Parathyroid Hormone-Receptor Interactions Identified Directly by Photocross-linking and Molecular Modeling Studies*

Direct mapping of the interface between parathyroid hormone (PTH) and its receptor (hPTH1-Rc) was carried out by photoaffinity scanning studies. Photoreactive analogs of PTH singularly substituted with a p-benzoylphenylalanine (Bpa) at each of the first six N-terminal positions have been prepared. Among these, the analog [Bpa1,Nle8,18,Arg13,26,27,l-2-Nal23,Tyr34]bPTH-(1–34)NH2(Bpa1-PTH-(1–34)) displayed in vitroactivity with potency similar to that of PTH-(1–34). The radioiodinated analog 125I-Bpa1-PTH-(1–34) cross-linked specifically to the hPTH1-Rc stably expressed in human embryonic kidney cells. A series of chemical and enzymatic digestions of the hPTH1-Rc–125I-Bpa1-PTH-(1–34) conjugate suggested that a methionine residue (either Met414 or Met425) within the contact domain hPTH1-Rc-(409–437), which includes the transmembrane helix 6 and part of the third extracellular loop, as the putative contact point. Site-directed mutagenesis (M414L or M425L) identified Met425 as the putative contact point. Molecular modeling of the hPTH1-Rc together with the NMR-derived high resolution structure of hPTH-(1–34), guided by the cross-linking data, strongly supports Met425, at the extracellular end of transmembrane helix 6, as the residue interacting with the N-terminal residue of the hPTH-(1–34). The photocross-linking and molecular modeling studies provide insight into the topologic arrangement of the receptor-ligand complex.

Understanding the molecular mechanism of ligand recognition and signal transduction by the PTH1-Rc may identify new directions for the design of novel hormone analogs for the treatment of diseases such as osteoporosis, hypercalcemia of malignancy and hyperparathyroidism (16). In order to directly identify the structural elements involved in PTH-PTH1-Rc interactions, we employed a photoaffinity scanning approach (17). The generation of covalently linked ligand-receptor conjugates and the identification of the cross-linked domains allows mapping of the interface between hormone and receptor. Photoaffinity cross-linking has been successfully applied in defining interactions between small peptides, such as substance P (18 -20), cholecystokinin (21), and vasopressin (22), and their receptors. Recently, we used this general approach to identify directly the interaction between position 13 of PTH and a 17-amino acid domain (residues 173-189) of the hPTH1-Rc (17).
We now report the evaluation of a series of photoreactive analogs obtained by a "p-benzoylphenylalanine (Bpa) scan" of the principal receptor activation domain (residues 1-6) of PTH- . A radiolabeled analog containing a photoreactive moiety at the N terminus, 125 I-[Bpa 1 ,Nle 8,18 ,Arg 13,26,27 ,L-2-Nal 23 , Tyr 34 ]bPTH-(1-34)NH 2 ( 125 I-Bpa 1 -PTH-(1-34)), maintained full potency and led to the identification of a second "contact domain" between PTH and hPTH1-Rc. This information allows us to create, for the first time, a model describing interactions of hPTH-(1-34) with its receptor based on direct identification of the interacting regions.
Adenylyl Cyclase Activity-HEK293/C-21 cells were subcultured in 24-well plates and grown to near confluence. COS-7 cells transiently expressing mutant receptors were subcultured 24 h following transfection at a density of 2 ϫ 10 5 /well in 24-well plates and assayed for adenylyl cyclase activity 72 h after transfection. Activation of adenylyl cyclase by PTH analogs was determined as described (23).
Intracellular Calcium Determinations-The stimulation of increases in intracellular calcium levels following treatment by PTH-(1-34) and the Bpa-containing analogs was assessed spectroscopically in Fura-2loaded HEK-293/C-21 cells as described (26).
