Functional analysis of the human D2 dopamine receptor missense variants.

The human dopamine D2 receptor gene (DRD2) has three polymorphic variants that predict the amino acid substitutions Val96 --> Ala, Pro310 --> Ser, and Ser311 --> Cys in the receptor protein. We have investigated the ligand binding and signal transduction properties of these human D2 receptor variants by stably expressing them in cultured mammalian cells. The Cys311 and Ser310 variants of the human D2 receptor, which involve substitutions located in the third cytoplasmic loop, were markedly less effective in inhibiting cAMP synthesis than the most prevalent form (Pro310, Ser311). Despite this difference, the Cys311 and Ser310 variants couple to G proteins in CHO-K1 (Chinese hamster ovary) cells. The impairment of the Cys311 and Ser310 variants to inhibit cAMP levels thus appears to result from a reduced ability of those variant receptors to activate the appropriate Gi-like protein. The demonstration of substantial functional differences between DRD2 gene variants found in the human population might have important pharmacological implications given the widespread use of D2 receptor blocking drugs in the treatment of schizophrenia and other psychiatric disorders.

The human dopamine D 2 receptor gene (DRD2) has three polymorphic variants that predict the amino acid substitutions Val 96 3 Ala, Pro 310  Dopamine receptors are members of a large superfamily of receptor proteins coupled to heterotrimeric guanine nucleotidebinding regulatory proteins (G proteins). The dopamine receptors are involved in motor control, neuroendocrine regulation, cognition, and emotion, and are crucial targets in the pharmacological therapy of schizophrenia, Parkinson's disease, Tourette's syndrome, tardive dyskinesia, and Huntington's disease (1). Five human dopamine receptors have been identified to date (D 1 , D 2 , D 3 , D 4 , and D 5 /D 1b ), which are classified into two subfamilies (D 1 -like and D 2 -like). These two subfamilies have different pharmacologies, signal transduction properties, and genomic organization (2). The receptors of the D 1 -like subfamily, D 1 and D 5 , stimulate cAMP synthesis through coupling with G s -like proteins, and their genes do not contain introns within their protein coding regions. Two of the members of the D 2 -like family, D 2 and D 4 , inhibit cAMP synthesis through their interaction with a G i -like proteins and all members (D 2 , D 3 , and D 4 ) share a similar genomic organization, which includes introns within their protein coding regions (3).
Further diversity in the human dopamine receptors can be achieved by alternative RNA splicing and by the existence of expressed polymorphic sequences (1)(2)(3). For instance, the DRD2 gene can generate two alternative RNA splicing isoforms, D 2L (long) and D 2S (short), that differ by the presence or absence of a 29-amino acid sequence in the third cytoplasmic loop of the receptor protein (4,5). The human DRD2 gene also has three identified DNA variants, which predict the amino acid substitutions Val 96 3 Ala, Pro 310 3 Ser, and Ser 311 3 Cys in the receptor protein (6). The prevalence of these DRD2 variants varies greatly in different human populations; in the Pima Native American population, the prevalence of the Cys 311 DRD2 allele is as high as 16% 1,2 ; in the Japanese, the prevalence of Cys 311 is 2.3% (7). In the Caucasian population the prevalence of the DRD2 variants was found to be 3% for Cys 311 , and 0.4% for Ser 310 , with only one reported case for Ala 96 (1:392) (6,8). Other human dopamine receptors also exhibit polymorphic variants. The D 4 receptor is the most polymorphic dopamine receptor with 20 different expressed variants identified to date (9,10). Five polymorphisms coding for amino acid changes have been recently identified for the D 5 receptor (11), and the D 3 receptor is known to have one variant (12). No protein sequence variant has been identified for the D 1 receptor to date (13). D 2 dopamine receptors are known to bind typical antipsychotic drugs with high affinity, and these drug affinities are correlated with clinical potencies in controlling psychotic symptoms in schizophrenia (14). Atypical antipsychotic drugs have been recently developed, which bind to serotonin as well as dopamine receptors; nevertheless, their improvement of positive psychotic symptoms has been suggested to result from blockade of D 