Regulation of bone turnover by calcium-regulated calcium channels

Abstract

Calcium plays multiple roles in osteoclast formation, survival, and activity. Intracellular calcium is determined both by the release of intracellular stores and the influx of extracellular calcium through a variety of calcium channels. Osteoclasts express several classes of calcium channels, including ryanodine receptors (RyRs), inositol-1,4,5-trisphosphate receptors (IP₃Rs), and calcium release-activated calcium channels (CRACs), which respond to depletion of intracellular stores. IP₃R2 is expressed in osteoclast precursors and activated by cytokines that stimulate osteoclast differentiation. In mature osteoclasts, the IP₃R1 isoform is highly expressed and is implicated in nitric oxide-cGMP-stimulated processes. RyR calcium channels may contribute to the release of intracellular calcium stores, while RyR2 in the plasma membrane may act to limit osteoclast activity based on extracellular calcium concentration. Orai, through regulation by endoplasmic reticular store-sensing proteins, including Stim-1, may also mediate calcium influx and act as a signal amplifier for calcium release by other calcium channels. Together, these receptors allow intracellular Ca²⁺ signals to modulate bone turnover and, through calcium-sensing functions, allow coupling of osteoclast activity to extracellular conditions and integrating additional cytokine and nitric oxide signals via transient intracellular calcium signals.

Introduction

The skeletal matrix in terrestrial vertebrates is continually resorbed and re-formed for growth and repair and is critical for the regulation of serum calcium. Osteoclasts resorb mineralized matrix in an acidic extracellular compartment at the bone surface and release calcium and phosphate into the extracellular space and circulation. Because calcium homeostasis must be tightly controlled, both osteoclast numbers, reflecting both formation and survival, and resorptive activity are highly regulated. Serum calcium is chiefly regulated by parathyroid hormone, which is released in response to signals from calcium sensors in the parathyroid, but there is also direct regulation of osteoclasts by extracellular calcium.

In addition, calcium plays a key role in the intracellular regulation of the osteoclast. Calcium is a ubiquitous second messenger involved in multiple signaling pathways mediating diverse physiological functions. For the osteoclast, calcium signaling has key roles not only in the regulation of mature cell function, but is also implicated in the regulation of osteoclast formation.

While the presence of calcium fluxes in osteoclasts have long been known,¹ the molecular mechanisms that govern calcium release and their links to osteoclast regulators and function are as yet only partially understood. In this review, we will focus on the role of inositol-1,4,5-trisphosphate receptor (IP₃R) and ryanodine receptor (RyR) channels in osteoclast formation and function. We will also discuss briefly the emerging evidence for mediators of calcium-release-activated calcium influx in osteoclasts.

1,4,5-trisphosphate receptors and osteoclast activity

The IP₃R channels are large tetrameric transmembrane proteins, found primarily in the endoplasmic reticulum (ER), where they mediate the release of intracellular calcium stores in response to extracellular signals.² IP₃R-mediated calcium release is regulated by the binding of its principal ligands, inositol-1,4,5-trisphosphate and calcium ions, as well as by interactions with other signaling proteins, including kinases for which IP₃R is a substrate. These may include tyrosine kinases of the src family and both cAMP and cGMP-dependent serine-threonine kinases. The classical pathway for ER calcium release by IP₃R involves receptor activation of phospholipase Cγ, which cleaves membrane phospholipids to release diacylglycerol and inositol-1,4,5-trisphosphate (IP₃). Although first identified in ER membranes, it is now known that some IP₃Rs are localized to other cellular membranes including the plasma membrane and may thus play roles, under some circumstances, in calcium influx, as well as release of calcium stores.³

In mammals and in other vertebrates, three genes encode inositol-1,4,5-trisphosphate receptor isoforms (ITPR1, ITPR2, and ITPR3), and splice variants exist for at least the type 1 and type 2 IP₃R isoforms.² IP₃R isoforms and splice variants show tissue-specific expression, with many cells expressing more than one IP₃R type.⁴ While studies of the biophysical properties of IP₃R channels have focused primarily on homo-tetrameric IP₃R, in vivo, hetero-tetramers involving at least two isoforms are known to occur. Evidence of osteoclast expression of all three IP₃R isoforms has been reported. In studies of rat osteoclasts in vitro, Morikawa et al.,⁵ reported labeling with antibodies raised to all three IP₃R isoforms by fluorescence microscopy. IP₃R2 and IP₃R3 were also identified in osteoclasts formed from mouse bone marrow mononuclear cells.⁶ We found IP₃R1 in CD14 human osteoclast precursors and in osteoclasts by Western blotting, gene expression profiling, and real-time PCR;⁷ we also found IP₃R2 in human osteoclasts and their monocytic precursors, with the mRNA for this second receptor being typically 20% of the quantity of IP₃R1 mRNA (unpublished data).

