Conformational studies on analogs of recombinant parathyroid hormone and their interactions with phospholipids.

Through the use of oligonucleotide-directed mutagenesis we have generated variants of a recombinant human parathyroid (PTH) hormone-(1-34)-homoserine (RPTH) in which a positively charged residue (Arg or Lys), a negatively charged residue (Glu), or a neutral residue (Gly) has been substituted at every position throughout the peptide. These 106 PTH analogs have been tested for their ability to stimulate cAMP production in the rat osteosarcoma cell line, UMR106. Analysis of these peptides led to the construction of several analogs containing multiple substitutions at sites of potential structural importance. Several of these analogs were shown to have 3-5-fold enhanced activity and receptor affinity. Circular dichroism (CD) and lipid binding studies were then performed on these analogs. Circular dichroism demonstrates enhanced helical content in the presence of lipid vesicles, particularly anionic lipid. The [Arg15,19,22,Lys29]RPTH (+6RPTH) analog requires higher concentrations of trifluoroethanol to attain enhanced helicity. The intrinsic tryptophan fluorescence of the peptides are blue shifted more in the presence of the anionic lipid dimyristoyl phosphatidylglycerol (DMPG) than with the zwitterionic lipid dimyristoyl phosphatidylcholine (DMPC). Effects of the peptides on the phase transition behavior of DMPC shows that +6RPTH has less effect on the lipid than does RPTH. This difference in lipid interaction is also exhibited with isothermal titration calorimetry, in which RPTH reacts exothermally with DMPG, while +6RPTH shows little or no heat change. The pH dependence of binding of the hydrophobic probe 1,1'-bis(4-anilino)-naphthalene-5,5'-trisulfonic acid, also shows a difference in exposure of hydrophobic sites between RPTH and +6RPTH. The +6RPTH has about a 5-fold greater affinity for receptor binding. We suggest that this enhanced activity is a consequence of the altered lipid interaction of +6RPTH, combined with increased conformational flexibility, particularly in the carboxyl-terminal region of the molecule.

Through the use of oligonucleotide-directed mutagenesis we have generated variants of a recombinant human parathyroid (PTH) hormone-(1-34)-homoserine (RPTH) in which a positively charged residue (Arg or Lys), a negatively charged residue (Glu), or a neutral residue (Gly) has been substituted at every position throughout the peptide. These 106 PTH analogs have been tested for their ability to stimulate cAMP production in the rat osteosarcoma cell line, UMR106. Analysis of these peptides led to the construction of several analogs containing multiple substitutions at sites of potential structural importance. Several of these analogs were shown to have 3-5-fold enhanced activity and receptor affinity. Circular dichroism (CD) and lipid binding studies were then performed on these analogs. Circular dichroism demonstrates enhanced helical content in the presence of lipid vesicles, particularly anionic lipid. The [Arg 15,19,22 ,Lys 29 ]RPTH (؉6RPTH) analog requires higher concentrations of trifluoroethanol to attain enhanced helicity. The intrinsic tryptophan fluorescence of the peptides are blue shifted more in the presence of the anionic lipid dimyristoyl phosphatidylglycerol (DMPG) than with the zwitterionic lipid dimyristoyl phosphatidylcholine (DMPC). Effects of the peptides on the phase transition behavior of DMPC shows that ؉6RPTH has less effect on the lipid than does RPTH. This difference in lipid interaction is also exhibited with isothermal titration calorimetry, in which RPTH reacts exothermally with DMPG, while ؉6RPTH shows little or no heat change. The pH dependence of binding of the hydrophobic probe 1,1-bis(4-anilino)naphthalene-5,5-trisulfonic acid, also shows a difference in exposure of hydrophobic sites between RPTH and ؉6RPTH. The ؉6RPTH has about a 5-fold greater affinity for receptor binding. We suggest that this enhanced activity is a consequence of the altered lipid interaction of ؉6RPTH, combined with increased conformational flexibility, particularly in the carboxyl-terminal region of the molecule.
