Structure-activity relation of NH2-terminal human parathyroid hormone fragments.

Human parathyroid hormone (hPTH) is involved in the regulation of the calcium level in blood. This hormone function is located in the NH2-terminal 34 amino acids of the 84-amino acid peptide hormone and is transduced via the adenylate cyclase and the phosphatidylinositol signaling pathways. It is well known that truncation of the two NH2-terminal amino acids of the hormone leads to complete loss of in vivo normocalcemic function. To correlate loss of calcium level regulatory activity after stepwise NH2-terminal truncation and solution structure, we studied the conformations of fragments hPTH-(2-37), hPTH-(3-37), and hPTH-(4-37) in comparison to hPTH-(1-37) in aqueous buffer solution under near physiological conditions by circular dichroism spectroscopy, two-dimensional nuclear magnetic resonance spectroscopy, and restrained molecular dynamics calculations. All peptides show helical structures and hydrophobic interactions between Leu-15 and Trp-23 that lead to a defined loop region from His-14 to Ser-17. A COOH-terminal helix from Met-18 to at least Leu-28 was found for all peptides. The helical structure in the NH2-terminal part of the peptides was lost in parallel with the NH2-terminal truncation and can be correlated with the loss of calcium regulatory activity.

Human parathyroid hormone (hPTH) is involved in the regulation of the calcium level in blood. This hormone function is located in the NH 2 -terminal 34 amino acids of the 84-amino acid peptide hormone and is transduced via the adenylate cyclase and the phosphatidylinositol signaling pathways. It is well known that truncation of the two NH 2 -terminal amino acids of the hormone leads to complete loss of in vivo normocalcemic function. To correlate loss of calcium level regulatory activity after stepwise NH 2

-terminal truncation and solution structure, we studied the conformations of fragments hPTH-(2-37), hPTH-(3-37), and hPTH-(4 -37) in comparison to hPTH-(1-37) in aqueous buffer solution under near physiological conditions by circular dichroism spectroscopy, two-dimensional nuclear magnetic resonance spectroscopy, and restrained molecular dynamics calculations. All peptides show helical structures and hydrophobic interactions between Leu-15 and Trp-23 that lead to a defined loop region from His-14 to
Ser-17. A COOH-terminal helix from Met-18 to at least Leu-28 was found for all peptides. The helical structure in the NH 2 -terminal part of the peptides was lost in parallel with the NH 2 -terminal truncation and can be correlated with the loss of calcium regulatory activity.
All known extracellular biological activity of human parathyroid hormone (hPTH) 1 is located in the NH 2 Ϫ terminus of this 84-amino acid peptide hormone (1). hPTH-  is the naturally occurring bioactive hormone extractable from human blood (2,3), and hPTH-(1-34) is known to maintain normocalcemia in blood via adenylate cyclase activation. To increase calcium flow into blood, the hormone acts directly on bone and kidney and indirectly on the intestine (1). In addition to the cyclic adenosine monophosphate (cAMP) pathway, involvement of the phosphatidylinositol hydrolysis signaling pathway is postulated for these functions (4). The receptor binding region mediating the calcium regulatory activity is located within sequence His-14 to Phe-34 (5,6). The complete NH 2 -terminal part of hPTH-  is required for stimulation of the cAMPdependent pathway (4), and the minimum sequence affecting bone and kidney comprises amino acids 2-27 (1,7). Adenylate cyclase activity is lost on deletion of the first NH 2 -terminal amino acid, whereas receptor binding capacity is not influenced, indicating that the activation region for cAMP production and the receptor binding region are located in two distinct domains (4,8). Adenylate cyclase activity measured in vitro does, however, not reflect the sequence-activity relationship indicated by various in vivo assays (4). hPTH-  is nearly inactive in an in vitro bioassay of cAMP stimulation, but in vivo the calcium level in blood is regulated with identical efficiency by hPTH-  and hPTH-(1-34) (Ref. 4 and references therein). This indicates that hPTH utilizes other second messengers in addition to cAMP for signal transduction and possibly additional receptors in vivo (9). Furthermore, hPTH is stimulating cell proliferation in skeletal derived cell cultures (10,11) as well as DNA synthesis in chondrocytes (12). Different sequence regions of the peptide are responsible for these functions; for stimulation of DNA synthesis, amino acids Asp-30 to Phe-34 are postulated as an indispensable region, but flanking residues seem to be required in addition for this function (12).