2 receptors in the brain limbic system (15). If variations of the DRD2 sequence found in the human population are physiologically or pharmacologically relevant, they would be expected to change an associated biochemical variable, such as agonist and/or antagonist binding, interaction with G proteins, or signal transduction. For instance, the substitution in Ala 96 is located in the putative second transmembrane domain and therefore might be expected to affect the ligand binding pocket of the D 2 receptor, since residues in the second and other transmembrane domains have been shown to regulate ligand binding (reviewed in Ref. 16). Substitutions in the Ser 310 and Cys 311 variants are located in the putative third cytoplasmic loop and might be expected to affect the interaction with G proteins, and therefore signal transduction, as studies with mutated G protein-coupled receptors have shown that this domain is a major determinant of G protein coupling (17,18). Therefore, we have analyzed the ligand binding and signal transduction properties of the variants of the human D 2 receptor by expression in mammalian cell lines, which do not naturally express endogenous D 2 receptors. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Mutagenesis, Cell Culture, and Transfection-A cDNA fragment coding for the short form of human D 2 receptor (4) (kindly provided by Dr. D. K. Grandy, Oregon Health Sciences University, Portland, OR) was subcloned into the eukaryotic expression vector pSVK3 (Pharmacia Biotech, Uppsala, Sweden) and used as a template for site-directed mutagenesis with the Transformer kit (Clontech Laboratories, Palo Alto, CA) using three different oligonucleotides containing the sequences from the Ala 96 , Ser 310 , and Cys 311 variants. Confirmation of the mutations was obtained by DNA sequencing. For stable transfections, the four D 2 receptor cDNA inserts were subcloned into pcDNA3 (Invitrogen Corp., San Diego, CA). African green monkey kidney (COS-7) and Chinese hamster ovary (CHO-K1) 3 cells were cultured in Ham's F-12 medium supplemented with 10% fetal bovine serum in a humidified atmosphere of 5% CO 2 at 37°C, and transfected with 30 g of plasmid DNA using LipofectAMINE reagent (Life Technologies, Inc.). Stable transfectant clones were selected in the presence of 1 mg/ml G418 (Life Technologies, Inc.). Positive receptor-expressing cell lines were identified by [ 3 H]methylspiperone binding and maintained in culture with constant selection in the presence of 0.3 mg/ml G418.
Radioligand Binding Assays-Cells were harvested with 5 mM EDTA in Ca 2ϩ /Mg 2ϩ -free EBSS and centrifuged at 300 ϫ g for 8 min. The cells were washed with EBSS, resuspended in 5 mM Tris, pH 7.4, at 4°C, 5 mM MgCl 2 , and homogenized. Cell membranes were collected by centrifugation twice at 34,000 ϫ g for 20 min with an intermediate resuspension and were finally suspended in binding buffer (50 mM Tris, pH 7.4, at 20°C, 4 mM MgCl 2 , 1 mM EDTA, 5 mM KCl, 1.5 mM CaCl 2 , 120 mM NaCl) at 0.3 mg of protein/ml. Membrane suspension (100 l) was added to triplicate assay tubes containing varying concentrations of [ 3 H]methylspiperone in a final volume of 1 ml for saturation analysis. Nonspecific binding was defined using 1 M (ϩ)-butaclamol. The assay tubes were incubated at room temperature for 1 h, and the assay was terminated by rapid filtration through GF/C filters pretreated with 0.3% polyethylenimine. The filters were rapidly washed with 3 ϫ 4 ml of 50 mM Tris-HCl, pH 7.4, at 4°C, and radioactivity bound to the filters was quantitated by scintillation counting. Protein concentrations were determined using the bicinchoninic acid protein reagent (Pierce).
Radioligand Competition Assays-Competition analyses of radioligand binding were carried out using a high magnesium, low ionic strength buffer consisting of 50 mM Tris-HCl, pH 7.4, at 20°C, 10 mM MgSO 4 , 1.2 mM EDTA, and 0.1% ascorbic acid. Membrane suspension (100 l) was added to triplicate or duplicate assay tubes containing 0.3 nM [ 3 H]methylspiperone and increasing concentrations of dopamine (10 nM to 1 mM) in a final volume of 1 ml. Nonspecific binding was defined using 1 M (ϩ)-butaclamol. To analyze the binding affinity of uncoupled receptors, dopamine competition assays were also performed with the addition of either 0.2 mM GTP or 0.1 mM Gpp(NH)p in the reaction buffer and also after overnight cell treatment with 0.1 g/ml pertussis toxin.
