Using human peripheral blood mononuclear cells as osteoclast precursors, we showed that dexamethasone stimulated osteoclast generation at a pharmacological concentration but did not affect the life span of human osteoclasts. Dexamethasone also dose-dependently increased signals for osteoclastogenesis.
Introduction: Glucocorticoid-induced osteoporosis is a common and serious disease. Glucocorticoids predominantly affect osteoblast proliferation and life span. Much of the bone loss is caused by reduced bone formation, but there is also an element of increased bone resorption.
Materials and Methods: Human peripheral blood mononuclear cells were cultured on whale dentine and induced to differentiate to osteoclasts by RANKL and human macrophage-colony stimulating factor (M-CSF). Osteoclast activity was quantified by pit area. RANKL and osteoprotegerin (OPG) expression in osteoblasts were measured by real-time RT-PCR.
Results: In the early phase of osteoclast generation (0-16 days), cultures from two different donors showed that dexamethasone at 10−8 M increased pit area by 2.5-fold, whereas lower concentrations had no effect. At the highest dexamethasone concentration (10−7 M), pit area was reduced. In 21-day cultures from three other donors, a similar increase was seen with dexamethasone at 10−8 M. There was, however, no evidence of increased life span of osteoclasts with dexamethasone. In human primary osteoblasts, dexamethasone dose-dependently reduced OPG and increased RANKL expression as measured by quantitative real time RT-PCR.
Conclusion: These data provide some explanation at a cellular and molecular level for the observed increase in bone resorption seen in patients treated with glucocorticoids and indicate that there are clear direct effects of glucocorticoids on bone resorption in human cell systems that may differ from other species.
GLUCOCORTICOID-INDUCED BONE LOSS is the most common form of secondary osteoporosis. It is usually the result of long-term treatment with glucocorticoids and results in loss of both cortical and cancellous bone,(1–4) thus increasing the risk of fracture.(5) Loss of bone is most rapid during the first year of glucocorticoid therapy, and significant reductions in bone mass can be seen as soon as 3 months after commencement of therapy.(6) Glucocorticoids predominantly affect osteoblast proliferation, differentiation, and life span.(7–9) Osteoclasts may also contribute to the loss of bone seen in patients with glucocorticoid-induced osteoporosis. An increase in osteoclast numbers and sites of lacunar resorption in the bones of these patients has been noted.(9,10) This increase in osteoclasts is thought to be caused by increased formation or increased survival of osteoclasts.(11)
Osteoblasts express two molecules that are essential for human osteoclastogenesis, macrophage-colony stimulating factor (M-CSF) and RANKL. M-CSF is important for macrophage maturation—it binds to its receptor, c-Fms, on early osteoclast precursors, thereby providing a signal for their survival and proliferation.(12) M-CSF is released into the extracellular bone environment by osteoblasts.(13) RANKL is expressed in abundance by activated T-lymphocytes as well as in osteoblasts, and it triggers osteoclastogenesis.(14) Osteoprotegerin (OPG), also produced by osteoblasts, acts as a decoy protein, binds to RANKL, and prevents osteoclastogenic activity. Osteoclastogenesis can now be achieved in vitro with pure cultures of osteoclast precursors exposed only to M-CSF and RANKL.
Reports of the effects of glucocorticoids on the generation of osteoclasts in various species have been inconsistent, with both enhancement and inhibition of osteoclast activity reported, and do not permit overall conclusions about the direct effects of glucocorticoids on osteoclast activity or life span.(15–18) This may be because of species differences. We sought to analyze the effects of glucocorticoids on signals for osteoclast generation as well as human osteoclast production and life span in the presence of fixed concentrations of added RANKL and M-CSF.
MATERIALS AND METHODS
Tissue culture media were purchased from Gibco BRL (Grand Island, NY, USA). FBS was from Commonwealth Serum Laboratories (Parkville, Australia). RANKL was kindly provided by Amgen (Thousand Oaks, CA, USA). Recombinant human M-CSF was from R&D systems (Minneapolis, MN, USA). Trypsin was from JRH Bioscience (Kansas, MI, USA). Dexamethasone was from Amersham Biosciences (Buckinghamshire, UK). All other chemicals were purchased from Sigma (St Louis, MO, USA) unless otherwise specified. Multiwell plates were from Becton Dickinson Labware (Rutherford, NJ, USA). Glass coverslips were from Menzel-Glaser (Braunschweig, Germany). Whale dentine was purchased from a private source.
