Inhibition of Serotonin 5-Hydroxytryptamine2C Receptor Function through Heterodimerization

Although dimerization appears to be a common property of G-protein-coupled receptors (GPCRs), it remains unclear whether a GPCR dimer binds one or two molecules of ligand and whether ligand binding results in activation of one or two G-proteins when measured using functional assays in intact living cells. Previously, we demonstrated that serotonin 5-hydroxytryptamine2C (5-HT2C) receptors form homodimers (Herrick-Davis, K., Grinde, E., and Mazurkiewicz, J. (2004) Biochemistry 43, 13963-13971). In the present study, an inactive 5-HT2C receptor was created and coexpressed with wild-type 5-HT2C receptors to determine whether dimerization regulates receptor function and to determine the ligand/dimer/G-protein stoichiometry in living cells. Mutagenesis of Ser138 to Arg (S138R) produced a 5-HT2C receptor incapable of binding ligand or stimulating inositol phosphate (IP) signaling. Confocal fluorescence imaging revealed plasma membrane expression of yellow fluorescent protein-tagged S138R receptors. Expression of wild-type 5-HT2C receptors in an S138R-expressing stable cell line had no effect on ligand binding to wild-type 5-HT2C receptors, but inhibited basal and 5-HT-stimulated IP signaling as well as constitutive and 5-HT-stimulated endocytosis of wild-type 5-HT2C receptors. M1 muscarinic receptor activation of IP production was normal in the S138R-expressing cells. Heterodimerization of S138R with wild-type 5-HT2C receptors was visualized in living cells using confocal fluorescence resonance energy transfer (FRET). FRET was dependent on the donor/acceptor ratio and independent of the receptor expression level. Therefore, inactive 5-HT2C receptors inhibit wild-type 5-HT2C receptor function by forming nonfunctional heterodimers expressed on the plasma membrane. These results are consistent with a model in which one GPCR dimer binds two molecules of ligand and one G-protein and indicate that dimerization is essential for 5-HT receptor function.

G-protein-coupled receptors (GPCRs) 2 represent one of the largest families of signaling proteins in the human genome and are targets for a wide variety of therapeutic agents. Recent studies investigating GPCR structure and function indicate that they form dimeric or oligomeric complexes. Homodimerization and heterodimerization between different families of GPCRs have been reported to regulate ligand binding, second messenger activation, and receptor trafficking (reviewed in Ref. 1). GPCR dimerization has been reported to involve the N-terminal domain, transmembrane domains, and the C-terminal domain of interacting receptors, and agonists have been reported to increase or decrease or to have no effect on GPCR dimerization (reviewed in Refs. [2][3][4]. Although these studies indicate that dimerization may be a common property of GPCRs, the specific mechanisms and functional consequences of dimerization may be vastly different from one GPCR family to another. In the case of ␥-aminobutyric acid type B receptors, the functional significance of dimerization is very clear. Heterodimerization between ␥-aminobutyric acid type B receptors 1 and 2 is essential for trafficking and expression of functional receptors on the plasma membrane (5)(6)(7). However, for the majority of GPCRs studied to date, the functional significance of dimerization remains unknown.
Serotonin (5-hydroxytryptamine (5-HT)) receptors represent one of the largest subfamilies of GPCRs. There are 14 different 5-HT receptor subtypes widely distributed throughout the peripheral and central nervous systems, representing therapeutic targets for drugs used to treat anxiety, depression, schizophrenia, obesity, irritable bowel syndrome, migraine, emesis, and other disorders (reviewed in Refs. 8 and 9). If dimerization plays a role in regulating the function of 5-HT receptors, it could have important implications for the development of novel serotonergic therapeutic agents. Immunoprecipitation and Western blot experiments suggest that 5-HT 1 receptors may form homodimers and heterodimers (10,11). Previously, we used biochemical and biophysical techniques to demonstrate that 5-HT 2C receptors are present as constitutive homodimers on the plasma membrane of living human embryonic kidney 293 (HEK293) cells (12). However, the functional significance of 5-HT receptor dimerization (if any) remains unknown. Also, it remains unclear whether a GPCR dimer binds one or two molecules of ligand and whether ligand binding results in receptor dimer activation of one or two G-proteins when measured using functional assays in an intact cell system.