cAMP Assays-D 2 receptor-expressing CHO cells were seeded in 96-well plates at 250,000 cells/ml (50,000 cells/well) in F-12 medium with 10% charcoal-treated fetal bovine serum and incubated at 37°C for 18 h. Cells were washed with EBSS and incubated in triplicate assay wells for 15 min at 37°C in EBSS containing 10 M forskolin, 100 M RO-20-1724 (a phosphodiesterase inhibitor), 1 mM sodium ascorbate, and serial dilutions of dopamine (50 pM to 1 M). After the drug incubations, the cells were extracted in 3% perchloric acid on ice for 15 min and precipitated by addition of 15% KHCO 3 , followed by centrifugation at 600 ϫ g at 4°C for 10 min. The amount of cAMP generated was quantitated by scintillation counting using a [ 3 H]cAMP assay kit (Diagnostic Products Corp., Los Angeles, CA).
Data Analysis-Radioligand binding data was analyzed with the nonlinear least-squares fitting program GraphPad Prism version 1. Competition curves were analyzed using equations for displacement of radioligand and competitor to one and two binding sites. Results from curve fitting using a two-site model were retained when the two-site model fit the data significantly better than a one-site model, as determined by an F test at a significance level of p Ͻ 0.05. Inhibition constants (K i ) were calculated from the EC 50 values by the method of Cheng and Prusoff (19).

D 2 Receptor-expressing Cell Lines-
The cDNAs encoding the human D 2 receptor variants were initially tested by transient expression in COS-7 cells. Receptor densities in transiently transfected COS-7 cells ranged from 5.3 to 7.2 pmol/mg of membrane protein, as determined by [ 3 H]methylspiperone binding. There were no differences between the variants with respect to the level of receptor expression or the affinities of the radioligand (data not shown).
In order to create stably transfected cells, CHO-K1 cells were transfected with the four D 2 receptor cDNAs. Single G418resistant clones were isolated and their D 2 receptor expression characterized by [ 3 H]methylspiperone binding. All D 2 receptorexpressing CHO-K1 cell clones displayed saturable, concentration-dependent [ 3 H]methylspiperone binding, with K d values (mean Ϯ S.E., n ϭ 5) of 76 Ϯ 12 pM for Ala 96 , 66 Ϯ 7 pM for Ser 310 , 64 Ϯ 7 pM for Cys 311 , and 73 Ϯ 10 pM for the most prevalent D 2 receptor variant (Val 96 , Pro 310 , Ser 311 ) or "wild type" (Fig. 1). Thus, there was no significant difference between the variants and wild type receptor for [ 3 H]methylspiperone binding (ANOVA, F (3,14) ϭ 0.32, p ϭ 0.80). For the purpose of further comparison among the different D 2 receptor variants, G418-resistant clones expressing the D 2 receptor variants were classified into two groups: low expression (0.4 -0.7 pmol/mg of membrane protein) and high expression (1-2 pmol/mg of membrane protein). Both groups were further tested for regulation of cAMP synthesis inhibition and ligand binding.
