Subtypes of the Somatostatin Receptor Assemble as Functional Homo- and Heterodimers*

The existence of receptor dimers has been proposed for several G protein-coupled receptors. However, the question of whether G protein-coupled receptor dimers are necessary for activating or modulating normal receptor function is unclear. We address this question with somatostatin receptors (SSTRs) of which there are five distinct subtypes. By using transfected mutant and wild type receptors, as well as endogenous receptors, we provide pharmacological, biochemical, and physical evidence, based on fluorescence resonance energy transfer analysis, that activation by ligand induces SSTR dimerization, both homo- and heterodimerization with other members of the SSTR family, and that dimerization alters the functional properties of the receptor such as ligand binding affinity and agonist-induced receptor internalization and up-regulation. Double label confocal fluorescence microscopy showed that when SSTR1 and SSTR5 subtypes were coexpressed in Chinese hamster ovary-K1 cells and treated with agonist they underwent internalization and were colocalized in cytoplasmic vesicles. SSTR5 formed heterodimers with SSTR1 but not with SSTR4 suggesting that heterodimerization is a specific process that is restricted to some but not all receptor subtype combinations. Direct protein interaction between different members of the SSTR subfamily defines a new level of molecular cross-talk between subtypes of the SSTR and conjugated SSTR Constructs and Transfections— The D 318 hSSTR5 C-tail dele- tion mutant and the second extracellular loop (ECL2) hSSTR5 mutant in which 7 of the 10 COOH-terminal residues of the second ECL were conservatively altered have been described previously (29, 30). Stable CHO-K1 cells overexpressing full-length HA-tagged hSSTR5 were ob- tained from K. Koller (31). Stable CHO-K1 cotransfectants expressing wt hSSTR1-pCDNA (32), wt hSSTR4-pRC/CMV (32), HA-hSSTR5 in a 1 12CA5-KH (31), D 318 hSSTR5-pTEJ8 or ECL2/ D 318 hSSTR5 mutants in pTEJ8 were prepared by Lipofectin transfection (Life Technol- ogies, Inc.). Neomycin-resistant clones were selected and maintained in F12 medium with 10% fetal bovine serum and 700 m g/ml G418. Binding Assays, Internalization, and Up-regulation Experiments— Binding studies were carried out for 30 min at 37 °C with cell mem- brane protein or whole cells with 125 I-labeled LTT-SST-28 radioligand or subtype-selective ligands as previously reported (29, 30, 32, 33). Receptor coupling to adenylyl cyclase was tested by incubating cells for 30 min with 1 m M forskolin with or without SST (10 2 10 –10 2 6 M ) at 37 °C as described previously (30). Cells were then scraped in 0.1 N HCl and assayed for cAMP by radioimmunoassay SSTRs in transfected CHO-K1 cells was determined by immunocytochemistry. Rabbit polyclonal anti- peptide antibodies directed against sequences in the amino-terminal segment of hSSTR1 (diluted 1:300) or mouse monoclonal anti-HA anti- bodies (diluted 1:300) were used as primary antibodies followed by reaction with rhodamine or fluorescein-conjugated secondary antibody as described previously (33). To demonstrate colocalization of hSSTR1 with hSSTR5, CHO cells stably cotransfected with wt hSSTR1 and HA-hSSTR5 were treated with 1 m M SMS for 12 h at 4 °C. For receptor localization on the plasma membrane, cells were fixed at 4 °C in 3.7% formalin for 15 min. For receptor localization in vesicles, cells were incubated for an additional 60 min at 37 °C to allow internalization, fixed, and permeabilized in methanol/acetone at 2 10 °C for 15 min. The fixed cells in both instances were processed for double label confocal fluorescence immunocytochemistry. Cells were mounted with immunofluor and viewed under a Zeiss LSM 410 confocal microscope. Images were obtained as single optical sections taken through the middle of cells and averaged over 32 scans/frame.

