|Home | About | Journals | Submit | Contact Us | Français|
This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The objective of the present study was to investigate the role of the stromal cell-derived factor 1 (SDF-1)/CXCR4 axis in TNF-induced mobilization of osteoclast precursors (OCPs) from bone marrow.
OCPs were generated from bone marrow cells of TNF-transgenic mice or wild-type mice treated with TNF or PBS. The percentage of CD11b+/Gr-1-/lo OCPs was assessed by fluorescence-activated cell sorting. OCP migration to the SDF-1 gradient and the osteoclast forming potency were assessed in chemotaxis/osteoclastogenic assays. SDF-1 expression was assessed by real-time RT-PCR, ELISA and immunostaining in primary bone marrow stromal cells, in the ST2 bone marrow stromal cell line, and in bones from TNF-injected mice.
OCPs generated in vitro from wild-type mice migrated to SDF-1 gradients and subsequently gave rise to osteoclasts in response to RANKL and macrophage colony-stimulating factor. TNF reduced SDF-1 expression by ST2 cells. Bone marrow stromal cells from TNF-transgenic mice produced low levels of SDF-1. TNF treatment of wild-type mice decreased the SDF-1 concentration in bone marrow extracts and decreased the SDF-1 immunostaining of bone marrow stromal cells, and it also increased the circulating OCP numbers. The percentage of bone marrow CXCR4+ OCPs was similar in TNF-transgenic mice and wild-type littermates and in TNF-treated and PBS-treated wild-type mice.
Systemically elevated TNF levels inhibit bone marrow stromal cell production of SDF-1 and increase the release of bone marrow OCPs to the peripheral blood. Disruption of the SDF-1/CXCR4 axis by TNF may play an important role in mediating OCP mobilization from the bone marrow cavity in chronic inflammatory arthritis.
TNF is a clinically validated etiological factor in inflammatory-erosive arthritis and is known to synergize with RANKL and macrophage colony-stimulating factor (M-CSF) to enhance the differentiation of osteoclast precursors (OCPs) into bone-resorbing osteoclasts in inflamed joints [1,2]. Patients with psoriatic arthritis  and mice with TNF-induced arthritis [4,5] have increased numbers of circulating OCPs, which correlate with systemically increased TNF concentrations and are reduced by anti-TNF therapy in association with clinical improvement. These findings suggest that OCP mobilization from the marrow may be involved in the pathogenesis of inflammatory arthritis. The factors that mediate OCP mobilization are currently unknown.
Stromal cell-derived factor 1 (SDF-1), a member of the C-X-C chemokine family also known as CXCL12, acts through its receptor CXCR4, and is the master chemokine that modulates trafficking of hematopoietic stem cells and progenitors [6,7]. Studies of knockout mice reveal that the SDF-1/CXCR4 axis is required for fetal B lymphopoiesis, bone marrow myelopoiesis and organogenesis [8-11]. Both SDF-1-deficient and CXCR4-deficient mice die perinatally and have very few hematopoietic stem cells and progenitors within their bone marrow. SDF-1 and CXCR4 have been implicated in OCP migration in vitro, and SDF-1 treatment of OCPs increases osteoclastogenesis and subsequent osteoclast bone-resorbing capacity [12,13].
SDF-1 is primarily produced by bone marrow stromal cells, such as osteoblasts and endothelial cells . Expression of SDF-1 is controlled by various factors including hypoxia , DNA damage  and cytokines, such as transforming growth factor beta (TGFβ)  and granulocyte colony-stimulating factor (G-CSF) . G-CSF is used clinically to stimulate the release of hematopoietic stem cells from the bone marrow into the bloodstream of patients with a variety of malignancies. The stem cells are then harvested from the blood as a source of stem cells to be returned to patients following chemotherapy or bone marrow transplantation. Whether or not inflammatory cytokines such as TNF affect the SDF-1/CXCR4 axis in vivo to control OCP mobilization, however, has not been studied.
We used TNF-transgenic (TNF-Tg) mice as a model of chronic TNF overexpression and also injected WT mice with TNF as an acute model to investigate the involvement of TNF in the SDF-1/CXCR4 axis control of OCP mobilization. We found that TNF directly inhibits SDF-1 production by bone marrow stromal cells and that it has little effect on CXCR4 expression by OCPs. A mechanism whereby TNF accelerates OCP mobilization in inflammatory erosive arthritis may therefore be to reduce bone marrow SDF-1 concentrations.