Inhibition of Intracellular cAMP Synthesis by D 2 Receptor Variants-We investigated whether the amino acid substitutions in the human D 2 receptor variants result in a functional change in their ability to inhibit cAMP synthesis in CHO-K1 cells. The cell clones expressing the D 2 variants showed no differences in basal cAMP levels (in picomoles per well): 0.33 Ϯ 0.10 pmol for Ala 96 , 0.35 Ϯ 0.08 pmol for Ser 310 , 0.34 Ϯ 0.07 pmol for Cys 311 , and 0.36 Ϯ 0.12 pmol for D 2 wild type (mean Ϯ S.E., n ϭ 5). After 15 min of treatment with 10 M forskolin, a drug that stimulates adenylyl cyclase, the intracellular cAMP levels were increased to 18 manner (Fig. 2). With a maximal stimulation of 1 M dopamine, the cAMP levels in forskolin-treated cells were 1.5 Ϯ 0.4 pmol for Ala 96 , 4.9 Ϯ 1 pmol for Ser 310 , 14.8 Ϯ 2.9 pmol for Cys 311 , and 1.4 Ϯ 0.3 pmol for D 2 wild type (mean Ϯ S.E., n ϭ 5). Therefore, the wild type and Ala 96 variant receptors were able to reduce the forskolin-induced cAMP levels to 7.6 Ϯ 1.7% and 5.2 Ϯ 2% of control, respectively (mean Ϯ S.E., n ϭ 5). In contrast, the variants Ser 310 and Cys 311 only reversed forskolin-induced cAMP levels to 24.3 Ϯ 5.1% and 58.6 Ϯ 10.5% of control, respectively (mean Ϯ S.E., n ϭ 5). The differences in the ability of the D 2 receptor variants to reduce forskolininduced cAMP levels were independent of their expression levels in the CHO-K1 cell clones. The variants Ser 310 and Cys 311 showed similar degrees of inhibition of cAMP levels when tested in cell clones expressing low (0.5 pmol/mg) versus high (1.5 pmol/mg) levels of the D 2 variants. The EC 50 values for dopamine to reduce forskolin-induced cAMP levels were 3 nM for Ala 96 , 2.3 nM for Ser 310 , 4.9 nM for Cys 311 , and 2.4 nM for D 2 wild type (averaged data from five experiments). These data indicate that the substitution of a proline by a serine at position 310 and of a serine by a cysteine at position 311 of the human D 2 receptor protein results in an impairment of the ability to inhibit intracellular cAMP levels upon dopamine stimulation.

Dopamine and [ 3 H]Methylspiperone Competition
Binding-Dopamine binding to the D 2 receptor variants was initially tested by [ 3 H]methylspiperone competition in the presence of 0.2 mM GTP, revealing single K i values (mean Ϯ S.E., n ϭ 3) of 228 Ϯ 26 nM for Ala 96 , 115 Ϯ 23 nM for Ser 310 , 222 Ϯ 28 nM for Cys 311 , and 120 Ϯ 9 nM for the D 2 wild type. Thus, the Ala 96 and Cys 311 variants exhibited approximately 2-fold lower affinities for dopamine than the Ser 310 variant and D 2 wild type receptor. These differences in K i values for dopamine between the human D 2 variants were also observed when the assays were performed in the reaction buffer containing sodium chloride (data not shown) and reached statistical significance (ANOVA, F (3,8) ϭ 7.11, p ϭ 0.012).
We next investigated the interaction of the human D 2 receptor variants with G proteins present in the CHO-K1 cells by performing dopamine displacement of [ 3 H]methylspiperone in parallel assays with and without 0.1 mM Gpp(NH)p and on membranes from cells treated with or without pertussis toxin. Gpp(NH)p and pertussis toxin both uncouple G protein-coupled receptors but through different molecular mechanisms. The nonhydrolyzable GTP analogue Gpp(NH)p binds to the ␣ subunit of any G protein heterotrimer, causing permanent activation. Pertussis toxin catalyzes ADP-ribosylation of the ␣ sub-units of G i , G o , and G t proteins, effectively blocking signal transduction. The conversion of two dopamine binding states (high and low affinity) to one state (low affinity) by both agents indicates the coupling of the D 2 receptor variant to a G protein present in the CHO-K1 cells. Competition curves were analyzed using equations for displacement of radioligand and competitor to one and two binding sites. Results from curve fitting using a two-site model were retained when this model fit the data significantly better than a one-site model, as determined by an F test at a significance level of p Ͻ 0.05. Averaged data from eight dopamine competition curves of all human D 2 receptor variants fit significantly better to the two-binding site