Structural and physiologic characterization of the mid-region secretory species of parathyroid hormone-related protein.

Parathyroid hormone-related protein (PTHrP) is initially translated as a preprohormone which is posttranslationally processed to yield a family of mature secretory forms. Most attention has focused on the amino-terminal portion of the molecule which is homologous to parathyroid hormone. It is clear, however, that a mid-region species of PTHrP is posttranslationally cleaved from the highly conserved mid-region of PTHrP, and that the amino terminus of this peptide is Ala38. The purposes of the current study were three: 1) to confirm that Arg37 immediately preceding Ala38 serves as a posttranslational processing site in the PTHrP precursor, 2) to determine the carboxyl terminus of the mid-region secretory species of PTHrP, and 3) to synthesize this authentic mid-region secretory form of PTHrP and determine whether it is biologically active. The results indicate that: 1) Arg37 is indeed a processing site in the PTHrP precursor; 2) three distinct mid-region PTHrP species are generated by posttranslational processing, PTHrP(38-94)amide, PTHrP(38-95), and most likely, PTHrP(38-101); and 3) synthetic mid-region PTHrP(38-94)amide is active in four different biological systems. These studies confirm the finding that PTHrP is a prohormone. More importantly, they define a novel, biologically active highly conserved mid-region secretory form of PTHrP.

cellular differentiation and proliferation, in the regulation of smooth muscle contraction, and in transepithelial calcium transport (for a complete review, see Ref. 5). It is now clear that the initial PTHrP translational product is a prohormone which is endoproteolytically cleaved during its passage through the secretory pathway in a fashion analogous to the prohormone processing of proopiomelanocortin, prosomatostatin, and other neurosecretory peptides to yield a group of mature daughter peptides (for detailed reviews, see Refs. 4 and 6). A peptide derived from the amino terminus of proPTHrP, PTHrP , contains the parathyroid hormone-like portion of the molecule ( Fig. 1) (6 -10).
A peptide derived from the mid-region of the PTHrP molecule has been detected in the circulation of patients with humoral hypercalcium of malignancy (see Ref. 11 and references therein). Mid-region PTHrP peptides have also been shown to be produced by a variety of cell types, to result from posttranslational processing of the preproPTHrP, and to be packaged into secretory granules prior to secretion via the regulated secretory pathway (7,8,12). Partial purification of this midregion secretory species of PTHrP has shown that it has an molecular mass, as assessed by SDS-PAGE, of approximately 7000 Da. Amino acid sequencing has shown that it begins at alanine 38 of the cDNA-predicted amino acid sequence ( Fig. 1) (7). This latter observation suggests that the arginine in position 37 is a processing site for a monobasic-specific prohormone convertase such as those which process prochromogranin A, prosomatostatin, and proatrionatriuretic hormone at single arginine or lysine residues (13). In contrast to the clear indication that mid-region PTHrP begins at Ala 38 , no information is available regarding the carboxyl terminus of the peptide. As a result, authentic mid-region PTHrP peptides have not been available for study.
The cDNA-predicted amino acid sequence of the mid-region of PTHrP is extremely highly conserved among species: the human, rat, mouse, baboon, chicken, and dog sequences vary by only one to three amino acids in the PTHrP(38 -101) region (1)(2)(3)(4)(5)(6). This striking evolutionary pressure to conserve sequence led to the prediction early on that peptides derived from this region would prove to be biologically active and developmentally important. Using synthetic peptides of arbitrary length derived from this region of PTHrP, several investigators have shown this prediction to be correct. For example, Care et al. (14) have shown that PTHrP(67-86)amide, PTHrP(75-84), and PTHrP(75-86)amide stimulate calcium transport across the placenta from the maternal to the fetal circulation in sheep. Orloff et al. (15) have reported that PTHrP(67-86)amide stimulates cytosolic calcium and inositol phosphates incrementally in human squamous carcinoma cells. Luparello et al. (16) have shown that PTHrP(67-86)amide inhibits the mitogenesis but stimulates metastatic potential of the human breast carcinoma line, 8701-BC. These findings are complemented by the observations of Karaplis et al. (17) demonstrating that disruption of the PTHrP gene produces a lethal outcome, although, since these studies disrupted the entire PTHrP coding region, the precise contribution of the mid-region secretory form(s) of the PTHrP to this lethality remain undefined. Thus, it seems clear that a mid-region secretory form of PTHrP exists and that it subserves a variety of biologic functions, but, as noted above, further study of the physiology of this peptide is hampered by the lack of information defining its precise structure.
The purposes of the current study were therefore 3-fold. First, we wanted to confirm that Arg 37 is indeed a posttranslational processing site in proPTHrP. We approached this issue through site-directed mutagenesis in which Arg 37 was replaced by other amino acids. Second, we wanted to determine the carboxyl terminus of mid-region secretory form of PTHrP, so that the complete structure of this peptide would be defined. We approached this question through standard protein purification techniques combined with tryptic digestion and mass spectroscopy. Finally, we wanted to determine whether this authentic mid-region secretory form of PTHrP contains biological activity. We approached this goal by synthesizing the peptide and examining its effects on cytosolic calcium and adenylyl cyclase in a panel of PTHrP-producing and PTHrP-responsive cell lines, and by examining its stimulatory effect on placental calcium transport in vivo. The results presented herein define in structural terms a new mid-region secretory species of PTHrP and demonstrate that this peptide is indeed biologically active in four different systems. These findings will permit the further elucidation of the normal physiologic roles of mid-region PTHrP.
