CXCL2 has been known to regulate immune functions mainly by chemo-attracting neutrophils. In this study, we show that CXCL2 can be induced by receptor activator of NF-κB ligand, the osteoclast (OC) differentiation factor, through JNK and NF-κB signaling pathways in OC precursor cells. CXCL2 in turn enhanced the proliferation of OC precursor cells of bone marrow-derived macrophages (BMMs) through the activation of ERK. Knockdown of CXCL2 inhibited both the proliferation of and the ERK activation in BMMs. During osteoclastogenesis CXCL2 stimulated the adhesion and the migration of BMMs. Moreover, the formation of OCs from BMMs was significantly increased on treatment with CXCL2. Conversely, the CXCL2 antagonist repertaxin and a CXCL2 neutralizing Ab potently reduced receptor activator of NF-κB ligand-induced osteoclastogenesis. Furthermore, CXCL2 evoked fulminant bone erosion in the in vivo mouse experiments. Finally, prominent upregulation of CXCL2 was detected in synovial fluids and sera from rheumatoid arthritis patients, suggesting a potential involvement of CXCL2-mediated osteoclastogenesis in rheumatoid arthritis-associated bone destruction. Thus, CXCL2 is a novel therapeutic target for inflammatory bone destructive diseases.

Bone is a dynamic tissue that is continuously formed and replaced via a process called bone remodeling. Osteoblasts, derived from mesenchymal stem cells, produce the bone matrix, and osteoclasts (OCs), originated from hematopoietic progenitor cells, dissolve (resorb) bone (1). The generation of OCs from progenitor cells is governed by the differentiation factor receptor activator of NF-κB ligand (RANKL) with the assistance of macrophage CSF (M-CSF) that supports cell survival during differentiation (2). A series of steps, including OC precursor development, migration and adhesion of precursors to bone surfaces, differentiation into mature OCs, and the secretion of protons and lysosomal enzymes into the resorption site regulate osteoclastic bone removal (3).

Some chemokines have been shown to be associated with the generation and function of OCs. RANKL induces the expression of C-C chemokines such as CCL2, CCL3, CCL5, and CCL9, and C-X-C chemokines, such as CXCL10 (47). CCL2/MCP-1, and CCL5/RANTES, induced by RANKL in human OCs, could promote the formation of tartrate-resistant acid phosphatase (TRAP)-positive multinucleated OCs without RANKL (4). CCL2 also rescues bone resorption suppressed by granulocyte M-CSF (4). RANKL upregulates the expression of CCL3 and its receptor CCR1 in an NFATc1-dependent manner, which enhances migration of differentiating OCs (5). CCL9/MIP-1γ stimulates cytoplasmic motility and cell spreading of OCs, and CXCL12/SDF-1 increases osteoclastogenesis and resorption activity (6, 8). CXCL10/IP-10, induced by RANKL, plays critical roles in bone-erosive experimental arthritis via a CD4+ T cell-mediated mechanism (7).

CXCL2 was first identified as a major chemokine produced by endotoxin-treated macrophages and acts as a mediator of inflammation (9). The principal roles of CXCL2 are chemotaxis of neutrophils (10), regulation of endotoxin-induced transmigration and extravascular tissue accumulation of leukocytes (10), proliferation and apoptosis protection of hepatocytes (10, 11), and regulation of ischemia/reperfusion-induced leukocyte adhesion (12).

In the current study, we show that the osteoclastogenic factor RANKL could promote CXCL2 expression in OC precursors. We further provide evidence that CXCL2 has critical roles in osteoclastogenesis in vitro and in bone erosion in vivo. Therefore, targeting CXCL2 might be a new therapeutic strategy for antiresorptive drug development.

Recombinant RANKL, M-CSF, and CXCL2 were purchased from Peprotech (London, U.K.). Polyclonal CXCL2 Ab was purchased from Abcam (Cambridge, U.K.). Monoclonal anti-CXCR2 Ab was purchased from R&D Systems (Minneapolis, MN). PE-conjugated anti-RANK Ab was from Biolegend (San Diego, CA). Anti–c-Jun and anti-p65 Abs for chromatin immunoprecipitation (ChIP) assays were from Upstate (Billerica, MA) and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. Polyclonal Abs against phospho-ERK, phospho-JNK, phospho-p38, phospho-Akt, and IκB were purchased from Cell Signaling Technology (Cambridge, MA). Anti-mouse and anti-rabbit IgG-conjugated HRP and anti-mouse actin Abs were obtained from Sigma-Aldrich (St Louis, MO). FITC-labeled mouse secondary Ab was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). Mouse and human CXCL2 ELISA kits were from R&D Systems and Peprotech, respectively. SB203580, SP600125, PD98059, LY294002, repertaxin, and BrdU assay kits were purchased from Calbiochem (San Diego, CA). Parthenolide (PAR) and Bay11-7082 were purchased from Alexis Biochemicals (Grünberg, Germany). The luciferase assay system kit was purchased from Promega (Madison, WI). Lipofectamine 2000 and Stealth RNAi were from Invitrogen (Carlsbad, CA). Transwell was purchased from Corning Costar (Cambridge, MA). Cy5-labeled phalloidin and the TRAP staining kit were obtained from Sigma-Aldrich. Materials for real-time PCR were purchased from Applied Biosystems (Foster City, CA).

Synovial fluid and serum were drawn from 25 patients with rheumatoid arthritis (RA) and from 16 patients with osteoarthritis (OA) after obtaining informed consent. All samples were stored at −70°C before use. The study was approved by the ethics committee of Seoul National University Hospital.

Mouse bone marrow cells were isolated by flushing the bone marrow space of femora and tibiae of 6- to 8-wk-old ICR mice (Orient Bio, Seongnam, Korea) (13). Cells were incubated with α-MEM (JBI, Daegu, Korea) containing 10% FBS, penicillin (100 U/ml) and streptomycin (100 μg/ml) for 24 h in 5% CO2 at 37C. Nonadherent cells were plated on noncoated petri dishes with M-CSF (20 ng/ml) for 3 d. Adherent cells were considered to be bone marrow-derived macrophages (BMMs) and were used as OC precursor cells. OC differentiation was induced by treatment of BMMs with M-CSF (20 ng/ml) and RANKL (150 ng/ml). After culturing for 2 d, most cells became TRAP-positive mononucleated prefusion OCs (pOCs). TRAP-positive multinucleated OCs were formed after culturing for 5 d.