model (Fig. 3). The D 2 wild type receptor had 44% of the dopamine binding sites in the high affinity state with K H ϭ 7.9 Ϯ 2.2 nM and a low affinity state with K L ϭ 403 Ϯ 96 nM (mean Ϯ S.E., n ϭ 5). The Ala 96 variant had 46% high affinity receptors with K H ϭ 11.7 Ϯ 1.4 nM and a K L ϭ 607 Ϯ 189 nM (mean Ϯ S.E., n ϭ 4), the Ser 310 variant had 56% high affinity receptors with K H ϭ 13.5 Ϯ 1.5 nM and a K L ϭ 617 Ϯ 73 nM (mean Ϯ S.E., n ϭ 4), and the Cys 311 variant had 47% high affinity receptors with K H ϭ 9.7 Ϯ 3.3 nM and a K L ϭ 493 Ϯ 67 nM (mean Ϯ S.E., n ϭ 5). Inclusion of 0.1 mM Gpp(NH)p in the reaction buffer resulted in dopamine competition curves with single K i values of 228 Ϯ 18 nM for Ala 96 , 212 Ϯ 40 nM for Ser 310 , 417 Ϯ 64 nM for Cys 311 , and 204 Ϯ 32 nM for D 2 wild type (mean Ϯ S.E., n ϭ 5). As previously observed, the affinity for dopamine of the Cys 311 variant was about 2-fold lower than the affinities of the Ser 310 variant and D 2 wild type receptor. Similar experiments were performed after overnight treatment with 0.1 g/ml pertussis toxin to identify the G proteins interacting with the D 2 receptor variants. This treatment completely abolished the inhibition of forskolin-stimulated cAMP levels (results not shown). After pertussis toxin treatment, the single K i values were: 297 Ϯ 108 nM for Ala 96 , 288 Ϯ 106 nM for Ser 310 , 380 Ϯ 161 nM for Cys 311 , and 320 Ϯ 131 nM for D 2 wild type (mean Ϯ S.E., n ϭ 3). DISCUSSION We have analyzed the ligand binding and signal transduction properties of the human D 2 receptor polymorphic variants Ala 96 , Ser 310 , and Cys 311 expressed in cultured mammalian cells. The affinities of the Ala 96 and Cys 311 variants for dopamine were approximately 2-fold lower than the affinities of the Ser 310 variant and D 2 wild type. No significant differences were detected in the binding affinity for the antagonist [ 3 H]methylspiperone between the D 2 wild type receptor and the expressed variants. Therefore, the amino acid changes in Ala 96 and Cys 311 seem to have a small effect on agonist but not antagonist binding. The amino acid substitution in Ala 96 is located in the second transmembrane domain and could affect the ligand binding pocket of the D 2 receptor. The substitution in the Cys 311 variant is located in the third cytoplasmic loop, and some sequence changes in this receptor domain have been shown to shift agonist affinity without affecting antagonist binding (20). However, the differences in binding affinity for dopamine detected between some of the human D 2 receptor variants are quantitatively small, and their biological significance is unknown.
We analyzed the modulation of cellular cAMP synthesis by the human D 2 receptor variants as a model for signal transduction. The CHO-K1 clones tested had similar basal cAMP concentrations, which were greatly increased by forskolin treatment; this indicates that none of the amino acid substitutions in the D 2 receptor variants results in a constitutive activation of the receptor. Dopamine stimulation of the D 2 receptor variants and the wild type receptor produced an inhibition of forskolin-stimulated cAMP levels in a concentration-dependent manner. Those changes in cAMP concentration are likely due to inhibition of adenylyl cyclase activity by a G i -like protein coupled to the D 2 receptors. There were marked differences in the ability of the different variants to inhibit cAMP synthesis. The D 2 wild type and Ala 96 variant receptors inhibited cAMP synthesis almost completely (to 6% of forskolin-stimulated cAMP levels). In contrast, the Ser 310 and Cys 311 variants were only able to inhibit cAMP synthesis to 24% and 58% of forskolin-induced levels, respectively (Fig. 2). These differences were observed in CHO clones expressing low levels (0.5-0.7 pmol/ mg) and high levels (1-2 pmol/mg) of the human D 2 receptor variants. These data also indicate that the results are not cell line-specific or are an artifact of the transfection process. Therefore the human D 2 receptor variants with amino acid substitutions in the third cytoplasmic loop, Cys 311 and Ser 310 , have an impairment in their ability to modulate adenylyl cyclase activity.
