Department of Pharmacology, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada
Correspondence: Christopher J. Sinal, Ph.D., Department of Pharmacology, Dalhousie University, 5850 College Street, Box 15000, Halifax, Nova Scotia, Canada B3H4R2. Telephone: 902-494-2347; e-mail: firstname.lastname@example.org
Author contributions: S.M., H.D., and J.R.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; N.M.: collection and/or assembly of data, data analysis and interpretation, and final approval of manuscript; C.S.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Bone is a dynamic tissue that is continuously remodeled through the action of formative osteoblasts and resorptive osteoclasts. Chemerin is a secreted protein that activates chemokine-like receptor 1 (CMKLR1), a G protein-coupled receptor expressed by various cell types including adipocytes, osteoblasts, mesenchymal stem cells (MSCs), and macrophages. Previously, we identified chemerin as a regulator of adipocyte and osteoblast differentiation of MSCs. Herein we examined the role of chemerin in Lin− Sca1+ c-kit+ CD34+ hematopoietic stem cell (HSC) osteoclastogenesis. We found that HSCs expressed both chemerin and CMKLR1 mRNA and secreted chemerin protein into the extracellular media. Neutralization of chemerin with a blocking antibody beginning prior to inducing osteoclast differentiation resulted in a near complete loss of osteoclastogenesis as evidenced by reduced marker gene expression and matrix resorption. This effect was conserved in an independent model of RAW264.7 cell osteoclastogenesis. Reintroduction of chemerin by reversal of neutralization rescued osteoclast differentiation indicating that chemerin signaling is essential to permit HSC differentiation into osteoclasts but following blockade the cells maintained the potential to differentiate into osteoclasts. Mechanistically, neutralization of chemerin blunted the early receptor activator of nuclear factor-kappa B ligand induction of nuclear factor of activated T-cells 2 (NFAT2), Fos, Itgb3, and Src associated with preosteoclast formation. Consistent with a central role for NFAT2, induction or activation of NFAT2 by forced expression or stimulation of intracellular calcium release rescued the impairment of HSC osteoclastogenesis caused by chemerin neutralization. Taken together, these data support a novel autocrine/paracrine role for chemerin in regulating osteoclast differentiation of HSCs through modulating intracellular calcium and NFAT2 expression/activation. Stem Cells2013;31:2172–2182
Constant remodeling of bone is essential for skeletal integrity as well as other vital aspects of physiology such as calcium and vitamin D homeostasis [1-3]. Osteoporosis and other bone loss disorders typically result from a dysregulation of bone turnover such that bone resorption exceeds bone formation leading to a deterioration of bone microarchitecture and reduction of bone mass and volume . Loss of the normal balance of bone formation/resorption is most commonly associated with aging , hormonal changes in postmenopausal women , and pathogenic disorders such as diabetes mellitus . Given that osteoporotic fractures are common (>9 million per year globally) and inflict a substantial economic burden in terms of lost productivity and health care expenditures , there is a present and growing need for more effective therapeutics to treat disorders of bone loss.
In healthy individuals, coordinated development of bone forming osteoblasts and bone resorbing osteoclasts governs a homeostatic balance that ultimately maintains a constant bone mass and volume . Osteoblasts and osteoclasts originate from distinct stem cell types within the bone marrow—mesenchymal stem cells (MSCs) and hematopoietic stem cells (HSCs), respectively [9-11]. Osteoblast differentiation of MSCs is orchestrated by a complex signaling cascade involving multiple pathways including the bone morphogenic proteins and winglesstype MMTV integration site (Wnt) proteins [10, 12]. Osteoclast differentiation of HSCs is largely under the control of immune modulators such as macrophage colony-stimulating factor (M-CSF) and receptor activator of nuclear factor-kappa B ligand (RANKL; also known as TNFSF11), a tumor-necrosis factor (TNF)-family cytokine [13-15]. As disorders of bone loss are commonly associated with a relative increase in bone resorption, the potential to treat bone loss by inhibiting osteoclastogenesis has promise . Osteoclast differentiation proceeds in a two stage differentiation process involving (a) induction of the FBJ osteosarcoma oncogene (Fos), nuclear factor of activated T-cells 2 (NFAT2 also called NFATc1 or NFATc) and nuclear factor-kappa B (NFκB) resulting in preosteoclast formation [17, 18] and (b) osteoclast-specific robust induction of NFAT2 resulting in terminal differentiation [18, 19]. Although a complex network of signaling events are reported to mediate these two phases [17-19], the link between the early signaling pathways activated by RANKL and the NFAT2-induced terminal osteoclast differentiation is not fully established.
