Homo- and Heterodimerization of Somatostatin Receptor Subtypes

Several recent studies suggest that G protein-coupled receptors can assemble as heterodimers or hetero-oligomers with enhanced functional activity. However, inactivation of a fully functional receptor by heterodimerization has not been documented. Here we show that the somatostatin receptor (sst) subtypes sst2A and sst3 exist as homodimers at the plasma membrane when expressed in human embryonic kidney 293 cells. Moreover, in coimmunoprecipitation studies using differentially epitope-tagged receptors, we provide direct evidence for heterodimerization of sst2A and sst3. The sst2A-sst3 heterodimer exhibited high affinity binding to somatostatin-14 and the sst2-selective ligand L-779,976 but not to the sst3-selective ligand L-796,778. Like the sst2A homodimer, the sst2A-sst3 heterodimer stimulated guanosine 5′-3-O-(thio)triphosphate (GTPγS) binding, inhibition of adenylyl cyclase, and activation of extracellular signal-regulated kinases after exposure to the sst2-selective ligand L-779,976. However, unlike the sst3 homodimer, the sst2A-sst3 heterodimer did not promote GTPγS binding, adenylyl cyclase inhibition, or extracellular signal-regulated kinase activation in the presence of the sst3-selective ligand L-796,778. Interestingly, during prolonged somatostatin-14 exposure, the sst2A-sst3 heterodimer desensitized at a slower rate than the sst2A and sst3 homodimers. Both sst2A and sst3 homodimers underwent agonist-induced endocytosis in the presence of somatostatin-14. In contrast, the sst2A-sst3 heterodimer separated at the plasma membrane, and only sst2A but not sst3 underwent agonist-induced endocytosis after exposure to somatostatin-14. Together, heterodimerization of sst2A and sst3results in a new receptor with a pharmacological and functional profile resembling that of the sst2A receptor, however with a greater resistance to agonist-induced desensitization. Thus, inactivation of sst3 receptor function by heterodimerization with sst2A or possibly other G protein-coupled receptors may explain some of the difficulties in detecting sst3-specific binding and signaling in mammalian tissues.

Although G protein-coupled receptors (GPCRs) 1 generally were believed to act as monomeric entities, a growing body of evidence suggests that they may form functionally relevant dimers. The existence of homodimers has been shown for several GPCRs including the ␤ 2 -adrenergic receptor (1), ␦and -opioid receptors (2,3), the metabotrobic glutamate receptor 5 (4), the calcium-sensing receptor (5), the m3 muscarinic receptor (6), and dopamine receptors (7). GPCRs seem to dimerize via different mechanisms. Whereas dimerization of the ␤ 2adrenergic receptor (1) and D2 dopamine receptor (8) occurs via their transmembrane regions, dimerization of the ␦-opioid receptor involves the carboxyl terminus (2). In contrast, the metabotrobic glutamate receptor 5 (4) and the calcium-sensing receptor (5,9,10) appear to be disulfide-linked dimers, and dimerization occurs via their large amino termini. The question of to what extent agonist binding affects dimerization remains controversial. Recent evidence obtained in living cells using bioluminescence resonance energy transfer suggests that the ␤ 2 -adrenergic receptor exists at the cell surface as a constitutive dimer that is stabilized by agonist binding (11). In contrast, agonist stimulation reduced the level of ␦-opioid receptor dimers suggesting that monomerization precedes agonist-induced internalization of this receptor (2). Biochemical and functional studies suggest that GPCRs can also assemble as heterodimers with enhanced functional activity (3,(12)(13)(14)(15)(16)(17)(18)(19)(20). Formation of heterodimers between two nonfunctional ␥-aminobutyric acid (GABA) receptors, GABA B R1 and GABA B R2, was necessary for a fully functional GABA B receptor (14 -19). ␦and -Opioid receptors form heterodimers with ligand binding and signaling properties resembling that of the 2 receptor (3). Finally, the somatostatin receptor (sst) sst 5 and the D 2 dopamine receptor heterodimerize to form a new receptor with enhanced activity (20).
The neuropeptide somatostatin (SS-14) is widely expressed throughout the central nervous system and periphery. SS-14 is involved in multiple functions including endocrine and exocrine hormone release, cognition, sleep, and motor activity. In addition, peptide derivatives of somatostatin have successfully been used in the treatment of neuroendocrine malignancies. The actions of SS-14 are mediated via five distinct somatostatin receptor subtypes, termed sst 1 -sst 5 , which belong to the superfamily of GPCRs. All somatostatin receptors bind SS-14 with high affinity and inhibit adenylyl cyclase. There is also evidence for different, although not mutually exclusive, pathways of intracellular signaling of somatostatin receptor subtypes, e.g. activation of extracellular signal-regulated kinases (ERK) via sst 1 , sst 3 , and sst 4 ; activation of phosphotyrosine phosphatases via sst 1 , sst 2 , and sst 3 ; activation of phospholipase A2 via sst 4 ; and modulation of K ϩ channels via sst 2 (21,22). Individual target cells often express more than one somatostatin receptor, raising the possibility that the functional diversity of these receptors could be expanded by heterodimerization among somatostatin receptor subtypes.
Recently, Rocheville et al. (23) provided evidence, based on photobleaching fluorescence resonance energy transfer, for homodimerization of the sst 5 somatostatin receptor. The sst 5 receptor also appears to form heterodimers with sst 1 but not sst 4 . In this study, we have examined dimerization of sst 2A and sst 3 somatostatin receptors. In coimmunoprecipitation studies using differentially epitope-tagged receptors, we show that sst 2A and sst 3 associate as dimers, both as homodimers and heterodimers. sst 2A -sst 3 heterodimerization resulted in a new receptor with enhanced sst 2A -like and diminished sst 3 -like activity.

