Evidence for a ligand interaction site at the amino-terminus of the parathyroid hormone (PTH)/PTH-related protein receptor from cross-linking and mutational studies.

Low resolution mutational studies have indicated that the amino-terminal extracellular domain of the rat parathyroid hormone (PTH)/PTH-related protein (PTHrP) receptor (rP1R) interacts with the carboxyl-terminal portion of PTH-(1-34) or PTHrP-(1-36). To further define ligand-receptor interactions, we prepared a fully functional photoreactive analog of PTHrP, [Ile5,Bpa23,Tyr36]PTHrP-(1-36)-amide ([Bpa23]PTHrP, where Bpa is p-benzoyl-L-phenylalanine). Upon photolysis, radioiodinated [Bpa23]PTHrP covalently and specifically bound to the rP1R. CNBr cleavage of the broad approximately 80-kDa complex yielded a radiolabeled approximately 9-kDa non-glycosylated protein band that could potentially be assigned to rP1R residues 23-63, Tyr23 being the presumed amino-terminus of the receptor. This assignment was confirmed using a mutant rP1R (rP1R-M63I) that yielded, upon photoligand binding and CNBr digestion, a broad protein band of approximately 46 kDa, which was reduced to a sharp band of approximately 20 kDa upon deglycosylation. CNBr digestion of complexes formed with two additional rP1R double mutants (rP1R-M63I/L40M and rP1R-M63I/L41M) yielded non-glycosylated protein bands that were approximately 6 kDa in size, indicating that [Bpa23]PTHrP cross-links to amino acids 23-40 of the rP1R. This segment overlaps a receptor region previously identified by deletion mapping to be important for ligand binding. Alanine scanning of this region revealed two residues, Thr33 and Gln37, as being functionally involved in ligand binding. Thus, the convergence of photoaffinity cross-linking and mutational data demonstrates that the extreme amino-terminus of the rP1R participates in ligand binding.

Current information suggests that PTH and PTHrP interact with the P1R through multiple sites and that these are dispersed throughout the extracellular surface of the receptor and some portions of the transmembrane helices (1). Studies with chimeras formed between P1Rs from different species or different receptor subtypes (P1R or P2R) indicate that there are interactions between the amino-terminal extracellular domain of the receptor and region 15-34 of the ligand and between the core region of the receptor and the amino-terminal portion of the ligand (9 -12). Furthermore, observations from studies on other members of this peptide hormone receptor family (13,14), and particularly with chimeras between the P1R and the calcitonin receptor (15), suggest that this general orientation of ligand-receptor interaction may apply to all members of this family of G protein-coupled receptors.
In addition to mutagenesis approaches, affinity cross-linking methods can provide valuable information on the location of ligand-receptor interactions sites in peptide hormone receptors (16,17). For the P1R, Zhou et al. (18) recently showed that a PTH-(1-34) analog containing a photoderivatized lysine 13 cross-linked to a 17-amino acid segment of the amino-terminal extracellular receptor domain that mapped close to the junction with the first membrane-spanning helix. In related experiments, this group showed that another PTH-(1-34) analog, which contains p-benzoyl-L-phenylalanine (Bpa) (19) at position 1, cross-linked to a region of the P1R containing transmembrane helix 6 and extracellular loop 3 (20). The results of these physicochemical analyses are in agreement with previous mutational studies that functionally identified similar regions of the P1R as candidate ligand-interaction sites (12,(21)(22)(23). In addition to these two putative ligand contact regions, mutational studies also identified segments within the large (Ϸ190 amino acids) amino-terminal domain of the receptor that appear to interact with region 15-34 of the ligand (10,24).
We have now performed cross-linking studies with a PTHrP analog that contains photoreactive Bpa at position 23 in place of the native phenylalanine, a residue recently shown to be involved in the ligand binding specificity of the PTH2 receptor (7,11). This new photoreactive ligand, [Ile 5 ,Bpa 23 ,Tyr 36 ]PTHrP-(1-36)-amide, cross-linked to a short segment between residue 40 and the amino-terminus, which is predicted to be Tyr-23. We also confirmed the importance of this amino-terminal receptor region by mutational methods and have identified two amino acid residues that contribute to ligand binding affinity. PTHrP-(7-34)-amide (PTHrP-(7-34)) were synthesized by the Protein and Peptide Core Facility at Massachusetts General Hospital (Boston, MA) by solid-phase method on Perkin-Elmer Model 430A and 431A synthesizers. The Fastmoc version of Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry was utilized, and peptides were purified by reversed-phase chromatography.