RNA (3 μg) extracted using an RNeasy mini kit (Qiagen, Valancia, CA) were reverse transcribed into cDNA using Superscript II (Invitrogen) in a total reaction volume of 20 μl. PCR was performed with 1 μl cDNA using gene-specific oligonucleotide primers; mCCL2, 5′-TCCCAATGAGTAGGCTGGAG-3′ and 5′-TCTGGACCCATTCCTTCTTG-3′; mCCL3, 5′-AGCCAGGTGTCATTTTCCTG-3′ and 5′-CTCAAGCCCCTGCTCTACAC-3′; mCCL4, 5′-CCTGACCAAAAGAGGCAGAC-3′ and 5′-GAGGAGGCCTCTCCTGAAGT-3′; mCCL5, 5′-GTGCCCACGTCAAGGAGTAT-3′ and 5′-CACTTCTTCTCTGGGTTGGC-3′; mCXCL2, 5′-ACAGAAGTCATAGCCACTCTC-3′ and 5′-CCTTGCCTTTGTTCAGTATC-3′; mCXCL10, 5′-CCCACGTGTTGAGATCATTG-3′ and 5′-GCTCTCTGCTGTCCATCCAT-3′; mCXCL11, 5′-AAGTCACGTGCACACTCCAC-3′ and 5′-ACAACGCAGAAATGAATCGT-3′; mGAPDH, 5′-AGGTCATCCCAGAGCTGAACG-3′ and 5′-CACCCTGTTGCTGTAGCCGTAT-3′; mCXCR2, 5′-CCTGGAAATCAACAGTTATGCTG-3′ and 5′-TCCTTCACGTATGAGAATATCTTGC-3′. PCR cycles were carried out for 15 s at 95°C and 60 s at 60°C. For CXCR2, PCR amplification was performed 30 s at 95°C, 30 s at 56°C, and 30 s at 72°C. PCR products were separated in a 2% agarose gel and stained with ethidium bromide. Real-time PCR was performed using an ABI prism 7500 Sequence Detection System with SYBRGreen PCR Master Mix. The PCR reaction was carried out for 40 thermal cycles. Expression of the target gene was analyzed by an absolute quantification method and normalized using GAPDH levels.

BMMs (5 × 105 cells per well in 6-well plates) were treated with M-CSF (20 ng/ml), RANKL (150 ng/ml), or M-CSF plus RANKL for 6 h. To detect CXCL2, 50 μl of the 1 ml culture supernatant was used for Western blotting (14). After treating BMMs with CXCL2 (50 ng/ml) in the absence or presence of M-CSF (10 ng/ml) for the indicated times, cell lysates were subjected to Western blotting to detect ERK and Akt phosphorylation. BMMs transfected with control- or CXCL2-small interfering RNA (siRNA) were also analyzed.

CXCL2 levels in synovial fluid and serum from patients with RA and OA, as well as levels in cell culture supernatants, were measured using human and mouse CXCL2 ELISA kits, following the manufacture’s instructions.

RAW264.7 cells were transfected with the CXCL2-luciferase construct (15) for 5 h in serum-free DMEM. After 24 h of incubation with DMEM containing 10% FBS, cells were stimulated with increasing doses of RANKL for 5 h. Cells were lysed with Glo lysis buffer and were subjected to a luciferase assay using a FLUOstar OPTIMA luminometer (BMG Labtech, Offenburg, Germany).

BMMs were treated with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) for 1 h. The procedure for ChIP assays was based on the description in the EZ ChIP kit (17-371, Upstate). The 5 μg anti–c-Jun or anti-p65 was used for immunoprecipitation. After eluting DNA from the precipitated immune complex, PCR reaction was performed using specific primers: c-Jun, 5′-CTCGTGCTCAGTACACCGCA-3′ and 5′-GGGAACTTTAGTCATTAGGACTGA-3′; p65, 5′-AACCCACTCAGCTTAGGGGC-3′ and 5′-TTGT-TGGAGGCACTGAGGC-3′. For the input control, 1% of the sonicated DNA was directly purifed before immunoprecipitation and subjected to PCR with the same primers.

To detect the surface expression of CXCR2, cells were incubated with monoclonal anti-mouse CXCR2 Ab for 20 min on ice after blocking nonspecific binding by treating with goat serum. Cells were washed with PBS three times before incubation with FITC-labeled secondary Ab for 20 min. To detect the surface expression of RANK, BMMs were incubate with PE-conjugated anti-RANK for 30 min. Flow cytometric analysis was performed using a FACSCalibur flow cytometer (BectonDickinson, San Jose, CA). Acquisition and analysis were performed using CellQuest software.

CXCL2 gene-specific double-stranded stealth RNAi was designed using software from Invitrogen (si-CXCL2 no. 1, UUGAAGUCAACCCUUGG-CAGGGUCU; si-CXCL2 no. 2, AGACAGAAGTCATAGCCACTCTCAA; and si-CXCL2 no. 3, GAACTGCGCTGTCAATGCCTGAAGA). BMMs were transfected with CXCL2-siRNA (20 nM) using lipofectamin 2000 after the siRNA transfection protocol described by the manufacturer and were cultured with M-CSF (10 ng/ml) for 24 h.

BMMs were cultured with increasing doses of CXCL2 in the presence of M-CSF (10 ng/ml) for 24 h. In experiments with transfected BMMs, cells were cultured with M-CSF (10 ng/ml) for 24 h and retreated with M-CSF (10 ng/ml) or M-CSF (10 ng/ml) plus RANKL (200 ng/ml) for 24 h. BrdU was incorporated for the last 2–4 h of the treatments. The BrdU assay was performed, following the manufacture’s instruction.

CXCL2-induced migrations of BMMs and pOCs were measured using transwell plates with 5-μM pore membranes. Serum-deprived 100 μl cell suspension (2 × 106 cells/ml) was added to the upper chamber. Increasing doses of recombinant CXCL2 in serum-free medium were added to the bottom chamber. After 3 h, nonmigrated cells were wiped off with a cotton swab and migrated cells were counted under the light microscope after fixing and staining with H&E. Adhesion assays were performed on the coverslip and on 96-well cell culture plates. BMMs and pOCs suspended in serum-free media were added onto the coverslip or cell culture plates in the presence of increasing doses of CXCL2. After vigorously washing off nonadherent cells three times with PBS, adherent cells were fixed and stained with phalloidin-Cy5 or H&E. Adherent cells were observed under either the confocal microscope (Olympus-FV300, ×100) or a light microscope (Olympus-CK30, ×100).

BMMs were cultured with M-CSF (20 ng/ml), RANKL (150 ng/ml), and increasing doses of CXCL2 for 6 d on dentin slices. After cells were removed by washing with distilled water, dentin slices were stained with H&E and photographed under a microscope. The pit areas of the dentin slices were determined from the photographed images using SPOT software version 4.6 (Diagnostic Instruments, Sterling Heights, MI).