Many membrane proteins such as ion channels, receptor tyrosine kinases, and receptors for growth hormone and cytokines associate as functional oligomeric complexes (1)(2)(3)(4). Although G protein-coupled receptors (GPCRs) 1 are generally believed to operate as monomers, several recent lines of evidence based on thermodynamic, biochemical, and functional studies suggest that this class of membrane proteins may also associate as dimers (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). However, the question of whether dimerization is a general property of GPCRs and whether it is necessary for GPCR function remains controversial (9,10,15,16,21). The GABA-B receptor associates as a heterodimer via the cytoplasmic C-tail in the endoplasmic reticulum and is targeted to the plasma membrane as a preformed dimer, independent of agonist regulation (11)(12)(13)(14)21). Whether other GPCR dimers are similarly preformed or whether they undergo dimerization at the plasma membrane in response to agonist activation is unclear (9,10,15,16,21). Dopamine and muscarinic receptors have been postulated to exist on the membrane as preformed dimers that are stabilized by ligand binding (9,19). The ␤-adrenergic receptor on the other hand undergoes ligand-dependent dimerization and activation, whereas agonists at the ␦ opioid receptor have been suggested to favor monomer formation that is required for agonist-induced internalization (10,16). In the case of somatostatin (SST) receptors (SSTRs), there are five distinct subtypes that bind the two natural ligands, SST-14 and SST-28, with comparable low nanomolar affinity (22). The five subtypes also share common signaling pathways such as the ability to inhibit adenylyl cyclase and to activate phosphotyrosine phosphatase (22)(23)(24). Furthermore, individual target cells typically express more than one SSTR subtype and often all five isoforms (25)(26)(27)(28) raising the question of whether multiple SSTRs in the same cell are redundant or whether they interact for greater functional diversity. By using pharmacological, biochemical, and physical methods, here we show that SSTRs associate as dimers, both as homodimers or heterodimers with other members of the SSTR family, and that dimerization alters the functional properties of the receptor such as ligand binding affinity, signaling, and agonist-induced regulation. We provide the first direct evidence based on the sensitive fluorescence resonance energy transfer (FRET) analysis that hSSTR5 exists as a monomer in the basal state and undergoes dose-dependent increase in dimerization when treated with SST-14 suggesting that dimerization is induced by agonist binding.

EXPERIMENTAL PROCEDURES
Peptides and Antisera-Peptides and antisera were obtained as follows: SST-14, SST-28 (Bachem); Leu 8 -D-Trp 22  This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Binding Assays, Internalization, and Up-regulation Experiments-Binding studies were carried out for 30 min at 37°C with cell membrane protein or whole cells with 125 I-labeled LTT-SST-28 radioligand or subtype-selective ligands as previously reported (29,30,32,33). Receptor coupling to adenylyl cyclase was tested by incubating cells for 30 min with 1 mM forskolin with or without SST (10 Ϫ10 -10 Ϫ6 M) at 37°C as described previously (30). Cells were then scraped in 0.1 N HCl and assayed for cAMP by radioimmunoassay (30,33). Internalization experiments were carried out by incubating cells overnight at 4°C with radioligand with or without SST (0.1 mM) (30,32,33). After washing, cells were warmed to 37°C for 15, 30, and 60 min to initiate internalization. At the end of each incubation, surface-bound radioligand was removed by acid wash, and internalized radioligand was measured as acid-resistant counts in 0.1 N NaOH extracts of acid-washed cells (30,32,33). The ability of long term treatment with SST to up-regulate surface SSTR binding was studied in cells cultured with 1 M SST or SMS for 22 h as described previously (32,33). After acid wash to remove surface-bound SST, whole cell binding assays were performed to determine total and nonspecific binding. Residual surface binding was calculated as the difference between control and experimental groups (32,33).