Buffy coat blood (Australian Red Cross, Sydney, Australia) was diluted 1:1 in PBS, layered over 10 ml Ficoll-Hypaque (Amersham Biosciences), and centrifuged at 900g for 30 minutes. A layer of monocytes was extracted from the interphase of the PBS and Ficoll-Hypaque and centrifuged at 400g for 5 minutes. The cell pellet was rinsed and washed in α-MEM medium with 2.2 g/liter NaHCO3, 0.03 g/liter penicillin, 0.01 g/liter streptomycin, and 10% (v/v) heat-inactivated FBS (10% α-MEM). Cells were counted on a hemocytometer to determine the number of mononuclear cells in the suspension.
Peripheral blood mononuclear cells (PBMCs) were seeded in 96-well plates containing either dentine slices (4 × 4 mm) or glass coverslips (5 mm diameter) at a concentration of 1 × 106 cells/well in α-MEM containing 10% FBS. After incubation at 37°C for 2 h, the cells were rinsed twice with α-MEM containing 10% FBS to remove any nonattached cells. Cells were cultured in 150 μl α-MEM with 10% FBS, containing RANKL (50 ng/ml), human M-CSF (25 ng/ml), and dexamethasone or vehicle alone. These concentrations were chosen as optimal on the basis of previous experiments with osteoclast generation using PBMCs. Each treatment was carried out in triplicate, and the medium was changed twice a week. Dexamethasone was serially diluted (10−7-10−11 M) and dissolved in spectroscopic grade ethanol so that the final concentration of ethanol did not exceed 0.1% v/v.
Histochemical characterization of cells
Histochemical staining for TRACP was carried out to identify the multinucleated TRACP+ cells. Cells were fixed with 1% (v/v) formalin/PBS for 10 minutes and stained for acid phosphatase using 10 mg/ml naphthol AS-BI phosphatase as substrate in the presence of acetate tartrate buffer at pH 5.0 (50 mM sodium acetate, 40 mM potassium-sodium tartrate), and the product was coupled with Fast garnet GBC salt.(19)
Preparation of whale dentine for culturing cells
Whale dentine was sliced to 400 μm thickness with a diamond saw. Dentine slices were treated with acetone for 15–20 minutes and 70% ethanol. Slices were cleaned by ultrasonication in water for 60 s. For sterilization, slices were passed through 70% ethanol and then filtered milli-Q water for 10 minutes and irradiated overnight.
Preparation of culture on dentine slices for examination by scanning electron microscope
Cells were removed from dentine slices by incubating in 0.1 M EDTA for 10 minutes at room temperature, sonicated with 0.25 M ammonium hydroxide for 30 s, passed through graded alcohol to absolute, and allowed to air dry. Dentine slices were sputter coated with gold and examined in a Philips 505 scanning electron microscope (FEI, Eindhoven).
Using Photoshop 6.0, a grid was placed over each dentine slice micrograph. The total number of points on the grid were counted as well as the number of points that lay on top of the pits. Percentage resorption was calculated using the number of pits, which fell over points on the grid as percentage of the total points.(20) For each dentine slice, four random areas were chosen to estimate the percentage area resorbed.
Osteoblast culture and RNA extraction
Human primary osteoblasts were obtained from the trabecular ends of fetal long bones and cultured in DMEM with 3.5 g/liter NaHCO3, 44 mg/liter phosphoascorbate, 0.292 g/liter l-glutamine, and 10% (v/v) heat-inactivated FBS as previously described by Slater et al.(21) The cells were seeded at a density of 2 × 105 cells/well in 6-well plates. Each treatment was carried out in three replicates. The cells were grown for 48 h before changing to serum-free DMEM (44 mg/liter phosphoascorbate, 1000 mg/liter bovine serum albumin, 10 mg/liter transferrin, and 10 nM NaSeO3) for 24 h to allow cells to adapt. Cells were treated with vehicle (ethanol) or dexamethasone in serum-free DMEM for 24 h. Cultured cells were harvested by trypsinization using 0.25% (w/v) trypsin. RNA was extracted using a commercially available kit (RNeasy Mini Kit; Qiagen, Hilden, Germany) and was carried according to the manufacturer's instructions. Purity and integrity of the RNA extracts were monitored both spectrophotometrically, assuming a 260:280 nm ratio of ∼2 for pure RNA, and on a 1% agarose gel containing ethidium bromide.