Proteins were precipitated by adding five volumes of cold acetone and redissolved in 25 mM Tris, pH 8.5, containing SDS (2% w/v). Proteins were reduced with 100 mM dithiothreitol for 2 h at 37°C and alkylated with 200 mM iodoacetamide for 15 min at room temperature. The solution was desalted and concentrated on Centricon 50 (Amicon), diluted with reducing Laemmli sample buffer, and loaded on a 7.5% (v/v) SDS-PAGE. After autoradiography, the radioactive 125 I-Bpa 1 -PTH-(1-34)-Rc conjugate was excised from the gel, passively eluted in 100 mM NH 4 HCO 3 /SDS (0.01% v/v), pH 7.5, and submitted to concentration and buffer exchange on Centricon 50 (Amicon) to 25 mM Tris, pH 8.5, containing Triton X-100 (0.1% v/v) and SDS (0.01% v/v). Small scale photoaffinity cross-linking of transiently transfected COS-7 cells, grown to overconfluence, was carried out in 24-well tissue culture plates. Cells were washed with D-MEM and were treated with 200 l of D-MEM and either 25 l of 10 Ϫ5 M PTH-  in vehicle (PBS, 0.1% BSA), or vehicle alone. Reactions were incubated 15 min at room temperature, 1-2 ϫ 10 6 cpm of 125 I-Bpa 1 -PTH-(1-34) (total volume 25 l) added to each well, and incubated an additional 15 min at room temperature. Plates were cross-linked in a Stratalinker for 30 min as described earlier. Each well was washed once with PBS, cells lysed with 0.5 ml of Laemmli sample buffer, shaken in dish for 10 -30 min, and harvested into Eppendorf tubes. Tubes were incubated on a rotating platform at room temperature for 2-3 h, and analyzed by SDS-PAGE.
Enzymatic and Chemical Digestions of the 125 I-Bpa 1 -PTH-(1-34)-Receptor Conjugate-Batches of SDS-PAGE-purified radiolabeled hormone-receptor conjugate and fragments were prepared in small volumes (typically 10 -20 l) of 25 mM Tris-HCl, pH 7.4, Triton X-100 (0.1% v/v), SDS (0.01% w/v). Endo-F digestions were carried out at 37°C for 24 h, according to the manufacturer's procedure. Lys-C digestions were performed by two 24-h treatments with 0.15 units (in 10 l of water) at 37°C. BNPS-skatole digestions were carried out with 2 mg/ml BNPSskatole in 70% acetic acid at 37°C for 24 -48 h in the dark. CNBr digestions were performed with 50 mg/ml CNBr in 70% formic acid at 37°C for 24 h in the dark. Samples were dried on Speed-Vac and dissolved in reducing sample buffer (27) prior to electrophoresis.
Electrophoresis and Autoradiography-Electrophoretic analyses were performed with 7.5% SDS-PAGE for the hormone-receptor conjugates and 16.5% Tricine/SDS-PAGE for the cleavage products. Appropriate molecular weght markers (Amersham Pharmacia Biotech and Bio-Rad) were included in each gel. Gels were dried and exposed to x-ray films (X-Omat, Eastman Kodak Co.) with intensifying screens (XAR-5, Eastman Kodak Co.). Following autoradiography, the radioactive fragments were excised from the dried gels, extracted in 100 mM NH 4 HCO 3 , pH 7.5, SDS (0.01% w/v) and concentrated on Speed-Vac.
Receptor Mutagenesis-Single mutations, M414L and M425L, were introduced into the hPTH1-Rc cDNA generating two mutated hPTH1-[M414L] and hPTH1[M425L] Rcs, respectively. Primer pairs (sense and antisense) were prepared containing these amino acid modifications (Life Technologies, Inc. custom primers): sense M414L (5Ј to 3Ј): CCA-CGCTGGTGCTCCTGCCCCTCTTTGGCGTC; sense M425L (5Ј to 3Ј): CACTACATTGTCTTCCTGCCACACCATACACC. Primer pairs were used in the polymerase chain reaction-based Quik-Change site-directed mutagenesis kit (Stratagene), using the hPTH1-Rc (28) in the pZeoSV2 (Invitrogen) mammalian expression vector as a template. Individual polymerase chain reaction reactions were used to transform DH5␣ competent cells (Life Technologies, Inc.). Transformations were plated on bacteriologic agar containing Zeocin, colonies identified and selected for plasmid isolation (Miniprep Kit, Quiagen). Plasmid preparations were cycle sequenced (Genomyx, Foster City, CA) to confirm the fidelity of the mutations, using oligonucleotide primers located 5Ј to the regions of the hPTH1-Rc targeted for mutation. The entire mutant Rc was subsequently completely sequenced, on both strands, to ensure the single Met 3 Leu mutation.
Transient Transfection-COS-7 cells were plated at 5-7.5 ϫ 10 5 cells/10-cm dish 24 h prior to transient transfections. Ten g of either mutant or native receptor construct were co-transfected with 10 g carrier DNA using calcium/phosphate (Life Technologies, Inc.). For adenylyl cyclase assay and photoaffinity cross-linking, transiently transfected cells were subcultured as described above.