In this study, radioligand binding, inositol phosphate (IP) signaling, confocal imaging of receptor endocytosis, and fluorescence resonance energy transfer (FRET) were evaluated in HEK293 cells coexpressing wild-type and inactive mutant S138R 5-HT 2C receptors to determine the effect of dimerization on receptor function and to determine the ligand/dimer/G-protein stoichiometry. The results of this study show that S138R receptors form heterodimers with wild-type 5-HT 2C receptors in living cells and have an inhibitory effect on 5-HT 2C receptor function. Previous studies have reported a dominant-negative effect of splice variant, truncated, or mutant receptors on wild-type receptor function through heterodimerization, resulting in trapping of wild-type receptors in the endoplasmic reticulum (13)(14)(15)(16), or through sequestration of G-proteins (17). In contrast, our study describes a dominant- and RR017926 (to J. E. M.). 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. 1  negative effect of an inactive mutant GPCR on wild-type receptor function through heterodimerization, resulting in the expression of nonfunctional receptor complexes on the plasma membrane. Although several studies have reported attenuation of receptor function following coexpression of different GPCR subtypes (18 -20), the ligand/dimer/Gprotein stoichiometry was not evaluated. Other studies using isolated membranes (21) and purified receptors (22) suggest that each GPCR dimer interacts with a single G-protein. We tested this model in intact living cells following coexpression of wild-type and inactive 5-HT 2C receptors. Our results suggest that each GPCR dimer binds two molecules of ligand, resulting in the activation of one G-protein, and indicate that dimerization is essential for 5-HT receptor function.

MATERIALS AND METHODS
Cell Culture-HEK293 cells (American Type Culture Collection) were cultured in Dulbecco's modified Eagle's medium (DMEM; Cell-Gro) with 10% fetal bovine serum (CellGro) at 37°C and 5% CO 2 . HEK293 cells were plated at 5 ϫ 10 6 cells/100-mm dish 24 h prior to transfection. Cells were transfected with the indicated plasmid DNAs using Lipofectamine reagent (Invitrogen) for 5 h at 37°C in serum-free medium and then returned to DMEM with serum. All experiments were performed 36 -42 h post-transfection, and cells were cultured in serum-free medium for the final 16 h prior to the experiment.
cDNAs-The human 5-HT 2C-VSV receptor cDNA (from Dr. Elaine Sanders-Bush) was used as the starting template for creating receptor fusion proteins and for site-directed mutagenesis to create the mutant S138R receptor. The S138R mutation in the 5-HT 2C receptor was created by PCR following the Stratagene protocol using complementary sense and antisense primers to nucleotides 404 -427 of the human 5-HT 2C receptor open reading frame, with the exception of nucleotides 413-415, which were changed to CGT to code for arginine instead of serine. The M1 muscarinic receptor was PCR-amplified from genomic DNA. Cyan fluorescent protein (CFP), yellow fluorescent protein (YFP), and hemagglutinin (HA) tags were attached to the C-terminal end of S138R and wild-type 5-HT 2C receptor cDNAs to create 5-HT 2C /CFP, 5-HT 2C /YFP, and 5-HT 2C /HA fusion proteins as described previously (12). S138R and wild-type 5-HT 2C receptor cDNAs were ligated into the pECFP-N1 and/or pEYFP-N1 vector (Clontech), and the M1 muscarinic receptor was ligated into pcDNA3 (Clontech). All constructs were confirmed by DNA sequencing (Center for Functional Genomics, Albany, NY). S138R 5-HT 2C /YFP-expressing Stable Cell Line-A stable cell line expressing S138R 5-HT 2C /YFP was created by transfecting 2 ϫ 10 6 HEK293 cells with 10 g of S138R 5-HT 2C /YFP plasmid DNA by calcium phosphate precipitation. Two days post-transfection, cells were cultured in DMEM containing 400 g/ml G418 for 2 weeks. Colonies expressing YFP-tagged S138R receptors on the plasma membrane were identified by confocal fluorescence microscopy.
Radioligand Binding-HEK293 cells (4 ϫ 10 6 cells/100-mm dish) were transfected with 0.5 g of S138R or wild-type 5-HT 2C receptor plasmid DNA using 20 l of Lipofectamine. For saturation experiments, HEK293 and S138R-expressing stable cells were transfected with 0.2 or 0.5 g of wild-type 5-HT 2C receptor plasmid DNA. Membranes were prepared from transfected cells as described previously (23). Saturation experiments were performed using [ 3 H]mesulergine (0.1-5 nM) in the presence and absence of 1 M mianserin in 0.5 ml of buffer (50 mM Tris-HCl, 10 mM MgSO 4 , 0.5 mM EDTA, and 0.1% ascorbate) with 0.5 ml of diluted cell membranes (10 g of protein/assay tube). The mixture was incubated at 37°C for 30 min, followed by filtration through glass fiber filters (Schleicher & Schüll) on a Brandel cell harvester and liquid scintillation counting in Ecoscint (National Diagnostics). Protein was determined using BCA (Pierce). Data were analyzed using GraphPad Prism software.
IP Assay-For IP assays, HEK293 or S138R-expressing stable cells were seeded at 2 ϫ 10 5 cells/well in 24-well plates (Biocoat, Inc.) and transfected 3 h later with 5-HT 2C or M1 muscarinic receptor plasmid DNA (25 ng/well) and Lipofectamine (1 l/well). For IP assays performed in parallel with the radioligand binding assays, 4 ϫ 10 6 cells/ 100-mm dish were transfected with 0.2 g of 5-HT 2C receptor plasmid DNA: half the dish was used for the radioligand binding assay, and the other half was replated at 2 ϫ 10 5 cells/well in 24-well plates for the IP assay. Twenty-four hours post-transfection, 0.5 Ci of [ 3 H]myoinositol in inositol-free, serum-free DMEM was added to each well for an additional 16 h. Cells were washed with phosphate-buffered saline and incubated in serum-free DMEM with lithium chloride in the absence or presence of agonist for 35 min. Total [ 3 H]IP production was measured by anion exchange chromatography as described previously (24). Data were analyzed using GraphPad Prism software.