Materials
Na 125 I (specific activity of 2000 Ci/mmol) and radioiodinated antimouse IgG Fab (Nex162) were purchased from NEN Life Science Products. Dulbecco's modified Eagle's medium, Ham's F-12 medium, trypsin/EDTA, penicillin G/streptomycin, and horse serum were from Life Technologies, Inc.. Fetal bovine serum, 3-isobutyl-1-methylxanthine, bovine serum albumin, Tricine, and Me 2 SO were from Sigma. Trifluoroacetic acid was from Pierce, and CNBr was from Serva Fine Chemicals/Boehringer Ingelheim (Heidelberg, Germany). 14 C-Methylated protein molecular mass markers for SDS-PAGE were purchased from Amersham Pharmacia Biotech, and peptide N-glycosidase F was from New England Biolabs Inc. (Beverly, MA). DEAE-dextran was from Pharmacia (Uppsala, Sweden), and X-Omat AR films for autoradiography were from Eastman Kodak Co. The monoclonal antibody 12CA5 was purchased from Berkeley Antibodies (Berkeley, CA).
Mutagenesis of the Rat PTH/PTHrP Receptors-Mutations were introduced into single-strand plasmid DNA encoding the wild-type rat PTH/PTHrP receptor (rP1R) (4) or, for cassette replacements and alanine point mutations, into the P1R with a hemagglutinin (HA) epitope in the receptor's E2 domain (rP1R-HA) (24) by oligonucleotide-directed site-specific mutagenesis as described (11,24,25). The oligonucleotide primers were synthesized on an Applied Biosystems Model 380A DNA synthesizer. Positive mutants were verified by nucleotide sequence analysis of plasmid DNA.
Cell Culture and DNA Transfection-COS-7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% heatinactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified atmosphere containing 95% air and 5% CO 2 . Cells were seeded in 24-well plates (200,000 cells/well) for radioreceptor and cAMP assays and for preliminary cross-linking experiments; all other cross-linking experiments were performed in 150-mm dishes (6 ϫ 10 6 cells). Once the cell monolayer reached 90 -100% confluency, cells were transfected by the DEAE-dextran method as described (22) using 200 ng of plasmid DNA/well or 2 g/150-mm dish. The medium was replaced daily, and 3 days after transfection, cells were used either for radioligand binding and cAMP accumulation assays or for cross-linking experiments. ROS 17/2.8 cells were maintained as described (26) in Ham's F-12 medium supplemented with 5% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin.
Radioligand-Receptor Binding Assays and cAMP Accumulation-Radiolabeled rNlePTH, PTHrP, and [Bpa 23 ]PTHrP were prepared by chloramine-T iodination, followed by high pressure liquid chromatography purification using a 30 -50% acetonitrile in 0.1% trifluoroacetic acid gradient over 30 min. Radioligand-receptor binding assays were performed in 24-well plates as described (11). In brief, each well (final volume of 500 l) contained binding buffer (50 mM Tris-HCl (pH 7.7), 100 mM NaCl, 5 mM KCl, 2 mM CaCl, 5% heat-inactivated horse serum, and 0.5% heat-inactivated fetal bovine serum), 125 I-labeled radioligands (100,000 -200,000 cpm), and varying concentrations of unlabeled peptide. After 4 h at 16°C, the binding mixture was removed, and the cells were rinsed with cold binding buffer and lysed with 1 M NaOH. The entire lysate was counted for ␥-irradiation. Specific binding was determined after subtracting radioactivity bound in the presence of maximal concentrations of unlabeled competing peptide (10 Ϫ6 M). Agonist-dependent accumulation of cAMP was determined by radioimmunoassay as described (11).