Molecular Modeling-The molecular model of the hPTH1-Rc was developed using the topological arrangement of the TM helices of rhodopsin (29). To identify the location of the TM portions of the hPTH1-Rc, assumed to be ␣-helices, a hydrophobicity profile (30) was calculated for the hPTH1-Rc sequence. Each of the initially identified transmembrane domains, expanded by approximately 15 amino acids on each side, was submitted to a BLAST search (31). The results helped to refine the location of the TM helices; there was good agreement with respect to the location of the helices from these two methods. After placing the identified helices of the hPTH1-Rc onto the rhodopsin template, the helices were rotated about their long axis to orient the hydrophobic moment toward the membrane environment. These orientations were then refined, requiring minor adjustments, following the substitution-table methodology reported by Donelly and co-workers (32). The loops connecting the TM helices were added to complete the model. In an attempt to develop the conformational preferences of the ectopic N-terminal portion of the receptor, the corresponding sequences were submitted to a BLAST search (31). There were numerous sequences of significantly high homology in the protein data bank, especially for a region of the a Molecular weights were measured by electronspray-mass spectrometry.
b Reverse-phase HPLC analyses were carried out on a Vydac C-18, 300-Å column (4.6 ϫ 150 mm, 5 m) at a flow rate of 1.0 ml/min and monitored at 220 nm. Linear gradient from 20 to 50% where t r and t 0 are the retention times of the peptide and the front, respectively, in the conditions described in footnote b. ectopic N-terminal tail of the hPTH1-Rc contiguous to the TM1 helix. The homologous regions of each of these protein structures were analyzed for secondary structural features and then incorporated into the molecular model.

Characterization of Bpa
Identification of the Ligand Binding Domain-The 87-kDa 125 I-Bpa 1 -PTH-(1-34)-Rc conjugate was purified from 7.5% SDS-PAGE and subjected to a series of chemical and enzymatic cleavages. The first digestion pathway (I) consisted of enzymatic cleavage at the carboxyl side of lysyl residues with Lys-C, followed by chemical cleavage at the carboxyl side of tryptophanyl residues with BNPS-skatole. Exhaustive Lys-C treatment of the 87-kDa ligand-receptor conjugate yielded a single radiolabeled band with apparent molecular mass of ϳ11 kDa (Ia) (Fig. 3A, lane 2). Similar treatment of the deglycosylated conjugate (ϳ66 kDa) yielded a band with the same apparent molecular mass (data not shown), confirming the absence of glycosylation within the Lys-C-generated fragment Ia. BNPSskatole treatment of the excised and eluted 11-kDa (Ia) fragment produced a single band with apparent mass of ϳ7 kDa (Ib) (Fig. 3A, lane 3).
A second digestion pathway (II), the reciprocal of I, initially yielded a single band migrating at ϳ14 kDa (IIa) (Fig. 3B, lane  2). Lys-C treatment yielded the final fragment migrating at ϳ7 kDa (IIb) (Fig. 3B, lane 3), similar to Ib obtained from pathway I.
Treatment with CNBr (III) of the purified intact ligandreceptor conjugate in 70% formic acid solution produced a band of very low apparent molecular mass (IIIa) (ϳ4 kDa, Fig. 3C, lane 2), with electophoretic mobility similar, if not identical, to that of the free radioligand 125 I-Bpa 1 -PTH-(1-34) (Fig. 3C). In addition, CNBr treatment of the 11-kDa (Ia) fragment obtained after Lys-C digestion produced a similar ϳ4-kDa band (data not shown).  (Fig. 5B, lanes 1  and 3, respectively) generating the anticipated ϳ87-kDa band corresponding to the 125 I-Bpa 1 -PTH-(1-34)-PTH1Rc conjugate. This cross-linking can be inhibited competitively by 10 Ϫ6 M PTH-(1-34) (Fig. 5B, lanes 2 and 4). In contrast, the functional transiently transfected hPTH1[M425L] does not crosslink to 125 I-Bpa 1 -PTH-(1-34) (Fig. 5B, lanes 5 and 6), suggesting that Met 425 may be involved in the cross-linking of 125 Table II. Equally important, in many of the protein structures examined, the topological location of the homologous helices was on the surface of the protein, dictated by the relative amphipathic nature of these ␣-helices. Given the apparent amphipathicity of the Lys 172 -Met 189 helix, we assume it is lying on the surface of the membrane, with the hydrophilic amino acids projecting into the aqueous phase. This ectopic helix is extended by an extracellular loop consisting of a few amino acids, Leu 187 -Gly 188 -Met 189 . An important point to emphasize is that the BLAST results clearly suggest a conformational discontinuity between the TM1 helix and this ectopic helix.
