MIP-1γ Promotes Receptor Activator of NF-κB Ligand-Induced Osteoclast Formation and Survival¹

Osteoclasts are bone-resorbing multinucleated giant cells that are derived from hemopoietic precursors of the monocyte-macrophage lineage. Osteoclastic bone resorption consists of multiple steps, including the differentiation of osteoclast precursors; the fusion of mononuclear cells to form mature multinucleated osteoclasts; activation to resorb bone; and finally, the survival of activated osteoclasts (1). Receptor activator of NF-κB ligand (RANKL),³ also known as TNF-related activation-induced cytokine or osteoprotegerin ligand, is a member of the TNF family and is one of key molecules that regulates both osteoclastogenesis and bone resorption (2, 3). RANKL expression by osteoblasts as well as by activated T cells has been shown to regulate these processes (3, 4). However, the participation of additional factors, including autocrine factors induced by RANKL stimulation, is less well characterized.

Chemokines play an important role in immune and inflammatory responses by inducing the migration and adhesion of leukocytes. It has been reported that several chemokines may regulate the migration and differentiation of osteoclasts, including MIP-1α and IL-8 (5, 6, 7). However, the cellular source(s) of these chemokines and their role in the overall regulation of bone mass remain unclear.

MIP-1γ is a C-C chemokine family member (8, 9). MIP-1γ induces the chemotaxis of CD4⁺ and CD8⁺ T cells and monocytes in vitro (10), and shows potent suppressive activity on the colony formation of murine bone marrow (BM) myeloid progenitor cells (11). MIP-1γ mRNA is widely expressed in most tissues of normal mice, except brain (9).

Using gene microarrays, we found that MIP-1γ mRNA expression was strongly up-regulated in RANKL-induced osteoclasts, suggesting its possible involvement in the regulation of this cell type. In this study, we report studies that indicate an important role for MIP-1γ in RANKL-induced osteoclast formation, survival, and activation in bone resorption, most likely via an autocrine pathway.

Three- to 5-wk-old BALB/c mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Mouse rRANKL and mouse rM-CSF were purchased from PeproTech (Rocky Hill, NJ). Recombinant mouse MIP-1γ, anti-mouse MIP-1γ Ab, and control IgG1 Ab were obtained from R&D Systems (Minneapolis, MN).

RAW264.7, a mouse macrophage/monocyte cell line, was purchased from American Type Culture Collection (Manassas, VA) (TIB-71). Cells were cultured in DMEM (JRH Biosciences, Lenexa, KS) supplemented with 10% FBS (Invitrogen Life Technologies, Grand Island, NY), 1.5 g/L sodium bicarbonate, and penicillin/streptomycin (Invitrogen Life Technologies). To generate osteoclasts, RAW264.7 cells were plated in 24-well plates at a density of 1 × 10⁴ cells/well. Cells were stimulated with 10 ng/ml mouse rRANKL for 5 days. Mouse BM cells were collected from femora and tibiae, as described (12). Briefly, 3- to 5-wk-old female mice were killed by cervical dislocation under light ether anesthesia. Femora and tibiae were dissected, and BM cells were flushed out and cultured in α-MEM (Cambrex, Walkersville, MD) supplemented with 10% FBS, 2.0 g/L sodium bicarbonate, and penicillin/streptomycin. BM cells were seeded into 24-well plates at a density of 2 × 10⁶ cells/well in medium supplemented with 20 ng/ml RANKL and 50 ng/ml mouse rM-CSF for 7 days. All cultures were maintained at 37°C in a humidified atmosphere of 5% CO₂ in air, with changes of medium every other day. Osteoclast numbers were evaluated by counting tartrate-resistant acid phosphatase (TRAP)-positive giant cells. At culture termination, cells were washed with PBS and fixed in 10% formalin for 5 min, followed by ethanol/acetone (1:1) for 1 min. Osteoclasts were stained for TRAP in the presence of 0.05 M sodium tartrate (Sigma-Aldrich, St. Louis, MO), napthol AS-MX phosphate (Sigma-Aldrich) as substrate, and fast red LB salt (Sigma-Aldrich). TRAP-positive multinuclear cells (three or more nuclei/cell) were counted under light microscopy.