Many extracellular signals to which osteoclasts respond activate phospholipase C and may thus act through IP3R-dependent calcium release,⁸ but in few cases has this pathway been explicitly demonstrated. Recently, however, Menteverri et al.⁹ demonstrated that a calcium-sensing receptor, through which high extracellular calcium can induce osteoclast apoptosis, specifically requires IP₃-dependent calcium release for signaling.

RANKL-stimulated osteoclastic differentiation itself is dependent on calcium-signaling pathways. In particular, activation of the osteoclast differentiation program through NFAT2 requires increases in intracellular calcium that stimulate calcineurin-dependent NFAT2 dephosphorylation and thus activation. The involvement of IP₃R was directly evaluated by Kuroda et al.,⁶ who found marked inhibition of RANKL-induced osteoclast formation from precursors lacking the type 2 IP₃R. The IP₃R2-deficient osteoclast precursors failed to show RANKL-stimulated calcium oscillations, but, consistent with the role of calcium in NFAT2 activation, osteoclastogenesis was restored in IP₃R2-deficient mice by introduction of constitutively active NFAT2. It is unclear whether the IP3R2 acts alone in regulating calcium oscillation, or whether other calcium-regulated calcium-release regulatory proteins function in the same pathway. Further studies are needed to evaluate how agents that modulate inositol-1,4,5-trisphosphate may affect osteoclast maturation.¹⁰

The mechanism by which IP₃R-mediated calcium release is stimulated in RANKL-treated osteoclast precursors is also incompletely understood. IP₃R-mediated calcium release may be a consequence of the costimulatory signals from ITAM-containing receptors, such as the FcRγ receptor and DAP12, which activate phospholipase C-γ.¹¹ Yang and Li have also identified the regulator of G-protein signaling 12 (RGS12) as a potential mediator of calcium signals involved in RANKL-induced osteoclast differentiation.¹² They demonstrate RGS12-dependent PLC-γ activation and calcium fluxes in RANKL-treated osteoclast precursors. A role for the calcium-sensing receptor in osteoclastic differentiation has also been suggested: in the absence of a calcium-sensing receptor that stimulates IP₃R-mediated calcium release, RANKL-stimulated osteoclastic differentiation of marrow precursors was also reduced.⁹

IP₃R1, nitric oxide, osteoclast motility, and additional regulatory proteins

Nitric oxide (NO) has important antiresorptive effects on bone that are mediated, in part, by NO regulation of osteoclast adhesion and motility (Fig. 1). Investigation into the mechanisms underlying nitric oxide effects on osteoclasts have identified IP₃R1 as a critical target, regulated by NO-dependent phosphorylation.^7,13

Figure 1.

Regulation of osteoclast motility by calcium and NO. NO is an autocrine and paracrine stimulus, being produced by mesenchymal cells including chondrocytes and osteoblasts, as well as a regulated product of monocyte family cells including osteoclasts. NO causes cGMP synthesis, activating PKG I and leading to reorganization of osteoclast attachment proteins. This can allow the cell to move, or it may precede apoptosis. PKG I mediates actin disassembly by phosphorylation of intermediate proteins. VASP is a major target; the integrin-related assembly includes αvβ3, migfilin, and other proteins.¹³ PKG I and VASP are essential for a local Ca²⁺ signal. This Ca²⁺ signal depends on the cytoskeletal assembly and cannot be produced without VASP and src. The inositol-1,4,5 trisphosphate receptor-1 is the key Ca²⁺ pulse generator, and Ca²⁺ pulses activate μ-calpain, which is also required for motility.⁷ Calmodulin activated proteins also include calcineurin, a phosphodiesterase, and a Ca2⁺-ATPase (not shown), which dephosphorylate proteins and reduce cGMP and Ca²⁺ terminating the motility cycle. By homology with other cells, PKG1β may downregulate the calcium signal as well by phosphorylating IP3R1,¹⁴ although intermediate protein-localized PKG1β activity in osteoclasts is uncharacterized.

