The role of osteoclast differentiation and function in skeletal homeostasis

Abstract

Osteoclasts are giant multinucleated cells that differentiate from hematopoietic cells in the bone marrow and carry out important physiological functions in the regulation of skeletal homeostasis as well as hematopoiesis. Osteoclast biology shares many features and components with cells of the immune system, including cytokine-receptor interactions (RANKL-RANK), intracellular signalling molecules (TRAF6) and transcription factors (NFATc1). Although the roles of these molecules in osteoclast differentiation are well known, fundamental questions remain unsolved, including the exact location of the RANKL-RANK interaction and the in vivo temporal and spatial information on the transformation of hematopoietic cells into bone-resorbing osteoclasts. This review focuses on the importance of cell-cell contact and metabolic adaptation for differentiation, relatively overlooked aspects of osteoclast biology and biochemistry.

Most of the skeleton in the human body is formed through the process of endochondral ossification using cartilage as the template (1). During growth, the extension of the long bones is dependent on the function of chondrocyte proliferation in the growth plate (2), while the width increases as a result of bone accrual at the outer edge (the periosteum) along with bone erosion at the inner surface (the endosteum); thus the size and shape of each bone element undergoes changes throughout the process of bone modelling. After reaching skeletal maturity, the quantity as well as the quality of bone is maintained by bone remodelling (3), a dynamic process in which bone is first resorbed by osteoclasts, followed by a filling in of the resulting lacunae by osteoblasts; thus the resorption and subsequent formation are necessarily tightly coupled in terms of both space and time.

Osteoclasts play critical roles in skeletal development and homeostasis (4). In other words, they are central in all three fundamental processes of bone biology, endochondral ossification during development, bone modelling during growth and bone remodelling in adulthood. Osteoclasts are found only on the surface of calcified matrix, and although the exact process of osteoclast development in vivo remains mostly an enigma, it is widely believed that the bone marrow is the site of osteoclast generation. For osteoclasts to be formed, both hematopoietic progenitor cells and cues from the microenvironment are required; the former are believed to be myeloid cells of the monocyte/macrophage lineage, and the latter constitutes macrophage colony-stimulating factor (M-CSF) and receptor activator of nuclear factor κB ligand (RANKL) (5) (Fig. 1). According to the results of our mouse genetics study, in vivo the RANKL expressed on bone surface osteoblasts seems to be an important source of osteoclastogenesis (6) (Fig. 1), while other investigators report that the RANKL produced by osteocytes embedded in the bone matrix is essential (8, 9). RANKL binding to RANK triggers the terminal differentiation into osteoclasts that takes place through intracellular kinase cascades and nuclear genetic programs coordinated by essential transcription factors such as c-Fos, NF-κB and NFATc1 (5). Several excellent reviews have highlighted the signalling molecules and pathways induced by the RANKL-RANK interaction (5, 10–12), so here we focus on the bioenergetics and biosynthetic aspects of osteoclastogenesis as well as the emerging concept of the ‘clastokines’, the number of which has continued to increase in recent years (13).

Fig. 1.

Mesenchymal and hematopoietic lineages for the ‘soil and seed’, respectively, of osteoclast formation. For osteoclast development, hematopoietic precursor cells with RANK that derive from hematopoietic stem cells (HSCs) need to interact with osteoblasts, which present RANKL on the cell surface (6). According to our analysis (7), on the way to differentiating into adipocytes from mesenchymal stromal cells (MSCs), pre-adipocytes transiently express RANKL, thereby contributing to osteoclastogenesis. Further, recent studies (8, 9) suggest that osteocytes, which terminally differentiate from osteoblasts and become embedded in the bone matrix, express RANKL in addition to OPG, a decoy receptor of RANKL.

The Relationship Between the Proliferation and Differentiation of Osteoclasts

It is generally accepted that terminal differentiation is coupled with permanent exit from the cell cycle (14). In fact, when we traced cell cycle dynamics during osteoclastogenesis ex vivo using Fucci indicators (15), we observed several waves of ‘green’ cells in the S/G2/M phase, followed by a progressive increase in ‘red’ cells corresponding to the G1 arrest that coincides with the appearance of mature osteoclasts having multiple nuclei; the nuclei of these mature osteoclasts were stained bright red (16), implying that at approximately the same time as cell fusion and osteoclast maturation, the cells had exited the cell cycle and the nuclei entered into the G0 state.

