Resorption Dendritic Cell-Mediated In Vivo Bone - The Journal of ...

The Journal of Immunology

Dendritic Cell-Mediated In Vivo Bone Resorption Radhashree Maitra,*,†,1 Antonia Follenzi,*,1 Arash Yaghoobian,† Cristina Montagna,*,‡ Simone Merlin,*,2 Elvira S. Cannizzo,* John A. Hardin,†,x Neil Cobelli,† E. Richard Stanley,{ and Laura Santambrogio*,‖ Osteoclasts are resident cells of the bone that are primarily involved in the physiological and pathological remodeling of this tissue. Mature osteoclasts are multinucleated giant cells that are generated from the fusion of circulating precursors originating from the monocyte/macrophage lineage. During inflammatory bone conditions in vivo, de novo osteoclastogenesis is observed but it is currently unknown whether, besides increased osteoclast differentiation from undifferentiated precursors, other cell types can generate a multinucleated giant cell phenotype with bone resorbing activity. In this study, an animal model of calvaria-induced aseptic osteolysis was used to analyze possible bone resorption capabilities of dendritic cells (DCs). We determined by FACS analysis and confocal microscopy that injected GFP-labeled immature DCs were readily recruited to the site of osteolysis. Upon recruitment, the cathepsin K-positive DCs were observed in bone-resorbing pits. Additionally, chromosomal painting identified nuclei from female DCs, previously injected into a male recipient, among the nuclei of giant cells at sites of osteolysis. Finally, osteolysis was also observed upon recruitment of CD11c-GFP conventional DCs in Csf1r2/2 mice, which exhibit a severe depletion of resident osteoclasts and tissue macrophages. Altogether, our analysis indicates that DCs may have an important role in bone resorption associated with various inflammatory diseases. The Journal of Immunology, 2010, 185: 1485–1491.

T

he osteoclast is a member of the monocyte/macrophage family for which the terminal differentiation is controlled by two critical cytokines, receptor activator of NF-kB ligand (RANK-L) and CSF-1 (1–4). Upon differentiation, precursor cells fuse into tartrate-resistant acid phosphatase (TRAP)-positive and cathepsin K-positive multinucleated osteoclasts (5). At sites of active bone resorption, the osteoclasts form a specialized cell membrane, the ruffled border, following organization of the actin cytoskeleton in conjunction with avb3 integrin-mediated matrix recognition. On attachment to bone, matrix-derived signals polarize intracellular secretory vesicles that deliver H+ATPase to the plasma membrane and release proteolytic enzymes into the resorptive microenvironment. Degradation of the bone Ca++/phosphate inorganic matrix is achieved through ATPase-mediated active secretion of protons, which create an acidic environment in the resorbing lacunae (6). In contrast, the organic matrix, mostly composed of collagen and elastin, is proteolized by secreted cystein and aspartic proteases, among which cathepsin K has a major role (7). *Department of Pathology, ‡Department of Molecular Genetics, xDivision of Rheumatology, Department of Medicine, {Department of Developmental Molecular Biology, and ‖Department of Microbiology and Immunology, Albert Einstein College of Medicine; and †Department of Orthopedic Surgery, Montefiore Medical Center, New York, NY 10461 1

R.M. and A.F. contributed equally to this work.

2

Current address: Department of Medical Sciences, University of Piemonte Orientale Medical School, Novara, Italy. Received for publication November 2, 2009. Accepted for publication May 20, 2010. This work was supported by the National Institutes of Health (to L.S. and E.R.S.) and Irene Diamond Professorships in Immunology (to L.S.). R.M. was supported by National Institutes of Health Training Grant NS07098. A.F. was partially supported by Regione Piemonte Ricerca Sanitaria Finalizzata 2008. Address correspondence and reprint requests to Dr. Laura Santambrogio, Department of Pathology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, New York, NY 10461. E-mail address: [email protected] Abbreviations used in this paper: CT, computerized tomography; DC, dendritic cell; EGFP, enhanced GFP; mgc, multinucleated giant cell; RANK-L, receptor activator of NF-kB ligand; Tg, transgenic; TRAP, tartrate-resistant acid phosphatase; UHMWPE, ultra-high m.w. polyethylene. Copyright Ó 2010 by The American Association of Immunologists, Inc. 0022-1767/10/$16.00 www.jimmunol.org/cgi/doi/10.4049/jimmunol.0903560

