Receptor Activator of NF-κB (RANK) Ligand Induces Ectodomain Shedding of RANK in Murine RAW264.7 Macrophages

Receptor activator of NF-κB (RANK) is one of the most crucial molecules involved in the differentiation, survival, and activation of osteoclasts (1–4). RANK is a type 1 transmembrane protein and is expressed predominantly in immune cells. Association with its cognate ligand, RANK ligand (RANKL), expressed on osteoblasts and stromal cells, is the key event in osteoclast development. As shown in studies on genetically engineered animals, the absence of either of these two molecules or inhibition of the RANKL–RANK association with an endogenous decoy receptor (osteoprotegerin) results in defective osteoclastogenesis in vivo, which is highlighted by the severe loss of osteoclasts and an osteopetrosis-like phenotype (5–7). Because RANKL is also a membrane-bound protein, osteoclast precursors expressing RANK must make cell–cell contact with osteoblasts and stromal cells through the RANK–RANKL association. Thus, the cell surface expression of these two molecules must be tightly regulated to properly sustain osteoclastogenesis during development and adulthood.

In the past decade, a posttranslational mechanism involving the proteolytic cleavage of membrane-bound proteins, also referred to as ectodomain shedding, emerged as a critical processing mechanism for modifying the functions of cell surface proteins (for reviews on ectodomain shedding, see Refs. 8–11). Recent studies showed that the cell surface availability of RANKL in osteoblasts and stromal cells is regulated, at least in part, by ectodomain shedding and that loss of shedding activity in vivo results in a decrease in bone mass due to increased osteoclastogenesis (12–14). Additionally, the cell surface isoform of CSF-1 (also known as M-CSF) and CSF-1R were shown to be proteolytically cleaved to become soluble (15–17). In contrast, it was unknown whether RANK expressed on osteoclasts and their precursors is subjected to ectodomain shedding.

In this study, we show that the extracellular domain of RANK is proteolytically cleaved and released as a soluble monomeric protein. In addition, we identified TNF-α–converting enzyme (TACE) as a crucial enzyme involved in this shedding activity, as well as the sites in the juxtamembrane domain that are required for efficient cleavage. We also found that the shedding activity is upregulated by rRANKL in macrophage-like cell line RAW264.7, indicating that the cell surface availability of RANK is negatively regulated by the association with its ligand. Furthermore, we found that recombinant soluble RANK, which contains the extracellular domain of RANK N terminus to the putative cleavage sites, suppresses RANKL-induced osteoclastogenesis, and that osteoclastogenesis is accelerated in osteoclast precursors lacking TACE. These observations reveal a previously unknown contribution of ectodomain shedding to the functions of RANK and may provide novel insights into the mechanisms involved in osteoclastogenesis.

TACE-deficient and wild-type mouse embryonic fibroblasts (mEFs) derived from E13.5 embryos were immortalized, as previously described (18, 19). RAW264.7 cells and COS-7 cells were obtained from the RIKEN cell bank. The anti-hemagglutinin (HA) mAbs were from Sigma-Aldrich (HA-7; St. Louis, MO) and Roche Diagnostics (3F10; Indianapolis, IN). Anti-Myc polyclonal Ab was purchased from Medical and Biological Laboratories (Nagoya, Japan). Anti-Myc mAbs were from Abgent (9E10; San Diego, CA) and Medical and Biological Laboratories (PL-14). TNFR-associated factor 6 (TRAF6) inhibitory peptide was from Imgenex (San Diego, CA). The fluorochrome-conjugated Abs and streptavidin used in immunostaining were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). Anti-human placental alkaline phosphatase Ab was from Sigma-Aldrich (8B6). U0126, SP600125, SB202190, and GM6001 were from Calbiochem (San Diego, CA). All other reagents were obtained from Sigma-Aldrich, unless otherwise indicated.