In osteoclasts, as in many other cell types, a major effect of nitric oxide is to stimulate guanylate cyclase, increase cGMP, and activate the cGMP-dependent kinase (PKG), which in turn acts on multiple targets. NO-treated osteoclasts also show increased tyrosine kinase activity, including PKG-dependent activation of src-family kinases.⁷ Both PKG and src family kinases can phosphorylate IP₃R1. Studies of the effects of src family kinases on IP₃R1 in lymphoid cells revealed phosphorylation of IP₃R1 Y353 that was associated with enhanced calcium release.^14,15 In contrast, PKG phosphorylation of IP₃R1 appears to be downregulatory, inhibiting calcium release stimulated by inositol trisphosphate.^16,17 We hypothesize that NO-activated PKG initiates IP₃R-dependent calcium release through its indirect activation of src family kinases.⁷ The tyrosine phosphorylation of IP₃R1 produces an initial enhancement of calcium release, activating calpain and cellular detachment. The downregulatory effects of the PKG phosphorylation of IP₃R1 would act to limit the duration of increased calcium release.

An emerging concept is that large calcium channels, including IP₃Rs, incorporate cassettes of binding domains, in addition to the calcium channel regions, to allow the proteins to assemble supramolecular complexes of modifying proteins, the composition of which vary from cell to cell and with cellular activity. Proteins complexed with IP₃Rs may include the phosphoinositide-regulated protein serine/threonine kinase mTOR (mammalian target of rapamycin), FK506-binding protein 12 (calstabin 1), and the calcium-regulated phosphatase calcineurin.¹⁸ Under some conditions the cAMP-dependent protein kinase is co-precipitated with IP₃R1, as are protein phosphatases PP1 and PP2a.¹⁹ In many cell types, the IP₃R1 is also predominately co-precipitated with an additional protein-binding partner, the product of the Mrvi1, commonly called the IP₃R-associated cGMP-dependent kinase substrate, or IRAG.^17,20 Mrvi1 was independently characterized as a putative tumor-suppressor gene in mice, identified at an integration site for a leukemia virus.²⁰ Two transcription start sites were identified for Mrvi1, resulting in transcripts that encode long and short forms of the protein. The two forms are distinguished only by the presence or absence of an 85-amino-acid N-terminal region predicted to target the larger gene product to the endoplasmic reticulum.²⁰ The function of the short form of IRAG is unknown. The interaction of IRAG with PKG1 is limited to one isoform of PKG1, PKG1β, which binds a region of IRAG that is common to both the long and short forms of IRAG.^20,21 IRAG is essential to cGMP regulation of calcium release in cell systems studied to date where IP₃R1, IRAG, and PKG1β are expressed.^17,22 We find that both the long and short forms of IRAG are expressed in osteoclasts (unpublished data), but the function of the protein is untested in this context. Because IRAG associates with proteins other than the IP₃R1, its functions in the osteoclast are by no means certain. Other possible functions include regulation of nuclear translocation, as detected in other cell types.²³ On the other hand, regulation of IRAG by PKGIβ is conserved; the key phosphorylation site of the bovine protein is serine 696, and human IRAG is phosphorylated by PKG at the homologous serine (S677) near the protein’s C-terminal.^21,24

Ryanodine receptors in osteoclasts

The ryanodine receptor-1 isoform, RyR1, was first characterized as a calcium-release channel in skeletal muscle sarcoplasmic reticulum; RyR2 and RyR3 isoforms occur in cardiac and smooth muscle, respectively. RyRs are among the largest naturally occurring proteins, having a molecular mass of 550–600,000. The gene RYR1 has 175 exons (GenBank NM˙000540), 106 of which occur in the protein-coding region. RyRs 2 and 3 have similar protein sizes and numbers of exons. RyRs function as homo-tetrameric complexes with molecular masses over 2 million. RYR1 and RYR2 mutations are associated with malignant hyperthermia and skeletal muscle myopathies and with cardiac arrhythmias and myopathies respectively. The RyRs include regions strongly homologous to IP₃R. However, they are regulated by different binding proteins and posttranslational modifications. These can involve cAMP-dependent protein kinase A, the phosphatases PP1 and PP2A, calstabins (FK506-binding protein 12 and FK506-binding protein 12.6), calmodulin, and phosphodiesterases.²⁵

Pharmacologic, biophysical, and immunochemical evidence demonstrates that osteoclasts express the RyR2 isoform at a unique surface membrane site (Fig. 2), with the channel-forming domain normally facing sarcoplasmic reticular lumen oriented extracellularly.²⁶ This thus results in a ryanodine-sensitive, single-channel calcium conductance.²⁷ These RyR2s has been implicated in the cellular retraction (“R-effect”) and increased intracellular calcium activity associated with sustained inhibition both of osteoclastic bone resorption and of enzyme secretion after elevations in extracellular calcium.²⁸ Such parallel findings would be consistent with a specific and causally related set of functional and morphologic events following such divalent ion elevations that would culminate in a reduction of bone resorptive activity driven by levels of local extracellular calcium.

Figure 2.