Two array systems were used to detect differences in gene expression between undifferentiated precursor cells and osteoclasts: the Atlas Mouse 1.2 array (BD Clontech, Palo Alto, CA) and the MG-U74 chip (Affymetrix, Santa Clara, CA). To enrich for differentiated osteoclasts, RANKL-stimulated cultures were briefly trypsinized for 1 min to remove other less adherent cells. This treatment generated a population of >80% osteoclasts.

For analysis of gene expression in osteoclasts, total RNA was isolated from both undifferentiated cells and purified osteoclasts using TRIzol reagent (Invitrogen Life Technologies). Total RNA was subsequently treated with DNase I (Ambion, Austin, TX) to remove contaminating genomic DNA and quantified by spectrophotometry. The Atlas array was hybridized to a radioactively labeled mixed cDNA probe obtained by reverse transcription of 4 μg of total RNA, according to the manufacturer’s instructions. After hybridization, the arrays were washed to remove unbound probe and exposed to x-ray film. The level of gene expression was analyzed and normalized using NIH Image software. The MG-U74 chip was hybridized to a biotinylated mixed cDNA probe, washed, and stained according to the standard Affymetrix GeneChip protocol. The level of gene expression was analyzed and normalized using statistical algorithms provided by Affymetrix.

MIP-1γ (0.1, 0.5, 2.0 ng/ml) was added to cultures of RAW264.7 and BM cells to examine its effect on RANKL-induced osteoclast differentiation. To eliminate possible effects of contaminating LPS, 1 μg/ml polymyxin B (Sigma-Aldrich) was simultaneously added to some cultures. To assess the effect of endogenous MIP-1γ, anti-MIP-1γ Ab or IgG1 control Ab (0.5, 5.0 μg/ml) was added to cultures of RAW264.7 and BM cells stimulated with RANKL.

Chemokines were measured in culture supernatants and cell lysates of RAW264.7 and BM cells. After supernatants were collected, cells were washed with PBS and lysed in 500 μl of protein extraction buffer (0.5% Triton X-100, 50 mM Tris-HCl, 0.3 M NaCl, and 5 mM EDTA). MIP-1γ, MIP-1α, and RANTES levels were determined using commercially available ELISA kits (R&D Systems).

Semiquantitative RT-PCR was used to examine CCR1 gene expression in undifferentiated cells and purified osteoclasts. For this, 1 μg of total RNA was reverse transcribed using Superscript II (Invitrogen Life Technologies) and random primers, according to the manufacturer’s instructions. cDNA was subjected to PCR amplification with Taq polymerase (Qiagen, Valencia, CA) using specific mouse CCR1 primers: sense, 5′-gtgttcatcattggagtggtgg-3′; antisense, 5′-ggttgaacaggtagatgctggtc-3′ (13).

For cell survival analysis, osteoclasts were generated from RAW264.7 cells for 5 days. Adherent osteoclasts were washed extensively with PBS to completely remove RANKL. Cells were subsequently cultured without or with 10 ng/ml RANKL in the presence/absence of 2.0 ng/ml MIP-1γ. Neutralizing anti-MIP-1γ Ab (5 μg/ml) or control IgG1 Ab (5 μg/ml) was simultaneously added with RANKL to determine the role of MIP-1γ in apoptosis of RANKL-induced osteoclasts. After 24, 48, and 72 h, survival was determined by counting adherent TRAP-positive osteoclasts.