Confocal Microscopy-HEK293 cells (4 ϫ 10 6 cells/100-mm dish) were transfected with 0.2 g of wild-type or S138R 5-HT 2C /YFP plasmid DNA using 20 l of Lipofectamine. Twenty-four hours post-transfection, cells were plated overnight in serum-free medium on polylysine-coated glass coverslips. The cells were viewed live in phosphatebuffered saline on a Noran Oz confocal microscope system using a Nikon Diaphot 200 inverted microscope with a ϫ60, 1.4-numerical aperture oil immersion objective. The 488-nm laser line from a krypton/ argon laser was used to excite YFP, and fluorescence was imaged with a 488-nm dichroic mirror and a 500-nm long-pass filter. Time-lapse images were collected every 15 s over a 10-min period. After 10 min, 10 M 5-HT was added to the cells, and images were collected every 15 s for 10 min. HEK293 cells coexpressing wild-type 5-HT 2C /CFP and S138R 5-HT 2C /YFP were viewed using a Zeiss LSM 510 META confocal imaging system with linear unmixing of CFP and YFP emission spectra following excitation at 458 nm using the same conditions and settings as described for the FRET experiments.
Immunoprecipitation and Western Blotting-HEK293 cells (4 ϫ 10 6 cells/100-mm dish) were transfected with 1 g of the indicated plasmid DNAs using 20 l of Lipofectamine. For control experiments, cells were transfected independently with either 5-HT 2C /YFP or 5-HT 2C /HA and mixed together after transfection. Cross-linking of receptors was performed by incubating intact cells in 5 mM bis(sulfosuccinimidyl) suberate (a cell membrane-impermeable cross-linker; Pierce) for 30 min at 37°C and 5% CO 2 . Cells were lysed, and membranes were solubilized in 10 mM CHAPS as described previously (12). Solubilized membrane proteins were immunoprecipitated overnight with 10 l of anti-HA (Y-11) antibody-agarose (Santa Cruz Biotechnology, Inc.) as described previously (12). Pellets were resuspended in 50 l of 1ϫ nonreducing Laemmli sample buffer and heated at 70°C for 10 min, and 10-l samples were run on a 10% Tris-HCl Ready Gel (Bio-Rad). Blots were probed with horseradish peroxidase-conjugated anti-green fluorescent protein (B-2) antibody (Santa Cruz Biotechnology, Inc.) diluted 1:3000 and visualized by enhanced chemiluminescence (Amersham Biosciences) according to the manufacturer's protocol.
FRET-HEK293 cells (4 ϫ 10 6 cells/100-mm dish) were cotransfected with S138R 5-HT 2C /CFP (donor) and wild-type 5-HT 2C /YFP (acceptor) or with wild-type 5-HT 2C /CFP (donor) and wild-type 5-HT 2C /YFP (acceptor) at a donor/acceptor ratio of 1:1 (0.4 g ϩ 0.4 g of plasmid DNA) or 1:2 (0.3 g ϩ 0.6 g of plasmid DNA) using 20 l of Lipofectamine. Twenty-four hours post-transfection, cells were plated overnight in serum-free medium on polylysine-coated glass coverslips prior to the FRET assay. The cells were viewed live in phosphate-buffered saline at room temperature using a Zeiss LSM 510 META confocal imaging system with a 30-milliwatt argon laser and a ϫ63, 1.4-numerical aperture oil immersion objective at ϫ2 zoom. CFP and YFP emission spectra were collected (from cells expressing 5-HT 2C /CFP or 5-HT 2C /YFP alone) following excitation at 458 nm and were used as reference spectra for linear unmixing of CFP and YFP emission spectra in cotransfected cells using the Zeiss META detector and Zeiss AIM software as described previously (12). FRET was measured by acceptor photobleaching (25). Confocal microscopy was used to visualize a 2-m optical slice through the middle of a living HEK293 cell expressing both 5-HT 2C /CFP and 5-HT 2C /YFP. A small region of plasma membrane (typically 3-4 m in length) was selected for photobleaching. Prebleach CFP and YFP images were collected simultaneously following excitation at 458 nm (38% laser intensity, detector gain ϭ 740). The selected region of plasma membrane was irradiated with the 514-nm laser line (100% intensity, 60 iterations, using a 458/514-nm dual dichroic mirror) for 10 s to photobleach YFP, typically resulting in a 95% decrease in YFP fluorescence. Post-bleach CFP and YFP images were collected simultaneously (at 458 nm) within 1 s following photobleaching. In an independent experiment, YFP fluorescence recovery after photobleaching was measured and found to be negligible 1 s following photobleaching. The short photobleach time and the automated rapid collection of pre-and post-bleach images make this technique suitable for use in living cells. FRET was measured as an increase in CFP fluorescence intensity (donor dequenching) following YFP (acceptor) photobleaching. FRET efficiency was calculated as 100 ϫ ((CFP postbleach Ϫ CFP pre-bleach)/CFP post-bleach) using the FRET macro in the Zeiss AIM software package, taking into account CFP and YFP background noise in each channel. FRET efficiencies were calculated in cells expressing both CFP and YFP, in cells expressing CFP alone, and in non-bleached regions of the cell membrane as a control.

