Calcitonin gene-related peptide (CGRP) and amylin are homologous 37-amino-acid peptides which have been demonstrated to have anabolic effects on bone. It is not clear whether these effects are mediated by a common receptor, nor is it known which ligand is the more potent. These questions are addressed in the present study using cultures of fetal rat osteoblasts. CGRP increased cell number when present in a concentration ≥10−9 M, but 10−8 M CGRP was required to stimulate thymidine and phenylalanine incorporation. Amylin was effective on these indices at 100-fold lower concentrations, and its maximal effects were about twice as great as those of CGRP. ED50's for the effects of amylin and CGRP on cell number were 10−12 M and 10−10 M, respectively. There was no additivity between maximal doses of the peptides on these indices. The effects of specific receptor blockers on the maximal stimulation of cell number by these peptides were also studied. The CGRP receptor-blocker, CGRP-(8–37), completely blocked the effect of CGRP at blocker concentrations ≥10−9 M. In contrast, the amylin receptor blocker, amylin-(8–37), completely blocked the effects of CGRP when the blocker was present in concentrations as low as 10−11 M. The KI of CGRP-(8–37) was 2 × 10−10 M and that of amylin-(8–37) was 7 × 10−12 M. In converse experiments studying the blockade of maximal doses of amylin, amylin-(8–37) 10−10 M was effective (KI 1 × 10−10 M), whereas a 100-fold greater concentration of CGRP-(8–37) was necessary to achieve the same effect (KI 6 × 10−9 M). It is concluded that amylin and CGRP probably act through a common receptor to stimulate osteoblast growth, and that this receptor has a higher affinity for amylin than for CGRP.
Calcitonin gene-related peptide (CGRP) and amylin are homologous 37-amino-acid peptides, the genes for which have a common ancestral origin.(1) CGRP-1 is generated by alternative processing of mRNA from the CALC1 gene, located on the short arm of chromosome 11. A second form of CGRP, CGRP-2, differs from CGRP-1 by only 3 amino acids in humans and 1 amino acid in rats. It is produced by a separate gene also on the short arm of chromosome 11, thought to have arisen as a result of exon duplication.(2) The peptides have in common a 6-amino-acid ring structure at the NH2 terminus created by a disulfide bond between positions 2 and 7. In addition, the COOH termini are amidated. Amylin is a further member of this peptide family and has 43% sequence identity with CGRP-1 and 49% with CGRP-2. Amylin was originally isolated from amyloid deposits in the pancreases of patients with insulinoma or diabetes mellitus.(3,4)
Sensory nerve fibers containing CGRP are widely distributed in bone, including bone marrow.(5–7) When defects are created surgically, the development of CGRP-containing nerves is noted several days later, often in association with new blood vessels,(8) suggesting a role in callus formation and bone healing. Similar responses are seen following fractures.(9) In contrast, amylin is produced principally in the β-cells of the islets and is cosecreted with insulin. It circulates in picomolar concentrations.(10)
There are believed to be separate specific receptors for amylin and CGRP, with some evidence pointing to more than one class of receptor for the latter.(11) The amino acid sequence of the amylin receptor remains unknown. Amylin, CGRP, and calcitonin are able to displace each other from specific binding sites, implying significant cross-reactivity of each with the receptors of the other peptides.(12) Whether this leads to significant biological effects at the peptide concentrations found in vivo is unknown.
Following the discovery of CGRP, its common origin and sequence homology with calcitonin led to an investigation of its effects on bone resorption, and similar studies followed with amylin.(13) Both peptides have calcitonin-like effects, probably mediated by the calcitonin receptor.(14) More recently, these peptides have been found to also have effects on osteoblasts, presumably via a different receptor since the calcitonin receptor is not found in osteoblasts.
A number of studies have found effects of CGRP on osteoblasts.(13) Michelangeli et al.(15) demonstrated that CGRP (but not calcitonin) increased cAMP formation in an osteogenic sarcoma cell line and in osteoblast-like cells from chicken, rat, or mouse calvariae.(16) CGRP regulates production of bone-active factors, increasing release of interleukin-6 from marrow stromal cells(17) and insulin-like growth factor-1 from osteoblasts(18) and decreasing osteoblast production of tumor necrosis factor-α.(19) In long-term osteoblast cultures, CGRP increases mineral formation.(20) Bernard and Shi(21,22) have demonstrated that the number and size of bone colonies developing in bone marrow cultures is increased by CGRP. Furthermore, when animals are pretreated with CGRP systemically, the number of bone colonies developing in cultures of their marrow is increased. Recently, it has been reported that transgenic mice with osteoblasts over-expressing the CGRP gene have a 5% increase in distal femoral bone density at the age of 12 weeks.(23) In contrast, local injection of CGRP over the calvariae of adult male mice did not produce any significant effects on bone histomorphometry.(24) Overall, however, these observations suggest that CGRP may have important actions on the osteoblast.