cDNA synthesis with spiked RNA
To normalize for differences in the RT reaction efficiencies between samples, RNA samples were spiked with exogenous RNA as previously described,(22) in this case with in vitro transcribed bacterial RNA as follows. Plasmid pSB955,(23) a gift from Dr Nick Dixon (Australian National University, Canberra, Australia), containing a portion of DnaB helicase gene was linearized with EcoRV restriction enzyme and transcribed with T7 RNA polymerase (Promega) to produce a run-off transcript of 581 nucleotides. RNA concentration was determined by optical density at 260 nm. One microgram of total RNA from osteoblasts, as measured by a specially modified spectrophotometer ultraspec 3100pro (Amersham Biosciences) as previously described,(24,25) and 106 molecules of bacterial RNA were reverse transcribed in a 20-μl reaction of Superscript III (Invitrogen, Carlsbad, CA, USA) using random hexamer primers according to the manufacturer's instructions.
Primers were chosen with assistance of the program Primer (Robert D Andersen and Greg Bristol, Laboratory of Biomedical Environmental Science, Los Angeles, CA, USA) for melting temperature and 2° structure, and Amply (Bill Engels, Department of Genetics, University of Wisconsin, Madison, WI, USA) was used to check for correct specificity for the PCR reaction. Primers are listed in Table 1. RANKL and OPG PCR product sequences were verified using an automatic sequencing analyzer at the Sydney University Prince Alfred Macromolecular Analysis Center (SUPAMAC).
Table Table 1. List of Primers Used for Quantitative Real Time RT-PCR Reactions
Each gene was amplified with BIOTAQ DNA polymerase (Astral Scientific, Gymea, Australia) according to the manufacturer's instructions. PCR products of each gene were purified using a Perfectprep Gel cleanup kit (Eppendorf, Berkhausenweg, Hamburg, Germany) according to the manufacturer's instructions. The purified product was cloned into pGEM-T easy vector (Promega). The concentrations of purified product were quantified by optical absorbance at 260 nm and linearized with EcoRI restriction enzyme. Serial dilutions of the linearized plasmid were made from 106 to 102 single-stranded DNA molecules/μl for the construction of a standard curve for each experiment.(22,26)
Quantitative real time PCR
PCR reactions were performed in a total volume of 20 μl containing 200 μM dNTPs, 200 nM forward/reverse primers, 2.5 mM MgCl2, 1:31,250 SYBR Green I, 1 μl cDNA or plasmid DNA standard, and 0.5U BIOTAQ DNA polymerase (Bioline, Randolph, MA, USA). Amplification was obtained by denaturing at 95°C for 2 minutes, followed by 40 cycles of denaturing at 95°C for 30 s, annealing at 60°C (65°C for RANKL) for 30 s, extension at 72°C for 30 s, and elongation was completed by heating at 72°C for 10 minutes in a Corbett Rotor Gene 3000 thermocycler (Sydney, Australia). Varying concentrations of plasmid DNA (102, 103, 104, 105, and 106 copies/μl) were used to create a standard curve. The amount of target gene in the sample was determined from the standard curves using Corbett Real Time software. Samples were run in triplicate, and the entire experiment was repeated three times. The RT efficiency for each treatment was determined by calculating the percentage of bacterial RNA expressed in each cDNA compared with 106 copies of RNA added to the RT reaction. The RT efficiency of each RNA was used to calculate the mRNA expression of OPG and RANKL in copies per microliter in each microgram of total RNA.