Cell Lines and Culture-The rat insulinoma (RIN) 1046-38 cell line is a pancreatic ␤ cell line, was generously provided by Dr. Michael Appel, and was cultured as described previously (7,8,12,20) using 10% fetal bovine serum in RPMI 1640 medium supplemented with penicillin, streptomycin, and glutamine. A-10 cells are rat fetal aortic vascular smooth muscle cells and were purchased from American Type Culture Collection (Rockville, MD). YCC SQ-1 cells are human squamous carcinoma cells derived from a carcinoma of the cervix and have been described in detail previously (21).
Creation and Expression of Arg 37 PTHrP(1-141) Mutants-Three 24-mer oligonucleotides encoding peptides in which Arg 37 was mutagenized to Lys 37 , Ala 37 , or Phe 37 were synthesized and used together with a second selection primer which converts a unique SspI site in pGEM to an EcoRV site (Transformer Mutagenesis Kit, Clontech, Palo Alto, CA). These primers were annealed to a pGEM-hPTHrP(1-141) construct. After standard heat denaturation, annealing, extension and ligation steps, the mixture of mutagenized and non-mutant plasmids were transformed into the repair-defective Escherichia coli strain, BMH 71-18 mut S, and plasmid preparations were made. Residual wild-type plasmids containing the SspI site, were cleaved with SspI, rendering them inefficient in transformation, and the remaining mutated plasmids were transformed into DH5␣ E. coli. Since the Arg 37 mutants also contained the new EcoRV site, the mutants could be identified by restriction mapping. Confirmation that the desired mutants were created was accomplished by direct DNA sequencing. The inserts were then directionally cloned into the pLJ vector which we have used previously to overexpress wild type PTHrP in RIN cells (7,8,12). Plasmid preparations of each of the three mutated plasmids were prepared, and the wild-type and the three mutant pLJ-PTHrP(1-141)s were stably transfected into RIN cells using LipofectAMINE as we have reported previously (7). Four to ten clones from each mutant RIN cell line were selected based on overexpression of PTHrP as determined using the PTHrP(37-74) RIA (see below).
Production of Mid-region PTHrP for Purification-Mid-region PTHrP for purification purposes was derived from RIN cells transfected with and overexpressing PTHrP(1-139). RIN cells were selected for they have been shown to faithfully process other neuroendocrine pep-FIG. 1. Panel A, the three initial PTHrP posttranslational products. Each begins with a common "prepro" sequence (amino acids Ϫ36 to Ϫ1), and each contains a common sequence in the 1-139 region. The three protein isoforms differ after amino acid 139, with the first isoform terminating at this point, the second extending an additional two amino acids, and the third extending an additional 34 amino acids. Note that each is rich in the basic amino acids arginine (R) and lysine (K) and that these are often arranged in clusters. The basic residues in positions Ϫ5 to Ϫ1 and at ϩ37 are known to be prohormone processing sites. Panel B, PTHrP functional domains. Amino acids 1-13 are homologous with parathyroid hormone. Amino acids 14 -36 contain no homology with PTH but share conformational homology with the corresponding region of PTH and permit interaction of PTHrP with the PTH/PTHrP receptor. Amino acids 38 -111 are very highly conserved among species and are presumed therefore to have important but as yet undefined physiological functions. Amino acids 107-139 are flanked by putative multibasic processing sites. A synthetic peptide from this region has been synthesized and shown to have antiresorptive properties in osteoclasts. This peptide has therefore been provisionally named "osteostatin." The 140 -173 region is unique to humans. No function has yet been identified for this region of the peptide. Panel C, the two radioimmunoassays used in this report are directed against PTHrP  and PTHrP . Panel D, current understanding of the posttranslational processing of PTHrP. As noted above, the Ϫ36 to Ϫ1 region contains a signal peptide (SP) and a "pro" peptide (P). The precise site of signal peptidase cleavage is unknown. Cleavage at the Ϫ5 to Ϫ1 site and at Arg 37 yields PTHrP . The cleavage at Arg 37 also yields a mid-region peptide which is the focus of this report. The carboxyl terminus of this peptide is shown as a dotted line to indicate that its termination has not yet been determined. PTHrP(107-139) is shown as a partially dotted box to indicate that, while a peptide derived from this region is known to exist, its precise amino terminus and carboxyl terminus have not been defined. A peptide from the PTHrP(141-173) region is shown as a dotted box to indicate the potential for the existence of such a tail-region PTHrP peptide. tides in general (22) and to process and secrete mid-region PTHrP in a manner identical with that of human cells such as keratinocytes and renal carcinomas (7). Cells were grown to confluence in RPMI 1640. Fresh medium (serum-free) was added to the cells and was harvested after 90 min of exposure to the cells. These conditions were selected for we have previously shown that RIN cells do not significantly degrade PTHrP within this time frame (7) and that the mid-region secretory form of PTHrP found in the medium under these conditions is chromatographically identical to that found within cells prior to secretion and prior to exposure to extracellular proteases. Thirteen liters of 90-min conditioned medium were harvested in this manner. The medium was promptly frozen and stored until ready for the first HPLC step described below.
Reversed-phase and Size Exclusion HPLC-HPLC was performed using a Waters HPLC system. Reversed-phase HPLC was performed using Vydac (Separations Group, Hesperia, CA) or Brownlee (Rainin Instruments, Emeryville, CA) analytical, preparative, and microbore HPLC columns, with flow rates and gradients as described in the figure legends and under "Results." Size exclusion HPLC was performed using either Waters I-125 Protein Pak or SW 300 columns or a Shodex KW 802.5 column (Waters Inc., Milford, MA) as described in the figure legends.