Mutation of Ser 138 to Arg Produces an Inactive Form of the 5-HT 2C
Receptor-Site-directed mutagenesis was used to create an inactive form of the 5-HT 2C receptor to test the hypothesis that 5-HT 2C receptors function as dimeric or oligomeric complexes. Computer modeling and mutagenesis studies have provided evidence that Ser 159 in transmembrane domain (TMD) III plays an important role in 5-HT binding to the 5-HT 2A receptor (26). Therefore, we mutated the analogous serine (amino acid 138) in the 5-HT 2C receptor to arginine (S138R) in an attempt to create an inactive receptor. [ 3 H]Mesulergine binding and IP production assays were used to evaluate receptor expression and function in HEK293 cells. Specific [ 3 H]mesulergine binding was detected in cells transfected with plasmid containing wild-type 5-HT 2C receptor cDNA, but not in cells transfected with plasmid containing S138R 5-HT 2C receptor cDNA (Fig. 1). Similarly, 5-HT stimulated [ 3 H]IP production in HEK293 cells transfected with plasmid containing wild-type 5-HT 2C receptor cDNA, but not in cells transfected with plasmid containing S138R 5-HT 2C receptor cDNA (Fig. 1). Although wild-type 5-HT 2C receptors had moderate levels of basal activity in the IP assay, S138R receptors were devoid of basal activity. These results demonstrate that mutation of Ser 138 to Arg eliminates the ability of the 5-HT 2C receptor to bind ligand and to stimulate phospholipase C.
Plasma Membrane Localization of S138R and Wild-type 5-HT 2C Receptors-Fluorescently tagged 5-HT 2C receptors were created to allow direct visualization of S138R and wild-type 5-HT 2C receptors in living cells by confocal microscopy. A previous study has shown that a fluorescent tag (YFP) expressed as a fusion protein on the C terminus of the 5-HT 2C receptor does not alter 5-HT binding affinity or potency (12). S138R or wild-type 5-HT 2C /YFP fusion protein was expressed in HEK293 cells, and live time-lapse images were collected (Fig. 2

, A-D).
To assess receptor trafficking activity subsequent to delivery to the plasma membrane, receptor movement was monitored in the absence and presence of 5-HT. Time-lapse images were captured every 15 s over a 10-min period. In cells expressing wild-type 5-HT 2C receptors, significant trafficking was observed between the plasma membrane and intracellular compartments in the form of bidirectional vesicular movement. Selected frames from a representative time-lapse movie are illustrated in Fig. 2A. Within the first minute following the addition of 5-HT, vesicular trafficking increased, with a net increase in the retrograde movement of receptor-containing vesicles as evidenced by in an increase in the fluorescence intensity of the intracellular compartment. This was followed by an overall cellular shape change that was clearly evident 3 min after the addition of 5-HT (Fig. 2B). The increase in intracellular fluorescence intensity suggests the induction of down-regulation of ligand-bound receptor in the form of endocytosis. HEK293 cells expressing S138R 5-HT 2C /YFP showed little or no observable receptor trafficking in the absence of 5-HT (Fig. 2C). Overall movement of the cell was observed, as would be expected from a living cell over a 10-min period. However, there was no significant receptor endocytosis subsequent to the addition 5-HT (Fig. 2D). When cells were cotransfected with wild-type 5-HT 2C /CFP and S138R 5-HT 2C /YFP, both wild-type and mutant receptors were expressed on the plasma membrane (Fig.  2E). The addition of 10 M 5-HT produced a small increase in intracellular wild-type 5-HT 2C /CFP fluorescence (Fig. 2F), in contrast to the large redistribution of wild-type 5-HT 2C receptors observed when expressed in the absence of S138R receptors (Fig. 2B). There was no change in the distribution of S138R 5-HT 2C /YFP fluorescence following the addition of 5-HT to the cotransfected cells (Fig. 2F). These results suggest that the ability of wild-type 5-HT 2C receptors to respond to 5-HT stimulation is impaired when coexpressed with inactive S138R 5-HT 2C receptors. S138R 5-HT 2C Receptors Inhibit the Function of Wild-type 5-HT 2C Receptors-Wild-type 5-HT 2C receptors were expressed in HEK293 cells and in a stable cell line expressing S138R 5-HT 2C receptors to determine whether wild-type 5-HT 2C receptor function is altered when expressed in the presence of inactive S138R receptors. 