Chemerin is a small (18 kDa) secreted protein that modulates inflammation, immune function, tumor progression, and recruitment of inflammatory cells through its interaction with chemokine-like receptor 1 (CMKLR1) [20-25]. Chemerin also serves as a ligand for two other receptors, G protein-coupled receptor 1  and chemokine (C-C motif) receptor-like 2 (CCRL2) . While CCRL2, the nonsignaling receptor for chemerin, can bind chemerin and increase local chemerin concentration to efficiently present it to CMKLR1 on nearby cells , all the biological effects ascribed to chemerin have been reported to be elicited through CMKLR1 activation [20-23]. In addition to the role in immune function, autocrine/paracrine activation of CMKLR1 by chemerin promotes adipogenesis of a variety of precursor cell types including 3T3-L1 fibroblasts and primary MSCs [29-31]. Given the importance of chemerin/CMKLR1 signaling to HSC-derived cells in the context of immune function, we hypothesized that this signaling pathway also regulates the differentiation of bone marrow HSCs to bone resorbing osteoclasts. Our experimental data provide evidence that supports a previously unrecognized autocrine/paracrine role for chemerin to regulate HSC differentiation and thereby, bone homeostasis.
Materials and Methods
All experiments were performed according to the guidelines of the Canadian Council on Animal Care and were approved by the Dalhousie University Committee on Laboratory Animals. The mice had free access to standard rodent chow and water, and were maintained at an ambient temperature of 20°C–22°C, relative humidity 18%–22%, and a 12 hours light/dark cycle.
Stem Cell Isolation and Differentiation
Isolation, culture, and adipocyte or osteoblast differentiation of MSCs were performed as described previously [29, 30]. HSCs were isolated from mouse bone marrow using a modification of the method described by de Vries et al. . Briefly, 2-month-old male C57BL/6J mice were euthanized with an overdose (90 mg/kg) of pentobarbital sodium (Ceva Sante Animale, LA Ballastiere, Libourne, France, http://www.ceva.com) and femurs and tibiae were removed. The marrow was flushed with HSC medium (α-MEM containing 1% l-glutamine, 10% fetal bovine serum, 100 U/ml penicillin, and 100 µg/ml streptomycin) and passed through a 23-gauge needle and filtered through a 70 µm pore size cell strainer (BD Biosciences, Bedford, MA, www.bdbiosciences.com). Lin− Sca1+ hematopoietic cells were isolated by sequential depletion of lineage positive (Lin+) cells (MagCellect Mouse Hematopoietic Cell Lineage Depletion Kit, R&D Systems) and isolation of Sca-1-positive stem cells (Mouse Sca-1+ Stem Cells kit, R&D Systems, Minneapolis, MN, www.rndsystems.com). The resulting Lin− Sca1+ cells were stained with 0.5 µg/ml fluorescein isothiocyanate-conjugated rat anti-mouse c-kit (CD117) (BD Biosciences, catalog number 561680) and 10 µg/ml Alexa Fluor 647 (AF 647)-conjugated rat anti-mouse CD34 antibodies (BD Biosciences, catalog number 560230), and sorted by fluorescence-activated cell sorting to obtain Lin− Sca1+ c-kit+ CD34+ pure HSCs. Osteoclast differentiation was induced by 30 ng/ml M-CSF (R&D Systems) and 50 ng/ml RANKL (R&D Systems) for 14 days. Undifferentiated control cells received M-CSF but no RANKL. Anti-chemerin antibody (CmAb) (catalog number AF2325, R&D Systems) was added to the medium to neutralize extracellular chemerin activity. Goat IgG (catalog number AB-108-C, R&D Systems) was used as a control.
Osteoclast Differentiation and Detection of NFκB Activation in RAW264.7 Cells
Osteoclast differentiation of monocyte RAW264.7 progenitors (American Type Culture Collection, Manassas, VA, www.atcc.org) was induced by 50 ng/ml RANKL. To monitor NFκB activity, the cells were stably transduced with the Cignal Lenti Reporter (luciferase) for NFκB (SABiosciences, Frederick, MD, www.sabiosciences.com). NFκB activity was assessed by measuring luciferase activity using a Luminoskan Ascent luminometer Luminoskan Ascent (Thermo Scientific, Waltham, MA, www.thermoscientific.com) and expressed as relative light units.