To determine the effect of MIP-1γ on bone-resorbing activity, osteoclasts were generated in three-dimensional collagen gels (Chemicon International, Temecula, CA). Dishes (60 mm) were covered with a collagen gel solution prepared according to the manufacturer’s instructions. RAW264.7 cells were seeded onto the gels and cultured with 10 ng/ml RANKL for 5 days. Cells were removed following digestion of gels with 1000 U/ml collagenase (Sigma-Aldrich) at 37°C for 30 min. Aliquots of the harvested cell suspension were seeded onto submicron calcium phosphate films (Osteologic; BD Biosciences, Bedford, MA) in 250 μl of medium. Cells were unstimulated (control), or stimulated with 10 μg of RANKL or MIP-1γ (2.0, 5.0, 10 ng/ml). After 24 h, the cells were removed with 5% sodium hypochlorite, and resorbed areas were visualized using light microscopy. The size of resorbed areas was quantified using NIH Image.

Osteoclasts, generated by RANKL stimulation of RAW264.7 cells for 5 days, were enriched by a brief trypsinization, which removed most mononuclear cells. Osteoclasts were washed twice with PBS, pH 7.4, followed by suspension in 800 μl of ice-cold lysis buffer (mmol/L: HEPES, 10; KCL, 10; EDTA, 0.1; EGTA, 0.1; DTT, 1.0; PMSF, 1.0; and 10 μg/ml aprotinin, 10 μg/ml pepstatin, and 10 μg/ml leupeptin). The collected samples were incubated on ice for 30 min, vortexed for 30 s after addition of 50 μl of 10% Nonidet P-40, and centrifuged for 10 min at 4°C at 5500 rpm. The nuclei-containing pellets were suspended in ice-cold buffer (mmol/L: HEPES, 20; NaCl, 400; EDTA, 1.0; EGTA, 1.0; DTT, 1.0; PMSF, 1.0; and 10 μg/ml aprotinin, 10 μg/ml pepstatin, and 10 μg/ml leupeptin), incubated on ice for 2 h with frequent mixing, and centrifuged for 10 min at 4°C at 14,000 rpm. The supernatants were collected as nuclear extract and stored at −70°C. The total protein concentration was determined using a protein assay kit (Pierce, Rockford, IL).

NF-κB-binding studies were performed using double-stranded oligonucleotides containing an NF-κB consensus binding site. The oligonucleotides were end labeled with [³²P]ATP using T4 polynucleotide kinase (Promega, Madison, WI) and incubated with the nuclear extract for 20 min at room temperature. The samples were loaded on a 4% nondenaturating polyacrylamide gel. After electrophoresis, the gel was dried and exposed to Kodak film.

Assays for caspase 3 activity and degradation of DNA (TUNEL) were used for detection of osteoclast apoptosis. Following RANKL stimulation of RAW264.7 cells for 5 days, osteoclasts were extensively washed, and restimulated with MIP-1γ or RANKL in the presence/absence of anti-MIP-1γ Ab. Twenty-four hours later, cells were fixed in 4% paraformaldehyde, and labeled with the caspase 3 substrate rhodamine 110 (100 μM) and 2.4 nM TOTO-3 for nuclear staining (Molecular Probes, Eugene, OR) at 37°C for 30 min. Cells were washed and viewed using a fluorescence microscope. Apoptotic cells exhibited bright green fluorescence.

Apoptosis was also assessed using TUNEL assays. Osteoclasts were derived from BM cells stimulated with RANKL and M-CSF for 7 days. Cultures were washed, and TUNEL assays were conducted using the In Situ Cell Death Detection kit, tetra-methyl-rhodamine red (Roche Applied Science, Indianapolis, IN), according to the manufacturer’s protocols. Apoptotic cells were identified by bright red fluorescence in this assay.

In all studies, differences between groups were analyzed using Student’s t test with the Bonferroni correction for multiple comparisons.

Osteoclast formation was induced by RANKL stimulation of RAW264.7 monocytes for 5 days, or by stimulation of normal mouse BM cells for 7 days with M-CSF and RANKL. As previously reported, RANKL-stimulated RAW264.7 and BM cells differentiated into TRAP-positive osteoclasts that expressed high levels of the osteoclast markers TRAP, cathepsin K, and the proton pump subunit ATP6I, and produced resorption pits on bone slices and calcium phosphate-coated slides (14).