Cell-surface Expression of PTH/PTHrP Receptors-Cell-surface expression was assessed as described (24) using antibody 12CA5 directed against the HA epitope in the rP1R-HA receptors and a second radiolabeled anti-mouse IgG Fab fragment. Relative specific binding of antibody to each mutant P1R was calculated by subtracting nonspecifically bound radioactivity (determined in mock-transfected COS cells; typically 0.1-0.2% of added radioactivity) from the total bound radioactivity divided by the radioactivity specifically bound to wild-type rP1R-HA (typically 1-2% of the added radioactivity).
Photoaffinity Labeling of PTH/PTHrP Receptors-In preliminary experiments, COS-7 cells that were grown and transfected in 24-well plates were rinsed twice with 1 ml of cold binding buffer, and the cell monolayer was then incubated for 6 h at 4°C with 125 I-[Bpa 23 ]PTHrP (1 ϫ 10 6 cpm) diluted in 0.5 ml of binding buffer with or without unlabeled ligand (10 Ϫ6 M bNlePTH or [Bpa 23 ]PTHrP). After incubation, cells were rinsed three times with 1 ml of cold binding buffer before adding 200 l of binding buffer and placing the dishes on ice under a UV light source for 20 min (Blak Ray long-wave lamp; 366 nm, 7000 microwatts/cm 2 ; UV Products, San Gabriel, CA; source-to-cell distance of Ϸ5 cm). After photolysis, cells were rinsed once with cold phosphatebuffered saline, twice with a cold acidic buffer (0.05 M glycine and 0.15 M NaCl (pH 2.5)) to remove noncovalently bound radioligand, and twice with cold phosphate-buffered saline before solubilization with 0.5 ml of SDS-PAGE sample buffer (4% (w/v) SDS, 80 mM Tris-HCl (pH 6.8), 20% (v/v) glycerol, 0.2% bromphenol blue, and 100 mM dithiothreitol). The lysate was then passed six times through a 19-gauge needle to shear genomic DNA.
To prepare larger amounts of the cross-linked ligand-receptor complex, a similar protocol was followed using COS-7 cells that were grown and transfected in 150-mm dishes. For each rinsing step, 30 ml of cold binding buffer were used, and incubation with 125 I-[Bpa 23 ]PTHrP (2-4 ϫ 10 7 cpm) was performed in a final volume of 20 ml of binding buffer. During UV light exposure, the cell monolayer was covered with 10 ml of binding buffer, and after photolysis and rinsing, cells were solubilized with 4 ml of SDS-PAGE sample buffer.
SDS-PAGE Analysis of the Ligand-Receptor Complex-After heating to 70°C, samples (and appropriate size markers) were either subjected to analytical SDS-PAGE analysis (5-20% acrylamide, 0.75-mm spacers) according to the method of Laemmli (27) or loaded onto a 16.5% (w/v) Tricine gel (0.75-mm spacers) according to the method of Schä gger and von Jagow (28), with subsequent autoradiography of the dried gels (1-14 days at Ϫ80°C with intensifying screens).

Purification of [Bpa 23 ]PTHrP⅐P1R
Complexes-To isolate larger amounts of radiolabeled ligand-receptor complexes from cells cultured in 150-mm dishes, we used preparative SDS-polyacrylamide gels (5-20% acrylamide, 3-mm spacers); the complexes were identified by autoradiography of the wet gels (exposure time of 4 -12 h at room temperature) and electroeluted from an excised gel slice using a Model 422 electroeluter (Bio-Rad). The isolated radiolabeled ligandreceptor complex was stored at Ϫ20°C in elution buffer (25 mM Tris, 192 mM glycine, and 0.02% SDS) before chemical/enzymatic treatment (see below).
CNBr Cleavage-CNBr was dissolved in 100% trifluoroacetic acid and then added to the partially purified radiolabeled ligand-receptor complex to give a final concentration of 100 mM CNBr in 70% trifluoroacetic acid. After an overnight, light-protected incubation, the digest was reduced in volume under a stream of nitrogen and then repeatedly lyophilized to remove trifluoroacetic acid and CNBr. Once the apparent molecular mass of each CNBr-derived ligand-receptor fragment had been established, trifluoroacetic acid and CNBr were eliminated more efficiently by ultrafiltration using a Microcon 3 concentrator (Amicon, Inc., Beverly, MA) and repeated dilution of the retentate with H 2 O.
