Molecular mechanisms of triggering, amplifying and targeting RANK signaling in osteoclasts

Abstract

Osteoclast differentiation depends on receptor activator of nuclear factor-κB (RANK) signaling, which can be divided into triggering, amplifying and targeting phases based on how active the master regulator nuclear factor of activated T-cells cytoplasmic 1 (NFATc1) is. The triggering phase is characterized by immediate-early RANK signaling induced by RANK ligand (RANKL) stimulation mediated by three adaptor proteins, tumor necrosis factor receptor-associated factor 6, Grb-2-associated binder-2 and phospholipase C (PLC)γ2, leading to activation of IκB kinase, mitogen-activated protein kinases and the transcription factors nuclear factor (NF)-κB and activator protein-1 (AP-1). Mice lacking NF-κB p50/p52 or the AP-1 subunit c-Fos (encoded by Fos) exhibit severe osteopetrosis due to a differentiation block in the osteoclast lineage. The amplification phase occurs about 24 h later in a RANKL-induced osteoclastogenic culture when Ca²⁺ oscillation starts and the transcription factor NFATc1 is abundantly produced. In addition to Ca²⁺ oscillation-dependent nuclear translocation and transcriptional auto-induction of NFATc1, a Ca²⁺ oscillation-independent, osteoblast-dependent mechanism stabilizes NFATc1 protein in differentiating osteoclasts. Osteoclast precursors lacking PLCγ2, inositol-1,4,5-trisphosphate receptors, regulator of G-protein signaling 10, or NFATc1 show an impaired transition from the triggering to amplifying phases. The final targeting phase is mediated by activation of numerous NFATc1 target genes responsible for cell-cell fusion and regulation of bone-resorptive function. This review focuses on molecular mechanisms for each of the three phases of RANK signaling during osteoclast differentiation.

Key Words: Receptor activator of nuclear factor-κB ligand, Tumor necrosis factor receptor-associated factor 6, c-Fos, Nuclear factor of activated T-cells cytoplasmic 1, Immunoreceptor tyrosine-based activation motif, Ca²⁺ oscillation

INTRODUCTION

Osteoclasts are bone-resorbing cells derived from hematopoietic precursor cells[1-3]. Macrophage-colony stimulating factor (M-CSF) stimulation up-regulates expression of receptor activator of nuclear factor-κB (RANK, encoded by Tnfrsf11a) in the osteoclast precursor cell[4]. RANK, a type I transmembrane receptor with a C-terminal cytosolic tail, is responsible for osteoclast differentiation and function. RANK signaling is induced by RANK ligand (RANKL, encoded by Tnfsf11), which is a type II transmembrane protein (i.e., with a cytoplasmic N-terminus and an extracellular C-terminus). Mice with genetic deletion of Tnfrsf11a or Tnfsf11 lack osteoclasts and exhibit severe osteopetrosis[5,6]. In humans, mutations in genes encoding RANK or RANKL are associated with osteoclast poor, autosomal recessive osteopetrosis[7,8]. RANK signaling is also modified by osteoprotegerin (encoded by Tnfrsf11b), a soluble decoy receptor of RANK that blocks RANKL binding to RANK[9,10].

RANK is a member of the tumor necrosis factor receptor (TNFR) superfamily consisting of 616 and 625 amino acid residues in human and mouse, respectively[11]. RANKL is produced by osteoblasts and osteocytes[12,13] and binds in a trimeric form to RANK, initiating signaling[14,15]. Like other TNFR superfamily members, RANK lacks intrinsic enzymatic activity and transduces intracellular signals by recruiting adaptor proteins including TNFR-associated factors (TRAFs), activating nuclear factor (NF)-κB and downstream mitogen activated protein kinase (MAPK) and Akt signaling[16-18]. RANK exhibits one of the longest cytoplasmic tails of any TNFR superfamily protein, and this domain is responsible for the osteoclast-specific signaling pathway[19,20].

Spatio-temporal control of signaling downstream of RANK[21] is divided into three phases in this review (Figure 1). In the triggering phase, NF-κB, activator protein-1 (AP-1), and MAPKs are rapidly activated within an hour of RANKL stimulation in a culture system[22]. Then, during the amplifying phase, nuclear factor of activated T-cells cytoplasmic 1 (NFATc1, encoded by Nfatc1) begins to accumulate approximately 24 h after RANKL stimulation as cytosolic Ca²⁺ levels begin to oscillate[23]. Finally, in the targeting phase, RANK signaling regulates multinucleation and bone resorptive function mainly through activation of NFATc1 target genes. Concerted action of RANK and its adaptor proteins as well as immunoreceptors and other co-stimulatory molecules drive these phases. Here we review literature relevant to the molecular mechanism of RANK signaling at each phase during osteoclast differentiation.