hPTH stimulates an increase of bone formation and axial bone mass after periodic administration of the hormone (13). Thus, hPTH is useful in the treatment of patients with hypoparathyroidism and, moreover, in the treatment of osteoporotic patients. Therefore, it would be highly desirable to construct a stable mimetic of this peptide hormone. Thus, recent studies focused on the determination of the three-dimensional structure of NH 2 -terminal peptides in solution by nuclear magnetic resonance (NMR) spectroscopy. In particular, hPTH-(1-34) is an intensely studied hormone fragment as it contains all functional domains (14 -17). From most experiments it was concluded that hPTH-(1-34) does not form secondary structure elements in the absence of TFE (14,16,18), but helix formation in TFE-free solution is nevertheless observed for hPTH-(1-34), residues 4 -13 and 21-29 (19), and for hPTH-(1-37), residues 5-10 and 17-28 (20). In TFE-containing solution hPTH-(1-34) displays helical regions from Ser-3 to Gly-12 and from Ser-17 to Lys-26 (16,18), but no tertiary interactions for hPTH-  are found under these conditions. It is commonly known that TFE stabilizes secondary structures, in particular helices (21)(22)(23)(24)(25)(26), but bears the risk of weakening hydrophobically stabilized tertiary structure domains (24), an effect also observed for hPTH-(1-34). 2 * 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.
Since hPTH is of considerable medical importance, drugs mimicking this structure could be useful as therapeutics. In a first step in this direction, we determine here the structures of the NH 2 -terminally truncated fragments hPTH-(2-37), hPTH- , and hPTH- (4 -37) in comparison with the biologically active fragment hPTH-(1-37) (20) under near physiological conditions to elucidate a possible correlation between the loss of calcium regulatory activity after stepwise truncation of NH 2terminal amino acids and structural features of the peptides.
Biological Activity-In vitro biological activity of the synthetic hPTH-(1-37), hPTH-(2-37), hPTH- , and hPTH-(4 -37) fragments was tested by observation of the stimulation of the cAMP generation in osteogenic cells (rat osteosarcoma cells) compared with synthetic hPTH-(1-34) fragment. ROS 17/2.8 cells were grown in 25-cm 2 plastic flasks at 37°C in a humidified atmosphere of air/CO 2 in Ham's F12/Dulbecco's modified Eagle's medium supplemented with 5% fetal calf serum, 50 mg of streptomycin/ml, and 50 units of penicillin/ml. The medium was changed on alternate days. The cells reached confluence within 3-4 days and were plated into 24-well dishes for experiments. Assays were performed on confluent cultures 1-2 days after change in medium. cAMP measurements were as follows. The cells were preincubated with 1 mM 3-isobutyl-1-methylxanthine for 15 min. The cells were then incubated for an additional 5 min in the presence of the agonists (hPTH-(1-34), hPTH-(1-37), hPTH-(2-37), hPTH- , and hPTH-(4 -37)). Incubation with forskolin was used as positive control. Supernatant was aspirated, and cAMP was extracted after addition of 70% chilled ethanol, evaporation, and redilution of the cells in cAMP buffer. The samples were kept at Ϫ20°C until cAMP levels were determined by a specific radioimmunoassay (29).
In vivo biological activity of these hPTH fragments was tested using Parsons' Chicken Assay (30) which is indicative of the Ca 2ϩ level homeostasis in blood. 6.25 g of hPTH fragment together with 20 mol of CaCl 2 was injected intravenously into 10 -14-day-old male chickens. After 60 min the chickens were anesthetized and then decapitated, and the blood was collected. The serum was diluted in a 1:50 ratio with 1% lanthanum nitrate solution. Atomic absorption spectroscopy was used for determination of serum calcium concentration. A hPTH-(1-34) sample served as a standard. Pure solvent without PTH was used as control.