Western Blots-CHO-K1 cells expressing HA-SSTR5 were analyzed for receptor protein by Western blots as reported previously (27). Membranes were incubated with or without SST-14 (10 Ϫ6 M) for 30 min at 37 o C and then solubilized in sample buffer containing 62.5 mmol/liter Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, and 50 mmol/liter dithiothreitol. 50-g samples of membrane protein were fractionated by electrophoresis on 10% SDS-polyacrylamide gels as described by Laemmli (34). The fractionated proteins were transferred by electrophoresis to nitrocellulose membranes in a transfer buffer containing 0.025 mol/liter Tris, 0.192 mol/liter glycine, and 15% methanol. The membranes were then probed for HA-SSTR5 using the mouse monoclonal antibody and the lumilight ϩ Western blotting Kit (Roche Molecular Biochemicals) (27). Blots were analyzed semi-quantitatively using the computer scanning software Masterscan.
Photobleaching (pb) FRET Microscopy-Generally, FRET efficiencies are determined indirectly by measuring changes in the quantum yield of any competitive donor deactivation process upon introduction of an acceptor molecule (35)(36)(37)(38)(39). Donor photobleaching represents such a competitive process that can be exploited in pbFRET microscopy. The effective FRET efficiency E is calculated from the photobleaching time constants of the donor (D) obtained in the absence ( DϪA ) and presence ( DϩA ) of acceptor (A) according to Equation 1.
In a two-state model, the minimal amount of receptor dimerization (␣ min ) is related to the fraction of acceptor labeled receptor (f A ) and E as shown in Equation 2, where f A is determined from the relative affinities of fluorescein-and rhodamine-conjugated mAbs and the concentration ratio used for incubation (39). pbFRET experiments were performed on CHO-K1 cells stably expressing HA-hSSTR5 using a Leica DMBL fluorescence microscope equipped with epi-illumination. An OSRAM HBO 100-watt mercury lamp was used as excitation light source. In order to separate fluorescein excitation from emission as well as to optimize fluorescein excitation while simultaneously blocking rhodamine excitation, the fol-lowing filters were used: Leitz BP 450 -490 (excitation), RKP510 (dichroic mirror), and BP 515-535 (emission). Digital images (8-bit) were generated with an Electrim-1000U CCD camera with a spatial resolution of 1134 ϫ 486 pixels of size 7.8 ϫ 13.6 m. Exposure as well as time delays were software controlled. IGOR Pro 3.13 (Wavemetrics, OR) was used for image analysis. Images were corrected for dark current, background, and flatness. Immunocytochemistry-Expression of SSTRs in transfected CHO-K1 cells was determined by immunocytochemistry. Rabbit polyclonal antipeptide antibodies directed against sequences in the amino-terminal segment of hSSTR1 (diluted 1:300) or mouse monoclonal anti-HA antibodies (diluted 1:300) were used as primary antibodies followed by reaction with rhodamine or fluorescein-conjugated secondary antibody as described previously (33). To demonstrate colocalization of hSSTR1 with hSSTR5, CHO cells stably cotransfected with wt hSSTR1 and HA-hSSTR5 were treated with 1 M SMS for 12 h at 4°C. For receptor localization on the plasma membrane, cells were fixed at 4°C in 3.7% formalin for 15 min. For receptor localization in vesicles, cells were incubated for an additional 60 min at 37°C to allow internalization, fixed, and permeabilized in methanol/acetone at Ϫ10°C for 15 min. The fixed cells in both instances were processed for double label confocal fluorescence immunocytochemistry. Cells were mounted with immunofluor and viewed under a Zeiss LSM 410 confocal microscope. Images were obtained as single optical sections taken through the middle of cells and averaged over 32 scans/frame.
Statistical Analysis-Results are presented as mean Ϯ S.E.