Mass Spectroscopy, Peptide Sequencing, Amino Acid Analysis, Tryptic Digestion-Peptide sequencing was performed using an ABI model 470A gas phase peptide sequencer (7). Mass spectroscopy was performed using a VG/Fisons Tof/Spec laser desorption mass spectroscope (VG Analytical, Manchester, UK). Tryptic digestion was performed on approximately 80 pmol of either PTHrP(1-108) or purified mid-region PTHrP as described under "Results." The digests were analyzed using a Vydac 2.1 ϫ 250-mm C18 microbore reversed-phase HPLC column as shown in Fig. 5. Amino acid analysis was performed as described previously (24).
Bioassays-Adenylyl cyclase assays were performed on confluent cells which had been serum-deprived for 24 h as we have described (9, 18 -20, 25). Briefly, cells were exposed to the peptides as described in Fig. 10 in the concentrations indicated for 10 min. The medium was then removed, and the cells were extracted using 4% perchloric acid and kept at 4°C for 30 min. The extracts were neutralized using 1 ml of 1:1 (v:v) Freon:tri-n-octylamine and vortexed. Phase separation was accomplished by microcentrifugation. The extracts were then diluted in 50 mM Tris-HCl, pH 7.4, before cAMP RIA. Cyclic AMP was quantitated by radioimmunoassay (Biomedical Technologies Inc., Stoughton, MA) and expressed as picomoles of cAMP produced per well.
Cytosolic calcium was measured using the calcium indicator fura-2 as described in detail previously (20,21,26). Briefly, cells were plated on glass coverslips and loaded with fura-2-AM (4 M) for 40 min at 37°C. The cells on the coverslip were placed in a perifusion system within a Perkin-Elmer LS-5B spectrofluorometer and perfused with bicarbonate buffer to which peptides were added in the concentrations as shown in Fig. 8. Excitation was at 340 and 380 nm. Fluorescence was monitored at 510 nm (5-nm bandwidth). The inset in panel A shows the position of Arg 37 in the precursor peptide. The extracts were resolved using a 15-38% acetonitrile: water gradient in 0.1% trifluoroacetic acid with a Vydac TP 218104 C18 reversed-phase HPLC column at a flow rate of 0.5 ml/min. The fractions were assayed for immunoreactivity using the PTHrP(1-36) RIA (closed symbols) and the PTHrP(37-74) RIA (open symbols). Note that the wild-type construct yields an early eluting peak with mid-region but no amino-terminal immunoreactivity. This mid-region PTHrP species has previously been purified and NH 2 -terminally sequenced and shown to begin at Ala 38 . Note also that none of the three mutant constructs yielded this mid-region peptide. Instead, in each case, mid-region immunoreactivity is shifted to a later elution position and co-migrates with amino-terminal immunoreactivity. This suggests that cleavage at Arg 37 did not occur in the mutants.
In Vivo Placental Perfusion-These experiments were performed as described previously (14,27). Briefly, pregnant ewes of known conception date were used. Under general anesthesia (halothane) each placenta was isolated from its fetus and perfused in situ via the umbilical vessels using a semiclosed system in which the flow rate and perfusion pressure were kept constant (27). Before catheterization of the umbilical vessels, the fetus was intravenously injected with 500 units of sodium heparin and 1 mg of acetyl promazine. Washout of fetal blood was accomplished by perfusion with sterile tissue culture medium, Medium 199 (Sigma). The placenta was then perfused with the blood substitute, Fluosol-43 (Green Cross Corp, Osaka, Japan) which contains a fluorocarbon as an oxygen carrier. The reservoir containing Fluosol-43 was stirred continuously throughout the perfusion. The calcium ion concentrations in the placental effluent were monitored at 15-min intervals until they had reached a plateau or constant rate of change. The plasma calcium ion concentration in the ewe was monitored at the same time intervals. Test and control peptides were dissolved in 30 l of 0.01 M acetic acid, added to 900 l of 140 mM sodium chloride containing 0.01% bovine serum albumin, and injected into the arterial inflow to the placenta after the establishment of a plateau. The alteration in the plateau or steady rate of change of calcium ion concentration in the effluent perifusate was calculated and is presented in Fig. 9 as the percent increment in placental calcium flux following administration of the test peptide.

Site-directed Mutagenesis of Arg 37 Prevents Prohormone
Cleavage-We have previously reported that the mid-region secretory form of PTHrP begins at Ala 38 , findings which would suggest that Arg 37 serves as a substrate for a prohormone convertase with monobasic specificity (7). In order to test this hypothesis more rigorously, we used site-directed mutagenesis to construct three PTHrP cDNAs in which the codon for Arg 37 was changed to encode Ala 37 , Phe 37 , or Lys 37 . These three mutant constructs as well as a fourth construct encoding wildtype PTHrP(1-141) were stably transfected into RIN cells, an islet cell line which has previously been shown to produce and secrete low levels of PTHrP and which appear to process PTHrP in a fashion representative of other cell types (7,8,12). RIN cell extracts were resolved using reversed-phase HPLC as shown in Fig. 2, and the resulting fractions were assayed using the PTHrP(1-36) and PTHrP(37-74) RIAs. While the midregion peptide is present in both conditioned medium and cell extracts, cell extracts were selected for study since the addition of guanidinium isothiocyanate would immediately terminate processing and would prevent artefactual proteolysis. As can be seen in Fig. 2A and as described previously (7,8,11,12), wild-type PTHrP-expressing cell lines contain an early eluting PTHrP peptide with mid-region immunoreactivity, as identified using the PTHrP(37-74) RIA, but which is devoid of aminoterminal PTHrP immunoreactivity, as identified using the PTHrP(1-36) RIA. This is the mid-region form of PTHrP which has previously been NH 2 -terminally sequenced and shown to begin at Ala 38 (7). In contrast to these findings using wild-type PTHrP-expressing RIN cells, each of the three mutant PTHrPexpressing RIN cell lines, while producing copious quantities of Note that the mid-region immunoreactivity from the wild-type constructs elutes with an apparent M r which is slightly less that of PTHrP  or PTHrP , and contains no amino-terminal immunoreactivity. In contrast, the mid-region immunoreactivity from the three mutants elutes with a larger apparent M r and co-elutes with amino-terminal immunoreactivity. These findings confirm that while processing of the wild-type PTHrP precursor does occur, processing at position 37 does not occur to a significant extent in the three mutants.