CD Spectroscopy-CD spectra were recorded at 25°C in 0.1-mm cells from 250 to 190 nm at 20 nm/min on a Jasco J 600A CD spectropolarimeter. Peptide concentrations ranged from 270 to 310 M in 50 mM phosphate buffer, pH 6.0, with 270 mM sodium chloride in 30 l volume. The reference sample contained buffer without peptide. Eight scans were accumulated from samples and reference, respectively.
NMR Spectroscopy-Two-dimensional NMR spectra were obtained on a commercial Bruker AMX600 spectrometer at 298 K with standard methods (31,32). For hPTH-(4 -37) an additional set of spectra was measured at 288 K to resolve frequency degeneracy. The measurements were carried out in 50 mM phosphate buffer with 270 mM sodium chloride. Peptide concentrations were 1.6 mM, pH 6.0 (hPTH-(2-37)), 2.1 mM, pH 6.0 (hPTH-(3-37)), and 1.9 mM, pH 5.8 (hPTH-(4 -37)). The H 2 O resonance was presaturated by continuous coherent irradiation at the H 2 O resonance frequency prior to the reading pulse. The sweep widths in 1 and 2 were 7042.3 Hz. Quadrature detection was used in both dimensions with the time proportional phase incrementation technique in 1 . 4 K data points were collected in 2 and 512 data points in 1 . Zero filling to 1 K data points was used in 1 . All twodimensional NMR spectra were multiplied with a squared sine bell function phase shifted by /4, /3, and /2, respectively, for the NOESY spectra, by /6 or /4 for the Clean-TOCSY spectra, and /8 or /4 for the double quantum filtered COSY spectra. Base-line and phase correction of the 6th order was used. Data were evaluated on X-Window work stations with the NDee program package (Software Symbiose GmbH, Bayreuth, Germany).
Data from the following 600 MHz spectra were employed for the sequence-specific assignment of spin systems and the evaluation of the NOESY distance constraints for the different PTH fragments: double quantum filtered COSY spectra, Clean-TOCSY spectra with mixing times of 80 ms, and NOESY spectra with mixing times of 200 ms. For the structure calculations only NOEs visible in the NOESY spectra at 298 K were taken into account.
Restrained Molecular Dynamics Calculations-Distance geometry and molecular dynamics (MD) calculations were performed with the XPLOR 3.1 program package (33) on an HP735 computer. The number of nontrivial interresidual NOESY cross-peaks used for structure calculation was 171 for hPTH-(2-37), 210 for hPTH- , and 159 hPTH-(4 -37) (Table I). These cross-peaks were divided into three groups according to their following relative intensities: strong, 0.2 to 0.3 nm, medium, 0.2 to 0.4 nm, and weak, 0.2 to 0.5 nm. 0.05 nm was added to the upper distance limit for distances involving unresolved methyl or methylene proton resonances (pseudoatom approach).
The structure calculations followed standard procedures employing a hybrid distance geometry-restrained MD approach with simulated an-  nealing refinement and subsequent energy minimization (protocol distance geometry simulated annealing (33)). For the refinement the dielectric constant was changed to ⑀ ϭ 4. Structure parameters were extracted from the standard parallhdg.pro and topallhdg.pro files (34). For each fragment 30 structures were calculated. Ten structures for every fragment were selected on the criteria of smallest number of NOE violations over 0.05 nm and lowest overall energy. Structure Analysis-The final structures were analyzed with respect to stable idealized elements of regular secondary structure using the DSSP (definition of secondary structure of proteins) program package (35). To elucidate the stability of the structures, we calculated the local root mean square deviations with a five-amino acid window (36). For visualization of structure data the SYBYL 6.0 (TRIPOS Association), the RASMOL V 2.6 (37), and the MOLSCRIPT program packages (38) were used.