Peptide N-Glycosidase F Digestion-The CNBr-cleaved and concentrated radioligand-receptor complex was treated with peptide N-glycosidase F (2500 units) for 1 h at 37°C in 30 l of 50 mM sodium phosphate (pH 7.5), 0.5% SDS, 1% ␤-mercaptoethanol, and 1% Nonidet P-40 according to the protocol provided by the supplier. 23 ]PTHrP was tested in competition binding studies performed with COS-7 cells expressing the native rP1R and was found to have an apparent binding affinity that is indistinguishable from that of bNlePTH and of other analogs of PTH and PTHrP (7, 10). The Bpa-containing PTHrP analog was also fully functional in cAMP accumulation assays and exhibited a potency that was indistinguishable from that of bNlePTH (data not shown).

Functional Characterization of [Bpa 23 ]PTHrP-[Bpa
Photoaffinity Labeling of Rat PTH/PTHrP Receptors-After binding and photoactivation, the covalent complex formed between radioiodinated [Bpa 23 ]PTHrP and the rP1R was visualized by analytical SDS-PAGE and subsequent autoradiography. The complex migrated as a single broad band corresponding to a glycosylated protein with a molecular mass of Ϸ80 kDa (Fig. 1, lane 1). This size of the ligand-receptor complex is comparable to that previously seen with other photoreactive PTH or PTHrP analogs using either cells expressing endogenous PTH/PTHrP receptors (29 -32)  To identify the region of the P1R that interacted with the Bpa 23 moiety, we isolated larger amounts of the radiolabeled ligand-receptor complex using preparative SDS-PAGE, cleaved the partially purified complex with CNBr, and separated the cleavage products on analytical gels. After CNBr cleavage, most of the radioactivity migrated on SDS-PAGE as a single sharp protein band corresponding to a size of Ͻ14 kDa (Fig. 1,  lane 2). A minor fraction migrated as a diffuse band at the 46-kDa size marker and probably corresponded to a partially cleaved glycosylated ligand-receptor complex. Tricine/SDS-PAGE analysis was used to achieve higher resolution in the low molecular mass range, and this suggested a molecular size of Ϸ9 kDa for the principal radiolabeled CNBr-generated fragment (see also Fig. 5, lane 1). Since [Bpa 23 ]PTHrP has a molecular size of 4.286 kDa, the receptor fragment contributing to the complex was estimated to have a molecular size of Ϸ5 kDa. The same results were obtained when analyzing the complex formed between radiolabeled [Bpa 23 ]PTHrP and the endogenous PTH/PTHrP receptor of ROS 17/2.8 cells (data not shown).
The above results suggested that Bpa at position 23 of PTHrP interacts with an Ϸ5-kDa non-glycosylated CNBr-generated portion of the receptor. Inspection of the amino acid sequence of the rP1R showed that several fragments delimited by methionine residues are within this molecular size range (Fig. 2). Because of the predicted overall architecture of ligandreceptor interaction (1,11,15), we considered the hypothesis that Bpa 23 interacts with the receptor segment defined by Met 63 and the amino-terminus, which is predicted to be Tyr 23 by a recently developed algorithm (33). A mutant rP1R was generated, rP1R-M63I, in which methionine at position 63 was replaced by isoleucine (Fig. 3B); this mutant receptor had functional (data not shown) and cross-linking (Fig. 1, lane 3) properties that were indistinguishable from the wild-type rP1R. CNBr cleavage of the covalently labeled rP1R-M63I mutant yielded to a broad radioactive band comigrating with the 46- kDa marker (Fig. 1, lane 4, and Fig. 4, lane 1); this cleavage product was reduced to a sharp protein band of Ͻ20 kDa upon further digestion with peptide N-glycosidase F (Fig. 4, lane 2). The increase in size from Ϸ9 to Ϸ46 kDa of the CNBr-derived ligand-receptor complex that occurred as a result of the M63I mutation and the subsequent size reduction of this larger receptor fragment upon deglycosylation indicated that, in the mutant receptor, Bpa 23 interacts with a receptor segment that extends from the amino-terminus to Met 174 . These results confirmed that cross-linking of [Bpa 23 ]PTHrP to the wild-type P1R occurred between Met 63 and the receptor's amino-terminal residue.