At steady state, osteoclasts are thought to arise from osteoclast progenitors present in the bone marrow as well as in the peripheral circulation (7). Two mouse mutants, CSF-1–deficient osteopetrotic (Csf1rop/op) and CSF-1R knockout (Csf1r2/2) mice, indicate that compromising myeloid differentiation into osteoclasts induces defective bone resorption with increased bone density and osteopetrosis (8, 9). In an osteopetrotic mouse (Csf1rop/op), a recessive mutation in the CSF1 gene compromises the development of monocyte-derived osteoclasts and tissue macrophages in several organs (8). Similarly, Csf1r2/2 mice have a profound reduction in the number of tissue macrophages and bone osteoclasts (9). However, in the Csf1rop/op mice, the development and total number of conventional dendritic cells (DCs) and Langerhans cells is not compromised (8), whereas in Csf1r2/2 mice, Langerhans cells are absent but the conventional DC population is not affected (9, 10). During conditions of acute and chronic inflammation, such as occur in osteoarthritis, rheumatoid , aseptic osteolysis, more inflammatory infiltrates are observed not only at the bone surface, but also in the soft tissue surrounding the bone. However, it is still uncertain whether the inflammatory infiltrate only participates in bone resorption by favoring osteoclastogenesis through RANK-L and MCSF production or whether the inflammatory infiltrates may play a more active primary role in bone resorption. In principle, because a CD11b+/GR1-low/c-fms2/+ common myeloid progenitor gives rise to macrophages, DCs, and osteoclasts, it is possible that DCs attracted to the site of inflammation could potentially be induced to express osteoclastic activity (11–13). The presence of the CSF-1R on conventional DC precursors (11), as well as expression of cathepsin K, TRAP, and RANK-L in immature DCs, supports this possibility (12). Indeed, several recent in vitro experiments indicate that both human (monocyte-derived) and mouse (bone marrowderived) DCs can function as osteoclasts in the presence of CSF1, RANK-L, and bone-like matrix (13–15). However, whether this occurs in vivo is not known. In the current study, a model of aseptic bone inflammation was used to determine the osteoclastic potential of conventional DCs. Among the possible experimental models that we could select to

1486 study the osteolytic potential of DCs, we chose ultra-high m.w. polyethylene (UHMWPE)-induced osteolysis for several reasons: 1) in this model, the bone inflammatory response is induced by injecting sterile alkane particles (UHMWPE) under the calvarial periostium to generate an aseptic inflammatory reaction that is easier to control compared with pathogen-induced osteolysis (16); 2) the resulting inflammatory reaction is localized to the site of injection and therefore easier to quantify than a more systemic reaction; and 3) the inflammatory infiltrates are mostly formed by local osteoclasts, recruited macrophages, and DCs, with very few T and B cells (16). Experiments performed in Csf1r2/2 mice, which are profoundly deficient in osteoclasts and tissue macrophages, conclusively demonstrated the in vivo osteolytic potential of CD11c+ conventional DCs recruited to the sites of bone resorption.

Materials and Methods UHMWPE-induced calvarial osteolysis UHMWPE-induced calvarial osteolysis was performed according to published protocols (16). Briefly, mice in each group (C57BL/6 female 8–12 wk old) were sedated with isoflurane gas and anesthetized with ketamine (20% ketamine HCl, 15% xylazine, and 65% saline at 0.1 ml per 20 g body weight). A 10-mm incision was made over the sagittal midline suture of the calvarium and a 1.0 3 1.0 cm area of the periosteum exposed. In sham control mice, the incision was closed without any further intervention, whereas experimental mice received 20 ml dried UHMWPE particles with an average size of 53–75 mm (Sigma-Aldrich, St. Louis, MO). The particles were distributed over the periosteum using a sterile sharp surgical spatula and the incision sutured. Twelve days postoperation, a microcomputed tomography (micro-CT) scan was performed to confirm and verify the development of particle-induced osteolysis. Micro-CT scans were performed on the Aloka LaTheta laboratory CT scan machine (Aloka, System Engineering Co., Tokyo, Japan). The x-ray voltage was set at high. For each skull, 162 slices were scanned with an average section-to-section distance of 0.03 nm at a slow acquisition speed using a facedown headfront orientation. Data were analyzed using LaTheta software (Aloka, System Engineering, Tokyo, Japan). All experiments were conducted in accordance with the Institute for Animal Care and Use Committee of Albert Einstein College of Medicine (New York, NY).