However, amylin also acts on the osteoblast, stimulating cell proliferation and protein synthesis, and causing increases in bone formation and bone mass in vivo after both local and systemic administration,(24,25) though in one study it did not reverse diabetic osteopenia.(26) Because of the homology of these peptides and their cross-reactivity with each others' receptors, it is possible that the effects of CGRP and amylin on osteoblasts are mediated by the same receptor. The present study addresses this issue.
MATERIALS AND METHODS
Osteoblast-like cell culture
Osteoblasts were isolated by collagenase digestion from 20-day fetal rat calvariae. Calvariae were dissected aseptically and the frontal and parietal bones were stripped of their periosteum. Only the central portions of the bones, free from suture tissue, were collected. The calvariae were treated twice with phosphate-buffered saline containing 4 mM EDTA (pH 7.4) for 15 minutes at 37°C in a shaking water bath. After washing once in phosphate-buffered saline, the calvariae were treated twice with 3 ml of 1 mg/ml collagenase for 7 minutes at 37°C. After discarding the supernatants from these two digestions, the calvariae were treated twice more with 3 ml of 2 mg/ml collagenase (30 minutes, 37°C). The supernatants of the latter two digestions were pooled, centrifuged, and the cells washed in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum (FCS), suspended in further DMEM/10% FCS, and placed in 75-cm2 flasks. After 48 h, the media were changed to minimal essential medium (MEM). Confluence was reached within 5–6 days, at which time the cells were subcultured. After trypsinization using trypsin-EDTA (0.05%/0.53 mM), the cells were rinsed in MEM with 5% FCS and resuspended in fresh medium, then seeded at 5 × 104 cells/ml in 24-well plates (0.5 ml cell suspension per well, i.e., 2.5 × 104 cells/well). The cells were incubated under 5% CO2 and 95% air at 37°C. Ascorbic acid in a concentration of 50 mg/l was added to the MEM used throughout. The osteoblast-like character of these cells has been established by demonstration of alkaline phosphatase staining in >95% (Fig. 1), osteocalcin production,(27) and a sensitive adenylyl cyclase response to parathyroid hormone and prostaglandin E2.(28)
Proliferation studies (cell counts and thymidine incorporation) were performed in subconfluent cell populations. Twenty-four hours after subculturing, cells were changed to serum-free medium with 0.1% bovine serum albumin for an additional 24 h prior to the addition of the experimental compounds. Cell numbers were analyzed 24 h after the addition of the peptide or vehicle by detaching cells from the wells by exposure to trypsin/EDTA (0.05%/0.53 mM) for ∼5 minutes at 37°C. Counting was performed in a hemocytometer chamber. Results are expressed per well. [3H]-thymidine incorporation into growth-arrested cells was assessed by pulsing the cells with [3H]-thymidine (0.5 μCi/well) 2 h before the end of the experimental incubation. [3H]-phenylalanine incorporation was assessed by pulsing the cells with [3H]-phenylalanine (1 μCi/well) for 4 h before the end of the experimental incubation. Experiments were terminated at 24 h by washing the cells in MEM followed by the addition of 10% trichloroacetic acid. The precipitate was washed twice with ethanol/ether (3:1) and the wells desiccated at room temperature. The residue was redissolved in 2 M KOH at 55°C for 30 minutess, neutralized with 1 M HCl, and an aliquot counted for radioactivity. Results were expressed as dpm per well. For cell counts, thymidine and phenylalanine incorporation, each experiment was performed at least four times using experimental groups consisting of at least 6 wells.
Similar methods of cell culture and for estimation of cell numbers were used for UMR 106–01 cells, an osteoblast-like cell line derived from an osteosarcoma.
Rat amylin, and human and rat CGRP were from Bachem California (Torrance, CA, U.S.A.). The rat amylin-(8–37) and human CGRP-(8–37) used in this study were COOH-terminal amides synthesized as described previously.(29) Because only rat amylin and rat amylin-(8–37) were used in these studies, their species of origin is not mentioned when they are referred to subsequently in this paper. EDTA and collagenase were obtained from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Trypsin-EDTA, MEM, DMEM, and FCS were from GIBCO Laboratories (Grand Island, NY, U.S.A.). [3H]-thymidine and [3H]-phenylalanine were from Amersham International (Little Chalfont, Buckinghamshire, U.K.).