Freeze-dried collagen was soaked with PBS or RANKL (5 μg) and transplanted onto the calvariae of 5- to 7-wk-old female ICR mice. Recombinant mouse CXCL2 (5 μg) was injected 1 d before transplantation, day 2, and day 4. At day 5, the mice were sacrificed, and calvariae were subjected to microcomputer tomography (μ-CT) analysis (1072 microtomograph, Skyscan Satellite Systems, Riverside, CA). For the neutralizing experiment, control or CXCL2 Ab (5 μg) was administered s.c. day 1 and day 3. At day 5, the mice were sacrificed, and calvariae were subjected to μ-CT analysis. The bone volume was assessed using CTAN software. Animal experiments were approved by the Institutional Animal Care and Use Committee of Seoul National University.

All experiments were repeated at least three times and are presented as the mean ± SD. The significance of the results was analyzed using Student t test or Mann-Whitney U test with p < 0.05 being considered significant.

We screened chemokines that could be induced by the osteoclastogenic factor RANKL in mouse primary OC precursor cells, BMMs. The expressions of CCL3 and CCL4 were slightly increased, whereas those of CCL5 and CXCL2, 10, and 11 were greatly induced by 150 ng/ml RANKL treatment for 3 h (Fig. 1A). In contrast to a previous report (4), CCL2 mRNA expression did not change in the presence of RANKL (Fig. 1A). Because the role of CXCL2 in osteoclastogenesis has not been addressed, we focused on CXCL2 for further studies. The induction of CXCL2 by RANKL was dependent on the concentrations of RANKL (Fig. 1B). Real-time PCR analysis showed that RANKL, not M-CSF, elicited CXCL2 mRNA induction in BMMs stimulated for 6 h (Fig. 1C). Interestingly, M-CSF and RANKL caused a synergistic increase in CXCL2 expression (Fig. 1C). However, M-CSF alone could also increase CXCL2 mRNA on long-hour treatment (data not shown). We examined the level of CXCL2 protein secretion in BMM culture medium. Similar synergistic increase in CXCL2 protein levels was detected by Western blotting and ELISA ana-lyses of BMM culture supernatants (Fig. 1D). The secretion of CXCL2 by BMMs was evident as early as 1 h after M-CSF and RANKL treatments, showing maximal stimulation at 24 h (Fig. 1E). The 24-h treatment of M-CSF could also elicit CXCL2 secretion, but to a much lower level (data not shown).

FIGURE 1.

Induction of chemokines in BMMs by RANKL. A, BMMs were treated with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) for the indicated times. mRNA expression for CCL2, 3, 4, 5, and CXCL2, 10, and 11 was analyzed by RT-PCR. B, CXCL2 was detected by RT-PCR after treatment with increasing doses of RANKL for 3 h. Hypoxanthine phosphoribosyltransferase was used as a loading control. C, BMMs were cultured with M-CSF (20 ng/ml), or RANKL (150 ng/ml), or M-CSF plus RANKL for 6 h. mRNA for CXCL2 was analyzed by real-time PCR. The fold induction of CXCL2 mRNA versus untreated BMMs is presented. D, BMMs were cultured in 6-well plates with M-CSF (20 ng/ml), RANKL (150 ng/ml), or M-CSF plus RANKL for 6 h. A total of 50 μl culture supernatant of 1 ml was subjected to Western blotting to detect secreted CXCL2 (toppanel). The amount of CXCL2 secreted into culture medium was also measured by ELISA (bottompanel). E, After BMMs were treated with M-CSF (20 ng/ml) plus RANKL (150 ng/ml), cultured media were collected at indicated times and subjected to ELISA. Data are expressed as means ± SD. *p < 0.05 versus control.

FIGURE 1.

Induction of chemokines in BMMs by RANKL. A, BMMs were treated with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) for the indicated times. mRNA expression for CCL2, 3, 4, 5, and CXCL2, 10, and 11 was analyzed by RT-PCR. B, CXCL2 was detected by RT-PCR after treatment with increasing doses of RANKL for 3 h. Hypoxanthine phosphoribosyltransferase was used as a loading control. C, BMMs were cultured with M-CSF (20 ng/ml), or RANKL (150 ng/ml), or M-CSF plus RANKL for 6 h. mRNA for CXCL2 was analyzed by real-time PCR. The fold induction of CXCL2 mRNA versus untreated BMMs is presented. D, BMMs were cultured in 6-well plates with M-CSF (20 ng/ml), RANKL (150 ng/ml), or M-CSF plus RANKL for 6 h. A total of 50 μl culture supernatant of 1 ml was subjected to Western blotting to detect secreted CXCL2 (toppanel). The amount of CXCL2 secreted into culture medium was also measured by ELISA (bottompanel). E, After BMMs were treated with M-CSF (20 ng/ml) plus RANKL (150 ng/ml), cultured media were collected at indicated times and subjected to ELISA. Data are expressed as means ± SD. *p < 0.05 versus control.

Close modal

To further delineate mechanisms for the induction of CXCL2, we next investigated the transcriptional regulation of CXCL2 promoter. The CXCL2 promoter-dependent luciferase activity was assessed in RAW264.7 macrophages. RANKL moderately increased the promoter activity, whereas cotreatment of M-CSF and RANKL further elevated the luciferase activity (Fig. 2A, 2B). The possibility of regulation of CXCL2 mRNA stability was also explored. As shown in Fig. 2C, mRNA stability of CXCL2 was increased by M-CSF plus RANKL (Fig. 2C). When we tested the single treatment of RANKL or M-CSF, CXCL2 mRNA stabilization was greater by RANKL than by M-CSF (data not shown). We next examined whether RANKL-induced CXCL2 expression would require new synthesis of other proteins. Interestingly, CXCL2 mRNA expression was not suppressed, rather increased, by the protein synthesis inhibitor cycloheximide (Fig. 2D), suggesting that RANKL directly stimulates CXCL2 expression and that RANKL might also promote synthesis of other proteins inhibitory to CXCL2 induction. To gain insight into the signaling pathways involved in RANKL-dependent CXCL2 induction, BMMs were stimulated with M-CSF plus RANKL in the presence of chemical inhibitors of MAPK, PI3K/Akt, and NF-κB pathways as those signaling molecules were activated by M-CSF plus RANKL (data not shown). CXCL2 mRNA induction by M-CSF and RANKL was almost completely abolished by the JNK inhibitor, SP600125, and NF-κB inhibitors, Bay 11-7082 and PAR, whereas the p38 inhibitor, SB203580, and the PI3K inhibitor, LY294002, slightly reduced CXCL2 mRNA expression (Fig. 2E). Similarly, JNK and NF-κB inhibitors abolished the upregulation of CXCL2 protein by M-CSF and RANKL in BMMs (Fig. 2F). The recruitment of JNK-dependent transcription factor c-Jun and the NF-κB subunit p65 to the CXCL2 promoter was also found to be stimulated by M-CSF plus RANKL (Fig. 2G, 2H). Taken together, these results demonstrate that JNK and NF-κB signaling pathways play a major role in RANKL-induced CXCL2 expression.

