Multiple Sites of Contact between the Carboxyl-terminal Binding Domain of PTHrP-(1–36) Analogs and the Amino-terminal Extracellular Domain of the PTH/PTHrP Receptor Identified by Photoaffinity Cross-linking

The carboxyl-terminal portions of parathyroid hormone (PTH)-(1--34) and PTH-related peptide (PTHrP)-(1-36) are critical for high affinity binding to the PTH/PTHrP receptor (P1R), but the mechanism of receptor interaction for this domain is largely unknown. To identify interaction sites between the carboxyl-terminal region of PTHrP-(1--36) and the P1R, we prepared analogs of [I(5),W(23),Y(36)]PTHrP-(1--36)-amide with individual p-benzoyl-l-phenylalanine (Bpa) substitutions at positions 22--35. When tested with LLC-PK(1) cells stably transfected with human P1R (hP1R), the apparent binding affinity and the EC(50) of agonist-stimulated cAMP accumulation for each analog was, with the exception of the Bpa(24)-substituted analog, similar to that of the parent compound. The radiolabeled Bpa(23)-, Bpa(27)-, Bpa(28)-, and Bpa(33)-substituted compounds affinity-labeled the hP1R sufficiently well to permit subsequent mapping of the cross-linked receptor region. Each of these peptides cross-linked to the amino-terminal extracellular domain of the P1R: [I(5),Bpa(23),Y(36)]PTHrP-(1-36)-amide cross-linked to the extreme end of this domain (residues 33-63); [I(5),W(23),Bpa(27),Y(36)]PTHrP-(1--36)-amide cross-linked to residues 96--102; [I(5),W(23),Bpa(28),Y(36)]PTHrP-(1--36)- amide cross-linked to residues 64--95; and [I(5),W(23), Bpa(33),Y(36)]PTHrP-(1--36)-amide cross-linked to residues 151-172. These data thus predict that residues 23, 27, 28, and 33 of native PTHrP are each near to different regions of the amino-terminal extracellular receptor domain of the P1R. This information helps define sites of proximity between several ligand residues and this large receptor domain, which so far has been largely excluded from models of the hormone-receptor complex.

34 amino acids of PTH and PTHrP are capable of fully activating the PTH/PTHrP receptor (P1R) (1,2). Studies with PTH and PTHrP ligand analogs and receptor chimeras have suggested that the ligands have two distinct functional domains: the amino-terminal residues, which are important for receptor activation; and the carboxyl-terminal residues, which are important for high affinity binding (3). These data furthermore indicate that the ligand's amino-terminal portion interacts with the extracellular loops and the membrane-spanning helices of the receptor, and the ligand's carboxyl-terminal portion interacts with the receptor's amino-terminal extracellular domain (3). A similar pattern of ligand-receptor interaction has been suggested for other members of this class II family of peptide hormone G-protein-coupled receptors, including the secretin receptor (4) and the PTH-2 receptor (5)(6)(7).
Some specific sites of interaction between the amino-terminal portions of PTH-(1-34)/PTHrP-  and the P1R have been identified by site-directed mutagenesis and photoaffinity cross-linking studies. For example, mutational analyses have shown that residues in extracellular loop 3 and the adjacent sixth membrane-spanning helix (TM6) are critical for mediating ligand-induced receptor activation (2,8). Consistent with these mutational data, Bpa introduced at position 1 of PTH-  or position 2 of either PTH-  or PTHrP-(1-36) was found to cross-link to methionine 425, at the extracellular end of TM6 in the P1R (9,10). Interactions between the aminoterminal portion of the ligand and the transmembrane domains/extracellular loops of the receptor are also suggested by a study showing that PTH- (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14) can stimulate cAMP accumulation in a mutant P1R missing most of the amino-terminal extracellular domain as efficiently as it does in the wild-type P1R (11). Furthermore, if a PTH fragment comprising the first 9 amino acids is covalently attached to the amino-terminal end of such a truncated receptor, the resulting ligand-receptor chimera displays constitutive activity, indicative of an intramolecular stimulation of the receptor's activation domain by the tethered ligand fragment (12).