5-HT dose-response curves for stimulation of IP production in HEK293 cells and in the S138R-expressing stable cell line transfected with equal amounts of wild-type 5-HT 2C receptor plasmid DNA are shown in Fig. 3A. Although the 5-HT EC 50 values for stimulation of IP production by wild-type 5-HT 2C receptors expressed in HEK cells (1.3 Ϯ 0.2 nM) and in S138R-expressing cells (2.3 Ϯ 0.4 nM) were similar, 5-HT-stimulated IP production was drastically reduced in the S138R-expressing cells. Radioligand binding studies were performed to measure wild-type 5-HT 2C receptor expression following transfection into the S138R-expressing stable cell line using the same transfection conditions as described for the IP assay. Additional radioligand binding studies were performed to compare wild-type 5-HT 2C receptor expression levels in HEK293 cells and in the S138R-expressing stable cell line. Four saturation experiments were performed in parallel: two with 0.2 g and two with 0.5 g of wild-type 5-HT 2C receptor plasmid DNA transfected into HEK293 cells or into the S138R-expressing stable cell line (Fig. 3B). B max values for 0.2 g of wild-type 5-HT 2C receptor plasmid DNA transfected into HEK293 and S138R-expressing stable cells were 4.9 and 4.1 pmol/mg of protein, respectively. B max values for 0.5 g of native 5-HT 2C receptor plasmid DNA transfected into HEK293 and S138R-expressing stable cells were 7.9 and 9.5 pmol/mg of protein, respectively. Because S138R receptors do not bind radioligand, the plasma membrane expression level of S138R receptors in the S138R 5-HT 2C /YFP-expressing stable cell line was determined by fluorescence microscopy. Plasma membrane YFP fluorescence intensity in the S138R-expressing stable cell line (1573 Ϯ 52, n ϭ 64) was similar to that in the wild-type 5-HT 2C /YFP-expressing stable cell line (1665 Ϯ 50, n ϭ 71) with a predetermined [ 3 H]mesulergine B max value of 33 pmol/mg of protein.
When wild-type 5-HT 2C receptor plasmid DNA (0.2 g) was transfected into the S138R-expressing stable cell line, the average wild-type 5-HT 2C receptor expression level was 4.5 pmol/mg of protein, equivalent to an approximate B max value of 15 pmol/mg of protein at a typical transfection efficiency of 30%. This would yield an approximate plasma membrane receptor expression ratio of 1:2 for wild-type/S138R 5-HT 2C receptors (following transfection of wild-type 5-HT 2C receptor plasmid into the S138R-expressing stable cell line) and would be expected to result in ϳ80% of the wild-type 5-HT 2C receptors forming heterodimers with S318R receptors. This is consistent with the observed 70 -80% decrease in 5-HT-stimulated [ 3 H]IP production for native receptors expressed in the S138R-expressing stable cell line (Figs. 3A and 4A).
When expressed in HEK293 cells, wild-type 5-HT 2C receptors displayed moderate levels of basal and 5-HT-stimulated IP production (Fig. 4A). However, when wild-type 5-HT 2C receptors were expressed in the S138R-expressing stable cell line, basal IP production was eliminated, and the maximal 5-HT-stimulated IP response was dramatically reduced. In contrast, both basal and carbachol-stimulated IP production were similar in HEK293 and S138R-expressing stable cells following transfection with M1 muscarinic receptor plasmid (Fig. 4B). These results indicate that the reduced wild-type 5-HT 2C receptor signaling observed in the S138R-expressing stable cell line is not due to decreased native receptor expression, abnormal function of the G␣ q proteins endogenously expressed in the S138R-expressing stable cell line, or sequestration of G␣ q proteins through binding to inactive S138R receptors.   DECEMBER 2, 2005 • VOLUME 280 • NUMBER 48

JOURNAL OF BIOLOGICAL CHEMISTRY 40147
Co-immunoprecipitation of S138R and Wild-type 5-HT 2C Receptors-The results of the coexpression studies raised the possibility that there is an interaction between S138R and wild-type 5-HT 2C receptors when they are coexpressed in the same cell. Co-immunoprecipitation and Western blot studies were performed to determine whether S138R 5-HT 2C receptors associate with wild-type 5-HT 2C receptors. HEK293 cells were transfected with 5-HT 2C /HA and/or 5-HT 2C /YFP plasmid DNA as indicated in Fig. 5. Previously, we demonstrated that 5-HT 2C receptor homodimers are detergent-sensitive (12). Therefore, in these experiments, intact cells were pretreated with the bis(sulfosuccinimidyl) suberate cross-linker prior to membrane solubilization and immunoprecipitation with anti-HA antibody. When the HA-immunoprecipitated proteins were separated by PAGE, YFP immunoreactive bands were not identified in samples prepared from cells individually expressing either 5-HT 2C /HA or 5-HT 2C /YFP, even when the transfected cells were mixed together prior to solubilization and immunoprecipitation. However, YFP immunoreactive bands the approximate predicted size of 5-HT 2C dimers were the predominant species identified in samples prepared from cells cotransfected with wild-type 5-HT 2C /HA and 5-HT 2C / YFP or with wild-type 5-HT 2C /HA and S138R 5-HT 2C /YFP plasmid DNAs.