A cloning vector containing murine RANK cDNA was generously provided by Dr. Akira Kudo (Tokyo Institute of Technology) and was used as a PCR template to generate epitope-tagged RANK expression vectors. HA-epitope and Myc/His-epitope dually tagged RANK (RANK^HA-Myc) was generated by inserting an HA-epitope sequence after the signal peptide sequence of the RANK cDNA and by cloning it into the pcDNA4/Myc-His expression vector (Invitrogen, Carlsbad, CA). Alkaline phosphatase (AP)-tagged RANK expression vector (RANK^AP) was generated by cloning RANK cDNA (Lys¹⁷¹-Ala²⁴²) into pAPtag5 (Genhunter, Nashville, TN) (see Fig. 1A for the schema of the constructs). Cell-expression of RANK^HA-Myc and RANK^AP was confirmed by Western blot and immunostaining (Fig. 1B and data not shown). RANK^AP vector was further used as a PCR template to generate cleavage-site mutant RANK expression vectors (Fig. 5B). Mutations in the juxtamembrane region of RANK were introduced by a PCR-based method using a KOD plus Mutagenesis kit (Toyobo, Tokyo, Japan), according to the manufacturer’s instructions.

COS-7 cells and mEFs were grown in DMEM supplemented with 5% FCS and antibiotics. RAW264.7 cells were grown in α-MEM supplemented with 10% FCS and antibiotics. Bone marrow cells collected from the tibiae and femurs of 6–10-wk-old Tace^flox/flox/LysM-Cre⁺ mice (Tace/LysM mice) (19) or littermate control mice were grown in α-MEM with 10% FCS, antibiotics, and 30 ng/ml recombinant mouse M-CSF-1 (WAKO, Osaka, Japan) for 3 d, and then used as bone marrow macrophages (BMMs). The cells were transfected with the expression vectors by using FuGENE HD (Roche Diagnostics), according to the manufacturer’s instructions. Fresh Opti-MEM (Invitrogen) medium, with or without PMA and/or GM6001, was added 18–24 h after transfection, and the cells were incubated for 1 h. The supernatants were collected and cleared by centrifugation to remove cell debris. The activity of AP in the supernatant, which reflects the amount of AP-tagged RANK released from the cell surface, was measured by colorimetry, as described previously (15, 20). The supernatant from nontransfected cells incubated with the substrate for the same amount of time was used as a spectrophotometric blank to offset background AP activity. In-gel visualization of AP activity in the supernatant was also performed, as previously described (20). The cleared samples were concentrated using ConA-Sepharose lectin beads (GE Healthcare Bio-Sciences, Piscataway, NJ) and separated by 10% SDS-PAGE. The AP activity in the gel was visualized with NBT/5-bromo-4-chloro-3-indolyl phosphate (Sigma-Aldrich). All experiments were repeated at least three times in duplicate with similar results.

Peptide cleavage assays were performed, as previously described (21). In brief, a peptide corresponding to the juxtamembrane region of RANK (SDVVCSSSMTLRRPPKEAQAY) was incubated with rTACE (R&D Systems, Minneapolis, MN) for 4–8 h at 37°C, and the cleavage products were analyzed by liquid chromatography MALDI time-off light mass spectrometry (BIFLEX3, Bruker, Billerica, MA).

Cells were lysed with 1% Triton-X 100/PBS containing a protease inhibitor mixture and 10 mM 1,10-phenanthroline (Sigma-Aldrich). The lysates were cleared by centrifugation and incubated with anti-Myc polyclonal Ab for 12 h, and then with protein G-coupled Sepharose beads for 1 h at 4°C with rotary agitation. After separating the lysates by 10% SDS-PAGE, Western blot analysis was performed using anti-HA and anti-Myc mAbs.

COS-7 cells transiently expressing RANK^HA-Myc were incubated at 37°C for 1 h with growth medium in the presence or absence of PMA and/or GM6001. The cells were washed twice with ice-cold PBS, trypsinized, and resuspended in ice-cold 5% BSA/PBS. Subsequently, the cells were incubated with anti-HA Ab at 4°C for 30 min and then washed with ice-cold 5% BSA/PBS. The Abs on the cell surface were detected with a fluorochrome-conjugated secondary Ab. RAW264.7 cells and BMMs were incubated with Opti-MEM containing PMA and/or GM6001 at 37°C for 1 h. The cells were collected using cell dissociation solution (Sigma-Aldrich) and incubated with ice-cold BSA/PBS containing 2 mM EDTA and anti-mouse CD16/32 Ab (clone, 93; BioLegend, San Diego, CA) for 15 min to block nonspecific binding of IgG to the Fc receptor. RAW264.7 cells were further incubated with biotin-labeled anti-mouse RANK Ab (R12-31, BioLegend), and BMMs were incubated with anti-mouse RANK Ab and anti-CD11b Ab (M1/70, BioLegend). The biotin-labeled anti-mouse RANK Ab was detected by allophycocyanin-streptavidin (BioLegend). Biotin-labeled rat IgG2aκ was used as an isotype control. Fluorochrome-labeled cells were analyzed by a laser flow cytometer (FACSCalibur system, BD Biosciences, San Jose, CA) and by FlowJo software (Tree Star, Ashland, OR).

