitization and trafficking of these receptors are similar. We have conducted a series of comparative studies of the cellular processes involved with the receptor sequestration for D 2 R, D 3 R, and chimeras made by interchanging the second and third cytoplasmic loops of each receptor. Our data show that significant differences exist in both desensitization and trafficking properties of these receptors. The swapping of cytoplasmic loops interchanges their phenotypes for phosphorylation, ␤-arrestin binding, and intracellular trafficking.

Materials
Human embryonic kidney cells (HEK-293) and COS-7 cells were provided by the American Type Culture Collection (Manassas, VA). Tissue culture media and fetal bovine serum were obtained from Life Technologies, Inc. Dopamine, spiperone, quinpirole, and sulpiride were purchased from RBI/Sigma. Anti-FLAG M2 antibody was from Eastman Kodak Co., anti-HA antibody was from Roche Molecular Biochemicals, rabbit anti-rat green fluorescent protein (GFP) IgG was from CLONTECH (Palo Alto, CA), and fluorescein-labeled goat anti-mouse secondary antibody was from Sigma.

Generation of Plasmid Constructs
Wild-type human dopamine D 2 R (short form) and D 3 R subtypes in mammalian expression vector pCMV5 have been described in our previous studies (32). The D 2 R was tagged at the amino terminus with the M2-FLAG epitope with the eight-residue sequence of the epitope (DYKDDDDA) inserted after Met 1 ; D 3 R was tagged at the amino terminus with the HA epitope (YPYDVPDYA). In addition, the D 2 R and D 3 R were fluorescently tagged at the carboxyl terminus using a mutant GFP (33). The entire D 2 R and D 3 R were amplified and subcloned into XhoI/HindIII sites of pEGFP-N1 (CLONTECH). Chimeric receptors between D 2 R and D 3 R have been described in our previous studies (11). The expression constructs for GRKs 2 and 3, ␤-arrestins 1 and 2, ␤-arrestin1-GFP, ␤-arrestin2-GFP, V53D-␤-arrestin1, and K44A-dynamin have been described previously (33)(34)(35).

Cell Culture and Transfection
The same experimental procedures were used as described previously (33).

Confocal Microscopy
For the ␤-arrestin translocation assays, HEK-293 or COS-7 cells were transfected with the ␤-arrestin1-GFP or ␤-arrestin2-GFP (33) and either D 2 R or D 3 R with or without GRK2 or GRK3. For the internalization assay, HEK-293 cells were transfected with D 2 R-GFP or D 3 R-GFP in the absence or presence of GRK2 or GRK3. One day after transfection, cells were seeded onto 35-mm dishes containing a centered, 1-cm well formed from a glass coverslip-sealed hole in the plastic and allowed to recover for 1 day. Cells were incubated with 2 ml of minimum essential medium containing 20 mM HEPES, pH 7.4, and viewed on a Zeiss laser scanning confocal microscope.

Sequestration Assay
Three different strategies were used. Method 1-The total receptor level was assessed using [ 3 H]spiperone, a hydrophobic ligand able to label both the intra-and extracellular complements of a receptor. Intracellular receptor levels were measured by displacing the extracellular binding with a high concentration of the hydrophilic ligand sulpiride. The intracellular and extracellular propor-tions of the D 2 R or D 3 R were then calculated before and after cell stimulation and used to calculate the percentage of sequestration. HEK-293 or COS-7 cells were seeded at a density of 3 ϫ 10 6 cells/ 100-mm dish. The following day cells were transfected with D 2 R or D 3 R in the absence or presence of GRKs and/or ␤-arrestins (as indicated in the figure legends) using the calcium phosphate method or Lipo-fectAMINE ® (Life Technologies, Inc.). After 48 -72 h, cells were rinsed once with warm medium containing 10 mM HEPES, pH 7.4, and 100 M ascorbic acid (anti-oxidant) at 37°C. Cells were stimulated with 1-10 M dopamine (DA) for 0 -120 min as indicated. Stimulation was terminated by quickly cooling the dishes on ice, washing the cells three times with ice-cold PBS, and rinsing the cells from the plate in 2 ml of ice-cold PBS/EDTA. Cell aliquots (200 l) were mixed with 50 l of [ 3 H]spiperone (final concentration, 3 nM) in the absence and presence of unlabeled competitive inhibitor. The hydrophobic properties of [ 3 H]spiperone allowed the measurement of total cellular levels of D 2 R or D 3 R (i.e. intra-and extracellular), while displacement of extracellular (i.e. plasma membrane-associated) [ 3 H]spiperone with unlabeled, hydrophilic sulpiride (3 M) allowed direct measurement of internalized receptor. Nonspecific binding was determined in the presence of 10 M spiperone. Assays to measure total, intracellular, and nonspecific binding were performed in triplicate for each time point. Binding assays were incubated for 3 h at 14°C to prevent receptor recycling and terminated by washing three times with vacuum filtration over Whatman GF/C glass-fiber filters using ice-cold Wash Buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl). Samples were mixed with 5 ml of Lefkofluor scintillation fluid and counted on a Packard 1900TR liquid scintillation analyzer.