While the multinucleated osteoclasts are thus formed through the fusion of mononuclear pre-osteoclasts and are in a post-mitotic state, prior cell proliferation or DNA synthesis has been considered to be an essential requirement for osteoclast differentiation, because of the fact that when DNA synthesis is inhibited, osteoclast formation is almost completely inhibited (17). Our recently reported findings suggest, however, that it is the resultant reduction in the cell number or density rather than the inhibition of DNA synthesis per se that underlies the apparently impaired osteoclast differentiation when DNA synthesis is inhibited, since the same HU-treated cells, when re-plated at higher cell densities, exhibited restored osteoclast formation in a cell density-dependent manner (16). In fact, when we varied either the number of bone marrow macrophages (BMMs) at the start of osteoclastogenic cultures or pre-osteoclasts (preOCs) halfway through the process, both of these modifications had a substantial impact on the number of osteoclasts that finally formed, as well as the timing of the peak of osteoclast formation (Fig. 2). These findings hint a need for caution regarding the interpretation of the results of widely used osteoclastogenic cultures ex vivo; any genetic or pharmacological intervention that affects the number of pre-osteoclasts could very well have a substantial impact on the number of osteoclasts, thus affecting the readout results used for the evaluation of osteoclast differentiation.

Fig. 2.

Cell density as a critical determinant of multinuclear osteoclast formation. In ex vivo osteoclastogenic cultures induced by M-CSF and RANKL, bone marrow macrophages (BMMs) proliferate and almost double the cell number every 24 hours before they become pre-osteoclasts (preOCs) on day 2, which then fuse with one another and finally become multinuclear mature osteoclasts (mOCs) between day 3 and 4. In our recent analysis (16), the cell density of BMMs at the start of the culture, as well as of preOCs halfway through the process, is a critical determinant of the number of mOCs that form and also the timing of the peak of mOC formation. As illustrated here, for example, the usual 5 × 10³ BMMs/96-well plate starts to exhibit the formation of mOCs on day 3, reaching a maximum mOC number on day 4 (middle). When the number of BMMs is lower to start with (top), only a few osteoclasts form on day 4. When 10 × 10³ or more BMMs/96-well plate are plated at the beginning (bottom), a number of mOCs form on day 3 and only a few mOCs are visible on day 4, by which time most of the mOCs that formed earlier had died, pointing to accelerated cell death. (slightly modified from the Graphic Abstract of reference 16)

It is generally believed that RANKL, the essential differentiation factor for osteoclasts, is cytostatic and causes cell cycle arrest (18). However, we have found that RANKL in the presence of M-CSF actually stimulates DNA synthesis and cell proliferation during the early proliferative phase of osteoclastogenesis, with the same cytokine exerting an anti-proliferative activity in the latter half (16). This phase-specific dimorphic effect of RANKL appears reasonable in the light of our hypothesis that RANKL optimizes osteoclast formation by potentiating the early proliferative phase, thereby securing a sufficient number of pre-osteoclasts for subsequent cell-cell contact and fusion events.

Bioenergetic and Biosynthetic Demands of Osteoclastogenesis

The functional identity of the osteoclast lies in the ability to resorb bone, i.e. dissolving both the inorganic and organic components of the matrix, by secreting protons and collagenolytic enzymes (5, 12), which is an energy-demanding process. The osteoclast evidently contains abundant mitochondria and is assumed to undergo metabolic adaptation during the course of differentiation so as to be able to meet the increased bioenergetics demand. We have previously demonstrated that the transcription of PGC-1β (PPARγ coactivator-1β) is induced during osteoclast differentiation by CREB via reactive oxygen species (ROS) and stimulates mitochondrial biogenesis (19). Concomitantly, transferrin receptor 1 (TfR1) expression is induced post-transcriptionally via iron regulatory protein (IRP) 2, due to the heightened iron demand, and the resulting increased uptake of the iron-transferrin complex contributes to the activation of mitochondrial respiration by providing iron to the respiratory proteins (19). The increased production of ROS further potentiates osteoclast differentiation, pointing to a potential positive feedback regulation loop (Fig. 3).