Figure 1 Three phases of receptor activator of nuclear factor-κB signaling during osteoclast differentiation. Osteoclast differentiation downstream of receptor activator of nuclear factor-κB (RANK) signaling is divided into triggering, amplifying and targeting phases, based on the nuclear factor of activated T-cells cytoplasmic 1 activation state. ITAM: Immunoreceptor tyrosine-based activation motif; TRAF6: Tumor necrosis factor receptor-associated factor 6; Gab2: Grb-2-associated binder-2; PLCγ2: Phospholipase C γ2; RANKL: RANK ligand; Ig: Immunoglobulin.

TRIGGERING PHASE

Once homotrimeric RANKL forms complex with its receptor RANK[14,15], a cascade of downstream signaling is initiated. RANK recruits adaptor proteins to specific motifs in its C-terminal cytoplasmic tail, which contains three TRAF6 binding sites near the transmembrane domain, the a highly conserved domain in RANK (HCR) motif, and two binding sites for TRAF2 or TRAF5 near the C-terminus[20] (Figure 2). These motifs have been analyzed using various mutant RANK proteins[16,22]. Inoue and colleagues generated a CD40/RANK chimeric receptor carrying the N-terminal extracellular domain of human CD40 (TNFRSF5) and the cytoplasmic tail of mouse RANK (Tnfrsf11a), which can be specifically activated by anti-CD40 antibody and found that TRAF6 binding sites, but not the HCR, are essential for RANK signaling in the immediate-early phase[16,22]. At least three molecules, TRAF6, Grb-2-associated binder-2 (Gab2) and phospholipase C (PLC)γ2, function as adaptor molecules for RANK. TRAF6 is a really interesting new gene (RING) E3 ubiquitin ligase and Lys63-linked auto-ubiquitination is necessary for the signal transduction to activate IκB kinase and NF-κB during osteoclast differentiation[24]. Mutational analysis of PLCγ2 revealed that catalytic activity of PLCγ2 is dispensable at the triggering phase but necessary for the amplifying phase of RANK signaling[25] (see below). These adaptor proteins activate diverse signaling molecules, phosphatidylinositol-3 kinase, Akt kinase, and MAPKs including c-Jun N-terminal kinase, p38, and extracellular-regulated kinase, leading to activation of the dimeric transcription factors NF-κB and AP-1[20,25,26]. Production of reactive oxygen species via a RANKL-TRAF6-Rac1-nicotinamide adenine dinucleotide phosphate oxidase-dependent pathway is also required for MAPK activation and osteoclastogenesis[27]. In osteoclast lineage cells, NF-κB and AP-1 are composed of two molecules among p65, RelB, p50 and p52 for NF-κB, and c-Fos (also Fra-1, Fra-2 or FosB) and c-Jun (also JunB or JunD) for AP-1. Double knockout mice lacking both p50 and p52 and single knockout mice lacking c-Fos lack osteoclasts and exhibit severe osteopetrosis[28-32]. Mice overexpressing dominant negative c-Jun also develop osteopetrosis[33]. These studies demonstrate the importance of NF-κB and AP-1 activation by RANK signaling in osteoclast differentiation.

Figure 2 Triggering phase. Trimerization of receptor activator of nuclear factor-κB (RANK) by binding of RANK ligand (RANKL) immediately activates mitogen-activated protein kinases (MAPKs), nuclear factor (NF)-κB, and activator protein-1 (AP-1). An adaptor molecule complex including tumor necrosis factor receptor-associated factor 6 (TRAF6), Grb-2-associated binder-2 (Gab2) and phospholipase C (PLC)γ2 on TRAF6 binding sites of RANK is essential to induce the triggering phase. HCR: Highly conserved domain in RANK; JNK: c-Jun N-terminal kinase; Erk: Extracellular signal-regulated kinase.