FRET between S138R and Wild-type 5-HT 2C Receptors-FRET experiments were performed to determine whether S138R 5-HT 2C receptors form heterodimers with wild-type 5-HT 2C receptors on the plasma membrane of living cells. FRET operates on the principle that, if two fluorescent proteins (donor and acceptor) with overlapping emission and excitation spectra are within 100 Å of each other and if their dipoles are oriented appropriately, energy emitted by the donor (in its excited state) will be transferred to the acceptor (27), resulting in acceptor excitation and quenching of donor fluorescence. When the acceptor is removed by photobleaching, the donor is dequenched, and an increase in donor fluorescence is observed (25). In this study, S138R 5-HT 2C / CFP was used as the donor, and wild-type 5-HT 2C /YFP was used as the acceptor. A pre-bleach image of a living HEK293 cell coexpressing S138R 5-HT 2C /CFP and wild-type 5-HT 2C /YFP was captured (Fig. 6, A,  C, and E). The selected region of plasma membrane (indicated by white boxes) was photobleached, and a post-bleach image was collected (Fig. 6,  B, D, and F). Acceptor photobleaching resulted in a FRET efficiency of 28.6% (Fig. 6B) and a 95% decrease in acceptor fluorescence (Fig. 6D).
Acceptor photobleaching experiments were performed on selected regions of plasma membrane in cells coexpressing S138R 5-HT 2C /CFP  . Co-immunoprecipitation of wild-type and S138R 5-HT 2C receptors. HEK293 cells were transfected with wild-type 5-HT 2C /HA (2C/HA) or 5-HT 2C /YFP (2C/YFP) cDNA or with S138R 5-HT 2C /YFP cDNA (S138R/YFP) as indicated. Cells were pretreated with the bis(sulfosuccinimidyl) suberate cross-linker prior to membrane solubilization in CHAPS. All samples were immunoprecipitated (Immppt) with anti-HA antibody, and the blot was probed with horseradish peroxidase-conjugated anti-green fluorescent protein (GFP) antibody. FIGURE 6. Confocal microscopy was used to visualize a 2-m-thick optical section through the middle of a living HEK293 cell coexpressing S138R 5-HT 2C /CFP (S138R/ CFP; donor) and wild-type 5-HT 2C /YFP (2C/YFP; acceptor). CFP fluorescence is shown in green, and YFP in red. CFP and YFP pre-bleach images were collected simultaneously following excitation at 458 nm (A, C, and E). A region of plasma membrane (indicated by white boxes) was photobleached at 514 nm for 10 s. CFP and YFP post-bleach images were collected simultaneously following excitation at 458 nm (B, D, and F). Plasma membrane CFP fluorescence increased from 1000 to 1400 arbitrary fluorescence intensity units following YFP photobleaching (B and F), producing a FRET efficiency of 28.6%. and wild-type 5-HT 2C /YFP or wild-type 5-HT 2C /CFP and wild-type 5-HT 2C /YFP. The mean FRET efficiencies were similar for both groups of cells (TABLE ONE). The plasma membrane fluorescence of wildtype 5-HT 2C /YFP was similar regardless of whether it was coexpressed with wild-type or S138R 5-HT 2C /CFP, indicating that coexpression with S138R does not influence the amount of wild-type 5-HT 2C receptor expressed on the plasma membrane (TABLE ONE). Fig. 7 (A and B) shows the relationship between FRET efficiency and the donor/acceptor ratio (unquenched post-bleach donor fluorescence divided by prebleach acceptor fluorescence (uD/A)). FRET efficiency was dependent on the uD/A ratio, but was independent of the receptor expression level as measured by acceptor fluorescence (Fig. 7C).
Because the FRET experiments were preformed in living cells, FRET was measured in cells expressing wild-type 5-HT 2C /CFP alone to determine whether receptor migration within the plasma membrane during the 10-s photobleaching period could give rise to a false-positive FRET signal. 5-HT 2C /CFP pre-and post-bleach plasma membrane fluorescence intensities were measured in 14 living cells, and the mean FRET efficiency was calculated in the same manner as described for cells coexpressing CFP-and YFP-tagged receptors. For cells expressing only 5-HT 2C /CFP, the mean FRET efficiency was Ϫ1.4 Ϯ 2.2%, indicating that receptor migration within the plasma membrane during the short photobleach period does not give rise to a false-positive FRET signal.