Osteoclast Detection and Characterization
Osteoclasts were fixed and TRAP-stained using a leukocyte acid phosphatase kit (Sigma-Aldrich, St. Louis, MO, www.sigmaaldrich.com) according to the manufacturer's protocol. Images were captured using a Zeiss Axiovert 200 inverted microscope equipped with an AxioCam camera system (Zeiss Canada, Toronto, ON, Canada, www.corporate.zeiss.com). TRAP activity at 37°C was quantified with a substrate buffer (125 µg/ml Naphthol AS-BI phosphoric acid [Sigma] in diazotized fast garnet base chemical solution (Sigma-Aldrich) containing 6.7 mM sodium tartrate in 100 mM acetate buffer pH 5.2) at 560 nm and expressed as mOD/minute. Actin ring formation was visualized using tetramethylrhodamine-phalloidin (Life Technologies, Carlsbad, CA, www.invitrogen.com) staining for 1 hour. The cell nuclei were counterstained with 1 µg/ml Hoechst 33258 (Sigma-Aldrich) in phosphate buffered saline for 5 minute. The number of actin rings in each well was counted and expressed as actin rings per well. Image analysis of the total area of plate surface contained within actin rings was performed with Axiovision rel 4.8 (Zeiss Canada).
Matrix Resorption Assays
HSCs were differentiated on calcified matrix using crystalline calcium phosphate coated polystyrene plates (Corning Osteo Assay Plate, Corning, NY, www.corning.com). After 14 days, cells were removed from the wells by washing with 10% bleach, followed by two washes in distilled water. The plates were air-dried and examined by light microscopy, in which resorption pits appear as light lacunae in the resorbed areas. Representative images were photographed under 10× magnification. The area of resorbed matrix was quantified using image analysis software (Axiovision rel 4.8, Zeiss Canada) and expressed as % resorbed area per well.
Cell lysates and media samples were separated by SDS-PAGE and transferred to nitrocellulose membranes as described previously . Immunostaining was performed using goat anti-chemerin (1:1,000; catalog number AF2325, R&D Systems), goat anti-matrix metalloproteinase (MMP)−9 (1:1,000; catalog number AF909, R&D Systems), rabbit anti-cathepsin K (1:500; catalog number ab19027, Abcam, Cambridge, MA, www.abcam.com), and mouse anti-β-actin (1:2,000; catalog number ab8226, Abcam). The appropriate secondary antibodies conjugated with a 680 nm or 800 nm infrared fluorophore (1:10,000, LI-COR Biosciences, Lincoln, NE, www.licor.com). The immunoreactive bands were detected by scanning at 700 nm or 800 nm, respectively, using a LI-COR Odyssey infrared scanner (LI-COR Biosciences).
Gene Expression Analysis
Total RNA was isolated using an RNeasy Plus Minikit (Qiagen, Hilden, Germany, www.qiagen.com) according to the manufacturer's instructions. Reverse transcription and quantitative PCR detection of genes were measured as described previously [29, 31, 33]. The Ct values for cyclophilin A were used to normalize the expression level of the gene of interest using the ΔΔCt method . Exon-spanning primers for quantitative real-time PCR (qPCR) were designed using the primerbank software algorithm (http://pga.mgh.harvard.edu/primerbank/) and are listed in supporting information Table S1.
Lentiviral Gene Transfer in Mouse HSCs
Lentiviruses were generated according to previously established methods . Briefly, HEK-293T/17 cells (ATCC CRL-11268) were transfected with lentiviral vectors, including psPAX2 (plasmid 12260, Addgene, Cambridge, MA, www.addgene.org) and pMD2.G (plasmid 12259, Addgene) along with the nonsilencing shRNA (Thermo Scientific GIPZ Lentiviral shRNAmir plasmid RHS_4346) a negative control that was verified to contain no homology to known mammalian genes, chemerin shRNA (Thermo Scientific GIPZ Lentiviral shRNAmir plasmid V2LMM_52383), or wild-type (WT) NFAT2 (plasmid 11101, Addgene), constitutively active (CA) mutant NFAT2 (plasmid 11102, Addgene), or green fluorescent protein (GFP) control (plasmid 11652, Addgene). Viral supernatants were collected at 24, 48, and 72 hours. The viruses for shRNA knockdown studies were purified by ViraBind Lentivirus Concentration and Purification kit (Cell Biolabs, San Diego, CA, www.cellbiolabs.com). Viral titer was determined according to the procedure described for Open Biosystems Expression Arrest GIPZ Lentiviral shRNAmir (Thermo Scientific) using GFP as a marker for detection of transduced cells. Chemerin knockdown was assessed using chemerin expression plasmids in HEK-293T cell line as well as endogenous chemerin expression in HSCs. Transduction of mouse HSCs with the lentiviral shRNA was performed in HSC medium containing 5% fetal bovine serum, 10 ng/ml M-CSF, and 10−4 M β-mercaptoethanol. All transductions were carried out at a multiplicity of infection of 10 supplemented with 4 µg/ml protamine sulfate and 50 µM deoxynucleoside triphosphates for 96 hours and the transduction medium was changed every 24 hours. Lentiviruses generated for NFAT2 expression studies were pelleted by ultracentrifugation and viral titer was determined by quantitative PCR amplification. NFAT2 transduction ability of the viruses was assessed using a nonosteoclastogenic HEK-293A cell line.