TRANSITION TO THE AMPLIFYING PHASE

The transition from triggering to amplifying phase requires induction of Nfatc1 transcription, which allows cooperation with signaling downstream of immune receptors. NFAT was first identified in nuclear extracts of activated T-cells as a transcription factor that binds to the interleukin-2 (IL-2) promoter[34]. NFAT regulates not only differentiation and activation of immune cells but also the development of tissues such as skeletal muscle, cardiac valve, and bone[35]. Since the promoter of the osteoclast-specific tartrate-resistant acid phosphatase (TRAP) gene carries an evolutionarily conserved AP-1/NFAT binding element similar to the cooperative AP-1/NFAT binding site in the IL-2 promoter, it was hypothesized that c-Fos/AP-1 is required for NFAT function in osteoclasts[36]. It was demonstrated that Nfatc1 itself is a major c-Fos target gene during osteoclast differentiation[37-39]. In cells lacking c-Fos, NF-κB activity is unexpectedly elevated[40], supporting the idea that Fos and Nfatc1 induction is downstream of NF-κB p50 and p52 activation in RANK signaling[41]. It is likely that NF-κB, c-Fos/AP-1 and NFATc2 mediate basal expression of Nfatc1 in preparation for the amplification phase[42].

In concert with RANK signaling, immunoglobulin-like receptors such as osteoclast-associated receptor (OSCAR) and the triggering receptor expressed in myeloid cells (TREM)-2 transduce Nfatc1 induction signals[43,44]. Both are associated with adaptor proteins containing the immunoreceptor tyrosine-based activation motif (ITAM), such as DNAX-activation protein 12 or the Fc receptor common γ subunit[45]. After ITAM tyrosine phosphorylation, a complex containing the tyrosine kinases Bruton’s tyrosine kinase and Tec and the adaptor molecules B cell linker protein and Src homology 2 domain-containing leukocyte protein of 76 kD may facilitate cooperation between RANK and ITAM signaling (Figure 3)[46]. This combined signaling selectively leads to PLCγ phosphorylation, suggesting that integration of RANK and ITAM signaling is required for efficient activation of PLCγ during the amplifying phase. Furthermore, following elevation of intracellular Ca²⁺ levels, prior to the beginning of Ca²⁺ oscillation, Nfatc1 transcription is enhanced by Ca²⁺/calmodulin-dependent kinase IV, which phosphorylates the cAMP response element-binding protein, inducing Fos expression[47].

Figure 3 Amplifying phase. Both Ca²⁺ oscillation-dependent and -independent nuclear factor of activated T-cells cytoplasmic 1 (NFATc1) amplification are induced. Highly conserved domain in receptor activator of nuclear factor-κB (RANK)-mediated RANK signaling and immunoreceptor tyrosine-based activation motif (ITAM) signaling lead to continuous phospholipase C (PLC)γ2 activation. Regulator of G-protein signaling 10 (RGS10) determines the Ca²⁺ oscillation pattern through control of PLCγ2 by competitive binding of Ca²⁺/calmodulin and phosphatidylinositol 3, 4, 5-trisphosphate (PIP₃). Sustained Ca²⁺ oscillation contributes to NFATc1 amplification mediated by transcriptional auto-induction. In the Ca²⁺ oscillation-independent pathway, Cot kinase enhances NFATc1 stabilization through direct phosphorylation and contributes to its accumulation. ER: Endoplasmic reticulum; BLNK: B cell linker protein; SLP76: Src homology 2 domain-containing leukocyte protein of 76 kD; TRAF6: Tumor necrosis factor receptor-associated factor 6; Gab2: Grb-2-associated binder-2; IP₃R: IP₃ receptor; CaM: Calmodulin; Btk: Bruton's tyrosine kinase.

AMPLIFYING PHASE

During the amplifying phase starting approximately 24 h after RANKL stimulation in osteoclastogenic cultures, intracellular Ca²⁺ levels oscillate, and activate the Ca²⁺/calmodulin-dependent phosphatase calcineurin, which dephosphorylates NFATc1 and induces its nuclear translocation. On the HCR of the RANK C-terminal tail, PLCγ2 forms a complex with the TRAF6 and Gab2 adapter proteins in a stimulation-dependent manner[20]. An HCR deletion mutant of CD40/RANK chimeric receptor does not alter NF-κB and MAPK activation in the triggering phase but abolishes Ca²⁺ oscillation, indicating that HCR-mediated signaling is indispensable for continuous PLCγ2 activation.

Both HCR-dependent RANK signaling and ITAM signaling lead to long-term induction of PLCγ2 catalytic activity. PLCγ2 increases intracellular Ca²⁺ levels by producing inositol-1,4,5-trisphosphate (IP₃). Since Ca²⁺ oscillation during osteoclast differentiation is abolished in IP₃ receptor (IP₃R) knockout cells, Ca²⁺ release from endoplasmic reticulum (ER) via IP₃Rs is required to generate Ca²⁺ oscillation[48]. The PLCγ family consists of PLCγ1, which is widely distributed, and PLCγ2, which is primarily limited to hematopoietic cells[49]. PLCγ2 null mice exhibit an osteopetrotic phenotype[25], indicating that PLCγ2, independent of PLCγ1, is required for osteoclastogenesis.

