Identification of the Cysteine Residues in the Amino-terminal Extracellular Domain of the Human Ca2+ Receptor Critical for Dimerization

We analyzed the effect of substituting serine for each of the 19 cysteine residues within the amino-terminal extracellular domain of the human Ca2+ receptor on cell surface expression and receptor dimerization. C129S, C131S, C437S, C449S, and C482S were similar to wild type receptor; the other 14 cysteine to serine mutants were retained intracellularly. Four of these, C60S, C101S, C358S and C395S, were unable to dimerize. A C129S/C131S double mutant failed to dimerize but was unique in that the monomeric form expressed at the cell surface. Substitution of a cysteine for serine 132 within the C129S/C131S mutant restored receptor dimerization. Mutation of residues Cys-129, Cys-131, and Ser-132, singly and in various combinations caused a left shift in Ca2+ response compared with wild type receptor. These results identify cysteines 129 and 131 as critical in formation of intermolecular disulfide bond(s) responsible for receptor dimerization. In a “venus flytrap” model of the receptor extracellular domain, Cys-129 and Cys-131 are located within a region protruding from one lobe of the flytrap. We suggest that this region represents a dimer interface for the receptor and that mutation of residues within the interface causes important changes in Ca2+ response of the receptor.

The Ca 2ϩ receptor (CaR) 1 regulates extracellular calcium ion ([Ca 2ϩ ] o ) homeostasis by controlling the rate of parathyroid hormone secretion from the parathyroid gland and the rate of calcium reabsorption by the kidney (1). [Ca 2ϩ ] o activates the CaR, leading to activation of phospholipase C␤ via the G q subfamily of G-proteins; this increases phosphoinositide (PI) hydrolysis and causes release of Ca 2ϩ from intracellular stores (2). Recent evidence suggests that the CaR is also involved in diverse cellular responses to extracellular Ca 2ϩ within microenvironments in other organs such as brain, skin, bone, and intestine (3).
Recently, it has been shown that both the mGluRs (13-15) and the CaR (12, 16 -19) are expressed at the cell surface as intermolecular disulfide-linked dimers. For mGluR1 (14), mGluR4 (15), and the CaR (12), the ECD of each receptor, purified as a secreted protein, exists as a disulfide-linked dimer, suggesting that one or more cysteines in the ECD is involved in receptor dimer formation. However, the cysteine(s) forming intermolecular disulfide bond(s) in the CaR or mGluR ECD have not yet been identified. Proteolysis of the mGluR5 receptor localized cysteine(s) critical for dimer formation to the first 17 kDa of the ECD (13). This region contains three cysteines conserved in all mGluRs and in the CaR. The human CaR (hCaR) ECD contains a total of 19 cysteines (20) all of which are highly conserved in bovine (2), rat (21,22), and rabbit (23) CaRs, and all but cysteine 482 (hCaR sequence numbering) are conserved in the chicken CaR (24). We showed previously that individual cysteine 3 serine mutations of 14 of these 19 cysteines (all but cysteines 129, 131, 437, 449, and 482) abolish or drastically reduce receptor cell surface expression and/or function, likely by causing misfolding and improper processing of the receptor (18). In the present study, we performed a detailed analysis of ECD cysteines responsible for dimer formation. We found that mutation of both cysteines 129 and 131, but not mutation of either alone, blocks dimer formation. Unlike other ECD cysteine mutations, however, mutation of both cysteines 129 and 131 results in a monomeric form of the hCaR expressed at the cell surface and with unique functional properties.

MATERIALS AND METHODS
Site-directed Mutagenesis of the hCaR-Site-directed mutagenesis was performed on hCaR cDNA in the pCR3.1 vector using a commercial kit (QuikChange TM site-directed mutagenesis kit, Stratagene Inc., La * 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. § To whom correspondence should be addressed: NIDDK, National Institutes of Health, 10/9N-222, Bethesda, MD 20892. Fax: 301-496-9943; E-mail: allens@amb.niddk, nih.gov. 1 The abbreviations used are: CaR, Ca 2ϩ receptor; hCaR, human Ca 2ϩ receptor; GPCR, G-protein-coupled receptor; [Ca 2ϩ ] o , extracellular calcium ion; PI, phosphoinositide; mGluRs, metabotropic glutamate receptors; VNR, vomeronasal organ receptors; TR, taste receptors; ECD, extracellular domain; TM1, hCaR truncation mutant containing ECD and 1st transmembrane domain; LIVBP, leucine/isoleucine/valine bacterial periplasmic binding protein; N-linked, asparagine-linked; HEK-293 cells, human embryonic kidney-293 cells; DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; PBS, phosphatebuffered saline; PAGE, polyacrylamide gel electrophoresis; Endo-H, endo-␤-N-acetylglucosaminidase H; BSA, bovine serum albumin; PIPES, 1,4-piperazinediethanesulfonic acid; Biotin-7-NHS, D-biotinoyl-⑀-aminocaproic acid-N-hydroxysuccinimide ester; GABA B , ␥-aminobutyric acid, type B; WT, wild type. Jolla, CA), according to the manufacturer's instructions. Briefly, a pair of complementary primers with 25-35 bases was designed for each mutagenesis, and the mutation to change cysteine to serine or alanine was placed in the middle of the primers. Parental hCaR inserted in pCR3.1 was amplified using Pyrococcus furiosus DNA polymerase with these primers for 12 cycles in a DNA thermal cycler (Perkin-Elmer). After digestion of the parental DNA with DpnI, the amplified DNA with the nucleotide substitution incorporated was transformed into Escherichia coli (DH-5␣ strain). The mutations were confirmed by automated DNA sequencing using a Taq DyeDeoxy Terminator Cycle Sequencing kit and ABI prism-377 DNA sequencer (Applied Biosystems, Inc., Foster City, CA). Most of the 19 single site cysteine to serine (Cys 3 Ser) mutants used in this study were described earlier (18). The C129S and S132C mutants and the Cys 3 Ala mutants including C60A, C101A, C358A, and C395A were newly generated. The C129S/C131S and C129A/C131A double mutants and the C129S/C131S/S132C triple mutant were created by changing cysteine at a given site to the desired amino acid and using this mutant DNA as template in the next round of mutagenesis. A truncation mutant containing the ECD and 1st transmembrane domain (henceforth termed TM1) was generated by introducing a stop codon in the first intracellular domain of the WT hCaR clone at amino acid position lysine 644. For all the newly generated mutants, we confirmed that two independent clones of the same mutant receptor cDNA showed identical properties.