RESULTS
Agonist-dependent Homodimerization of hSSTR5-We first demonstrated SSTR homodimerization by functional complementation of two partially active mutants of human SSTR5 (hSSTR5) that we have previously described (29,30) (Fig. 1). One is a conservative segment exchange mutant of the second extracellular loop, ECL2 hSSTR5 which fails to bind SST-14/ SST-28 but which is correctly targeted to the plasma membrane as shown by immunocytochemistry (29). The other is a cytoplasmic tail (C-tail) deletion mutant ⌬318 hSSTR5 that displays complete loss of adenylyl cyclase coupling while retaining full agonist binding potency and the ability to undergo agonist-dependent internalization (30). We wondered whether loss of adenylyl cyclase coupling by the C-tail deletion mutant could be rescued by cotransfection, whereby the binding competent mutant would associate and signal through the C-tail of the binding-deficient mutant. The two mutants were stably cotransfected in CHO-K1 cells (B max 119 Ϯ 36 fmol/mg protein) and compared with individual ⌬318 hSSTR5 (B max 126 Ϯ 43 fmol/mg protein) or ECL2 hSSTR5 (no binding) stable monotransfectants. Coupling to adenylyl cyclase was determined by the ability of SST-28 to inhibit forskolin-stimulated cAMP. In the cotransfectant, SST-28 produced dose-dependent inhibition of forskolin-stimulated cAMP (31 Ϯ 2.5% at 1 M agonist) which was completely abolished by pertussis toxin treatment (Fig. 2B). This suggests that the two mutant receptors assemble as homodimers to constitute a functional G protein-linked effector complex. Competition analysis showed a significant 4-fold increase in the binding affinity of SST-14 for the putative ⌬318-hSSTR5/ECL2-hSSTR5 dimeric receptors (K i 3.1 Ϯ 1.9 nM) compared with ⌬318 hSSTR5 alone (K i 12.1 Ϯ 2.5 nM) suggesting physical association leading to a change in receptor conformation ( Fig. 2A). We next investigated the effect of coexpression of the mutant receptors on receptor internalization ( Fig. 2C). CHO-K1 cells expressing either ⌬318 hSSTR5 or ⌬318 hSSTR5 and ECL2 hSSTR5 mutants were incubated at 37°C for different times with 125 I-labeled LTT-SST-28 (a nonselective radioligand for all five SSTR subtypes) with or without 0.1 M SST-28 (30,32). The ⌬318 hSSTR5 mutant displayed time-dependent internalization with a maximum of 41 Ϯ 2.8% at 60 min (Fig. 2C). Coexpression of the ECL2 mutant markedly inhibited internalization of the putative dimeric complex to only 13 Ϯ 3.9% at 60 min. This suggests that the association of the ECL2 hSSTR5 mutant with the binding competent ⌬318 hSSTR mutant, although promoting ligand binding affinity, impairs internalization of the dimeric receptor complex perhaps because only one of the receptor subunits is in a ligand-bound state. Physical association of SSTRs was investigated by Western blot analysis of CHO-K1 cell membranes expressing full-length hSSTR5 tagged at the amino terminus with a nonapeptide of the hemagglutinin (HA) sequence (31). These cells expressed a high level of membrane hSSTR5 (B max 800 Ϯ 90 fmol/mg protein) that existed in the basal state as a mixture of broad monomeric 55-65-kDa and dimeric 105-115-kDa bands (ratio 73 Ϯ 3%: 27 Ϯ 3%) (Fig. 3). Treatment with SST-14 resulted in a dose-dependent saturating increase in the proportion of dimers (monomer:dimer ratio 54 Ϯ 1%:46 Ϯ 1% with 1 M ligand). The dose-response curve for hSSTR5 dimerization paralleled that for ligand-induced receptor signaling (Fig. 2B).