Peptide hormones interact with specific receptor proteins on cell surfaces. This specific interaction may be modulated by the target cell surface which can cause the accumulation of the hormone near the site of the receptor as well as inducing a particular folding and orientation of the peptide (1). An interesting example to study in this regard is parathyroid hormone. This hormone plays an important biological role in calcium metabolism (2). Intact parathyroid hormone is an 84-amino acid peptide, but a 34-amino acid fragment, comprising the amino-terminal segment of the hormone, maintains virtually full activity (2). While the biological activity of this peptide will be determined by its interaction with its receptor, certain conformational and membrane-binding properties of the peptide will likely be required for maximal biological activity. This 34-amino acid peptide has regularly spaced hydrophobic amino acid residues which could form either a 3 10 helix or an ␣-helix with a hydrophobic domain that twists around the helix axis. This peptide can interact with phospholipids to attain a more structured conformation (3,4). There also have been several NMR studies on the conformation of the 34-amino acid fragment of the human parathyroid hormone (hPTH- ). 1 These studies demonstrated an increased ␣-helical content in trifluoroethanol with the formation of two helical segments, one in the region of residues 3-13 and the other around residues 16 -27 (5-7). More recently, a detailed NMR and molecular dynamics study of hPTH-(1-37) has appeared (8). This work concluded that this peptide in buffer contained an ␣-helical region between residues 5 and 10, a flexible link at residues 12 and 13 followed by a well defined turn comprising residues 14 to 17. There is another ␣-helical segment between residues 17 and 28.
Results from a series of hPTH-(1-34) analogs indicate that a number of amino acid residues can be substituted without loss of biological activity and in fact, in several cases there is enhanced activity. Some of the modified forms of the peptide hormone with enhanced activity have non-conserved substitutions. We wished to study how these alterations in sequence affected the interaction of these hormone analogs with both phospholipids and the PTH receptor and to determine if there is a relationship between changes in biological activity and differences in structural or lipid binding properties of these peptides.

MATERIALS AND METHODS
Reagents-Restriction enzymes, mammalian cell culture media, and Escherichia coli cell line DH10B (F Ϫ mcrA ⌬(mrr Ϫ hsdRMS Ϫ mcrBC) ⌽80dlacZ⌬M15 ⌬lacX74 deoR recA1 araD139 ⌬(ara,leu)7697 galU galK Ϫ rpsL endA1 nupG) were purchased from Life Technologies, Inc./BRL * This research was supported by the Natural Science and Engineering Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
PTH Expression Plasmid-The expression vector consisted of a modified form of the ara B-containing plasmid, pBAD18 (9), with the following sequence inserted downstream from the ara B promoter. The sequence contains a Shine-Dalgarno ribosome binding site (S&D) and TrpLE leader peptide. This 17-amino acid leader sequence has previously been shown to enhance expression of small proteins and may promote the sequestering of fusions into inclusion bodies (10,11). The polyhistidine site allows for rapid purification of recombinant parathyroid hormone (RPTH), and RPTH analogs via nickel chelation chromatography.
A RPTH gene, with the following coding sequence, was designed using high-use E. coli codons and constructed from partially overlapping oligonucleotides.

SEQUENCE II
This RPTH gene was cloned into the SalI and BglII sites of the modified pBAD vector. RPTH analog genes were constructed in a manner similar to that described for the RPTH gene. Each amino acid codon was individually replaced with ((A/G)(A/G)G), which codes for lysine, arginine, glutamate, or glycine. Transformants expressing the desired PTH variant genes were selected by plasmid DNA sequence analysis.
PTH Analog Expression and Purification-PTH analogs were expressed and purified as described previously (12) with the following modifications. Analogs were expressed in groups with concurrent expression of RPTH as a control. Following elution from His-bind resin, analogs were precipitated with 4 volumes of absolute ethanol. This precipitate was dissolved in 1 ml of 70% formic acid and a 200-fold molar excess of CNBr was added. The reaction was allowed to proceed for 2 h at room temperature in the dark under argon after which time the sample was loaded onto a Sep-Pak column, washed with 5 volumes of 20% acetonitrile, 0.1% trifluoroacetic acid, and then eluted with 3 ml of 60% acetonitrile, 0.1% trifluoroacetic acid. The RPTH analog was then lyophilized and stored at Ϫ20°C.
cAMP Assay-UMR106 cells were seeded at 2.5 ϫ 10 5 /well in 48-well microtiter plates and allowed to grow to confluence. Media (Dulbecco's modified Eagle's medium/F12, L-Glu, penicillin-streptomycin, 5% fetal bovine serum) was replaced daily. Assays were performed on cells 3-5 days post-confluence. Media was removed and replaced with 0.5 ml of fresh media containing PTH (or an analog) at the indicated concentration and 1 mM 3-isobutyl-1-methylxanthine. After 5 min of incubation, cells were washed once with ice-cold phosphate-buffered saline and the cAMP was extracted twice with 1 ml of absolute ethanol. The two extractions were combined and ethanol removed by evaporation under vacuum. The dry residue was resuspended in 1 ml of scintillation proximity assay ("SPA") buffer (Amersham), and cAMP concentration determined by using a commercially available SPA kit (Amersham).