Two gene array systems (Atlas and Affychip) were used to study gene expression following RANKL induction of osteoclast formation. Total RNA was extracted from RANKL-induced osteoclasts, and was used as a template to generate mixed cDNA probes. In the Atlas system (1100 genes), we observed a highly significant up-regulation of mRNA for the chemokine MIP-1γ in RANKL-stimulated osteoclasts derived from RAW264.7 cells, compared with unstimulated precursor cells (Table I). This result was confirmed and extended using the Affymetrix system (32,000 genes) and mRNA derived from osteoclasts induced from normal BM as well as RAW264.7 cells (Table I). The induction of MIP-1γ was more significant than any other chemokine or cytokine gene represented on these arrays, suggesting a possible role in osteoclast development and/or function.

MIP-1γ binds to CCR1, which is also activated by related C-C chemokines MIP-1α and RANTES. We therefore examined the production of MIP-1γ, MIP-1α, and RANTES proteins during the process of RANKL-induced osteoclastogenesis from RAW264.7 cells. As shown in Fig. 1 A, RANKL strongly stimulated the production of MIP-1γ in both cell lysates and culture supernatants over the 5-day culture period. MIP-1α was also induced by RANKL, as previously reported (15), albeit at 10-fold lower levels. Interestingly, the levels of MIP-1α in supernatants peaked by day 2, although cell-associated levels continued to increase to day 5. RANTES was also weakly induced by RANKL, but its levels were only 1–2% of those of MIP-1γ at any time point (maximum: 50 pg/ml).

MIP-1γ was similarly induced following RANKL stimulation of normal BM cells (Fig. 1 B). However, compared with RAW264.7 cells, RANTES was expressed initially at somewhat higher levels, but expression declined with increasing times after culture induction, and MIP-1α was nearly undetectable. Taken together, these results suggest that MIP-1γ is the predominant C-C chemokine produced by RANKL-stimulated precursor cells during osteoclastogenesis.

We next examined RAW264.7 and BM cells for the expression of CCR1 mRNA upon RANKL stimulation. As shown in Fig. 2, CCR1 mRNA was undetectable in unstimulated RAW264.7 cells, but was present at low levels in unstimulated BM cells. RANKL stimulation strongly induced CCR1 mRNA in RAW264.7 and, to a lesser extent, BM cells. Taken together, these results show that RANKL induces both MIP-1γ and its receptor during osteoclast differentiation, suggesting the operation of an autocrine pathway.

We then investigated the role of endogenously produced MIP-1γ in RANKL-induced osteoclastogenesis in RAW264.7 and normal BM cells. A neutralizing anti-MIP-1γ Ab was added to cultures beginning on day 0, and was replenished periodically throughout the osteoclast induction period. As seen in Fig. 3, in both cell systems, the addition of anti-MIP-1γ Ab resulted in a decreased number of TRAP-positive osteoclasts, relative to control IgG1 Ab. The reduction was ∼60% in cultures treated with 5 μg/ml anti-MIP-1γ Ab, and 45% in cultures treated with 0.5 μg/ml Ab. These findings were replicated in two additional experiments (data not shown).

The effect of adding exogenous rMIP-1γ on osteoclast formation was also determined. As seen in Fig. 4, TRAP-positive osteoclasts were not induced by MIP-1γ alone, nor did exogenous MIP-1γ have a synergistic effect with RANKL on osteoclastogenesis in either cell system. These results are perhaps not unexpected, given the high level of endogenous MIP-1γ production. Nevertheless, they indicate that endogenous MIP-1γ increases RANKL-induced osteoclast formation, but has no independent ability to induce osteoclastogenesis.