Mean and SE were calculated for each treatment. ANOVA was performed to determine whether there were significant (p ≤ 0.05) differences between treatments. When an ANOVA indicated any difference among the means, Students t-test was used to determine which means were significantly different. In PCR studies, the SD was calculated using CVs of normalizing gene and the gene of interest, using the method of Colquhoun.(27)
Formation and characterization of osteoclasts in monocyte cultures
Figures 1A and 1C show control dentine slices with PBMCs that were cultured without RANKL and M-CSF and were stained for TRACP or assayed for resorption. Under these conditions, no multinucleated TRACP+ cells or pit formation occurred. Osteoclasts were present in Fig. 1B, where the cells were treated with RANKL and M-CSF for 21 days, and these cells were able to resorb dentine (Fig. 1D).
Time course study for human osteoclast formation
Positive staining for TRACP is not an accurate criterion for differentiating multinucleated giant cells from osteoclasts, because macrophages contain and develop the enzyme both in vivo and in vitro.(28–30) Multinuclearity is also not specific to osteoclasts, because there are a number of other cells that are multinucleated in bone marrow cultures.(31–34) Resorption is, however, the gold standard for identifying osteoclasts, because no other cell has been reported to have the capacity to resorb bone. For this reason, osteoclast generation was studied by examination of resorption pits (Fig. 1D).
Figure 2 shows a time course of human osteoclastic resorption activity, generated from PBMCs cultured with RANKL (50 ng/ml) and human M-CSF (25 ng/ml) over a period of 14, 21, and 27 days. In the time course study of human osteoclast activity, there was a steep increase in pit formation from 14 to 27 days. Osteoclasts began to be active around 14 days of culture, and at this time, there was about 1 ± 0.8% of the dentine resorbed. By 27 days, a sharp increase to 70 ± 18.9% was noted.
Effects of dexamethasone on osteoclast generation
To examine the effect of dexamethasone on osteoclast formation, PBMCs were cultured with vehicle or dexamethasone, RANKL, and M-CSF from days 1 to 16. The time-point of 16 days was chosen for this study because of the previous finding that osteoclasts begin to resorb dentine at ∼14 days. The mean percentage area resorbed by these cells in the vehicle-treated wells was 6.85 ± 0.94% at 16 days of culture. Figure 3A shows that lower concentrations of dexamethasone, ≤10−9 M, had no significant effects on osteoclast activity compared with vehicle-treated cultures. At the pharmacological concentration of dexamethasone (10−8 M), however, there was a significant 2.5-fold increase in the mean percentage area resorbed compared with that of vehicle control (p < 0.01). In contrast, in the presence of the highest concentration of dexamethasone, 10−7 M, osteoclast activity, as measured by bone resorbing activity, was decreased to 25% of control. Pit formation at 16 days in the presence of vehicle or dexamethasone at 10−8 M is shown in Fig. 4.
PBMCs were also cultured for a longer period to observe whether the early effects of dexamethasone persisted. PBMCs from three different donors were treated with dexamethasone in the presence of RANKL and M-CSF for 21 days. Figure 3B shows that, again, lower concentrations of dexamethasone (≤10−9 M) had no significant effects on osteoclast activity. As seen in the early phase of osteoclast generation at 16 days, dexamethasone at 10−8 M caused a 2.3-fold increase (p < 0.01) in human osteoclast activity at 21 days of culture. At the highest concentration of dexamethasone, 10−7 M, however, no statistically significant inhibition of resorption was observed, unlike that seen in 16-day cultures.
Effects of dexamethasone on life span of human osteoclasts
The effects of dexamethasone on human osteoclast survival were examined using PBMC precultured for 16 days in the presence of RANKL and M-CSF. By this time, mature human osteoclasts had formed. As shown in Fig. 5A, cells were treated with dexamethasone or vehicle in the continued presence of RANKL and M-CSF for a further 6 days. There was no significant difference in the area resorbed with all doses of dexamethasone tested compared with vehicle (Fig. 5A). Similarly, when the cells were stressed by the removal of the differentiating and survival factors, RANKL and M-CSF, from the cultures after day 16 and cultured for a further 6 days, there were no significant differences in resorption activity with dexamethasone compared with vehicle (Fig. 5B). Under similar conditions (culture for 16 days with RANKL and M-CSF and then for a further 6 days without these agents), the number of multinucleated TRACP+ cells was also not significantly different in the presence of dexamethasone compared with vehicle controls (Fig. 5C)
Effects of dexamethasone on the expression of RANKL and OPG
To assess the effects of dexamethasone on the expression of OPG and RANKL mRNA, human primary osteoblasts were treated with vehicle or dexamethasone for 24 h in serum-free DMEM. The expression of the OPG gene is 10- to 100-fold greater compared with RANKL expression in human primary osteoblasts (Fig. 6C). It was found that dexamethasone inhibited the expression of OPG mRNA (Fig. 6A) and stimulated the expression of RANKL mRNA (Fig. 6B) in a dose-dependent manner. Dexamethasone increased the RANKL:OPG ratio in a dose-dependent fashion (Fig. 6C).