FIGURE 2.

Induction of CXCL2 by RANKL through JNK and NF-κB signaling pathways. A and B, RAW264.7 cells were transfected with a CXCL2 promoter construct and cultured with M-CSF (20 ng/ml) plus RANKL (200 ng/ml) (A) or increasing doses of RANKL (B) for 6 h. Cell lysates were analyzed for luciferase activity. The fold activation represents the ratio of luciferase activity in stimulated cells versus unstimulated cells. C, BMMs were cultured with M-CSF (20 ng/ml) or M-CSF plus RANKL (150 ng/ml) for 3 h. Actinomycin D (1 μg/ml) with or without M-CSF plus RANKL was subsequently to the cells for the indicated times. CXCL2 mRNA levels were determined by RT-PCR. GAPDH was used as a loading control (toppanel). The band intensity of RT-PCR products was determined using image J program and expressed as the percentage relative to the mRNA level in cells before actinomycin D addition. D, BMMs were treated with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) in the presence or absence of cycloheximide (1 μg/ml) for the indicated times. CXCL2 mRNA was detected by RT-PCR. GAPDH was used as a loading control. E and F, BMMs were cultured with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) in the presence of indicated inhibitors for 6 h. A total of 20 μM SB (p38 inhibitor), SP (JNK inhibitor), PD (MEK1 inhibitor), LY (PI3K inhibitor), or 10 μM BAY or PAR (both NF-κB inhibitors) were used. CXCL2 levels were analyzed by real-time PCR (E) and ELISA (F). SB, SB203580; SP, SP600125; PD, PD98059; LY, LY294002; BAY, Bay11-7082. G and H, Recruitment of c-Jun and p65 transcription factors to CXCL2 promoter. BMMs were treated with M-CSF plus RANKL for one h. The cross-linked chromatin was immunoprecipitatied with c-Jun or p65 Ab. PCR was performed using specific primers for the consensus binding site of each transcription factor. 1% chromatin before immunoprecipitation was used as an input control. Data are expressed as means ± SD. *p < 0.05 versus control.

FIGURE 2.

Induction of CXCL2 by RANKL through JNK and NF-κB signaling pathways. A and B, RAW264.7 cells were transfected with a CXCL2 promoter construct and cultured with M-CSF (20 ng/ml) plus RANKL (200 ng/ml) (A) or increasing doses of RANKL (B) for 6 h. Cell lysates were analyzed for luciferase activity. The fold activation represents the ratio of luciferase activity in stimulated cells versus unstimulated cells. C, BMMs were cultured with M-CSF (20 ng/ml) or M-CSF plus RANKL (150 ng/ml) for 3 h. Actinomycin D (1 μg/ml) with or without M-CSF plus RANKL was subsequently to the cells for the indicated times. CXCL2 mRNA levels were determined by RT-PCR. GAPDH was used as a loading control (toppanel). The band intensity of RT-PCR products was determined using image J program and expressed as the percentage relative to the mRNA level in cells before actinomycin D addition. D, BMMs were treated with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) in the presence or absence of cycloheximide (1 μg/ml) for the indicated times. CXCL2 mRNA was detected by RT-PCR. GAPDH was used as a loading control. E and F, BMMs were cultured with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) in the presence of indicated inhibitors for 6 h. A total of 20 μM SB (p38 inhibitor), SP (JNK inhibitor), PD (MEK1 inhibitor), LY (PI3K inhibitor), or 10 μM BAY or PAR (both NF-κB inhibitors) were used. CXCL2 levels were analyzed by real-time PCR (E) and ELISA (F). SB, SB203580; SP, SP600125; PD, PD98059; LY, LY294002; BAY, Bay11-7082. G and H, Recruitment of c-Jun and p65 transcription factors to CXCL2 promoter. BMMs were treated with M-CSF plus RANKL for one h. The cross-linked chromatin was immunoprecipitatied with c-Jun or p65 Ab. PCR was performed using specific primers for the consensus binding site of each transcription factor. 1% chromatin before immunoprecipitation was used as an input control. Data are expressed as means ± SD. *p < 0.05 versus control.

Close modal

CXCR2 is known to be the major receptor for CXCL2. FACS analysis using anti-CXCR2 Ab revealed that BMMs expressed CXCR2 at a high level (Fig. 3A). Thus, we hypothesized that CXCL2 released on RANKL stimulation of BMMs could function in an autocrine manner. To explore this possibility, we first investigated the effect of CXCL2 on the proliferation of BMMs. During the early stage of osteoclastogenesis, enhanced proliferation was observed when BMMs were cultured with M-CSF plus RANKL compared with that of M-CSF alone (Fig. 3C). To test whether RANKL-induced CXCL2 production is involved in the proliferation, BMMs were transfected with CXCL2-specific siRNA oligonucleotides. Two functional (no. 1 and no. 3) and one nonfunctional (no. 2) siRNAs were used (Fig. 3B, data not shown). CXCL2 knockdown significantly reduced BMM proliferation on M-CSF stimulation or M-CSF plus RANKL treatment (Fig. 3C, data not shown). As the ERK and Akt signaling pathways have been implicated in cell proliferation, we next examined the effects of CXCL2 siRNA on ERK and Akt phosphorylation. Although RANKL stimulation of BMMs induced sustained activation of Akt and ERK, knockdown of CXCL2 inhibited ERK phosphorylation (Fig. 3D, data not shown). The activation of Akt was indistinguishable between si-control and si-CXCL2 transfected cells. Next, we tested whether CXCL2 could augment M-CSF–induced BMM proliferation in the absence of RANKL. As shown in Fig. 3E, the addition of CXCL2 significantly elevated BMM proliferation in the presence of M-CSF. Again, CXCL2 enhanced sustained ERK phosphorylation, with little effect on Akt activation in BMMs treated with M-CSF (Fig. 3F).

FIGURE 3.

