Annexin II Stimulates RANKL Expression Through MAPK



We report that AX-II, in addition to inducing GM-CSF expression, also increases membrane-bound RANKL synthesis by marrow stromal cells and does so through a previously unreported MAPK-dependent pathway. Thus, both GM-CSF and RANKL are required for AX-II stimulation of OCL formation.

Introduction: Annexin II (AX-II) is an autocrine/paracrine factor secreted by osteoclasts (OCLs) that stimulates human OCL formation and bone resorption in vitro by inducing bone marrow stromal cells and activated CD4+ T cells to produce granulocyte-macrophage colony-stimulating factor (GM-CSF). GM-CSF in turn increases OCL precursor proliferation and further enhances OCL formation. However, the induction of GM-CSF by AX-II cannot fully explain its effects on OCL formation. In this study, we tested the capacity of AX-II to induce the expression of RANKL and the corresponding signaling pathways AX-II employs in human marrow stromal cells to induce RANKL. We also showed that both GM-CSF and RANKL are required for OCL formation induced by AX-II.

Materials and Methods: Real-time RT-PCR and Western blot analysis were used to detect RANKL and osteoprotegerin (OPG) mRNA and protein expression in unfractionated human bone marrow mononuclear cells stimulated with AX-II. Soluble RANKL in the conditioned medium was analyzed by ELISA. Activation of the MAPK pathway by AX-II was tested by Western blot. The effects of OPG and anti-GM-CSF on AX-II-induced OCL formation were also examined.

Results and Conclusion: In addition to upregulating GM-CSF mRNA, AX-II increased RANKL mRNA expression dose-dependently in unfractionated human bone marrow mononuclear cells and modestly increased soluble RANKL in unfractionated human bone marrow mononuclear cell conditioned medium. However, AX-II markedly increased membrane-bound RANKL on human bone marrow stromal cells. Treatment of marrow stromal cells with AX-II activated MAP-kinase (ERKs) and PD 98059 abolished the effect but did not block the increase in GM-CSF. Interestingly, OPG, a natural decoy receptor for RANKL, or anti-GM-CSF partially inhibited OCL formation by AX-II in human bone marrow cells, and the combination of OPG and anti-GM-CSF completely blocked AX-II-induced OCL formation. These data show that AX-II stimulates both the proliferation and differentiation of OCL precursors through production of GM-CSF and RANKL respectively.


MULTIPLE FACTORS REGULATE osteoclast (OCL) differentiation from hemopoietic cells of the monocyte/macrophage lineage. These factors can be systemic [e.g., 1,25(OH)2D3], calcitonin and parathyroid hormone, or can be produced locally, such as interleukin-6 (IL-6) and RANKL, by cells in the bone microenvironment such as osteoblasts, stromal cells, and immune cells,(1) and act at different stages of OCL differentiation.

There are two major stages in osteoclastogenesis. In the first stage, the proliferation and differentiation of immature OCL precursors occurs, which is stimulated by factors such as IL-6 and granulocyte-macrophage colony-stimulating factor (GM-CSF) that increase the OCL progenitor pool for subsequent OCL formation.(1) These immature precursors differentiate to become committed postmitotic OCL precursors that then fuse to form OCL. RANKL acts at the second stage of OCL formation(2) and is expressed on osteoblasts/stromal cells. It can be upregulated by bone resorbing factors such as glucocorticoids, 1,2-(OH)2D3, IL-1, IL-6, IL-11, IL-17, TNF-α, prostaglandin E2 (PGE2), or PTH.(3) OCL precursors express RANK, a receptor for RANKL that binds RANKL through cell-to-cell contact with osteoblasts/stromal cells, and differentiate into OCLs.(4)