Transient Transfection of Wild Type and Mutant Receptors in HEK-293 Cells-For transfection, a given amount of the plasmid DNA was diluted in Dulbecco's modified Eagle's medium (DMEM) (BioFluids Inc., Rockville, MD), mixed with diluted LipofectAMINE, and the mixture was incubated at room temperature for 30 min. The DNA-Lipo-fectAMINE complex was further diluted in serum-free DMEM, and 8 -15 g of DNA was added to 80 -90% confluent HEK-293 cells plated in 75-cm 2 flasks. For 6-well plates, transfections were performed using 2 g of DNA for single plasmid transfection or 1 g of DNA of each of two different plasmids in cotransfection experiments, with 25 l of LipofectAMINE per well at dilutions described above. After 5 h of incubation, equal volume of DMEM containing 20% fetal bovine serum (FBS) (BioFluids Inc., Rockville, MD) was added, and the media were replaced 24 h after transfection with complete DMEM containing 10% FBS. Membrane protein extraction for immunoblotting, whole cell enzyme-linked immunoassay, or PI hydrolysis assay were performed 48 h after transfection.
Biotinylation of the Cell Surface hCaR-48 h after transfection, cell surface proteins of the intact HEK-293 cells were labeled with membrane-impermeant Biotin-7-NHS using the cellular labeling kit (Roche Molecular Biochemicals). Briefly, adherent cells were washed once with ice-cold phosphate-buffered saline (PBS) and treated with 50 g/ml Biotin-7-NHS in biotinylation buffer (50 mM sodium borate, 150 mM NaCl) for 15 min at room temperature to biotinylate cell surface proteins. The reaction was stopped by adding 50 mM NH 4 Cl for 15 min on ice. The cells were washed twice with ice-cold PBS and solubilized with 1 ml of buffer B per well containing 1% Triton X-100, 20 mM Tris-HCl (pH 6.8), 150 mM NaCl, 10 mM EDTA, 1 mM EGTA with freshly added protease inhibitor mixture.
Immunoprecipitation of hCaR Receptors-300 l (approximately 600 g of total protein) of the whole cell lysate prepared by scraping cells from 6-well plates in buffer B as described above was further diluted with 300 l of buffer B and incubated with either 5 l of 7F8 mouse monoclonal hCaR-specific antibody (made against the purified hCaR ECD (12); 1 mg/ml stock) or 7 l of affinity purified rabbit polyclonal hCaR-specific antibody GGD (made against a synthetic peptide corresponding to amino acids 1037-1050 of the hCaR protein; 1 mg/ml stock) for 1-2 h at 4°C. Subsequently, 25 l of protein A/G (for 7F8) or protein A (for GGD)-agarose (Santa Cruz Biotechnology, Santa Cruz, CA) was added, and the incubation was continued for an additional 1-2 h. The protein A/G-or A-agarose was washed three times with buffer B containing 0.5% SDS, and the immunoreactive proteins were eluted in 120 l of 1ϫ sample buffer containing either no ␤-mercaptoethanol or 300 mM ␤-mercaptoethanol at room temperature for 5 min. 50 l of sample was loaded per lane, and immunoblotting was performed as described below.
Immunoblotting Analyses with Detergent-solubilized Whole Cell Extracts-Confluent cells in 75-cm 2 or 6-well plates were rinsed with ice-cold PBS and scraped on ice in buffer B containing 20 mM Tris-HCl (pH 6.8), 150 mM NaCl, 10 mM EDTA, 1 mM EGTA, 1% Triton X-100 with freshly added protease inhibitors mixture. The protein content of each sample was determined by the modified Bradford method (Bio-Rad), and 40 -60 g of protein per lane was separated on 5% SDS-PAGE. The proteins on the gel were electrotransferred to nitrocellulose membrane and incubated with 0.1 g/ml protein A-purified mouse monoclonal anti-hCaR antibody ADD (raised against a synthetic peptide corresponding to residues 214 -235 of hCaR protein (25)). Subsequently, the membrane was incubated with a secondary goat antimouse antibody conjugated to horseradish peroxidase (Kirkegaard and Perry Laboratories, Gaithersburg, MD) at a dilution of 1:5000. The hCaR protein was detected with an enhanced chemiluminescence system (Amersham Pharmacia Biotech Corp.). Biotinylated proteins were detected using peroxidase-conjugated streptavidin followed by visualization of the biotinylated bands using BM chemiluminescence kit (Roche Molecular Biochemicals).