H&E staining of mouse calvaria Post-CT scan animals were euthanized utilizing CO2. Calvaria were excised and fixed in 4% paraformaldehyde in PBS (pH 7.4) and placed at 4˚C. After 24 h, samples were placed in 10% EDTA in PBS (pH 7.4) for 48– 72 h at 4˚C until decalcified. The calvaria were then processed for paraffin embedding, sectioned to 5 mm, and stained with H&E.

In vitro bone marrow cell differentiation Bone marrow cells were purified using the lineage-negative magnetic bead kit, which depletes cells expressing the following lineage Ags: CD5, B220, CD11b, Gr-1, CD-3, and Ter-119 (Miltenyi Biotec, Auburn, CA). Lineage negative (Lin2) precursors were then cultured in DMEM (Life Technologies, Rockville, MD), supplemented with 50 U/ml penicillin, 50 mg/ml streptomycin, 2 mM L-glutamine, 0.1 mM nonessential amino acids, 1 mM pyruvate, and 10 mM HEPES containing either GM-CSF (10 ng/ml) for 8 d to obtain immature conventional DCs or further differentiated with RANK-L (5 ng/ml) and M-CSF (10 ng/ml) for 8 d to obtain osteoclasts.

Intravenous injection of GFP+ DCs and FACS analysis of cells recovered at sites of osteolysis In some experiments, 2 wk postsurgery, mice (8–12-wk-old female C57BL/ 6) were injected i.v. with 107 bone marrow-derived GFP+-CD11c+ DCs prepared from C57BL/6-transgenic (Tg) (CAG-EGFP)10sb/J mice (The Jackson Laboratory, Bar Harbor, ME). The CD11c+ cells were purified from bone marrow cells that had been cultured in GM-CSF for 7 to 8 d, using CD11c+ Ab conjugated magnetic beads (Miltenyi Biotec). Cell purity following magnetic bead separation was assessed by FACS. Only preparations with a purity $99% were used. In some experiments, 96 h after DC injection, calvaria were isolated from sham and UHMWPEimplanted mice and directly observed as a whole mount using differential contrast imaging and fluorescence microscopy. In other experiments, 96 h following GFP+-DC injection, cells were retrieved from the calvaria by extensively washing the bone in DMEM and gently scraping the bone

DCs AND BONE RESORPTION surface. Retrieved cells were either analyzed directly by FACS (to quantify the percentage of GFP+ cells) or stained for cathepsin K. For staining, cells were fixed for 10 min in 2% paraformaldehyde, washed in 10 mM glycine, and permeabilized for 30 min in 0.05% saponin. Cells were incubated on ice for 30 min with 10 mg/ml Ab to mouse cathepsin K (Santa Cruz Biotechnology, Santa Cruz, CA) followed by an anti-mouse IgG conjugated to Alexa Fluor 568. Cells were analyzed on an FACScan flow cytometer (BD Biosciences, San Jose, CA).

Western blot analysis Cultured cells (DC or osteoclasts) were collected, washed in PBS, and lysed in 150 mM Tris-HCl (pH 8), 150 mM NaCl, 5 mM EDTA, and 1% Nonidet P-40 supplemented with a mixture of protease inhibitors (Roche, Basel, Switzerland). Eighty micrograms protein was run on a 12% SDSPAGE. After blotting, membranes were probed with the following primary mAbs: cathepsin F, K, L, S, and TRAP (Santa Cruz Biotechnology). Secondary Abs were HRP-labeled, and detection was performed using the Super Signal West Pico Chemiluminesent substrate (Pierce, Rockford, IL).

Confocal microscopy The calvaria from sham and UHMWPE-implanted mice were extracted, decalcified as described above, and embedded in either OCT-compound (Tissue-Tek, Sakura, Torrance, CA) for frozen sections or in paraffin according to standard procedures. Slides were stained for 30 min at room temperature with 10 mg/ml Ab to mouse cathepsin K (Santa Cruz Biotechnology) followed by Alexa Fluor 568-conjugated goat anti-mouse secondary Ab. Stained slides were examined using a scanning confocal microscope (Leica AOBS system, Leica Microsystems, Deerfield, IL). All images were collected using identical fluorescence settings.