Proliferation of OC precursors by CXCL2. A, The surface expression of CXCR2 in BMMs was analyzed by flow cytometry as described in 1Materials and Methods. The black line indicates the negative control with the primary Ab omitted. B, BMMs were transfected with control or CXCL2-siRNA for 24 h, followed by stimulation with M-CSF (20 ng/ml) and RANKL (150 ng/ml) for 6 h. CXCL2 expression was determined by RT-PCR (bands presented in the upperpanel), and real-time PCR (histogram). C, Control or CXCL2-siRNA transfected BMMs were cultured with M-CSF (10 ng/ml) for 24 h. The BrdU assay was performed when the culture media were changed and cells were retreated with M-CSF (10 ng/ml) or M-CSF (10 ng/ml) plus RANKL (200 ng/ml) for 24 h. D, BMMs were transfected with control or CXCL2-siRNA for 24 h. Cell lysates were subjected to Western blotting to determine the activations of ERK and Akt after incubation with M-CSF (10 ng/ml) and RANKL (150 ng/ml) for 24 h. E, BMMs were cultured with M-CSF (10 ng/ml) and increasing doses of CXCL2 for 24 h. The proliferation of BMMs was determined by BrdU assay. F, BMMs were cultured with M-CSF (10 ng/ml) or combination of CXCL2 (50 ng/ml) and RANKL (150 ng/ml) for 24 h. The activations of ERK and Akt were assessed by Western blotting using respective phospho-ERK and phospho-Akt specific Abs. Actin was used as a loading control. Data are expressed as means ± SD. *p < 0.05 versus control.

FIGURE 3.

Proliferation of OC precursors by CXCL2. A, The surface expression of CXCR2 in BMMs was analyzed by flow cytometry as described in 1Materials and Methods. The black line indicates the negative control with the primary Ab omitted. B, BMMs were transfected with control or CXCL2-siRNA for 24 h, followed by stimulation with M-CSF (20 ng/ml) and RANKL (150 ng/ml) for 6 h. CXCL2 expression was determined by RT-PCR (bands presented in the upperpanel), and real-time PCR (histogram). C, Control or CXCL2-siRNA transfected BMMs were cultured with M-CSF (10 ng/ml) for 24 h. The BrdU assay was performed when the culture media were changed and cells were retreated with M-CSF (10 ng/ml) or M-CSF (10 ng/ml) plus RANKL (200 ng/ml) for 24 h. D, BMMs were transfected with control or CXCL2-siRNA for 24 h. Cell lysates were subjected to Western blotting to determine the activations of ERK and Akt after incubation with M-CSF (10 ng/ml) and RANKL (150 ng/ml) for 24 h. E, BMMs were cultured with M-CSF (10 ng/ml) and increasing doses of CXCL2 for 24 h. The proliferation of BMMs was determined by BrdU assay. F, BMMs were cultured with M-CSF (10 ng/ml) or combination of CXCL2 (50 ng/ml) and RANKL (150 ng/ml) for 24 h. The activations of ERK and Akt were assessed by Western blotting using respective phospho-ERK and phospho-Akt specific Abs. Actin was used as a loading control. Data are expressed as means ± SD. *p < 0.05 versus control.

Close modal

CXCL2 has been suggested as a chemoattractant for various types of cells (10). Thus, the effect of CXCL2 on the migration of OC precursors was examined using the transwell system. CXCL2 sig-nificantly increased the number of migrating BMMs (Fig. 4A). In a bone microenvironment, adhesion of OC precursors is another critical step during the early stages of osteoclastogenesis. Notably, increasing concentrations of CXCL2 dramatically stimulated the adhesion of BMMs on the glass cover slips (Fig. 4B), as well as on tissue culture plastics (data not shown). In addition, CXCL2 treatment at higher concentrations enhanced the cell spreading of the BMMs (data not shown).

FIGURE 4.

Migration and adhesion of OC precursors by CXCL2. A, A migration assay was performed using the transwell system. Serum-deprived BMMs were added to the upper chamber and serum-free α-MEM containing CXCL2 was added to the lower chamber. After a 3-h incubation, migrated cells were fixed, stained with H&E, and counted under a microscope. B, CXCL2 was added to serum-starved BMMs on the cover slip for 10 min. Nonadherent cells were washed vigorously with PBS. Adherent cells were fixed, stained with phalloidin-Cy5, and observed under a confocal microscope (toppanel). Mean fluorescence intensity was measured using an LSM program (bottompanel). Data are expressed as means ± SD. *p < 0.05 versus untreated control.

FIGURE 4.

Migration and adhesion of OC precursors by CXCL2. A, A migration assay was performed using the transwell system. Serum-deprived BMMs were added to the upper chamber and serum-free α-MEM containing CXCL2 was added to the lower chamber. After a 3-h incubation, migrated cells were fixed, stained with H&E, and counted under a microscope. B, CXCL2 was added to serum-starved BMMs on the cover slip for 10 min. Nonadherent cells were washed vigorously with PBS. Adherent cells were fixed, stained with phalloidin-Cy5, and observed under a confocal microscope (toppanel). Mean fluorescence intensity was measured using an LSM program (bottompanel). Data are expressed as means ± SD. *p < 0.05 versus untreated control.

Close modal

We next investigated the effect of CXCL2 on OC differentiation from mouse BMMs. When CXCL2 was added to BMMs in osteo-clastogenic medium containing RANKL and M-CSF, the generation of TRAP-positive multinucleated OCs was enhanced (Fig. 5A, 5B). Moreover, the size of the OCs formed was also increased by CXCL2 (Fig. 5A), and the number of nuclei per cell was also higher in CXCL2-treated OCs (Fig. 5C). To further confirm the role of CXCL2 during RANKL-induced OC differentiation, CXCL2-neutralizing Ab was included during in vitro osteoclastogenesis. As shown in Fig. 5D, addition of a CXCL2-neutralizing Ab almost completely inhibited OC differentiation. In addition, the CXCR2-specific antagonist repertaxin also blocked osteoclastogenesis (Fig. 5E). These data indicate that the CXCL2 induced by RANKL in OC precursors acts on its receptor in an autocrine fashion to augment osteoclastogenesis. To examine whether the OCs generated in the presence of CXCL2 are functionally competent, we performed bone resorption assays. BMMs were cultured with M-CSF plus RANKL in the presence of increasing doses of CXCL2 on dentin slices. The resorption pit areas were increased in CXCL2-treated dentine slices (Fig. 6A, 6B). We next investigated whether CXCL2 could also regulate the survival of differentiated OCs. It has been reported that RANKL sustains the survival of mature OCs (16). When a neutralizing CXCL2 Ab was added in the presence of RANKL to differentiated mature OCs, the number of live cells was significantly decreased (Fig. 6C).

FIGURE 5.