Preparation of Sonicated Vesicles-Lipids in powder form were suspended in PIPES buffer and vortexed vigorously; then placed in a bath-type sonicator (Cole Parmer Ultrasonic Model 08849 -00) and sonicated to clarity.
Circular Dichroism-The circular dichroism (CD) spectra of the peptides were recorded on an AVIV model 61 DS solid-state CD instrument (AVIV Associates, Lakewood, NJ). The instrument was interfaced with a computer, which was used for all mathematical calculations. A 1-mm sample cell, maintained at 25°C with a thermostated cell holder, was used for all spectral studies. Peptide concentrations in solution, prior to the addition of lipid, were determined from the ultraviolet absorption at 280 nm and by the Pierce BCA Protein assay. The CD spectra were corrected for baseline and normalized per mol of amino acid residue. These spectra were then used for estimation of secondary structure by the procedure of Chang et al. (15) and reported as percent of ␣-helix (%␣), ␤-structure (%␤), turns (%T), and random coil or other structures (%R) present, respectively.
Intrinsic Fluorescence Spectra-Tryptophan fluorescence emission spectra were recorded on an SLM-Aminco Series 2 luminescence spectrometer, using 1-cm path length quartz cuvettes and 4-nm slits.
Differential Scanning Calorimetry (DSC)-Lipid films were made from DMPC or DMPG dissolved in chloroform/methanol (2/1, v/v) to which varying quantities of peptide dissolved in methanol were added. After solvent evaporation with nitrogen, final traces of solvent were removed in a vacuum chamber for 90 min. The lipid films were suspended in buffer by vortexing at 45°C for 30 s. The final lipid concentration was 1.5 mM at a lipid to peptide ratio of about 15, in buffer at pH 7.40. Two different buffers were used, one at physiological salt concentration was composed of 20 mM PIPES, 0.15 M NaCl, 1 mM EDTA, 0.002% NaN 3 and the other at low ionic strength was 10 mM sodium phosphate. The lipid suspensions were put through three freeze-thaw cycles and degassed under vacuum before being loaded into an MC-2 high sensitivity scanning calorimeter (Microcal Co., Amherst, MA). A heating scan rate of 39 K/h was employed. The bilayer to hexagonal phase transition was fitted to a single van't Hoff component and the transition temperature reported as that for the fitted curve.
Titration Calorimetry-Isothermal heats of reaction were measured using the Omega cell of a Microcal titration calorimeter (16). Solutions were degassed under vacuum prior to use. A 0.1 mM peptide solution was placed in the 1.3-ml reaction cell and thermally equilibrated before each run. DMPG or DMPC as small (or sonicated) unilammelar vesicles were made in PIPES buffer or in 10 mM sodium phosphate, pH 7.4. A solution of 100 mM lipid was used to fill a 100-l motor-driven syringe and 2.5-, 5-, or 10-l aliquots were delivered at 5-min intervals. The peptide solution in the reaction vessel was stirred at 400 rpm and maintained at 35°C. The observed enthalpies were corrected for the small heat of dilution of lipid into buffer. The calorimeter was calibrated electrically. The data were analyzed using the OMEGA v1.1 EMF Scientific Software (1987) from Microcal or with ORIGIN TM Scientific Software for Microcal Omega data. Data were fitted to a one independent binding site model.
Bis-ANS Titration-The fluorescence study of the pH dependence of bis-ANS binding was performed by a procedure similar to that described by Butko et al. (17). Fluorescence was excited at 395 nm and emission was observed at 533 nm, with a 420-nm emission cut-off filter. Measurements were made in 2 ml of PIPES buffer in a cuvette containing either 1.5 or 3 M peptide and 15 M bis-ANS. The pH was changed by adding HCl or NaOH and the fluorescence intensity at 533 nm recorded at each pH.

RESULTS
Native PTH-(1-34) contains methionine residues at positions 8 and 18. Oxidation of methionine at position 8 results in substantial changes in secondary structure as determined by circular dichroism whereas oxidation of Met 18 has little effect on secondary structure (18). The mono-or dioxidized forms of the peptide display significantly reduced affinity and biological activity (19). Previous reports (20) indicated that substitution of norleucine for methionine decreases the in vitro activity of the peptide to approximately 43% of the unsubstituted peptide. However, a peptide containing the phenylalanine to tyrosine substitution at position 34 in combination with Nle at positions 8 and 18 has an in vitro activity only slightly less (76%) than the unmodified peptide. Given the necessity of avoiding internal methionine residues to prevent fragmentation of the peptide following cyanogen bromide cleavage of the fusion protein, the methionine residues were changed to leucine and Phe 34 has been changed to Tyr in the recombinant form of the peptide. Cleavage of the fusion peptide with cyanogen bromide results in the addition of a homoserine/homoserine lactone (Hse) residue at the carboxyl-terminal of the peptide (21). These substitutions, Leu 8,18 ,Tyr 34 ,Hse35, when combined into hPTH to yield the recombinant peptide (RPTH), have little effect on the adenylate cyclase stimulation or receptor binding of the peptide when compared to hPTH (Fig. 1).