Prolonged and excessive exposure to glucocorticoids causes a number of unwanted effects. Osteoporosis, with the attendant hazard of fractures, is probably one of the main limitations to long-term glucocorticoid therapy.(35) It is estimated that as many as 50% of patients receiving glucocorticoid treatment will ultimately suffer fractures. Glucocorticoid-induced bone loss has been largely considered to be caused by alterations in bone remodeling associated with decreased bone formation.(35–37) The decreased bone formation in glucocorticoid-induced bone loss is caused by inhibition of proliferation and enhancement of differentiation of pre-osteoblasts at the expense of proliferation and enhancement of apoptosis.(7,8,38) It has been proposed that any increase in bone resorption is caused by secondary hyperparathyroidism.(39–41) A number of studies, however, have shown that patients with glucocorticoid-induced bone loss do not always show an increase in parathyroid hormone concentrations in serum.(42–45)
Several studies have shown an increase in osteoclast numbers as well as an increase in osteoclastic activity in bone of patients receiving glucocorticoid treatment.(1,9,11) In vitro studies addressing the possibility of a direct role of glucocorticoids in bone resorption have yielded contradictory results, with both inhibition and enhancement of bone resorption reported. Studies on fetal rat limb bones in culture(15,46) and isolated rodent osteoclasts(18) showed a decrease in osteoclast function and inhibition of bone resorption. In contrast, a number of other studies using models such as a co-culture system of bone marrow stromal cells and spleen cells(47) or a mouse bone marrow model(48) reported an enhancement of bone resorption.
The requirement for co-culture with stromal/osteoblast cells to generate osteoclasts has been overcome by the identification of RANKL, a member of the TNF superfamily, which is expressed on the cell membranes of stromal/osteoblast cells, among others. There are many systemic hormones and local cytokines that are involved in regulating osteoclast differentiation, although RANKL is thought be the primary molecule in osteoclastogenesis.(14,49–52) This has allowed the use of recombinant RANKL to replace the requirement for the presence of stromal/osteoblastic cells for osteoclastogenesis to occur.(50,53,54) This system has an advantage in that it allowed us to look at the effects of dexamethasone directly on osteoclast function, generation, and life span without the influence of osteoblasts.
The results of the time course study were consistent with a previous study by Hirayama et al,(55) which also showed that in the human model multinucleated TRACP+ cells do not exhibit active resorption before day 14. Similar to the results reported here, previous studies carried out in this laboratory on murine myeloid osteoclast precursors (RAW264.7) also showed that dexamethasone at 10−7 M suppressed early stage osteoclast formation,(56) although this concentration is generally not reached in vivo. In contrast, in the murine study, lower concentrations of dexamethasone (≤10−8 M) had no effect on the differentiation of RAW264.7 cells into mature osteoclasts. The peculiar effect of dexamethasone at 10−8 M, a concentration achieved by pharmacological doses,(57,58) that was seen at 16 days culture with cells from two separate donors was also seen with cells from three other donors at a longer time course of 21 days. Hirayama et al.(55) also reported significantly increased responsiveness of human osteoclast precursors to RANKL in the presence of dexamethasone at 10−8 M but not at lower concentrations.