EXPERIMENTAL PROCEDURES
Materials-The sst 2 -selective ligand L-779,976 and the sst 3 -selective ligand L-796,778 were kindly provided by Dr. Susan Rohrer (24) (Merck). The radioligand 3-[ 125 I]iodotyrosyl-SS-14 (74 TBq/mmol) was from Amersham Pharmacia Biotech, and [ 35 S]GTP␥S (46.25 TBq/mmol) was from PerkinElmer Life Sciences. Mouse monoclonal anti-T7 tag antibody was obtained from Novagen (Madison, WI), and polyclonal rabbit anti-T7 and anti-c-Myc antibodies were obtained from Gramsch Laboratories (Schwabhausen, Germany). The rabbit anti-sst 2A antibody (6291) and the guinea pig anti-sst 2A antibody (GP3) were generated to the peptide ETQRTLLNGDLQTSI, which corresponds to residues 355-369 of the carboxyl terminus of the rat/mouse/human sst 2A and have been characterized extensively (25,26). The anti-sst 3 antibody (7986) was generated to the peptide TAGDKASTLSHL, which corresponds to residues 417-428 of the carboxyl terminus of the rat/mouse sst 3 and has also been characterized previously (27). All polyclonal rabbit antisera were affinity-purified against their immunizing peptides using the Sulfo-Link coupling gel (Pierce) according to the instructions of the manufacturer.
Cell Culture and Transfections-The wild-type rat sst 2A and sst 3 receptors were tagged at their amino termini either with the c-Myc epitope tag sequence MEEQKLISEEDLLR or the T7 epitope tag sequence MASMTGGQQMG using polymerase chain reaction. Two more amino acids, KL, were added representing a HindIII cloning site encoded by the nucleotide sequence AAGCTT. The resulting fragments were then subcloned into pcDNA3.1 expression vectors containing either a neomycin or a hygromycin resistance (Invitrogen, Groningen, The Netherlands). The integrity of all constructs was verified by dideoxy sequencing. Human embryonic kidney (HEK) 293 cells were obtained from ATCC and grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum in a humidified atmosphere containing 10% CO 2 . Cells were first transfected with plasmids containing the neomycin resistance using the calcium phosphate precipitation method. Stable tranfectants were selected in the presence of 500 g/ml G418 (Life Technologies, Inc.). To generate lines coexpressing two differentially epitope-tagged receptors, cells were subjected to a second round of transfection using Effectene (Qiagen, Hilden, Germany) and selected in the presence of 500 g/ml G418 and 300 g/ml hygromycin B (Invitrogen). Three clones expressing sst 2A alone, four clones expressing sst 3 alone, and two clones coexpressing sst 2A and sst 3 were generated. Receptor expression was monitored using saturation ligand binding assays as described below. The B max values of all selected clones were in the range of 800 -1200 fmol/mg of protein, and K D values were in the range of 0.16 -0.24 nM. In addition, quantitative Western blot analysis was carried out to ensure that clones coexpressing ϳ1:1 ratio of sst 2A and sst 3 were selected and used throughout this study. Finally, double immunofluorescent staining was performed to validate that sst 2A and sst 3 were coexpressed within the same cells.
Immunoprecipitation and Western Blot Analysis-Cells were plated onto poly-L-lysine-coated 150-mm dishes and grown to 80% confluence. When indicated, cells were exposed to agonist, treated with reducing agents, or incubated with cross-linking agents. Agonist exposure was performed with either SS-14, L-779,976, or L-796,778 at a concentration of 1 M for 10, 30, or 60 min at 37°C. Treatment with reducing agents was performed with 1 mM DL-dithiothreitol (DTT) for 30 min at 37°C. Incubation with cross-linking agents was performed with either 2 mM bis(sulfosuccinimidyl)suberate (BS 3 ) or 5 mM dithiobis-(succinimidylpropionate) (both from Pierce) for 30 or 60 min in phosphate-buffered saline (PBS) at 4°C. The reaction was quenched by the addition of 50 mM Tris (pH 7.5) for 15 min. Cells were then washed twice with PBS and harvested into ice-cold lysis buffer (10 mM Tris-HCl, pH 7.6, 5 mM EDTA, 3 mM EGTA, 250 mM sucrose, 10 mM iodoacetamide, and the following proteinase inhibitors: 0.2 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 1 g/ml pepstatin A, 1 g/ml aprotinin, 10 g/ml bacitracin). Iodoacetamide was included in each buffer used for protein preparation to prevent nonspecific disulfide linkages. Cells were swollen for 15 min on ice and homogenized. The homogenate was spun at 500 ϫ g for 5 min at 4°C to remove unbroken cells and nuclei. Membranes were then pelleted at 20,000 ϫ g for 30 min at 4°C, and pelleted membranes were lysed in detergent buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM EDTA, 3 mM EGTA, 4 mg/ml ␤-dodecylmaltoside, 10 mM iodoacetamide, and proteinase inhibitors as above) for 1 h on ice. The lysate was centrifuged at 20,000 ϫ g for 30 min at 4°C. The protein content of the resulting supernatant was determined using the BCA protein assay (Pierce); samples containing equal amounts of protein (300 g) were then subjected to immunoprecipitation, or glycoproteins were purified using wheat germ lectin.
For enrichment of glycoproteins, one ml of the supernatant was incubated with 100 l wheat germ agglutinin-agarose beads (Amersham Pharmacia Biotech) for 90 min at 4°C with continuous agitation. Beads were washed five times with detergent buffer, and adsorbed glycoproteins were either subjected to enzymatic deglycosylation or directly eluted into 200 l of SDS-sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 20% glycerol, 100 mM DL-dithiothreitol, 0.005% bromphenol blue) at 60°C for 20 min. Deglycosylation experiments were performed using peptide N-glycosidase F according to the manufacturer's protocol (New England Biolabs, Beverly, MA).