Treatment of Detergent-solubilized Crude Cell Membrane Extracts with Endoglycosidase H-For cleavage with endoglycosidase H (Endo-H) (Roche Molecular Biochemicals), cell extracts (20 l) were diluted in 20 l of 50 mM sodium acetate (pH 4.8). Samples were incubated with 0.5 milliunits of Endo-H for 2 h at 37°C.
Intact Cell Enzyme-linked Immunoassay to Determine Cell Surface Expression-This method has been described (26). Briefly, intact transfected HEK-293 cells in 75-cm 2 flasks were detached with 1 mM EDTA in PBS containing 0.5% bovine serum albumin and incubated with 0.5 ml of DMEM containing 10% FBS and 1 g/ml monoclonal anti-hCaR antibody 7F8 at 4°C for 2 h. After incubation in peroxidase-conjugated anti-mouse secondary antibody, cells were washed and, after adding peroxidase substrate, precipitated by centrifugation; absorbance of the supernatant was measured at 405 nm using a Thermo max microtiter plate reader (Molecular Devices, Sunnyvale, CA).
PI Hydrolysis Assay-PI hydrolysis assay has been described (26,27). Briefly, 24 h after transfection, transfected cells from a confluent 75-cm 2 flask were replated in two 12-well plates in medium containing 3.0 Ci/ml [ 3 H]myoinositol (NEN Life Science Products) in complete DMEM for another 24 h, followed by 1-h incubation with 1ϫ PI buffer (120 mM NaCl, 0.5 mM CaCl 2 , 5 mM KCl, 5.6 mM glucose, 0.4 mM MgCl 2 , 20 mM LiCl in 25 mM PIPES buffer, pH 7.2). After removal of PI buffer, cells were incubated for an additional 1 h with different concentrations of [Ca 2ϩ ] o in PI buffer. The reactions were terminated by addition of 1 ml of acid/methanol (1:1000 v/v) per well. Total inositol phosphates were purified by chromatography on Dowex 1-X8 columns.
Molecular Modeling-The amino acid sequence for the hCaR ECD (20) was aligned with that of the rat mGluR1 (20) and the 344-residue E. coli leucine/isoleucine/valine periplasmic binding protein (LIVBP) based on the original alignment performed by O'Hara et al. (11). A model for the three-dimensional structure of this portion of the hCaR ECD was generated using the SEGMOD algorithm (28) of the program LOOK-version 3.5 (29) using the LIVBP coordinates as a template from the Brookhaven Protein Data base. The modeled coordinates were oriented in RASMOL (30) and depicted using MOLSCRIPT (31) and RASTER3D (32).

Intermolecular Disulfide Bridge Involves ECD of the hCaR-
To determine the role of the ECD in dimer formation of the membrane-bound hCaR, a mutant construct (TM1) containing the whole ECD and first transmembrane domain of hCaR and truncated at lysine 644 in the first intracellular loop was prepared and transiently expressed in HEK-293 cells. Cell surface proteins were labeled with membrane-impermeant Biotin-7-NHS prior to lysing the cells as described before (33). To prevent nonspecific disulfide bond formation during protein extraction, the intact cells were incubated and washed in PBS containing 50 mM iodoacetamide, and 10 mM iodoacetamide was included in the lysis buffer. Both the wild type hCaR and TM1 were then immunoprecipitated with receptor-specific 7F8 monoclonal antibody and eluted with gel loading sample buffer either containing ␤-mercaptoethanol as reducing agent or with no ␤-mercaptoethanol. Immunoprecipitates were run on SDS-PAGE and analyzed on immunoblots stained either with streptavidin to detect biotinylated cell surface proteins or with anti-hCaR monoclonal antibody ADD to detect total hCaR immunoreactive species.
As shown in Fig. 1 (ADD blot, 1st lane), under nonreducing conditions, ADD antibody detected two major dimeric bands of hCaR ϳ260 -300 kDa in size; ϳ150and 130-kDa monomeric forms appeared only after reducing the samples (ADD blot, 3rd lane). Previous studies have shown that the monomeric ϳ150-kDa band represents hCaR forms expressed at the cell surface and modified with N-linked, complex carbohydrates; the ϳ130-kDa band represents high mannose-modified forms, trapped intracellularly and sensitive to Endo-H digestion (26,33,34). showing that only the upper form of TM1 is expressed at the cell surface, as with the wild type receptor. These results indicate that TM1 is capable of forming a dimer expressed at the cell surface like the wild type hCaR but lacking the full seven transmembrane domain does not stimulate PI hydrolysis in response to [Ca 2ϩ ] o (data not shown). The ability of TM1 to form dimers, together with our previous observation that the secreted, purified ECD of the hCaR is a disulfide-linked dimer (12), indicates that the determinants including the cysteines important for dimer formation are present in the ECD of the hCaR.