Agonist-dependent Homodimerization of hSSTR5 in Intact
Cells by Photobleaching FRET Analysis-To obtain direct evidence for the association of SSTRs in intact cells, we probed for receptor homodimerization by pbFRET microscopy (35)(36)(37)(38)(39). HA hSSTR5 was visualized in CHO-K1 cells using fluorescein (donor)-and rhodamine (acceptor)-conjugated monoclonal antibody (mAb) against HA. Both fluorophore-tagged antibodies exhibited clear plasma membrane staining (Fig. 4I, a-c) as well as competitive antigen binding (Fig. 4I, d), from which their relative affinities could be determined. The decrease in donor fluorescence intensity due to photobleaching during prolonged exposure to excitation light was monitored in the absence (Fig.  4II, a) or presence (Fig. 4III, a) of acceptor, i.e. in the potential presence of an additional donor deactivation process, FRET. The photobleaching decay was analyzed for the plasma membrane regions, both on a pixel-by-pixel basis (Fig. 4, IIb and IIIb) as well as averaged over each image (Fig. 4II, c and d). We observed a significant slow down of the photobleaching process (as described by an increase in the photobleaching time constant) upon addition of rhodamine-labeled antibody to the cells suggesting that a large proportion of rhodamine molecules are in close enough proximity to fluorescein to act as acceptors for energy transfer. Given that the two fluorophores are associated with different receptor molecules, this finding suggests receptor association. In the basal state, we found effective FRET efficiencies of approximately 11 and 15% for donor:acceptor  (Table I). For a two-state model, i.e. for receptors existing either in a monomeric or dimeric state, as suggested by the Western blot data (Fig. 3), these FRET efficiencies relate to a minimal amount of receptor dimerization of 24 and 30%, respectively (39). Treatment with agonist resulted in increased FRET efficiencies of 21 and 27%, corresponding to higher levels of dimerization of at least 42 and 48% under saturation conditions (Table I), in good agreement with values obtained by Western blot.
To determine whether the high level of basal dimerization was caused by receptor overexpression, and to explain the relationship between monomers and dimers, we took advantage of the high sensitivity of FRET analysis for detecting dimerization to investigate a second CHO-K1 cell line expressing a 5-fold lower concentration of HA-hSSTR5 receptors (B max 160 Ϯ 30 fmol/mg protein). In contrast to cells overexpressing HA-hSSTR5, these cells displayed insignificant effective FRET efficiencies of 0 -1% in the basal state suggesting that monomers predominate in the absence of agonist when the receptor is expressed at levels in the range of endogenous SSTR concentrations (Table I) (40). Treatment with SST resulted in a dosedependent increase in FRET efficiencies (Fig. 5) suggesting that dimerization is induced by agonist binding.
Agonist-dependent Heterodimerization of hSSTR5 with hSSTR1-We next investigated SSTR heterodimerization and selected hSSTR1 and hSSTR5 to take advantage of their different pharmacological properties. Both receptors bind SST-14 and SST-28, but only SSTR5 and not SSTR1 binds the octapeptide SMS-(201-995) (SMS, Octreotide) (22). SSTR5 is internalized by acute agonist exposure, whereas hSSTR1 is resistant to internalization and is instead up-regulated at the membrane by prolonged agonist treatment (30,32,33). As previously Table I   The symbols used are as follows: D:A, concentration ratio donor/acceptor; D Ϫ A and D ϩ A, corresponding to donor in absence and presence of acceptor, respectively; avg , mean of n photobleaching time constants (each being the pixel-based average of a cell membrane), Ϯ S.E.; n, number of cells analyzed (with an average number of ϳ1500 pixels per cell); nϪ1 , standard deviation of avg ; E, average effective FRET efficiency; ␣ min , corresponding minimal amount of receptor dimerization. Absolute photobleaching time constants are not comparable between different data sets as they were measured on different days and