Although these data support the presence of an amphipathic ␣-helix, the orientation of this helix with respect to the bundle of TM helices is not clear. Molecular dynamics simulation of many different starting orientations of the helix were carried out using the two phase simulation cell (33). Throughout the molecular dynamics simulations, the ectopic helix always tended to move away from the bundle of TM helices. Therefore, for the purpose of docking the ligand hPTH-(1-34), a conformation of hPTH1-Rc with the ectopic amphipathic helix projecting away from the TM bundle was utilized.
Molecular modeling of hPTH-(1-34) with the hPTH1-Rc model was then performed, guided by the contact domain identified by previous photoaffinity cross-linking studies (17) (Fig.  6). The conformation of hPTH-(1-34) used in the molecular modeling was obtained from our high resolution NMR studies performed in different environments, including aqueous saline conditions and the presence of dodecylphosphocholine as a membrane mimetic (37). DISCUSSION The characterization of the bimolecular interaction between the activation domain of PTH-(1-34) and the hPTH1-Rc is of fundamental importance for elucidating the molecular mechanism of signal transduction. To this end, photoaffinity crosslinking of bioactive analogs enables direct identification of "contact domains" and/or "contact points" between ligand and receptor (17-21, 38 -42). The goal of this study was to identify the region of the hPTH1-Rc which is in direct contact with the principal "activation domain" of PTH-(1-34). Residues 1 through 6 (43), particularly positions 1 and 2, have been shown to be essential for full agonist activity. Stepwise deletion of N-terminal residues from PTH-(1-34) yields the antagonist/ partial agonist PTH-  and the antagonist PTH-(7-34) (16,44) with little diminishment of binding affinity, but with progressive loss of agonist bioactivity.
The photoreactive [Bpa 1-6 ]PTH-(1-34) analogs were specifi-  1 and 2), and receptor containing mutations M414L (lanes 3 and 4), and M425L (lanes 5 and 6) in the absence (lanes 1, 3, and 5) or presence of 10 Ϫ6 M PTH-(1-34) (lanes 2, 4, and 6). Size markers (in kDa) are also shown. The arrow to the right of lane 6 indicates the location of the 125 I-Bpa 1 -PTH-PTH1-Rc conjugate (ϳ87 kDa). cally designed for this study. Both Met 8 and Met 18 were replaced by the isosteric Nle residue, rendering the ligand resistant to cyanogen bromide treatment. Replacement of Trp 23 with 2-naphthylalanine (2-Nal) introduces stability toward digestion by Trp-specific reagents. Replacement of all Lys residues with Arg provides resistance to Lys-C-mediated cleavage, and replacement of Phe 34 with Tyr generates a reactive site for incorporation of radioiodine. These modifications were introduced singularly and found to be well-tolerated (23,43). As demonstrated previously, the combination of all the modifications is, in many cases, also well tolerated (17,23).
Exhaustive cyanogen bromide digestion of the intact 125 I-Bpa 1 -PTH-(1-34)-PTH1-Rc conjugate (87 kDa) (III) (Fig. 4) and of the Lys-C-generated fragment Ia (11 kDa) yields a similar band IIIa with a low apparent molecular mass (ϳ4 kDa) (Fig. 3C, lane 2). The electrophoretic mobility of this band is similar to that of the ligand itself (mass ϭ 4,487 Da) and distinctly lower than any potential covalent ligand-receptor complex produced by cyanogen bromide cleavage.
Photocross-linking of a benzophenone-containing ligand through insertion into a C-H bond of the S-CH 3 group in Met residues will generate upon cyanogen bromide treatment a ligand-CH 3 SCN adduct, which increases its molecular mass by only 73 Da (19). Electrophoretically, the CH 3 SCN-125 I-Bpa 1 -PTH-(1-34) adduct will be indistinguishable from the nonmodified photoreactive radioligand (Fig. 3C). This adduct could be generated by cross-linking to either one of the two methionine residues present in the minimal contact domain hPTH1-Rc-(409 -437) and included in the conjugate fragment (Ib/IIb) (Figs. 4, 3A, lane 3, and 3B, lane 3). Both methionines, Met 414 and Met 425 , are located in TM6 and are therefore potential contact points between the N-terminal residue of PTH and the receptor.