Annexin II (AX-II) is a member of the annexin family of homologous proteins. Similar to other annexins, the C-terminal half of AX-II contains four repeats of 70-80 amino acids and an annexin consensus sequence, which is responsible for the Ca2+-dependent binding of the protein to phospholipids.(5) In contrast, the N-terminal region of the annexins is highly variable and may contribute to the specific functions of the different annexins. AX-II, -V, and -VI can form Ca2+ channels in phospholipid bilayers,(6) and we have identified a putative surface AX-II receptor on marrow stromal cells and CD4+ T cells. Several functions for AX-II have been documented recently. Intracellular AX-II is associated with membrane organization, membrane trafficking, membrane-cytoskeleton linkage, and ion conductance across membranes.(6) Extracellular AX-II has recently been identified as a potential receptor for a number of polypeptide ligands, such as tenascin-C, plasminogen, human cytomegalovirus,(7) and 1,25(OH)2D3.(8)

Our group first identified AX-II as an OCL-secreted factor that stimulates human and murine OCL formation. AX-II enhances the stimulatory effects of suboptimal concentrations of 1,25(OH)2D3 on OCL formation in marrow cultures and on bone resorption in long bones of fetal rats.(9) We further showed that AX-II induces proliferation of human OCL precursors by stimulating GM-CSF production by activated T cells and marrow stromal cells(10) through binding to a putative receptor. Although GM-CSF is one of the primary factors that modulate OCL formation induced by AX-II, these data do not fully explain the effects of AX-II on OCL formation, because time-course studies suggested that AX-II also affected the final stages of differentiation of OCL precursors.(10) Thus, in this study, we tested the capacity of AX-II to induce the expression of RANKL, an essential factor for OCL formation that acts as the later stage of OCL formation by human marrow cells. We determined the molecular signaling mechanisms by which AX-II induces RANKL expression and determined if either induction of GM-CSF or RANKL or both were required for AX-II to induce OCL formation.



The 23c6 mAb that identifies human OCLs was generously provided by Dr Michael Horton (St Bartholomew's Hospital, London, UK). Highly purified bovine lung AX-II was purchased from Biodesign International (Kennebunk, ME, USA). Recombinant human osteoprotegerin (OPG)/TNFRSR11B, RANKL, and anti-GM-CSF were purchased from R&D Systems (Minneapolis, MN, USA); the soluble RANKL (sRANKL) ELISA was from Biomedica group (Windham, NH, USA). RANKL polyclonal antibody and actin antibody were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Phospho-p44/42 MAPK antibody and p44/42 MAPK antibody were obtained from Cell Signaling Technology (Beverly, MA, USA), and the MAPK inhibitor, PD 98059, was purchased from Calbiochem (San Diego, CA, USA). RNA-Bee RNA isolation solvent was purchased from Tel-test (Friendswood, TX, USA). Random hexamer, human RANKL, human 18S primers, and probes were synthesized by Integrated DNA Technologies (Coralville, IA, USA). All reagents used in the reverse transcription (RT) and DNaseI treatment were purchased from Invitrogen (Carlsbad, CA, USA). TaqMan universal PCR master mix and MicroAmp optical 96-well reaction plates for real-time PCR were from Applied Biosystems (Foster City, CA, USA).

Human bone marrow mononuclear cell cultures and isolation of human bone marrow stromal cells from human bone marrow mononuclear cells

After obtaining informed consent, 2 ml of bone marrow was aspirated from the posterior superior iliac crest of healthy normal volunteers into a 20-ml syringe containing 1 ml α-MEM and 1000 U/ml of preservative-free heparin. The bone marrow was processed as described previously.(11) Briefly, the mononuclear cell fraction was obtained by density gradient centrifugation over Ficoll-Hypaque (Sigma, St Louis, MI, USA), and the mononuclear cells were incubated in α-MEM/20% FCS at 5 × 106 cells/ml overnight at 37°C in 100-mm tissue culture plates (Fisher Scientific, Houston, TX, USA) to separate nonadherent and adherent cells. The nonadherent cells were collected and used in cultures as the source of immature OCL precursors for formation of OCL-like multinucleated cells as described below.

