Cortical bone loss caused by glucocorticoid excess requires RANKL production by osteocytes and is associated with reduced OPG expression in mice

Abstract

Glucocorticoid excess is a major cause of low bone mass and fractures. Glucocorticoid administration decreases cortical thickness and increases cortical porosity in mice, and these changes are associated with increased osteoclast number at the endocortical surface. Receptor activator of NF-κB ligand (RANKL) produced by osteocytes is required for osteoclast formation in cancellous bone as well as the increase in cortical bone resorption caused by mechanical unloading or dietary calcium deficiency. However, whether osteocyte-derived RANKL also participates in the increase in bone resorption caused by glucocorticoid excess is unknown. To address this question, we examined the effects of prednisolone on cortical bone of mice lacking RANKL production in osteocytes. Prednisolone administration increased osteoclast number at the endocortical surface, increased cortical porosity, and reduced cortical thickness in control mice, but none of these effects occurred in mice lacking RANKL in osteocytes. Prednisolone administration did not alter RANKL mRNA abundance but did reduce osteoprotegerin (OPG) mRNA abundance in osteocyte-enriched cortical bone. Similarly, dexamethasone suppressed OPG but did not increase RANKL production in cortical bone organ cultures and primary osteoblasts. These results demonstrate that RANKL produced by osteocytes is required for the cortical bone loss caused by glucocorticoid excess but suggest that the changes in endocortical resorption are driven by reduced OPG rather than elevated RANKL expression.

therapeutic administration of glucocorticoids causes loss of cortical bone, increased cortical porosity, and fractures (28). However, the mechanisms responsible for these changes are only partially understood. Nonetheless, we and others have shown previously that administration of glucocorticoids to adult mice increases osteoclast number at the endocortical surface and that this is associated with reduced cortical thickness and increased cortical porosity (7, 24).

Receptor activator of nuclear factor-κB ligand (RANKL) is encoded by the Tnfsf11 gene and is a member of the TNF cytokine superfamily (14). RANKL is essential for osteoclast generation, as demonstrated by the osteopetrotic phenotype of mice and humans that lack a functional version of the Tnfsf11 gene (12, 25). Upon binding to its receptor RANK on the surface of osteoclast precursors, RANKL stimulates their differentiation, activity, and survival (14). Osteoprotegerin (OPG) is a soluble decoy receptor that binds RANKL and prevents it from interacting with RANK, thereby limiting bone resorption (1).

We and others have shown previously that osteocyte-derived RANKL is required for osteoclastogenesis in normal bone remodeling and for the loss of bone associated with mechanical unloading and hyperparathyroidism (18, 29, 31). RANKL is produced by a variety of different cell types, including bone marrow stromal cells, osteocytes, chondrocytes, and lymphocytes, all of which have the potential to serve as sources of RANKL during pathological conditions. For instance, RANKL produced by B lymphocytes does not participate in physiological bone remodeling but is required for the increase in osteoclasts and decrease in cancellous bone caused by estrogen deficiency (21). Similarly, RANKL produced by synovial fibroblasts is the major contributor to the formation of osteoclasts and the bone erosion caused by inflammatory arthritis (2). Changes in both RANKL and OPG production have been implicated in the pathogenesis of glucocorticoid-induced osteoporosis (8). However, the cellular sources of the RANKL involved in the increase in endocortical osteoclasts that occurs with glucocorticoid excess are unknown.

The goal of this study was to determine whether RANKL produced by osteocytes is required for the cellular and structural changes in cortical bone caused by glucocorticoid excess. For this purpose, prednisolone was administered to mice with conditional deletion of the Tnfsf11 gene in osteocytes or their control littermates. We found that mice lacking RANKL in osteocytes were completely protected from the cortical bone loss caused by glucocorticoid excess. Gene expression analysis of these mice revealed reduced OPG production but no changes in RANKL production. This latter finding suggests that, although osteocyte-derived RANKL is required for the increase in osteoclasts, glucocorticoids increase endocortical resorption by suppressing OPG production rather than by stimulating RANKL production.

MATERIALS AND METHODS

Animal studies.