For immunoprecipitation, the lysates were precleared with 50 l of protein A-agarose beads (Calbiochem) for 2 h. After immunoprecipitation with 10 g of either mouse monoclonal anti-T7, affinity-purified rabbit anti-c-Myc, affinity-purified rabbit anti-sst 2A (6291), or affinitypurified rabbit anti-sst 3 (7986) antibodies, immunocomplexes were collected using 100 l of protein A-agarose beads. Beads were washed five times with detergent buffer, and immunoprecipitates were eluted from the beads with 200 l of SDS-sample buffer at 60°C for 20 min. Equal amounts of protein of each sample were then loaded onto regular 6% SDS-polyacrylamide gels, which contain 0.1% SDS. When indicated gels containing 2-fold (0.2%) or 4-fold (0.4%) higher SDS concentrations were run to test the sensitivity of dimers to stronger denaturing conditions. After electroblotting, membranes were incubated with either mouse monoclonal anti-T7, affinity-purified rabbit anti-c-Myc, affinitypurified rabbit anti-sst 2A (6291) or affinity-purified rabbit anti-sst 3 (7986) antibody at a concentration of 1 g/ml for 12 h at 4°C, followed by detection using an enhanced chemiluminescence detection system. Densitometric analysis of Western blots exposed in the linear range of the x-ray film was performed as described (31). The amount of immunoreactive material in each lane was quantified using NIH Image 1.57 software.
Immunocytochemistry-Cells were grown on poly-L-lysine-treated coverslips overnight and then exposed to agonists. Cells were fixed with 4% paraformaldehyde and 0.2% picric acid in phosphate buffer, pH 6.9, for 40 min at room temperature and washed several times in 10 mM Tris, 10 mM phosphate buffer, 137 mM NaCl, and 0.05% thimerosal, pH 7.4 (TPBS). Specimens were then incubated for 3 min in 50% methanol and for 3 min in 100% methanol and subsequently washed several times in TPBS and preincubated with TPBS and 3% normal goat serum for 1 h at room temperature. For single immunofluorescence, cells were then incubated with either mouse monoclonal anti-T7, affinity-purified rabbit anti-c-Myc, affinity-purified rabbit anti-sst 2A (6291), or affinitypurified rabbit anti-sst 3 (7986) antibody at a concentration of 1 g/ml in TPBS and 1% normal goat serum overnight. Bound primary antibody was detected with biotinylated secondary antibodies (1:100; Vector, Burlingame, CA) followed by cyanine 3.18-conjugated streptavidin (Amersham Pharmacia Biotech). For double immunofluorescence, cells were incubated either with a mixture of mouse monoclonal anti-T7 and affinity-purified rabbit anti-c-Myc or guinea pig anti-sst 2A (GP3) and affinity-purified rabbit anti-sst 3 (7986) antibodies. Bound primary antibodies were detected with biotinylated anti-rabbit antibodies, followed by a mixture of cyanine 2.18-conjugated streptavidin and cyanine 5.18conjugated anti-mouse or anti-guinea pig antibodies (1:200; Jackson ImmunoResearch, West Grove, PA). Cells were then dehydrated, cleared in xylol, and permanently mounted in DPX (Fluka, Neu-Ulm, Germany). Specimens were examined using a Leica TCS-NT laserscanning confocal microscope (Heidelberg, Germany) equipped with a krypton/argon laser. Cyanine 2.18 was imaged with 488-nm excitation and 500 -560-nm band pass emission filters, cyanine 3.18 with 568-nm excitation and 570 -630-nm band pass emission filters, and cyanine 5.18 with 647-nm excitation and 665-nm long pass emission filters. Confocal micrographs were taken by a person blinded to the treatments who was instructed to randomly select one colony of 4 -12 cells per coverslip.
Radioligand Binding Assays-Cells were harvested into PBS and stored at Ϫ80°C. After thawing, cells were centrifuged at 20,000 ϫ g for 5 min at 4°C and then homogenized in lysis buffer (50 mM Tris-HCl, 3 mM EGTA, 5 mM EDTA, pH 7.4). Cell membranes were pelleted by centrifugation at 50,000 ϫ g for 15 min at 4°C, washed with lysis buffer, and resuspended in binding buffer (10 mM HEPES, 5 mM MgCl 2 , 5 g/ml bacitracin, pH 7. Filters were rinsed twice with washing buffer (50 mM Tris-HCl, pH 7.4) and air-dried. Bound radioactivity was determined using a ␥-counter. Protein content was determined by the Lowry method.
GTP␥S Binding Assays-Cells were harvested and lysed as described above except that a lysis buffer containing 50 mM Tris and 10 mM EDTA (pH 7.4) was used. The resulting pellet was resuspended in assay buffer (20 mM HEPES, 100 mM NaCl, 10 mM MgCl 2 , pH 7.4), and aliquots containing 25 g of protein were incubated with 3 M GDP and 0.05 nM [ 35 S]GTP␥S in the presence or absence of either SS-14, L-779,976, or L-796,778 in concentrations ranging from 10 Ϫ12 to 10 Ϫ6 M. Assays were carried out in a final volume of 1 ml for 30 min at 30°C under continuous agitation. Nonspecific binding was determined in the presence of 10 M unlabeled GTP␥S. The incubation was terminated by vacuum filtration through glass fiber filters as described above. Filters were rinsed twice with washing buffer (20 mM HEPES, pH 7.4). Scintillation mixture was added, and radioactivity was determined using a ␤-counter. The maximal effector response (E max ) was determined by stimulation with sst agonists at a concentration of 1 M.