To further define the site of cross-linking, we introduced methionine substitutions at either Leu 40 or Leu 41 in the rP1R-M63I mutant to yield the double mutants rP1R-M63I/ L40M and rP1R-M63I/L41M, respectively (Fig. 3C). Like rP1R M63I, these two mutants had biological properties that were indistinguishable from those of the wild-type rP1R (data not shown). Leu 40 and Leu 41 were chosen because their substitution with methionine is a conservative replacement and because CNBr cleavage at these positions would yield ligand-receptor conjugates whose size upon SDS-PAGE analysis would easily distinguish between the two possible sites of interaction with [Bpa 23 ]PTHrP. Thus, cross-linking to a site amino-terminal of Met 40 or Met 41 would yield non-glycosylated, low molecular mass complexes corresponding to receptor residues 23-40 and 23-41, respectively, whereas cross-linking to a site carboxyl-terminal of either mutation would yield glycosylated, high molecular mass complexes corresponding to residues 41-174 and 42-174, respectively. As shown in Fig. 5 (lane 3), CNBr cleavage of the affinity-labeledM63I/L40M mutant yielded a small radiolabeled, nonglycosylated complex of Ϸ6 kDa, as did cleavage of the M63I/ L41M mutant (data not shown). This indicated that the covalent interaction between Bpa 23 and the rP1R occurred between the receptor's amino-terminus and Leu 40 .

Effects of Point Mutations in the Amino-terminal Extracellular Domain of the PTH/PTHrP Receptor on Ligand Binding-
The amino-terminal receptor fragment identified by the above physicochemical approach overlaps a P1R region previously shown by functional studies to be important for ligand binding (24). Two mutant receptors with deletions of residues 26 -60 (the E1 region) or 31-47 (E1a) were shown to have only moderately reduced receptor expression levels in COS-7 cells (22 Ϯ 1 and 36 Ϯ 3% of the wild type, respectively) and little or no capacity to bind radiolabeled PTH (24). To further examine the functional importance of residues in this amino-terminal E1a region, we first made four cassette mutant receptors, termed E1a-1 through E1a-4, in which four or five adjacent residues were replaced by either alanine or valine (Fig. 6, A and B). Each mutant receptor was adequately expressed on the surface of COS-7 cells (Ͼ35% of the wild-type) (Fig. 6C). The two mutants in which residues 31-35 and 36 -39 were altered displayed diminished 125 I-rNlePTH binding capacity, whereas the two mutants with substitutions of residues 40 -43 and 44 -47, respectively, maintained high levels of PTH binding (Fig. 6D).
To further localize candidate binding residues within region 31-39 (E1a-1 and E1a-2), an alanine-scanning approach was used. Several of the individual alanine substitutions in this region, which had little or no effect on cell surface expression (Fig. 7A), resulted in small reductions in 125 I-rNlePTH binding capacity (Fig. 7B). A reduction in PTH binding of Ͼ25% occurred with two substitutions, T33A and Q37A (Fig. 7B). In addition, each of these two point mutations had a more severe effect on 125 I-PTHrP binding than on 125 I-rNlePTH binding (Fig. 7C). In competition binding studies with 125 I-rNlePTH as tracer radioligand, the apparent binding affinity of rNlePTH for wild-type and mutant P1Rs was comparable (Fig. 8A). The apparent binding affinity of bNlePTH for the T33A and Q37A mutant receptors was 5.0-and 2.3-fold weaker, respectively, than it was for the wild-type receptor (Fig. 8B). Consistent with the reduced maximal binding of radiolabeled PTHrP, the apparent binding affinity of PTHrP-(1-36) for these two mutant receptors was 14-and 48-fold weaker, respectively, than it was for the wild-type receptor (Fig. 8C). Both receptor mutations also abolished binding of an amino-terminally truncated PTHrP analog, [Leu 11 ,D-Trp 12 ]PTHrP-(7-34)-amide, indicating that Thr 33 and Gln 37 affect interactions with region 7-34 of the ligand rather than with region 1-6 ( Fig. 8D and Table I carboxyl-terminal region 15-34 of either PTH or PTHrP; a similar architecture of ligand-receptor interaction may well apply