The adherent marrow cells were cultured in α-MEM/10% FCS until they were confluent.(12) These cells were used as the source of human bone marrow stromal cells (hBMCs) and did not contain detectable hematopoietic cells. Only cells from the first three passages were used for detecting RANKL and GM-CSF mRNA and protein expression.

OCL formation

OCL-like multinucleated cells were formed by culturing 1 × 105 nonadherent bone marrow mononuclear cells in 0.1 ml of α-MEM/20% horse serum for 3 weeks in 96-well plates (Fisher Scientific). One-half the media was changed twice weekly. Cells were incubated with 100 ng/ml AX-II, 200 ng/ml OPG, media alone, or 30 ng/ml RANKL as a positive control for 3 weeks. After 3 weeks, the cultures were fixed and stained with the 23c6 anti-vitronectin receptor mAb, which identifies multinucleated cells that express calcitonin receptors and form resorption lacunae on calcified matrices, as described previously.(13) Cells that contained three or more nuclei and reacted with the 23c6 antibody were scored as OCL-like multinucleated cells. OCLs were scored from three independent experiments.

Detecting soluble RANKL in the conditioned media by ELISA

Unfractionated hBMCs, which contain both hematopoietic and stromal components (1 × 105) and were obtained by density gradient centrifugation over Ficoll-Hypaque, but not subsequently depleted of adherent cells, were treated with different concentrations of AX-II for 1 week, and the conditioned media were collected. One hundred microliters of conditioned media was assayed using an sRANKL ELISA kit as described in the manufacturer's manual. Optical density was read at 450- and 690-nm wavelengths with a microplate reader. Results were calculated using the standard curves created in each assay. Concentrations were given as picomolar.

Western blot analysis

hBMCs were incubated in lysis buffer (20 mM Tris [pH 7.5], 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 1:1000 proteinase inhibitor) for 10 minutes on ice, and the cells were sonicated until the sonicate appeared clear. Cell lysates were separated by SDS-PAGE and transferred to nitrocellulose membrane. Nonspecific binding sites were saturated by incubating the membranes in blocking buffer (5% nonfat milk in PBS and 0.5% Tween-20) for 1 h at room temperature. Membranes were incubated with rabbit polyclonal anti-RANKL (1:500) and anti-Phospho-p44/42 MAPK (1:2000) overnight. After washing, the anti-rabbit secondary antibody conjugated to horseradish peroxidase was added in blocking solution and incubated for 1 h at room temperature. Specific bands were visualized by chemiluminescence. The membrane was stripped and probed with goat polyclonal anti-actin (1:1000) and rabbit polyclonal anti-p44/42 MAPK (1:2000) separately, and the actin bands were used as an internal control for the loading.

Quantitative real-time RT-PCR and RT-PCR

Total RNA was extracted with RNA-Bee reagent according to the manufacturer's protocol.(14) The RNA was quantified spectrophotometrically. The RNA was treated with DNaseI at 37°C for 30 minutes and followed by DNaseI inactivator for 2 minutes at room temperature to remove DNA contamination. Two micrograms of total RNA was applied to the 20-μl RT reaction with random hexamer primers. The basis for PCR quantitation in the ABI7700 (TaqMan; Applied Biosystems) instrument is to continuously measure PCR product accumulation using a dual-labeled fluorogenic oligonucleotide probe.

Primers and probes were designed for human RANKL and human 18S using the primer express program (Applied Biosystems). The sequences of the forward primer, reverse primer, probe, and gene bank accession number for each gene are shown in Table 1. Each reaction (20 μl) contained 5 μl template and was performed in duplicate. 18S was used in a similar manner to quantitate the presence of 18S mRNA as an endogenous mRNA control in the samples. Each sample was normalized on the basis of its 18S mRNA content. The universal thermal cycling program for real-time PCR consisted of 40 cycles of 95°C for 15 s and 60°C for 60 s The results are expressed as the percentage of control (mean ± SE). GM-CSF mRNA was measured by RT-PCR. The primers for amplifying GM-CSF were 5′-AGCCCTGGGAGCATGTGAAT-3′ (forward) and 5′-GTTGCACAGGAAGTTTCCGG-3′ (reverse). The primers for GAPDH were 5′-ACCACAGTCCATGCCATCAC-3′ (forward) and 5′-TCCACCACCCTGTTGCTGTA-3′ (reverse). The parameters for amplifying GM-CSF and GAPDH were 94°C for 1 minute, 60°C for 1 minute for 30 cycles, and 60°C for 7 minutes as the final elongation step.