were therefore affected by the decrease in excitation intensity of the UV lamp. reported, up-regulation of SSTR1 is time-and temperature-dependent, reaches saturation at 22 h, and is dependent on molecular signals in the cytoplasmic C-tail of the receptor (33). To test for heterodimerization, the ⌬318 hSSTR5 mutant which binds SMS was cotransfected with wt hSSTR1, which does not bind SMS (⌬318 hSSTR5, B max 112 Ϯ 12 fmol/mg protein; wt hSSTR1, B max 96 Ϯ 17 fmol/mg protein). We wondered whether the C-tail of hSSTR1 in this case would confer adenylyl cyclase responsiveness to the C-tail deletion mutant. Immunocytochemistry with antipeptide antibodies directed against the amino-terminal segment of hSSTR1 and hSSTR5 confirmed the expression of both receptor proteins in the plasma membrane of the majority of transfected cells (data not shown). Binding analysis showed a small but significant increase in the binding affinity of SST-14 for the cotransfectants (Fig. 6A). In contrast to the ⌬318 hSSTR5 mutant that shows no adenylyl cyclase coupling, the hSSTR1/⌬318 hSSTR5 cotransfectants displayed significant dose-dependent maximum 17 Ϯ 1.6% inhibition of forskolin-stimulated cAMP with 1 M SMS (Fig. 6B). A slightly greater 23 Ϯ 2.4% maximum inhibition was seen when SST-14, a common agonist, was applied. The reduced effect of SMS compared with SST-14 may be explained by putative ⌬318 hSSTR5 dimers that would be expected to bind but not inhibit adenylyl cyclase. The SSTR1-selective agonist SCH275 (41) inhibited adenylyl cyclase to a similar extent to SST-14 suggesting that the C-tail of hSSTR1 is the limiting factor in effecting maximum forskolin-stimulated cAMP response. In contrast to the ability of hSSTR1 to rescue adenylyl cyclase coupling of the ⌬318 hSSTR5 mutant, the related SSTR subtype hSSTR4, which is also SMS-insensitive (22), failed to heterodimerize with ⌬318 hSSTR5 in similar cotransfection experiments. This suggests that heterodimerization of SSTRs is a specific process that is restricted to some but not all receptor subtype combinations.

FIG. 5. Dose-dependent increase in effective FRET efficiency by SST-14 in CHO-K1 cells expressing relatively low density of HA-hSSTR5 (see
Agonist-dependent Internalization of hSSTR1 through Heterodimerization with hSSTR5-We next looked at the internalization of hSSTR1/⌬318 hSSTR5 cotransfectants using as ligands 125 I-Tyr 3 SMS, which binds ⌬318 hSSTR5 but not hSSTR1, and 125 I-SCH288, which is selective for SSTR1 (41). Compared with 55 Ϯ 5.6% internalization of 125 I-Tyr 3 SMS at 60 min by the ⌬318 hSSTR5 mutant alone, the cotransfectants showed 11 Ϯ 3.8% internalization of this ligand (Fig. 6C). hSSTR1 when expressed alone showed no internalization of its selective ligand 125 I-SCH288 consistent with the known inabil-ity of this receptor to undergo agonist-induced internalization as a monotransfectant (32,33). hSSTR1 cotransfected with ⌬318 hSSTR5, however, displayed 15 Ϯ 5% internalization of 125 I-SCH288 at 60 min. Since hSSTR1 alone cannot internalize 125 I-SCH288, the presence of this radioligand intracellularly must reflect internalization of hSSTR1/⌬318 hSSTR5 heterodimers. This was further demonstrated by confocal fluorescence immunocytochemistry using CHO-K1 cells cotransfected with wt hSSTR1 and HA-hSSTR5. Both receptors were colocalized on the plasma membrane of nonpermeabilized cotransfected cells (Fig. 7, a-c). As previously reported, hSSTR1 was predominantly localized over the cell surface when expressed alone in CHO-K1 cells (Fig. 7g) (33). The hSSTR1 cells permeabilized after 60 min treatment at 37°C with agonist SST-14 (1 M) showed very poor labeling of cytoplasmic vesicular structures (Fig. 7h) (33). In contrast, when hSSTR1 was cotransfected with HA-hSSTR5, it underwent internalization in the presence of agonist and was indeed colocalized with hSSTR5 in cytoplasmic vesicles of permeabilized cells (Fig. 7, d-f). These results suggest that although hSSTR1 does not internalize when expressed alone, it does so when coexpressed with an appropriate partner, in this case hSSTR5.