PTH analogs were expressed as TrpLe fusion peptides in E. coli with expression levels ranging from 5 to 180 mg/liter depending upon the analog and appeared to be independent of any single amino acid substitution. After isolation of inclusion bodies and purification by nickel affinity chromatography, the fusion peptide is greater than 95% pure as determined by SDS-PAGE ( Fig. 2A). After cyanogen bromide cleavage, the peptide is desalted, partially purified on a Sep-Pak column, and assayed for the ability to stimulate adenylate cyclase as de-scribed. A typical HPLC chromatogram (OD 280 ) of a Sep-Pakpurified sample, corresponding to lane 6 "CNBr cleaved" in Fig.  2A, is shown in Fig. 2B. Individual peaks were isolated and analyzed by electrospray mass spectrometry to determine the composition. The two minor peaks are the result of incomplete cyanogen bromide cleavage and represent the PTH-His 6 peptide (peak 1) and the I-N-M-PTH peptide (peak 3). The major peak (peak 2) comprises greater than 80% of the total protein and is RPTH. Comparison of the "crude" Sep-Pak-purified sample with an HPLC purified RPTH showed no significant difference in adenylate cyclase stimulatory ability (data not shown). Dual point cAMP assays were performed using the crude Sep-Pak purified samples. Dose-response and receptor binding curves were performed on RPTH analogs which were purified to Ͼ95% purity by reverse phase HPLC on a Vydac C18 column. Purity and composition were determined by electrospray mass spectroscopy.
The ability to stimulate adenylate cyclase activity was tested in the rat osteosarcoma cell line, UMR106. Initially, 106 RPTH analogs (a positive, negative, and neutral charge at each position) were functionally tested at a concentration of 5 nM, slightly higher than the EC 50 of hPTH (EC 50 ϳ 2.8 nM). Those analogs with activities greater than 50% of wild type were then retested at a peptide concentration of 1 nM. Analyzing peptides at two concentrations ensured that, for active peptides, one of the concentrations would lie on the linear portion of the doseresponse curve and thereby yield an accurate measure of the relative activity.
Substitution of any of the four amino acids (Lys, Arg, Glu, and Gly) in positions 1-8 in all cases decreased the activity of the peptide to less than 20% of wild type activity and in many cases to below detectable levels (Fig. 3A). Positions 9 and 10 were somewhat more tolerant of substitution. Leu 11 may play a role in ligand receptor interaction since the Arg 11 analog, which has a side chain backbone similar to leucine but which may also form ionic interactions retains complete activity but Glu 11 with its negative charge is inactive.
Chou-Fasman (22) calculations suggest a disordered region at positions 10 -13 of hPTH-(1-34) while Garnier-Robson calculations predict a turn. Barden and Kemp (23) have shown by NMR the presence of a type-I ␤-turn at positions 10 -13 and 16 -19 in PTH-related peptide (PTHRP). NMR studies by Klaus et al. (24) using slightly different solution conditions, however, suggest that this region in hPTH is disordered and does not form a ␤-turn. A more recent study (8) suggests a flexible link at positions Gly 12 and Lys 13 and a well defined turn from His 14 to Ser 17 . Substitutions at Gly 12 (Fig. 3B) either eliminate or significantly reduce (Lys Ͻ 10%) activity. This loss in activity could be due to the disruption in peptide structure. Chorev et al. (25) have shown, however, that substitutions of Ala, D-Ala, or ␣-aminoisobutyric acid at position 12 are well tolerated, suggesting that some flexibility at this position is tolerated but substitution with Pro decreases receptor binding by 2 orders of magnitude. Substitution at Lys 13 is well tolerated. The Arg 13 and Glu 13 analogs are fully active and the Gly 13 analog retains greater than 40% activity. Since substitutions at this position have little effect on cAMP stimulation, it is unlikely that this residue interacts with the receptor. Deletion of this amino acid, however, results in an analog with about 5% of the activity of that found in the complete peptide (26) and NMR analysis of this analog supports the hypothesis that it is important in the folding of the peptide. With the exception of Leu 15 , the remainder of the loop region, positions 14 through 17 (Fig. 3B), is tolerant of substitution with all analogs showing some activity. At position Leu 15 , the positively charged amino acids, Lys and Arg, retain greater than 50% activity while the negative and neutral substitutions, Glu and Gly, are completely inactive.