Measurements of cAMP Accumulation-Transfected cells were seeded at a density of 1.5 ϫ 10 5 /well onto poly-L-lysine-treated 22-mm 12-well dishes. On the next day, cells were either not preincubated or preincubated with 1 M SS-14 for 1, 2, 4, or 6 h in Opti-MEM 1 (Life Technologies). The medium was then removed and replaced with 0.5 ml of serum-free RPMI medium containing 25 M forskolin or 25 M forskolin plus either SS-14, L-779,976, or L-796,778 in concentrations ranging from 10 Ϫ12 to 10 Ϫ6 M. The cells were incubated at 37°C for 15 min. The reaction was terminated by removal of the culture medium and the subsequent addition of 1 ml of ice-cold HCl/ethanol (1 volume of 1 N HCl, 100 volumes of ethanol). After centrifugation, the supernatant was evaporated, the residue was dissolved in TE buffer (50 mM Tris-EDTA, pH 7.5), and the cAMP content was determined using a commercially available radioimmunoassay kit (Amersham Pharmacia Biotech).
ERK Assays-Cells were seeded at a density of 1.5 ϫ 10 5 /well onto poly-L-lysine-treated 22-mm 12-well dishes and grown for 2 days in Dulbecco's modified Eagle's medium containing 0.5% fetal calf serum. Cells were then exposed to either SS-14, L-779,976, or L-796,778 in concentrations ranging from 10 Ϫ12 to 10 Ϫ6 M for 5 min in RPMI medium without fetal calf serum at 37°C. Incubation was terminated by removal of the culture medium and the subsequent addition of 300 l of boiling SDS-sample buffer. Samples were heated to 95°C for an additional 5 min period. Equal amounts of protein of each sample were separated on 10% SDS-polyacrylamide gels and electroblotted onto nitrocellulose membranes. The protein content was determined using the BCA method. After blocking with 5% low fat dried milk dissolved in PBS containing 0.1% Tween 20, membranes were incubated with mouse monoclonal phosphospecific anti-ERK1/2 antibody clone E10 (1:1000; New England Biolabs). Blots were developed using peroxidase-conjugated secondary antibodies and enhanced chemiluminescence. Densitometric analysis of phospho-ERK1/2 levels on Western blots exposed in the linear range of the x-ray film was performed using NIH Image 1.57 software.
Data Analysis-Data from ligand binding, GTP␥S, cAMP, and ERK assays were analyzed by nonlinear regression curve fitting using GraphPad Prism 3.0 software. 2A Receptor-To examine dimerization of the sst 2A receptor, we used HEK 293 cells stably expressing either the T7-tagged sst 2A alone (B max ϭ 820 Ϯ 19 fmol/mg of membrane protein) or stably coexpressing T7tagged sst 2A and Myc-tagged sst 2A receptors were used. Western blot analysis of membrane extracts from these cells with the anti-sst 2A antibody (6291) revealed a predominant receptor band migrating at 80 kDa (Fig. 1A). We also observed an additional band with a higher molecular mass migrating at 160  (Fig. 1A). These bands were not only detected with antibodies directed against the carboxyl terminus (6291, GP3) but also with antibodies directed against the amino-terminal added T7 or Myc tag (not shown). A similar ratio of these two immunoreactive bands was observed when the cells had been subjected to cross-linking using the cell-impermeable cross-linker BS 3 prior to cell lysis (Fig. 1A). However, enzymatic deglycosylation reduced the size of the 80-kDa protein to 55 kDa and the size of the 160-kDa protein to 110 kDa, suggesting that the band with the higher molecular weight may consist of two sst 2A receptor proteins (Fig. 1B). Exposure to SS-14 did not grossly modulate the dimer/monomer ratio (Fig. 1C).

Homodimerization of the sst
The sst 2A dimer was stable under reducing conditions. Similar levels of sst 2A dimers were detected after incubation of the cells with 1 mM DTT for 30 min prior to cell lysis (Fig. 1D). The addition of DTT to the SDS-sample buffer also did not reduce the level of sst 2A dimers detected on Western blots. sst 2A dimers were also stable in the presence of 2% SDS (e.g. heating to 60°C for 20 min in SDS-sample buffer and loading onto regular SDS-polyacrylamide gels). However, when these samples were run on SDS-polyacrylamide gels containing 2-or 4-fold higher SDS concentrations, the sst 2A dimer was destabilized in a concentration-dependent manner, and only the monomeric form of the receptor was detectable under these conditions (Fig. 1D). This suggests that the sst 2A dimer is formed by noncovalent interactions of two receptor proteins.
To directly examine the presence of sst 2A receptor dimers, we used coimmunoprecipitation and Western blotting of differentially epitope-tagged sst 2A receptors. HEK 293 cells coexpressing Mycsst 2A and T7sst 2A were lysed, and the receptors from these cells were immunoprecipitated using anti-c-Myc antibody. Immunoprecipitates were resolved by SDS-PAGE and immunoblotted with anti-T7 antibody. As depicted in Fig. 1E, anti-T7 antibody detected a single band migrating at 160 kDa, which represents the Mycsst 2A -T7sst 2A dimer. Receptor monomer was not detectable, suggesting that the coprecipitated sst 2A dimers were stable during sample preparation and SDS-PAGE. The fact that T7sst 2A and Mycsst 2A were coimmunoprecipitated as part of a dimer complex from membrane extracts of untreated HEK 293 cells indicates that the sst 2A receptor exists as a constitutive homodimer in these cells. No bands were detectable in immunoprecipitates prepared under identical conditions from cells expressing only Mycsst 2A (Fig. 1E).