At least two large and highly regulated calcium channels mediate osteoclast retraction and detachment. The RyR2 is, in the osteoclast, a plasma membrane receptor that under some circumstances colocalizes with cell attachment.²⁶ In contrast, the IP3R1 is localized to endoplasmic reticulum, where one of its major functions is transducing calcium signals secondary to nitric oxide.⁷ It is probable, although yet poorly studied, that calcium-activated calcium release, including that via Stim1 and Orai, may contribute to downstream calcium activation in some cases.

This effect is modified or replicated by other known physiological and pharmacologic RyR modulators. These include ryanodine itself, its known regulator cyclic adenosine diphosphate ribose (cADPr), and the known pharmacologic agonists perchlorate, caffeine, and ruthenium red.^29,30 Osteoclasts thus are included in the increasing number of cell types that show a capacity to detect and respond to alterations in the extracellular calcium. Admittedly, involvement of such a mechanism in the local regulation specifically of osteoclast activity would predict a calcium receptor that would sense changes over considerably higher, millimolar, local calcium concentrations than in some of the other examples in which this phenomenon has been described. This specific response contrasts with the quiescence (“Q-effect”) in cell motility, granule movement, and margin ruffling that follow other stimuli in the absence of altered intracellular calcium often following systemic hormonal as opposed to local calcium stimuli.³¹

These findings implicate the surface RyR in the chain of currently incompletely characterized events connecting elevations of extracellular calcium to release of intracellularly stored calcium. On the one hand, this might involve its directly sensing extracellular ionized calcium conceivably through its extracellularly facing low-affinity calcium binding site while also permitting modulation by cADPr produced by the ADP-ribosyl cyclase, CD38. On the other hand, there could be couplings, whether direct or involving intracellular messengers with or without kinase modulations or a calcium-induced calcium release, between the RyR and calcium detection and release proteins. In addition, the extent to which RyR, IP₃R, and other calcium-release mechanisms interact remains to be be clarified, again whether through calcium-activated calcium release or shared kinase activators.³²

Other channels including Orai and its regulator Stim1

While a number of mechanisms are undoubtedly involved in calcium release-activated calcium currents, two proteins that are clearly involved in this regulatory mechanism, often downstream of primary signals activating IP₃Rs, are Orai and Stim1. Stim1 is a calcium-sensor protein that is redistributed within the endoplasmic reticulum as a function of calcium concentration, while Orai is the functional calcium channel activated by Stim1.

Although not well studied in osteoclasts and precursor monocyte-macrophage family cells, other calcium-activated calcium-release proteins including Orai and Stim1 are expressed,³³ and we have consistent findings from gene-expression profiles of monocytes and osteoclasts (unpublished data). These are potentially important because they may function in a variety of receptor-mediated calcium-releasing pathways related to cellular differentiation, and they may also act in concert with IP₃Rs in pathways involving inositol phosphate synthesis.³⁴ Orai and Stim1 potentially may interact with other calcium-release mechanisms such as the RyR, to co-ordinate the activity and differentiation of osteoclasts in response to primary calcium signals, since Orai is also a calcium-activated calcium-release mediator. Indeed, the popular paradigm in differentiation of lymphocytes is for Stim1 to couple store-operated calcium release by IP₃Rs or RyRs to Orai activity.³⁵

There are many additional calcium transport proteins in osteoclasts that participate in “housekeeping” functions, such as calcium pumps that maintain calcium stores and the cytoplasmic extracellular calcium gradient,³⁶ and calcium transport proteins involved in clearing bulk calcium, which include TRPV5.³⁷ However, the voltage-dependent calcium channels associated, in excitable cells, with store-dependent calcium release appear not to be present in osteoclasts.³⁸ Our experience with large-scale gene-screening of CD14 cells and osteoclasts is in keeping with this view and suggests that, if L-type voltage-dependent calcium channels are present in osteoclasts, they are present in small amounts and likely would have highly specialized functions.

Conclusion

Osteoclast activity is modulated by calcium via at least two, and probably three, types of calcium channels. The RyR2 is, in the osteoclast, a plasma membrane-associated channel that functions in the sensing of extracellular ionized calcium. IP₃Rs expressed in osteoclasts include IP₃R1 and IP₃R2. IP₃R2 expressed in osteoclast precursors appears to be important to calcium oscillations that are required, under some circumstances, for normal osteoclast differentiation. IP₃R1 is expressed at higher levels in mature osteoclasts; IP₃R1 calcium release downstream of nitric oxide stimulation appears to be required for NO-induced osteoclast motility. IP₃R1 in osteoclasts probably is also regulated by other mechanisms, however. The Orai-Stim1 system may act to amplify calcium release initiated by other mechanisms, and may be activated by other signals, but this has not yet been studied adequately in bone-resorbing cells.