Enhanced OC differentiation by CXCL2. A, BMMs were cultured with M-CSF (10 ng/ml) and RANKL (150 ng/ml) in the presence of CXCL2 during the whole period of osteoclastogenesis. Cells were fixed and stained for TRAP. B, TRAP-positive multinucleated OCs shown in A were counted. C, The number of nuclei per OC shown in A was counted. D, BMMs were cultured with M-CSF (20 ng/ml) and RANKL (150 ng/ml) with a CXCL2 neutralizing Ab during the whole period of osteoclastogenesis. Isotype-matched Ig was used as a control. The number of TRAP-positive multinucleated OCs was counted. E, BMMs were treated with M-CSF (20 ng/ml) and RANKL (150 ng/ml) in the absence or presence of the CXCR2 antagonist repertaxin (2 μM). After 5 d, TRAP-positive multinucleated OCs were counted. Data are expressed as means ± SD. *p < 0.05 versus control.

FIGURE 5.

Enhanced OC differentiation by CXCL2. A, BMMs were cultured with M-CSF (10 ng/ml) and RANKL (150 ng/ml) in the presence of CXCL2 during the whole period of osteoclastogenesis. Cells were fixed and stained for TRAP. B, TRAP-positive multinucleated OCs shown in A were counted. C, The number of nuclei per OC shown in A was counted. D, BMMs were cultured with M-CSF (20 ng/ml) and RANKL (150 ng/ml) with a CXCL2 neutralizing Ab during the whole period of osteoclastogenesis. Isotype-matched Ig was used as a control. The number of TRAP-positive multinucleated OCs was counted. E, BMMs were treated with M-CSF (20 ng/ml) and RANKL (150 ng/ml) in the absence or presence of the CXCR2 antagonist repertaxin (2 μM). After 5 d, TRAP-positive multinucleated OCs were counted. Data are expressed as means ± SD. *p < 0.05 versus control.

Close modal
FIGURE 6.

Effect of CXCL2 on function and survival of OCs. A, Resorption assay was performed with BMMs cultured with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) and increasing doses of CXCL2 for 6 d. B, Pit formation shown in A was analyzed using SPOT software after staining with H&E. C, Mature OCs were treated with or without RANKL for overnight. RANKL-treated OCs were cultured in the presence of increasing doses of CXCL2 Ab. Survived OCs were shown after staining with TRAP (upper panel). The number of live OCs was counted (bottompanel). Data are expressed as means ± SD. *p < 0.05 versus control.

FIGURE 6.

Effect of CXCL2 on function and survival of OCs. A, Resorption assay was performed with BMMs cultured with M-CSF (20 ng/ml) plus RANKL (150 ng/ml) and increasing doses of CXCL2 for 6 d. B, Pit formation shown in A was analyzed using SPOT software after staining with H&E. C, Mature OCs were treated with or without RANKL for overnight. RANKL-treated OCs were cultured in the presence of increasing doses of CXCL2 Ab. Survived OCs were shown after staining with TRAP (upper panel). The number of live OCs was counted (bottompanel). Data are expressed as means ± SD. *p < 0.05 versus control.

Close modal

We investigated the effect of CXCL2 on bone resorption in vivo. PBS- or RANKL-soaked collagen sponges were implanted onto mice calvariae, and CXCL2 or a control vehicle was injected three times onto the calvariae to assess bone erosion. The μ-CT analyses revealed that RANKL significantly increased bone loss in calvariae (they appeared black in the μ-CT image) compared with PBS (Fig. 7A). Notably, the addition of CXCL2 elicited dramatic bone loss both in the absence and the presence of RANKL (Fig. 7B). To confirm the involvement of CXCL2 in the RANKL-induced bone resorption, a CXCL2 neutralizing Ab was administered into calvariae. The CXCL2 Ab significantly reduced the bone loss induced by RANKL (Fig. 7C, 7D). To gain further insights into the role of CXCL2 in bone-destructive diseases, we measured CXCL2 levels in synovial fluids and sera from RA patients. RA synovial fluids contained significantly higher levels of CXCL2 compared with those of OA (Fig. 7E). A similar increase in the CXCL2 protein levels was also evident in serum samples from RA patients (Fig. 7F). These data suggest that increased CXCL2 in synovial fluids might have a role in bone destruction during RA pathogenesis.

FIGURE 7.

In vivo bone erosion by CXCL2. A, Mice calvariae that received PBS, RANKL, and CXCL2 were subjected to μ-CT analysis. Three-dimensonal reconstructed images are shown. B, The bone volume of calvariae was analyzed with CTAN software. C, Mice calvariae that received RANKL plus a control Ab or a CXCL2 Ab (5 μg/mouse) were subjected to μ-CT analysis. D, The bone volume of calvariae shown in C was analyzed with CTAN software. E and F, CXCL2 was measured by ELISA in synovial fluids (C) and sera (D) from 25 patients with RA and 16 patients with OA. Data are expressed as means ± SD. *p < 0.05 between indicated groups.

FIGURE 7.

In vivo bone erosion by CXCL2. A, Mice calvariae that received PBS, RANKL, and CXCL2 were subjected to μ-CT analysis. Three-dimensonal reconstructed images are shown. B, The bone volume of calvariae was analyzed with CTAN software. C, Mice calvariae that received RANKL plus a control Ab or a CXCL2 Ab (5 μg/mouse) were subjected to μ-CT analysis. D, The bone volume of calvariae shown in C was analyzed with CTAN software. E and F, CXCL2 was measured by ELISA in synovial fluids (C) and sera (D) from 25 patients with RA and 16 patients with OA. Data are expressed as means ± SD. *p < 0.05 between indicated groups.

Close modal

The pathogenesis of bone destruction in RA is initiated by the upregulation of RANKL as well as TNF-α and IL-1 in the synovial tissue. These cytokines induce other cytokines that promote the differentiation and activation of OCs and recruit immune cells, leading to massive destruction of cartilages and bones (17). To uncover the role of chemokines during this process, we investigated the regulation of chemokines by the osteoclastogenic factor RANKL in primary OC precursor cells. RANKL rapidly induced the C-C chemokines CCL3, CCL4, and CCL5 as well as the C-X-C chemokine CXCL10 (Fig. 1). A number of studies have shown that some of these chemokines and their receptors play roles in OC differentiation and activation (48). Interestingly, RANKL also significantly increased the expression of CXCL2 that has never been associated with OCs. In this study, we clearly showed that the RANKL-induced CXCL2 in OC precursors play key roles in the attachment, migration, differentiation, and func-tion of OCs. Because mouse bone marrow OC precursors (BMMs) highly expressed the CXCL2 receptor CXCR2 (Fig. 3), it was likely that the CXCL2-dependent augmentation of osteoclastogenesis was mediated via an autocrine/paracrine mechanism.