DISCUSSION
This study was performed to determine whether dimerization plays a functional role in regulating the activity of 5-HT receptors and to determine the ligand/dimer/G-protein stoichiometry. We addressed these issues by examining the function of wild-type 5-HT 2C receptors when coexpressed with an inactive form of the receptor. Mutation of Ser 138 to Arg eliminated [ 3 H]mesulergine binding and IP production. Time-lapse confocal fluorescence imaging of HEK293 cells expressing S138R receptors fused to YFP revealed a predominantly plasma membrane localization. In contrast to wild-type 5-HT 2C receptors, the S138R receptors displayed minimal constitutive receptor cycling, and 5-HT did not promote internalization of the S138R receptors. When 5-HT was added to cells coexpressing S138R and wild-type 5-HT 2C receptors, intracellular trafficking of wild-type receptors was diminished, and S138R receptors remained on the plasma membrane. These results suggest that, in the presence of S138R receptors, wild-type 5-HT 2C receptors lose the ability to respond to 5-HT stimulation and appear to be trapped on the plasma membrane.
IP production was measured in cells coexpressing S138R and wildtype 5-HT 2C receptors to test the hypothesis that wild-type 5-HT 2C receptor function is compromised in the presence of S138R receptors. When wild-type 5-HT 2C receptors were expressed in the S138R-expressing stable cell line, both basal and 5-HT-stimulated IP production were greatly reduced, with little change in 5-HT potency. Coexpression of S138R and wild-type 5-HT 2C receptors did not decrease wild-type receptor expression on the plasma membrane, as demonstrated by confocal fluorescence microscopy and radioligand binding, indicating that the decrease in IP production observed for wild-type receptors following expression in the S138R-expressing cell line is not due to a decrease in the amount of wild-type receptor expressed on the plasma membrane. In addition, M1 muscarinic receptors exhibited normal IP signaling in the S138R-expressing cell line. These results indicate that G␣ q proteins in the S138R-expressing cell line function normally and that the mutant S138R receptors do not sequester G␣ q proteins, in contrast to studies with mutant ␣ 1B -adrenergic receptors that exhibit a dominant-negative effect on native receptor function through G-protein sequestration (17).
FRET (measured by acceptor photobleaching) demonstrated coex-   DECEMBER 2, 2005 • VOLUME 280 • NUMBER 48 pression of both S138R and wild-type 5-HT 2C receptors in close proximity on the plasma membrane of living cells. Laser scanning confocal microscopy allowed the photobleaching to be confined to a small region of plasma membrane, thereby minimizing the time required for photobleaching and making the technique suitable for living cells. A positive FRET signal resulting from random proximity (overexpression) of donor and acceptor has been reported to be dependent on the amount of acceptor expressed on the plasma membrane, whereas FRET produced by clustered proteins (such as dimers or oligomers) is predicted to be independent of acceptor expression levels and dependent on the uD/A ratio (28,29). FRET efficiencies measured in cells coexpressing S138R and wild-type 5-HT 2C receptors were independent of the acceptor expression level and dependent on the uD/A ratio, as predicted for receptors in a clustered distribution. These results indicate that S138R receptors form heterodimers with wild-type 5-HT 2C receptors when they are coexpressed in the same cell.

Ligand/Dimer/G-protein Stoichiometry
The results of the immunoprecipitation and FRET studies suggest that S138R receptors decrease the function of wild-type 5-HT 2C receptors by forming inactive heterodimers that are expressed on the plasma membrane. As a result, the effective concentration of active wild-type receptor homodimers is reduced, resulting in decreased IP signaling. Therefore, heterodimerization between inactive and active GPCRs can have a dominant-negative effect on receptor function when the heterodimers are expressed on the plasma membrane. This is in contrast to previous studies demonstrating a dominant-negative effect of inactive splice variants or mutant receptors on wild-type receptor function through heterodimerization, resulting in trapping of the wild-type receptors in the endoplasmic reticulum (13)(14)(15)(16). Although our study examined S138R and wild-type 5-HT 2C receptor interactions only on the plasma membrane, it is possible that the heterodimers form in the endoplasmic reticulum and are transported to the plasma membrane, as described previously for other GPCRs (16,30).
GPCR activation has been reported to involve the coordinated movements of TMDs III and VI (17,(31)(32)(33). The S138R mutation is located in TMD III of the 5-HT 2C receptor. As a result, this mutation may directly interfere with the ability of TMD III to adopt the proper conformation to achieve an active state of the receptor. This is supported by the observation that the S138R mutation eliminates receptor basal activity. This mutation does not appear to alter the ability of 5-HT 2C receptors to dimerize, as similar FRET efficiencies were observed for homodimerization of wild-type receptors and heterodimerization between S138R and wild-type receptors. Presently, the site(s) of interaction between 5-HT receptor dimers/oligomers remains unknown. Previous studies using cysteine cross-linking and mutagenesis have provided evidence for the involvement of TMD IV in the formation of D 2 -dopamine receptor homodimers (34,35), and atomic force microscopy has provided evidence for rhodopsin receptor homodimer interfaces at TMDs IV and V (21,36). Studies of the ␣ 1b -adrenergic receptor suggest that the homodimerization interface for this receptor may involve TMDs I and IV (37,38). FRET experiments performed using chimeric ␣ 1b /␤ 2 -adrenergic receptors suggest that TMD III does not play a role in ␣ 1b -adrenergic receptor homodimerization (37). These findings are consistent with the results of our study, in which the S138R mutation in TMD III did not appear to alter 5-HT 2C receptor dimerization.