HA-epitope and Myc/His-epitope dually tagged soluble RANK (sRANK^HA-Myc) expression vector was generated by cloning RANK^HA-Myc cDNA (Met¹-Ser¹⁹⁸) into pcDNA4.0 (Fig.7A, left panel). The construct was introduced into COS-7 cells by using FuGENE HD, and the cells were cultured for 2 d. At the end of the incubation, the cells were lysed in lysis buffer, and sRANK^HA-Myc in the lysates was collected using a c-Myc-tagged protein purification kit (Medical and Biological Laboratories), as instructed by the manufacturer. Expression of sRANK^HA-Myc in COS-7 cells was confirmed by Western blot (Fig. 7A, right panel).

BMMs were plated on 48-well plates at 1 × 10⁵ cells/well and incubated with 30 ng/ml CSF-1 and 50 ng/ml rRANKL in the presence or absence of sRANK^HA-Myc (∼50 ng/ml) for 5 d. The cells were stained for tartrate-resistant acid phosphatase (TRAP), and the number of osteoclasts (defined as TRAP-positive multinucleated cells with more than three nuclei) was counted under the microscope.

Total RNA was extracted using an RNeasy Mini Kit (Qiagen, Valencia, CA) and was reverse-transcribed by ReverTra ACE (Toyobo). PCR amplification and quantification were done using SYBR Premix ExTaq II (Takara Shuzo, Otsu, Japan) and the LightCycler Quick System (Roche Diagnostics). Relative mRNA expression levels were obtained by normalizing to β-actin expression. Results are representative of at least three individual experiments.

The Student t test for two samples, assuming equal variances, was used to calculate the p values. The p values <0.05 were considered statistically significant. All data are presented as mean ± SD.

We first investigated whether RANK could be proteolytically cleaved and released into the supernatant by conducting cell-based assays. To facilitate the detection of cleaved RANK, we generated an HA-epitope (inserted between the signaling sequence and extracellular domain) and Myc/6×His (added to the C terminus of the intracellular domain) dually tagged RANK construct (Fig. 1A) and introduced it into COS-7 cells. As shown in Fig. 1B, an ∼85-kDa protein was detected in the lysate from RANK^HA-Myc-transfected COS-7 cells. Cell surface expression of the RANK^HA-Myc protein introduced into COS-7 cells was confirmed by immunostaining (data not shown). We next examined the supernatant from RANK^HA-Myc-transfected COS-7 cells for the presence of cleaved soluble RANK. The supernatants were concentrated with ConA-Sepharose lectin beads and subjected to Western blot analysis with anti-HA Ab, which recognizes the extracellular domain of RANK^HA-Myc. Under unchallenged conditions, almost no protein was found, but an ∼25-kDa band was detected in the supernatant from the cells stimulated with PMA (Fig. 2A). Moreover, the intensity of the band was suppressed by the addition of a broad metalloprotease inhibitor, GM6001, indicating that the proteolytic activity is metalloprotease dependent. Accordingly, when the cell surface expression level of RANK^HA-Myc in COS-7 cells was investigated by flow cytometry, a significant decrease was observed in the cells treated with PMA, whereas treatment with GM6001 abolished the effects of PMA stimulation (Fig. 2B). To confirm these observations, we attempted to determine whether the cleaved stub of RANK^HA-Myc remaining in the cells could be detected in the cell lysates. The lysates from the cells incubated with or without PMA were immunoprecipitated with rabbit anti-Myc sera and probed with murine anti-HA Ab and then with murine anti-Myc Ab. As expected, full-length RANK was found when the sample was analyzed with anti-HA Ab, whereas when the membrane was washed and reprobed with anti-Myc Ab, which recognizes the cytoplasmic end of RANK^HA-Myc, an ∼60-kDa band, in addition to the full-length RANK, was obtained (Fig. 2C). Identical results were obtained with a murine anti-Myc Ab from a different clone (data not shown). Taken together, these observations suggest that RANK^HA-Myc that has been transfected into COS-7 cells can be proteolytically processed and released into the supernatant and that the PMA-stimulated proteolytic activity is metalloprotease dependent.