Induction of Calcium Release
To induce calcium release from intracellular stores, Thapsigargin (3 nM; catalog number T9033, Sigma-Aldrich) and Ionomycin (300 nM; catalog number I9657, Sigma-Aldrich) were used.
MTT Assay of Cellular Viability
Cell viability after 8 days of treatment with CmAb was measured by a colorimetric 3-(4,5-dimethyl-thiazol-2-yl) 2,5-diphenyl tetrazolium bromide (MTT) assay originally developed by Mosmann . To measure cell viability, 100 µl MTT dye solution (0.5 mg/ml) was added to each well of a 96-well plate. The formazan product produced by viable cells was solubilized by the addition of 100 µl of dimethyl sulfoxide to the wells and the absorbance at 540 nm was determined. Absorbance values of control cells were considered as 100% cell viability.
Results are expressed as mean ± SEM. Comparisons between values were performed using a Student's t test (two-tailed) as in Figures 1, 6A. For comparison of multiple groups as in Figures 2-7, a one-way or two-way analysis of variance with Bonferroni post hoc analysis was used. For all statistical analyses, p < .05 was considered statistically significant.
Chemerin/CMKLR1 Expression During HSC Osteoclast Differentiation
Freshly isolated HSCs from mouse bone marrow, negative for markers of lineage commitment (Lin−) and positive for the HSC markers [37, 38] Sca1 (Sca1+), c-kit (c-kit+), and CD34 (CD34+) expressed mRNA for both chemerin and the cognate receptor CMKLR1 at levels comparatively lower and higher, respectively, to bone marrow MSCs (supporting information Fig. S1). Following 14 days exposure to an osteoclastogenic medium containing M-CSF and RANKL, the Lin− Sca1+ c-kit+ CD34+ HSCs exhibited robust osteoclast differentiation, as evidenced by the presence of phalloidin-stained actin rings (Fig. 1A), TRAP-stained osteoclasts (Fig. 1B) and significant TRAP secretion (Fig. 1C). Consistent with this, qPCR analysis revealed increased mRNA levels for various osteoclast marker genes including TRAP, MMP-9, MMP-10, cathepsin K (CatK), calcitonin receptor (CalR), nuclear factor of activated T-cells 2 (NFAT2) also called NFATc1 or NFATc, osteoclast-associated receptor (OSCAR), and TNF receptor-associated factor 6 (TRAF6) (Fig. 1D). In contrast to the osteoclast marker genes, chemerin mRNA levels (Fig. 1E) were not significantly changed while CMKLR1 mRNA levels (Fig. 1F) decreased with osteoclast differentiation. Compared to other bone marrow cell types such as MSCs, adipocytes, or osteoblasts, the absolute amount of chemerin secreted by HSCs was lower (Fig. 1G). Moreover, in contrast to the differentiation dependent increase with MSC adipogenesis, chemerin protein secretion by HSCs did not increase with osteoclast differentiation (Fig. 1G).
Neutralization of Chemerin in HSCs
To characterize the role of chemerin in the osteoclastogenic program of HSCs, we blocked chemerin/CMKLR1 signaling with a chemerin-neutralizing antibody (CmAb) beginning prior to application of osteoclastogenic stimulus (day 0) and persisting throughout 14 days of osteoclastogenic stimulation. Goat IgG was used as an isotype control for the goat anti-mouse chemerin-neutralizing antibody. Both vehicle (phosphate buffered saline) and IgG (50 µg/ml) treated cells exhibited robust osteoclast differentiation as evidenced by actin ring formation and the presence of large, TRAP positive osteoclasts (Fig. 2A). Neutralization of extracellular chemerin by CmAb markedly reduced actin ring formation and the presence of TRAP positive cells (Fig. 2A). Quantitative analysis verified ∼50% and ∼75% loss of actin ring number and area, respectively, with 25 µg/ml CmAb, while a higher concentration (50 µg/ml) caused >95% loss of both actin ring number and area (Fig. 2B). Consistent with this, TRAP secretion was reduced in a concentration-dependent manner with >50% loss at 25 µg/ml and >95% loss with 50 µg/ml CmAb compared to vehicle or IgG-treated cells (Fig. 2C). Both protein and mRNA levels for NFAT2, a critical transcription factor that regulates the expression of numerous osteoclastogenic genes , were markedly suppressed by chemerin neutralization (Fig. 2D, 2E). Moreover, chemerin neutralization resulted in a dramatic loss of MMP9 and CatK protein expression (Fig. 2D) and a concentration-dependent reduction of TRAP, MMP9, MMP10, CatK, CalR, OSCAR, and TRAF6 mRNA levels (Fig. 2E). To examine whether the loss of osteoclast differentiation would impact osteoclast-associated function, matrix resorption activity was measured. Direct or phase-contrast microscopic examination of the calcified matrix following 14 days of osteoclast differentiation revealed a near complete matrix loss in the wells treated with vehicle or IgG (50 µg/ml; Fig. 3A, 3B). In contrast, chemerin neutralization produced a marked decrease of matrix resorption with the highest concentration of 50 µg/ml CmAb reducing resorption activity to levels comparable to the undifferentiated M-CSF control (Fig. 3A, 3B). Quantitative analysis verified these qualitative observations and revealed >50% and >90% loss of matrix resorption with 25 and 50 µg/ml CmAb, respectively (Fig. 3C). Importantly, both the apparent potency and efficacy of CmAb with respect to inhibiting matrix resorption were very similar to that seen for other measures of osteoclast phenotype such as actin ring formation, TRAP secretion, and marker gene expression (Fig. 2).