In contrast to the amino-terminal portion of the ligand, less is known about the interaction of the mid-and carboxyl-terminal region of PTH-  and PTHrP-(1-36) with the P1R. Initial studies with different P1R chimeras indicated that the carboxyl-terminal portion of PTH-  interacts with the receptor's amino-terminal extracellular domain (13). This domain in the P1R is quite large (ϳ190 amino acids, including the 22-amino acid signal sequence) and can be further subdivided based on the exons that encode it: S-(1-25), E1-(26 -60), E2-(61-105), E3-(106 -141), and G-(142-181) (14, 15) (see Fig. 1 below). Ligand residue 23 (Phe in PTHrP and Trp in PTH) has been found to be important for high affinity binding to the P1R and to the PTH-2 receptor (16,17). Replacement of this residue with a Bpa photoreactive group results in affinity cross-linking to residues 23-40 of the rat P1R, at the extreme end of the amino-terminal extracellular domain. Two residues within this region, Thr 33 and Gln 37 , were shown to be important for binding of PTHrP-   (18). Residues 24 -34 of PTHrP-   (19) and PTH-   (20) also contribute to high affinity binding to the P1R, and three hydrophobic amino acids within this region of PTH- , residues Leu 24 , Leu 28 , and Val 31 , have been found to be intolerant to substitution with polar residues (17). A PTH analog with a benzoyl group attached to the epsilonamino group of lysine 27 has been shown to cross-link to extracellular loop 1 of the human P1R (21), a result that would not have been predicted by the previous receptor mutagenesis and chimera studies (3,8,13). Affinity cross-linking studies using a PTH-  analog with a photoreactive benzoyl group attached to Lys 13 , in the center of the ligand, demonstrated an interaction to P1R residue Arg 186 , located near the first transmembrane helix (22).
Based upon the available physical data, two computer models have been constructed for the complex formed between PTH-(1-34) and the P1R, and these models include interactions between residues in the carboxyl-terminal portion of PTH-  or PTHrP-  and residues in the amino-terminal extracellular domain of the P1R (23,24). The first model, by Rölz et al. (23), utilized a solution-phase NMR structure of the ligand, which indicates two ␣-helices separated by a flexible linker region. Using the previously described cross-link between residue 13 in PTH-  and Arg 186 of the P1R as a constraint (22), the model aligns the carboxyl-terminal helix of PTH-(1-34) with a putative ␣-helix comprising residues 180 -189 of the receptor, at the proximal portion of the aminoterminal extracellular domain (23). This model was later modified to include the cross-link sites of ligand positions 23 and 27, but the revised model was not assessed by molecular dynamic simulations (25). A second model was reported with the recent x-ray crystallographic structure of PTH-(1-34) (24). The crystal structure showed the peptide to contain an extended ␣-helix spanning nearly the entire chain length (residues 3-32), with a 15°bend in the center (24). Using this structure of the ligand, along with the previously determined cross-linking sites of ligand positions 1 (9) and 13 (22), a detailed alignment was suggested for the carboxyl-terminal region of PTH-(1-34) and the amino-terminal extracellular domain of the P1R. This alignment predicted a hydrophobic interaction between residues Trp 23 , Leu 24 , and Leu 28 of PTH-(1-34) and Phe 173 , Leu 174 of the P1R; and a polar interaction between Lys 27 of PTH-  and Glu 169 of the P1R. The accompanying computer-based model of PTHrP-(1-34) complexed with the P1R predicts the same ligand-receptor interactions, with the exception that Leu 27 of PTHrP-(1-34) also participated in the hydrophobic interactions with residues Phe 173 and Leu 174 of the P1R (24).
To help further elucidate the interactions that occur between the carboxyl-terminal region of the ligand and the P1R, we undertook a photoaffinity cross-linking study utilizing PTHrP-(1-36) analogs with individual Bpa substitutions of residues 22-35. Although not all of the analogs were informative, the overall results establish that multiple regions of the aminoterminal extracellular domain of the P1R are in close proximity to the carboxyl-terminal portion of PTHrP-  Mutagenesis of the Human P1R-Mutations were introduced into single-strand plasmid DNA encoding the wild-type human P1R by oligonucleotide-directed site-specific mutagenesis as described previously (5,26). The oligonucleotide primers were synthesized on an Applied Biosystems model 380A DNA synthesizer. All mutations were verified by nucleotide sequence analysis of plasmid DNA.