Table Table 1.. Taqman Primer and Probe Sequences
original image

Statistical analysis

Data are expressed as the mean ± SE for three independent experiments and were compared using the two-tailed t-test. Statistical significance was assigned at p < 0.05.


RANKL mRNA expression is upregulated by AX-II in hBMCs

To determine if AX-II could induce RANKL expression in hBMCs, we measured RANKL expression at both the mRNA and protein levels in unfractionated marrow mononuclear cells treated with AX-II. As shown in Fig. 1A, quantitative real-time PCR showed that RANKL mRNA expression was increased 1.8-fold (1.8 ± 0.4, p < 0.05) over the control when bone marrow cells were treated with 250 ng/ml AX-II for 24 h, 2.1-fold (2.1 ± 0.2, p < 0.05) at 48 h, and 1.5-fold (1.5 ± 0.6, p < 0.05) at 72 h. The increase in RANKL mRNA levels was significantly different at 24, 48, and 72 h compared with the control. AX-II increased RANKL mRNA expression in a dose-dependent fashion. RANKL mRNA expression was increased 1.9- (1.9 ± 0.4, p < 0.05), 2.5- (2.5 ± 0.4, p < 0.05), and 3.8-fold (3.8 ± 0.2, p < 0.05) at 24 h compared with the control (p < 0.05) by 100, 250, and 500 ng/ml AX-II, respectively (Fig. 1B).

Figure FIG. 1..

AX-II Induces RANKL mRNA in unfractionated hBMCs. (A) Time-course of RANKL mRNA expression induced by AX-II. Approximately 1 × 106 primary hBMCs were treated with 250 ng/ml AX-II for 12, 24, 48, and 72 h. RANKL mRNA expression was detected by quantitative real-time PCR assay. (B) AX-II dose-dependently increased RANKL mRNA levels. hBMCs (1 × 106) were treated with 100, 250, and 500 ng/ml AX-II for 24 h. RANKL mRNA expression was examined by quantitative real-time RT-PCR assay. (C) AX-II induces unfractionated hBMCs to secrete soluble RANKL. After 1 week of incubation with 50, 100, and 250 ng/ml AX-II, the conditioned medium in the unfractionated hBMC (1 × 105) culture was collected to test the soluble RANKL expression by ELISA assays. *p < 0.05 compared with control culture; #p < 0.05 compared with 500 ng/ml AX-II. Data are expressed as mean ± SE for three independent experiments.

There are two forms of RANKL: a soluble form mainly produced by nonadherent cells (e.g., T cells) and a membrane-bound form primarily expressed on osteoblast/stromal cells.(15) To determine which form of RANKL was produced after exposure to AX-II, we performed ELISA assays to detect sRANKL in unfractionated hBMC cultures treated with varying concentrations of AX-II. As shown in Fig. 1C, sRANKL modestly increased by 7% and 9%, respectively, when treated with 50 and 100 ng/ml AX-II for 1 week, but returned to basal level when AX-II was increased to 250 ng/ml. OPG in RNA expression was not consistently decreased by AX-II (data not shown).

Increased RANKL protein expression by AX-II on hBMCs

We then examined membrane-bound RANKL expression on the marrow stromal cells by Western blot assay. Although real-time PCR assay clearly showed a significant difference in RANKL mRNA expression (see Fig. 1A) with 24 h of AX-II treatment, RANKL protein expression was not significantly changed until 48 h. Because of the delay in the change of RANKL protein compared with mRNA expression, the marrow stromal cells were treated with different doses of AX-II for 48 h. As shown in Fig. 2, at 100 ng/ml AX-II, RANKL protein expression was 1.2-fold of the control; AX-II significantly stimulated RANKL protein level by 2.0-fold at 250 ng/ml and 1.8-fold at 500 ng/ml. sRANKL was not detected by ELISA in media conditioned by marrow stromal cells (data not shown).