The mouse monocytic cell line, RAW 264.7, readily undergoes osteoclast differentiation upon exposure to RANKL and has been widely used to study the development and function of this cell type . To determine whether the role for chemerin in osteoclastogenesis of primary HSCs was conserved in an independent model of osteoclast differentiation, chemerin neutralization was performed with RAW264.7 cells. RANKL treatment of RAW264.7 cells induced robust osteoclast differentiation, as evidenced by extensive actin ring formation (Fig. 4A, 4B) and TRAP secretion (Fig. 4C) under vehicle or IgG-treated condition. However, similar to HSCs, chemerin neutralization with 25 or 50 µg/ml CmAb markedly reduced actin ring formation (>50% loss with 25 µg/ml; >90% loss with 50 µg/ml) and TRAP secretion (>50% loss with 25 µg/ml; >90% loss with 50 µg/ml) compared to vehicle treated cells (Fig. 4B, 4C).
We developed lentiviral delivered shRNA to knockdown chemerin as a complementary experimental approach to study the role of this protein in HSC osteoclastogenesis. Transfection of the lentiviral plasmid achieved effective knockdown in HEK293T cells overexpressing chemerin mRNA (supporting information Fig. S2A). However, transduction with lentivirus particles expressing the same shRNA sequence (supporting information Fig. S2B) was less effective despite high transduction efficiency as assessed by a GFP marker (data not shown). Similarly, we were able to achieve only a partial knockdown (∼50%) of HSC chemerin mRNA or protein levels with lentivirus transduction (supporting information Fig. S2C, S2D). However, in agreement with the antibody neutralization studies, this incomplete knockdown resulted in a partial impairment of osteoclastogenesis as evidenced by reduced actin ring area and TRAP staining compared to vehicle and nonspecific shRNA expression (supporting information Fig. S2E). Moreover, while there was an overall trend toward reduced osteoclast marker gene expression, only MMP10 was reduced significantly (supporting information Fig. S2F).
Reversal of Chemerin Neutralization and Blockade of HSC Osteoclastogenesis
In order to determine whether the inhibitory effect of chemerin neutralization on HSC osteoclastogenesis was reversible, after 7 days of blockade the differentiation medium containing CmAb was replaced with fresh differentiation medium containing IgG. Under this protocol, both actin ring number and area (Fig. 5A, 5B) as well as TRAP secretion (supporting information Fig. S3A) reached levels similar to that of IgG control and that was dramatically higher (∼50-fold) compared to cells that were continuously exposed to CmAb for 17 days (Fig. 5A). To assess the ability of chemerin to rescue the osteoclastogenic deficits induced by CmAb, the cells were exposed to a recombinant form of bioactive chemerin after 7 days antibody neutralization. Neither recombinant mouse chemerin (mCm) nor human chemerin (hCm) affected HSC osteoclastogenesis in the presence of IgG (Fig. 5C, 5E). However, in the presence of CmAb both species of chemerin were able to increase actin ring formation (Fig. 5C--5F) and TRAP secretion (supporting information Fig. S3B) albeit not to levels observed in the absence of CmAb. Exposure of HSCs to CmAb at concentrations up to 100 µg/ml caused no significant cell toxicity (supporting information Fig. S4) indicating that the loss of osteoclastogenesis was not a consequence of cell death and that the recovery of osteoclast formation after removal of CmAb was a consequence of recruitment of viable multipotent HSCs to the osteoclastogenic program.