RANKL seemed to increase the mRNA level of CXCL2 not only by direct stimulation of CXCL2 promoter activity, but also by stabilizing the mRNA. Although the CXCL2 promoter-dependent luciferase activity was only 1.5-fold enhanced by RANKL, there was >4-fold induction of CXCL2 mRNA (Figs. 1, 2). Indeed, the presence of RANKL significantly delayed the degradation of CXCL2 mRNA (Fig. 2). Interestingly, inhibition of protein synthesis by cycloheximide seemed to increase CXCL2 mRNA level, suggesting that there might exist RANKL-induced protein(s) that negatively regulate CXCL2 mRNA. Further studies are required to reveal whether these proteins are repressors that regulate transcription, or involved in the mRNA degradation.

CXCL2 induction by RANKL in OC precursors was significantly impaired in the presence of JNK and NF-κB inhibitors (Fig. 2). In fact, c-Jun and p65 transcription factors were recruited to CXCL2 promoter by RANKL stimulation. These data concur with a previous report that showed the existence of two AP-1 binding sites and one NF-κB binding site in the promoter region of the mouse CXCL2 gene (15). Because c-Jun and NF-κB–dependent pathways also regulated LPS-dependent CXCL2 induction (15), it is likely that the same signaling mechanisms are shared by TLR and RANK receptors in terms of CXCL2 production. Interestingly, we found that p38 and PI3K inhibitors slightly inhibited CXCL2 mRNA expression, but they did not affect CXCL2 protein production. The extent of mRNA reduction by the p38 and PI3K inhibitors might have not been sufficient to mitigate CXCL2 protein levels. Alternatively, these inhibitors may have additional suppressive effects on a mechanism that destabilizes the CXCL2 protein.

Osteoclastogenesis requires proper regulation of sequential processes, including the proliferation, adhesion, migration, and fusion of precursors and differentiating cells. Our data showed that CXCL2 promoted the proliferation of BMMs (Fig. 3). The knockdown of CXCL2 significantly reduced BMM proliferation and ERK activation on M-CSF or RANKL stimulation, whereas exogenous CXCL2 augmented both proliferation and ERK activity. A similar ERK-dependent stimulation of cell proliferation by CXCL2 was observed in esophageal cancers that highly expressed both CXCL2 and CXCR2 compared with adjacent normal esophageal tissues (18). CXCL2 has been shown to potently chemo-attract neutrophils (19). Similarly, CXCL2 not only stimulated the adhesion but also induced the migration of OC precursors in our study (Fig. 4). Moreover, cell spreading was enhanced by CXCL2 treatment (data not shown). Thus, CXCL2 likely plays a crucial role by regulating various cellular responses during the early stages of osteoclastogenesis. Indeed, CXCL2-neutralizing Ab and the CXCR2 antagonist repertaxin almost completely inhibited osteoclastogenesis when they were included during the early stages of osteoclastogenic culture (Fig. 5). In addition, a significant increase in the number of nuclei per OC was observed by CXCL2 treatment (Fig. 5), suggesting that CXCL2 might also be involved in the fusion of OC precursors. Furthermore, CXCL2 increased the number of active OCs (Fig. 6). Finally, CXCL2 blockade by neutralizing Ab significantly decreased OC survival. All these data suggest that the RANKL-induced production of CXCL2 has influence on various cellular responses involved in OC differentiation and function in vitro.

The osteoclastogenesis-enhancing effects of CXCL2 were further corroborated by our investigation with an in vivo bone resorption model. Notably, CXCL2 alone induced significant bone loss in mice calvariae similar to that by RANKL (Fig. 7). Because CXCL2 alone in the absence of RANKL did not induce OC differentiation in vitro, this might be explained by the CXCL2 stimulation of osteoclastogenesis from pre-existed RANKL-primed precursors. However, we cannot rule out the possibility that CXCL2 might have caused the recruitment of inflammatory immune cells that contributed to osteoclastogenesis. The RANKL-induced bone destruction was significantly blocked by administration of CXCL2 neutralizing Ab (Fig. 7). Therefore, it was likely that CXCL2 might be one of the important mediators in response to RANKL stimulation that promote bone destruction. In support of this notion, CXCL2 was significantly upregulated in synovial fluids and sera from RA patients (Fig. 7). Combined with the data demonstrating the ability of CXCL2 to evoke bone resorption in mice calvariae, our study suggests that CXCL2 might be involved in bone destruction during the pathogenesis of RA. Therapies targeting CXCL2/CXCR2 have been tested in animal models of arthritis with concomitant reduction in neutrophil recruitment and TNF-α production (20), and immunization against CXCL2 was efficient in delaying the onset of arthritis and reducing the disease severity in a murine collagen-induced arthritis model (21). In addition, an antagonist of CXCR2 inhibited arthritis in rabbits (22). Our results imply that targeting CXCL2 as a strategy to treat RA would be beneficial in protection from bone destruction by directly inhibiting bone resorption, in addition to the already suggested anti-inflammatory effects.

In conclusion, RANKL induced CXCL2 in OC precursors in both JNK- and NF-κB–dependent mechanisms. CXCL2 promoted the proliferation, adhesion, and migration of OC precursors and augmented the formation of OCs in vitro. Furthermore, CXCL2 induced significant bone destruction on in vivo administration, and its level was found to be high in RA patient fluids. These results uncover a previously unappreciated role of CXCL2 during osteoclastogenesis and suggest CXCL2 blockade as a novel strategy for the treatment of inflammatory bone destructive diseases.

Disclosures The authors have no financial conflicts of interest.

This work was supported by grants from the 21C Frontier Functional Proteomics Project (FPR08B1-170) and Science Research Center (2009-0063269) to H-H. K.

Abbreviations used in this paper:

BMM

bone marrow-derived macrophages

ChIP

chromatin immunoprecipitation

M-CSF

macrophage CSF

μ-CT

microcomputer tomography

OA

osteoarthritis

OC

osteoclast

PAR

parthenolide

RA

rheumatoid arthritis

pOC

prefusion OC

RANKL

receptor activator of NF-κB ligand

siRNA

small interfering RNA

TRAP

tartrate-resistant acid phosphatase.