It is generally accepted that PTH initiates multiple intracellular signals, for example cAMP formation, phosphatidylinosi-tol hydrolysis, and release of intracellular calcium by activating G protein-linked receptors in bone and kidney (4). A single receptor was shown to stimulate intracellular accumulation of both cAMP and inositol triphosphates (39). PTH has the concentration-dependent ability to stimulate two separate signal pathways (9), and different sequential regions of the hormone may be responsible for initiation of the adenylate cyclase and the phospholipase C activating pathway. The existence of these multiple pathways is possibly reflected by the fact that hPTH-(2-37) is virtually inactive in the adenylate cyclase assay but can induce substantial hypercalcemia in the in vivo model (Fig.  1, a and b).
CD Spectroscopy-To compare the overall content of helical structure of the different peptides, far UV CD spectroscopy was used (Fig. 2) with peptide concentrations ranging from 270 to 310 M. The overall shape of the spectra of the different peptides indicates the presence of both ␣-helical and random coil structural elements (40,41). With the stepwise truncation of the NH 2 -terminal amino acids the ellipticity at 222 nm changes to less negative values. The evaluation of the helix content of the different peptides by standard methods (42) shows the following approximate fractional helix contents: hPTH-(1-37), 29%; hPTH-(2-37), 24%; hPTH-(3-37), 23%; and hPTH-(4 -37), 22%.
After truncation of the first two amino acids resulting in FIG. 1. Activity tests. a, in vivo activity test of hPTH-(1-37), hPTH-(2-37), hPTH- , and hPTH-(4 -37) relative to hPTH-(1-34) as a standard using Parsons' Chicken Assay (30). Pure solvent without PTH served as control. The variation of the calcium level in blood is used as indicator for PTH activity.  -37) is at 203 nm, indicating a structural transition between hPTH-(2-37) and hPTH- . These changes in the shape of the spectra may be interpreted as relative increase of random coil structure upon truncation of the first two amino acids (40,41,43).
Analysis of C-␣ Proton Chemical Shifts-To allow an initial mutual comparison of the truncated fragments and hPTH-(1-37), we used the chemical shift data available from our experiments to perform a secondary structure estimation based on the chemical shift index strategy (44,45). The procedure depends on a direct correlation between the chemical shifts of C-␣ proton resonances of consecutive amino acids and the local secondary structure: an upfield shift of the C-␣ proton resonances relative to the corresponding "random coil" values indicates local ␣-helical structure (negative value in Fig. 3), and a downfield shift of C-␣ proton resonances compared with the corresponding random coil values indicates a local ␤-sheet structure (positive value in Fig. 3). Only deviations from the random coil values by more than 0.1 ppm are taken into account for secondary structure estimation. For hPTH-(1-37) and hPTH-(2-37), the chemical shifts of C-␣ proton resonances suggest two helical regions extending from Ser-17 to at least Gln-29 and around Glu-4 to His-9. In contrast, no indication of an NH 2 -terminal helix is found for hPTH-(3-37) and hPTH-(4 -37), although the helical region in the COOH-terminal part can clearly be derived (Fig. 3). No other elements of regular secondary structure were evidenced by this procedure.
From the difference values of the observed C-␣ proton chemical shifts relative to the random coil values (45), one can estimate the stability of structural elements, as there is not only a correlation between the existence of secondary structure elements and chemical shifts of C-␣ proton resonances but also a correlation between the inherent main chain flexibility of these structure elements and the chemical shift data (46). The difference values in the NH 2 -terminal region of hPTH-(1-37) and hPTH-(2-37) are much less negative than in the COOHterminal part around Leu-24, indicating that the NH 2 -terminal helix is less stable than the COOH-terminal helix. Chemical shifts of the C-␣ proton resonances of residues 4 -37 of hPTH-  are nearly equal to that of hPTH-(1-37), indicating that the distribution of secondary structure elements is identical for these two fragments. The two NH 2 -terminal amino acids cannot be taken into account as the lack of flanking residues does not allow their C-␣ proton resonance chemical shift values to be compared with those of the amino acids with more than one flanking residue. The C-␣ proton resonances of Ile-5 to Met-8 of hPTH-(3-37) and Gln-6 to His-9 of hPTH-(4 -37) are shifted downfield by 0.05 to 0.15 ppm relative to the corresponding values of hPTH-(1-37), suggesting a loss of the NH 2 -terminal helix after deletion of the first two amino acids of hPTH-(1-37) (Fig. 3).