Screening of Single Cysteine Mutants for Their Ability to Form Homodimers or Heterodimers with TM1-We previously generated single Cys 3 Ser mutants of all 19 cysteines in the hCaR ECD (18). We now coexpressed each of these Cys 3 Ser mutants and the wild type hCaR with the TM1 mutant and tested for their ability to heterodimerize with TM1 in a coimmunoprecipitation assay. A polyclonal antibody "GGD" made against a peptide from the carboxyl-terminal tail region of the hCaR is able to immunoprecipitate full-length wild type hCaR and full-length Cys 3 Ser mutants but not TM1 because TM1 lacks the GGD epitope. After immunoprecipitation with GGD antibody, samples were run on SDS-PAGE under reducing conditions and immunoblotting performed with ADD antibody whose epitope within the ECD is contained in both TM1 and full-length forms of hCaR. GGD antibody fails to immunopre-cipitate TM1 when it is transfected by itself, as no immunoreactivity is detected on ADD immunoblots of such immunoprecipitates (data not shown). Fig. 2A shows the results for wild type and seven of these mutants. When coexpressed with wild type hCaR, TM1 immunoreactivity is detected with ADD on blots of the immunoprecipitate. Both the 95-and 85-kDa bands of TM1 were detected under reducing conditions as were the 150-and 130-kDa forms of the wild type hCaR ( Fig. 2A, 1st  lane). These results indicate that TM1 and wild type hCaR heterodimerize, allowing TM1 to be coprecipitated with wild type hCaR by GGD antibody. We suggest that the upper forms of TM1 and wild type detected on ADD blot reflect heterodimers expressed at the cell surface and the lower forms, heterodimers of the respective incompletely processed, intracellular forms of TM1 and wild type hCaR. Similarly, with C129S and C131S mutants, both 95-and 85-kDa forms of TM1 were coimmunoprecipitated along with 150-and 130-kDa forms of the C129S and C131S mutant receptors. C236S mutant expressed primarily as the 130-kDa form and coimmunoprecipitated mainly with the lower 85-kDa form of TM1. A small amount of the upper 95-kDa TM1 monomeric band was detected corresponding to the faint 150-kDa band detected for C236S ( Fig. 2A, 6th lane). In contrast, C60S, C101S, C358S, and C395S failed to immunoprecipitate the 95-kDa form of the TM1 mutant and little if any of the 85-kDa form. Each of these mutants was expressed primarily as the incompletely processed, 130-kDa form.
The Cys 3 Ser mutants were further analyzed by determining their homodimerization patterns on ADD immunoblots run under nonreducing conditions. Cells were treated with iodoacetamide as described under "Materials and Methods" to prevent aggregates forming secondary to nonspecific disulfide bond formation. Fig. 2B shows that C60S, C101S, C358S, and C395S mutant receptors remained mostly as a 130-kDa monomeric form and showed very little or no dimeric forms. In contrast, wild type hCaR, C129S, and C131S mutant receptors formed two homodimeric bands with little or no monomeric forms visible on immunoblot. C236S mutant receptor showed a strong dimeric band with mobility differing from either wild type hCaR, C129S, or C131S mutant receptors. Taken together, the results suggest that substituting serine (or alanine; data not shown) for cysteines 60, 101, 358, or 395 may directly or indirectly block dimerization, whereas serine substitution for either Cys-129 or Cys-131 has minimal effect on dimer formation. Substitution of serine for cysteine 236 does not block dimer formation, but the conformation of the C236S dimeric forms based on different mobility on SDS-PAGE appear to differ from those of wild type hCaR. Of the other Cys 3 Ser mutants, C447S, C449S, and C482S showed essentially the same pattern of hetero-and homodimerization as the wild type hCaR. The C542S, C546S, C561S, C562S, C565S, C568S, C582S, C585S, and C595S mutant receptors showed similar heterodimerization patterns with TM1 as the C236S mutant; their homodimerization patterns also resembled C236S with variable degrees of dimer mobility differences from wild type hCaR on non-reducing SDS-PAGE (data not shown).