Cell Culture/DNA Transfection-LLC-PK 1 cells stably expressing P1R at a density of ϳ950,000 receptors per cell (HKrk-B7) (27) or COS-7 cells were cultured in Dulbecco's modified Eagle's medium (Mediatech, Washington, DC) supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified atmosphere containing 95% air and 5% carbon dioxide. Cells were seeded in 24-well plates for radioreceptor and cAMP assays. For crosslinking experiments, cells were seeded in either 6-well plates (for COS-7 cells transfection) or 15-mm plates (HKrk-B7 cells). For transfection, once the monolayer of COS-7 cells reached ϳ80% confluency, cells were transfected by the DEAE-dextran method as described (28) using 200 ng of plasmid DNA/well in 24-well plates or 800 ng of plasmid/well in 6-well plates. After 3 days, cells were used for the experiments. HKrk-B7 cells were used in experiments once the monolayer reached confluency.
Radiolabeling of Peptides-Radiolabeled peptides were prepared by chloramine-T iodination followed by high pressure liquid chromatography purification using a 30 -50% acetonitrile gradient in 0.1% trifluoroacetic acid over 30 min as previously described (29).
Radioligand-Receptor Binding and cAMP Accumulation Assays-Binding assays were performed as previously described (5). In brief, 125 I-labeled radioligands, rPTH-(1-34) or [I 5 ,W 23 ,Y 36 ]PTHrP-(1-36)amide, were incubated with cells expressing P1R in the presence of varying concentrations (0 -10 Ϫ6 M) of unlabeled peptide. After 4-h incubation at 15°C, the binding mixture was removed and cells were rinsed twice with cold buffer. Cells were lysed, and the entire lysate was counted for gamma irradiation. Intracellular cAMP accumulation was measured by radioimmunoassay as previously described (5).
SDS-PAGE Analysis/Purification-Samples were incubated with SDS-PAGE sample buffer at room temperature for at least 2 h and separated on 10% SDS-PAGE gels. For visualization of intact crosslinking products, dried gels were subjected to autoradiography at Ϫ80°C. For partial purification of cross-linked ligand-receptor complexes, wet gels were cut into strips and counted for gamma irradiation. Gel strips with peak radioactivity were electroeluted in a dialysis bag at 100 V for 2 h. Eluted samples were concentrated using Centricon-10 filters (Millipore Co., Bedford, MA).
Chemical/Enzymatic Cleavage-For cleavage at methionine residues, the gel-purified, radiolabeled ligand-receptor complexes were incubated with CNBr (100 mM) in 70% formic acid at 20°C for 24 h. After digestion, CNBr and formic acid were removed by repetitive lyophilization. For cleavage at lysine or arginine residues, the gel-purified ligandreceptor complexes were incubated with endopeptidase LysC or ArgC, respectively, at 37°C for 24 h. For removal of N-linked glycosyl groups, samples were treated with N-glycosidase F (2500 units) for 3 h at 37°C.
Size Analysis by SDS-PAGE-Digested or mock-digested samples were suspended in SDS-PAGE sample buffer and incubated at room temperature for 2 h. Samples were analyzed by SDS-PAGE (10% acrylamide) according to the method of Laemmli (30) or Tricine gels according to the method of Schagger and von Jagow (31). Dried gels were subjected to autoradiography at Ϫ80°C with an intensifying screen.

