Human parathyroid hormone (hPTH)-(1–31)NH2 (Ostabolin), which only stimulates adenylyl cyclase (AC) instead of AC and phospholipase-C as do hPTH(1–84) and hPTH(1–34), strongly stimulates femoral cortical and trabecular bone growth in ovariectomized (OVX) rats. Two side-chain lactams have been introduced in the hydrophilic face of the receptor-binding region of the fragment's Ser17-Val31 amphiphilic α-helix in an attempt to develop improved analogs for the treatment of osteoporosis. Replacing the polar Lys27 with an apolar Leu on the hydrophobic face of this α-helix and stabilizing the helix with a lactam between Glu22 and Lys26 produced a fragment, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2, which had six times the AC-stimulating ability of hPTH(1–31)NH2 in ROS 17/2 rat osteosarcoma cells, but the other helix-stabilizing lactam derivative [Leu27]-cyclo(Lys26-Arg30)-hPTH(1–31)NH2 did not have a greater AC-stimulating ability than hPTH(1–31)NH2, to stimulate AC in ROS 17/2 rat osteosarcoma cells. As expected from AC stimulation being responsible for PTH's anabolic action, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 was, depending on the experimental design, a 1.4 to 2 times better stimulator of trabecular bone growth in the OVX rat model than either hPTH(1–31)NH2 or [Leu27]-cyclo(Lys26-Arg30)-hPTH(1–31)NH2. Thus, there is now a more potently anabolic derivative of hPTH(1–31)NH2, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2, which might ultimately prove to be one of the more effective therapeutics for osteoporosis.
It has been known since Selye's report in 19321 that intermittent injections of low doses of parathyroid hormone (PTH), particularly the human PTH (hPTH)-(1–34) fragment introduced by Tregear et al. in 1974,2 stimulate the formation of biomechanically normal or supranormal bone in various parts of the skeletons of several animals, particularly ovariectomized (OVX) rats, and osteoporotic humans.3–9 Since the PTH(1–84) holohormone and hPTH(1–34) stimulate both adenylyl cyclase (AC) and phospholipase-Cβ (PLC), their osteogenic actions could have been triggered by either one or both of these signaling enzymes. The reports of Rixon et al.10 and Strein11 that PTH fragments, such as hPTH(8–84), hPTH(28–48), and 1-desamino-hPTH(1–34) that only stimulated PLC in vitro and in vivo did not stimulate bone growth in OVX rats established AC as an essential part of the osteogenic trigger. But the question remained as to whether the osteogenic mechanism was triggered by AC alone or by AC plus PLC. To answer this question, we made hPTH(1–31)NH2 (Ostabolin) which could not stimulate PLC but stimulated AC in ROS 17/2 differentiation-competent rat osteosarcoma cells as strongly as the standard hPTH(1–34).12,13 This small, AC-only stimulator strongly stimulated cortical and trabecular bone growth in both 3-month-old and 1-year-old OVX rats.10,14 Thus, hPTH(1–31)NH2 became a new candidate for treating osteoporosis with the twin advantages of being the smallest of the osteogenic fragments and being unable to trigger the osteogenically unnecessary cascade of PLC-induced reactions. Among these PLC-triggered reactions are bursts of protein kinase C activity that might promote tumor development like the protein kinase C (PKC)-activating tumor-promoter TPA15 in the many PTH/parathyroid hormone related protein (PTHrP) target tissues such as skin16 during persistent long-term treatment.
The design of hPTH(1–31)NH2 was based on PTH's PLC-activation domain being in the 29–32 region of the molecule.12 Therefore, a fragment having only the first 31 of PTH's 84 amino acids could not activate PLC because it would lack the necessary His32.12 Such a fragment could still stimulate AC because it would have both the intact N terminus needed for AC stimulation and a key portion of the PTH/PTHrP receptor-binding region, the hydrophobic face of an amphiphilic α-helix between Ser17 and Val31.13,17–21 This fragment with its C-terminal amide had a greater AC-stimulating ability than hPTH(1–31)-OH in vitro, and the C-terminal amide would also partially protect the fragment from peptidase attacks in vivo.13,22
We have since been exploring ways to increase hPTH(1–31)NH2's anabolic activity and therewith its potential for treating osteoporosis. First, we have replaced the polar Lys27 with the apolar Leu27 in the fragment's α-helix to increase the hydrophobicity of the amphiphilic helix's receptor-binding hydrophobic face. Then we conformationally restricted the fragment with strategically placed side-chain lactam (cyclic amide) linkages. Small linear peptides such as hPTH(1–31)NH2 are highly flexible in solution, can assume a number of nonhelical, hence nonbioactive, configurations, and therefore have little helical structure that can be detected by circular dichroism (CD) or nuclear magnetic resonance.17,21,23 An appropriate cyclization can promote receptor-binding conformation.17,23–25 Cyclization by side-chain lactams could also increase the resistance to proteolytic digestion, as has been found with cyclic analogs of the thrombin inhibitor, hirudin.26 There are two possible lactam linkages in hPTH(1–31)NH2, between Glu22 and Lys26 or between Lys26 and Asp30, on the helix's hydrophilic surface which would not compromise the receptor-binding hydrophobic face and should promote the α-helical conformation postulated to bind to the receptor.17 According to CD spectra,17 either of these lactams would stabilize the α-helix. However, we will show that only one of them, Glu22-Lys26, produces a significantly more osteogenic peptide, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2.