Cys 3 Ser Mutation of Both Cys-129 and Cys-131 Blocks Functional Dimer Formation and Generates a Monomeric Form of hCaR Expressed at the Cell Surface-
The first 17 kDa of the mGluR5 was shown to be critical for dimer formation (13). The mGluRs all have three conserved cysteines within this region. Two correspond to Cys-60 and Cys-101 of the hCaR, but the hCaR has two cysteines, 129 and 131, in the position corresponding to the third mGluR-conserved cysteine (20). This led us to examine the possibility that the lack of effect of serine substitution for either Cys-129 or Cys-131 on hCaR dimer formation could be due to the ability of the remaining, nearby cysteine to substitute in a putative intermolecular disulfide bond for that mutated to serine. We therefore created a C129S/ C131S double mutant and compared its expression and dimer formation pattern with wild type, C129S, and C131S mutant receptors. As seen in Fig. 3A (ADD blot), under nonreducing conditions, a pair of immunoreactive bands in the Ͼ200-kDa size range is detected for the C129S/C131S double mutant as for the wild type and C129S and C131S mutants. The mobility of these bands, however, differs from that of the two major dimeric bands of the wild type hCaR or C129S or C131S mutant receptors. Moreover, whereas wild type hCaR showed no monomeric forms and C129S and C131S showed only a small amount of the intracellular 130-kDa monomeric form under nonreducing conditions, the C129S/C131S double mutant generated significant amounts of both 150-and 130-kDa monomeric forms. Streptavidin blot under nonreducing conditions (Fig. 3A, Biotin-Strep) showed that the wild type hCaR, C129S, and C131S mutants are expressed at the cell surface as the ϳ300-kDa upper dimer form. In contrast, the C129S/C131S double mutant was expressed at the cell surface only as the monomeric 150-kDa form. Importantly, neither of the apparent dimeric forms detected for the C129S/C131S double mutant with ADD appear to be expressed at the cell surface as judged by lack of streptavidin staining.
This suggested that the dimeric forms detected for the C129S/C131S double mutant could represent an aggregate of improperly processed forms that do not reach the cell surface.
To determine further the biochemical identity of the ϳ300-kDa dimeric band of the C129S/C131S double mutant, we tested for sensitivity to Endo-H digestion to distinguish between the fully processed hCaR forms that are modified with complex carbohydrates (Endo-H-resistant) and high mannose-modified forms (Endo-H-sensitive) that have not trafficked from the endoplasmic reticulum to the Golgi (26,33,34). As shown in Fig. 3B, for samples run under nonreducing conditions, digestion with Endo-H caused a decrease in the size of the lower dimeric form of the wild type hCaR, and the upper form remained mostly resistant to Endo-H digestion. For the C129S/C131S double mutant, however, the ϳ300-kDa dimeric and 130-kDa monomeric forms were sensitive to Endo-H digestion, whereas the 150-kDa form was resistant. When these samples are analyzed under reducing conditions, the upper, monomeric 150-kDa band of both the wild type hCaR and C129S/C131S double mutant is Endo-H-resistant, whereas the respective 130-kDa monomeric forms are Endo-H-sensitive. These data strongly suggest that the ϳ300-kDa band of the C129S/C131S double mutant represents an aggregate that remains intracellularly trapped.
To assess further the ability of the C129S/C131S double mutant to form dimers, the double mutant and TM1 mutant receptors were coexpressed in HEK-293 cells, and coimmunoprecipitation was performed with GGD antibody as described above. As shown in Fig. 3C, as for wild type hCaR, both the 150and 130-kDa monomeric forms of the C129S/C131S double mutant were detected by ADD after GGD immunoprecipitation. Unlike for the wild type hCaR, however, the 95-kDa form of TM1 fails to coprecipitate with the double mutant receptor, and only a small amount of the lower monomeric 85-kDa form of TM1 coprecipitates. This result indicates that despite the fact that the monomeric 150-kDa form of the C129S/C131S double mutant is fully processed and resistant to Endo-H digestion (Fig. 3B) and expressed at the cell surface (Fig. 3A), it is incapable of forming a heterodimer with the fully processed 95-kDa form of TM1. The small amount of the 85-kDa intracellular form of TM1 that does coprecipitate with the C129S/ C131S double mutant presumably reflects interaction (aggregation?) with the form of the double mutant that remains intracellularly trapped and is seen as an ϳ300-kDa band under nonreducing conditions. A C129A/C131A double mutant showed the same changes seen with the C129S/C131S double mutant (data not shown) indicating that loss of cysteine rather than specifically serine substitution was responsible for the changes observed.
The inability of the C129S/C131S double mutant to dimerize compared with the relatively unimpaired ability of either C129S or C131S mutants to dimerize suggested that a cysteine residue in either position is sufficient to form an intermolecular disulfide bond critical for dimerization. To explore further the requirements for CaR dimerization, we constructed a triple mutant receptor, C129S/C131S/S132C, in which the serine normally found at amino acid position 132 is changed to a cysteine in the context of the C129S/C131S double mutant. Fig. 4 shows that under nonreducing conditions, ADD antibody detects two major dimeric bands of the C129S/C131S/S132C triple mutant receptor like the wild type hCaR, and both triple mutant receptor and wild type show little or no monomeric forms in contrast to the C129S/C131S double mutant which generates significant amounts of both the 150-and 130-kDa monomeric forms. The streptavidin blot (Fig. 4) also shows that like the wild type hCaR, the upper dimeric form of the C129S/C131S/ S132C triple mutant is expressed at the cell surface. This contrasts with the C129S/C131S double mutant which shows the 150-kDa monomer form but not the dimeric form expressed at the cell surface.