Homodimerization of the sst 3 Receptor-To examine dimerization of the sst 3 receptor, we used HEK 293 cells stably expressing the T7-tagged sst 3 (B max 1182 Ϯ 31 fmol/mg of membrane protein). Western blot analysis of membrane extracts from these cells with the anti-sst 3 antibody (7896) revealed a predominant receptor band migrating at 80 kDa ( Fig.  2A). We also observed an additional band with a higher molecular mass migrating at 160 kDa ( Fig. 2A). These bands were also detected with antibodies directed against the amino-terminal added T7 tag (not shown). Treatment of cells with the cross-linker BS 3 prior to cell lysis resulted in an increase of the 160-kDa band, suggesting that cross-linking can stabilize the sst 3 dimer ( Fig. 2A). However, no further increase in the intensity of the 160-kDa band was found when cells were exposed to SS-14 prior to cross-linking. (Fig. 2A). Enzymatic deglycosylation reduced the size of the 80-kDa protein to 70 kDa and the size of the 160-kDa protein to 140 kDa, suggesting that the band with the higher molecular weight may consist of two sst 3 receptor proteins (Fig. 2B).
Similar to that observed for the sst 2A dimer, the sst 3 dimer was stable under reducing conditions. The levels of sst 3 dimers were unchanged after incubation of cells with 1 mM DTT for 30 min prior to cell lysis (Fig. 2C). The addition of DTT to the SDS-sample buffer also did not reduce the levels of sst 3 dimers detected on Western blots. sst 3 dimers were also stable in the presence of 2% SDS (e.g. heating to 60°C for 20 min in SDSsample buffer and loading onto regular SDS-polyacrylamide gels). However, when these samples were run on SDS-polyacrylamide gels containing 2-or 4-fold higher SDS concentrations, the sst 3 dimer was destabilized in a concentration-dependent manner, and only the monomeric form of the receptor was detectable under these conditions (Fig. 2C). This suggests that the sst 3 dimer is formed by noncovalent interactions of two receptor proteins.
To examine the role of the carboxyl terminus for sst 3 dimerization, we performed Western blotting of HEK 293 cells expressing an sst 3 receptor with a deletion of the entire cytoplasmic tail (T7sst 3 ⌬C75). In membrane extracts from these cells, we detected a predominant receptor band migrating at 70 kDa and a second band migrating at 140 kDa, suggesting that the carboxyl terminus was not absolutely required for sst 3 receptor homodimerization (Fig. 2D).

FIG. 2. Characterization of sst 3 somatostatin receptor homodimers by immunoblotting.
A, HEK 293 cells expressing T7 epitope-tagged sst 3 receptor were incubated in the absence (Ϫ) or presence (ϩ) of 2 mM BS 3 and/or SS-14 (ϩ). Membrane proteins were extracted and immunoblotted using the anti-sst 3 antibody (7986) as described under "Experimental Procedures." B, membrane proteins from T7sst 3 -expressing cells were extracted and subjected to enzymatic deglycosylation using peptide N-glycosidase F (PNGF). C, cells were incubated in the absence (Ϫ) or presence (ϩ) of 1 mM DTT for 30 min. Membrane proteins were extracted, and samples were run on regular SDS-polyacrylamide gels (lanes 1 and 2). Alternatively, samples were run on SDS-polyacrylamide gels containing a 4-fold higher concentration of SDS (lane 3). DTT was also present in the SDS-sample buffer. Note that the sst 3 dimer is detectable without cross-linking and stable under reducing conditions; however, it is sensitive to higher concentrations of detergents. D, HEK 293 cells expressing either the wild-type sst 3 receptor (T7sst 3 ) or a C-terminal truncated sst 3 receptor (T7sst 3 ⌬75) were immunoblotted using mouse monoclonal anti-T7 antibody. The positions of molecular mass markers are indicated on the left (in kDa). The arrows point to the dimeric and monomeric forms of the receptor. Four additional experiments gave similar results.
Heterodimerization of the sst 2A and sst 3 Receptors-First, we determined the levels of sst 2A and sst 3 receptor proteins in coexpressing HEK 293 cells using quantitative Western blot analysis. Since sst 2A and sst 3 were tagged with different epitopes, it was not possible to compare their expression levels directly using different antibodies. Therefore, equal amounts of membrane protein extracted from HEK 293 cells stably expressing the T7-tagged sst 2A receptor alone (B max 820 Ϯ 19 fmol/mg) or the T7-tagged sst 3 receptor alone (B max 1182 Ϯ 31 fmol/mg) or coexpressing T7-tagged sst 2A and c-Myc-tagged sst 3 receptors (B max 1088 Ϯ 66 fmol/mg) were subjected to SDS-PAGE and subsequently probed with both anti-sst 2A (6291) and anti-sst 3 (7986) antibodies (Fig. 3A). Densitometric scanning of the resulting Western blots revealed that virtually identical levels of sst 2A receptor protein were present in HEK 293 cells coexpressing sst 2A and sst 3 as compared with cells expressing sst 2A alone (Fig. 3A, left panel). Conversely, virtually identical levels of sst 3 receptor protein were present in HEK 293 cells coexpressing sst 3 and sst 2A as compared with cells expressing sst 3 alone (Fig. 3A, right panel). Given the fact that T7sst 2A and T7sst 3 cells expressed similar numbers of somatostatin binding sites, we conclude that cotransfected cells expressed T7sst 2A and Mycsst 3 receptor proteins in a ratio of ϳ1:1. However, it should be noted that the higher levels of total somatostatin receptor proteins (sst 2A plus sst 3 ) present in coexpressing cells were not associated with an equivalent increase in the total number of somatostatin binding sites detected in saturation binding assays (B max for T7sst 2A -Mycsst 3 1088 Ϯ 66 fmol/mg as compared with B max for T7sst 2A 820 Ϯ 19 fmol/mg and B max for T7sst 3 1182 Ϯ 31 fmol/mg).