to other members of this family of G protein-coupled receptors (9 -15). In this study, we confirmed and extended these predictions for the P1R with a PTHrP analog containing photoreactive Bpa at position 23, a residue with apparent functional significance based on its ability to determine ligand binding specificity in the P2R (7,11). After CNBr digestion of [Bpa 23 ]PTHrP⅐P1R complexes, an Ϸ9-kDa radiolabeled protein was detected upon Tricine/SDS-PAGE analysis. This fragment was likely to represent 125 I-[Bpa 23 ]PTHrP covalently coupled to a receptor fragment extending from Tyr 23 , the first residue after the predicted cleavage site for the signal peptide (33), to Met 63 , the first methionine in the mature receptor sequence. We confirmed this assignment and refined the mapping further by using site-directed mutagenesis to introduce or remove methionines at strategic sites in the receptor. First, the rP1R-M63I mutant was generated and shown to be fully functional. When the ligand-receptor complex formed with this receptor was cleaved with CNBr, the Ϸ9-kDa band was replaced by an Ϸ46-kDa glycosylated band corresponding to the receptor fragment extending from the amino-terminus to Met 174 . This receptor segment contains three of the four potential N-linked glycosylation sites, and glycosylation is consistent with the broadness of the Ϸ46-kDa complex on SDS-PAGE and its reduction to a smaller, non-glycosylated protein band by peptide N-glycosidase F treatment. These results confirmed that crosslinking between BPA 23 and the rat PTH/PTHrP receptor involved residues that are located between the amino-terminus and Met 63 . The M63I mutation allowed us to exclude other CNBr-derived receptor fragments of similar size, such as Ala 426 -Met 450 . Two additional, fully functional receptor double mutants, rP1R-M63I/L40M and rP1R-M63I/L41M, were prepared to further refine the cross-linking site. Both mutants contained the M63I mutation to eliminate the natural CNBr cleavage site at position 63. CNBr cleavage of the complexes formed between 125 I-[Bpa 23 ]PTHrP and either of these two mutant receptors resulted in low molecular mass radiolabeled protein conjugates (Fig. 5). This result established that Bpa 23 cross-linked to a side amino-terminal to Met 41 in the rP1R and clearly excluded segment 41-174 as the site of interaction.
Earlier mutagenesis studies had indicated that deletion of a portion of the rP1R that included 17 residues (the E1a region) close to the amino-terminus of the mature receptor abolished binding of radiolabeled PTH or PTHrP, with only moderate effects on receptor expression (24). To further map functional binding residues in this region, we constructed four "cassette" mutants in which four or five adjacent amino acids were replaced by alanine or valine. Two of these cassette mutants, E1a-1 and E1a-2, showed normal cell-surface expression, but little or no binding of radiolabeled PTH or PTHrP. These results suggested that residues within segments 31-35 and 36 -39 contribute to ligand interaction. The replacement of each of these nine residues with individual alanine substitutions confirmed this hypothesis. Two mutants, rP1R T33A and rP1R Q37A, exhibited the weakest capacity to bind the radioligand. Interestingly, the effects of these mutations on ligand binding were more pronounced with PTHrP than with PTH; this pattern might be attributable to the divergence in region 15-34 of these two ligands, a hypothesis supported by the observation that the mutations also impaired PTHrP-(7-34) binding.
In summary, our physicochemical observations indicate that Bpa 23 (and presumably Phe 23 in the native PTHrP molecule) interacts with residues at the extreme amino-terminus of the PTH/PTHrP receptor. Mutational analysis of this receptor region supported this conclusion and identified two amino acid residues, Thr 33 and Gln 37 , as possible sites for ligand interaction. The combined use of the two techniques, photoaffinity cross-linking and receptor mutagenesis, should enable the definition of other receptor segments that comprise contact points for PTH and PTHrP.