Function of Cysteine Mutants in PI Hydrolysis Assay-Because the monomeric form of the C129S/C131S double mutant reaches the cell surface, we tested whether the mutant receptor is capable of signal transduction using the intact cell [Ca 2ϩ ] ostimulated PI hydrolysis assay. Since we sought to compare the signaling properties of the single cysteine mutants, C129S and C131S, and the C129S/C131S double mutant at similar expression levels as the wild type receptor, we first determined the levels of cell surface expression for each receptor by whole cell enzyme-linked immunoassay after transfecting HEK-293 cells with equal amounts of plasmid DNA. We found that for a given amount of plasmid DNA transfected, the C129S/C131S double mutant showed a reduced cell surface expression compared with wild type hCaR or other mutants used in this study (data not shown). To assess comparable levels of expression, we varied the amount of plasmid DNA used for transfection, and we achieved comparable levels of expression by transfecting HEK-293 cells with 12 g of C129S/C131S double mutant receptor DNA and 8 g of wild type hCaR or other mutant hCaR DNA. Intact cell enzyme-linked assay (Fig. 5, lower inset) showed that cell surface expression was comparable for wild type and mutant receptors transfected with these DNA amounts. Streptavidin blot (Fig. 5, upper inset) confirmed that all the mutants expressed at the cell surface as dimers except for the monomeric C129S/C131S double mutant. We then compared the PI hydrolysis response to [Ca 2ϩ ] o of all the mutant receptors, including an S132C mutant, with the wild type hCaR ( Whole cell extracts were immunoprecipitated with monoclonal antibody 7F8 and then fractionated on 5% SDS-PAGE under non-reducing conditions (NR). Total hCaR immunoreactivity was detected by immunoblot with ADD antibody and cell surface-expressed forms by streptavidin blot. Molecular mass standards are indicated at the right of the blots. C129S/C131S, 1.0 Ϯ 0.2; C129S/C131S/S132C, 2.2 Ϯ 0.2; S132C, 1.9 Ϯ 0.1.

DISCUSSION
The major reduction in size observed on SDS-PAGE following disulfide reduction for both the CaR and mGluRs suggests that these receptors are dimers linked by one or more intermolecular disulfide bonds (13, 16 -18). The cysteine(s) involved in the relevant intermolecular disulfide bond(s) are localized to the ECD of the CaR and mGluRs (12,14,15), and in mGluR5 are localized within the amino-terminal 17 kDa of the ECD (13). To identify the specific cysteine(s) in the CaR ECD involved in intermolecular disulfide linkage, we tested the ability of a series of cysteine mutants to homodimerize and to heterodimerize with a truncation mutant, TM1. TM1, truncated within the first intracellular loop, was found to express well at the cell surface, in contrast to mutants truncated within the second or third intracellular loops which we previously showed fail to be processed normally and fail to reach the cell surface (26). This difference suggests that the number of transmembrane domains in a given CaR construct (one and seven for TM1 and wild type, respectively; three and five for second and third intracellular loop truncation mutants, respectively) is critical for folding and normal processing. TM1 also was able to homodimerize, in agreement with recently reported results for a similar truncation mutant (19), and to heterodimerize with wild type CaR. The dimerization ability of TM1 is further evidence for the importance of the ECD in CaR dimer formation.
Only C60S, C101S, C358S, and C395S of the single cysteine to serine mutants we tested showed substantial reduction in homodimerization and in heterodimerization with TM1. Because these cysteine mutant receptors are expressed primarily as incompletely processed 130-kDa monomers, it is possible that mutation of these cysteines blocks dimerization by causing misfolding of the protein rather than because such cysteines are directly involved in intermolecular disulfide-linked dimer formation. Our results are similar to those recently reported in another study (19) for C131S (similar to wild type) and C101S (reduced total expression and reduced ability to dimerize) but differ importantly for several other cysteine mutants. Unlike our results, that study reported that the C60S was similar in all respects to wild type and that the C236S mutant was largely unable to dimerize. These authors also reported that a C101S/ C236S double mutant was well expressed and appeared exclusively as a monomer. Since we did not study a C101S/C236S double mutant, we cannot comment on the results with that mutant. We are unable to explain the differences seen for mutants such as C60S and C236S examined in both studies except that the other study, unlike ours, involved green fluorescent protein-tagged CaR constructs and did not directly assess cell surface expression as we did with the biotinylation method. Given the similarity between the CaR and mGluRs and the evidence for mGluR5 that the first 17 kDa of the ECD is the region critical for dimerization, we decided to test the effect of substituting serine for both Cys-129 and Cys-131. The resultant double mutant failed to dimerize, like C60S, C101S, C358S and C395S mutants, but unlike these, C129S/C131S formed a  6. A, alignment of a portion of the amino acid sequence of the CaR ECD and mGluR1 ECD with that of the LIVBP. The entire 344-residue LIVBP sequence is shown aligned with a portion of the sequence of the rat mGluR1 ECD and the human CaR ECD. Amino acid numbering on the right refers to the full-length LIVBP, mGluR1 and hCaR. Identical amino acid residues are shown in bold. The alignment places Gly-36 of the hCaR at residue Glu-1 of LIVBP and ends with Val-513 of the hCaR. Four insertions in the mGluR1 and CaR sequences that do not align with LIVBP are labeled I-IV. In the largest of these insertions, III, the hCaR sequence from Phe-347 to Ser-403 has been omitted. The secondary structure (arrows, ␤ sheet; cylinders, ␣ helix; thin lines, turns and loops) is superimposed above the alignment and is based on the three-dimensional structure of LIVBP. B, venus flytrap model of the three-dimensional structure of part of the ECD of the hCaR monomer. A ribbon diagram of the hCaR ECD model is shown with ␣ helices in red, ␤ sheets in yellow, and loops and turns in cyan. N represents the amino-terminal amino acid monomer that was normally processed (Endo-H resistance) and expressed at the cell surface (biotinylation experiment). These results strongly suggest that mutation of both Cys-129 and Cys-131 does not block dimerization by causing misfolding and abnormal processing of the receptor but rather by disrupting intermolecular disulfide linkage. The results do not allow us to distinguish between three possibilities as follows: (a) dimer formation involves intermolecular disulfide linkage between both Cys-129 and Cys-131 on respective monomers; mutation of either cysteine fails to block dimer formation because the intermolecular disulfide bond between the remaining cysteines is sufficient to maintain dimerization; (b and c) dimer formation normally involves a single intermolecular disulfide bond between either Cys-129 or Cys-131 and their counterparts on the other monomer; mutation of either cysteine fails to block dimer formation either because that cysteine is ordinarily uninvolved in intermolecular disulfide linkage or because the nearby, unmutated cysteine substitutes for the mutated one in disulfide bond formation. It is interesting in this respect that creation of a cysteine at adjacent position 132, normally a serine, restores normal dimer formation to the monomeric C129S/C131S mutant. This suggests that residues 129, 131, and 132 are all located within a putative dimer interface that permits intermolecular disulfide bond formation.