RESULTS
Analogs of [I 5 ,W 23 ,Y 36 ]PTHrP-(1-36)-amide with p-benzoyl-L-phenylalanine (Bpa) substituted at positions 22-35 were prepared for photoaffinity cross-linking to the hP1R. The apparent binding affinity and cAMP-stimulating potency of each of these analogs was assessed in HKrk-B7 cells. Most Bpa-substituted analogs competed with the radiolabeled agonist rPTH-(1-34) for hP1R binding with apparent affinities comparable with that of the parent peptide (IC 50 values: 7-74 nM, Table I) and stimulated cAMP accumulation as efficiently as did the parent peptide (EC 50 values: 4 -48 nM, Table I This receptor mutant has been previously shown to have PTH binding and activation properties similar to those of the wild-type receptor (8,13). In contrast to the radiolabeled band seen with COS-7 cells expressing the wild-type hP1R, which was indistinguishable from the band observed with HKrk-B7 cells (data not shown), the CNBr-digested complex obtained with the delta-E2 mutant showed a large, broad band at ϳ50 kDa (Fig. 2, right panel). This size was consistent with a cross-link to the glycosylated receptor fragment that comprises residues 33-60/106 -189. The smaller CNBrgenerated band observed with the wild-type hP1R affinity labeled by 125    Each of these digested ligand-receptor complexes showed a broad band of ϳ50 kDa upon Tricine gel analysis, indicating that the CNBr-generated ligand/receptor conjugate fragment was glycosylated (Fig. 3). These findings were consistent with crosslinking of each of the three Bpa groups to the region of the amino-terminal extracellular domain of the receptor delimited by Glu 64 and Met 189 ; a region that is carboxyl to the [I 5 ,Bpa 23 ,Y 36 ]PTHrP-(1-36)-amide cross-linking site (see Fig. 1).
[I 5 ,W 23 ,Bpa 33 ,Y 36 ]PTHrP-(1-36)-amide contained the most carboxyl-terminal Bpa-substitution analyzed to date for PTH or PTHrP. The complex between radiolabeled [I 5 ,W 23 ,Bpa 33 , Y 36 ]PTHrP-(1-36)-amide and the hP1R was evaluated by enzymatic digestion with LysC or ArgC. Each enzyme digest revealed cross-linking to a large, glycosylated fragment, along with a smaller band comigrating with the 3.5/2.35-kDa size markers (Fig. 4, lanes 1 and 3). The mobility of the smaller band was indistinguishable from that of the free ligand (not shown), suggesting that some cross-linked radioligand was released from the gel-purified complex during the digestion process. After Endoglycosidase-F treatment, the larger band in the LysC-digested sample was reduced to ϳ5 kDa (Fig. 4, lane 2). Only two glycosylated fragments are predicted after complete LysC digestion of the hP1R; these fragments consist of residues 142-172 (LysC-1) and residues 173-240 (LysC-2). Deglycosylated LysC-2, when cross-linked to the [I 5 ,W 23 ,Bpa 33 , Y 36 ]PTHrP-(1-36)-amide, has a calculated mass of 11 kDa, considerably larger than the observed radiolabeled ϳ5 kDa band. The calculated mass of deglycosylated LysC-1 crosslinked to the ligand is 6.5 kDa, much closer to that of the observed band. The LysC digest thus narrowed the [I 5 ,W 23 ,Bpa 33 ,Y 36 ]PTHrP-(1-36)-amide cross-link site to the hP1R region between residues 142-172. Endoglycosidase-F treatment of the ArgC-digested sample eliminated the large, more diffuse band but a prominent new band could not be detected (Fig. 4, lane 4). The faint band at ϳ16 kDa, which appeared after deglycosylation, presumably derived from a partial ArgC or Endoglycosidase-F digestion, because complete ArgC digestion of the P1R is predicted to generate two glycosylated fragments, residues 151-162 or residues 163-179, and the calculated sizes of these receptor fragments (after deglycosylation) when cross-linked to [I 5 ,W 23 1 and 2), position 28 (lanes 3 and 4), or position 33 (lanes 5 and 6) were bound to LLC-PK 1 cells stably transfected with hP1R. Bound ligand was photoaffinity cross-linked to the receptor as described under "Experimental Procedures." Cells were lysed, the ligand-receptor complex was gel-purified, and the resulting sample was placed in formic acid for 24 h in the absence (Ϫ; lanes 1,4,5) or presence (ϩ; lanes 2, 3, 6) of CNBr. Samples were analyzed on Tricine/SDS-PAGE followed by autoradiography at Ϫ80°C. The positions of size markers are indicated in kDa. Below the gel is shown a schematic representation of the amino-terminal, extracellular domain of the hP1R with the location of three methionine residues (q) and the four sites for potential N-linked glycosylation (¦). The shaded region indicates the cross-linked region in the wild-type hP1R for each of the three ligands.
cross-link site lies in a region of the amino-terminal extracellular domain that is encoded by exon G (see Fig. 1).