Figure FIG. 2..

AX-II dose-dependently increases RANKL protein expression by hBMCs. The primary hBMCs (1 × 106) were treated with 100, 250, and 500 ng/ml AX-II for 48 h, and RANKL protein was analyzed by Western blot assay. Actin was used as an internal control for loading. Data are expressed as mean ± SE for three independent experiments.

MAPK activation by AX-II in hBMCs

Activation of the MAPK pathway is one of the major routes leading to cell survival and/or proliferation.(16) Previous studies have shown that AX-II can act as an autocrine/paracrine factor on OCL formation(10,11) that AX-II induces GM-CSF.(10) Therefore, we examined the effects of AX-II on extracellular signal-regulated kinase (ERK) activation to assess if the MAPK pathway is involved in RANKL induction by AX-II. ERK phosphorylation was elevated rapidly within 5 minutes after AX-II treatment and further increased at the15- and 30-minute time-points (Fig. 3).

Figure FIG. 3..

AX-II activates p42/44 MAPK in hBMCs. Primary hBMCs (1 × 106) were cultured in α-MEM medium without FCS for 2 h. After serum starvation, the cells were treated with 250 ng/ml AX-II for 5, 15, and 30 minutes. The cells were extracted for protein and analyzed for phospho-MAPK(p42/44) with a phospho-specific antibody by Western blot analysis. MAPK was used as an internal control for loading.

Effect of PD 98059, an inhibitor of MAPK, on RANKL expression in hBMCs

To further confirm that the MAPK pathway participates in the induction of RANKL by AX-II, we tested the effect of PD 98059 on RANKL induction by AX-II. At 20 μM, PD 98059 partially inhibited RANKL mRNA expression (1.47 ± 0.04, p < 0.05) compared with AX-II alone. At 40 μM, PD 98059 completely suppressed RANKL transcripts to the control levels (0.91 ± 0.21, p < 0.05; Fig. 4A). Similarly preincubation of cells with PD 98059 followed by treatment with AX-II neutralized the induction of RANKL protein by 250 ng/ml AX-II after 48 h (Fig. 4B). As previously reported, AX-II increased GM-CSF mRNA expression. However, PD 98059 did not block the upregulation of GM-CSF (Fig. 4C).

Figure FIG. 4..

The MAPK inhibitor PD 98059 inhibits RANKL expression. (A) PD 98059 blocks AX-II-induced RANKL mRNA expression in hBMCs. Primary hBMCs (1 × 106) were preincubated with 20 and 40 μM PD 98059 for 30 minutes and treated with 0 or 250 ng/ml AX-II for 24 h. RANKL mRNA expression was detected by quantitative real-time PCR. (B) PD 98059 inhibited RANKL protein expression induced by AX-II in hBMCs. hBMCs (1 × 106) were preincubated with 20, 40, and 60 μM of PD 98059 for 30 minutes and treated with 0 or 250 ng/ml of AX-II for 48 h. The cells were collected for Western blot analysis. Actin was used as an internal control for loading. *p < 0.05 compared with control; #p < 0.05 compared with 250 ng/ml AX-II. (C) hBMCs (1 × 106) were preincubated with or without 40 μM PD 98059 for 30 minutes and treated 0 or 250 ng/ml AX-II for 24 h. GM-CSF mRNA was detected by RT-PCR. GAPDH mRNA expression was used as the control for equal loading. Data are expressed as the mean ± SE for three independent experiments.