Mutations at these positions, singly or in combination, cause a significant left shift in receptor sensitivity (Fig. 5). The EC 50 for [Ca 2ϩ ] o is reduced by a factor of 2 compared with wild type in each of the mutants except for the C129S/C131S double mutant whose EC 50 is reduced nearly 4-fold. In our previous study of individual C129S and C131S mutants (18), we failed to discern a clear left shift in comparison with wild type in part because we did not measure response at several [Ca 2ϩ ] o concentrations Ͻ2.0 mM. The left shift in sensitivity seen with the mutants is not due to lack of dimer formation since only the C129S/C131S double mutant fails to dimerize. Interestingly, five mutations in the hCaR (A116T, N118K, L125P, E127A, and F128L) shown to cause a left shift in receptor sensitivity and identified in subjects with the disease autosomal dominant hypocalcemia (3) are located close to Cys-129 and Cys-131 in the putative dimer interface. None of these mutations disrupted dimer formation (data not shown). The ability of so many different missense mutations involving residues between 116 and 132 of the hCaR to enhance sensitivity to [Ca 2ϩ ] o suggests that this region is involved in some critical but as yet undefined way in receptor activation. The more profound increase in sensitivity seen with the monomeric C129S/C131S double mutant may be a result of an inability to form the dimer. Note also (Fig. 5) that despite its enhanced sensitivity to [Ca 2ϩ ] o , the C129S/C131S double mutant fails to reach wild type levels of activation even at the highest concentrations tested.
A recent study (35) provided functional evidence for the importance of CaR dimer formation in that coexpression of inactive but cell surface-expressed mutants led to heterodimer formation and some reconstitution of activity. The present results show that dimer formation is not essential for CaR function per se but may be essential for normal activation. The differences in activity between wild type dimer and the C129S/ C131S mutant monomer lead us to speculate that dimer formation may constrain activation by low [Ca 2ϩ ] o concentrations. Missense mutations at a number of residues within the dimer interface reduce this constraint resulting in receptor activation at "inappropriately" low agonist concentrations, as observed in subjects with autosomal dominant hypocalcemia. Complete loss of dimer formation in the C129S/C131S mutant causes an even more extreme left shift in receptor sensitivity but appears to compromise the maximum level of signal transduction achieved.
It is instructive to consider the present results in the context of a model of the three-dimensional structure of the hCaR ECD (Fig. 6). This model is based on the original alignment of mGluRs with bacterial periplasmic binding proteins by O'Hara et al. (11), a reasonable extrapolation given the high sequence identity between CaR and mGluRs (2,20). The alignment (Fig.  6A) places glycine 36 of the hCaR at residue 1 of LIVBP and ends with valine 513 of the hCaR. There are four insertions in both the CaR and the mGluR sequence (labeled I-IV in Fig. 6) that cannot be aligned with the LIVBP, and hence their structure cannot be modeled. The model shows the hCaR ECD monomer as a bilobed, venus flytrap-like structure with three strands connecting the two globular lobes each of which consists of ␤ sheets flanked by ␣ helices (Fig. 6B). The four insertions are all contained in the amino-terminal lobe 1 as in the mGluR model (11).
The carboxyl terminus in the model corresponds to a part of the ECD just before the cysteine-rich region that contains 9 of the 19 cysteines in the CaR ECD. Interestingly, the GABA B receptor ECD that completely lacks the cysteine-rich region found in other GPCR family 3 members including CaR, mGluRs, TR, and VNR has recently also been modeled as a venus flytrap-like structure, consistent with the idea that the cysteine-rich region represents a separate domain of the ECD (10). Mutation of any of the nine cysteines of the hCaR cysteine-rich region did not block dimer formation but led to intracellularly trapped proteins that were incompletely processed. We speculate that mutation of these cysteines likely causes misfolding of the receptor, perhaps by disrupting intramolecular disulfide bond formation, but this occurs only after dimers involving the venus flytrap-like portion of the ECD have already been formed.