Mapping of the [I 5 ,W 23 ,Bpa 27 ,Y 36 ]PTHrP-(1-36)-amide cross-link site was next undertaken. Digestion of the covalent ligand-receptor complex with either LysC or ArgC revealed nonglycosylated fragments of ϳ6.5 and ϳ4 kDa, respectively (Fig. 5, lanes 1 and 2). The best candidate segment for the LysC fragment comprised residues Glu 96 -Lys 141 , even though the predicted size of this ligand-receptor complex was 8.1 kDa, i.e. ϳ2 kDa larger than that estimated from the Tricine gel. To assess this plausible assignment of the LysC fragment, a mutant receptor with Lys 141 mutated to arginine (K141R) was prepared. When transiently expressed in COS-7 cells, this mutant receptor showed binding and cAMP activation responses to rPTH-(1-34) that were indistinguishable from those of the wild-type receptor (data not shown). After cross-linking of 125 I-[I 5 ,W 23 ,Bpa 27 ,Y 36 ]PTHrP-(1-36)-amide and LysC digestion of the covalent ligand-receptor complex, COS-7 cells expressing the wild-type P1R showed a radiolabeled band similar to that observed with HKrk-B7 cells (Fig. 5, lane 3). COS-7 cells expressing the K141R mutant receptor yielded a radiolabeled fragment that was much larger (ϳ50 kDa) than that obtained with the wild-type receptor. In addition, this fragment was likely to be glycosylated, as indicated by its mobility and the broad banding pattern (Fig. 5, lane 4). This experiment confirmed that cross-linking of Bpa 27 occurred within the LysC digest fragment comprising residues Glu 96 -Lys 141 . The relative size of the ArgC-derived fragment, in comparison to the LysC fragment, further narrowed the cross-link interval to between receptor residues Glu 96 and Arg 102 . This 7-amino acid segment resides in the portion of the amino-terminal extracellular domain encoded by exon E2, which has been shown to be nonessential for ligand interaction (8,13).
Our findings with [I 5 ,W 23 (Fig. 6). Glycosylation was confirmed by sequential CNBr/Endoglycosidase-F treatment, which resulted in a smaller Ͻ14-kDa fragment (data not shown), consistent with cross-linking to residues 64 -189 in the amino-terminal extracellular domain of the receptor. Thus, each of our three Bpa 27 compounds cross-linked to the same or overlapping segments within the amino-terminal domain of the hP1R, regardless of whether they were PTH-or PTHrP-based analogs. The difference in the cross-linked sites identified by Greenberg et al. (21) and by our current investigations is therefore most likely due to the different configuration of the benzophenone adduct in the ligands used in the two studies.