Fig. 3.

Glycolysis and glutaminolysis in osteoclast differentiation and bone-resorbing function. Due to the bioenergetic needs for osteoclastogenesis and bone resorpion, osteoclasts contain abundant mitochondria, the biogenesis of which is stimulated by PGC-1β according to our analysis (19). In addition, through the action of HIF-1α, the expression of Glut1 and glycolytic genes is stimulated toward the maturation stage, which contributes to the bone-resorbing function (20). The expression of Slc1a5 and glutaminase is induced via c-MYC early during the course of differentiation, which contributes to osteoclastogenesis and the bone-resorbing function (20). On net balance, osteoclast differentiation and function are positively and negatively regulated by mTOR and AMPK, respectively (20).

Osteoclast maturation is associated with a substantial increase in biomass, even after DNA synthesis ceases at the preOC stage. Upon global gene expression analysis, we found that the expression of glucose transporter 1 (Glut1) is increased in the course of progression toward the mOC stage. The expression of glycolytic genes, such as hexokinase, phosphofructokinase and pyruvate kinase, was also progressively increased along with the expression of lactate dehydrogenase and vascular endothelial growth factor (20). Further, glucose deprivation as well as knockdown of Hif-1α inhibited the bone-resorbing function of osteoclasts in association with a decreased expression of Glut1 and glycolytic enzymes. Thus, as a result, we have proposed that HIF-1α activates the transcription of Glut1 and glycolytic enzymes in osteoclasts, thereby promoting glucose uptake and glycolysis, which are prerequisite for the bone-resorbing function (20) (Fig. 3). The importance of HIF-1α in osteoclast activation has been reported by other investigators in the context of estrogen deficiency (21).

We have made the serendipitous finding that L-glutamine is required for osteoclast differentiation when α-MEM lacking L-glutamine was used for culture by mistake. Subsequent analysis revealed that the expression of Slc1a5, a transporter of L-glutamine, was up-regulated at the transition from BMMs to preOCs, and GPNA, an inhibitor of SLC1A5-regulated transport, dose-dependently inhibited the formation of mOCs (20). Glutaminase, which converts glutamine to glutamate, is also upregulated during osteoclastogenesis, suggesting that following glutamine uptake, osteoclasts are able to convert glutamine to glutamate and then to α-ketoglutarate (α-KG) (Fig. 3). The importance of α-KG supply through glutaminolysis is underscored by the finding that the inhibition of osteoclast formation by glutamine deprivation is rescued by the addition of a membrane-permeable α-KG analog, dimethyl-α-KG (DM-α-KG). It was proposed that osteoclasts secrete L-glutamate through transcytosis (22); however, the relationship between the glutamate signalling and glutamine uptake in osteoclasts remains to be elucidated.

JQ1, which is known to displace bromodomain and extra-terminal (BET) proteins from chromatin by competitively binding to the acetyl-lysine recognition pocket of the BET bromodomains, thereby downregulating Myc transcription (23, 24), suppresses osteoclast differentiation as well as the bone-resorbing function of mOCs. Taken together, it is suggested that c-MYC activates the transcription of Slc1a5 and glutaminase, promoting glultamine uptake and glutaminolysis, thus contributing to osteoclast differentiation and function (20) (Fig. 3). The importance of MYC in osteoclastogenesis has been confirmed by other investigators (25).

A metabolic shift to aerobic glycolysis in certain cancers is known as the Warburg effect (26, 27). An emerging concept suggests that immune cell development, behavior and fate are under the tight control of intermediary metabolism, termed immunometabolism (28). Our data indicate that glucose and glutamine are essential carbon sources for osteoclast differentiation and/or function, and that this metabolic adaptation is orchestrated by two transcription factors, HIF-1α and c-MYC, and is dependent upon a coordinated balance between the nutrient and energy sensors, mTOR and AMPK (20) (Fig. 3).