It was shown that RANK forms a trimer without ligand binding and that the self-association of RANK is mediated through its cytoplasmic tail (22). Based on these observations, we hypothesized that, upon cleavage, the extracellular domain of RANK is released in the supernatant as a monomer protein. To test this hypothesis, we analyzed the cell lysates and supernatants of COS-7 cells transfected with RANK^HA-Myc by Western blot with or without the addition of a reducing agent, DTT. Under nonreducing conditions (without the addition of DTT), as much as half of the RANK^HA-Myc in the cell lysate appeared in the form of a >200-kDa protein, presumably representing a RANK trimer, whereas, under reducing conditions, the >200-kDa band disappeared, and only the 85-kDa band was detected (Fig. 3A). In contrast, the protein species detected in the supernatant were insensitive to DTT and seemed to be identical under nonreducing and reducing conditions (Fig. 3B). These results suggest that RANK cannot form a trimer after it is released from the cell surface and that the cleaved extracellular domain is present in the supernatant in the form of a monomer.

Because the combination of a response to short-term PMA stimulation and the sensitivity to metalloprotease inhibitors is a hallmark of TACE activity (23), we next attempted to determine whether TACE was involved in the ectodomain shedding of RANK. To facilitate the transfection and the detection of cleaved RANK in the supernatant, we used the AP reporter system (20) and generated a truncated RANK construct with an AP module added to its N terminus (RANK^AP, Fig. 1A). We first investigated whether this construct could reproduce the shedding properties of RANK^HA-Myc by introducing RANK^AP into wild-type mEFs and measuring AP activity in the lysates and supernatants by in-gel staining, as described in 1Materials and Methods. We obtained results similar to those observed with RANK^HA-Myc, indicating that RANK^AP has shedding properties similar to those of RANK^HA-Myc (Fig. 4A); therefore, we used this construct for the subsequent studies. To investigate the possible involvement of TACE in the processing of RANK, we next used immortalized mEFs derived from Tace^−/− embryos (19). As shown in Fig. 4B, the PMA-stimulated activity was completely abolished in Tace^−/− mEFs, and reintroduction of TACE fully restored it. Furthermore, we evaluated the cell surface expression of endogenous RANK in BMMs derived from wild-type mice and the mutant mice in which TACE is inactivated in monocytes/macrophages [Tace/LysM mice (19)] by flow cytometry. In wild-type BMMs, there was a significant decrease in the RANK expression levels after PMA treatment, whereas in BMMs derived from Tace/LysM mice the majority of the cell population was unaffected by PMA stimulation and maintained the RANK expression at a comparable level to the cells that were not treated with PMA (Fig. 4C). Taken together, these observations indicate that TACE is primarily responsible for the PMA-stimulated cleavage of RANK in osteoclast precursors.

To identify the potential RANK cleavage site(s) in more detail and determine whether TACE directly cleaves RANK, we synthesized a peptide covering the juxtamembrane domain of RANK (Ser¹⁹¹-Tyr²¹¹) and incubated it with rTACE. Based on the results of the mass spectrometry analysis, we concluded that rTACE directly cleaved the peptide and deduced two potential cleavage sites in the juxtamembrane domain of RANK: a major site between Thr²⁰⁰-Leu²⁰¹ and a minor site between Met¹⁹⁹-Thr²⁰⁰ (Fig. 5A and data not shown). We next generated several mutant RANK^AP constructs (Fig. 5A, Mut1–6) and investigated how the mutations in the putative cleavage sites affected the cleavage efficiency. Cell surface expression of the mutants (Mut1–6) was confirmed by immunostaining and confocal microscopic analysis (data not shown). As shown in Fig. 5B, when Met¹⁹⁹-Thr²⁰⁰-Leu²⁰¹ was replaced with Ile-Ser-Pro (Mut1) or with Ile-Thr-Pro (Mut5), there was a significant decrease in the AP activity in the supernatant. Although cleavage of these mutants was still increased by PMA stimulation, overall shedding efficiency was significantly diminished. In contrast, the shedding efficiency of Mut2–4 was comparable to that of the wild-type control. The deletion mutant (Mut6), in which Met¹⁹⁹-Thr²⁰⁰-Leu²⁰¹ had been removed, was as resistant to shedding as Mut1 and Mut5. Taken together, these observations suggest that, although RANK shedding by TACE is not absolutely dependent on the Met¹⁹⁹-Thr²⁰⁰-Leu²⁰¹ sequence in the juxtamembrane domain, it is required for efficient cleavage.