Effect of OPG on osteoclastogenesis induced by AX-II

OPG is a natural soluble decoy receptor of RANKL, which is involved in the negative regulation of OCL formation.(17) As shown in Fig. 5, 200 ng/ml OPG suppressed 71% OCL formation induced by AX-II in human bone marrow culture, whereas 1 μg/ml anti-GM-CSF only blocked OCL formation by 38%. When OPG and anti-GM-CSF were combined, AX-II-induced OCL formation was totally inhibited (Fig. 5).

Figure FIG. 5..

Inhibition effect of OPG on OCL formation by AX-II. Nonadherent hBMCs (1 × 105) were cultured in 0.2 ml of α-MEM/20% horse serum for 3 weeks with 100 ng/ml AX-II, 200 ng/ml OPG, RANKL (30 ng/ml), and anti-GM-CSF (1 μg/ml). After 3 weeks, the cultures were fixed and stained. Cells that contained three or more nuclei and reacted with the 23c6 antibody were scored as OCL-like multinucleated cells. OCLs were scored from three independent experiments. *p < 0.05 compared with control; #p < 0.05 compared with 100 ng/ml AX-II alone. Data are expressed as the mean ± SE for three independent experiments.


We previously found that AX-II is an autocrine/paracrine factor produced by OCLs that stimulates OCL formation and showed that it does so by binding a putative receptor on marrow stromal cells and inducing GM-CSF production.(10) However, GM-CSF primarily acts at the proliferative stage of OCL precursor differentiation and is not a potent inducer of mature OCLs.(18) Therefore, we examined the capacity of AX-II to induce RANKL expression, because many factors that act on the later stages of OCL differentiation, such as 1,25(OH)2D3 and PGE2, do so indirectly by inducing RANKL(3)

In this report, we showed that AX-II increased RANKL expression in adherent hBMCs. AX-II primarily increased membrane-bound RANKL rather than sRANKL in cell cultures, suggesting that the sRANKL we detected was most likely derived by cleavage of the full-length form or the 40- to 45-kDa cellular membrane-bound form on the surface of stromal cells or produced by T cells.(15) The membrane-bound RANKL was increased 2-fold on the stromal cells, whereas sRANKL expression was significantly but only slightly enhanced by AX-II. These results show that AX-II is similar to osteotropic hormones and factors, such as 1,25(OH)2D3, PTH, PGE2, and IL-11.

AX-II can stimulate the expression of membrane-bound RANKL on osteoblasts/stromal cells. OPG partially blocked OCL formation induced by AX-II, confirming that OCL formation induced AX-II is caused in part by induction of RANKL expression on hBMCs.

The MAPK cascade is an important signaling system shared by various types of mammalian cells. The ERK cascade is stimulated in response to signals from tyrosine kinases, hematopoietic growth factor receptors, and some heterotrimeric G protein-coupled receptors and seems to mediate signals promoting cell proliferation, differentiation, and survival.(19) Our results show that the ERK pathway is involved in RANKL expression induced by AX-II, and the MAPK pathway inhibitor, PD 98059, abolished AX-II-induced RANKL expression. We further showed that AX-II does not use this pathway to induce GM-CSF. There are four other independent signaling pathways documented that regulate RANKL expression in osteoblasts/stromal cells. These include the vitamin D receptor pathway [1,25(OH)2D3], the protein kinase A pathway (PTH and PGE2), the gp130 pathway (IL-6 and IL-11), and the calcium/protein kinase C pathway (high extracellular Ca2+).(20–23) We now report that RANKL expression is also induced by activation of MAPK.

These data suggest that AX-II binds its receptor on marrow stromal cells, and this receptor signals through the MAPK pathway to induce RANKL but uses another pathway to induce GM-CSF. Both GM-CSF and RANKL contribute to the stimulation effects of AX-II on OCL formation.


This work was supported by Research Funds from National Institutes of Health RO1 AG13625 (04/01/2001 to 03/31/2006). The authors thank Judy Anderson for preparation of the human mononuclear marrow cells and other technical support. We also thank Donna Gaspich and Theresa Casciato for preparing the manuscript.

Dr Roodman serves as a consultant for Salmedix and SCIOS, Inc. All other authors have no conflict of interest.