Signalling Molecules from Osteoclasts to Osteoblasts

Once formed, osteoclasts are short-lived and die of apoptosis within a few days. In this short period of time, however, osteoclasts produce and secrete important extracellular signalling molecules that are relayed to bone-forming osteoblasts. In each bone remodelling cycle, osteoclastic bone resorption is followed by bone formation, thereby ensuring that both the quality and quantity of bone is maintained throughout the course of life. The concept that osteoclastic bone resorption is coupled to the recruitment of osteoblast lineage cells and new bone formation has led to intensive studies on the factors and mechanisms mediating the ‘bone coupling’ (13). Historically, the release and activation of matrix-embedded growth factors during bone resorption, such as TGF-β and IGF-I, was first proposed as ‘coupling factors’, and has recently gained experimental support from studies in mouse genetics (29, 30). More recently, sphingosine-1-phosphate (S1P), the product of phosphorylation of sphingosine by sphingosine kinase (SPHK), was identified as a secretory product of osteoclasts, along with the expression of SPHK 1 and 2 in osteoclasts (31). S1P stimulates the migration of osteoblasts (31), while the stimulation of mineralization by osteoclast conditioned media is inhibited by S1P receptor antagonist VPC23019 (32), leading to the notion that the S1P secreted from osteoclasts attracts osteoblastic cells and stimulates their bone-forming function, thereby playing an active role in the communication between osteoclasts and osteoblasts (Fig. 4).

Fig. 4.

‘Clastokines’, the secreted products of osteoclasts, regulate osteogenesis and angiogenesis. As bone marrow macrophages (BMMs) become committed pre-osteoclasts (preOCs) and terminally differentiate into mature osteoclasts (mOCs), they produce and secrete various molecules, as illustrated in this figure. Different isoforms of PDGF are secreted at distinct differentiation stages (33), then binding to PDGFR-α on mesenchymal stromal cells (MSCs) or PDGFR-β on vascular cells, thereby regulating osteogenesis and angiogenesis. Along with differentiation, S1P production is increased, which acts on osteoblasts (OBs) through S1P₃. The C3a that is locally processed from osteoclast-derived C3 stimulates osteoblastogenesis through C3aR (34). Finally, the mOCs engaged in bone resorption (active mOCs) produce particular molecules, such as Cthrc1 (collagen triple helix repeat containing 1), which favors osteoblast differentiation, diverting the cells from adipogenesis (35).

Osteoclasts produce a variety of cytokines themselves, such as PDGF and VEGF. Recently it was shown that the PDGF-BB secreted by preOCs stimulates bone formation indirectly through angiogenesis (36). Subsequently we have reported that all three of the isoforms, i.e. PDGF-AA, AB and BB are secreted, and in this temporal order as osteoclast differentiation proceeds (33) (Fig. 4). Thus, the PDGF-AA derived from early osteoclast progenitors exerts an effect on mesenchymal stromal cells (MSCs), which express PDGFR-α and have the potential to differentiate into osteoblasts (Fig. 4). The PAGF-BB secreted by mOCs as well as preOCs exerts an effect on smooth muscle cells and pericytes, which evidently express PDGFR-β (37) (Fig. 4). It has recently been suggested that CD31^hiEndomucin^hi endothelial cells in the bone marrow are important for osteogenesis as well as angiogenesis (38), and that endothelial notch activity contributes to osteogenesis through the release of Noggin (39). Whether osteoclast-derived PDGF-BB promotes osteogenesis by expanding this particular population of endothelial cells remains to be elucidated.