PTHrP, failed to contain the mid-region PTHrP fragment. Instead, each of the mutant PTHrP-expressing cell lines contained a later-eluting PTHrP(37-74) peak which co-eluted with PTHrP(1-36) immunoreactivity. These findings suggest that cleavage at position 37, which would yield two separate PTHrP peptides recognized by the two RIAs, did not occur in the three mutant PTHrPs.
In order to confirm that cleavage at Arg 37 failed to occur, the fractions which contained mid-region immunoreactivity in Fig.  2 were pooled and rechromatographed using size exclusion HPLC. As can be seen in Fig. 3A, the mid-region PTHrP peaks from Fig. 2A, when analyzed by size exclusion HPLC, migrated with an M r similar to or smaller than synthetic amino-terminal PTHrP and mid-region PTHrP. In addition, as expected, the peak contained no amino-terminal PTHrP immunoreactivity. In striking contrast to the pattern observed with wild-type PTHrP, mid-region immunoreactivity for each of the three mutant PTHrP lines migrated on size exclusion HPLC with an molecular mass similar to that of PTHrP(1-86) (approximately 10,000 Da). In addition, for each mutant, amino-terminal and mid-region PTHrP immunoreactivity co-eluted. Taken together, these findings provide strong evidence that Arg 37 is indeed a processing site which is employed in PTHrP-secreting cells during biosynthesis.
Purification of Mid-region PTHrP-In order to determine the carboxyl terminus of the mid-region species, it was necessary to purify the peptide to homogeneity. With this goal in mind, RIN cells stably transfected with a wild-type PTHrP(1-139) construct described previously (7,8,12) were grown to confluence and their medium harvested under conditions of protease protection. Thirteen liters of mid-region PTHrP-containing medium were subjected to a purification scheme which is shown in Table I and which included two sequential preparative reversed-phase HPLC steps followed by an analytical reversedphase HPLC step. At the conclusion of this third step, a 2000fold purification had been achieved (Table I). The peak of midregion PTHrP immunoreactivity was then resolved using a size exclusion HPLC column as shown in Fig. 4 (inset). As can be seen in Fig. 4, mid-region PTHrP immunoreactivity co-migrated with the 6500 M r standard. The three peak fractions from this size exclusion step were then further resolved using a microbore C18 reversed-phase HPLC column as shown in Fig.  4. This step yielded three optical density peaks; mid-region PTHrP immunoreactivity co-migrated with the large middle protein peak (Fig. 4, peak b). This apparently homogeneous mid-region PTHrP species was then subjected to amino-terminal sequencing. Thirteen cycles of NH 2 -terminal sequence were obtained which identified the NH 2 terminus of the mid-region peptide as Ala 38 . These findings confirm those reported previously (7) and would appear to provide further support for the hypothesis that Arg 38 is a prohormone convertase site in pro-PTHrP. Moreover, since only a single sequence was observed, these findings would support the suggestion from Fig. 5 that the major protein peak (peak b) is mid-region PTHrP and that it is homogeneous.
This apparently homogeneous mid-region PTHrP secretory peptide was then examined using laser desorption mass spectroscopy. The results are shown in Fig. 5 and were surprising: instead of containing a single mass, three distinct peptide were observed, with molecular masses of 6352, 6409, and 7195 Da. These masses are consistent with identification as PTHrP (38 -94) or PTHrP(38 -94)amide, which have predicted masses of 6355 and 6356 Da, respectively; with PTHrP(38 -95), which has a predicted mass of 6413 Da; and with PTHrP(38 -101), which has a predicted mass of 7212 Da. The observed masses of the first two peptides are sufficiently close to their predicted masses to confirm their structure. The observed mass of the third peptide, 7195 Da, is suggestive but not conclusive of identity with PTHrP(38 -101).