Regulation of Osteoclastogenesis by Chemerin Through NFAT2
The marked loss of NFAT2 expression with antibody neutralization suggested that modulation of the expression and/or activity of this transcription factor was relevant to the role of chemerin to osteoclastogenesis. However, it was not clear whether the defect in osteoclast differentiation with chemerin neutralization was mediated through a loss of NFAT2 expression or whether the loss of NFAT2 was a consequence of effects on later stages of differentiation. To investigate this, we assessed the RANKL-induced changes in early-stage gene expression. Following 72 hours of RANKL stimulation, an 8–10-fold induction of mRNA levels of NFAT2 and its transcriptional partner, Fos, was observed (Fig. 6A). Moreover, NFAT2 target genes including integrin beta 3 (Itgb3)  and Src-kinase (Src)  were induced in HSCs within 72 hours of RANKL treatment (Fig. 6A). Consistent with the association of CMKLR1 with the early osteoclastogenic events, a loss of CMKLR1 was coincidental with these changes in the 72 hours period. NFκB is another important transcription factor that plays a pivotal role in osteoclast differentiation by activating NFAT2 and Fos [42-45]. RANKL increased NFκB luciferase reporter activity >20-fold within 24 hours after stimulation (Fig. 6A). To test whether chemerin mediated the RANKL-induced early-stage gene expression, we neutralized chemerin and assessed mRNA expression in the presence of RANKL for 72 hours. The CmAb markedly abrogated (>75% loss) the RANKL-induced mRNA expression of NFAT2, Fos, Itgb3, and Src in a dose-dependent manner (Fig. 6B). Moreover, the RANKL-induced loss of CMKLR1 expression was significantly reversed by CmAb (Fig. 6B). In contrast, chemerin neutralization did not significantly affect RANKL-induced NFκB reporter activation (Fig. 6B). To determine whether reduced NFAT2 expression was a major factor underlying the inhibition of osteoclastogenesis observed with chemerin neutralization, we tested whether NFAT2 expression alone was sufficient to rescue osteoclast differentiation of HSCs. Forced expression of WT-NFAT2 coincident with chemerin neutralization produced a partial rescue of expression of osteoclast markers such as MMP9 and CatK while the CA form restored the expression of these markers to levels similar in magnitude to that for phosphate buffered saline or IgG control (Fig. 6C). Consistent with this, CA-NFAT2 expression significantly restored actin ring formation (>75%) but the WT failed (<25%) to produce significant restoration of actin ring formation (Fig. 6D, 6E). However, both WT- and CA-NFAT2 significantly rescued TRAP secretion (>50% rescue for WT; total rescue for CA; Fig. 6F). The levels of NFAT2 protein were apparently higher with CA- versus WT-NFAT2 rescue (Fig. 6C). This was likely a consequence of a greater feed-forward amplification of endogenous NFAT2 expression provoked by CA-NFAT2 rather than differences in the inherent expression of lentivirus expressed WT-NFAT2 and CA-NFAT2 transgenes. In support of this, transduction of the nonosteoclastogenic 293A cell line resulted in similar levels of NFAT2 expression with both the WT-NFAT2 and CA-NFAT2 transgenes (supporting information Fig. S5).
Intracellular calcium levels are an important factor governing the nuclear translocation of NFAT2 during osteoclastogenesis . To determine whether changes in intracellular calcium contributed to the effects of chemerin neutralization on NFAT2 expression and osteoclastogenesis, we induced intracellular calcium release using the Ca2+-ATPase inhibitor (thapsigargin) or the Ca2+-ionophore (ionomycin). Stimulation of intracellular calcium release by either compound partially rescued (25%–50%) actin ring formation (Fig. 7A, 7B) as well as NFAT2, MMP9, and CatK protein expression (Fig. 7C) in the face of persistent chemerin neutralization. However, the rescue by calcium release was generally lower in magnitude compared to the rescue achieved with NFAT2 overexpression on MMP9, CatK, and NFAT2 expression (Fig. 7C). Taken together, these data indicate that chemerin regulates osteoclastogenesis primarily through modulating NFAT2 expression and activity, likely in part by affecting intracellular calcium levels.