Our attempts to identify new ‘clastokines’, or secretory products of osteoclasts, that stimulate the differentiation of osteoblast lineage cells and their bone-forming function have yielded two interesting molecules to date. By carefully following the osteoblastogenesis-stimulating biological activity in the conditioned medium of high-purity osteoclast preparations biochemically through a series of column chromatography, we identified complement component C3a by LC-MS/MS analysis (34). C3a was detected by ELISA in the conditioned medium of osteoclasts, and when C3 gene expression in osteoclasts was knocked down, the original biological activity that stimulated osteoblastogenesis was substantially reduced. When the expression of a seven transmembrane G protein-coupled receptor of C3a in osteoblasts was knocked down, the stimulation of osteoblast differentiation by the osteoclast conditioned medium was abrogated, while the application of a C3a receptor agonist stimulated osteoblastic differentiation. These results led us to conclude that the C3a derived from osteoclasts is involved in the stimulation of bone formation acting through C3a receptor (Fig. 4).

Collagen triple helix repeat containing 1 (Cthrc1) was first identified as a gene induced in the aorta after balloon injury (40) and also as a gene induced in a chondrogenic cell line in response to BMP2 (41). We came upon Cthrc1 as a gene the expression of which was induced in mOCs only when they were placed on mineralized matrix, i.e. only when engaged in bone resorption (35). Recombinant Cthrc1 protein stimulated marrow stromal cells toward osteoblastic differentiation while inhibiting adipogenic differentiation. Most importantly, specific deletion of the Cthrc1 gene in osteoclasts using a cathepsin K-Cre mouse was found to result in a lower bone mass along with reduced bone formation (35), suggesting that osteoclast-derived Cthrc1 exerts the physiological functions of stimulating bone formation and maintaining bone mass (Fig. 4). During mouse development, Cthrc1 is involved in the Wnt/planar cell polarity signalling pathway through the formation of the Cthrc1-Wnt-Fzd/Ror2 complex (42), and since Wnt is a potent anabolic signal in bone (43), the cross-talk between the Cthrc1 and Wnt signals is of considerable interest. We have recently identified a cell surface molecule that binds to Cthrc1 and mediates its osteoblastogenesis-stimulating activity (Matsuoka K, Takeshita S, unpublished observations). Interestingly, this molecule also seems to converge on Wnt signalling, and our current work is focused on the pathophysiological relevance of the interaction between this molecule and Cthrc1 using mouse genetics.

Conclusion

Osteoclast differentiation is considered to take place in a special bone marrow niche comprised of osteoblast lineage cells and other cell types, although the precise nature of the niche in vivo remains to be identified. It is certain that osteoclast differentiation is not a self-contained process, but rather, the osteoclasts themselves, by secreting a variety of extracellular signalling molecules, have an active impact on the local microenvironment by regulating osteogenesis, angiogenesis and hematopoiesis. Future work should shed light on this type of molecule-based cell–cell communication in various pathophysiological settings and in the understanding of the mechanisms of action of osteoclast-targeted drugs such as bisphosphonates and cathepsin K inhibitors, which are known to affect bone coupling in opposite directions (44).

Glossary

Abbreviations

RANKL

receptor activator of nuclear factor-κB ligand

RANK

receptor activator of nuclear factor-κB

TRAF6

TNF receptor-associated factor 6

NFATc1

nuclear factor of activated T-cells, cytoplasmic 1

M-CSF

macrophage colony-stimulating factor

NF-κB

nuclear factor-κB

Fucci

fluorescent ubiqutination-based cell cycle indicator

HU

hydroxyurea

BMMs

bone marrow macrophages

preOCs

pre-osteoclasts

mOCs

mature osteoclasts

PGC-1β

peroxisome proliferator-activated receptor gamma coactivator 1β

CREB

cAMP response element binding protein

ROS

reactive oxygen species

TfR1

transferrin receptor 1

IRP2

iron regulatory protein 2

Glut1

glucose transporter 1

HIF-1α

hypoxia-inducible factor 1α

Slc1a5

solute carrier family 1 (neutral amino acid transporter), member 5

α-KG

α-ketoglutarate

BET

bromodomain and extra-terminal

mTOR

mammalian target of rapamycin

AMPK

AMP-activated protein kinase

S1P

sphingosine-1-phosphate

SPHK

sphingosine kinase

PDGF

platelet-derived growth factor

VEGF

vascular endothelial growth factor

Cthrc1

collagen triple helix repeat containing 1

C3

complement component 3

LC-MS/MS

liquid chromatography coupled with tandem mass spectrometry