Method 2-Sequestration of D 2 R was measured according to Itokowa et al. (29) using the hydrophilic properties of sulpiride. HEK-293 or COS-7 cells overexpressing D 2 R in the absence or presence of GRKs and/or ␤-arrestins were seeded 1 day after transfection at a density of 1.5 ϫ 10 5 cells/well in 24-well plates. The following day, cells were rinsed once and preincubated for 15 min with 0.5 ml of prewarmed serum-free medium containing 10 mM HEPES, pH 7.4, at 37°C. Cells were stimulated with 10 M DA for 0 -120 min as indicated. Stimulation was terminated by quickly cooling the plates on ice and washing the cells three times with ice-cold serum-free medium containing 20 mM HEPES, pH 7.4. Cells were incubated with 250 l of [ 3 H]sulpiride (final concentration, 2.2 nM) at 4°C for 150 min in the absence and presence of unlabeled competitive inhibitor (10 M haloperidol). Cells were washed two times with the same medium, and 1% Triton X-100 was added. Samples were mixed with 5 ml of Lefkofluor scintillation fluid and counted on a Packard 1900TR liquid scintillation analyzer.
Method 3-Sequestration of D 2 R or D 3 R was measured using flow cytometry by incubation of plated cells with anti-FLAG M2 IgG or anti-HA antibody (1:500 dilution) for 1 h on ice. Wells were washed two times with ice-cold PBS followed by incubation with Fc-specific, fluorescein-labeled goat anti-mouse antibody (1:250 dilution) for 1 h on ice. Cells were washed three times, rinsed from the plate with 400 l of PBS/EDTA, and fixed in 3.6% formaldehyde. Samples were analyzed within 24 h on a Becton-Dickinson flow cytometer. Baseline fluorescence was determined from a sample of HEK-293 cells transfected with empty vector and was subtracted from each sample.

RESULTS
Ligand Binding Properties of Wild-type D 2 R, FLAG-D 2 R, and D 2 R-GFP-Binding studies to characterize agonist and antagonist binding properties were performed with cell membranes prepared from HEK-293 cells transiently expressing the wild-type D 2 R, the amino-terminally tagged FLAG-D 2 R, or the carboxyl-terminally conjugated D 2 R-GFP. In the presence of increasing concentrations of the D 2 R-selective antagonist spiperone, K i values for displacement of [ 125 I]iodosulpride from D 2 R, FLAG-D 2 R, and D 2 R-GFP were 0.35, 0.71, and 0.11 nM, respectively (Fig. 1A). Similarly, DA and the D 2 R-selective agonist quinpirole (data for DA is not shown), demonstrated the following K i values for the three receptors: DA: 4.49, 6.52, and 1.35 M; quinpirole: 0.59, 0.72, and 0.37 M for D 2 R, FLAG-D 2 R, and D 2 R-GFP, respectively. These values are similar to those previously reported for cloned D 2 R (38, 39), suggesting that these constructs retain characteristic D 2 R ligand binding properties. Pharmacological properties of HA-tagged D 3 R and six different D 2 /D 3 chimeric receptors in which the second or third cytoplasmic loop or both loops were replaced with those of the other receptor are described in detail in our previous studies (32,40).
Comparison of the Phosphorylation of D 2 R and D 3 R-As an initial step to investigate the roles played by GRKs in the regulation of D 2 R and D 3 R, whole-cell phosphorylation assays were conducted in HEK-293 cells overexpressing receptors and/or kinases (Fig. 1B). The major phosphorylated protein in D 2 R-and D 3 R-expressing cells runs as a broad band of 55-80 kDa and is not present in immunoprecipitates of mock-transfected cells (data not shown). As revealed by immunoprecipitation using anti-FLAG-Sepharose beads for FLAG-D 2 R, moderate agonist-induced phosphorylation of D 2 R was evident in the absence of exogenous GRK2 or GRK3 and could be enhanced by co-expression of GRKs (Fig. 1B, upper panel). D 2 R-GFP was also phosphorylated by endogenous kinase (Fig. 1B, lower panel), and overexpression of GRK2 with D 2 R-GFP increased the agonist-induced phosphorylation.