Mice harboring a conditional Tnfsf11 allele (RANKL-f/f) and mice harboring a Dmp1-Cre transgene have been described previously (17, 29). In the present study, experimental animals were obtained by crossing RANKL-f/f mice with RANKL-f/f mice that were also hemizygous for the Dmp1-Cre transgene, as described previously (29). All mice used in these experiments were in the C57BL/6 genetic background. Offspring were genotyped by PCR using the following primer sequences: Cre forward 5′-GCGGTCTGGCAGTAAAAACTATC-3′, Cre reverse 5′-GTGAAACAGCATTGCTGTCACTT-3′ (product size 102 bp); RANKL-flox forward 5′-CTGGGAGCGCAGGTTAAATA-3′, RANKL-flox reverse 5′-GCCAATAATTAAAATACTGCAGGAAA-3′ [product size 108 (wild type) and 251 bp (floxed allele)]. Seven-month-old male conditional knockout mice (RANKL^ΔOt) and their control littermates were subcutaneously implanted with slow release pellets releasing placebo or prednisolone (2.1 mg·kg⁻¹·day⁻¹) (Innovative Research of America, Sarasota, FL). Mice were euthanized and tissues collected 28 days after pellet implantation. All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Arkansas for Medical Sciences.

Microcomputed tomography and histomorphometry.

Cortical and trabecular architecture was measured by microcomputed tomography (μCT) of the femur and fourth lumbar vertebra, as described previously (21). Femurs were fixed in Millonig's 10% formalin for 24 h, decalcified in 14% EDTA, pH 7.1, and embedded in paraffin to obtain 5-μm longitudinal sections. After removal of paraffin and rehydration, sections were stained for tartrate-resistant acid phosphatase (TRAP) activity and counterstained with methyl green. Osteoclasts were enumerated at the endocortical surface using an Olympus BX53 microscope and Olympus DP73 camera (Olympus, Tokyo, Japan) interfaced with a digitizer tablet with Osteomeasure software version 4.1.0.2 (OsteoMetrics, Decatur, GA). The terminology used in this study has been recommended by the Histomorphometry Nomenclature Committee of the American Society for Bone and Mineral Research (22).

Primary calvaria cell culture and osteocyte-enriched organ culture.

Primary calvaria cells were derived from 5-day-old wild-type mice, as described previously (9). For osteocyte-enriched organ culture, femurs and tibias from 2-mo-old female C57BL/6 mice were harvested, and muscle tissue was removed using sterile gauze. After the distal and proximal epiphyses were removed, the bone marrow was flushed out with Hanks' balanced salt solution. The periosteum was removed by scraping the bones with a scalpel while keeping the bones submerged in Hanks' solution. The resulting osteocyte-enriched shafts or 75 × 10³ calvaria cells were cultured in 1.5 ml of α-MEM supplemented with 10% charcoal-stripped FBS and 1% penicillin-streptomycin-glutamine in a 12-well plate for 18 h, followed by treatment with vehicle or 10⁻⁷ dexamethasone for 3, 12, and 24 h or 1, 3, and 12 h, respectively.

Nucleic acid isolation and gene expression analysis.