Intracellular Ca²⁺ levels (approximately 100 nmol/L) are 20 000-fold lower than outside the cell (approximately 2 mmol/L)[50]. Ca²⁺ oscillation in osteoclasts is tightly controlled by the regulator of G-protein signaling 10 (RGS10)[51]. RGS10 is competitively bound by phosphatidylinositol 3, 4, 5-trisphosphate (PIP₃) and Ca²⁺/calmodulin, and intracellular Ca²⁺ concentration shifts the balance between RGS10-PIP₃ and RGS10-Ca²⁺/calmodulin complexes[51] (Figure 3). PIP₃ is required for membrane localization and subsequent activation of PLCγ2. As the first peak formation of Ca²⁺ oscillation, PLCγ2 activation induces transient release of Ca²⁺ from the ER, elevating intracellular Ca²⁺ concentration (Figure 3, arrows 1 and 2). RGS10 forms a complex with the Ca²⁺/calmodulin complex and increases levels of free PIP₃, further activating PLCγ2 until the intracellular Ca²⁺ level reaches its peak (Figure 3, arrow 1 and 3). Empty ER Ca²⁺ stores reload through sooth endoplasmic reticular Ca²⁺ ATPase, decreasing intracellular Ca²⁺, increasing RGS10-PIP₃, and reducing PLCγ2 activity (Figure 3, arrows 4-6). A repeat of these processes may generate Ca²⁺ oscillation through oscillatory regulation of PLCγ2 activation[51]. RGS10 knockout mice exhibit severe osteopetrosis caused by a defect in osteoclasts in vivo, indicating that Ca²⁺ oscillation is a crucial mechanism underlying NFATc1 activation and amplification during osteoclast differentiation[51].

NFATc1 is also activated by an osteoblast-induced Ca²⁺ oscillation-independent pathway. When osteoclast precursors are co-cultured with osteoblasts, osteoblasts increase NFATc1 levels in osteoclast precursors, and promote osteoclast differentiation even in the presence of the calcineurin inhibitor FK506. Furthermore, wild-type osteoblasts induce differentiation of osteoclast precursors derived from IP₃R type 2 and type 3 double knockout mice without detectable RANKL-induced Ca²⁺ oscillation[48]. Indeed, Cot (cancer osaka thyroid) serine/threonine kinase, also known as tumor progression locus 2, is activated by cell-cell interaction of osteoclasts with osteoblasts and promotes Ca²⁺ oscillation/calcineurin-independent osteoclastogenesis[52]. Furthermore, Cot increases NFATc1 protein levels through phosphorylation-dependent protein stabilization thereby amplifying NFATc1 activity in the absence of Ca²⁺ oscillation. Cot likely phosphorylates residues that differ from those targeted by calcineurin-mediated dephosphorylation required for nuclear translocation. At present, the identity of osteoblast-derived molecules that activate Cot in osteoclasts is unknown, but Cot-mediated NFATc1 stabilization clearly contributes to osteoclastogenesis in vivo. Collectively, NFATc1 amplification is achieved by both upregulated expression and enhanced stability.

TARGETING PHASE

NFATc1 induction and amplification regulate mRNA levels of target genes driving osteoclast differentiation, fusion and function. While forced NFATc1 expression directs osteoclast differentiation, NFATc1-deficient embryonic stem cells fail to differentiate into osteoclasts following RANKL-stimulation[23,42].

In osteoclast differentiation, the immunoglobulin-like receptor OSCAR, but not TREM-2, is an NFATc1 target gene[53,54]. During differentiation, positive regulators of NFATc1 are enhanced while negative regulators are suppressed. The transcriptional repressor B-lymphocyte-induced maturation protein-1 (Blimp1) is induced by RANKL-stimulation and down-regulates three negative regulators: the v-maf musculoaponeurotic fibrosarcoma oncogene family, protein B; interferon regulatory factor-8; and B cell lymphoma 6. All of these proteins repress Nfatc1 transcription[55-58] (Figure 4). Evidence showing that Blimp1 is a direct NFATc1 target[55] suggests that NFATc1 maintains expression of itself via NFATc1/Blimp1 signaling.