Using site-directed mutagenesis to produce both M414L and M425L mutated Rcs, the Met residue involved in the crosslinking can be assigned to Met 425 . The M414L mutant is fully active and like the native PTH1-Rc it photocross-links to 125 I-Bpa 1 -PTH- . In contrast, M425L mutant, although fully active, does not cross-link to 125 I-Bpa 1 -PTH-(1-34). Therefore, this mutation perturbs the close spatial proximity required for effective cross-linking with an N-terminal benzophenone-containing ligand. The identification of Met 425 as the contact point for position 1 of PTH-(1-34) provides an additional, and important constraint in defining the topology of the PTH-(1-34)-hPTH1-Rc complex.
The structure of the hPTH1-Rc obtained by homology modeling and MD using a two-phase solvent cell (33) suggests that the segment Arg 179 -Arg 189 consists of an amphipathic ␣-helix whose axis is parallel to the membrane surface and directed away from the helical bundle of the receptor. The structure of hPTH-(1-34) used in the molecular modeling was determined by NMR in a zwitterionic, micellar environment (37) as a mimetic of the cellular membrane. Throughout the MD simulations, deviations from the experimentally determined structure of hPTH-(1-34), consisting of ␣-helices for residues 4 -10 and 20 -32, were not allowed. A number of starting structures with Lys 13 of hPTH placed at different locations along the 17-amino acid cross-linking domain (hPTH1-Rc-(173-189)) (17) were used for MD simulations. During these simulations, utilizing the biphasic solvent mixture to mimic the membrane environment (33), the interactions between the amphipathic helix of the receptor, segment 179 -186, just exterior to TM1, and the C-terminal helix (residues 20 -32) of hPTH were optimized. Throughout these simulations, the N-terminal residue of hPTH could be easily placed in close proximity to Met 425 , which is on the surface of the membrane at the C-terminal end of TM6 (Fig.  6). In contrast, all attempts to place position 1 of hPTH in close proximity to Met 414 , while maintaining the experimentally determined conformation of hPTH and Lys 13 of hPTH close to its cross-linking domain, failed. In our model, Met 414 is on the intracellular half of TM6, projecting toward the membrane, a full three helical turns removed from Met 425 . Thus, the biochemical analysis of the 125 I-Bpa 1 -PTH-(1-34)-PTH1-Rc conjugate, site-directed mutagenesis of the hPTH1-Rc, and molecular modeling simulation strongly suggest that Met 425 is the "contact point" of the hPTH1-Rc and the cross-linking site for 125 I-Bpa 1 -PTH- .
Previous mutagenesis studies have implied that TM6 and the third extracellular loop are important for hormone binding and signal transduction. Homologous substitution of these regions in the rat Rc with the corresponding portions of either the opossum PTH1-Rc (45) or the secretin Rc (46) identified several residues (i.e. Leu 427 , Trp 437 ) which affect hormone binding and/or signaling. Interestingly, mutation of Thr 410 in TM6 generated a constitutively active receptor associated with the clinical skeletal disorder, Jansen's metaphyseal chondrodysplasia (47). TM6 seems to be directly involved in signaling in other G protein-coupled receptors. Mutations in this region of the m5 muscarinic (48) and ␣-factor (49) receptors result in constitutive receptor activation. In addition, TM6 is contiguous with the third intracellular loop, which has been implicated in the interaction of G proteins in several seven TM-domain-containing receptors, including the PTH1-Rc (50).
The identification of Met 425 in the extracellular end of TM6 as the contact point for the N terminus of PTH, together with the emerging model of ligand-receptor interaction, offers new insights into the nature of hormone-receptor interactions and signal transduction in this system. The contact of the principal activation domain of PTH-(1-34) with residues in the extracellular end of TM6, which in turn is connected to the third intracellular loop (considered to contain a G protein contact domain) suggests a possible relay mechanism that communicates an extracellular stimulus (i.e. agonist binding) into an intracellular signaling event (i.e. activation of the G protein).
Our modeling of the bimolecular PTH-hPTH1-Rc interaction potentially can be greatly enhanced by identification of the specific amino acid in hPTH1-Rc involved in cross-linking with residue 13 of hPTH. Narrowing the 17-amino acid contact domain to a smaller fragment, plus identification of additional contact domains in the Rc with other amino acids of the hormone, will refine our experimentally based model and provide greater detail regarding the hPTH-hPTH1-Rc bimolecular interface.