Data are presented as mean ± SEM. The significance of differences between groups was determined using Student's t-test (two-tailed). The comparisons to be made in each experiment were specified a priori, so no adjustment for multiple comparisons was necessary. Where several experiments have been shown in one figure, data are expressed as the ratio of results in treatment groups to those in the control group and the p values shown were calculated using the data from the individual experiments, before the data were pooled.
Dose–response curves were analyzed using a simple empiric model chosen based upon the Hill formulation. Observed values for the effect (E) after dose (D) were calculated from the time integral of response from 0 using a trapeziodal rule. Predicted values of E(D) were calculated using the sigmoid Emax model(30)
where E(D) is the integrated response after a dose D, Eo is the integrated response in the absence of peptide, Emax is the asymptotic maximum integrated effect attributable to peptide, ED50 is the dose producing a response of Eo + 0.5 × Emax, and N is a steepness factor (Hill coefficient).
Parameter estimation was performed using a nonlinear least-squares regression procedures (NLIN, SAS/STAT Guide for personal computers; SAS Institute, Inc., Cary, NC, U.S.A.) with the multivariate secant iterative method DUD.(31) The goodness of fit for each model was assessed using the Schwartz information criterion.(32)
Figures 2 and 3 show dose–response curves for the effects of rat CGRP and rat amylin on cell number, thymidine incorporation, and phenylalanine incorporation. The minimally effective concentration of CGRP for stimulation of cell number was 10−9 M and 10−8 M for the other two indices. In contrast, amylin was effective at 100-fold lower concentrations and its maximal effects were about twice as great as those of CGRP (Emax for cell number, 0.21 vs. 0.09; for thymidine, 0.21 vs. 0.14). The respective ED50's for amylin and CGRP were 10−12 M and 10−10 M for cell number, and 5 × 10−11 M and 10−9 M for thymidine incorporation. The maximal effects of amylin on cell proliferation in this culture system are similar to those of other osteoblast mitogens (Fig. 4), suggesting that these effects might be of physiological significance. Very similar dose–response curves for amylin and CGRP are found with the clonal rat osteoblastic cell line UMR 106–01 (Fig. 5), indicating that these are osteoblastic growth effects and do not represent an effect on any contaminating cells that may be present.
The effects of blockers of CGRP and amylin receptors were then studied. Human CGRP was used in these studies because its maximal effect tends to be greater than that of the rat peptide in these osteoblast cultures, facilitating the quantitation of an inhibitory effect. The reason for this surprising finding is unknown. Figure 6 shows the effect of a maximal dose of CGRP on cell number and the blockade of this with the CGRP antagonist, CGRP-(8–37). The antagonist in a concentration of 10−10 M had no significant effect on the proliferative action of CGRP 10−8 M, but antagonist concentrations of 10−9 M and greater were adequate to block CGRP's effect completely. In contrast, the amylin receptor blocker, amylin-(8–37), blocked the effects of CGRP when the blocker was present in concentrations as low as 10−11 M. The KI of CGRP-(8–37) was 2 × 10−10 M and that of amylin-(8–37) was 7 × 10−12 M. The thymidine data from these experiments showed the same effects (data not shown). Treatment of cells with these (8–37) fragments alone (in concentrations up to 10−6 M) had no effect on either cell number or thymidine incorporation (data not shown).
For comparison, the effects of the two receptor blockers on the actions of amylin are shown in Fig. 7. The effect of amylin 10−9 M was completely blocked by amylin-(8–37) in concentrations of 10−10 M and greater, the KI being 1 × 10−10 M. However, a 100-fold greater concentration of CGRP-(8–37) was necessary to achieve the same effect, the KI being 6 × 10−9 M. The thymidine data from these experiments showed the same pattern of results (data not shown).
Finally, the additivity of the maximal effects of CGRP and amylin on osteoblast proliferation were assessed as a further way of assessing whether the two peptides had a common mechanism of action (Fig. 8). For both cell number and thymidine incorporation (latter data not shown), the combination of the agonists was no more effective than either one alone. In contrast, the maximal effects of transforming growth factor-β (TGF-β) showed clear additivity to those of amylin, implying that TGF-β's mechanism of action is substantially independent from that of amylin.
