Minireview: Live and Let Die: Molecular Effects of Glucocorticoids on Bone Cells

Abstract

Glucocorticoids (GCs) are efficient drugs that are used to treat various immune-mediated diseases, but their long-term administration is associated with multiple metabolic side effects, including osteoporosis. Molecular analyses of the mechanisms exerted by the GC receptor have resulted in the development of GC receptor agonists that selectively repress or activate GC target genes. This review summarizes the cellular and molecular effects of GCs on bone cells and highlights the critical signaling pathways that may evolve into future therapeutic strategies.

While traditional glucocorticoids have the drawback of impairing bone cell function through several distinct molecular mechanisms, newer selective receptor agonists may have bone-sparing effects.

Glucocorticoids (GCs) are used as potent immunomodulatory drugs for a broad range of immune-mediated diseases such as rheumatoid arthritis, asthma, and multiple sclerosis. The prolonged application of high doses of GCs, however, is limited by deleterious adverse effects on multiple organ systems, including the skeleton (1). Severe bone loss occurs within the first months of systemic GC use, and long-term GC treatment results in a slower but persistent decrease in bone mass and quality, leading to an increased risk of fractures (2). Up to 50% of the patients develop vertebral fractures under prolonged use of systemic GCs (3, 4). Although studies have shown that bone mineral density increases again after cessation of GC therapy, persistent inflammatory activity of the underlying disease frequently does not allow complete discontinuation of GCs (5).

Cells of the skeletal system (osteoclasts, osteoblasts, and osteocytes) express GC receptors (GRs) and GC-modifying enzymes that determine the susceptibility to GCs. Previous research on GC actions through the classical GR has highlighted the existence of various distinct pathways that either activate or repress target genes. The fact that these pathways may be selectively activated has resulted in the development of selective GR agonists (SEGRAs) that target the GR and are able to dissociate the immunosuppressive actions from some of the adverse effects. This review focuses on the current cellular and molecular understanding of GC effects on bone cells, the underlying pathways, and the bone-protective potential of SEGRAs.

Physiological bone remodeling

Bone mass and bone quality are maintained by continuous remodeling, a physiological process that adapts bone to changing functional, mechanical, and metabolic demands. The remodeling process is balanced between cells from the osteoblastic and osteoclastic lineages. Osteoblasts derive from mesenchymal stem cells and primarily secrete a collagenous bone matrix and orchestrate its subsequent mineralization. Osteocytes are terminally differentiated osteoblasts with characteristic dendrites. Using gap junctions that are located on their dendrites, osteocytes form a network throughout the bone tissue that allows them to communicate, to sense bone damage, and to direct its repair (6). Osteoclasts, on the other hand, are of hematopoietic origin and resorb bone by creating an acidic environment and producing collagen-degrading enzymes. Whereas receptor activator of nuclear factor-κB ligand (RANKL) is an essential cytokine for osteoclastogenesis, osteoprotegerin (OPG) acts as a decoy receptor for RANKL and suppresses osteoclastogenesis (7). Intriguingly, both proteins are produced by osteoblasts, thereby underlining the intimate communication between osteoblasts and osteoclasts in the cyclic process of bone remodeling (8).

The GRs and their actions

In principle, GCs exert their biological actions through genomic mechanisms using the GR. Although nongenomic mechanisms, i.e. interactions with cellular membranes and signal transduction via membrane-bound GRs have been reported, their relevance on bone cells remains currently elusive (9). Thus, this review will only focus on genomic mechanisms.

Alternative splicing of exon 9 of the human GR results in the production of two isoforms, GRα and GRβ. Bone cells express both GRs, but GRα is more abundantly expressed than GRβ (10).