Recombinant murine SDF-1, TNFα, and RANKL were from R&D Systems (Minneapolis, MN, USA). Allophycocyanin–anti-murine CD11b (M1/70) was from eBiosciences (San Diego, CA, USA). FITC–anti-murine Gr-1 (RB6-8c5), biotin–anti-CXCR4 (2B11/CXCR4) and streptavidin–PE-Texas Red conjugate were from BD PharMingen (San Diego, CA, USA). Mouse SDF-1/CXCL12 DuoSet Development system was from R&D Systems.
TNF-Tg mice in a CBA × C57BL/6 background (3647 TNF-Tg line) were obtained originally from Dr G. Kollias and were characterized by our group previously . TNF-Tg mice have been bred with C57/B6 mice for eight generations. Cxcr4 floxed and CD11b+/Cre mice were obtained from Dr YR Zou  and Dr J Vacher , respectively. Both types of mice are in a C57BL/6 background.
TNF was given by subcutaneous injection, as described previously . The University Committee on Animal Resources of the University of Rochester approved all studies.
Freshly isolated bone marrow cells were cultured with M-CSF in α-modified essential medium (Invitrogen, San Francisco, CA, USA) supplemented with 10% fetal bovine serum (Invitrogen) for 3 days, and adherent cells were used as OCPs. Assays were performed using transwell chemotaxis inserts with 5-μm-pore polycarbonate filters (Corning Costar, Acton, MA, USA). OCPs were labeled with Calcein AM (Molecular Probes, Carlsbad, CA, USA) at a final concentration of 2 μg/ml, and 100 μl (106 cells) cell suspension were loaded into the upper chamber of a transwell insert. The transwell inserts were immediately moved to wells of a 24-well tissue culture dish containing different doses of SDF-1α (1, 10 or 100 ng/ml). After 3 hours of incubation, the migrated cells in the bottom wells were collected, centrifuged and solubilized (in 100 μl Hank's Buffered Salt Solution with 1% SDS/0.2 N NaOH). The calcein label was read in a 96-well FluoroNunc plate (Nalge Nunc International, Rochester, NY, USA) and quantified in a Gemini XS microplate spectrofluorometer (Molecular Devices, Sunnyvale, CA, USA) at 485 nm/530 nm.
The number of cells that migrated was calculated according to a standard curve generated by plotting the calcein intensity of serially diluted labeled cells versus the cell numbers. The percentage of migrated cells was calculated as follows: (migrated cell number/total loaded cell number) × 100%. The cells in the upper and lower chambers of the transwell were collected and cultured with M-CSF and RANKL to determine whether they could differentiate into osteoclasts, as described previously . These treated cells were fixed and stained for tartrate-resistant acid phosphatase activity to identify osteoclasts. Tartrate-resistant acid phosphatase-positive cells containing ≥ 3 nuclei were counted as mature osteoclasts.
Bone marrow cells or peripheral blood were freshly isolated, stained with various fluorescence-labeled antibodies, and subjected to fluorescence-activated cell sorting (FACS) analysis, as described previously [4,20].
Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and cDNA was synthesized by the RNA PCR Core Kit (Applied Biosystems, Branchburg, NJ, USA). Quantitative PCR amplification was performed with gene-specific primers using an iCycler iQ Multiple-Color Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA), as described previously .
The primer sequences are as follows: SDF-1, forward 5'-GCTCTGCATCAGTGACGG TA-3' and reverse 5'-TAATTACGGGTCAATGCACA-3' ; CXCR4, forward 5'-CTTTGTCATCACACTCC-CCTT-3' and reverse 5'-GCCCACATAGACTGCCT-TTTC-3' ; TGF-β, forward 5'-TCACTGGAGTTGTACGGCAG-3' and reverse 5'-TCTCTGTGGAGCTGAAGCAA-3' ; G-CSF, forward 5'-GCTGCTGCTGT-GGCAAAGT-3' and reverse 5'-AGCCTGACAGTGACCAGG-3' ; and actin, forward 5'-ACCCAGATCATGTTTGAGAC-3' and reverse 5'-GTCAGGATCTTCATGA-GGTAGT-3'.
