MIP-1δ Activates NFATc1 and Enhances Osteoclastogenesis: Involvement of Both PLCγ2 and NFκB Signaling

Pathological bone resorption is a source of significant morbidity in diseases affecting the skeleton such as rheumatoid arthritis, periodontitis, and cancer metastasis to bone. Evidence indicates that elevated levels of inflammatory mediators such as IL-1, IL-6, and TNF-α play a role in this process by promoting the formation of bone-resorbing osteoclasts. Additionally, current studies have identified inflammatory chemokines of the macrophage inflammatory protein (MIP) family as potential mediators of pathological bone resorption, where both MIP-1α and -3α have been shown to enhance osteoclast (OCL) development. In this study we provide evidence that MIP-1δ, whose expression is associated with renal cell carcinoma bone metastasis and rheumatoid arthritis, enhances OCL formation in vitro via a direct effect on OCL precursors. Consistent with this ability, exposure of OCL precursors to MIP-1δ resulted in the activation of PLCγ2 and NF-κB, two signaling pathways known to regulate OCL differentiation. Moreover, MIP-1δ induced expression and nuclear translocation of NFATc1, a master regulator of osteoclastogenesis, which was dependent on activation of both the PLCγ2 and NFκB signaling pathways. Lastly, consistent with in vitro studies, in vivo administration of MIP-1δ significantly increased OCL number and resorption area as determined using a murine calvarial bone resorption model. Taken together, these data highlight the potential of MIP-1δ as a mediator of pathological bone resorption and provide insight into the molecular mechanism through which MIP-1δ enhances osteoclastogenesis.


Introduction
Pathological bone resorption occurs in skeletal diseases such as rheumatoid arthritis, periodontitis, and cancer, resulting in substantial bone pain and loss of function. In the case of rheumatoid arthritis and periodontitis, bone loss occurs following chronic inflammation. Inflammatory mediators such as interleukin (IL)-1, IL-6, and tumor necrosis factor (TNF)-a have been shown to elevate levels of the osteoclastogenic cytokine, receptor activator of nuclear factor kappa-B ligand (RANKL), enhancing the development of bone resorbing osteoclasts (OCL) [1], thereby disrupting the delicate balance of bone resorption and formation. Evidence also supports a role for inflammatory mediators (eg. IL-3, IL-6, and IL-8) in the OCL-mediated bone resorption observed in metastatic breast cancer and multiple myeloma [2,3,4].
Recent studies indicate that inflammatory chemokines of the macrophage inflammatory protein (MIP) family may also play a role in mediating pathological bone resorption. Currently, the MIP family consists of six members: MIP-1a, MIP-1b, MIP-1d, MIP-1c, MIP-3a, and MIP-3b. MIP-3a, whose expression is increased in bone biopsies from rheumatoid arthritis patients, has been shown to enhance OCL development by stimulating OCL precursor proliferation [5]. Similarly, it has also been detected in periodontitis where elevated expression was positively correlated with disease status [6,7,8]. Elevated levels of another family member, MIP-1a, were reported in bone marrow of multiple myeloma patients as compared to healthy adults [9]. Further studies indicate that MIP-1a is able to stimulate OCL development [10], while inhibition of MIP-1a significantly reduces bone destruction in a mouse model of multiple myeloma [11].
In line with these findings, we recently found levels of another MIP family member, MIP-1d (CCL15), to be significantly elevated in human renal cell carcinoma bone metastasis (RBM) tissues relative to bone marrow from healthy adults [12]. Further, consistent with the osteolytic nature of RBM, we provided in vitro evidence that MIP-1d stimulates chemotaxis of OCL progenitors and enhances OCL differentiation in response to RANKL. Here, we demonstrate the ability of MIP-1d to directly enhance the differentiation of OCL precursors in vitro, elucidate its effect on the signaling pathways and transcription factors regulating osteoclastogenesis, and provide the first evidence that MIP-1d can stimulate osteoclastogenesis and bone resorption in vivo, highlighting its potential as a mediator of pathological bone loss.