Analysis of Medium Range NOEs-The NOEs observed for the various hPTH fragments were determined from the 200-ms NOESY spectra at 298 K (Table I and Fig. 4). The d ␣N (i,i ϩ 3) and d ␣␤ (i,i ϩ 3) NOESY cross-peaks fully corroborate the existence of two helical regions for hPTH-(1-37) and hPTH- . Indications for an NH 2 -terminal helix for hPTH-(3-37), however, are weak and are entirely missing for hPTH-(4 -37), thus confirming the results from the chemical shift index procedure. In particular, helix typical (i,i ϩ 3) NOEs are clustered from Ile-5 to Leu-11 and Ser-17 to Phe-34 for hPTH-(1-37) and hPTH-(2-37), respectively. For hPTH-(1-37), two helical regions were found earlier, a short one from Ile-5 to Asn-10 and a longer one from Ser-17 through at least Leu-28 (20). For hPTH-(3-37) two weak helix typical NOEs are found in the NH 2 -terminal region, and for hPTH-(4 -37) no helix typical NOE could be found in the NH 2 -terminal region, and frequency degenerations of possible (i,i ϩ 3) NOEs were not present. In contrast, clear evidence of the COOH-terminal helix in these Structures of hPTH Fragments 4312 two fragments is found from Ser-17 to Phe-34 and His-32, respectively. To investigate whether the missing (i,i ϩ 3) NOEs in the NH 2 -terminal region of hPTH-(4 -37) can be accounted for by the lower concentration of this peptide (1.9 mM), twodimensional NMR spectra of a sample of hPTH-(1-37) with 1.8 mM concentration were measured with the same buffer, temperature, and spectrometer conditions. From the 200-ms NOESY spectrum of this sample, two d ␣N (i,i ϩ 3) and five d ␣␤ (i,i ϩ 3) NOEs in the NH 2 -terminal region of hPTH-(1-37) could be assigned. Thus, the lower concentration cannot account for the missing (i,i ϩ 3) NOEs in the NH 2 -terminal region of hPTH- (4 -37). The loss of the NH 2 -terminal helix on removal of the first two amino acids was fully confirmed by the NOESY cross-peak patterns (Fig. 4).
Relative NOE Intensities-Relative intensities of sequential and medium range NOEs may be used to estimate the perfection and stability of helical structures, in addition to the upfield shift of the ␣-proton resonances. For an ideal ␣-helix the d ␣N (i,i ϩ 1) and d ␣N (i,i ϩ 3) distances should be nearly identical, whereas the d NN (i,i ϩ 1) distances should be shorter, yielding higher intensity NOEs (47)(48)(49). For all PTH fragments employed in our experiments, most of the sequential d ␣N (i,i ϩ 1) NOEs are of higher intensity than the corresponding d NN (i,i ϩ 1) and d ␣N (i,i ϩ 3) NOEs (Fig. 4), indicating that the helices are not ideal but are in an equilibrium with a more extended conformation, possibly a 3 10 -helix. The helices of the PTH fragments are clearly more stable than nascent helices that do not show (i,i ϩ 3) NOEs (48,50). Simultaneous observation of d ␣N (i,i ϩ 2) and d ␣N (i,i ϩ 4) NOEs may arise from a mixture of 3 10 -and ␣-helix type structures (50). This effect was observed for the COOH-terminal region of the different PTH fragments. Karle and Balaram (51,52) suggest that six-residue sequences are equally likely to form 3 10 -or ␣-helices. Both helices in the hPTH fragments we studied seem to represent an equilibrium between an ␣-helix and 3 10 -helical conformation, the NH 2terminal helix having a higher tendency to a more extended 3 10 -helical conformation. This phenomenon is also reflected by values of the upfield shift of the ␣-proton resonances (Fig. 3).