A relative standard curve method was used to calculate the amplification efficiency. The standard curve was made from six points corresponding to 10-fold cDNA dilution series. For each sample, the relative amount was calculated from its respective standard curve. Standards and samples were run in triplicate.
Culture supernatants were collected from primary stromal cells and from the ST2 stromal cells. ELISA was performed with the Mouse SDF-1/CXCL12 DuoSet Development system. Ninety-six-well EIA/RIA plates (Costar, Corning, NY, USA) were coated with a capturing monoclonal antibody to SDF-1 and were then blocked with a mixture of 1% bovine serum albumin, 0.05% NaN3 and 5% sucrose in PBS. Culture supernatants were diluted in reagent diluent (1% bovine serum albumin in PBS) and incubated for 2 hours at room temperature. The detection antibody was diluted in reagent diluent and incubated for 2 hours at room temperature. Antibody binding was detected with streptavidin-conjugated horseradish peroxidase and developed with a substrate solution (1:1 mixture of H2O2 and tetramethylbenzidine).
A standard curve was generated for each set of samples assayed and was made from seven points of a twofold dilution series. Each standard or sample was assayed in duplicate.
Long bones from mice treated with TNF or PBS were fixed in 10% phosphate-buffered formalin, decalcified in 10% ethylenediamine tetraacetic acid and embedded in paraffin wax. Deparaffinized sections were quenched with 3% hydrogen peroxide and were treated for antigen retrieval for 30 minutes. Sections were then stained with a rabbit anti-SDF-1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and immunostaining was performed.
Cxcr4 floxed female mice were bred with CD11b+/Cre male mice to generate the Cxcr4+/f/CD11b+/Cre F1 generation. Cxcr4+/f/CD11b+/Cre male mice were then crossed with Cxcr4f/f female mice to produce Cxcr4f/f/CD11b+/Cre conditional knockout mice (CXCR4 CKO). Each litter comprised five to eight pups, indicating that deletion of CXCR4 in CD11b+ cells does not cause embryonic death. CXCR4 CKO mice were identified by PCR genotyping. The efficiency of CXCR4 deletion in the bone marrow CD11b+ cells was assessed by FACS analysis using FITC–anti-CD11b and allophycocyanin–anti-CXCR4 antibodies.
All results are presented as the mean ± standard error of the mean. Comparisons were made by analysis of variance and Student's t test for unpaired data. P < 0.05 was considered to represent statistical significance.
We and other workers have demonstrated that patients or mice with chronic inflammatory arthritis have an increased frequency of OCPs in peripheral blood and spleens, and that TNF promotes the release of bone marrow OCPs into the bloodstream [3-5]. To investigate whether the SDF-1/CXCR4 axis – the master chemokine system controlling mobilization of hematopoietic stem cells and progenitors – mediates TNF-induced OCP mobilization, we first verified that OCPs express functional CXCR4 and migrate toward a SDF-1 gradient in a combined chemotaxis/osteoclastogenesis assay. M-CSF-dependent bone marrow mononuclear cells were generated in vitro and were used as the source of OCPs. To confirm that these cells are enriched for OCPs, we compared their surface expression of CD11b and Gr-1 proteins, cell surface markers for OCPs , with primary bone marrow mononuclear cells isolated from the same mice.
As we reported previously , more than 10% of primary bone marrow cells are CD11b+/Gr-1-/lo. After 3 days of culture with M-CSF, more than 90% of adherent bone marrow cells become CD11b+/Gr-1-/lo – cells indicating enrichment of OCPs (Figure (Figure1a).1a). We then demonstrated that these OCPs migrated to SDF-1 gradients in a dose-dependent manner (Figure (Figure1b,1b, left panel). The maximum chemotaxic response was observed at 100 ng/ml SDF-1 and did not increase further with up to 250 ng/ml SDF-1 (data not shown). No migration occurred when SDF-1 was included in both the upper and lower chambers.
To confirm their osteoclast forming potency, we cultured the OCPs that had migrated to the SDF-1 gradient with M-CSF and RANKL, and demonstrated that they formed tartrate-resistant acid phosphatase-positive osteoclasts (Figure (Figure1b,1b, right panel). We also cultured nonmigrated OCPs from the upper chamber with M-CSF and RANKL, and compared their osteoclast forming potency with those that have migrated to SDF-1 gradient (100 ng/ml) in the lower chamber. Both nonmigrated and migrated OCPs can give rise to mature osteoclasts, but the cells from the lower chamber formed more osteoclasts (Figure (Figure1c,1c, left panel). In contrast, cells that were freely migrated to the lower chamber without a SDF-1 gradient did not form osteoclasts under the same condition (Figure (Figure1c,1c, right panel). These findings suggest that both nonmigrated and SDF-1 migrated cells can differentiate into osteoclasts but that CXCR4-positive cells have more osteoclast forming potency.