Treatment with dexamethasone did not affect the life span of mature human osteoclasts, whether assessed by resorption assay or by multinucleated TRACP+ cell numbers. In contrast, a study using a murine osteoclast precursor cell line showed that dexamethasone enhanced osteoclast survival.(56) Studies on mouse and rat osteoclasts reported contradictory results, with glucocorticoids shown to cause inhibition of apoptosis in mouse osteoclasts(10) and enhancement of apoptosis in rat osteoclasts.(59) This study provides evidence that there may be species and system differences concerning the effects of dexamethasone on osteoclasts.
Our results on osteoclast generation and life span show that glucocorticoids are likely to have a direct role in human osteoclast generation and activity, at least at a pharmacological concentration of 10−8 M.(57,58) It is likely that the effects of dexamethasone on these cells are through direct interactions with glucocorticoid receptors that have been identified on multinucleated osteoclast-like cells in human giant cell tumors of bone.(60)
The key cytokines that regulate osteoclastogenesis are RANKL, a stimulator of osteoclast differentiation, activity and survival, and OPG, an inhibitor of osteoclastogenesis. Osteoblast/stromal cells express these two genes. RANKL is a membrane bound protein, whereas OPG is cleaved from the membrane and is secreted in a soluble form. Osteoclast lineage cells express RANK on their membranes, and this binds to RANKL, forming a cell-to-cell interaction with osteoblasts that results in their differentiation into mature osteoclasts.(14,50,61) OPG also attaches to RANKL, thereby blocking osteoclastogenesis. What determines the numbers of osteoclasts generated, in part, is the balance between RANKL:OPG expression.
A method of analyzing the expression of RANKL and OPG by real time RT-PCR was developed. This method accurately measured the changes in gene expression and the exact copy numbers of RANKL and OPG expressed in human osteoblasts when treated with varying concentrations of dexamethasone. It was seen that OPG mRNA copy numbers were 10-fold higher than those of RANKL. Other studies that examined OPG and RANKL expression used real time RT-PCR with a housekeeping gene to normalize mRNA expression of OPG and RANKL and therefore are not comparable with this method. They cannot be used to measure the copy number expressed.(62,63) A possible explanation for the markedly higher OPG copy number may be that OPG is released in a soluble form and therefore is diluted by extracellular fluid, whereas RANKL is mainly membrane bound. RANKL attaches to its receptor, RANK, on osteoclast precursors by cell-to-cell interaction to cause the precursors to differentiate into mature osteoclasts.
Previous studies that examined the two cytokines by semiquantitative methods showed that dexamethasone inhibits OPG mRNA(64,65) and stimulates RANKL mRNA expression.(65) In this study, using quantitative methods, it was confirmed that dexamethasone inhibited OPG mRNA expression, stimulated RANKL mRNA, and increased the RANKL:OPG ratio in a dose-dependent manner in human primary osteoblasts. The importance of the balance between RANKL and OPG is shown by observations that overexpression of OPG in transgenic mice causes osteopetrosis, and administration of OPG to normal mice results in blockade of ovariectomy-induced bone loss.(66) In contrast, OPG gene knockout mice developed severe osteoporosis.(50,67)
The observations reported here may help to explain the clinical course of bone resorption in glucocorticoid-treated patients.(68) Although there is some variation in the reported parameters of resorption under these conditions, Rubin and Bilezekian(68) describe an early phase of accelerated bone resorption followed by considerable slowing of osteoclast-mediated resorption in the presence of continuing major suppression of osteoblasts. If the dominant effect of glucocorticoids on resorptive activity in human systems is to increase osteoblastic signals for osteoclastogenesis, accompanied by an enhancement of osteoclast precursor responsiveness to these signals, rather than an influence on lifespan of mature osteoclasts, it is reasonable to propose that these effects would become considerably less important once osteoblast numbers and activity were suppressed with prolonged glucocorticoid dosing.
In summary, while there may be species differences, there is clear evidence for a direct influence of the glucocorticoid dexamethasone on resorptive parameters in human bone cells through increased signals for osteoclastogenesis and increased responsiveness to those signals.
We thank Dr Neal Williams (Institute for Biomedical Research, University of Sydney, NSW, Australia) for assistance with real-time RT-PCR methods. This study was partially funded by AstraZeneca (Sydney, NSW, Australia), including a postgraduate scholarship provided to SS.