Agonist-dependent Up-regulation of hSSTR1 through Heterodimerization with hSSTR5-We further investigated agonist-induced up-regulation of hSSTR1 through heterodimerization with ⌬318 hSSTR5. hSSTR1 is up-regulated at the membrane by prolonged (22 h) exposure to SST-14 (33). SMS does not bind and therefore does not up-regulate this receptor (Fig. 8A). The ⌬318 hSSTR5 mutant bound both SST-14 and SMS, but neither induced up-regulation. Treatment of the coexpressed receptors with SMS 1 M for 22 h induced 110 Ϯ 16% up-regulation of cell surface binding comparable with that obtained with 1 M SST-14 (113 Ϯ 23%). Pharmacological analysis of the up-regulated receptors with radioligand selective for SSTR1 ( 125 I-SCH288) or SSTR5 ( 125 I-Tyr 3 SMS) showed 92 Ϯ 12.5% increase in 125 I-SCH288 binding without any change in 125 I-Tyr 3 SMS binding, thereby identifying hSSTR1 as the receptor subtype that was up-regulated at the cell surface by chronic SMS treatment (Fig. 8B). Since SMS does not bind SSTR1, its ability to up-regulate this receptor must be through binding to ⌬318 hSSTR5 and association with hSSTR1. Although cross-talk between the receptors at the level of signaling cannot be entirely excluded, this appears unlikely since deletion of the C-tail of hSSTR5 blocks signaling, at least via adenylyl cyclase (30). The ability of a ligand that interacts selectively with one SSTR subtype to induce up-regulation of another, noninteractive subtype was also demonstrated in the case of endogenous SSTRs (Fig. 8C). MCF-7 human breast cancer cells were found (by RT-PCR) (28) to express mRNA for several SSTRs with the following relative abundance (compared with actin mRNA): SSTR1 (ϩϩϩ), SSTR2 (ϩ), SSTR3 (Ϯ), SSTR4 (Ϫ), and SSTR5 (ϩϩϩ). Confocal fluorescence immunocytochemistry (26,27) confirmed the protein expression of SSTR1,2,3,5 (but not SSTR4) in these cells. Treatment of MCF-7 cells with 1 M SMS for 22 h induced a significant 109 Ϯ 15% increase in membrane SSTRs assessed by whole cell binding with 125 I-LTT-SST-28. Analysis of the up-regulation response with 125 I-SCH288 indicated surface recruitment of hSSTR1 (118 Ϯ 22% increase in binding) (Fig. 8C). Since SMS binds only to SSTR2,3,5, these results, taken together with those from the hSSTR1/⌬318 hSSTR5 cotransfection experiments, suggest functional cross-talk in MCF7 cells between SSTR1 and SSTR5 (and likely between SSTR1 and SSTR2) through receptor dimerization. DISCUSSION The existence of receptor dimers has been proposed for several GPCRs, based on studies of cross-linked or solubilized receptors or on functional complementation of mutant and chimeric receptors (5-10, 19 -21). However, the question of whether GPCR dimers are necessary for activating or modulating normal receptor function has remained unclear (16,21). By using FRET to monitor dimerization directly, as well as pharmacological and biochemical studies of both mutant and wild type receptors, here we provide strong evidence that members of the SSTR family, undergo agonist-dependent homo-and heterodimerization and that dimeric association alters SSTR functions such as ligand binding affinity, internalization, and up-regulation. We show that hSSTR5 forms heterodimers with hSSTR1 but not with hSSTR4 suggesting that heterodimerization of SSTRs is a specific process that is restricted to some but not all receptor subtype combinations.