MATERIALS AND METHODS
Synthesis of PTH fragments
hPTH(1–31)NH2 was synthesized using the same Fmoc protocol and continuous-flow peptide synthesizer (PerSeptive Biosystems Model 9050, Framingham, MA, U.S.A.) described earlier.10,13 Its molecular weight by electrospray mass spectrometry was 3717.77 ± 0.13 (calculated for M + 1, 3717.14).
[Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 and [Leu27]- cyclo(Lys26-Asp30)-hPTH(1–31)NH2 were synthesized as described by Barbier et al.17 The molecular masses were 3685.46 ± 0.46 (expected M + 1, 3685.12) and 3685.61 ± 0.36 (expected M + 1, 3685.12), respectively. The positions of the lactams were confirmed by sequencing.17
Bone growth assay
In this study, we used sexually mature OVX'ed 3-month-old rats. This model is widely used because it combines the easy manageability of rats with the main features of human postmenopausal osteoporosis: increased estrogen depletion-induced bone remodeling; rapid and selective trabecular bone loss; and skeletal responsiveness to bisphosphonates and, most important in the present context, PTH fragments.14 It is also one of the two animal models required by the United States Food and Drug Administration for the preclinical assessment of drugs for treating osteoporosis.27 Normal Sprague-Dawley rats were bought from Charles River Breeding Laboratories (St. Constant, QC, Canada). They were randomly distributed into six groups (normal, sham-OVX, vehicle-treated OVX, hPTH(1–31)NH2-treated OVX, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2-treated OVX, and [Leu27]-cyclo(Lys26-Asp30)-hPTH(1–31)NH2-treated OVX) of eight animals each. The rats were OVXed or sham-OVXed in this Institute as described by Rixon et al.10 They were given Purina rat chow (1.00% calcium, 0.61% phosphorus) and water ad libitum. It is important to note that there were no adverse reactions to the treatments or unscheduled deaths in this study. All experiments were approved by this Institute's Animal Care Committee.
The various fragments were dissolved in an acidic saline vehicle (0.15 M NaCl in water containing 0.001 N HCl) and injected subcutaneously at a dose of 0.6 nmol/100 g of body weight. This dose was chosen because it lies close to the middle of the range of effective doses of hPTH(1–31)NH2.14 We treated rats in two ways. In a prevention protocol, the rats were OVXed and a 6-week series of once-daily injections (6 days/week) of vehicle or fragment was started at the end of the second week after OVX before there had been a significant loss of trabecular bone. In a restoration protocol, the 6-week series of once-daily injections was started at the end of the ninth week after OVX when the trabecular bone in the distal femur had been severely depleted.
Although hPTH(1–31)NH2 can strongly stimulate cortical bone growth in the OVX rat,10 we focused on trabecular bone to compare the osteogenic potencies of hPTH(1–31)NH2 and its two lactam derivatives because its responses to OVX and a PTH fragment, such as hPTH(1–31)NH2, are proportionally much larger and therefore more easily and accurately measurable.10,14 Femurs were removed from euthanized rats, the distal halves stripped of their epiphyses, and the remaining bone fixed in acetate-buffered 10% formalin for at least 24 h. The distal femurs were demineralized in 4% trichloroacetic acid at room temperature for 6–8 weeks and then dehydrated, cleared, and embedded in paraffin. Serial, 10−μm sections were cut with a Leica RM-2035 microtome and stained for 30 s with Sanderson's rapid bone stain (Surgipath Medical Industries Inc., Winnipeg, MB, Canada).