Mapping of the [I 5 ,W 23 ,Bpa 28 ,Y 36 ]PTHrP-(1-36)-amide cross-link site was of particular interest, because this Bpa substitution is located at an important functional position in the ligand (17). LysC digestion of the covalent complex obtained with this ligand revealed two overlapping nonglycosylated fragments of 4 -5 kDa (Fig. 7, lane 1) Fig. 3, lanes 3 and 4), the cross-link interval for [I 5 ,W 23 ,Bpa 28 ,Y 36 ]PTHrP-(1-36)-amide can be reduced to the region between Glu 64 and Lys 95 . Within this region, complete LysC digestion of the P1R is predicted to generate five fragments; three of which, after cross-linking to 125 I-[I 5 ,W 23 ,Bpa 28 ,Y 36 ]PTHrP-(1-36)-amide, have a predicted size of 4 -5 kDa. The two bands observed experimentally could thus be due to cross-linking to two separate sites in the Glu 64 -Lys 95 interval (although partial enzymatic digestion of this region cannot be excluded). Although we were unable to resolve this mapping further, the results clearly indicate that, similar to the site of [I 5 ,W 23 2 and 4) was bound to LLC-PK 1 cells stably transfected with hP1R. Bound ligand was photoaffinity cross-linked to the receptor as described under "Experimental Procedures." Cells were lysed, the ligand-receptor complex was gel-purified, and the resulting sample was placed in formic acid for 24 h in the absence (Ϫ; lanes 1 and 2) or presence (ϩ; lanes 3 and 4) of CNBr. Samples were analyzed on SDS-PAGE (10% acrylamide) followed by autoradiography at Ϫ80°C. The positions of size markers are indicated in kDa. Below the gel is shown a linear representation of the aminoterminal, extracellular domain of the hP1R with the location of the three methionine residues (q) and the four sites for potential N-linked glycosylation (¦). The shaded region indicates the cross-linked region in the wild-type hP1R. all of these contacts were found to occur in the receptor's amino-terminal extracellular domain. All of the Bpa-substituted analogs activated the P1R with potencies similar to that of [I 5 ,W 23 ,Y 36 ]PTHrP-(1-36)-amide (Table I). Thus, although we cannot determine the actual state of the receptor at the time of photolabeling, it is at least possible that ligand cross-linking occurred to the P1R in its activated, G-protein-coupled conformation. Binding to the P1R was only minimally affected by most of the Bpa substitutions, with the exception of [I 5 ,W 23 , Bpa 24 ,Y 36 ]PTHrP-(1-36)-amide, which showed an ϳ10-fold reduced ability to compete with 125 I-labeled rPTH-(1-34) at the hP1R (Table I). The conserved Leu at position 24 in PTH was previously shown to be intolerant to substitution with polar residues (ϳ1000-fold reduced binding affinity) (17). In contrast, other mutationally intolerant ligand residues nearby (e.g. at positions 23, 28, and 31 (16,17)) tolerated the Bpa substitution with no apparent reduction in binding affinity. At position 23, Bpa replaced a tryptophan, and the structural similarity between these two residues may explain why the substitution was tolerated. At positions 28 and 31, Bpa replaced similarly hydrophobic isoleucine residues; the tolerance of these positions to substitutions with Bpa and intolerance to substitutions with polar residues (17) suggests that these two residues are involved in important hydrophobic interactions.
Among the compounds generated for our studies, the Bpa 23 -, Bpa 27 -, Bpa 28 -, and Bpa 33 -substituted analogs yielded sufficient gel-purified material to permit digestive mapping of the cross-link site in the P1R. Our previous studies with [I 5 ,Bpa 23 ,Y 36 ]PTHrP-(1-36)-amide and the rat P1R (18) demonstrated that cross-linking occurred to a site between the amino-terminal residues 23 and 40. We now extended these studies to include the human P1R, for which we determined that cross-linking occurred to a site within residues 33-63, an extreme amino-terminal receptor segment encoded mostly by exon E1 (see Fig. 1). Taken together, the results with rat and human receptors indicate that residue 23 of PTHrP-  likely contacts the receptor between residues 33-40; a conclusion that is consistent with alanine-scan studies, which identified Thr 33 and Gln 37 as important determinants for PTHrP-  binding (18).