It has been reported that mature osteoclasts rapidly undergo apoptosis in the absence of bone-resorptive stimuli such as RANKL, LPS, or IL-1α (16). We therefore determined whether MIP-1γ might also play a role in maintaining osteoclast viability. In these experiments, RANKL was removed from cultures of differentiated osteoclasts by extensive washing of the cells on day 5. Cells were then recultured for an additional 24–72 h in the presence or absence of MIP-1γ or RANKL as a positive control. As shown in Fig. 5, the number of surviving osteoclasts was reduced by 90% after 24 h in the absence of a stimulating agent (medium alone). RANKL restimulation promoted osteoclast survival, as indicated by only a 30% reduction in viable TRAP-positive cells after 24 h. Interestingly, the addition of MIP-1γ alone also prevented cell death, albeit somewhat less effectively than RANKL itself. We also determined whether the survival-promoting effect of RANKL was dependent on its ability to induce MIP-1γ expression. As indicated (Fig. 5), the prosurvival activity of RANKL was reduced by ∼60% in the presence of anti-MIP-1γ Ab, whereas an isotype-matched control IgG1 Ab had no effect.

That the loss of osteoclasts was the result of apoptosis rather than necrosis was demonstrated using a fluorescence-based assay for the specific activity of caspase 3 (Fig. 6,A). Osteoclasts cultured in the presence of RANKL exhibited minimal apoptosis (panel 2), compared with cells from which RANKL was removed (panel 1). Cells cultured with MIP-1γ alone were partially protected from apoptosis (panel 3). The prosurvival effect of RANKL was again demonstrated to be partially dependent on its ability to induce MIP-1γ, as shown by anti-MIP-1γ Ab blockade of the protective effect of RANKL (panel 4 vs 5). These results were further confirmed in similar studies using osteoclasts derived from normal BM cells and TUNEL assays (Fig. 6 B).

Given that many of the factors that promote osteoclast survival, including RANKL, act by stimulating NF-κB, we examined the effect of MIP-1γ on this transcription factor. As shown in Fig. 6,C, following extensive washing of mature osteoclasts, RANKL restimulation strongly induced NF-κB DNA-binding activity in mature osteoclasts, as assessed by EMSA. MIP-1γ by itself also stimulated NF-κB, but less strongly than RANKL, which correlated with the level of its prosurvival activity (Fig. 6, A and B). Taken together, these results support the conclusion that a primary function of MIP-1γ may be to promote the survival of mature osteoclasts by preventing apoptosis.

Finally, we determined the effect of MIP-1γ on the bone-resorbing activity of mature osteoclasts. RANKL-induced RAW264.7 cells were cultured in three-dimensional collagen gels, isolated by enzymatic digestion, and replated onto submicron calcium phosphate films for an additional 24 h in the presence/absence of MIP-1γ or RANKL as a positive control. As shown in Fig. 7, osteoclasts were stimulated to form numerous resorption pits in the presence of added RANKL, but not in its absence. Of interest, the addition of MIP-1γ alone also resulted in a marked stimulation of resorption, to a level similar to that seen with RANKL-stimulated cells. These results suggest that MIP-1γ stimulates the activation as well as the survival of mature osteoclasts.

It is increasingly evident that extensive cross talk occurs between the immune and skeletal systems. In particular, the differentiation and activity of osteoclasts, and hence bone mass, can be modulated by an ever-expanding number of cytokines/chemokines, many of which derive from immune cells. In the present investigation, we used gene arrays to identify mediators that were up-regulated following RANKL stimulation of osteoclast precursor cells. Our data show that of all the mediators screened, MIP-1γ was the most strongly up-regulated in osteoclasts derived from RANKL-stimulated monocytes/macrophages. CCR1, the high affinity receptor for MIP-1γ, was also increased following RANKL stimulation. Inhibition of MIP-1γ resulted in decreased osteoclast formation and reduced resorptive activity. Furthermore, MIP-1γ promoted osteoclast survival and prevented apoptosis, and was responsible for a major proportion of the prosurvival activity of RANKL itself. To our knowledge, this is the first report that MIP-1γ may play an important role in regulating osteoclastic bone resorption, via effects on cell differentiation, survival, and activation.