In our experiments, the S138R receptor was incapable of adopting the appropriate active conformation necessary to generate basal or agoniststimulated IP production. Our observation that coexpression of S138R and wild-type 5-HT 2C receptors eliminated basal activity and greatly diminished agonist-stimulated IP production is consistent with a model in which a single G-protein interacts with one GPCR dimer (Fig. 8). In this model, the S138R receptor in the wild-type/S138R 5-HT 2C receptor heterodimer is not capable of forming the complementary G-protein interaction required for heterodimer activation of a single G-protein.
Thus, the heterodimer remains silent with respect to G-protein activation and IP signaling.
If two G-proteins bound to each dimer (in other words, one G-protein/receptor monomer), then equal wild-type 5-HT 2C receptor expression levels in the HEK293 and S138R-expressing stable cells would be predicted to result in equal IP production. However, IP signaling was 70 -80% lower when wild-type 5-HT 2C receptors were expressed in the S138R-expressing stable cell line. Expression of wild-type 5-HT 2C receptors in the S138R-expressing stable cell line resulted in a plasma membrane receptor expression ratio of 1:2 for wild-type/S138R 5-HT 2C receptors, predicting that ϳ80% of the wild-type 5-HT 2C receptors would be engaged as heterodimers with S138R receptors. This corresponds to the 70 -80% decrease in IP signaling observed when wild-type 5-HT 2C receptors were expressed in the S138R-expressing stable cell line, consistent with the wild-type/S138R 5-HT 2C receptor heterodimers being inactive with respect to G-protein coupling and signaling. If the wild-type/S138R 5-HT 2C receptor heterodimers exhibited impaired G-protein coupling or activation, then the 5-HT dose-response curve for IP production would be shifted to the right, with 5-HT displaying lower potency upon expression of wild-type 5-HT 2C receptors in the S138R-expressing stable cell line. However, 5-HT potency for IP stimulation was the same when wild-type 5-HT 2C receptors were expressed in HEK293 cells or in the S138R-expressing stable cell line. These results indicate that the heterodimers are inactive with respect to G-protein coupling and signaling and are consistent with the model in Fig. 8, showing a receptor dimer/G-protein stoichiometry of 1:1.
This model is consistent with studies based on the crystalline structure of rhodopsin and atomic force microscopy studies suggesting that rhodopsin receptor dimerization may be necessary to form a G-proteinbinding site that is large enough to accommodate a single heterotrimeric G-protein (21,36). In addition, studies using chemical cross-linking and purified leukotriene B 4 receptors indicate that one G-protein associates with each leukotriene B 4 receptor homodimer (22). These results are supported by functional studies showing that D 1 /D 2 -dopamine receptor heterodimers can form a novel signaling complex capable of stimulating a different G␣ protein than D 1 -or D 2 -dopamine receptor homodimers (39).
Given that the S138R receptor is incapable of binding ligand, if wildtype/S138R 5-HT 2C receptor heterodimers formed a single binding pocket for one molecule of ligand, it would be predicted that the heterodimer would have a lower affinity for ligand or would be incapable of binding ligand, and a significant decrease in K D and/or B max would be predicted for wild-type 5-HT 2C receptors expressed in the S138R-expressing stable cell line. However, radioligand binding studies revealed similar [ 3 H]mesulergine K D and B max values for wild-type 5-HT 2C receptors expressed in HEK293 cells and in the S138R-expressing stable cell line. These results suggest that a GPCR heterodimer contains two separate ligand-binding pockets, consistent with a model in which ligand binds to both receptors in the dimer, as illustrated in Fig. 8.
In conclusion, Ser 138 plays a critical role in ligand binding and the ability of the 5-HT 2C receptor to adopt the active conformation required for G-protein activation. Plasma membrane expression of an inactive mutant receptor with wild-type receptors can have a dominantnegative effect on the function of the wild-type receptors, indicating that dimerization is essential for 5-HT receptor function. These results are consistent with a model in which one GPCR dimer binds two molecules of ligand, resulting in the binding and activation of one G-protein. Studies with the inactive S138R receptor suggest that dimerization may be independent of the active state of the 5-HT 2C receptor. Additional studies comparing dimerization of 5-HT 2C receptors with high basal activity and no basal activity (INI and VGV isoforms, respectively) (23) are in progress to determine the relationship between dimer formation and the active state of the 5-HT 2C receptor.