Osteocyte-enriched shafts were homogenized in Trizol Reagent (Life Technologies, Grand Island, NY) to extract total RNA according to the manufacturer's instructions. RNA was quantified using a Nanodrop instrument (Thermo Fisher Scientific), and RNA integrity was verified by resolution on 0.8% agarose gels. Five hundred nanograms of RNA was then used to synthesize cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystem, Foster City, CA) according to the manufacturer's directions. Transcript abundance in the cDNA was measured by quantitative PCR using TaqMan Universal PCR Master Mix (Life Technologies) and Taqman assays, as described previously (19). The following TaqMan assays from Life Technologies were used: tartrate-resistant acid phosphatase (Mm00475698_m1), calcitonin receptor (Mm00432271_m1), cathespin K (Mm00484036_m1), RANKL (Mm00441908_m1), osteoprotegerin (Mm00435452_m1), osteocalcin, (forward 5′-GCTGCGCTCTGTCTCTCTGA-3′, reverse 5′-TGCTTGGACATGAAGGCTTTG-3′, probe 5′-FAM-AAGCCCAGCGGCC-NFQ-3′), glucocorticoid-induced leucine zipper (GILZ; forward 5′-GTGGTGGCCCTAGACAACAAG-3′, reverse 5′-TCACAGCGTACATCAGGTGGTT-3′, probe 5′-FAM-CACGAGGTCCATGGCCTGCTCA-NFQ-3′), and the housekeeping gene ribosomal protein S2, (forward 5′-CCCAGGATGGCGACGAT-3′, reverse 5′-CCGAATGCTGTAATGGCGTAT-3′, probe 5′-FAM-TCCAGAGCAGGATCC-NFQ-3′). Gene expression was calculated using the ΔC_T method (16), and ribosomal protein S2 (ChoB) levels were used for normalization. For genomic DNA isolation, humeral cortical bone fragments were decalcified in 14% EDTA for 1 wk after the bone marrow was removed and the periosteum was scraped. Decalcified osteocyte-enriched bone or soft tissues (liver, muscle, or spleen) were digested with proteinase K (0.5 mg/ml in 10 mM Tris, pH 8.0, 100 mM NaCl, 20 mM EDTA, and 1% SDS) at 55°C overnight. Genomic DNA was then isolated by phenol-chloroform extraction and ethanol precipitation. To quantify Tnfsf11 gene deletion efficiency, we obtained a custom TaqMan assay from Life Technologies specific for sequences between the loxP sites (forward 5′-GCCAGTGGACTTACTCAAACCTT-3′, reverse 5′-GGTAGGGTTCAACTGAAGGGTTTA-3′, probe 5′-FAM-CCTCCTCCTCATGGTTTAGT-NFQ-3′). The custom Tnfsf11 assay was used in combination with a Taqman copy number reference assay, Tfrc (catalog no. 4458367).

Statistics.

Data were analyzed using two-way ANOVA to detect statistically significant treatment effects after it was determined that the data were distributed normally and exhibited equivalent variances. In some cases, transformations were used to obtain normally distributed data and equal variance. This was followed by all pairwise comparisons using Tukey's procedure. For experiments involving comparison of only two groups, Student's t-test or two-sample test with multiple comparisons was used. P values <0.05 were considered as significant. Results in all graphs represent means ± SD.

RESULTS

Deletion of RANKL from osteocytes blocks the cortical bone loss caused by glucocorticoid excess.

To determine whether RANKL expression in osteocytes is required for the cortical bone loss caused by glucocorticoid administration, we deleted the Tnfsf11 gene from osteocytes by crossing RANKL-f/f mice, which contain a conditional Tnfsf11 allele, with Dmp1-Cre transgenic mice (17, 29), which express the Cre recombinase in osteocytes. Quantitative analysis of the region of DNA flanked by loxP sites in Dmp1-Cre;RANKL-f/f mice, hereafter referred to as RANKL^ΔOt mice, confirmed that the Dmp1-Cre transgene caused deletion of this region in osteocyte-enriched cortical bone and not in liver or spleen (Fig. 1, A and B). Deletion was also detected in muscle tissue, consistent with a previous report of Dmp1-Cre activity in this tissue (10).

Fig. 1.

Confirmation of receptor activator of NF-κB ligand (RANKL) deletion in bone. A and B: quantitative PCR of loxP-flanked RANKL exons in genomic DNA from humeral cortical bone and the indicated soft tissues. All mice were 8-mo-old male RANKL-f/f and RANKL^ΔOt littermates. Values represent means of 9 animals/group. All statistical comparisons were performed using a t-test. *P < 0.05.

Seven-month-old male RANKL^ΔOt mice and their RANKL-f/f littermates were implanted with placebo pellets or pellets releasing prednisolone at a dose of 2.1 mg·kg⁻¹·day⁻¹ for 28 days. Analysis of the skeletal architecture revealed that cancellous bone in the femur was higher in placebo-treated RANKL^ΔOt mice compared with placebo-treated RANKL-f/f mice (Fig. 2A) but that cortical thickness and cortical area were lower and cortical porosity higher (Fig. 2, B–E). Measurement of cortical thickness and porosity in femurs of RANKL^ΔOt mice was complicated by the presence of much more cancellous bone than RANKL-f/f littermates, which makes it difficult to distinguish the boundary between cortical and cancellous bone in the conditional knockout mice (Fig. 2D). In vertebral bone, placebo-treated RANKL^ΔOt mice had high cancellous and cortical bone compared with placebo-treated RANKL-f/f mice (Fig. 2, F and G).