Positions 18 -34 (Fig. 3, B and C), have been reported to have an amphipathic helical structure in the presence of TFE or lipid vesicles (27). Mutations which alter the hydrophobic face of the amphipathic helix, positions 21, 24, 28, and 31, dramatically reduce the activity of the peptide. Leu 18 is predicted to initiate the helix, but substitution of either Glu or Arg at this position has only a moderate effect on the activity suggesting that this position is not critical to the overall structure. The hydrophilic surface of the helix has a highly charged character with 5 positively charged and three negatively charged residues, as well as a critical Trp at position 23. Conversion of the negatively charged residues, Glu 19 , Glu 22 , or Asp 30 , to positively charged residues either maintains or slightly enhances the activity of the peptide. Amino acid substitution at Trp 23 dramatically decreases bioactivity of the peptide suggesting that this residue makes a critical hydrophobic contact. There is also good evidence to suggest (26) that positions 19 and 21 may play a significant role in intramolecular stabilization between the amino and carboxyl-terminals of the peptide.
Based on NMR analysis, Klaus et al. (24) have suggested that Arg 20 is involved in a salt bridge with Glu 4 . Substitution of any of the amino acids (Lys, Arg, Glu, Gly) at either of these positions significantly reduces the cAMP stimulatory ability of the peptide. If the side chains at these two positions are primarily involved in salt bridge formation, then one might expect that reversing them would have little effect on the structure and activity of the peptide. The peptide [Arg 4 ,Glu 20 ]RPTH at 500 nM failed to stimulate a cAMP response suggesting that either salt bridge formation alone is not sufficient or that one or both of these amino acids also plays a direct role in ligandreceptor interaction.
Based on studies with model peptides (28), a salt bridge has also been postulated to occur between Glu 22 and Arg 25 or Lys 26 , which could help stabilize the amphipathic helix. Since the charge reversal analog E22R shows full activity, this potential salt bridge is unlikely to be important for activity. Charge reversal at positions Arg 25 , Lys 26 , Lys 27 , or His 32 , decrease the peptide activity to less than 10% of wild type. In NMR studies of PTHRP, Barden and Kemp (23) have suggested the possibility that His 32 forms a salt bridge with Glu 4 thereby helping to stabilize the peptide structure. Since these two residues are highly conserved in both PTH and PTHRP through a variety of species, they may be necessary for proper folding of the peptide. Our data are not completely consistent with this supposition since Lys 32 or Glu 32 analogs are completely inactive while Arg 32 and Gly 32 retain activity. Since the Gly 32 analog retains FIG. 2. A, total cellular protein at time 0 (T ϭ 0) and 3 h after induction with arabinose (T ϭ 3). Inclusion bodies were isolated and Trp-LEPTH purified by nickel chelation chromatography (His-bind). Purified TrpLEPTH was subjected to CNBr cleavage. Bovine PTH (BPTH) as a control. B, CNBr cleavage resulted in three peptide species. These peptides were separated by reverse phase chromatography and analyzed by electrospray mass spectroscopy. Peak 1 is PTH-His 6 , peak 2 is RPTH, and peak 3 is I-N-M-PTH.
nearly complete activity, it is unlikely that His 32 is involved in a salt bridge with Glu 4 , We cannot rule out the possibility, however, that additional structural changes in the Gly 32 analog may allow other biologically active structures. Position 25 is particularly sensitive to substitution in that even a conservative substitution, R25K, decreases the activity by over 80%. Arg 25 may take part in a charge-charge interaction with the receptor whereas Lys 25 may be sterically hindered from performing the same function. Positions Asn 33 and Tyr 34 do not appear to be important to structure/function since substitutions at these two positions have very little effect on activity. It has been shown previously (29), however, that serial deletion of these amino acids has a detrimental effect on bioactivity. Taken together, the data suggests a possible role in metabolic stabilization for these residues. Overall, addition of a positive charge to the hydrophilic surface of the amphipathic helix tends to maintain or slightly enhance activity suggesting that this surface may be important in binding either to a negatively charged surface on the receptor or by interacting with negatively charged lipid head groups.