We next examined the ability of sst 2A and sst 3 to assemble as heterodimers in coexpressing cells. As shown in Fig. 3B, the C-terminal anti-sst 2A antibody (6291) detected a band migrating at 160 kDa in material immunoprecipitated using the Cterminal anti-sst 3 antibody (7986), suggesting that this band represents a T7sst 2A -Mycsst 3 heterodimer. sst 2A -sst 3 Heterodimers can be immunoprecipitated under a variety of conditions (e.g. using the anti-c-Myc antibody or the anti-T7 antibody and immunoblotted vice versa) (Fig. 3C). All of these preparations revealed a single 160-kDa band, suggesting that stable sst 2A -sst 3 heterodimers were precipitated under these conditions and that sst 2A -sst 3 heterodimers were not subject to monomerization during sample preparation. However, when T7-tagged sst 2A receptors were immunoprecipitated from T7sst 2A -Mycsst 3 cells using anti-sst 2A antibody (6291) and then immunoblotted with anti-sst 2A antibody (6291), both dimeric and monomeric forms of the receptor could be detected (Fig.  3B). Essentially identical results were obtained when Myctagged sst 3 receptors were immunoprecipitated from T7sst 2A -Mycsst 3 cells using anti-sst 3 antibody (7986) and then immunoblotted with anti-sst 3 antibody (7986) (not shown). These results are in agreement with those seen on Western blots from membrane extracts of T7sst 2A -Mycsst 3 cells, suggesting that the receptor proteins existed in both monomeric and dimeric form in these cells (Fig. 3A).
As shown in Fig. 3C, anti-T7 antibody detected a band migrating at 160 kDa in material immunoprecipitated using antic-Myc antibodies specific for the sst 3 receptor, suggesting that this band represents a T7sst 2A -Mycsst 3 heterodimer. In contrast, no bands were detectable in immunoprecipitates prepared under identical conditions from cells expressing only T7sst 2A or T7sst 3 or from a mixture of T7sst 2A -and Mycsst 3expressing cells (Fig. 3C). These data strongly suggest that sst 2A -sst 3 heterodimers preexisted in cells prior to cell lysis and were not artificially formed during sample preparation.
Agonist exposure had little effect on the level of sst 2A -sst 3 heterodimer (Fig. 3D). However, we observed that after exposure of coexpressing cells, not only heterodimers but also traces of the monomeric receptor forms became detectable (Fig. 3D). The simplest explanation for this finding is that during agonist exposure a proportion of the sst 2A -sst 3 heterodimer was destabilized, presumably due to a change in the conformation and/or phosphorylation of the receptor that still allowed coimmunoprecipitation of sst 2A -sst 3 heterodimers but facilitated separation of the receptors during sample preparation and gel electrophoresis.
Endocytotic Trafficking of the sst 2A -sst 3 Heterodimer-We In cells coexpressing T7sst 2A and Mycsst 3 , both sst 2A -like immunoreactivity and sst 3 -like immunoreactivity were seen at the cell surface, revealing extensive colocalization of sst 2A and sst 3 receptor proteins (Fig. 5, Control). Interestingly, after 30 min of SS-14 exposure, the sst 2A receptor underwent robust internalization, whereas the sst 3 receptor remained almost exclusively confined to plasma membrane (Fig. 5, SS-14). However, internalization of the sst 2A receptor was not complete, which may explain the fact that sst 2A -sst 3 heterodimers could also be coimmunoprecipitated after treatment with SS-14 (Fig.  3C). The sst 3 receptor did not undergo endocytosis even after prolonged SS-14 exposure (up to 4 h, not shown). Similarly, after treatment with the sst 2 -selective ligand L-779,976, only the sst 2A and not the sst 3 receptor was internalized (Fig. 5,  L-779,976). In contrast, treatment with sst 3 -selective ligand L-796,778 did not induce substantial internalization of either sst 2A or sst 3 (Fig. 5, L-796,778).
Ligand Binding Properties of the sst 2A -sst 3 6, A-C). Specifically, the cells coexpressing sst 2A and sst 3 exhibited a 100-fold lower affinity for the sst 3 -selective agonist L-796,778 than cells expressing sst 3 alone (Table I). These findings imply that sst 2A -sst 3 heterodimerization results in a new binding site with a pharmacological profile resembling that of the sst 2A receptor. Although sst 2A and sst 3 receptor proteins were expressed in a 1:1 ratio, the sst 2A -sst 3 heterodimer showed no significant affinity for the sst 3 (Fig. 6, D-F) (Fig. 6E). In contrast to that seen in T7sst 3 cells, the sst 3 -selective agonist L-796,778 produced no substantial [ 35 S]GTP␥S binding in cells coexpressing T7sst 2A and Mycsst 3 (Fig. 6, E and F). hibition of adenylyl cyclase with half-maximal inhibitory concentrations (IC 50 ) in the low nanomolar range in T7sst 2A cells and T7sst 3 cells as well as in cells coexpressing both T7sst 2A and Mycsst 3 (Fig. 6, G-I, Table I). The sst 2 -selective agonist L-779,976 produced similar robust responses in T7sst 2A cells as in cells coexpressing T7sst 2A and Mycsst 3 . In contrast, sst 3selective agonist L-796,778 inhibited forskolin-stimulated cAMP accumulation in T7sst 3 cells only, but essentially no inhibition was found in cells coexpressing T7sst 2A and Mycsst 3 (Fig. 6, H and I, Table I). We also investigated the possibility of a partial agonism by L-796,778. If the sst 3 -selective ligand L-796,778 would be a partial agonist at the sst 3 dimer or antagonist at the sst 2A -sst 3 heterodimer, it would be expected to block the SS-14-mediated inhibition of cAMP accumulation. However, SS-14-mediated responses were not attenuated in the presence of L-796,778 either on T7sst 3 cells or on cells coexpressing both T7sst 2A and Mycsst 3 , suggesting that L-796,778 is a pure sst 3 agonist (data not shown).