The phosphorylation of D 3 R was different from that of D 2 R. D 3 R was not phosphorylated by agonist stimulation in the absence of exogenous GRK (data not shown), and overexpression of GRK caused only a subtle increase in the level of phosphorylation (Fig. 1C, left panel). When the second and third cytoplasmic loops of D 3 R were replaced with those of D 2 R, D 3 -(IL23-D 2 ), phosphorylation properties of D 3 R became similar to those of D 2 R (Fig. 1C, right panel). These results suggest that the pattern of receptor phosphorylation is determined by cytoplasmic loops of the receptor.
␤-Arrestin-GFP Translocation in Response to D 2 R or D 3 R Activation and Localization of the Regions Involved with ␤-Arrestin Binding-The involvement of ␤-arrestins in the regulation of D 2 R and D 3 R was investigated using a GFP-conjugated ␤-arrestin (␤-arrestin-GFP) translocation assay (33). Prior to agonist addition, cells expressing D 2 R or D 3 R and ␤-arrestin-GFP in the absence or presence of GRKs displayed a uniform distribution of fluorescence throughout the cytosol.
In D 2 R-expressing HEK-293 cells, translocation of ␤-arres-tin2-GFP was observed in response to agonist stimulation regardless of the co-transfection of additional GRKs ( Fig. 2A, left  panel). As expected from the phosphorylation profiles, D 3 R caused only a subtle translocation of ␤-arrestin2-GFP to the plasma membrane only in the presence of overexpressed GRKs ( Fig. 2A, right panel). Consistently GRK2 caused more reliable ␤-arrestin translocation of the D 2 R, while GRK3 was the more effective enzyme for the D 3 R, although quantitative studies were not conducted. As reported for the ␤ 2 -adrenergic receptor (␤ 2 AR) (41), ␤-arrestin2-GFP showed a better translocation than ␤-arrestin1-GFP to activated D 2 R and D 3 R activation (data not shown). In the case of the D 3 R, ␤-arrestin1-GFP did not translocate regardless of overexpression of GRKs. For both the D 2 R and D 3 R, translocated ␤-arrestins co-localized with the receptors and remained on the membrane even when receptors were stimulated for 1 h indicating that they belong to class A G protein-coupled receptors (41).
In our previous studies, it was demonstrated that the second and third cytoplasmic loops of D 2 R and D 3 R are involved with the inhibition of adenylyl cyclase (40). Considering that G proteins and ␤-arrestins usually compete for the same region in the cytoplasmic loops of G protein-coupled receptors, it is likely that both loops are involved in ␤-arrestin binding. We used chimeric D 2 /D 3 receptors (40) to determine which regions are involved with the translocation of ␤-arrestins in the presence of overexpressed GRK3. The stepwise replacement of loops in the D 2 R with those of D 3 R converts it to a D 3 R phenotype for ␤-arrestin translocation (Fig. 2B1), and conversely the removal of D 3 R loops from the D 3 R and replacement by loops from the D 2 R converts it to a D 2 R phenotype (Fig. 2B2). The same pattern was observed in the absence of exogenous GRK2 except that the translocations of ␤-arrestin2 were less intense for all the chimeric receptors in response to agonist activation (data not shown).

Comparison of the Sequestration of D 2 R or D 3 R and Localization of the Regions Involved with the Receptor Sequestra-
tion-Using confocal microscopy the cellular distribution and trafficking of the D 2 R-GFP and D 3 R-GFP were visually assessed in HEK-293 cells. The fluorescence distributions of the receptors were almost exclusively localized to the plasma membrane in unstimulated cells, and this was not altered by coexpression of GRKs (Fig. 3, A and B). Stimulation of D 2 R-GFP with DA resulted in a time-dependent redistribution of fluorescence from the plasma membrane to cytosolic compartments. As shown in Fig. 3A, sequestration of D 2 R was evident in the absence of overexpression of GRK but was significantly enhanced by exogenous GRKs. In contrast, the internalization of D 3 R was very subtle even in the presence of overexpressed GRKs (Fig. 3B).
The sequestration of D 2 R and D 3 R was compared using the [ 3 H]spiperone in HEK-293 cells. As shown in Fig. 3C, dopamine induced a low extent of sequestration of the D 2 R in the absence of exogenous GRKs, and this was enhanced by co-expression of GRK2 or GRK3. Unlike D 2 R, a subtle sequestration of the D 3 R was observed only in the presence of exogenous GRK. In the absence of GRK overexpression, no sequestration of receptor was observed.