To extend our findings, we next investigated whether RANK is cleaved in a macrophage-like cell line (RAW264.7) that expresses endogenous RANK and differentiates into multinucleated osteoclasts in response to RANKL stimulation. Expression of TACE in this cell line was confirmed by Western blot (data not shown). RAW264.7 cells transfected with RANK^AP were incubated with PMA and/or GM6001, and the supernatant was analyzed for AP activity by colorimetry, as described in 1Materials and Methods. As shown in Fig. 6A, the results were comparable to those observed in wild-type mEFs (Fig. 4A), indicating that RANK is similarly processed in RAW264.7 cells. Based on an analogy to the induction of the shedding of vascular endothelial growth factor receptor by its ligand, vascular endothelial growth factor-A (24), we next investigated whether RANKL could trigger the cleavage of RANK. As shown in Fig. 6A, RANK shedding was upregulated by rRANKL in RAW264.7 cells to a level comparable to that induced by PMA, and the increased activity was abolished by the addition of GM6001. It has been well established that following ligand binding, RANK activates various signaling pathways essential for osteoclast development and that the initial step in the signaling involves binding of TRAF6 to the cytoplasmic domain of RANK. To explore whether RANKL-induced RANK shedding was TRAF6 dependent, we used a TRAF6 inhibitory peptide, which functions as a TRAF6 decoy by binding to the TRAF6-binding motif of RANK (25), and examined how it would affect RANKL-induced RANK shedding activity. When RANK^AP-transfected RAW264.7 cells were incubated in the presence of TRAF6 inhibitory peptide, it had a dose-dependent inhibitory effect on RANKL-induced shedding activity (Fig. 6B). The inhibitory effect was specific to TRAF6, because PMA-induced shedding activity, which does not involve TRAF6 activation, was insensitive to the inhibitory peptide at a concentration at which RANKL-induced shedding was completely abolished (Fig. 6C). We also found that RANKL-induced shedding was sensitive to U0126 (MEK inhibitor) and SB202190 (p38 inhibitor) but not to SP600125 (JNK inhibitor), indicating that RANKL-induced shedding activity is TRAF6 and MAPK dependent (Fig. 6D). To further confirm the observation that RANK shedding can be induced by RANKL stimulation, we examined the cell surface expression of endogenous RANK by flow cytometry. As shown in Fig. 6E, incubation with PMA or RANKL significantly decreased the cell surface expression of RANK in RAW264.7 cells, and this activity was abolished by the addition of GM6001.

To gain insight on the physiological relevance of RANK shedding, we generated recombinant soluble RANK composed of the extracellular region of RANK N terminus to the putative cleavage site (Met¹-Ser¹⁹⁸, Fig. 7A), and examined how the cleaved extracellular domain of RANK would affect RANKL-induced osteoclastogenesis in vitro. RAW264.7 cells were incubated with rRANKL in the presence or absence of sRANK^HA-Myc for 5 d, and the number of TRAP-positive multinucleated cells was evaluated. As shown in Fig. 7B, the addition of sRANK^HA-Myc almost completely abolished RANKL-induced osteoclastogenesis, indicating that cleaved soluble RANK functions as an antagonist to RANKL in a similar manner to osteoprotegerin, an endogenous decoy receptor for RANKL. Furthermore, we found that TACE-deficient BMMs derived from Tace/LysM mice exhibited increased levels of NFATc1, a transcription factor essential for osteoclastogenesis, after RANKL stimulation (Fig. 7C) and that RANKL-induced osteoclastogenesis is accelerated compared with the wild-type control (Fig. 7D). These observations indicate that RANK shedding functions as a negative regulator for osteoclastogenesis, presumably via the production of soluble RANK and by decreasing cell surface RANK availability in osteoclast precursors.

Several membrane-bound molecules involved in osteoclast development, including RANKL (12), CSF-1R (16), TNF-α (26), and the membrane-bound isoform of CSF-1 (15), are proteolytically cleaved to become soluble. The current study shows for the first time that RANK, a molecule essential for osteoclastogenesis, also undergoes ectodomain shedding and is released as a monomeric protein. We also found by means of cell-based assays, using TACE-deficient cells, that TACE is involved in RANK shedding activity and that this activity can be upregulated by RANKL in macrophage-like cell line RAW264.7. These results indicate a potential involvement of ectodomain shedding of RANK in the regulation of osteoclastogenesis.