ERK Activation by sst 2A -sst 3 Heterodimers-The activation of somatostatin receptors by agonists results in a rapid and transient stimulation of ERK1/2 phosphorylation. We next examined the ability of several agonists to increase the levels of phosphorylated ERK1/2 in cells expressing either T7sst 2A or T7sst 3 or coexpressing both T7sst 2A and Mycsst 3 . As shown in Fig. 7, 5-min exposure to 100 nM SS-14 produced a robust increase in ERK1/2 phosphorylation in T7sst 2A cells and T7sst 3 as well as in cells coexpressing both T7sst 2A and Mycsst 3 . The sst 2 -selective agonist L-779,976 (10 nM) promoted a similar response in T7sst 2A cells as in cells coexpressing T7sst 2A and Mycsst 3 . In contrast, the sst 3 -selective agonist L-796,778 (100 nM) stimulated ERK1/2 activity in T7sst 3 cells only but not in cells coexpressing T7sst 2A and Mycsst 3 (Fig. 7, Table I). These findings imply that sst 2A -sst 3 heterodimerization results in a new receptor with a functional profile (e.g. [ 35 S]GTP␥S binding, inhibition of adenylyl cyclase, and ERK activation) resembling that of the sst 2A receptor. Although sst 2A and sst 3 receptor proteins were expressed in a 1:1 ratio, the sst 2A -sst 3 heterodimer showed no significant functional response to the sst 3specific ligand L-796,778.
Desensitization of the sst 2A -sst 3 Heterodimer-Finally, we compared agonist-induced desensitization of sst 2A dimer and sst 2A -sst 3 heterodimer. Cells expressing T7sst 2A or coexpressing both T7sst 2A and Mycsst 3 were preincubated with 1 M SS-14 for 0, 1, 2, 4, or 6 h. The medium was removed, and the ability of SS-14 to inhibit forskolin-stimulated cAMP accumulation was examined. The sst 2A dimer underwent a rapid timedependent loss of coupling to adenylate cyclase with a maximum desensitization at 6 h (Fig. 8A). In contrast, the sst 2A -sst 3 heterodimer appeared to be more resistant to agonist-induced desensitization with a strong receptor response retained as long as 6 h (Fig. 8B). To test the possibility that the delayed desensitization of the sst 2A -sst 3 heterodimer may be related to agonist-induced monomerization and subsequent gain of function of the sst 3 receptor, cells coexpressing T7sst 2A and Mycsst 3 were challenged with the sst 3 -selective ligand L-796,778 after extended exposure with SS-14. However, the sst 3 -selective ligand was not able to inhibit forskolin-induced cAMP accumulation in T7sst 2A -Mycsst 3 cells whether or not these cells had been preincubated with SS-14 (data not shown).

DISCUSSION
The Western blot and coimmunoprecipitation experiments carried out in the present study clearly demonstrate that the sst 2A receptor as well as the sst 3 receptor exist as constitutive homodimers at the plasma membrane when expressed alone and as heterodimers when coexpressed in HEK 293 cells. Both sst 2A and sst 3 dimers were resistant to reducing agents but sensitive to higher concentrations of detergents, suggesting that dimerization involves noncovalent hydrophobic interactions of the receptor proteins (Figs. 1D and 2C). Formation of homodimers between truncated sst 3 receptors suggests that the cytoplasmic tail was not necessary for dimerization (Fig.  2D). Several dimerization interfaces have been proposed for other GPCRs such as the extracellular amino-terminal domain for the glutamate and calcium-sensing receptors, the intracellular third loop, and the VIth transmembrane region for the dopaminergic and ␤ 2 -adrenergic receptor and the C-terminal tail for the ␦-opioid and GABA B receptor (1, 2, 4, 5, 7, 14 -19).
To study the sst 2A -sst 3 heterodimer directly, we performed coimmunoprecipitation experiments using HEK 293 cells coexpressing T7sst 2A and Mycsst 3 . These studies clearly showed that the sst 2A -sst 3 heterodimers can be coimmunoprecipitated using a variety of antibody combinations (Fig. 3). To exclude the possibility that sst 2A -sst 3 heterodimers were artifactually formed during the preparation of cell lysates and sample processing, coimmunoprecipitation studies were also carried out using a mixture of cells expressing T7sst2A and Mycsst3 individually. Under these conditions, heterodimers could not be immunoprecipitated, strongly suggesting that sst 2A -sst 3 heterodimers were formed in vivo prior to cell lysis (Fig. 3A). Interestingly, when the material coimmunoprecipitated using an anti-sst 2A antibody was probed with an anti-sst 3 antibody, we detected only a single high molecular weight band corresponding to the sst 2A -sst 3 heterodimer. This suggests that het-erodimers were stable during cell lysis and reducing SDS-PAGE. When material coimmunoprecipitated using an antisst 2A antibody was probed with an anti-sst 2A antibody, we detected an additional lower molecular weight band that corresponds to the sst 2A monomer, indicating that, similar to the results seen on Western blots, both monomeric and heterodimeric forms of the receptor proteins were present at the plasma membrane. However, the possibility that sst 2A and sst 3 may exist as high molecular weight hetero-oligomeric arrays in coexpressing cells cannot be excluded.