GRα

GC actions mediated through the classical receptor GRα are well defined, and studies over the last decade have yielded important mechanistic insights. The unligated receptor is located in the cytoplasm and is part of a multiprotein complex that contains several heat-shock proteins, MAPKs, and chaperones. Ligation of GRα leads to its dissociation from the multiprotein complex. After translocation into the nucleus, GRα modifies gene expression through multiple pathways. To classify the various entities of GRα-mediated mechanisms, the terms “transactivation” and “transrepression” were coined. Accordingly, genes are activated by direct binding of GRα homodimers to GC response elements (e.g. phosphoenolpyruvate carboxykinase) or other DNA-bound transcription factors (e.g. signal transducer and activator of transcription 5). Transrepression, on the other hand, occurs upon binding to negative GC response elements (e.g. osteocalcin) or the interaction of GRα monomers with transcription factors (e.g. Smad3 in the case of type I collagen), coactivators, or corepressors (1, 11). Because many proteins that regulate metabolic functions are activated by GCs, transactivation has been suggested to account for most of the adverse effects. By contrast, transrepression inhibits many inflammatory mediators (activator protein 1, nuclear factor-κB) and is thought to mediate most of the desired antiinflammatory GC effects. As a caveat, however, this model does not hold true for all GC-regulated genes, because some genes that contribute to the antiinflammatory effects of GCs e.g. type II IL-1 receptor or lipocortin, are transactivated, whereas other genes that contribute to metabolic processes, e.g. osteocalcin are transrepressed.

GRβ

GRβ differs from the classical receptor GRα in that it has a distinct sequence of helix 11 and lacks helix 12 of the ligand-binding domain. Thus, earlier studies reported that GRβ does not bind agonists and only affects gene transcription through blocking the actions of GRα. However, this concept has been challenged by a recent study showing that, after binding of certain ligands such as the GR antagonist RU-486, GRβ is able to translocate into the nucleus independently of GRα and modify the gene expression pattern of target cells (12). Although little is known about the function and regulation of the GRβ in bone or other cells, clinical and in vitro studies propose a functional role of GRβ as a determinant of GC sensitivity because its expression correlates with GC resistance in patients with inflammatory conditions and cancer (13, 14). Thus, GRβ may play a role in fine tuning intraindividual GC susceptibility, but further studies are required to assess this in detail.

Pathogenesis of GC-inducedosteoporosis (GIO)

The pathogenesis of GIO is complex and involves many organ systems. GCs impair bone metabolism through extraskeletal effects including decreased calcium resorption in the gut, increased calcium loss through the kidneys, a decreased production and/or action of sex steroids and GH, decreased muscle strength, and impaired musculoskeletal interactions, all of which contribute to loss of bone mass, decreased bone quality, and a higher propensity to falls (15).

In addition, GCs have profound (physiological and pathological) direct effects on all cell types found in bone tissue. GCs suppress bone formation and stimulate bone resorption. Histomorphometric studies of patients and experimental animal models revealed a reduced bone formation rate and trabecular wall thickness due to reduced numbers of osteoblasts as well as enhanced apoptosis of mature osteoblasts and osteocytes (16, 17). Although the number of osteoclasts and parameters of bone erosion are within the normal range or slightly above, GCs have been shown to extend the life span of osteoclasts, providing a possible explanation for the sustained bone loss during prolonged GC treatment.

GC effects on the osteoblastic lineage

In contrast to physiological amounts of endogenous GCs that are necessary for efficient osteoblast differentiation (18, 19), GC excess shifts the osteoblast-adipocyte balance toward enhanced adipogenesis (Fig. 1, orange box), inhibits osteoblast differentiation and function (blue box), and increases both osteoblast and osteocyte apoptosis (blue and green box).

Fig. 1.