The fragments' osteogenic potencies were indicated by how much they had increased the metaphyseal trabecular volume and the mean thickness (area [μm2] ÷ perimeter [μm]) of 80–100 individual trabeculae (and a total of 421–856 trabeculae for each experimental group) within that volume. To calculate trabecular volumes in a 100−μm-thick slice of distal femur, the sum of the longitudinal trabecular areas in the entire metaphysis in 10 serial 10−μm sections were added to obtain the trabecular volume in a 100−μm slice of each distal femur. The trabecular areas and perimeters were measured with a M4 imaging system from Imaging Research Inc. (St. Catherine's, ON, Canada). The computer software used was Imaging Research's morphometric version 1.2. All of the measurements were made by the same person to minimize variation from the unavoidably subjective component of such an analysis.
Adenylyl cyclase measurement
ROS 17/2 rat osteosarcoma cells were used to compare the adenylyl cyclase-stimulating ability of the various PTH fragments. Adenylyl cyclase activity was measured according to Jouishomme et al.12
All data are expressed as means ± SEMs. Statistical comparisons were made by one-way analysis of variance (ANOVA). When significant differences were observed, Scheffe's test was used for multiple comparisons and p < 0.05 was considered to be significant.
Since it is intermittent AC stimulation by hPTH, or a hPTH fragment such as hPTH(1–31)NH2, that triggers and drives the osteogenic mechanism,8–10,14 we first compared the abilities of hPTH(1–31)NH2, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2, and [Leu27]-cyclo(Lys26-Arg30)-hPTH(1–31)NH2 to stimulate ROS 17/2 cells' AC. AC-stimulating ability was indicated by the concentration of a fragment that induced 50% of the maximal AC activity, the EC50, and the stimulatory ability of a fragment relative to that of hPTH(1–31)NH2 was expressed as EC50 hPTH(1–31)NH2/EC50, frag. [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 was a 6-fold more effective AC stimulator than [Leu27]-cyclo(Lys26-Arg30)-hPTH(1–31)NH2 which, in turn, was twice as effective as hPTH(1–31)NH2 (Fig. 1). Since it is AC that triggers the osteogenic mechanism, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 should also have been the most effective stimulator of trabecular bone growth in OVX rats.
To test this prediction, we first used a prevention protocol in which rats were OVXed and 6 weeks of once-daily (6 days/week) subcutaneous injections of acidified saline vehicle or 0.6 nmol of fragment/100 g of body weight were started at the end of the second week after the operation, which was before a significant loss of trabecular bone would have occurred.10,14 The initial (normal) trabecular volume in the distal femurs was 0.70 ± 0.06 mm3. By the end of the eighth week after sham-OVX, the trabecular volume was about the same, 0.72 ± 0.04 mm3, but it had dropped by 51% to 0.34 ± 0.02 mm3 (p < 0.01) in the distal femurs of vehicle-treated OVX rats (Fig. 2). hPTH(1–31)NH2 and [Leu]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 prevented the loss and significantly (p < 0.01) increased the trabecular volume to supranormal levels of 1.2 ± 0.2 and 1.1 ± 0.1 mm3, respectively (Fig. 2). [Leu27]-cyclo(Lys26-Asp30)-hPTH(1–31)NH2 also prevented the trabecular volume from falling below the normal and sham values, but it did not raise the volume above the normal level (Fig. 2).
As expected from the results of previous studies, the mean trabecular thickness did not drop during the first 8 weeks after OVX (Fig. 3).14,28 Although [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 injections did not increase the trabecular volume more than hPTH(1–31)NH2, they caused a much greater increase in the mean trabecular thickness than injections of hPTH(1–31)NH2 or [Leu27]-cyclo(Lys26-Asp30)-hPTH(1–31)NH2 (Fig. 3). Thus, by the end of the eighth week after OVX and 36 injections of [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 between the ends of the second and eighth weeks, the mean trabecular thickness had increased 3.1 times (p < 0.01) from the normal 26.3 ± 0.4 μm to 82.3 ± 2.6 μm (Fig. 3). During the same time hPTH(1–31)NH2 injection increased the mean trabecular thickness 1.9 times to 49.4 ± 1.1 μm (p < 0.01), and [Leu27]-cyclo(Lys26-Asp30)-hPTH(1–31)NH2 injections increased the thickness 1.6 times to 42.2 ± 0.7 μm (Fig. 3). The superiority of [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 can be seen in the typical demineralized bone sections in Fig. 4.