To study the effect of SDF-1 on OCP differentiation and activation, OCPs were cultured with M-CSF and RANKL in the presence or absence of SDF-1 (200 ng/ml) for 9 days on bone slices. SDF-1 did not affect osteoclast numbers, but slightly increased osteoclast resorptive activity (Figure (Figure1d).1d). SDF-1 had no effect on OCP production of TNF. In contrast, RANKL significantly increased TNF expression under the same culture conditions (fold induction of TNF over PBS: RANKL, 11.6 ± 0.9 versus SDF-1, 0.7 ± 0.1). The major role of SDF-1 in the regulation OCPs therefore appears to affect their mobilization through chemotaxis.
Since an SDF-1 gradient determines the direction of mobilization of hematopoietic stem cells and progenitors , we examined whether SDF-1 levels are decreased in bone marrow stromal cells and long bone samples from TNF-Tg mice to account for the increased OCP mobilization from their bone marrow to their peripheral blood. SDF-1 mRNA and protein levels were significantly reduced in the bone marrow stromal cells (Figure (Figure2a)2a) and in the long bones from TNF-Tg mice compared with wild-type littermates (Figure (Figure2b2b).
To examine whether TNF directly affects SDF-1 production, we treated ST2 cells – a bone marrow-derived cell line – with TNF, and found that SDF-1 expression decreased within 8 hours and with a relatively low dose of TNF (0.1 ng/ml) (Figure (Figure3a).3a). TNF-reduced SDF-1 production was also confirmed at protein levels (Figure (Figure3a).3a). Other osteoclastogenic cytokines, including IL-1 and RANKL, had no effect on SDF-1 expression (Figure (Figure3b),3b), while TGFβ significantly reduced SDF-1 mRNA expression, as reported previously .
To determine whether the reduction in bone marrow expression of SDF-1 induced by TNF leads to OCP mobilization, we treated wild-type mice with TNF using a subcutaneous injection protocol shown previously to increase the OCP frequency in the blood [4,20]. As expected, TNF increased the blood OCP numbers (Figure (Figure4a).4a). It also decreased SDF-1 protein levels in bone marrow extracts (Figure (Figure4b).4b). The concentration of SDF-1 in bone marrow was thus reduced significantly. Consistent with the SDF-1 protein data, SDF-1 mRNA expression was significantly decreased in the bone marrow cells of TNF-treated mice (Figure (Figure4c).4c). As a control, TGFβ mRNA levels did not change in the same samples (data not shown). Immunostaining with an anti-SDF-1 antibody showed that SDF-1 is strongly expressed by osteoblasts on endosteal and trabecular bone surfaces of murine long bones (Figure (Figure5,5, arrows). TNF treatment was associated with loss of SDF-1-positive staining of these cells without affecting the cell morphology (Figure (Figure5,5, arrow heads), indicating that TNF inhibits SDF-1 production by marrow osteoblasts.
CXCR4 is the sole receptor for SDF-1, and CXCR4 knockout mice die during embryonic development due to impaired cell homing in the bone marrow . To determine whether the number of CXCR4+ cells is altered in TNF-Tg mice, the percentage of bone marrow CXCR4+/CD11b+/Gr-1-/lo OCPs was examined by FACS analysis. No difference was observed in the percentage of CXCR4+/CD11b+/Gr-1-/lo cells between TNF-Tg mice and wild-type littermates (data not shown). TNF pretreatment of wild-type OCPs in vitro had no effect on CXCR4 expression on the cell surface (Figure (Figure6a),6a), and OCP migration to SDF-1 gradients was similar between PBS-pretreated and TNF-pretreated cells (Figure (Figure6b).6b). TNF therefore does not appear to influence the expression of CXCR4 by OCPs.