The present data confirm that both CGRP and amylin stimulate the proliferation of fetal rat osteoblasts in culture. The minimal effective concentration of CGRP is 100-fold higher than that of amylin, the maximal effect of amylin is greater, the amylin receptor blocker is 100-fold more potent in blocking the effects of both agonists, and there is no additivity of the effects of maximal doses of the peptides. These results are consistent with a model in which the actions of both CGRP and amylin are mediated by the same receptor. The relative potencies of both the agonists and antagonists assessed in the present studies suggest that this common receptor has a much higher affinity for amylin than for CGRP.
Definitive identification of the receptor involved is complicated by the fact that substantial uncertainty still surrounds the identities of the receptors for this family of peptides. The amylin receptor has not yet been isolated. A CGRP receptor was identified in 1995,(33) and a cell line expressing the calcitonin receptor-like receptor has been shown to display CGRP responsiveness.(34) Luebke et al.(35) have identified a small peptide that confers CGRP responsiveness on Xenopus oocytes, possibly by forming a part of a receptor complex. The recent work of McLatchie et al.(36) has carried the area forward substantially. They suggest that the CGRP receptor is indeed the so-called calcitonin receptor–like receptor, originally described by Njuki et al.(37) and Chang et al.(38) Cells expressing this receptor alone are relatively unresponsive to CGRP, but in the presence of a peptide they have termed receptor-activity-modifying protein 1 (RAMP1), the calcitonin receptor–like receptor is both glycosylated and translocated to the cell surface, conferring CGRP sensitivity. A separate RAMP interacts with the calcitonin receptor-like receptor to produce an adrenomedullin receptor.
The relevance to osteoblast biology of these findings from Xenopus oocytes transfected with RNA from a neuroblastoma cell line remains to be determined since neither of these receptors showed significant cross-reactivity with the other ligand nor with amylin, whereas the present findings suggest substantial cross-reactivity of amylin and CGRP with the same receptor, and we have previously demonstrated similar cross-reactivity for amylin and adrenomedullin.(39) Thus, the receptor complexes identified by McLatchie et al. have a much greater specificity than the receptors functioning in primary osteoblast cultures and further studies will be necessary to determine whether the calcitonin receptor-like receptor or any of the RAMP proteins are present in osteoblasts. We have recently demonstrated mRNA for the adrenomedullin receptor identified by Kapas et al.(40) in the osteoblasts used in the present studies (Naot D, Cooper GJS, Cornish J, unpublished observations). The data available at the present time would be consistent with the actions on osteoblasts of all three peptides being mediated by a single receptor which could be either an amylin or adrenomedullin receptor, since its sensitivities to both peptides are comparable.
One way to carry this matter forward would be to supplement the present assessments of cell proliferation with competitive binding studies of these peptides in osteoblasts. This is made difficult by the fact that currently available radiolabeled amylins are not biologically active in osteoblasts and do not show specific binding to bone cells (Naot D, Cooper GJS, Cornish J, unpublished observations). Efforts are under way at the present time to produce a radiolabeled amylin which is biologically active, and further studies of the receptor interactions of these peptides will need to await this development.
It is important to distinguish between the effects of these peptides on osteoblasts and osteoclasts. CGRP, amylin, and adrenomedullin all stimulate osteoblast proliferation, an effect inhibited by amylin receptor blockers(29) and associated with only minimal changes in cAMP production (Cornish et al., unpublished observations). In contrast, inhibition of osteoclast activity by CGRP and amylin probably involves cAMP as a second messenger(41) and is not affected by amylin receptor blockers.(29) It is a property shared by calcitonin which is active at much lower concentrations,(42) but not by adrenomedullin(39) nor by fragments of amylin which retain activity on osteoblast proliferation.(29) Furthermore, calcitonin does not affect cell proliferation in our osteoblast model.(24) Together, these observations suggest that a different receptor mediates the effects of these peptides on osteoblasts from that involved in their actions on osteoclasts. The data are consistent with the receptor in osteoclasts being a calcitonin receptor and that in osteoblasts being an amylin or adrenomedullin receptor.
In conclusion, the present data provide further evidence of an anabolic effect of both amylin and CGRP on osteoblasts and suggest that these effects are both mediated through the same receptor, which has a higher affinity for amylin than for CGRP. Work to definitively identify the receptors for this family of peptides is now necessary followed by a determination of which are present in osteoblasts. These peptides show potential as therapies for osteoporosis, since their systemic administration increases bone volume in adult mice(25) and their activity on osteoblasts can be isolated to fragments that are without known effects on other cells and tissues.(29,39) The identification of their molecular target in osteoblasts may aid the honing of their therapeutic potential.
This study was supported by the Health Research Council of New Zealand.