The analog [I 5 ,W 23 ,Bpa 27 ,Y 36 ]PTHrP-(1-36)-amide crosslinked to a site between receptor residues 96 -102, within the region of the amino-terminal extracellular domain encoded by exon E2. The E2-encoded region is not present in the P1R of lower vertebrate species (32,33) and can be removed without affecting receptor functioning (8,13). In addition, residue Lys 27 of PTH has been shown to be moderately tolerant to substitution (17,34). Thus the combined data do not support a critical functional role for the E2-encoded P1R domain or the residue at position 27 of PTH-  or PTHrP- . The cross-linking site recently identified by Greenberg et al. (21) for a PTH analog benzoyl-tagged at Lys 27 in the first extracellular loop was not identified with our Bpa 27 -containing PTH-  or PTHrP-(1-36) analogs, each of which cross-linked to the aminoterminal extracellular domain of the receptor. The divergence of these results most likely reflects the different spatial properties of the photoreactive benzoyl group in the ligands used by the two groups, because it is attached distally to the lysine side chain (Bpz) in the ligand used by Greenberg et al. (21), whereas it is incorporated into the peptide chain as p-benzoyl-L-phenylalanine (Bpa) in each of our analogs. Similar to the lack of functional effect that occurs with the deletion of the E2-encoded region, removal of a large portion of extracellular loop 1 (including the Bpz 27 cross-link site) from the P1R has little or no effect on receptor functioning (28). Thus, the photoreactive moiety at position 27, in either the Bpa or Bpz configuration, cross-links to one of two receptor segments that have been found to be nonessential. Although these data do not identify a critical ligand-receptor contact, they do lead to the important conclusion that the first extracellular loop and portions of the amino-terminal extracellular domain of the P1R are close to each other, at least when the receptor is occupied by agonist ligand.
The analog [I 5 ,W 23 ,Bpa 28 ,Y 36 ]PTHrP-(1-36)-amide crosslinked to a site between receptor residues 64 -95, and thus in a different portion of the exon E2-encoded region from that which reacted with [I 5 ,W 23 ,Bpa 27 ,Y 36 ]PTHrP-(1-36)-amide. However, unlike PTH residue Lys 27 , Leu 28 is intolerant to many amino acid substitutions (17). The reason why Bpa located at an apparently critical position in the ligand (e.g. position 28) crosslinked to a nonessential domain of the receptor is not clear. One possibility is that residue 28 (Leu in PTH, Ile in PTHrP) contacts a critical hydrophobic pocket that is located close to the FIG. 8. Representation of ligand-P1R interaction based on photoaffinity cross-linking studies with different PTH and PTHrP analogs. A, schematic diagram of known cross-link sites between PTH or PTHrP and the P1R (for details see text). Filled circles with adjacent bold numbers indicate ligand residues that were either photoderivatized or substituted with Bpa for affinity cross-linking studies; open circles represent the cysteine residues that are likely involved in the formation of disulfide bonds; small numbers indicate the receptor residues or regions to which cross-linking occurs. B, combining structural information obtained by photoaffinity cross-linking, a rhodopsin/ bacteriorhodopsin-based model of the P1R transmembrane domains, and the disulfide bridging pattern described by Grauschopf et al. (38), a representation of PTH or PTHrP bound to the P1R is shown; disulfide bonds between cysteine residues are indicated by the ∧ symbol.
E2-encoded domain such that the Bpa-substituted analog can cross-link to the E2 region. Another possibility is that residue 28 initially contacts a critical hydrophobic pocket in the P1R, but this contact is not maintained during the time of photolabeling, perhaps because of a conformational change in the receptor. A third possibility relates to the hypothesis that some amphipathic peptide ligands, including PTH and PTHrP (35), associate initially with the cell membrane and then diffuse two-dimensionally to their cell surface receptors, where a more specific high affinity interaction occurs (36). Residue 28 is located on the hydrophobic face of the carboxyl-terminal amphipathic helix and thus, according to this hypothesis, may participate in a nonspecific but important interaction with the lipid component of the cell membrane. If such an interaction with the cell membrane is the primary role of residue 28, then [I 5 ,W 23 ,Bpa 28 ,Y 36 ]PTHrP-(1-36)-amide would not necessarily be expected to cross-link to an essential P1R region. Our current data cannot distinguish between these possible explanations. In any case, it seems unlikely that the proximity between the Bpa moiety in [I 5 ,W 23 ,Bpa 28 ,Y 36 ]PTHrP-(1-36)-amide and receptor residues 64 -95 indicates a role of this receptor region in determining high affinity ligand binding. Nevertheless, the apparent proximity between these specific sites in the ligand and the receptor does provide important topographical information about the receptor's amino-terminal extracellular domain when occupied by agonist ligand.