Agonist treatment had little effect on the levels of sst 2A and sst 3 homo-and heterodimers detected on Western blots (Figs. 1C, 2A, and 3D). Interestingly, agonist exposure of the ␦-opioid receptor was found to induce monomerization, which precedes receptor internalization (2). In contrast, recent studies using fluorescence resonance energy transfer suggest that dimers formed by the ␤ 2 -adrenergic receptor and the sst 5 somatostatin receptor are stabilized by agonist binding (1,11,20). On the other hand, agonist stimulation of the m3 muscarinic receptor and the calcium-sensing receptor neither promotes nor destabilizes receptor dimer formation (5,6,9,10). It remains to be clarified whether the observed differences in agonist regulation of receptor dimerization reflect intrinsic differences in the structural properties of the studied receptor proteins or are due to variations in experimental conditions.
Although sst 2A and sst 3 receptor proteins were expressed  Table I). These findings suggest that heterodimerization of sst 2A and sst 3 results in a new receptor with a pharmacological profile resembling that of the sst 2A receptor. However, the sst 2A -sst 3 heterodimer and the sst 2A homodimer differ in that the sst 2A -sst 3 heterodimer appears to be more resistant to agonist-induced desensitization of coupling to adenylyl cyclase than the sst 2A homodimer (Fig. 8). Several recent studies have reported on heterodimerization among GPCRs (e.g. the GABA B heterodimers, Ϫ␦ heterodimers, and D2R-sst 5 heterodimers) (3,(12)(13)(14)(15)(16)(17)(18)(19)(20). In each case, heterodimerization results in a new receptor with enhanced functional activity. Here, we provide the first evidence for inactivation of a fully functional sst 3 receptor by heterodimerization with the sst 2A receptor.
The fact that the sst 3 receptor is rendered nonfunctional after heterodimerization with sst 2A may provide a plausible explanation for the unexpectedly low number of somatostatin binding sites on HEK 293 cells coexpressing sst 2A and sst 3 . Comparative Western blot analysis revealed that virtually identical levels of sst 2A and sst 3 receptor protein were present in coexpressing cells as compared with cells expressing these receptors alone. Saturation binding assays revealed that T7sst2A (B max ϭ 820 Ϯ 19 fmol/mg) and T7sst3 cells (B max ϭ 1182 Ϯ 31 fmol/mg) exhibited similar densities of somatostatin binding sites. This suggests that sst 2A and sst 3 receptor proteins were expressed approximately in a 1:1 ratio in T7sst 2A -Mycsst 3 cells. This also suggests that higher levels of total somatostatin receptor proteins (sst 2A plus sst 3 ) were present in coexpressing cells. However, this doubling of somatostatin receptor protein was not associated with an equivalent increase in the total number of detectable somatostatin binding sites in these cells (B max ϭ 1088 Ϯ 66 fmol/mg).
The functional inactivation of sst 3 by heterodimerization with sst 2A may also explain the observed destabilization of the sst 2A -sst 3 heterodimer and selective internalization of sst 2A after SS-14 treatment. Agonist exposure may induce a conformational change of the sst 2A -sst 3 heterodimer that favors G protein-coupled receptor kinase-mediated phosphorylation of sst 2A while sst 3 remains in a nonphosphorylated state. Phosphorylation of intracellular domains of the sst 2A receptor would then be expected to facilitate binding of ␤-arrestin and preferential targeting of this receptor to the endocytotic machinery. Nevertheless, internalization of the sst 2A receptor was not complete (Fig. 5), and sst 2A -sst 3 heterodimers could also be coimmunoprecipitated from membrane preparations of SS-14treated cells (Fig. 3D). However, the agonist-induced conformational change appeared to destabilize the sst 2A -sst 3 heterodimer to an extent that still allowed coimmunoprecipitation of sst 2A -sst 3 heterodimers but facilitated separation of a proportion of heterodimers during sample preparation and gel electrophoresis and, hence, permitting detection of monomeric  3

heterodimers
Radioligand binding studies and cAMP and ERK assays were carried out as described under "Experimental Procedures." The half-maximal inhibitory concentrations (IC 50 ) for competition binding and cAMP assays and the half-maximal effector concentration (EC 50 ) for ERK activation were analyzed by nonlinear regression curve fitting using the computer program GraphPad Prism 3.0. Data are presented as the mean of three or four independent experiments. S.E. values were smaller than 15%.  receptor forms on Western blots (Fig. 3D). sst 2A and sst 3 are widely distributed throughout the central nervous system and periphery (25)(26)(27). sst 2A -sst 3 heterodimerization may also occur in vivo, since colocalization of these receptors has been observed in several tissues including pancreatic islands and the anterior lobe of the pituitary. Interestingly, particular high levels of sst 3 expression have been detected in the cerebellum; however, radioligand binding studies largely failed to detect sst 3 receptor binding sites in this region (27)(28)(29). Thus, functional inactivation of the sst 3 receptor by heterodimerization with sst 2A or other GPCRs may provide a possible explanation for some of the difficulties in detecting sst 3 -specific binding and signaling in mammalian tissues.
The physical interaction between somatostatin receptors may have direct implications for the treatment of neuroendocrine malignancies that frequently overexpress several subtypes of somatostatin receptors (30,31). Based on the present findings, a tumor coexpressing sst 2A and sst 3 would be expected to respond to treatment with sst 2 -selective but not with sst 3selective agonists. In contrast, a tumor with isolated expression of sst 3 would be expected to respond to treatment with sst 3selective but not with sst 2 -selective agonists. Evaluation of the somatostatin receptor status in a given tumor may therefore provide valuable predictive information for the treatment of human tumors with somatostatin receptor subtype-specific ligands (24,31).
In conclusion, we provide biochemical and functional evidence for homo-and heterodimerization of the sst 2A and sst 3 somatostatin receptors. We show that heterodimerization results in a new receptor with a pharmacological and functional profile resembling that of the sst 2A receptor, however with a greater resistance to agonist-induced desensitization. The sst 2A -sst 3 receptor is the first heterodimer that results in inactivation of a fully functional receptor.