Tryptic Digestion and Analysis of Mid-region PTHrP-The findings described thus far were consistent with two general possibilities. First, it was possible that peak b in Fig. 4 contained three mid-region PTHrP species all of which began with Ala 38 , but each of which terminated at a different amino acid as described in the preceding paragraph. Alternatively, it was possible that this peak contained three unrelated peptides, one being mid-region PTHrP, and the two other being contaminating peptides which are NH 2 -terminally blocked and therefore failed to sequence but which were readily observed on mass spectroscopy. In order to decide between these two possibilities, 80 pmol of the mid-region peptide shown in Fig. 4 was subjected to tryptic digestion, and the resulting tryptic digest was resolved on microbore reversed-phase HPLC (Fig. 6, middle  panel). The resulting cleavage products were compared to those resulting from the tryptic digest performed under identical conditions of 80 pmol of recombinant PTHrP(1-108) (Fig. 6,  upper panel). As can be seen in Fig. 6, the tryptic digest of the purified mid-region peptide revealed four dominant fragments. In each case, corresponding fragments were generated by the tryptic digest of PTHrP(1-108). These peaks were identified by mass spectroscopy and by amino acid sequencing as PTHrP-(38 -53), PTHrP(54 -58), PTHrP(59 -65), and PTHrP(67-79). Importantly, no extraneous peaks unrelated to the PTHrP(1-108) digest were observed in the purified mid-region PTHrP digest. Collectively, these findings are most consistent with the hypothesis that three, or perhaps four, distinct mid-region secretory forms of PTHrP are produced by RIN cells: PTHrP(38 -94) (with either a free carboxyl terminus or an amidated carboxyl terminus), PTHrP(38 -95), and PTHrP(38 -101). These are shown schematically in Fig. 7.
Physiologic Characterization of Synthetic Mid-region PTHrP-We chose one of these peptides, PTHrP(38 -94)amide, for synthesis in order to determine if this peptide would prove to be biologically active. Three different cell lines were selected as target cell lines for bioassay: RIN cells which correspond to beta cells of the pancreatic islet, A-10 cells which are fetal aortic vascular smooth muscle cells, and YCC SQ-1 cells which are human squamous carcinoma cells. These three cell lines were selected since they represent tissue types which have all been demonstrated to produce PTHrP and to have receptors and physiologic responses to the amino-terminal secretory form of PTHrP, PTHrP(1-36) (1-6, 20, 21, 28). The results of these studies are shown in Fig. 8. Each of the three cell lines was examined for adenylyl cyclase response to PTHrP(38 -94) amide. Each cell line was exposed to a known agonist of adenylyl cyclase as a positive control: PTHrP(1-36) for A-10 cells (28), glucagon-like peptide-1 (7-36) for RIN cells (20), and isoproterenol for YCC cells (21). Each cell line responded appropriately to control agonists of adenylyl cyclase. In contrast, none of the cell lines demonstrated a response to PTHrP(38 -94) amide at 10 Ϫ6 M as shown, nor at doses from 10 Ϫ12 to 10 Ϫ7 M (not shown).
In contrast to the failure of PTHrP(38 -94)amide to act as an agonist of adenylyl cyclase in the three cell lines examined, it was a potent agonist of cytosolic calcium in each of the three cell lines, as shown in Fig. 8, and acted at doses which are within the physiologic range. As can be seen in Fig. 8, doses as low as 10 Ϫ12 M stimulated transients in cytosolic calcium in RIN and YCC cells. A-10 cells were responsive to 10 Ϫ9 M midregion PTHrP. Collectively, these findings demonstrate that PTHrP(38 -94)amide is active biologically, and is capable of signaling through changes in intracellular calcium pathways in at least three different cell types, all of which produce PTHrP and all of which also respond to amino-terminal PTHrP.
Finally, since it had been suggested that an authentic midregion secretory form of PTHrP would prove to stimulate placental calcium transport in vivo (14), we determined the effects of PTHrP(38 -94)amide on transplacental calcium flux in placentae perfused in situ in sheep. The results of these studies are shown in Fig. 9 and indicate that in all six of six placentae perfused, PTHrP(39 -94)amide was indeed a potent agonist of transplacental calcium flux, whereas control peptides, including the closely related peptide, PTHrP(37-74), had no effect. DISCUSSION Confirmation that Arg 37 Is a Cleavage Site for PTHrP in RIN Cells: Mutagenesis Studies-A number of prohormones are posttranslationally cleaved at a single basic residue, typically at arginine but occasionally at lysine, by prohormone convertases during biosynthesis. This subject has been extensively reviewed recently (13). Substrate prohormone sites for monobasic-specific prohormone convertases identified to date number more than 100 and include single basic residues in prosomatostatin, procholecystokinin, proatrial natriuretic peptide, prochromogranin A, and many others. The specific prohormone convertase(s) responsible for these cleavages have yet to be fully characterized, although it has been suggested that a mammalian homologue of yeast aspartyl protease-3 may be such an enzyme. As described earlier, the studies of Soifer et al. (7) have shown that the mid-region PTHrP peptide begins at Ala 38 , and that, by inference, Arg 37 is a monobasic endoproteolytic cleavage site. That Arg 37 is truly a cleavage site is strongly supported by the fact that the Arg 37 -Ala 38 cleavage follows precisely the rules and tendencies of Devi (13) for cleavage at monobasic residues, including the presence of a basic amino acid (histidine in PTHrP) in position Ϫ5 relative to the cleavage site. On the other hand, formal proof that this is a cleavage site requires demonstration that the peptide is not cleaved if this arginine is changed to a nonbasic amino acid. In addition, it is important to determine whether the PTHrP monobasic cleaving enzyme is specific for arginine or whether it can cleave at lysine residues as well. This is critical in trying to decipher whether the PTHrP monobasic cleaving enzyme is similar to or different from the monobasic enzymes which cleave somatosta- FIG. 4. Inset, size exclusion HPLC of the partially purified mid-region PTHrP species derived from 13 liters of RIN cell medium. The M r markers are albumin (68,000), cytochrome C (12,300), aprotinin (6,500), and PTHrP(107-138) (3,300). The column used was a Shodex KW 802.5 (Waters Inc., Milford, MA), and the mobile phase was the same as that shown in Fig. 3. Note that the partially purified mid-region PTHrP co-elutes with the aprotinin (6,500) standard. In the main figure, microbore reversed-phase HPLC of the three most active fractions from the inset was performed using a Brownlee C8 Aquapore RP-300 10 ϫ 250-mm column in the same mobile phase as described in the previous figures at a flow rate of 0.1 ml/min. The thick line indicates the chromatogram resulting from a blank injection immediately prior to the injection of the mid-region fractions from the inset shown as the thinner line. Note that three OD peaks are observed in the latter, designated as fractions a, b, and c, all eluting in the vicinity of 70 -75 min. Mid-region immunoreactivity was present in fraction b, but absent from fractions a and c. tin, cholecystokinin, and atrial natriuretic peptide and which have been partially characterized in intestine and in cardiac atria. For example, the intestinal somatostatin monobasic cleavage enzyme which has been partially characterized by Bourdais et al. (29) cleaves at a single arginine but not at a single lysine.