The analog [I 5 ,W 23 ,Bpa 33 ,Y 36 ]PTHrP-(1-36)-amide crosslinked to a site between residues 151-172 of the hP1R, within the glycosylated region (37) encoded by exon G (see Fig. 1). Within this cross-link interval, the cassette replacement and deletion-based mutational analyses that have been performed so far are largely uninformative, because the receptor modifications tested eliminated or severely diminished cell surface expression (8,28). Our current studies suggest that point mutational analysis of this region may ultimately reveal important functional roles for individual residues in the cross-linked interval.
The cross-linking data presented in our study do not provide support for the components of the two previously mentioned computer-based models of the PTH/PTHrP ligand-receptor interaction that concern the amino-terminal extracellular domain of the receptor and the carboxyl-terminal portion of the ligand. The model by Rölz et al. (23) aligns the carboxyl-terminal helix of the ligand with a putative ␣-helix comprising residues 180 -189 of the receptor. However, none of the crosslinking sites identified in our studies were mapped to this receptor domain. The model proposed by Jin et al. (24) predicts specific receptor interactions for ligand residues Trp 23 , Lys 27 , and Leu 28 of PTH-(1-34), and for residues Phe 23 , Leu 27 , and Ile 28 of PTHrP-(1-34) to an ␣-helical region comprising residues 169 -177 of the receptor. This receptor region is also distant from the cross-linking sites of our corresponding Bpasubstituted PTHrP-(1-36) analogs. In light of these discrepancies between the reported computer models and the new crosslinking data provided herein, it now seems likely that alternative modes of interaction between the carboxyl-terminal helix of PTH-(1-34)/PTHrP-(1-36) and the amino-terminal extracellular domain of the P1R need to be considered.
The cross-linking intervals identified in our studies are well separated from each other in the receptor's primary structure (Fig. 8A). These separations therefore predict that the aminoterminal region of the P1R forms a folded structure in which the identified ligand-receptor contact points converge. Additionally, there are six highly conserved cysteine residues in the amino-terminal extracellular domain of the hP1R (positions 48, 108, 117, 131, 148, and 170) that are likely to be involved in intramolecular disulfide bonds. A plausible pattern of disulfide bonding has recently been determined using an amino-terminal fragment of the P1R, which, although generated in Escherichia coli, binds PTH-(1-34) with the expected affinity (38). Based on these studies, disulfide bridges are likely to occur between residue pairs Cys 48 /Cys 117 , Cys 108 /Cys 148 , and Cys 131 / Cys 170 (38). Although these findings have yet to be confirmed in the full-length P1R, or in an hP1R fragment produced in a mammalian expression system, we find that the proposed pattern of disulfide bridging is at least compatible with the current cross-linking data. Reconciliation of these two sets of data can occur with the receptor's amino-terminal extracellular domain assuming a spiral shape that partially encircles the ligand's carboxyl-terminal helix, as illustrated in Fig. 8B. However, it is important to note that the cross-linking data could also be compatible with other patterns of disulfide bridging, which would presumably give a different overall topology. The Bpz cross-link between ligand residue 27 and extracellular loop 1 of the P1R (21) helps to orient the ligand's terminal helix with respect to the juxta-membrane region of the receptor. Satisfying this constraint, together with those imposed by the crosslinking sites for ligand positions 1 (9) and 13 (22), appears to require greater flexibility between the amino-terminal and carboxyl-terminal regions of the ligand than that observed in the crystalline structure of PTH-(1-34) (24); indeed the degree of folding of the ligand may be greater than that pictured in Fig.  8B. Note that a considerable degree of flexibility between the amino-terminal and carboxyl-terminal domains of PTH-(1-34) has been observed in solution-phase NMR studies (39).
In summary, results obtained in our studies identified four new sites of contact between residues in the carboxyl-terminal portion PTHrP-(1-36) and the amino-terminal extracellular domain of the human P1R. One of these contact sites was also confirmed with a PTH-(1-34) analog. The findings should help constrain 3-dimensional structural models of ligand-P1R interaction, particularly in regards to the receptor's large aminoterminal extracellular domain, which now appears to contain multiple sites that contribute to ligand interaction. Additional studies, aimed at further refining the topology of the P1R binding pocket, may ultimately assist in understanding the fundamental mechanism(s) by which this, and potentially other related family B receptors, bind and respond to their specific ligands.