Consistent with the results of a previous study (22), RANK^HA-Myc introduced into COS-7 cells self-assembled in the absence of RANKL (Fig. 3A), whereas the cleaved extracellular domain of RANK released in the supernatant was in the form of a monomeric protein, not as a trimer, indicating that the cytoplasmic domain or juxtamembrane domain is required for the self-association of RANK and that once cleaved from the cell surface, RANK is no longer able to form a trimer. In contrast, it is possible that RANK is resistant to ectodomain shedding when it is in the trimetric conformation and that only the unassociated monomer proteins are cleaved from the cell surface.

The consequences of ectodomain shedding of receptors can be complex in some cases. Ectodomain shedding of receptors can result in downregulation of receptor availability on the cell surface and can simultaneously give rise to soluble receptors, which have a potential to function as decoy receptors and to interfere with ligand–receptor association. Moreover, as in the case of CD44 and Notch (27–29), ectodomain shedding of receptors may be necessary to trigger intramembranous cleavage by presenilin and the release of the cytoplasmic domain into the cytoplasm. For example, it is well established that cleaved TNFR1 and TNFR2 are capable of binding to TNF-α and suppressing TNF-α–TNFR signaling and that dysregulation of the shedding of these receptors results in aberrant immunoreactions (30–32). Similarly, recombinant soluble RANK (Glu³⁰-Pro²¹³) and RANK-Fc were demonstrated to inhibit RANKL-induced osteoclastogenesis in vitro and to prevent the formation of metastatic bone lesions in xenograft tumor models (33–36). We also generated soluble RANK composed of the extracellular domain N terminus to the putative cleavage sites and observed a similar inhibitory effect on RANKL-induced osteoclastogenesis in vitro. These observations suggest that the cleaved extracellular domain of RANK is biologically active and potentially functions as a decoy receptor that blocks RANKL binding in a manner similar to that of osteoprotegerin. Because osteoprotegerin can bind not only to RANKL but also to the TNF-related apoptosis-inducing ligand (TRAIL/Apo2 ligand) (37), soluble RANK may function as a more specific inhibitor of RANKL.

Based on the observations that cell surface expression of RANK decreases following RANKL stimulation, it is tempting to hypothesize a negative feedback loop in which RANKL induces RANK shedding and, thereby, downregulates RANKL–RANK signaling during osteoclast development. In accordance, we found that TACE-deficient BMMs form multinucleated cells more rapidly than did the control BMMs and that the expression of NFATc1 is upregulated in TACE-deficient BMMs, indicating that RANK shedding negatively regulates osteoclastogenesis. However, in contrast, no overt bone defects were found in Tace/LysM mice, in which TACE is inactivated in monocytes and macrophages (19), at least under unchallenged conditions (data not shown). Thus, other enzyme(s) may compensate for the lack of TACE in the cleavage of RANK, or the regulation of cell surface RANK availability may also be regulated by different mechanisms that do not involve ectodomain shedding (e.g., via endocytosis). Furthermore, it is also possible that the lack of TACE leads to an increase in the amount of other cell surface proteins, which may function in an antiosteoclastogenic manner, and offsets the effects of enhanced RANK signaling. Further research, including a generation of uncleavable mutant RANK knock-in mice, is needed to elucidate the significance of RANK shedding and its impact on osteoclastogenesis.

In conclusion, the current study demonstrated that RANK is subjected to ectodomain shedding by TACE and identified potential cleavage sites in the juxtamembrane domain of RANK. The cleavage of RANK should decrease its availability on osteoclasts and their precursors and simultaneously generate soluble decoy receptors that may inhibit the RANKL–RANK association. Moreover, the observation that RANKL–RANK signaling induces RANK shedding suggests a possible negative-feedback mechanism regulating the cell surface availability of RANK on osteoclasts and their precursors. Therefore, upregulation of RANK shedding via TACE activation may be beneficial in suppressing overt osteoclastogenesis under pathological conditions, including rheumatoid arthritis, bone metastasis, and osteoporosis.

We thank Dr. Akira Kudo for murine RANK cDNA and Shizue Tomita for excellent technical assistance.

Disclosures The authors have no financial conflicts of interest.