The sequestration experiments were also conducted for chimeric receptors and for two wild-type receptors using [ 3 H]sulpiride in COS-7 cells. As shown in Fig. 3C, the chimeric receptors showed the sequestration patterns that might be predicted from the ␤-arrestin2-GFP translocation assay. The D 3 R chimeric receptor whose second and third cytoplasmic loops were replaced with those of D 2 R (D 3 -(IL23-D 2 )) had a similar extent of sequestration to wild-type D 2 R. These results clearly demonstrate that the second and third cytoplasmic loops of D 2 R regulate the translocation of ␤-arrestins and sequestration of the receptor.
Effects of GRKs and ␤-Arrestins on the Basal and Agoniststimulated Sequestration of the D 2 R-Effects of GRKs and ␤-arrestins on the sequestration of the D 2 R were studied in HEK-293 cells using the hydrophobic radioligand [ 3 H]spiperone method, which provided detailed information regarding the receptor proportions present inside the cell or on the cell surface. Table I shows the levels of sequestered D 2 R (proportion of intracellular D 2 R in the basal state) and maximal agonist-induced sequestration levels above basal sequestration in cells transfected with GRKs and/or ␤-arrestins. Co-expression  of GRK2 or GRK3 significantly enhanced both the internalization rate (the half-time was 86 Ϯ 11 and 28 Ϯ 11 min for control and GRK2/␤-arrestin1 group, respectively) and maximal sequestration level (about 4-fold). Although co-expression of ␤-arrestins alone also demonstrated a modest enhancement in both the rate and maximally sequestered receptor levels, the overexpression of GRK was sufficient to elicit maximum sequestration of D 2 R, a situation different from that observed in COS-7 cells. It should also be noted that no change in the total cellular complement of receptor was observed during the 2-h time course for these experiments indicating that down-regulation of receptor did not contribute to decreases in cell surface receptor levels (data not shown).
Internalization Pathway of the D 2 R and D 3 R-The sequestration of the D 2 R in COS-7 cells was assessed by measuring the disappearance of [ 3 H]sulpiride binding sites from the cell surface after agonist stimulation. As shown in Fig. 4A, the sequestration of D 2 R was essentially GRK-/␤-arrestin-dependent, and the greatest potentiation was achieved by the combination of GRK2 and ␤-arrestin2 (Fig. 4A).
It is now well established that many G protein-coupled receptors internalize via a clathrin-coated vesicle-mediated endocytic pathway (42)(43)(44)(45) and that the GTPase dynamin is a major component and marker for this pathway (46 -49). Co-expression of K44A-dynamin markedly inhibited the GRK-/␤arrestin-dependent sequestration of D 2 R (Fig. 4B), suggesting that functional dynamin is required for sequestration of D 2 R and implicating the clathrin-coated endocytic pathway in D 2 R internalization. In the case of the D 3 R, the extent of receptor sequestration was very low, and it was not possible to determine a role for dynamin using analogous approaches to D 2 R. As discussed in Fig. 2A, D 3 R caused a subtle agonist-activated translocation of ␤-arrestin2-GFP only in the presence of exogenous GRKs. This is probably because the affinity of ␤-arres-tin2-GFP for the D 3 R is weak, and translocation is not robust enough for a substantial amount to accumulate at the plasma membrane in the absence of exogenous GRK. If this were the case, the blockade of receptor sequestration would increase the accumulation of translocated ␤-arrestin2-GFP on the plasma membrane. Therefore, HEK-293 cells were transfected with K44A-dynamin in the absence of exogenous GRK with D 3 R and ␤-arrestin2-GFP. As shown in Fig. 4C, K44A-dynamin did not affect the distribution of ␤-arrestin2-GFP in the resting state; however, translocation of ␤-arrestin2-GFP was obvious when cells were treated with dopamine. This result suggests that, although it is very subtle, D 3 R utilizes a clathrin-coated endocytic pathway for sequestration that can be enhanced by exogenous GRKs.