MIP-1γ is a relatively recently described C-C chemokine with a predicted length of 100 aa, which is identical with CCF18 (8). MIP-1γ is constitutively expressed by a wide variety of tissues (9), and exclusively binds to CCR1 on mouse neutrophils in vitro (9, 11). MIP-1γ was recently reported to be increased in rat BM cells stimulated with RANKL, although its function was not determined (17). Two other C-C chemokines, MIP-1α and RANTES, also bind to CCR1 (18), and are therefore potentially able to modulate osteoclast development and function. However, we found that MIP-1γ was by far the predominant chemokine produced by RANKL-stimulated RAW264.7 and BM cultures, compared with relatively minor amounts of MIP-1α and RANTES. Additionally, the function of MIP-1α and RANTES in osteoclasts is somewhat controversial. Fuller et al. (5) found that MIP-1α and IL-8 inhibited the resorption of isolated rat osteoclasts, and concluded that these chemokines participate in osteoclast migration, but not resorption and survival. Conversely, it was reported that levels of MIP-1α, which is produced and secreted by osteoblasts, correlated with an increase in the number of osteoclasts in a porcine BM culture system (6). Han et al. (15) also demonstrated that MIP-1α stimulated osteoclast formation in human marrow cultures and also enhanced formation induced by parathyroid hormone-related peptide and RANKL. It was also reported that the RANTES gene was up-regulated in RANKL-induced osteoclasts derived from mouse BM cells using gene arrays (19). Given our finding that neutralization of MIP-1γ reduced osteoclast formation by ∼60%, it remains possible that MIP-1α and RANTES could also participate in RANKL-induced osteoclastogenesis following interaction with CCR1, albeit with a more modest effect. Alternatively, residual MIP-1γ-independent RANKL-induced osteoclast formation may proceed via pathways independent of these chemokines.

Most of our findings with RAW264.7 cells were replicated using normal BM cells, confirming the role of MIP-1γ in a nontransformed osteoclast precursor (i.e., macrophage) population. At the same time, a concern in using BM cells is that they are heterogeneous and contain a number of cell types that may express RANKL following activation, including T cells, B cells, and osteoblasts (2, 20, 21, 22). We examined the phenotype of nonosteoclastic cells after 7 days’ culture, and found that T cells (CD3⁺, 0.7%), B cells (CD19⁺, 0.5%), and osteoblasts/stromal cells (alkaline phosphatase positive, <0.5%) constituted only minor components after 7 days in culture (H. Sasaki, unpublished observations). Monocytes (CD14⁺), which are the precursors of osteoclasts, comprised most of the remaining nonosteoclastic cells. Thus, although these contaminating populations may have the capacity to express RANKL and modulate the observed MIP-1γ response, we believe that this effect is minimal given the small numbers of such cells and the correspondence in findings between the two cell systems.

Osteoclasts rapidly undergo apoptosis unless stimulated by exogenous mediators, which may include M-CSF, RANKL, IL-1, fibroblast growth factor 2 (FGF2), or LPS (16, 23, 24, 25, 26). With the exception of M-CSF and FGF2, these survival-promoting stimuli act by inducing NF-κB (26, 27), which is well established as an antiapoptogen (28, 29). FGF2, which directly stimulates activation and survival of mature osteoclasts, mediates its effects through p42/p44 MAPK. We confirmed that MIP-1γ, like RANKL, stimulated NF-κB activity, and furthermore demonstrated that the prosurvival activity of RANKL was partially dependent on its ability to induce MIP-1γ. Of interest, MIP-1γ also stimulated the activation of mature osteoclasts, although it is difficult to separate this effect from its ability to promote cell viability. Thus, in contrast to most survival factors that act in a paracine manner, our results suggest that osteoclasts protect themselves from apoptosis through production of MIP-1γ as an autocrine survival factor. At present, the signal transduction pathways involved in MIP-1γ-induced osteoclast survival via CCR1 have not been characterized.

In conclusion, our data indicate that MIP-1γ represents a new and potentially important factor in the bone microenvironment that regulates osteoclastic bone resorption. The role that MIP-1γ may play in both normal bone turnover and osteolytic diseases remains to be established.