The studies shown in Figs. 2 and 3 confirm that Arg 37 is a prohormone convertase substrate site. While the wild-type PTHrP protein expressed in RIN cells is cleaved as we have previously reported in RIN cells, in human keratinocytes, human renal carcinoma cells and in Chinese hamster ovary fibroblasts (7), none of the three Arg 37 mutant PTHrP precursors was so processed. The amino acids selected for substitution were selected to be either very similar to arginine (i.e. the lysine mutant), to be hydrophobic (the phenylalanine mutant), or to be minimally disruptive (the alanine mutant). Since the requirements for a basic amino acid is central to the specificity of monobasic-specific prohormone convertases, it was not surprising to observe that the Ala 37 and Phe 37 mutants were not processed normally. It was somewhat surprising, however, to find that the Lys 37 mutant was not processed. These findings emphasize the specificity of the putative monobasic-specific PTHrP Arg 37 prohormone convertase. With the clear demonstration that Arg 37 is a prohormone convertase substrate site, studies can now focus on the characterization of the responsible enzyme.

Determination of the Carboxyl Terminus of Mid-region
PTHrP-Prior studies had suggested that the mid-region PTHrP peptide has a molecular mass, as determined by SDS-PAGE, of approximately 7,000 daltons, and that it extends, as determined by direct amino acid sequencing, at least to amino acid 71 (7). In an effort to precisely define the carboxyl terminus of the peptide, we undertook a large scale purification of this low abundance peptide. Sequential purification steps yielded a peptide with an apparent M r as determined using size exclusion HPLC of approximately 7,000 (Fig. 4), agreeing nicely with prior estimates derived from SDS-PAGE (7). Microbore RP-HPLC yielded a major symmetrical OD peak (peak b in Fig. 4) suggesting that the peptide in this fraction might be homogeneous, and NH 2 -terminal amino acid sequencing demonstrated that the peptide was indeed mid-region PTHrP, beginning as reported previously (7), at Ala 38 . Further, NH 2terminal sequencing studies identified only a single sequence, suggesting again that the peptide was homogeneous. In contrast, mass spectroscopy (Fig. 5) indicated the presence of three different peptides. The masses identified were sufficiently close in mass to define two of the peptides as 1) either PTHrP (38 -94) or PTHrP(38 -94)amide, and 2) PTHrP (38 -95). The mass of the third peptide was compatible with PTHrP(38 -101), but the predicted versus observed masses were insufficiently close to confirm identity. Taken together, the mass spectroscopic and the NH 2 -terminal sequencing findings were consistent with two major possibilities. First, it was possible that peak b in Fig.  4 contained three peptides, one or two of which were a PTHrP  Fig. 4. In contrast to the observations in Fig.  4 suggesting that peak b might represent a single peptide, these findings suggest that peak b contains peptides with three different masses, all of which are in the 7000 range, consistent with the size exclusion HPLC findings shown in Fig. 4. The first has a mass of 6352 and is very close to the predicted masses of PTHrP(38 -94)amide (6356) and PTHrP(38 -94) (6355). The second with a mass of 6409 is very close to the predicted mass of PTHrP(38 -94) (6413). The third has an apparent mass of 7195 and may represent PTHrP(38 -101) which has a predicted mass of 7212. These findings are compatible with either of two interpretations as described in the text. species beginning at Ala 38 and the remaining peptide or peptides were contaminating peptides with no homology to PTHrP and which were amino-terminally blocked and therefore failed to yield an NH 2 -terminal amino acid sequence. Second, it was possible that the three masses observed represented three PTHrP species, all of which began at Ala 38 , but each of which terminated at a different amino acid. In order to determine which of these possibilities was correct, tryptic digest studies were performed on both recombinant PTHrP(1-108) and on peak b. As shown in Fig. 6, the digests of both PTHrP(1-108) and of peak b yielded four major peptides. These could be identified on the basis of mass and sequence as being PTHrP(38 -53), PTHrP(54 -58), PTHrP(59 -65), and PTHrP-(67-79). Inspection of the lower panel of Fig. 6 indicates that the other tryptic fragments of PTHrP in the 80 -101 region would not likely have been identified. Conversely, the peaks which are present in the digest of PTHrP(1-108) but absent in mid-region PTHrP most likely represent amino-terminal fragments of PTHrP derived from the 1-37 region. Importantly, the tryptic digest studies failed to identify tryptic peptides derived FIG. 7. Summary of posttranslational processing of preproP-THrP. The top diagram represents preproPTHrP as described in Fig. 1. The complete posttranslational processing as currently understood is presented below. See Fig. 1 for details.   