Our previous studies identified the contribution of chemerin/CMKLR1 signaling as a determinant of the developmental fate of bone marrow-derived MSCs to the adipocyte and osteoblast lineage. In this study, we have identified a further role for this signaling pathway in regulating the osteoclastogenic differentiation of Lin− Sca1+ c-kit+ CD34+ HSCs and thus revealed an expanded influence in regulating bone homeostasis. Similar to our previous findings [29, 30] that MSCs are both a source and target of chemerin, we found that HSC-secreted chemerin also functions in an autocrine/paracrine fashion to promote osteoclastogenesis. Within the bone, crosstalk between osteoblast and osteoclast progenitors is essential for coupling of bone formation to bone resorption. One way this is achieved is through the paracrine actions of osteoblast-secreted RANKL to promote coordinated osteoclast differentiation of RANK receptor-expressing HSC progenitors [47, 48]. Following from this, chemerin secretion by other cell types within the bone marrow milieu may also influence HSC differentiation. For example, dynamic secretion of chemerin from MSCs and MSC-derived osteoblast may contribute to the homeostatic coupling of MSC osteoblastogenesis with HSC osteoclastogenesis through the paracrine actions of osteoblast-secreted chemerin on CMKLR1 receptor-expressing HSCs. Of all bone marrow cell types examined in this study, adipocytes secreted the highest amounts of chemerin. Many disorders of bone loss such as osteoporosis are characterized by both an increase in bone resorption and a preponderance of bone marrow adipocytes [4, 10]. Thus, given the pro-osteoclastogenic effect of chemerin/CMKLR1 toward HSCs, it is certainly possible that elevated levels of chemerin within the bone marrow milieu contribute to pathogenic disruptions of bone homeostasis that ultimately lead to the destruction of bone in osteoporotic and rheumatic diseases.
It is generally believed that M-CSF and RANKL are both necessary and sufficient for osteoclast differentiation [14, 49-52]. However, we have demonstrated that the autocrine/paracrine actions of chemerin are also essential for osteoclast differentiation of both HSCs and RAW264.7 cells. This was demonstrated most convincingly in the context of antibody neutralization of extracellular chemerin. When lentiviral shRNA-mediated knockdown of chemerin mRNA was used, the effects were qualitatively similar to that of antibody neutralization. The comparatively smaller impact of the latter approach on osteoclastogenesis was likely a consequence of the inability to achieve a substantial (>90%) loss of chemerin expression. Indeed, at lower antibody concentrations (e.g., 10 µg/ml) where only partial (30%–50%; data not shown) chemerin neutralization is achieved, incomplete inhibition of HSC osteoclastogenesis was observed (Figs. 2C, 3C, 4C, 6B). Importantly, the anti-osteoclastogenic effect of CmAb was not due to cytotoxicity and was almost completely reversible. This reversibility indicates that the cells did not lose the potential to differentiate but rather that chemerin was necessary for progression down the differentiation pathway. Unexpectedly, recombinant chemerin did not completely rescue the osteoclast differentiation blockade imposed by the CmAb. Chemerin is initially translated as a 163 aa preproprotein that is believed to be secreted as a relatively inactive 143 aa proprotein (Chem163K) after removal of the N-terminal signal peptide sequence [23, 53]. Subsequently, extracellular proteolytic processing of the C-terminal is the major determinant of the biological activity of chemerin. This processing is complex and involves a number of enzymes including plasmin, carboxypeptidases, chymase, cathepsins, and elastase that are variously involved in bioactivating/deactivating chemerin to forms that vary with respect to intrinsic biological activity . A limitation of this study is that the specific isoform(s) of chemerin present in the extracellular media of the HSCs and which promotes osteoclastogenesis is unknown. This is an important consideration as experimental data indicate that the actions of chemerin are highly contextual and that the abundance of chemerin isoforms varies based upon anatomical site. For example, while Chem163K is the major circulating form of chemerin in humans, the dominant isoform in synovial fluid is Chem158K . A further consideration is that nothing is known regarding other post-translational modifications (e.g., glycosylation and oligomerization) that affect bioactivity, stability, or receptor selectivity of chemerin. Thus, supplementation of the media with bacterially expressed recombinant human Chem157S or murine Chem156S may not effectively reproduce the endogenous chemerin environment and thereby fully rescue osteoclastogenesis in the presence of the CmAb.
Experimental evidence indicates that both HSCs and RAW264.7 cells exhibit two phases of osteoclast differentiation involving the formation of a progenitor preosteoclast cell prior to the development of a mature, multinucleated osteoclast [18, 56]. The majority of the genes required for osteoclast differentiation are induced during the early-stage of commitment to the osteoclast lineage . An important early event is the induction of NFAT2 expression, which is mediated through Fos and NFκB acting directly on the NFAT2 promoter [57, 58]. Our finding that chemerin neutralization substantially blunted the induction of NFAT2 and Fos gene expression by RANKL is consistent with a very early influence of chemerin/CMKLR1 signaling on HSC osteoclastogenesis. RANKL has also been shown to promote integrin-mediated cell adhesion of HSCs during osteoclastogenesis . Similarly, chemerin has previously been reported to promote adhesion of macrophage cell lines . In this study, chemerin neutralization resulted in reduced levels of RANKL-induced itgb3 and Src expression, both of which are NFAT2 target genes [40, 41]. Taken together, these findings indicate that chemerin/CMKLR1 signaling is involved in key early events of osteoclastogenesis including NFAT2 induction and cell adhesion. RANKL-induced NFκB activation has also been shown to be a key early event in promoting NFAT2 and Fos expression and thereby, osteoclastogenesis. The absence of effect of the CmAb on NFκB activation indicates that chemerin modulates NFAT2 expression and osteoclastogenesis independent of this signaling pathway.