Figure 4 Targeting phase. Negative regulation of nuclear factor of activated T-cells cytoplasmic 1 (NFATc1) is indicated by blue lines and activation of its targets, shown in boxes, is indicated by red arrows. RANK: Receptor activator of nuclear factor-κB; ITAM: Immunoreceptor tyrosine-based activation motif; OSCAR: Osteoclast-associated receptor; Blimp1: B-lymphocyte-induced maturation protein-1; DC-STAMP: Dendritic cell-specific transmembrane protein; TRAP: Tartrate-resistant acid phosphatase; IRF-8: Interferon regulatory factor-8; MafB: V-maf musculoaponeurotic fibrosarcoma oncogene family, protein B; Bcl6: B cell lymphoma 6; CLC: Chloride channel.

NFATc1 target genes encode proteins crucial for osteoclast cell-cell fusion such as a dendritic cell-specific transmembrane protein (DC-STAMP), vacuolar proton pump subunit Atp6v0d2 and the c-Src substrate Tks5 (tyrosine kinase substrate with five SH3 domains)[59-62]. Tks5 appears to be required not only for fusion but for circumferential podosome (actin ring or sealing ring) formation. Following Tks5 knockdown in osteoclasts, multinucleation is abolished although mononuclear osteoclasts still express and amplify NFATc1 in the presence of M-CSF and RANKL[62]. Furthermore, defects of c-Src knockout osteoclasts can be partially rescued by expression of a form of Tks5 carrying glutamate substitutions that mimic constitutive phosphorylation at c-Src phosphorylation target tyrosines[62]. The c-Src-Tks5 axis illustrates an additional signaling pathway induced by RANK signaling beyond NFATc1. In conjunction with ITAM-bearing proteins, c-Src also phosphorylates the tyrosine kinase Syk when integrin α_Vβ₃ is activated by adhesion to bone matrix, in particular, vitronectin[63,64]. In these c-Src-Syk signaling components, integrin β₃ and c-Src are NFATc1 target gene products[65,66], suggesting that NFATc1 target genes include those critical for osteoclast-adhesion.

To resorb bone, osteoclasts secrete acid hydrogen chrolide and various hydrolases. Several NFATc1 target genes encode proteins required for acidification and proteolysis, such as the CLC-7 chloride channel (Clc7)[66], a late endosomal/lysosomal chloride channel localizing in ruffled borders, cathepsin K[67], which degrades collagens, and TRAP[23,39], which dephosphorylates the bone matrix phosphoproteins osteopontin and bone sialoprotein. Mice lacking Clc7 or the V0-ATPase subunit a3 show severe osteopetrosis reminiscent of osteoclast-rich osteopetrosis in humans[68,69]. These mice show TRAP-positive osteoclasts with apparently normal NFATc1 amplification. Expression of the calcitonin receptor depends on NFATc1[23,39,53,70], and calcitonin receptor signaling inhibits both osteoclast formation and function independently of transcriptional regulation by RANK signaling[71].

Finally, NFATc1 induces transcription of ephrinB2[72]. Eph receptors and ephrin ligands are increasingly recognized as important in bone biology[73]. Reverse signaling into ephrinB2-expressing osteoclast lineage cells suppresses osteoclast differentiation by downregulating c-Fos and NFATc1, while forward signaling into receptor EphB4-expressing osteoblast lineage cells enhances osteoblastic differentiation and bone formation. Therefore, ephrinB2 is considered as a coupling factor inducible by RANK signaling[73].

In conclusion, RANK signaling appears to be a straightforward transcriptional cascade of “NF-κB/c-Fos induces NFATc1 induces target genes”. Numerous signaling molecules including receptors, adaptors, kinases and lipases reinforce this cascade. Oscillation of intracellular Ca²⁺ levels drives the cascade, but a Ca²⁺ oscillation-independent mechanism also contributes to amplification of NFATc1 activity. RANK signaling stimulates the cell-cell fusion machinery (specifically, DC-STAMP and Tks5) and activates proteins located on or secreted from the osteoclast ruffled border (CLC-7 and cathepsin K, respectively). Numerous questions remain unanswered about RANK signaling, such as whether and how RANK signaling is connected to microRNA control[74-77] or to long noncoding RNAs (such as competing endogenous RNAs, or ceRNAs)[78]. Components of the RANK signaling pathway will continue to provide not only topics for investigation but novel therapeutic targets to prevent osteoporosis and other bone loss diseases.

ACKNOWLEDGMENTS

We thank Tsukasa Oikawa, Ichiro Takada and Elise Lamar for critical reading of the manuscript.