FIG. 6. Upper panel, trypsin digest of 80 pmol of recombinant PTHrP(1-108). Middle panel, trypsin digest of 80 pmol of purified mid-region PTHrP from peak b in Fig. 4 (top line) and a "blank" HPLC chromatogram performed immediately prior to loading the tryptic digest of peak b (bottom line). Bottom panel, schema showing the potential and actual cleavage products iderntified by the two trypsin digests in the upper and middle panels. Peaks a, b, c, and d as well as aЈ, bЈ, cЈ, and dЈ in the two top panels were identified by both amino acid sequencing and mass spectroscopy as being tryptic fragments a, b, c, and d in the bottom panel. Peptides w, x, y, and z in the bottom panel are too small to have been identified by this method but were presumably generated. The findings indicate that the four labeled peaks in the upper and middle panels are the same and that they represent all of the peptides that would be expected to be present in a trypsin digest of a midregion PTHrP derived from the (38 -101) region. The additional tryptic fragments in the upper panel not present in the middle panel presumably represent aminoterminal tryptic fragments of PTHrP which include portions of the 1-36 region. These were not subjected to identification by mass or sequence. Finally, the finding that no peptides resulted from tryptic digest of peak b (middle panel) other than those expected from PTHrP and found in the upper panel indicates that peak b contains only mid-region PTHrP and no contaminating peptides. from peak b, which were not observed in the PTHrP(1-108) digest, making it unlikely that peak b was contaminated with peptides not derived from the mid-region of PTHrP. Taken together, these observations provide strong support for the second hypothesis presented above, namely that peak b contains three mid-region PTHrP species all of which begin at Ala 38 and all of which terminate at a different amino acid, and that these peptides are: 1) PTHrP (38 -94) or PTHrP(38 -94)amide, 2) PTHrP (38 -95), and 3) (most likely) PTHrP (38 -101). It is important to note that the difference in predicted mass between PTHrP(38 -94) and PTHrP(38 -94)amide is sufficiently small so that current mass spectroscopic techniques cannot distinguish between these two peptides. Direct verification as to whether or not the carboxyl terminus is amidated will require further studies. For the reasons described below, however, we believe that PTHrP(38 -94)amide will prove to be the authentic form of the peptide.
These observations are consistent with the suggestion that the multibasic amino acids in the region of amino acids 87-106 ( Fig. 1) are substrate sites for the subtilisin family of prohormone convertases such as furin, PACE 4, PC1/3, and PC2 (5,30). In support of this possibility, Diefenbach-Jagger et al. (31) have reported that recombinant kexin is able to cleave recombinant PTHrP(1-141) following residues 97, 105, 106 and 108. However, these studies were performed in vitro using recombinant peptides. In contrast, the current studies were performed in intact cells. These studies indicate that prohormone cleavage in the PTHrP precursor in the region of Lys 96 Arg 97 Lys 98 , perhaps followed by trimming of the resulting peptide by carboxypeptidase H, is very likely to occur and would appear to result in the production of PTHrP (38 -95), one of the peptides which was observed. As discussed below, this peptide may be a processing intermediate in the production of PTHrP(38 -94)amide, or could be a glycine-extended form of the peptide as occurs for gastrin. Similarly, cleavage between Glu 101 and Lys 102 , or cleavage in the 102-106 region followed by trimming by carboxypeptidase H, would yield PTHrP (38 -101). Importantly, all of the requisite enzymes (furin, PACE 4, PC1/3, PC2, and carboxypeptidase H) have been reported to be present and operative in RIN cells (31)(32)(33)(34). It is somewhat surprising, in the above context, that a peptide with a mass consistent with PTHrP(38 -86)amide or with PTHrP(38 -87) was not observed, since the 88 -91 region (Fig. 1) could possibly serve as a prohormone convertase substrate. The current studies do not preclude the existence of these peptides, nor of other candidate peptides such as PTHrP (38 -96): it is possible that they do exist, but that they were lost as "shoulders" of chromatographic peaks were "shaved" during purification. However, it is likely that such peptides, if they were to be generated during PTHrP biosynthesis, are present in lesser amounts than the three peptides described above.
In addition to the subtilisin family of prohormone convertases and to monobasic prohormone convertases, RIN cells also contain another posttranslational processing enzyme, peptidyl ␣-amidating mono-oxygenase or PAM (35,36). PAM performs carboxyl-terminal amidation of peptides which contain the amino acid sequence "X-Gly-dibasic." Inspection of the PTHrP sequence indicates that two potential amidation sequences are present at amino acids 86 -89 (Pro-Gly-Lys-Lys) and at 94 -97 (Pro-Gly-Lys-Arg) (Figs. 6 and 7). As noted in the preceding paragraph, there is no current evidence for the existence of PTHrP (38 -87). However, the studies described herein did identify PTHrP (38 -94). While the current studies do not define whether or not the carboxyl-terminal proline of this peptide is amidated, it seems most likely that the carboxyl terminus of PTHrP(38 -94) is amidated for several reasons: 1) there is no likely processing mechanism which would yield nonamidated PTHrP(38 -94) (i.e. in the absence of modification by PAM, there is no likely posttranslational mechanism for removal of a carboxyl-terminal glycine, although it is remotely possible that an angiotensin-converting enzyme-like carboxy dipeptidase could cleave PTHrP (38 -96) to PTHrP(38 -94) with a free carboxyl terminus); 2) carboxyl-terminally amidated peptides are generally (but not always) more active than their nonamidated counterparts (36 -39); 3) PTHrP(38 -95), a likely substrate for PAM, is produced; and 4) PTHrP-expressing cells co-express PAM (39).