Fig. 2.

Deletion of RANKL using Dmp1-Cre prevents glucocorticoid-induced cortical (Ct.) bone loss and porosity. A–C: cancellous bone volume, Ct. thickness, and Ct. area were measured at the femoral metaphysis by microcomputed tomography (μCT). D and E: μCT images of femoral Ct. bone and Ct. porosity measurement at the femoral metaphysis by μCT. Scale bar, 1 mm. F and G: cancellous bone volume and Ct. thickness measured in the 4th lumbar vertebra. All mice were 8-mo-old male RANKL-f/f and RANKL^ΔOt mice treated with placebo or prednisolone for 28 days. Values represent means of 9–13 animals/group. All statistical comparisons were performed using 2-way ANOVA. *P < 0.05. BV/TV, bone volume/tissue volume.

Glucocorticoid administration did not cause a significant change in femoral or vertebral cancellous bone volume or architecture in either RANKL-f/f or RANKL^ΔOt mice (Fig. 2, A and F), as reported previously for mice in the C57BL/6 genetic background (24). In contrast, glucocorticoid administration reduced cortical thickness in the femur and spine of RANKL-f/f mice, and this effect was completely prevented in RANKL^ΔOt mice (Fig. 2, B and G). In addition, prednisolone increased cortical porosity in the femur of RANKL-f/f mice but not RANKLΔ^Ot mice (Fig. 2, D and E). Taken together, these results demonstrate that loss of RANKL in osteocytes prevents the deleterious effects of exogenous glucocorticoids on cortical bone.

Osteocyte RANKL is required for the increase in endocortical osteoclast number caused by glucocorticoids.

We and others have reported previously that the loss of cortical bone mass caused by glucocorticoid excess is associated with an increase in resorption at the endocortical surface (7, 24). Therefore, we measured osteoclast number and the percentage of bone surface covered by osteoclasts at endocortical surface of the femur in this cohort of mice. Deletion of RANKL from osteocytes reduced osteoclast number and perimeter by ∼30% at the endocortical surface when RANKL^ΔOt and RANKL-f/f mice treated with placebo were compared (Fig. 3, A–C). Excess glucocorticoids increased osteoclast number and perimeter at the femoral endocortical surface of RANKL-f/f mice, but this increase did not occur in mice lacking RANKL in osteocytes (Fig. 3, A–C). Consistent with the histological results, prednisolone stimulated expression of the osteoclast-specific genes TRAP, calcitonin receptor, and cathepsin K in osteocyte-enriched cortical bone from RANKL-f/f mice, but these changes did not occur in RANKL^ΔOt mice (Fig. 4, A–C). In contrast, osteocalcin mRNA was potently suppressed by prednisolone in both genotypes (Fig. 4F), suggesting that the suppression of bone formation by glucocorticoids was unaffected in RANKL^ΔOt mice. The low level of osteocalcin mRNA in placebo-treated RANKL^ΔOt mice is consistent with the low level of bone remodeling observed previously in these mice (18).

Fig. 3.

Osteocyte-derived RANKL is required for the increase in endocortical osteoclast number caused by glucocorticoid excess. A and B: osteoclast number per endocortical bone perimeter (Oc.N/B.Pm) and osteoclast surface per endocortical bone perimeter (Oc.S/B.Pm) measured by histomorphometric analysis starting from the region under the primary spongiosa of decalcified femurs. C: images of histological sections of femurs stained for tartrate-resistant acid phosphatase (TRAP) activity (osteoclasts are stained red) and counterstained with methyl green in endocortical bone. Scale bar, 20 μm. All mice were 8-mo-old male RANKL-f/f and RANKL^ΔOt mice treated with placebo or prednisolone for 28 days. Values are means of 5–6 animals/group. All statistical comparisons were performed using 2-way ANOVA. *P < 0.05.

Fig. 4.

Prednisolone suppresses osteoprotegerin (OPG) expression in cortical bone. A–F: quantitative RT-PCR of TRAP, calcitonin R, cathepsin K, RANKL, OPG, and osteocalcin mRNA in osteocyte-enriched cortical bone. All samples were from 8-mo-old male RANKL-f/f and RANKL^ΔOt mice treated with placebo or prednisolone for 28 days. All mRNA levels were normalized to ribosomal protein S2 mRNA levels. Values represent means of 9–13 animals/group. All statistical comparisons were performed using 2-way ANOVA. *P < 0.05.