Cellular effects and molecular pathways employed by GCs in the homeostasis of bone. GCs, as illustrated by cortisol, enhance the expression of adipogenic and osteogenic transcription factors in mesenchymal stem cells. Long-term exposure may increase adipogenesis (orange). The GC-mediated arrest of various signaling pathways, including IGF-1, TGF-β/bone morphogenetic protein, and Wnt suppresses the osteoblastic transcription factors Runx2 and β-catenin and attenuates osteoblast differentiation. Arrest of osteoblast differentiation translates into a lower production of extracellular bone matrix proteins (collagen type I, osteocalcin). Stimulation of the mineralization inhibitors Dmp1 and Phex impairs subsequent matrix mineralization (blue). GCs increase caspase-mediated osteoblast and osteocyte apoptosis through up-regulation of the Bim and Fas/FasL systems. Some osteoprotective drugs, e.g. PTH analogs and bisphosphonates specifically prevent GC-mediated apoptosis. This involves survival pathways steroid receptor coactivator and ERK, Wnt signaling, and gap junctional communication via connexin-43 (green). GCs directly prolong osteoclast survival via caspase 3 but reduce osteoclast activity through blocking macrophage-colony stimulating factor (M-CSF)-mediated pathways. Also, osteoclast number and activity are modulated by enhancing the RANKL/OPG ratio and by inhibiting the proosteoclastic cytokines IL-1, -6, and TNF-α. Denosumab, a human anti-RANKL antibody, is able to prevent GC-induced enhanced osteoclastogenesis (purple). BMP, Bone morphogenetic protein; DKK-1, dickkopf-1; SRC, src proto-oncogene; sFRP-1, secreted frizzled-related protein-1.

Osteoblast fate

The shift toward adipogenesis is mainly mediated by the transactivation of CCAAT/enhancer binding protein δ, which increases the expression of the pivotal adipogenic transcription factor peroxisome proliferator-activated receptor (PPAR)γ2. Concomitantly, the expression of the osteogenic transcription factor runt-related protein 2 (Runx2) is repressed (Fig. 1, orange box) (20). This phenomenon may be recapitulated in patients receiving long-term GC treatment, who concurrently lose bone and accumulate bone marrow fat (21). Further studies have identified additional molecules such as the GC-induced leucine zipper or FHL2 that are activated by GCs to control the expression of PPARγ2 and Runx2 (22, 23). In fact, these molecules are capable of promoting osteogenic differentiation through Runx2 and Wnt signaling (Fig. 1, orange and blue box). Thus, whereas physiological doses of GCs may regulate PPARγ2 and Runx2 in such a way that osteoblast differentiation is favored, pharmacological doses of GCs may preferentially activate PPARγ2 and induce adipogenic differentiation.

Osteoblast maturation and function

Excess GC levels further inhibit osteoblast differentiation as they repress bone-anabolic factors such as bone morphogenetic proteins, IGF-1, and TGFβ, which activate the essential osteoblastic transcription factors Runx2 and β-catenin (Fig. 1, blue box) (15). The canonical Wnt pathway, in particular, is crucial for osteoblastogenesis and mediates its actions via stabilizing cytosolic β-catenin, which then translocates into the nucleus and induces the expression of osteoblast-specific genes. This process is promoted by the destabilization of glycogen synthase kinase 3β (GSK3β), and can be specifically inhibited by secreted proteins such as dickkopf-1 and secreted frizzled-related protein-1. GCs have been shown to suppress Wnt signaling by increasing the expression of these inhibitors and by stabilizing GSK3β (Fig. 1, blue box) (24). However, also in the case of Wnt signaling, GC effects are bivalent, depending on the differentiation state and anatomical site. A recent study has demonstrated that endogenous GCs are in fact necessary for proper skeletal development of the cranium (25). This phenomenon was attributed to decreased canonical Wnt signaling in osteoblasts. Interestingly, long bone development was not affected in these mice, indicating that GCs differentially regulate intramembranous and endochondral bone formation.

In addition to osteoblast differentiation, GCs also impair bone matrix synthesis and composition. Gene expression levels of type I collagen and osteocalcin, two major proteins of the extracellular matrix, are transrepressed by GCs through transcriptional and posttranscriptional mechanisms (15). Moreover, expression levels of mineralization inhibitors such as Dmp-1 and Phex were increased in bone tissue of mice after GC treatment (26). Of note, the osteoporosis drugs, teriparatide, a PTH analog, and the bisphosphonate risedronate, have been shown to improve bone quality in a mouse model of GC-induced osteoporosis partially by counteracting GC effects on mineralization inhibitors (Fig. 1, blue and green box) (26).