Other NOEs-For each fragment, four to six long range (͉i Ϫ j͉ Ͼ 5) NOEs could be assigned (Table I). All fragments show several long range NOEs between Leu-15 and Trp-23. For hPTH-(1-37), five NOEs were found between Leu-15 and Trp-23 (20), four NOEs for hPTH-(2-37) and hPTH- , and five NOEs for hPTH- (4 -37). Additionally, two NOEs between Leu-15 and Val-21 were observed in hPTH- . These NOEs indicate a spatial proximity between Leu-15 and Trp-23, probably due to hydrophobic interactions between these two residues. The observed NOEs are responsible for a clear restriction of the conformational space of the calculated structures and lead to a defined loop region around His-14 to Ser-17. Furthermore, due to the ring current field of the spatial neighboring aromatic ring system of Trp-23 the ␦ proton resonances of Leu-15 are shifted upfield in comparison to the analogous resonances of other leucines for all four fragments.
Structure Calculation and Analysis-159 -210 interresidual NOEs per fragment were collected from 200-ms NOESY spectra at 298 K and used in restrained MD calculations (Table I).
For structure calculation of the NH 2 -terminally truncated fragments, the combined distance geometry/simulated annealing protocol described earlier (20,33) was used. For each fragment a family of 30 structures was calculated, and the 10 structures with lowest energy values and lowest number of NOE viola-tions were selected from each group. To resolve frequency degenerations of proton resonances in the spectra of hPTH-(4 -37) an additional set of spectra was obtained at 288 K. From this NOESY spectrum, 175 unambiguous interresidual NOEs could be assigned. Only NOEs were taken into account for structure calculation that were also observed, albeit ambiguously, in the NOESY spectrum at 298 K. For each of the four fragments the COOH-terminal helix extending from Met-18 to at least Leu-28 is found by DSSP analysis. For hPTH-(1-37) an NH 2 -terminal helix from Gln-6 to His-9 exists. For hPTH-(2-37), five structures show an NH 2 -terminal ␣-helix around Leu-7; the others show turns or 3 10 -helix in this region. None of the 10 calculated structures of hPTH-(3-37) displays an NH 2 -terminal ␣-helix, and only two structures exhibit a 3 10 -helix from Glu-4 to Gln-6. No structure of hPTH-(4 -37) shows an NH 2 -terminal helix, and only in one case a turn is indicated by DSSP in this region. The extension of the COOH-terminal helix of hPTH-(4 -37) is virtually identical to that of the corresponding helix in the other fragments. The loss of the NH 2 -terminal helix after truncation of the first two amino acids is corroborated by the structure calculations.
Local RMSD Values-To elucidate the stability of the structures in the helical regions and the defined loop, we calculated the local root mean square deviations (RMSD) using a fiveamino acid window (36) (Fig. 5). The upper trace represents the local RMSD values for all heavy atoms, and the lower trace represents the values for the peptide backbone. The regions with defined structure show substantially reduced local RMSD values compared with the flexible regions at the termini and around Gly-12. For hPTH-(1-37) and hPTH-(2-37) two regions with local backbone RMSD values lower than 0.07 nm were found from Gln-6 to His-9 and Asn-16 to Lys-26 for hPTH-(1-37) and from Leu-7 to His-9 and Asn-16 to Asp-30 for hPTH-(2-37), respectively. Comparatively high RMSD values for the amino acids Leu-11 to Lys-13 for fragments hPTH-(1-37) and hPTH-(2-37) indicate a flexible hinge region between the NH 2terminal helix and the loop region followed by the COOHterminal helix. For hPTH-(3-37) and hPTH-(4 -37) a decrease of the RMSD values is found in the region of the COOHterminal helix from Ser-17 to Gln-29. Compared with the fragments hPTH-(1-37) and hPTH-(2-37) the NH 2 -terminal region is structurally less well defined for the fragments hPTH-  and hPTH-(4 -37) (Fig. 5).