Glucocorticoids suppress OPG expression in cortical bone.

Glucocorticoid treatment has been shown to regulate the production of RANKL and OPG in osteoblastic cells cultured in vitro as well as osteocytic cell lines (8, 11, 26, 27). Therefore, we measured the mRNA levels of these two factors in osteocyte-enriched cortical bone from the placebo- and prednisolone-treated mice. As expected, deletion of Tnfsf11 from DMP1-Cre-expressing cells decreased RANKL mRNA abundance compared with RANKL-f/f mice, but prednisolone did not change RANKL mRNA abundance in either genotype (Fig. 4D). Under basal conditions (placebo treated), OPG mRNA levels were lower in RANKL^ΔOt compared with control littermates (Fig. 4E). Nonetheless, prednisolone administration significantly reduced OPG mRNA in both genotypes (Fig. 4E). Prednisolone did not alter osteocyte density in the cortical bone of either genotype (data not shown), suggesting that these changes in gene expression were not the result of altered osteocyte number.

The gene expression analysis of osteocyte-enriched bone suggested that glucocorticoids stimulate osteoclast formation by suppressing OPG in osteocytes without affecting RANKL expression. To investigate this issue further, we analyzed gene expression in osteocyte-enriched cortical bone cultured and treated with vehicle or dexamethasone. Expression of a known target gene of glucocorticoids, GILZ, was robustly elevated after 3, 12, and 24 h of dexamethasone exposure (Fig. 5A). RANKL expression was unchanged between vehicle and dexamethasone at each time point (Fig. 5B), consistent with the in vivo gene expression results. OPG mRNA abundance was suppressed by dexamethasone at 3 but not 12 or 24 h (Fig. 5C). The lack of suppression at 12 and 24 h may be due in part to a decline in OPG mRNA in vehicle-treated cells over time.

Fig. 5.

Dexamethasone suppresses OPG in cortical bone organ culture and primary calvaria cells. A–C: quantitative RT-PCR of glucocorticoid-induced leucine zipper (GILZ), RANKL, and OPG mRNA in osteocyte-enriched cortical bone shafts from 2-mo-old female C57BL/6 mice cultured in vitro. D–F: quantitative RT-PCR of GILZ, RANKL, and OPG mRNA in primary calvaria cells. All mRNA levels were normalized to ribosomal protein S2 mRNA levels. Values represent means of 3 samples/group. Statistical comparisons between vehicle and dexamethasone at each time point were performed using 2-sample test with multiple comparison. *P < 0.05.

We also measured the expression levels of these genes in primary cultures of calvaria cells. GILZ was elevated after dexamethasone exposure at all time points (Fig. 5D). RANKL expression was almost two orders of magnitude lower than that measured in vivo, but dexamethasone suppressed expression at 3 and 12 h (Fig. 5E). OPG mRNA abundance was unchanged after 1 h but strongly suppressed by dexamethasone at 3 and 12 h (Fig. 5F).

DISCUSSION

Abundant evidence supports the idea that glucocorticoids decrease bone formation via direct effects on osteoblasts, and this is often cited as the most important mechanism underlying the bone loss associated with their therapeutic use (4, 6). We and others have recently noted that glucocorticoid administration also causes a profound increase in osteoclast number at the endocortical surface of mice and that this is associated with decreased cortical thickness and increased cortical porosity (7, 24). Since cortical bone is a major contributor to overall bone strength, this increase in bone resorption may play an important role in the increased fracture risk caused by glucocorticoid excess.

In the current study, we found that RANKL produced by osteocytes is essential for the increase in endocortical osteoclast number, the decrease in cortical thickness, and the increase in cortical porosity caused by prednisolone administration. Thus, although other cell types produce RANKL, none of them can compensate for the loss of RANKL in osteocytes for these effects of glucocorticoids on cortical bone. Importantly, we did not observe any changes in RANKL production after glucocorticoid treatment either in vivo or in vitro, but we did find reduced OPG production in both situations. Previously, we have observed the decrease in OPG with no change in RANKL expression in vivo, demonstrating the reproducibility of this finding in independent experiments (24). Based on this, we propose that glucocorticoids increase bone resorption at the endocortical surface primarily by suppressing OPG production, which allows normal levels of RANKL production by osteocytes to drive increased osteoclast differentiation.