EC 50 values are highly dependent upon the passage number of the UMR106 cell line. For example, EC 50 values for RPTH ranged from 1.9 nM in early passage cells to 4.4 nM for cells of later passage. Similar results were observed with both synthetic bPTH and hPTH. In order to properly assess the activities of individual RPTH analogs on cells of different passage number, it was therefore necessary to use both bPTH and RPTH as internal controls. As such, the data for each analog is presented as an approximate value and must be compared to the activities of both bPTH and RPTH which had been prepared and assayed at the same time. Within a given experiment, hPTH, bPTH, and RPTH maximally stimulated cAMP production to the same extent. The maximal amount of cAMP produced in these experiments ranged from 11 to 30 pmol depending upon the passage number of the cell line. K d values appeared to be independent of passage number.
Full dose-response and receptor binding curves were prepared for a number of RPTH analogs and for all of those RPTH analogs which contained several substitutions (Fig. 1, Table I).
In order to further investigate the effect of charge reversal substitutions on the hydrophilic surface of the amphipathic helix, 13 combination analogs were expressed, HPLC purified, and tested in the adenylate cyclase assay. Positions for modification were chosen based on the bioactivity data of the single point mutations. Five positions were chosen for further analysis with 4 sites located in the proposed amphipathic helical region (18 -32) and 1 site located in the proposed turn (position 15). In all cases these individual substitution analogs have EC 50 values comparable to (2.6 nM) or slightly lower (1.2 nM) than hPTH (2.8 nM) or RPTH (1.9 nM) (Fig. 1). These same analogs were tested for receptor binding and were shown to have dissociation constants slightly lower than either hPTH or RPTH.
Combining these individual analogs into a combination mutant, ϩ6RPTH, results in a more active peptide as judged by its lowered K d and increased adenylate cyclase activity. Fig. 1 shows the dose-response curves and receptor binding curves, for the ϩ6RPTH, RPTH, bPTH, and hPTH. The ϩ6RPTH analog has significantly enhanced bioactivity with an EC 50 ϳ 0.9 nM and a K d ϳ 1.5 nM. These values are approximately 3-and 5-fold lower, respectively, than the corresponding values for either hPTH or RPTH and are comparable to the values for bPTH.
In separate experiments, all possible combinations of the mutations comprising the ϩ6RPTH analog were constructed along with several which contained substitutions at Asp 30 . Expression levels of these analogs ranged from 1 to 10 mg/liter as compared to ϳ100 mg/liter for either RPTH or ϩ6RPTH. After HPLC purification, dose-response curves were generated as in Fig. 1 for each analog and EC 50 values determined by linear regression best fit from the curves ( Table I) 22,30 ]RPTH had EC 50 values comparable to that of the ϩ6RPTH analog. The Arg 22 substitution was contained within each of these peptides yet, when assayed individually, it displayed an EC 50 value significantly lower than that observed for RPTH. While the EC 50 values of these three analogs appeared to be enhanced, the receptor affinity was not significantly changed. Other analogs of this series which contain the Arg 22 substitution are only slightly less active than the ϩ6RPTH, with EC 50 values of approximately 1.8 nM. This suggests that within this series of substitutions, the Glu 22 to Arg substitution is necessary but not sufficient for enhanced activity. The E22R substitution does not appear to have as dramatic of an effect when combined in other series. It has been previously reported (30) that an D30R substitution significantly enhanced the adenylate cyclase stimulatory ability of the peptide. However, when both Glu 22 and Asp 30 are converted to arginine, the EC 50 values are comparable to or slightly lower (2-fold) than for RPTH. In this case the E22R substitution had very little effect on the bioactivity of the peptide.
Other substitutions had a much more modest effect on the adenylate cyclase activity. For example, unless combined with the Arg 22 substitution, the Arg 15 , Lys 29 combination appears to only slightly enhance the adenylate cyclase activity of the peptide. All combinations yielded peptides with adenylate cyclase activation activities intermediate between RPTH and the ϩ6RPTH analog. Interestingly, only the ϩ6RPTH peptide had both an EC 50 value and K d value comparable to that observed with bPTH. The cause of this observation is currently unknown and warrants further investigation.