1
Hadjidakis
D. J.
,
Androulakis
I. I.
.
2006
.
Bone remodeling.
Ann. N. Y. Acad. Sci.
1092
:
385
396
.
2
Suda
T.
,
Takahashi
N.
,
Udagawa
N.
,
Jimi
E.
,
Gillespie
M. T.
,
Martin
T. J.
.
1999
.
Modulation of osteoclast differentiation and function by the new members of the tumor necrosis factor receptor and ligand families.
Endocr. Rev.
20
:
345
357
.
3
Nakamura
I.
,
Rodan
G. A.
,
Duong
T.
.
2003
.
Regulatory mechanism of osteoclast activation.
J. Electron Microsc. (Tokyo)
52
:
527
533
.
4
Kim
M. S.
,
Day
C. J.
,
Morrison
N. A.
.
2005
.
MCP-1 is induced by receptor activator of nuclear factor-kappaB ligand, promotes human osteoclast fusion, and rescues granulocyte macrophage colony-stimulating factor suppression of osteoclast formation.
J. Biol. Chem.
280
:
16163
16169
.
5
Ishida
N.
,
Hayashi
K.
,
Hattori
A.
,
Yogo
K.
,
Kimura
T.
,
Takeya
T.
.
2006
.
CCR1 acts downstream of NFAT2 in osteoclastogenesis and enhances cell migration.
J. Bone Miner. Res.
21
:
48
57
.
6
Lean
J. M.
,
Murphy
C.
,
Fuller
K.
,
Chambers
T. J.
.
2002
.
CCL9/MIP-1gamma and its receptor CCR1 are the major chemokine ligand/receptor species expressed by osteoclasts.
J. Cell. Biochem.
87
:
386
393
.
7
Kwak
H. B.
,
Ha
H.
,
Kim
H. N.
,
Lee
J. H.
,
Kim
H. S.
,
Lee
S.
,
Kim
H. M.
,
Kim
J. Y.
,
Kim
H. H.
,
Song
Y. W.
,
Lee
Z. H.
.
2008
.
Reciprocal cross-talk between RANKL and interferon-gamma-inducible protein 10 is responsible for bone-erosive experimental arthritis.
Arthritis Rheum.
58
:
1332
1342
.
8
De Klerck
B.
,
Geboes
L.
,
Hatse
S.
,
Kelchtermans
H.
,
Meyvis
Y.
,
Vermeire
K.
,
Bridger
G.
,
Billiau
A.
,
Schols
D.
,
Matthys
P.
.
2005
.
Pro-inflammatory properties of stromal cell-derived factor-1 (CXCL12) in collagen-induced arthritis.
Arthritis Res. Ther.
7
:
R1208
R1220
.
9
Wolpe
S. D.
,
Cerami
A.
.
1989
.
Macrophage inflammatory proteins 1 and 2: members of a novel superfamily of cytokines.
FASEB J.
3
:
2565
2573
.
10
Li
X.
,
Klintman
D.
,
Liu
Q.
,
Sato
T.
,
Jeppsson
B.
,
Thorlacius
H.
.
2004
.
Critical role of CXC chemokines in endotoxemic liver injury in mice.
J. Leukoc. Biol.
75
:
443
452
.
11
Ren
X.
,
Carpenter
A.
,
Hogaboam
C.
,
Colletti
L.
.
2003
.
Mitogenic properties of endogenous and pharmacological doses of macrophage inflammatory protein-2 after 70% hepatectomy in the mouse.
Am. J. Pathol.
163
:
563
570
.
12
Riaz
A. A.
,
Schramm
R.
,
Sato
T.
,
Menger
M. D.
,
Jeppsson
B.
,
Thorlacius
H.
.
2003
.
Oxygen radical-dependent expression of CXC chemokines regulate ischemia/reperfusion-induced leukocyte adhesion in the mouse colon.
Free Radic. Biol. Med.
35
:
782
789
.
13
Ha
J.
,
Choi
H. S.
,
Lee
Y.
,
Lee
Z. H.
,
Kim
H. H.
.
2009
.
Caffeic acid phenethyl ester inhibits osteoclastogenesis by suppressing NF kappaB and downregulating NFATc1 and c-Fos.
Int. Immunopharmacol.
9
:
774
780
.
14
Chang
E. J.
,
Ha
J.
,
Huang
H.
,
Kim
H. J.
,
Woo
J. H.
,
Lee
Y.
,
Lee
Z. H.
,
Kim
J. H.
,
Kim
H. H.
.
2008
.
The JNK-dependent CaMK pathway restrains the reversion of committed cells during osteoclast differentiation.
J. Cell Sci.
121
:
2555
2564
.
15
Kim
D. S.
,
Han
J. H.
,
Kwon
H. J.
.
2003
.
NF-kappaB and c-Jun-dependent regulation of macrophage inflammatory protein-2 gene expression in response to lipopolysaccharide in RAW 264.7 cells.
Mol. Immunol.
40
:
633
643
.
16
Jimi
E.
,
Akiyama
S.
,
Tsurukai
T.
,
Okahashi
N.
,
Kobayashi
K.
,
Udagawa
N.
,
Nishihara
T.
,
Takahashi
N.
,
Suda
T.
.
1999
.
Osteoclast differentiation factor acts as a multifunctional regulator in murine osteoclast differentiation and function.
J. Immunol.
163
:
434
442
.
17
Boyce
B. F.
,
Schwarz
E. M.
,
Xing
L.
.
2006
.
Osteoclast precursors: cytokine-stimulated immunomodulators of inflammatory bone disease.
Curr. Opin. Rheumatol.
18
:
427
432
.
18
Wang
B.
,
Hendricks
D. T.
,
Wamunyokoli
F.
,
Parker
M. I.
.
2006
.
A growth-related oncogene/CXC chemokine receptor 2 autocrine loop contributes to cellular proliferation in esophageal cancer.
Cancer Res.
66
:
3071
3077
.
19
Wang
L.
,
Fuster
M.
,
Sriramarao
P.
,
Esko
J. D.
.
2005
.
Endothelial heparan sulfate deficiency impairs L-selectin- and chemokine-mediated neutrophil trafficking during inflammatory responses.
Nat. Immunol.
6
:
902
910
.
20
Coelho
F. M.
,
Pinho
V.
,
Amaral
F. A.
,
Sachs
D.
,
Costa
V. V.
,
Rodrigues
D. H.
,
Vieira
A. T.
,
Silva
T. A.
,
Souza
D. G.
,
Bertini
R.
, et al
.
2008
.
The chemokine receptors CXCR1/CXCR2 modulate antigen-induced arthritis by regulating adhesion of neutrophils to the synovial microvasculature.
Arthritis Rheum.
58
:
2329
2337
.
21
Kasama
T.
,
Strieter
R. M.
,
Lukacs
N. W.
,
Lincoln
P. M.
,
Burdick
M. D.
,
Kunkel
S. L.
.
1995
.
Interleukin-10 expression and chemokine regulation during the evolution of murine type II collagen-induced arthritis.
J. Clin. Invest.
95
:
2868
2876
.
22
Podolin
P. L.
,
Bolognese
B. J.
,
Foley
J. J.
,
Schmidt
D. B.
,
Buckley
P. T.
,
Widdowson
K. L.
,
Jin
Q.
,
White
J. R.
,
Lee
J. M.
,
Goodman
R. B.
, et al
.
2002
.
A potent and selective nonpeptide antagonist of CXCR2 inhibits acute and chronic models of arthritis in the rabbit.
J. Immunol.
169
:
6435
6444
.