Comparison of D 2 R and ␤ 2 AR in the Utilization of Endocytic Vesicles-Since the endocytic characteristics of the D 2 R were similar to those of ␤ 2 AR, it would be interesting to further test whether D 2 R uses the same endocytic vesicles as the ␤ 2 AR. HEK-293 cells were transfected with D 2 R-GFP and ␤ 2 AR-RFP in the presence of overexpressed GRK2 and stimulated with DA and isoproterenol (1 M each). Both receptors formed clusters on the membrane within 5 min (data not shown); thereafter, sequestration of ␤ 2 AR was a fast process and reached maximum in 20 min. Sequestration of D 2 R, however, continued to increase over the 1-h time course. ␤ 2 AR-RFP and ␤ 2 AR-GFP colocalized on the plasma membrane, in small endocytic vesicles, and in large endosomes as reported previously (50). Internalized D 2 R-GFP could be found on the plasma membrane and in small vesicles but did not co-localize with the ␤ 2 AR-RFP (Figs. 3A and 5). Therefore, although biochemical studies show that ␤ 2 AR and D 2 R both sequester in a clathrin-dependent manner, they are not sorted to the same population of endocytic For the immunoblotting, cells were washed twice with ice-cold PBS and lysed in radioimmunoprecipitation assay buffer for 1 h and centrifuged for 30 min at 45,000 ϫ g. The supernatant containing 20 g of protein was analyzed by SDS-polyacrylamide gel electrophoresis, and the blot was probed with antibodies specific for either GRK2/3 or GRK4/5/6. B, effects of the dominant-negative mutant of dynamin, K44A, on the sequestration of D 2 R. COS-7 cells were transfected with D 2 R, GRK, and ␤-arrestin2 as described in Fig. 4A without or with 2 g of K44Adynamin. C, effects of K44A-dynamin on the translocation of ␤-arres-tin2-GFP in HEK-293 cells expressing D 3 R. Cells were transfected with D 3 R-pCMV5 (3 pmol/mg of protein) and 1 g of ␤-arrestin2-GFP with or without 2 g of K44A-dynamin. Cells were treated with 10 M DA for 5 min. vesicles when simultaneously activated by their respective ligands. The same studies were repeated for the D 3 R, however, few small-sized sequestered vesicles were observed, and we could not reach a clear conclusion whether they co-localize with ␤ 2 AR-RFP. DISCUSSION The D 2 R and D 3 R have long cytoplasmic loops but very short carboxyl tails (10 amino acids) unlike the ␤ 2 AR that has a long carboxyl tail and short cytoplasmic loops. The tail of the ␤ 2 AR is known to undergo GRK phosphorylation and regulate recruitment of ␤-arrestins. Since the short carboxyl tails of the D 2 R and D 3 R are unlikely to be capable of regulating these processes, we tested whether their second and third cytoplasmic loops, which are involved in agonist binding and inhibition of adenylyl cyclase (32,40), could be responsible for these regulatory events. The phosphorylation, arrestin translocation, and sequestration assays clearly show that the intracellular trafficking properties of the D 2 R and D 3 R are mediated by the composition of their second and third cytoplasmic loops.
There seems to be some uncertainty regarding the role of clathrin pathways in mediating D 2 R endocytosis that may have arisen from the methodology used to assess sequestration (31,51). In our experiments the sequestration of the D 2 R was significantly increased by co-expression of GRK and ␤-arrestin but significantly reduced by dominant-negative dynamin (29,51). These results indicate that the D 2 R internalizes in a clathrin-dependent manner in agreement with previous studies (31,34) despite the fact it does not co-localize with the ␤ 2 AR in large endosomes.
The characteristics of D 2 R sequestration varied with cell type. In COS-7 cells the sequestration of the D 2 R was observed only in the presence of exogenous GRKs or ␤-arrestins. In contrast, in HEK-293 cells we could observe a moderate degree of sequestration of the D 2 R without overexpression of GRK or ␤-arrestins. This cellular difference is most likely due to higher levels of GRKs and ␤-arrestin expression in HEK cells as reported previously (52).
Differences in the signaling or brain distribution of the D 2 R and D 3 R could ultimately affect their physiological roles. Numerous unsuccessful efforts have been made to identify a difference in G protein signaling between the D 2 R and D 3 R. If signaling behavior is not responsible for differences in their physiological roles, perhaps differences in their regulation are. We have observed that the D 3 R has a much reduced ability to translocate ␤-arrestin to the plasma membrane and sequesters much less than the D 2 R. Thus, upon agonist stimulation a proportionally larger fraction of D 2 R would desensitize and disappear from the plasma membrane compared with the D 3 R. Perhaps their dissimilar distributions in the brain are necessary consequences of these differences and could have therapeutic implications. For example, the D 3 R is found mainly in the limbic system where antipsychotic drugs target dopamine signaling. Our results showing a difference in their regulatory properties may provide a new rationale for the development of selective antipsychotic drugs.