Qualitatively the RPTH and RPTH analogs behave similarly. They exhibit more structure in the presence of lipid than in its absence and DMPG promotes more structure than DMPC. The secondary structure content of the various peptides in the presence and absence of lipid has been calculated from the CD spectra (Table II) as described under "Materials and Methods." The ␤-structure content is least accurately assessed by this criterion. Furthermore, ␤-structure aggregates may be formed by peptide-peptide interactions in the presence of lipid, rather than by lipid effects on the conformation of the peptide. We will therefore limit our discussion to changes in ␣-helical content under different conditions. Some differences among the various cases are that the unmodified RPTH has greater helical structure in the presence of DMPC, compared with the other peptides. In addition, the Arg 15 and the Lys 29 analogs are the only ones which exhibit a marked dependence of structure on peptide concentration, both in buffer and in the presence of DMPG. In other cases there is essentially no dependence on peptide concentration and data at only one peptide concentration is shown. TFE titrations were carried out in mixtures of TFE with 10 mM phosphate buffer, pH 7.4. RPTH shows an increased ␣-helical structure at a lower TFE concentration than ϩ6RPTH, but both peptides achieve a full helical conformation by 50% TFE (Fig. 4), at 25°C. At this concentration of TFE, when heating up to 50°C, the RPTH retains, almost entirely, its helical structure. In contrast, the ϩ6RPTH analog loses part of its ␣-helix to only 60% helicity.
The peptides have a Trp emission at or close to 350 nm in buffer, indicative of this residue being exposed to solvent in aqueous solution. Addition of small (or sonicated) unilammelar vesicles of DMPC did not greatly affect this emission maximum (not shown). Thus, although CD and DSC results indicate that some of these peptides interact with DMPC, this interaction does not lead to a sequestering of the Trp residue into a less polar environment. Addition of DMPG, however, caused a marked blue shift in the fluorescence emission (Table III), indicating that the tryptophan side chain (residue 23) becomes partially buried in a hydrophobic environment upon binding to DMPG. This observation is of particular interest because of the important role this residue has for receptor binding. There was no time dependence of the shifts observed. It has recently been suggested that Trp 23 interacts hydrophobically with Leu 15 (8).
DSC of mixtures of DMPC and the peptides was studied in two buffers, one of low ionic strength and the other of higher, physiological ionic strength. High ionic strength would suppress the ionic interactions between the cationic peptides and anionic charges on the lipid. In the presence of the low ionic strength buffer, 10 mM sodium phosphate, pH 7.4, the RPTH and the analogs greatly broaden the phase transition of the phospholipid, except for the ϩ6RPTH and the Arg 15 mutant which have little effect on the phase transition properties of this lipid (Fig. 5). When the DSC was done in PIPES buffer, pH 7.4, at the higher salt concentration of 0.15 M NaCl, none of the peptides had any effect on the phase transition of DMPC. In comparison (Fig. 5), all the peptides studied broaden the phase transition of DMPG in PIPES buffer.
Titrating a solution of peptide with sonicated vesicles of DMPG gave markedly different results for the RPTH and the ϩ6RPTH analog (Fig. 6). The RPTH gave an exothermic reac-   peptide binding to a cluster of 10 lipid molecules, in phosphate buffer. In contrast, little or no enthalpy was observed under these conditions for the reaction of DMPG with the ϩ6RPTH mutant. The binding of either RPTH or the ϩ6RPTH mutant to DMPC by titration calorimetry showed only small heats of reaction, which could not be readily fitted into a binding isotherm. No corrections for the effective charge of the peptides at the lipid interphase were made. The fluorescent probe bis-ANS has been used to identify hydrophobic binding sites on proteins since it exhibits an increased quantum yield in a less polar environment (31). Bis-ANS is negatively charged and therefore tends to bind better to proteins at low pH (17) where the protein is more cationic. This is observed for RPTH and two of the mutants but not with the ϩ6RPTH or the Arg 19,22,30 analogs which exhibit an anomalous pH dependence of the fluorescence of bis-ANS (Fig. 7). DISCUSSION A number of mutants of RPTH (net charge ϩ1), with additional positive charges placed at key positions of potential structural importance, were studied. The amino acid substitutions in the analogs studied in this work did not greatly affect the conformational properties of the peptides in buffer, as exhibited by CD, and the presence of anionic lipid enhanced the helical content of all of the peptides in a similar manner (Table  II). In addition, all of the peptides inserted into phospholipid bilayers to a similar extent as measured with shifts in the fluorescence emission spectra of Trp (Table III). However, there are marked differences in the nature of the interaction of the ϩ6RPTH analog (net charge ϩ7) with lipid, compared with the other peptides.
It is interesting to compare the equilibrium constants for dissociation of the hormone from its receptor, for the various derivatives (Table I). The ϩ6RPTH mutant stands out as having the highest affinity for the receptor (lowest K D ). A priori one would expect an analog with greatly altered charge and sequence to be less active than the native hormone. We would like to speculate that the enhanced receptor binding affinity observed for the ϩ6RPTH derivative is related to its altered conformational property. This may allow it to insert into membranes in a different manner, facilitated by its structure. This altered structure is not primarily a consequence of a difference in secondary structure.