To determine whether specific deletion of CXCR4 protein in the OCPs affects TNF-induced OCP mobilization, we generated Cxcr4f/f/CD11b+/Cre conditional knockout (CXCR4 CKO) mice. FACS analysis of bone marrow CD11b+cells from adult CXCR4 CKO mice indicate that more CD11b+ cells from CXCR4 CKO mice are CXCR4-negative (49% in CXCR4 CKO mice versus 22% in control mice; Figure Figure7b).7b). We administered TNF (0.5 μg/injection, 4/day for 3 days) to CXCR4 CKO mice and their Cxcr4f/f/CD11b-/Cre control mice, and assessed the blood OCP frequency by FACS analysis. No clear difference in the percentage of CD11b+/Gr-1-/lo OCPs between TNF-treated CXCR4 CKO mice and control littermates was observed (Figure (Figure77).
Increased numbers of OCPs have been reported in the peripheral blood of mice in several animal models of arthritis [4,5] and in patients with arthritis , but the mechanisms that mediate this increase have not been elucidated. In the present study, we investigated whether the SDF-1/CXCR4 axis is involved in TNF-mediated OCP mobilization because this chemokine system plays an essential role in hematopoietic stem cell and progenitor homing . We found that TNF directly inhibits bone marrow stromal cell production of SDF-1 and reduces SDF-1 levels in the bone marrow, which is accompanied with an increase in the egress of OCPs from the marrow. Decreased SDF-1 production by bone marrow stromal cells in response to TNF overexpression may therefore be one of the mechanisms mediating release of OCPs to the peripheral blood in mice with TNF-induced arthritis or in patients with inflammatory arthritis.
SDF-1-regulated cell mobilization is determined by local SDF-1 gradients and/or CXCR4 expression on target cells. Although alternation of either of these could lead to impaired cell mobilization and homing, external factor regulation of SDF-1 expression levels appears to be the major mechanism. For example, hypoxia , DNA damage , proteases  and cytokines – including TGFβ  and G-CSF  – all reduce SDF-1 levels and stimulate hematopoietic stem cell release from bone marrow. Regulation of CXCR4 expression by external factors has been studied less and the results have been inconsistent. This inconsistency may be related to small numbers of CXCR4-expressing cells and low expression levels by these cells, making it difficult to reliably detect a change in the number of CXCR4-positive cells.
Our findings that TNF significantly decreases SDF-1 levels but has little effect on OCP CXCR4 expression suggest that, like most hematopoietic cell mobilizers, TNF also promotes OCP mobilization through regulation of SDF-1 rather than through CXCR4 expression. TNF-mediated OCP mobilization, however, is different from stem cell and precursor mobilization induced by SDF-1, G-CSF or other agents because TNF also has a strong stimulatory effect on OCP generation. This represents a unique pathologic situation in chronic inflammatory arthritis, in that the entire process of generation of OCPs and their egress from the bone marrow is accelerated in response to TNF. This situation leads to increased numbers of OCPs in both bone marrow and blood, whereas SDF-1 or G-CSF administration triggers a rapid release of cells from the bone marrow – and the total bone marrow cell number is consequently reduced.
We do not currently know the molecular mechanisms by which TNF inhibits SDF-1 production. SDF-1 is regulated at both transcriptional and post-translational levels [16,21]. We found that TNF induced massive apoptosis of ST2 cells when a transcription or translation inhibitor was used with TNF (data not shown). In these circumstances it is therefore difficult to investigate the mechanism of action for TNF. Protease degradation is one of the major mechanisms to reduce SDF-1 protein levels , and protease release from neutrophils and other myeloid cells can be stimulated by TNF. However TNF may also inhibit SDF-1 expression at the RNA level within 8 hours of treatment as shown by our data (Figure (Figure3a3a).
TGFβ at concentrations as low as 0.01 ng/ml decreases SDF-1 mRNA expression in stromal cells , implying that a small change in TGFβ could alter SDF-1 concentrations. We found that TNF increases TGFβ mRNA expression in ST2 cells. TNF administration to wild-type mice had no effect on TGFβ expression, however, although it significantly decreased SDF-1 expression in bone marrow stromal cells. Therefore it is unlikely that TGFβ mediates TNF-induced bone marrow SDF-1 downregulation in vivo. G-CSF is another cytokine that downregulates SDF-1 mRNA expression in osteoblasts . TNF did not increase G-CSF in ST2 cells (data not shown), however, suggesting that the reduction in SDF-1 induced by TNF in vitro is not mediated by G-CSF. Furthermore, the SDF-1 promoter does not contain binding sites typically present in the other CXC chemokine promoters, especially for NF-κB, interferon regulatory factor recognition elements or NF-IL6, which are associated with transcriptional activation in response to proinflammatory extracellular signals, such as TNF, IL-6 or interferons . These data suggest that studying SDF-1 regulation may be more complicated than studying other CXC chemokines.