The interaction between the human D 2 receptor variants and G proteins was further investigated with the use of Gpp(NH)p and pertussis toxin. No quantitative differences in coupling to G proteins expressed by the CHO cells were found between the D 2 receptor variants and wild type. This G protein coupling was blocked by membrane incubation with Gpp(NH)p and by cell treatment with pertussis toxin. These data indicate that the impairment in inhibiting cAMP synthesis by the Cys 311 and Ser 310 variants is not due to a defect in G protein coupling but rather to a lower efficiency in activating the ␣ subunit of the G protein heterotrimer. Activation of G protein-coupled receptors by an agonist induces a conformational change in the receptor, which is then able to activate the ␣-subunit of the G protein (G␣). This activation is achieved by inducing a conformational change in the G␣ subunit that decreases its affinity for GDP (reviewed in Ref. 21). The amino acid substitutions Cys 311 and Ser 310 , located in the third cytoplasmic loop of the human D 2 receptor, may render it less efficient in inducing the conforma- tional change required to activate G␣.
An alternative explanation would be that the amino acid substitutions in the third cytoplasmic loop affect the G protein coupling specificity of the variant receptors. If the sequence changes in the Cys 311 and Ser 310 variants result in a fraction of these receptors coupling to G s , their stimulatory effect on cAMP synthesis would decrease the inhibitory G i -mediated effect. Coupling to a pertussis toxin-insensitive G protein (such as G s ) would explain a lower K i for dopamine of the Cys 311 variant after pertussis toxin treatment than in the presence of Gpp(NH)p (Fig. 3). To test this hypothesis, we performed cAMP assays after overnight cell treatment with 0.1 g/ml pertussis toxin; under these conditions, G i -mediated signal transduction is blocked while G s is unaffected. However, there was no detectable increase of cAMP synthesis by dopamine with any of the variants and wild type D 2 receptors after pertussis toxin treatment without forskolin stimulation (results not shown).
The role of the diversity in dopamine receptor sequences in the human population poses an interesting question. It has been hypothesized that polymorphisms in dopamine receptors may correlate with individual differences in phenotype, including behavior, response to drugs, or susceptibility to disease. However, it cannot be predicted what effect, if any, a small functional difference between receptor variants may have in the dopaminergic systems in the brain and on the behavior of an individual carrying such a variant. Interestingly, an association between D 4 alleles and a human behavioral trait has recently been reported (22,23). Variants of the human D 4 receptor have been shown to have small differences in affinity for clozapine in the presence of sodium chloride, and in the EC 50 for dopamine to inhibit cAMP synthesis (10,24). Those reported differences between D 4 variants for inhibiting cAMP were smaller than the differences observed for the D 2 variants Cys 311 and Ser 310 .
Genetic associations between polymorphisms in the human DRD2 gene and several pathologies have been claimed, including alcoholism, polysubstance abuse, and schizophrenia (25)(26)(27)(28). As several groups have failed to replicate them independently, those reports remain unconfirmed (6, 29 -33). Mutations in G protein-coupled receptor genes have been shown to be the cause of several diseases. Mutations can cause receptor inactivation, as in the V2 vasopressin receptor in X-linked nephrogenic diabetes insipidus or constitutive activation as for rhodopsin in severe retinitis pigmentosa and congenital night blindness (reviewed in Ref. 34). Mutations in G protein-coupled receptors, although not implicated in a disease, could be associated with secondary effects to pharmacological treatment. For example, asthmatic patients who are homozygous for a mutation in the ␤ 2 adrenergic receptor were found to be more likely to be steroid-dependent (35). In addition, the Ser 310 variant of DRD2 was found in a patient with a clinical history of neuroleptic malignant syndrome (NMS) (8), and a larger series of patients with NMS is being investigated to establish whether this allele is associated with an increased susceptibility to NMS. Mutations in G protein-coupled receptors do not necessarily produce clinical pathology, e.g. the most common variation in human color vision is due to a polymorphism in the red opsin gene (36).
Given the allele frequencies of the DRD2 variants that predict amino acid substitutions in the human D 2 receptor protein in the Caucasian population, most individuals carrying the variant receptors will be heterozygous. In those individuals, the co-expression of both alleles might complicate the evaluation of any pharmacological and behavioral effect that could be associated with a variant allele. However, in the Pima Native American population, the prevalence of the Cys 311 DRD2 allele was found to be 16%, and other ethnic groups could also have higher frequencies of some of the DRD2 variant alleles. The potential effect of an impairment in D 2 receptor signal transduction in the behavior or response to neuroleptic treatment in individuals expressing the Cys 311 or Ser 310 variants raises interesting questions that require further investigation.