Fluorescence in situ hybridization Fluorescence in situ hybridization was performed using a painting probe specific for the murine Y chromosome. DNA for the Y chromosome (M. A. Ferguson-Smith, University of Cambridge, Cambridge, U.K.) was labeled by degenerate oligonucleotide-primed PCR with Spectrum Orange-dUTP (Vysis, Abbott Molecular, Berkshire, U.K.). Five-micron formalin-fixed paraffin embedded tissue sections were baked overnight at 56˚C, dipped twice in xylene for 5 min at room temperature, and rehydrated in ethanol (100–90–70%). Slides were then washed in 43 SSC/0.5% Tween 20 at room temperature for 30 min in a rotating shaker. Pretreatment to digest cytoplasm was performed with pepsin (final concentration 0.4 mg/ml) at 37˚C for 30 min. Slides were denatured in formamide for 1 min and 45 s, and then a probe was applied and incubated overnight at 37˚C in a humidified chamber. After washing, nuclei were counterstained with DAPI for 10 min. Interphase cells were imaged with an Olympus BX61 microscope equipped with a Cooke SensicamQE camera with IPLab for image acquisition (Olympus, Melville, NY). An IPLab script was generated to acquire images of interphase cells for the Spectrum Orange dye and for a differential contrast image. Multiple focal planes (13) were acquired to ensure that signals on different focal planes were included. Images were analyzed with tools available through ImageJ (National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov/ij/).

Generation and subcloning of the CD11c promoter Two hundred nanograms genomic DNA from a 129 mouse was amplified using the following primers: forward primer 59-CATGCTCGAGGATCCCAGAGCTGGCCTGCTG-39 (designed with XhoI and BamHI restriction sites at the 59); reverse primer 59-CATGACCGGTCAGAAGCAGGCTCTGAGCAA-39 (designed with the AgeI restriction site at the 39). The amplified 2013-bp CD11c promoter corresponds to bases 3790–5802 of the published CD11c promoter sequence (Gene Bank DQ658851.1). Following PCR amplification, the promoter was digested with XhoI and AgeI restriction enzymes and cloned into the pCCL.PPT.PGK.GFP.PRE lentiviral vector upon removal of the PGK promoter (17).

Lentiviral vector production and concentration Lentiviral particles expressing either the non–tissue-specific PGK promoterGFP construct or the tissue-specific mCD11c promoter-GFP construct were generated by calcium phosphate-mediated cotransfection of 293T cells with four plasmids: 1) a CMV promoter-driven packaging construct expressing the gag and pol genes; 2) a Rous sarcoma virus promoter-driven construct expressing rev; 3) a CMV promoter-driven construct expressing the VSV-g envelope; and 4) the previously generated 2013-bp CD11c promoter controlling GFP expression. Thirty hours following transfection, the culture supernatant containing the packaged viral particles was collected and concentrated by ultracentrifugation. Collected viral particles were titrated on


1487 1:100, 1:1000, and 1:10,000). For each dilution, positively transduced DCs were quantified by FACS as a percentage of GFP+ cells. Calculation from the titration analysis indicated 1 3 109 transducing viral particles per milliliter.

Bone marrow transplant experiments Six to 8-wk-old male mice (Csf1r2/2 on an outbred C57BL/6J, C3Heb/FeJa/a, 129SvJ background) were killed with CO2 and the bone marrow collected by flushing femurs and tibia. Lineage negative (Lin2) cells were purified as described above (Miltenyi Biotec MACS Lin2 cells). Cells were transduced with either the PGK prom-GFP lentiviral vector or the mCD11c prom-GFP lentiviral vector at a concentration of 108 TU/ml for 106 cells. Transduced cells were cultured for 24 h in StemSpan media (StemCell Technologies, Vancouver, British Columbia, Canada) supplemented with IL-3, IL-6, stem cell factor, and Flt-3 ligand cytokines. For bone-marrow transfer experiments, transduced cells were injected into the tail vein (106 cells/mouse) of 8-wk-old mice (C57BL/6 or Csf1r+/2) that had been previously lethally irradiated (2 3 6 Gy). Full engraftment of the transplanted hematopoietic stem cells was observed 2 mo following bone marrow transplantion.