Of the remaining 10 cysteines within the ECD, only Cys-236 is located within the carboxyl-terminal lobe 2 of the flytrap model (Fig. 6B). Mutation of this residue does not impair dimer formation but causes expression of an intracellularly trapped, presumptively misfolded protein. On the basis of the model, in which Cys-236 is remote from other cysteines in the venus flytrap structure, it appears unlikely that Cys-236 is involved in intramolecular disulfide bond formation with other cysteines in the venus flytrap. We cannot, however, exclude its involvement in formation of a disulfide with one of the nine cysteines of the cysteine rich-region. Thus misfolding caused by mutation of Cys-236 could either be due to disruption of an intramolecular disulfide or to a more subtle effect of substitution of serine for this cysteine.
In contrast to Cys-236, Cys-60, Cys-101, Cys-358, Cys-395, Cys-437, Cys-449, and Cys-482 are all located within the amino-terminal lobe 1 of the flytrap. Mutation of the first four of these residues does block dimer formation. We speculate that this is due to misfolding of lobe 1 and prevention of formation of the dimer interface region (region II in the model) that normally protrudes from this lobe. These residues may be involved in intramolecular disulfide bonds critical for achieving Gly-36, and C represents the carboxyl-terminal amino acid Val-513. I-IV designate the four regions of the hCaR ECD that do not align with LIVBP (see A). These "insertions" in the hCaR ECD sequence are shown in the model as purple loops, except for III in which only the first (His-338) and last (Asp-410) residues colored beige and golden, respectively, are shown. The positions of individual cysteines found in the model are shown in green (numbered according to hCaR sequence), and the putative dimer interface is indicated. the correct tertiary structure of lobe 1, but verification of this speculation awaits direct biochemical evidence. In the model (Fig. 6B), Cys-60 and Cys-101 are in close proximity suggesting that they may indeed form a disulfide. Whether Cys-358 and Cys-395 are likewise in close proximity cannot be predicted from the model since both of these cysteines are located in the large insertion III which cannot be modeled. Interestingly, mutation of any of the other three cysteines located in lobe 1 (Fig. 6B), Cys-437, Cys-449, and Cys-482, neither disrupts dimerization nor normal processing of the receptor. This suggests that none of these cysteines is critical in terms of disulfide bond formation.
The putative dimer interface containing cysteines 129 and 131 corresponds to region II in the model (Fig. 6B). Because this insertion cannot be modeled on the basis of the threedimensional structure of LIVBP, we cannot predict accurately the location of these two cysteines in relation to the rest of the protein. Indirect evidence, however, suggests that they are located at the surface of the protein, since we have shown previously that the intervening residue, Asn-130, is glycosylated (33). This region, in addition to containing the cysteine(s) involved in intermolecular disulfide bond formation, may be involved in specific recognition of homodimerization partners. Alignment of this region in CaR and mGluR shows substantial sequence diversity, far greater in particular than in the regions of the ECD encompassed in the venus flytrap model. The mGluR5 was shown to homodimerize but not to heterodimerize with mGluR1 (13), reflecting the importance of specific recognition sequences in addition to the conserved cysteine necessary for dimerization. If our model is correct, we predict that chimeras of a given mGluR containing region II of a different mGluR may be able to heterodimerize with that other mGluR. We cannot, of course, exclude the involvement of additional parts of the ECD, for example region IV, in forming the dimer interface.
An alignment of the ECD of family 3 GPCR shows that all mGluR subtypes have a single cysteine corresponding to cysteines 129 and 131 of the hCaR. Our data would suggest that mutation of this single cysteine would disrupt dimer formation and result in a cell surface-expressed monomer. Several VNR (6) also have a single cysteine in this position and are therefore predicted to be dimers. TR, lacking this cysteine, may either be monomers or may have another basis for dimer formation. The GABA B receptor is interesting in this respect in that it lacks not only this cysteine but others conserved in the ECD of family 3 GPCR. Indeed, a recent modeling and mutagenesis study shows a lack of intermolecular disulfides in the GABA B receptor ECD (10). The native receptor, however, forms a heterodimer formed by a coiled-coil interaction involving a region within the intracellular carboxyl terminus of two GABA B receptor subtypes (4).
There is evidence for dimerization for members of other subfamilies of the GPCR superfamily (see Ref. 4 for review), but the region responsible for dimerization may be within the seven transmembrane domain in those cases, and the functional significance for the respective native receptors is as yet unclear. The present results identify two specific cysteines within a specific region necessary for CaR dimerization and provide evidence for the functional importance of this dimerization. Further biochemical and structural studies are needed to assess the present model of the CaR ECD and of the cysteines contained therein. Also important will be studies directed at identifying the site(s) of Ca 2ϩ binding to the receptor and the differences if any in binding between receptor dimers and monomers. Unlike for mGluRs (11,14,15) and GABA B receptor (10), where ligand binding assays are available and putative binding sites within the respective ECDs have been identified, ligand binding to the ECD has not been demonstrated for the CaR. Only very indirect evidence (36,37) and analogy to mGluRs and bacterial periplasmic binding proteins suggests binding within the ECD.