A best fit superposition of the peptide backbone atoms of the 10 final structures selected from the MD calculation of hPTH-  shows two well defined regions linked by a hinge region around Gly-12: a loop region from His-14 to Ser-17 followed by the COOH-terminal helix up to at least Leu-28 (Fig. 6, a and b) and a short NH 2 -terminal helix around Leu-7 (Fig. 6c). The loop region and the following COOH-terminal helix is very similar to the region from His-14 to Leu-28 of hPTH-(1-37) (20). The same is true for hPTH-  and hPTH-(4 -37) (Fig.  6d), indicating that the truncation of the first amino acids only influences the NH 2 -terminal structure, whereas the loop region and the COOH-terminal helix remain unimpaired. This is also confirmed by similar RMSD values for the region His-14 to Leu-28 of the four peptides (Table I). For most of the calculated structures the COOH-terminal helix ends at Leu-28, but the (i,i ϩ 3) NOE pattern may be interpreted to indicate that the helical region extends to Phe-34 or His-32, respectively, for all four PTH fragments. The upfield shifted C-␣ proton resonances also extend to Gln-29 or His-32.
Helix Content-From NMR and structure calculation data the following helix contents for the different peptides were estimated. The secondary structure analysis using the DSSP program (35) result in a helix content of 43% for hPTH-(1-37), 44% for hPTH-(2-37), 37% for hPTH- , and 32% for hPTH- (4 -37). For this calculation only amino acids that reside in a helical conformation in more than 50% of the calculated structures were taken into account. Under the assumption that residues which contribute to medium and strong (i,i ϩ 3) NOEs are part of helical structures (Fig. 4), the helix content is 59% for hPTH-(1-37), 55% for hPTH-(2-37), 43% for hPTH- , and 38% for hPTH- (4 -37). The helix content according to the chemical shift indexing procedure is 46 and 50% for hPTH-(1-37) and hPTH-(2-37), respectively, 31% for hPTH-(3-37), and 32% for hPTH-(4 -37) (Fig. 3). From the NMR results a clear decrease in the helix content is derived between hPTH-  and hPTH- . Assuming that the length of helical regions is reflected correctly by the combined NMR results, there is a significant underestimation of the helix content from CD spectra (22-29%), which is also reported for other peptides (54,55). One explanation for the apparent lower helix content estimated from CD spectra is that the helical sequences are in helical conformation in 50 -70% on time average in the case of PTH. Other explanations are the absolute length of the helices and the associated end group effects (56 -60) as well as a possible contribution of the aromatic side chain of Trp-23 to the far UV CD signal (59,61). Additionally, the shape and intensity of the CD signal depends on the geometry of a peptide helix. An ideal ␣-helix has a stronger CD signal than a 3 10 -helix (59) with a different shape (50,58). These phenomena lead to a lower percentage of helicity estimated from the [] 222 value. Thus, changes in the short NH 2 -terminal helix could not be detected on the basis of the CD signal at 222 nm alone. The possibility of an equilibrium with a 3 10 -helix is also reflected by the values of the upfield shifts of the C-␣ proton resonances (Fig. 3).
Progressive destabilization of a helix by successive removal of residues at the NH 2 -terminal end was observed earlier (62). Among others, backbone hydrogen bonding, loss of conformational entropy, interactions between side chains, electrostatic interactions between polar and charged groups at the termini with the helix macrodipole, and capping interactions at the helix termini influence helix stability (54,63,64). The unfavorable interaction of the positively charged NH 2 -terminal ]NH 3 ϩ group with the helix macrodipole is known as a destabilizing factor (63,65,66). This terminal charge gets closer to the region of the NH 2 -terminal helix by successive deletion of the NH 2terminal amino acids, which would provide an explanation for the destabilization of the NH 2 -terminal helix of hPTH.