Following the initial phase of NFAT2 induction, NFAT2 cooperates with Fos and NFκB to promote transcriptional activation of genes associated with terminal osteoclast differentiation. Also characteristic of this response is an auto-amplification of NFAT2 expression whereby NFAT2 itself binds and transactivates the NFAT2 promoter. Consistent with modulation of NFAT2 being an effect of chemerin/CMKLR1 signaling, forced expression of NFAT2 rescued osteoclastogenesis to varying degrees in the presence of CmAb. CA-NFAT2 was particularly effective and completely rescued HSC osteoclastogenesis in the face of persistent chemerin neutralization. Intriguingly, the protein level of NFAT2 in the osteoclasts expressing CA-NFAT2 was greater than the cells expressing WT-NFAT2. This difference in protein level was not related to differences in the inherent expression of the lentivirus transgenes but rather due to enhanced auto-amplification of endogenous NFAT expression by CA-NFAT2. Consistent with this, previous studies have identified that FK506, an inhibitor of NFAT2 nuclear translocation, blocked the auto-amplification of NFAT2 and that overexpression of NFAT2 induced the expression of endogenous NFAT2 gene .
The incomplete rescue of osteoclastogenesis achieved by expression of WT-NFAT2 also suggests that chemerin/CMKLR1 is involved in modulating both the expression and activation of NFAT. In order to achieve nuclear translocation, NFAT2 must be first dephosphorylated by calcineurin, a calcium-dependent phosphatase . Intriguingly, one of the earliest biological actions ascribed to chemerin was to promote calcium mobilization in immature dendritic cells and macrophages . Thus, it is possible that chemerin promotes NFAT2 activation through increasing intracellular calcium concentrations. Our finding that increasing intracellular calcium levels with a Ca2+-ATPase inhibitor (thapsigargin) or a Ca2+-ionophore (ionomycin) partially restored RANKL-induced osteoclastogenesis in the face of chemerin neutralization is consistent with this proposition. The failure to achieve a complete rescue may relate to differences in the pattern of calcium release achieved with thapsigargin or ionomycin compared to chemerin. For example, a previous study reported that a sustained calcium oscillation but not transient activation of calcium spike promoted robust NFAT2 activation during terminal osteoclast differentiation . Taken together, these data indicate that chemerin/CMKLR1 signaling is important early in osteoclastogenic differentiation of HSCs and that NFAT2 expression and activation are key mediators of this effect.
In summary, our data demonstrate that blockade of chemerin signaling abrogates osteoclastogenesis of HSCs and thereby reveals a hitherto unrecognized autocrine/paracrine role for this secreted protein. MSCs and MSC-derived cell types (e.g., osteoblasts and adipocytes) also express and secrete chemerin at levels higher than HSCs, while CMKLR1 receptor expression is comparatively highest in HSCs. Thus, chemerin may contribute, in a similar fashion to RANKL, to the paracrine cross talk between osteoblasts and osteoclasts and contribute to the homeostatic balance of bone formation/resorption. It is also possible that higher levels of chemerin secretion associated with changes in the bone marrow cell population (e.g., more adipocytes) may be pathogenic and contribute to the relative increase of bone resorption characteristic of disorders of bone loss. Denosumab is a therapeutic antibody that neutralizes RANKL and has been shown in clinical trials to improve bone mineral density and reduce the risk for vertebral fractures in postmenopausal women with osteopenia or osteoporosis [60, 61]. Our finding that chemerin neutralization reduces osteoclast differentiation reveals the possibility that chemerin-neutralizing antibodies will also have therapeutic value for treating disorders of bone loss.
We are grateful for on-going collaborations with Eugene C. Butcher and Brian A. Zabel that were essential to this work. HJD is supported by a National Sciences and Engineering Research Council of Canada (NSERC) Canada Graduate Scholarship and HJD and JLR are supported by Izaak Walton Killam Predoctoral scholarships. This work was supported by funding from the Canadian Institutes of Health Research.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.