In the next experiment, we used a restoration protocol in which the 6 weeks of vehicle or fragment injections did not start until the end of the ninth week after OVX, by which time the animals were expected to have lost a large part of the trabecular bone in their distal femurs.10,14 By the end of the 15th week after OVX and 36 vehicle injections from the end of the 9th to the end of the 15th week, the trabecular volume in the distal femurs had dropped by 75% from a normal 0.72 ± 0.05 mm3 to 0.18 ± 0.02 mm3 (p < 0.01) (Fig. 5). By the end of the 15th week after OVX and 36 injections of hPTH(1–31)NH2 or [Leu27]-cyclo(Lys26-Asp30)-hPTH(1–31)NH2 from the end of the 9th to the end of the 15th week, the trabecular volume had risen to 0.50 ± 0.01 and 0.49 ± 0.04 mm3, respectively (p < 0.05) (Fig. 5). But the [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 injections were significantly more effective (p < 0.01) and had returned the trabecular volume to a normal value of 0.74 ± 0.04 mm3 (Fig. 5).
[Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 also stimulated the largest increase in trabecular thickness in the 9-week depleted femoral trabecular bone. By the end of 15 weeks after OVX, the mean trabecular thickness had dropped 20% from the normal starting value of 43.4 ± 1.0 μm to 34.5 ± 0.6 μm, while in the sham-operated rats it was 51.2 ± 0.9 μm (Fig. 6). Treatment with hPTH(1–31)NH2 or [Leu27]-cyclo(Lys26-Asp30)-hPTH(1–31)NH2 between the end of the 9th and 15th weeks equally increased the mean trabecular thickness about 1.6 times to 68.5 ± 0.9 or 69.0 ± 1.4 μm, respectively (p < 0.05) (Fig. 6). However, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 was significantly (p < 0.01) more effective; it increased the mean trabecular thickness 2.3 times to 98.8 ± 1.8 μm (Fig. 6). The superior trabecula-thickening ability of this lactam fragment can be seen in the typical demineralized bone sections of Fig. 7.
hPTH(1–31)NH2, like other PTH molecules, is highly flexible and in solution only a fraction of hPTH(1–31)NH2 molecules are in the bioactive configuration with a large receptor-binding α-helix in the Ser17-Val31 region at any given time.13,17–21 As shown by the circular dichroism analyses of Barbier et al.,17 the fraction of hPTH(1–31)NH2 molecules with an α-helix can be equally increased by producing either a Glu22-Lys26 or a Lys26-Asp30 lactam linkage, which would limit the molecular flexibility and hold the 17–31 region in an α-helical configuration. However, according to Barbier et al.17 and the results of the present experiments, it is the Glu22-Lys26 lactam linkage that stabilizes the more perfect α-helix (probably because it is near the center of the region while the Lys26-Asp30 is at the periphery of the region) and produces the greater bioactivity as indicated by a 6-fold greater ability to stimulate adenylyl cyclase in cultured ROS 17/2 osteoblast-like cells and as much as a doubling of the ability to stimulate trabecular bone growth in OVXed rats. At first sight, this difference between the increases in adenylyl cyclase-stimulating ability and bone-anabolic action could be interpreted to mean that cyclic adenosine monophosphate (cAMP) may not be the trigger of trabecular bone growth. But this would be a mistake because unlike responses of cultured cells, there are limits to the growth of trabecular bone in the animal and a 6-fold increase in 6 weeks would be unattainable outside of a tumor.
In conclusion, we have made hPTH(1–31)NH2 into a potentially more potent therapeutic for osteoporosis specifically by stabilizing its α-helical configuration with a Glu22-Lys26 lactam linkage. The position of the lactam found to be useful in improving the antiosteoporotic activity of hPTH is strikingly similar to that used in creating a more potent vasoactive intestinal peptide (VIP) analog for the treatment of bronchial asthma.24 The VIP receptor is a member of the same family as PTH, and VIP has an amphiphilic helix at its C-terminal end. The problem with the AC-stimulating linear VIP was that it is a potent smooth muscle relaxant in guinea pig airway tissue but only a weak relaxant in human airway tissue. A key step in making VIP into a potent human bronchorelaxant was to stabilize an optimally bioactive configuration with a Lys21→Asp25 lactam.24 In an analogous manner, [Leu27]-cyclo(Glu22-Lys26)-hPTH(1–31)NH2 will likely prove to be an improved therapeutic for osteoporosis.
We acknowledge the excellent assistance of Sasha Bidwell in carrying out several of the key experiments in this study.