Some of the altered features observed with ϩ6RPTH as compared to RPTH are that the ϩ6RPTH has: (a) no effect on the phase transition of DMPC; (b) no heat of reaction with DMPG as detected by titration calorimetry; (c) an anomalous pH titration with bis-ANS fluorescence; (d) increased hydrophilicity in the region between residues 11-18 as seen in hydropathy plots (data not shown); (e) decreased structure in DMPC, observed with CD; and (f) decreased structural stability in TFE, observed with CD.
In the Arg 15 mutant (net charge ϩ2), some of these altered features were also present. This analog has a small effect on the phase transition of DMPC and its hydrophilicity between residues 11 and 18 is increased, but its activity is similar to that of the native RPTH. This indicates that a single charge replacement in the turn region, which would confer altered conformational properties in the surrounding areas by increasing its hydrophilicity, is by itself not sufficient for enhanced activity. If the charge replacement is made in the second hinge region only, around residues 29 -30 (7), we obtain the mutant Lys 29 (net charge ϩ2) that behaves structurally like RPTH and does not show enhanced receptor affinity.
In the mutant Arg 19,22,30 (net charge ϩ7), three negatively charged residues were replaced by positively charged ones. These substitutions would greatly affect the stability of the carboxyl-terminal helix, retaining the integrity of the turn.
However, this mutant behaved conformationally like RPTH including anomalous pH titration with bis-ANS and did not show any enhanced receptor binding affinity, indicating that the changes in charge groups by themselves are not sufficient for altered activity and that the conformation of the molecule is not greatly altered. Concomitant changes seem to be necessary in more than one region of the molecule for increased activity. This result is consistent with reports by Avnur et al. (32) that complete substitution of positions 24 -34 in PTHRP with a model amphipathic helix yields a peptide which has only a slightly enhanced bioactivity. Biotinylation of Lys 26 or Lys 27 (33) had no effect on receptor binding affinity and single site mutations of K26Q (34), H32R, or D30Y (35) had no effect on either receptor binding affinity or adenylate cyclase activity. Taken together, this data suggests that the hydrophilic surface of the amphipathic helix in native PTH may play a limited role in ligand-receptor interactions, but that this interaction can be enhanced by the further addition of positively charged residues to the surface. This additional positive charge may allow the peptide to bind more tightly to a negatively charged surface on the receptor or with negatively charged lipid head groups in the cellular membrane. The data also suggests a possible long range effect on the amino-terminal domain of the peptide.
One of the predominant features of the ϩ6RPTH mutant is its decreased affinity for binding to lipid. This is likely to be a consequence of its conformational properties since lipid affinity is not correlated simply with the overall charge on the peptide. The critical factor determining the overall conformation for this series of analogs is likely to be the conformation near the middle of the molecule which permits the hydrophobic areas in the helices at the amino end and near the carboxyl terminus to become sequestered, preventing their interaction with lipids. The Arg 15 mutant by itself does not show enhanced activity indicating that the carboxyl-terminal substitutions present in ϩ6RPTH, may lead to enhanced activity through alternative conformational arrangements. The conformational alteration seen in ϩ6RPTH is also indicated by the temperature sensitivity of the CD of this peptide in 50% TFE. The Arg 19,22,30 mutant by itself, with 3 Arg substitutions in the carboxyl-terminal region, has somewhat lower affinity for lipid than RPTH as exhibited by the broadening seen on the phase transition of DMPC; but this broadening is much greater than that exhibited by ϩ6RPTH or Arg 15 , which hardly affect the DMPC main transition. Thus, the extent of interaction of Arg 19,22,30 is intermediate between the RPTH and the ϩ6RPTH mutant. This is likely to be a consequence of differences in the bend region.
Generally the ability to bind lipid has been associated with increased hormone binding potency. In the case of RPTH, the decrease in lipid binding affinity may limit nonspecific binding to the membrane and therefore be a factor favoring partitioning into the receptor.
Functions other than receptor binding and adenylate cyclase activation may also be mediated by PTH. Jouishomme et al. (36) and Somjen et al. (37) have suggested that the carboxylterminal of the peptide may play a role in the stimulation of intracellular Ca 2ϩ /protein kinase C pathway leading to both increased DNA synthesis and creatine kinase activity. The analogs described here have not yet been tested for their ability to stimulate the protein kinase C pathway.