MIP-1d Enhances Osteoclastogenesis in vitro
Previously, we reported the first evidence that MIP-1d enhances RANKL-mediated OCL differentiation in vitro using murine bone marrow mononuclear cells (BM-MNC) [12]. Since BM-MNC is a heterogeneous population containing a minor fraction of OCL progenitors, it was unclear whether MIP-1d affected OCL differentiation via a direct effect on OCL progenitors or through indirect effects on other cells within the population (eg. marrow stromal cells). Thus, we examined the ability of MIP-1d to promote OCL differentiation in vitro using macrophage colonystimulating factor (M-CSF)-dependent bone marrow macrophages (BMM), a refined population of committed OCL precursors. Consistent with our previous findings [12], while insufficient to stimulate OCL differentiation alone, MIP-1d significantly enhanced OCL differentiation in response to RANKL, evidenced by increased numbers of TRAP-positive multinucleated cells (Fig. 1A). In addition, OCL formed in the presence of MIP-1d exhibited a significantly higher fusion index (Fig. 1B) compared to those treated with RANKL alone, highlighted by a greater than 7-fold increase in OCL containing more than 11 nuclei per cell (Fig. 1C, D). Notably, similar results were observed using the macrophage cell line RAW 264.7 (Fig. 1A, B, D), indicating that our findings using BMM were not likely affected by any minor stromal elements remaining within the BMM population. Taken together, these data suggest that MIP-1d enhances OCL formation via a direct effect on OCL precursors.

MIP-1d Activates Key Pathways Regulating Osteoclastogenesis
Prior studies identified several signaling pathways through which OCL differentiation is promoted, including ERK [13], JNK [14], PLCc2 [15], and NF-kB [16]. To investigate the mechanism whereby MIP-1d enhances OCL differentiation, we first determined its ability to activate each of these signaling pathways in OCL precursors. Although exposure of BMM to MIP-1d did not induce phosphorylation of ERK or JNK, increased phosphorylation of PLCc2 was observed, indicating activation of PLCc2 signaling ( Fig. 2A). In addition, MIP-1d demonstrated an effect on both the canonical and non-canonical NF-kB signaling pathways. While activation of the canonical NF-kB pathway depends on phosphorylation and subsequent degradation of IkBs, largely IkBa, the non-canonical pathway depends on processing of p100 to p52 [17]. In addition, p50 traditionally partners with RelA or c-Rel in canonical signaling, while p52 partners with RelB in noncanonical signaling. Following exposure of BMM to MIP-1d, we observed phosphorylation and degradation of IkBa ( Fig. 2A). Further, analysis of nuclear extracts revealed the translocation of p50 and its binding partners RelA and c-rel to the nucleus, indicating activation of canonical NF-kB signaling (Fig. 2B). It is also interesting to note that the kinetics of p50 and c-Rel nuclear translocation following stimulation with MIP-1d appear shorter than that of RelA. While this apparent shift in kinetics may be due to a number of events (eg. differences in NF-kB subunit degradation kinetics, nuclear translocation of p50 homodimers, etc.), this data suggests that MIP-1d may have a unique effect on NF-kB signaling, which may be of interest in subsequent studies. Examination of non-canonical NF-kB signaling revealed processing of p100 to p52 following stimulation with MIP-1d, however, nuclear translocation of the p52 binding partner RelB was not detected (Fig. 2C), suggesting that MIP-1d may favor activation of the canonical pathway. Lastly, using an oligonucleotide probe containing an NF-kB binding site, we observed increased NF-kB activity in response to MIP-1d stimulation of BMM by EMSA (Fig. 2D). Collectively, these data demonstrate the ability of MIP-1d to activate both PLCc2 and NF-kB signaling in OCL precursors, each having the potential to contribute to the effects of MIP-1d on OCL formation.