Osteoblast and osteocyte apoptosis

Among the various skeletal effects of GCs, the induction of apoptosis of osteoblasts and osteocytes appears to be predominant. This aspect substantially contributes to impaired bone quality because the main function of osteocytes is to sense local microdamage and to induce bone repair. GC-induced apoptosis is associated with enhanced activities of caspases 3, 7, and 8 that are downstream effectors of the activated proapoptotic gene Bim and the Fas/FasL death receptor pathway (Fig. 1, green box) (27, 28, 29). GCs also induce apoptosis in osteoblasts by stabilizing GSK3β activity. Recent studies indicate that bisphosphonates and PTH can reverse the proapoptotic effects of GCs through activating antiapoptotic pathways, including ERK and steroid receptor coactivator as well as Wnt signaling. Interestingly, both drugs enhance osteocyte life span by increasing connexin-43-mediated communication through gap junctions, a critical mechano-sensing pathway (6, 30). Of note, administration of the antiarrhythmic rotigaptide, which modulates gap junctional communication, has been shown to maintain bone mass in the ovariectomized rat, a model of postmenopausal osteoporosis, which is also characterized by enhanced osteoblast/osteocyte apoptosis (31). Thus, modulating cell-cell communication through gap junction and hemichannel formation may represent a novel therapeutic principle.

GC effects on osteoclasts

GCs exert direct and indirect actions on osteoclasts. Although GCs inhibit the expression of proosteoclastic cytokines such as IL-1 and -6 in osteoblasts and stromal cells, the increase of the RANKL/OPG ratio seems to be a key mechanism of GC-induced bone loss (Fig. 1, purple box) (32). Multiple pathways have been proposed to inhibit OPG expression. GCs have been found to transrepress OPG by decreasing the amount of the phospho-c-Jun protein, which maintains steady state transcription of the OPG gene. Promoter activity studies revealed that GC-mediated suppression of OPG production is dependent on canonical Wnt signaling (33). The significance of the RANKL/OPG pathway in the pathogenesis of GC-induced osteoporosis has been underscored in mice expressing a chimeric RANKL protein, which is neutralized by the human monoclonal antibody denosumab. Treatment of these mice with denosumab maintained bone mass and bone strength and suppressed bone resorption during a 4-wk course of GC treatment (34).

The direct effects of GCs on osteoclasts in vitro are ambiguous. In vivo studies using mice overexpressing 11β-hydroxysteroid dehydrogenase 2 (11β-HSD2) or lacking the GR specifically in osteoclasts revealed that GCs prolong osteoclast survival by inhibiting caspase 3-dependent apoptosis (Fig. 1, purple box) (35, 36). GCs also reduce osteoclast activity by interfering with the formation of the ruffled border, an organelle that is indispensible for bone resorption. Disruption of the osteoclastic cytoskeleton due to decreased activation of rhoA, rac, and vav3 through macrophage colony-stimulating factor was identified as the underlying mechanism (36). Intriguingly, mice containing the osteoclast-specific knockout of the GR were partially protected from GC-suppressed bone formation, indicating that some of the GC-mediated effects on osteoblasts are indirectly mediated by osteoclasts.

Susceptibility to GC-induced osteoporosis

Interestingly, some individuals are more susceptible to develop GIO than others, indicating that GC bioavailability may vary among individuals. However, predicting the individual GC sensitivity is difficult, because multiple determinants may contribute, including polymorphisms in the GR gene, the activity of the underlying disease, the distribution and metabolism of steroids, the choice of exogenous steroids, the availability and affinity of GRs, the interactions with nuclear coactivators and corepressors, and the presence and local activity of GC-modifying enzymes such as 11β-HSD (37). Especially the 11β-HSD system has emerged as a key regulator of active GC availability because 11β-HSD1 catalyzes the conversion of the inactive cortisone into the active cortisol, and 11β-HSD2 isoform catalyzes the reverse reaction. Of note, exogenously administered GCs differ in their affinities for the 11β-HSD isoforms (38). For example, oxidation of steroids by 11β-HSD2 is diminished in fluorinated (dexamethasone) or methylated (methylprednisolone) GC compounds, and the reduction by 11β-HSD1 depends upon the steroids structure. Thus, it is important to consider the metabolism of synthetic GCs by 11β-HSD as well as the target tissue that may preferentially express 11β-HSD1 or 11β-HSD2 (e.g. colon, kidney) to optimize systemic GC therapy.