At pH 6.0, as was used in our experiments, the side chains of Glu-4 and His-9 are charged. These charges stabilize the helix macrodipole (64 -66) and are of importance for stabilizing short helices in particular (66). The negative charge of the Glu-4 side chain that interacts favorably with the helix macrodipole may be screened by the closer proximity of the positively charged NH 3 ϩ group in the truncated fragments hPTH- , and even more so in hPTH- (4 -37). The truncation of Ser-3 leads to a complete loss of the NH 2 -terminal helix. Ser-3 may possibly serve as an N-cap (54,63,64) as its side chain may form a hydrogen bond to the main chain of the NH 2 -terminal helix, thus stabilizing the NH 2 -terminal helix. Gly-12 is at the very COOH-terminal end of the NH 2 -terminal helix. Indeed, Gly has a propensity to function as a helix C-cap (64). Stabilizing effects within the COOH-terminal helix were discussed earlier (20).
Most structural investigations of PTH fragments so far were carried out on hPTH-(1-34) using NMR spectroscopy in TFE containing solution to stabilize the helical regions (14 -16, 18, 67). Under these conditions no long range NOEs were found, and thus no tertiary interactions could be derived. Spatial proximity of Leu-15 and Trp-23, indicated by several long range NOEs between the side chain protons of these residues, was not observed in the TFE containing solution due to the lower polarity of TFE compared with H 2 O. hPTH-  in TFE free aqueous buffer solution at pH 4.1 (19) also exhibits NOEs between Leu-15 and Trp-23. Additional long range NOEs and a longer and more stable NH 2 -terminal ␣-helix extending from Glu-4 to Lys-13 for hPTH-(1-34) as proposed by Barden and Kemp (19) could not be confirmed in our present and earlier (20) work.
Conclusion-After deletion of the NH 2 -terminal two or three amino acids, PTH's biological activity is lost, but its receptor binding ability remains unimpaired (8,53). Table II  rizes the results of the activity tests and the structure calculations. All fragments show the loop region and the COOHterminal helix. His-14 to Leu-28 (loop and COOH-terminal helix) comprises the major part of the receptor binding region that is known to reside within His-14 to Phe-34 (5,6,68). The NH 2 -terminal helix, however, is present only in the in vivo bioactive fragments hPTH-(1-37) and hPTH-(2-37), but not in the inactive fragments hPTH-(3-37) and hPTH-(4 -37) (Table  II). This may indicate that the NH 2 -terminal helix is correlated with the in vivo bioactivity of the PTH fragments concerning the calcium level in blood. Existence of the NH 2 -terminal helix, however, cannot be connected to the ability to stimulate adenylate cyclase, as hPTH-(2-37) is nearly inactive in the cAMP assay. This result may imply different structural requirements for triggering the different signal transduction pathways (4), and may thus indicate the occurrence of different PTH receptors as discussed in the literature (69). To decide whether or not the in vivo biological activity is determined on a structural level by the NH 2 -terminal helix or depends on a direct functional role of the first two amino acids, structure calculations and activity tests of stabilized PTH fragments are currently under investigation.
FIG. 6. a, best fit superposition of the backbone atoms of His-14 to Leu-28 of the 10 final structures selected from the MD calculation of hPTH- . Only the backbone atoms of amino acids His-14 to Leu-28 are shown. b, same structures and superposition as in a, but all backbone atoms are shown to illustrate the inherent flexibility of the peptide. c, best fit superposition of the backbone atoms of Gln-6 to His-9 of the 10 final structures selected from the MD calculation of hPTH- . Only the backbone atoms of amino acids Val-2 to Lys-13 are shown. d, best fit superposition of the backbone atoms of His-14 to Leu-28 of the 10 calculated and selected structures of hPTH- (4 -37). Only the backbone atoms of amino acids His-14 to Leu-28 are shown.