MIP-1d Induces Critical Regulators of Osteoclastogenesis via activation of PLCc2 and NF-kB
Recent studies suggest that the importance of the PLCc2 and NF-kB signaling pathways to osteoclastogenesis lies in their ability to stimulate expression and nuclear translocation of the transcription factor NFATc1, considered the ''master regulator of OCL differentiation'' [18,19]. NFATc1 auto-regulates its promoter together with c-fos, a component of the dimeric transcription factor AP-1, and controls expression of genes critical to OCL differentiation and function [20]. Because MIP-1d activated both the PLCc2 and NF-kB signaling pathways, we next examined its effect on expression and activity of c-fos and NFATc1. Treatment of BMM with MIP-1d significantly increased expression of c-fos ( Fig. 3A) and, correspondingly, increased activation of AP-1 as determined by EMSA analysis (Fig. 3B). Moreover, similar to its effect on c-fos expression, MIP-1d significantly increased expression ( Fig. 3A) and translocation of NFATc1 to the nucleus (Fig. 3C). Correspondingly, increased mRNA expression of the NFATc1 downstream targets cathepsin K (CTSK), tartrateresistant acid phosphatase (TRAP), and dendritic cell-specific transmembrane protein (DC-STAMP) was also observed (Fig. 3D).
To determine the contribution of the NF-kB and PLCc2 signaling pathways to MIP-1d-induced expression of c-fos and NFATc1 we examined the effect of MIP-1d in the presence of chemical inhibitors of each pathway. Inhibition of both canonical and noncanonical NF-kB signaling using an IKK-2 inhibitor prevented cfos and NFATc1 induction by MIP-1d (Fig. 3E). Similarly, exposure of BMM to the PLC inhibitor U73122 also inhibited the ability of MIP-1d to induce c-fos and NFATc1 expression (Fig. 3F). Taken together, these data indicate that MIP-1d induces the expression and activation of c-fos and NFATc1, critical for OCL differentiation, involving stimulation of both the PLCc2 and NF-kB pathways.
It is interesting to note that although able to induce NFATc1 expression, MIP-1d was unable to induce OCL formation in the absence of RANKL ( [12], data not shown). While seemingly incongruous, NFATc1 is reported to be necessary for OCL differentiation, yet insufficient in the absence of RANKL [21], indicating an additional effect of RANKL beyond the induction of NFATc1. Intriguingly, while both MIP-1d and RANKL activate NF-kB, MIP-1d appears to incompletely stimulate the non-canonical pathway, evidenced by the lack of RelB nuclear translocation (Fig. 2C). Given the importance of the noncanonical pathway for OCL differentiation [22], it is tempting to speculate that the limited ability of MIP-1d to stimulate this pathway contributes to its failure to induce OCL differentiation independently of RANKL. Despite a limited capacity to stimulate non-canonical NF-kB signaling, MIP-1d robustly activated the canonical pathway as evidenced by the nuclear translocation of RelA and c-Rel (Fig. 2B). Notably, canonical NF-kB signaling through induction of RelA has been reported to promote survival of OCL precursors, translating to increased OCL formation [23]. Thus, although unable to induce OCL differentiation alone, MIP-1d may enhance OCL formation not only by promoting differentiation via NFATc1 induction, but also by promoting survival of OCL progenitors via activation of RelA.