The density of endogenous SSTR expression in receptor-rich tissues such as the brain, pituitary, pancreas, and adrenals in the rat measured with a nonspecific radioligand such as 125 I-LTT-SST-28 (which detects all five SSTR subtypes) ranges between 220 and 360 fmol/mg protein (40). In addition, because these tissues express all five SSTR isoforms, the concentration of individual subtypes is likely to be a fraction of this amount. To determine whether the level of receptor expression influences dimerization, we initially studied recombinant hSSTR5 overexpressed in CHO-K1 cells and found significant dimerization of this receptor in the basal state, both by Western blots and FRET analysis. At lower levels of transfection corresponding to endogenous SSTR concentrations, however, hSSTR5 oc-  (d-f). a, fluorescein immunofluorescent images showing HA-hSSTR5 localized on the plasma membrane (green). b, rhodamine immunofluorescent images of wt hSSTR1 localized on the plasma membrane (red). c, merged image to show colocalization of the two receptors on the plasma membrane (yellow). d-f, in permeabilized cells, hSSTR5 (d, green label) and hSSTR1 (e, red label) are colocalized (f, yellow image) in cytoplasmic vesicular structures. g and h, rhodamine fluorescence of hSSTR1 expressed alone in CHO-K1 cells. hSSTR1 is distributed on the plasma membrane (g) but does not appear in cytoplasmic vesicles in agonist-treated permeabilized cells (h). hSSTR1 is therefore localized intracellularly with hSSTR5 when coexpressed but not alone. curred only as a monomer in the basal state as determined by FRET. Activation by ligand induced receptor homodimerization in a dose-dependent manner. The parallel dose-response curves for ligand-induced dimerization and signaling by hSSTR5 suggest that dimerization is obligatory for receptor activation. Furthermore, hSSTR5 formed heterodimers with hSSTR1. This means that a given SSTR exists in different states as a monomer (probably inactive), a homodimer, or a heterodimer with one or more SSTR subtypes. Our results suggest that the use of high density receptor expression systems for detecting dimers by Western blots in several earlier studies may account for the high level of basal dimerization as an artifact of receptor overexpression and may help to explain some of the difficulties in interpreting the functional relationship between monomers and dimers (9,10,19,20). The structural requirements for GPCR dimerization are unknown although several dimerization interfaces have been proposed such as the extracellular amino-terminal domain for the glutamate and calcium-sensing receptors (42,43), the intracellular third loop and the VIth transmembrane domain for the ␤-adrenergic and dopamine receptors (9,15,16), and the C-tail for the GABA-B receptor (11)(12)(13)(14)21). In the case of SSTRs, the C-tail is clearly not required given the ability of SSTR1 and SSTR5 to form homo-and heterodimers with the C-tail deletion mutant of hSSTR5.
There are a number of functional consequences of dimerization by SSTR and other GPCRs. A given agonist may bind with different affinities to a given SSTR depending on its oligomeric configuration. A receptor may undergo regulatory responses in the absence of ligand, for instance by an agonist that binds selectively to one subtype and that modulates the internalization or up-regulation responses of another subtype(s) through heterodimerization. hSSTR1 internalized only as a heterodimer but not when expressed alone suggesting that coexpression of this receptor with hSSTR5 or possibly another subtype(s), as occurs endogenously, is a crucial determinant of its agonist-dependent regulatory responses. The ability of SMS to up-regulate endogenous hSSTR1 by binding to hSSTR5 as shown here may explain the clinical observation of why prolonged treatment of individuals with SMS leads to an escape from the acute effects of the drug (44). This is because upregulation of receptors such as hSSTR1 may compensate for the desensitized responses of other subtypes interacting with SMS to maintain normal SST responsiveness in target cells such as those in the pituitary and islets that coexpress all of these subtypes (26,27). Such direct protein interaction between different members of the SSTR subfamily, and possibly between SSTR and related receptor families, defines a new level of molecular cross-talk between GPCRs for greater functional diversity.