The Dmp1-Cre transgene used in our study also causes Cre-mediated recombination in myocytes and osteoblasts (10, 29). Consistent with the former, we detected recombination of the Tnfsf11 conditional allele in muscle tissue. However, RANKL expression in murine muscle is more than 300-fold lower than in osteocyte-enriched bone (Xiong J and O'Brien CA, unpublished observations). Thus it is unlikely that myocyte-derived RANKL contributes to osteoclast formation either under basal conditions or after prednisolone administration. Similarly, we have shown that RANKL produced by osteoblasts does not contribute to osteoclast formation at the endocortical surface under basal conditions (30). Although it is possible that this situation changes with glucocorticoid excess, our observation that RANKL levels were unchanged by prednisolone administration argues against this scenario. A recent report demonstrates that the Dmp1-Cre transgene causes recombination in a subset of stromal cells designated Cxcl12-abundant reticular (CAR) cells (32). The Osx1-Cre transgene also causes recombination in CAR cells (5), but we have shown that deletion of RANKL using Osx1-Cre after 4 mo of age does not alter osteoclast number (29), indicating that CAR cells are not an important source for the RANKL involved in osteoclast formation.

The cellular source of the OPG that regulates osteoclast number is uncertain, but osteoblasts, osteocytes, and B lymphocytes have been implicated (13, 15). Our detection of OPG mRNA in osteocyte-enriched bone and its regulation by glucocorticoids are consistent with this gene being expressed by osteoblasts, osteocytes, or both. The reasons for reduced OPG mRNA in the conditional knockout mice are unclear but may be related to the very low bone turnover under basal conditions. Importantly, Henneicke et al. (7) have shown that blocking glucocorticoid action on osteoblasts and osteocytes prevents the increase in endocortical osteoclast formation and loss of cortical bone caused by glucocorticoid excess, indicating that these effects are mediated via action on cells of the osteoblast lineage. The results of the current study suggest a molecular explanation for how osteoblast lineage cells mediate the effects of glucocorticoids on bone resorption.

Glucocorticoid modulation of OPG expression occurs in different cell types, and several potential mechanisms have been identified. Previous studies have suggested that glucocorticoids can directly regulate OPG expression by binding of the glucocorticoid receptor to the OPG promoter (23) as well as indirectly by inhibiting binding of the AP1 transcription factor (11). Modulation of Wnt/β-catenin signaling has also been proposed (20). The suppression of OPG within 3 h in our organ culture experiments is consistent with direct regulation by glucocorticoids. Importantly, these results are consistent with the early increase in bone resorption markers that occurs in patients treated with glucocorticoids (3).

In conclusion, the results presented here provide evidence that the loss of cortical bone and the increase in porosity caused by glucocorticoid excess depend on the presence of RANKL produced by osteocytes and that suppression of OPG in osteoblast lineage cells may be the primary mechanism for increased cortical resorption. Further studies will be required to understand the cellular sources of OPG production as well as the mechanism by which glucocorticoids suppress it.

GRANTS

This work was supported by the National Institutes of Health Grants P01-AG-13918 and R01-AR-49794, Department of Veterans Affairs Biomedical Laboratory Research and Development Service Grant I01-BX000294, the University of Arkansas for Medical Sciences Translational Research Institute (NIH UL1-TR-000039), and Tobacco Settlement funds.

DISCLOSURES

C. A. O'Brien owns stock in Radius Health. The remaining authors have no conflicts, financial or otherwise, to declare.

AUTHOR CONTRIBUTIONS

M.P. and C.A.O. conception and design of research; M.P., J.X., and Y.F. performed experiments; M.P., J.X., Y.F., J.D.T., and C.A.O. analyzed data; M.P., J.X., Y.F., J.D.T., and C.A.O. interpreted results of experiments; M.P. and J.X. prepared figures; M.P. and C.A.O. drafted manuscript; M.P., J.X., Y.F., J.D.T., and C.A.O. edited and revised manuscript; M.P., J.X., Y.F., J.D.T., and C.A.O. approved final version of manuscript.