MIP-1d Enhances Osteoclastogenesis and Bone Resorption in vivo
Although MIP-1d levels are elevated in several conditions causing pathological bone loss [12,24] and we have demonstrated its ability to enhance OCL differentiation in vitro, the capacity of MIP-1d to induce osteolysis in vivo has not been investigated. To determine whether MIP-1d affects osteoclastogenesis and bone resorption in vivo, we utilized a murine model wherein MIP-1d or PBS was injected subcutaneously over the calvariae. Consistent with the ability of MIP-1d to enhance OCL differentiation in vitro, calvariae from mice administered MIP-1d showed a significant increase in OCL number and resorption area relative to that observed in mice administered PBS (Fig. 4). These data indicate that MIP-1d is able to increase osteoclastogenesis and bone resorption, underscoring its potential to contribute to pathological bone loss.
While we have demonstrated the ability of MIP-1d to directly influence OCL precursors in vitro, it is possible that MIP-1d promotes OCL formation and bone resorption in vivo via additional mechanisms. Interestingly, expression of the receptors for MIP-1d, CCR1 and CCR3, is increased in mature OCL relative to OCL precursors [12]. Therefore it is conceivable that MIP-1d exerts effects on mature OCL in addition to OCL precursors, possibly influencing survival and activity. In addition to effects on bone-resorbing OCL, it is also possible that MIP-1d modulates the activity of bone-forming osteoblasts (OB). OB play a major role in OCL differentiation by producing RANKL, and its decoy receptor, osteoprotegerin (OPG). Since OB also express the receptors for MIP-d (S.L.K., unpublished observations/data), it may indirectly promote OCL formation and bone loss in vivo by influencing the production of RANKL and OPG by OB as well as by regulating OB survival and activity. Thus, in future studies it will be important to examine the effects of MIP-1d on mature OCL as well as pre and mature OB.
Collectively, these data highlight the potential of MIP-1d to contribute to pathologic bone resorption associated with skeletal diseases such as rheumatoid arthritis and cancer metastasis to bone, where MIP-1d levels are elevated [12,25]. This work is timely as small molecule antagonists against CCRs are being developed and evaluated in clinical trials [26]. While this approach is promising, caution must be taken as many CCRs, including CCR1 and CCR3, are widely expressed in the body and serve as receptors for multiple ligands raising the potential for toxicity. Further investigation into the molecular mechanisms through which MIP-1d influences OCL formation and bone resorption in vivo are warranted and may lead to the discovery of more specific methods of inhibiting its effects.

Ethics Statement
All experiments were carried out in accordance with the National Research Council's ''Guide to the Care and Use of Laboratory Animals''. Animal use was approved by the Johns Hopkins Animal Care and Use Committee, animal welfare assurance #A3272-01, protocol #MO07M149.

Generation of Bone Marrow Macrophages
Bone marrow was isolated from the tibiae, femur, and humerus of 4-8 week old NIH Swiss mice. The ends of each bone were removed and the marrow collected by flushing the bone with aMEM containing 15% FBS and 1% penicillin and streptomycin

Osteoclast Differentiation Assay
BMM were cultured in complete media containing M-CSF (20 ng/ml) and RANKL (5 ng/ml) (R&D Systems) in the presence or absence of full-length recombinant human MIP-1d (0.1 pg/ml) (R&D Systems). 50% of the culture media was replaced with fresh complete media containing cytokines as appropriate every 3 days. After 7 days, the development of mature osteoclasts (OCL) was determined by positive tartrate-resistant acid phosphatase (TRAP) staining (bright red color) and the presence of $3 nuclei under light microscopy. To determine the fusion index, the number of nuclei per OCL was counted in 15 to 20 random fields at 100X magnification.

In vivo Calvarial Bone Resorption
Following a previously established procedure [27], recombinant human MIP-1d (1 ng) or PBS was injected in a volume of 50 ml into the subcutaneous tissue overlying the right side of the parietal bones of calvariae in female 6-week-old CD-1 mice twice each day for 5 days. Four days following the last injection, calvariae were harvested, fixed in formalin for 24 hours, and decalcified in 10% EDTA, pH 7.4 for 48 hours. Paraffin-embedded sections (5 mm) were generated laterally from the sagittal suture at 100 mm intervals. A minimum of three sections were stained with orange G and TRAP, and light microscopic images were acquired. The number of OCL/mm of marrow cavity surface and total resorption area (the total area occupied by marrow cavities) between the coronal and lamboid sutures was measured using MetaMorph image analysis software (Universal Imaging Corp., Sunnyvale, CA).