These studies should be interpreted with caution for several reasons. First, as noted above, the amidation state of PTHrP(38 -94) has not been unequivocally determined at present. Second, whether PTHrP(38 -95) is a processing intermediate or whether it is a mature secretory form is unclear. Third, it is uncertain whether processing events observed for human peptides expressed in rat cell lines can reliably be extrapolated to events ocurring in human cells. Available data would suggest that RIN cells reliably process human prohormones such as insulin, proopiomelanocortin, glucagon, and somatostatin (22). Further, the human and rat PTHrP sequences are so highly conserved in the PTHrP(1-111) region that aberrant processing seems unlikely (1)(2)(3)(4)(5)(6). In addition, we have previously reported that mid-region PTHrP peptides chromatographically indistinguishable from those produced by RIN cells are produced by human cell lines such as renal carcinoma cells and human keratinocytes (7). Fourth, Henderson et al. (4) have made the fascinating observation that the multibasic amino acids in the (87-106) region of PTHrP ( Fig. 1) are nucleolar targeting sequences. Thus, evidence that these same sequences are used in the cytosol for nuclear and nucleolar targeting as well as within the cisterns of the endoplasmic reticulum, the Golgi apparatus, and the secretory granule as prohormone processing sites will require explanation.
Preliminary Biologic Characterization of Synthetic PTHrP-(38 -94)amide-Studies with amino-terminal PTHrP have shown that it is able to stimulate adenylyl cyclase in bone, kidney, and vascular smooth muscle cells (5,6,8) and that it induces increments in cytosolic calcium in islet cells, squamous carcinoma cells, and lymphocytes (20,21,26,41). Examination of PTHrP(38 -94)amide in three different cell lines which are target cells for amino-terminal PTHrP indicated that even large doses of PTHrP(38 -94) amide (10 Ϫ6 M) failed to stimulate adenylyl cyclase, even though adenylyl cyclase was stimulated by appropriate control peptides. In striking contrast, PTHrP(38 -94)amide proved to be a potent activator of the intracellular calcium signaling pathway in each of the three cell lines examined. This is reminiscent of the calcium signaling induced by PTH  and PTHrP  in RIN cells and lymphocytes. In addition, as had been predicted from prior studies, PTHrP(38 -94)amide was active in vivo as an agonist of placental calcium transport from the maternal to the fetal circulation. These findings further document the biological activity of PTHrP(38 -94)amide and support the concept that FIG. 9. Transplacental calcium flux in the perfused ovine placenta. Placentae were perfused as described under "Materials and Methods" with either 10 g (approximately 6 nM in the 250 ml of perifusate, assuming that no degradation occurs) of PTHrP(38 -94)amide or with 10 g of the closely related control peptide, PTHrP(37-74). As can be seen, PTHrP(38 -94)amide reproducibly stimulated transplacental calcium transport, while PTHrP(37-74) had no effect. Additional negative controls have been reported previously and include 10 -24 g PTHrP(1-34) (14, 43) and 25 g of teleocalcin (14). In addition to the data shown, a single experiment was performed using 2.5 g of PTHrP(38 -94)amide, which resulted in a 30% increase in placental calcium transport. Finally, perfusion of a single placenta from a previously parathyroidectomized fetus with 10 g of PTHrP(38 -94)amide yielded a 160% increment in transplacental calcium flux. Thus, six of six PTHrP(38 -94)amide-perfused placentae demonstrated clear-cut calcium transport responses to PTHrP(38 -94)amide, whereas none of three placentae perfused under the identical conditions with the control peptide, PTHrP(37-74), demonstrated such a response. mid-region PTHrP derived from either the fetal parathyroid glands, the fetal placenta, or from both, plays a critical role in calcium delivery from the maternal circulation to the circulation of the developing fetus.
Collectively, these observations support the following conclusions: 1) the authentic secretory forms of PTHrP now include PTHrP(38 -94)amide, and possibly PTHrP (38 -95) and PTHrP- (38 -101), in addition to PTHrP , and longer amino-terminally intact PTHrP species as shown in Fig. 7; 2) demonstration of prohormone cleavage in the (96 -106) multibasic region supports the existence of PTHrP(107-139) as an authentic secretory form of PTHrP, as proposed by Fenton et al. (42); 3) cell lines representative of beta cells of the pancreatic islet, of vascular smooth muscle and of squamous epithelia respond to physiologically achievable concentrations of mid-region PTHrP; 4) PTHrP(38 -94)amide is active in vivo as well, as demonstrated by its ability to stimulate transplacental calcium transport; 5) the responses observed in vitro in cell lines and in vivo in the placenta may be mediated by the calcium-coupled receptor activated by PTHrP(67-86)amide identified in squamous carcinoma cells by Orloff et al. (15); 6) mid-region PTHrP appears to be capable of acting in both a paracrine/autocrine fashion as well as in a classical endocrine fashion; 7) in situations in which PTHrP functions in an autocrine/paracrine manner, given that the majority of cells process and secrete multiple forms of PTHrP as outlined in Fig. 7 and also have receptors for these multiple secretory forms of PTHrP, it will be critical to define the coordinate response to combinations of these secretory forms of PTHrP in order to fully understand the normal physiological functions of PTHrP; and, 8) the ultimate physiological consequence of PTHrP secretion by a given cell type will be a function not only of the amount of PTHrP precursor produced, but also the types of intracellular prohormone convertases present in that cell and the repertoire of PTHrP secretory forms produced. With the characterization of a second secretory form of PTHrP, further studies designed to examine the coordinate responses the multiple secretory forms of PTHrP can now be undertaken.