Bone-sparing GCs on the horizon?

In the last decade, considerable effort has been made to improve the risk-benefit ratio of GCs. Among these, SEGRAs appear to be the most promising development. By preferring GR-protein interactions rather than GR-DNA binding, these drugs have been found to dissociate transactivating from transrepressing activities. As a result, many compounds have been developed that display similar antiinflammatory actions as conventional GCs but are accompanied with fewer metabolic side effects (Table 1). ZK 216575 and AL-438 are two particular SEGRAs that have been investigated regarding their bone-sparing potential. Although AL-438 was as efficient as prednisolone in reducing inflammation in a rat model of adjuvant arthritis and paw edema, it did not reduce the bone mineral apposition rate or cancellous bone mass compared with prednisolone (39). Also, ZK 216575 has been shown to exert a similar reduction of ear inflammation in rats and mice compared with prednisolone, but a reduced induction of diabetes mellitus and skin atrophy (40). Although the effects on bone were not assessed in this study, in vitro experiments revealed that ZK 216575 may be bone sparing, because it was a poor inducer of RANKL (41). Of note, a recent in vivo study demonstrated that GR dimerization-deficient mice were not protected from GC-induced bone loss (11). This suggests that GC-mediated suppression of bone formation is not entirely dependent on transrepression through GR-DNA binding, but may also depend on protein-protein interactions. Because the structural and functional mechanisms that are required to dissociate transrepression from transactivation to achieve maximal antiinflammatory activity while minimizing side effects are incompletely understood, development of the perfect SEGRA will remain a challenge. Nevertheless, a better understanding of the GR-mediated actions may eventually result in the generation of more specific (nonsteroidal) antiinflammatory ligands.

Table 1.

Profile of SEGRAs

Compound	Manufacturer	Dissociating effects in vitro	Dissociating effects in vivo	Reference
AL-438	Abbott-Ligand	Yes	Yes	(39 42 )
BI-115	Boehringer Ingelheim	Yes	Yes	(11 43 )
Compound A	University of Ghent	Yes	Yes	(44 45 )
Compound 15	Merck Research Laboratories	Yes	AI, not tested for SEs	(46 )
Compound 25	UC-SF	Yes	AI, not tested for SEs	(47 )
Compound 28	Merck Research Laboratories	Yes	AI, not tested for SEs	(48 )
Compound 60	GSK	Yes	AI, not tested for SEs	(49 )
LGD-5552	Ligand	Yes	Yes	(50 )
RU 24858	Sanofi Aventis	Yes	No	(51 )
ZK 216348	Bayer Schering Pharma	Yes	Yes	(40 )

Modified from Ref. 11 . AI, Antiinflammatory; SEs, side effects.

Conclusion

Long-term treatment with GCs is associated with profound bone loss and increased bone fragility. Across various species and model systems, GCs control the birth (differentiation), life (activity), and death (apoptosis) of bone cells. These fundamental effects translate into insufficient matrix production by osteoblasts, impaired matrix mineralization, increased osteoblast and osteocyte apoptosis, and prolonged life span of osteoclasts, which in sum contribute to bone loss. Some of the molecular signaling cascades underlying these cellular phenomena have been defined and provide novel targets for future therapies such as interfering with the Wnt or RANK pathway. A deeper understanding of the regulation of bone-specific proteins by the GR would facilitate the evaluation of bone-protective SEGRAs and their usefulness in clinical medicine.

NURSA Molecule Pages:

• Ligands: Dexamethasone | Hydrocortisone;

• Nuclear Receptors: GR.