The present study did not provide a direct association between TNF-reduced SDF-1 production and OCP mobilization in vivo. We have attempted to answer this question using mice with CXCR4 specifically deleted in OCPs by generating CXCR4 CKO mice via crossing CXCR4 floxed mice  with CD11b-Cre mice . We injected TNF to these CXCR4 conditional mice to determine whether TNF-induced increased OCP release is altered when CXCR4 expression has theoretically been deleted in CD11b-expressing OCPs. Unfortunately, we found that only about 50% of bone marrow CD11b+ cells have no CXCR4 surface expression in these CXCR4 CKO mice (Figure (Figure7a),7a), suggesting a low excision frequency of the Cre recombinase in our system. With this leaky system, the blood OCP frequency was similar between CXCR4 CKO mice and wild-type mice (Figure (Figure7b).7b). Our results suggest that CD11b-Cre mice appear not a good system to delete the gene encoding cxcr4 in bone marrow CD11b-positive cells.
The importance of TNF-mediated reduction in SDF-1 production in increased OCP mobilization in vivo needs to be further confirmed using a model where SDF-1 concentration in the bone marrow is maintained in the presence of TNF. Since rheumatoid arthritis and other forms of inflammatory bone disorders are chronic diseases, however, multiple factors may contribute to promote OCP release from the bone marrow. For example, we have demonstrated that TNF-stimulated OCP formation could increase the OCP pool in bone marrow and push cell egression . Kindle and colleagues reported that TNF activates endothelial cells and increases the attachment of OCPs to vascular endothelium in vitro. They speculated that this could increase the ability of OCPs to enter the bloodstream . It has been reported recently that RANKL-stimulated osteoclastogenesis promotes the mobilization of hematopoietic progenitor cells by cleaving SDF-1 through bone-resorbing proteinase, cathepsin K . TNF stimulates osteoclastogenesis synergistically with RANKL , and this mechanism may also apply to TNF-induced OCP mobilization. The regulation of OCP mobilization is therefore a complicated process, and decreased SDF-1 expression by bone marrow stromal cells may represent another important mechanism.
Our findings demonstrate that TNF directly inhibits bone marrow stromal cells to produce SDF-1, which is associated with increased release of OCPs from the bone marrow. The SDF1/CXCR4 axis therefore may not only control hematopoietic cell homing, but may also contribute to the accelerated OCP mobilization in inflammatory arthritis where systemic TNF levels are elevated.
CKO = conditional knockout; ELISA = enzyme-linked immunosorbent assay; FACS = fluorescence-activated cell sorting; FITC = fluorescein isothiocyanate; G-CSF = granulocyte colony-stimulating factor; IL = interleukin; M-CSF = macrophage colony-stimulating factor; NF = nuclear factor; OCPs = osteoclast precursors; PBS = phosphate-buffered saline PCR = polymerase chain reaction; RANKL = receptor activator of nuclear factor-B ligand; RT = reverse transcriptase; SDF-1 = stromal cell-derived factor 1; Tg = transgenic; TGFβ = transforming growth factor beta; TNF = tumor necrosis factor.
The authors declare that they have no competing interests.
LX had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study design was by LX, QZ, EMS, and BFB. LX, QZ, and RG were responsible for acquisition of data. Analysis and interpretation of data were performed by LX, QZ, RG, EMS, and BFB. LX, EMS, BFB, and QZ prepared the manuscript. Statistical analysis was performed by QZ and RG.
The murine TNF used in this study was provided by Amgen Inc. The authors thank Dr YR Zou (Columbia University College of Physicians and Surgeons, New York, USA) for providing breeders of Cxcr4 floxed mice, Dr J Vacher (Institut de Recherches Cliniques de Montréal, Québec, Canada) for providing CD11b+/Cre mice, Ms Xiaoyun Zhang for technical assistance with the histology, and Yan Lu for assistance with osteoclast bone resorption assay. The present work is supported by research grants from the National Institute of Health (PHS AR 48697 to LX and AR43510 to BFB).