Characterization of GFP+ DCs in blood, bone marrow, and spleen of the transplanted mice Spleen, blood, and bone marrow samples were harvested from each mouse 2 mo following bone marrow transplantion. Single-cell suspensions were prepared from each organ and RBCs removed by lysis. For immunostaining, 105 cells were incubated in staining buffer (PBS, 2% FBS) for 30 min at 4˚C with 1 to 2 mg/ml of the following PE-conjugated primary Abs: CD19, CD11b, CD11c, Gr-1, CD3, and NK1.1 (BD Pharmingen, San Diego, CA). Following incubation, cells were washed three times in staining buffer and analyzed on an FACScan flow cytometer (BD Biosciences). FIGURE 1. Implant of UHMWPE particles in mouse calvaria induces osteolysis. a, H&E staining of mouse calvaria in sham or UHMWPEimplanted mice (upper panels 320 magnification; lower panels 340 magnification). Arrows point to areas of osteolysis observed in the UHMWPEimplanted mice but not in the sham controls. b, H&E staining of pericalvarial tissue implanted with UHMWPE. mgcs surround the UHMWPE particles (asterisks) (original magnification 3100). c, Micro-CT scan of mouse calvaria (a representative frontal section is shown out of a total of 162 sections). CT scan was performed in both sham or UHMWPE-implanted mice pre- and postsurgery. Arrow indicates an area of osteolysis in UHMWPE-implanted mice. d, Analysis of pre- and postsurgery bone density parameters in sham and UHMWPE-implanted mice. One of three experiments is shown. In each experiment, three to four mice for each experimental condition (sham or UHMWPE) were used. bone marrow DCs using limiting dilution analysis. Briefly, 100,000 DCs were cultured in DMEM with progressively lower dilutions of either the PGK promoter-GFP or the murine CD11c prom-GFP lentivirus (1:10, FIGURE 2. DCs are recruited to sites of osteolysis. a, Differential interference contrast image of the calvarium from a sham (left panel) and a UHMWPE-implanted (right panel) mouse (original magnification 35). b, Fluorescence microscopic images of the same calvaria shown in a demonstrating the recruitment of GFP+ DCs in a mouse implanted with UHMPE (right panel) but not in the sham-treated mouse (left panel) (original magnification 35). c, FACS quantification of GFP+-DCs recruited to the site of osteolysis in sham (left panel) or UHMWPE (right panel) implanted mice. d, Differential interference contrast image of frozen sections of the calvaria from sham (left panel) and UHMWPE-implanted (right panel) mice (original magnification 363). e, Fluorescence microscopy of the same frozen sections from sham and UHMWPE-implanted mice shown in d to determine the presence of recruited GFP-labeled DCs (original magnification 363). f, Overlay of images shown in d and e. One of two experiments is shown. In each experiment, three mice for each experimental condition (sham or UHMWPE) were used.

Results UHMWPE calvarial model The mouse model of osteolysis was established by implanting microgram amounts of UHMWPE particles adjacent to bone in mouse calvaria as described previously (16). In sham-operated mice, the calvaria were surgically uncovered and the periosteum incised with no further intervention, whereas in the UHMWPE group, particles were deposited directly between the bone and the periosteum. Two weeks postsurgery, a strong cellular infiltrate was observed in the UHMWPE-implanted mice, and small pit-like erosions were present on the calvarial surfaces (Fig. 1a). Cellular fusion around the particles was also evident from the presence of multinucleated giant cells (mgcs) exhibiting 6–12 nuclei (Fig. 1b). To quantify the extent of calvarial osteolysis, mice were analyzed with micro-CT (Fig. 1c). In mice receiving UHMWPE particle implants, a statistically

1488 significant decrease in cortical and total bone mass density was observed in the postsurgery specimens. In contrast, no significant differences were observed in the sham controls pre- and postsurgery (Fig. 1d). We therefore sought to adapt this model to gain further insight into how DCs might be involved in the bone-resorptive process. DCs are recruited to the site of osteolysis In the next set of experiments, we addressed the question of whether peripheral myeloid DCs were recruited to the tissue surrounding the area of osteolysis. Ten million CD11c+ immature bone marrow- derived DCs were prepared from C57BL/6-Tg (CAG-EGFP) mice, which express the enhanced GFP (EGFP) under the control of the b-actin promoter. Cells were injected into the tail veins of UHMPE-implanted and sham-operated mice. Ninety-six hours later, the whole calvarium was observed using differential contrast imaging and fluorescent microscopy (Fig. 2a, 2b). Green fluorescent areas were observed in the frontal area of the calvaria in UHMWPE-implanted mice but not in the sham control (Fig. 2b). The calvarium area adjoining the site of surgery was then excised, decalcified, and 5-mm frozen sections generated for further fluorescent microscopic analysis. Extensive cellular infiltrates were observed under light microscopy in UHMWPEimplanted mice but not in the sham controls (Fig. 2d). Among the infiltrates, many fluorescent cells could be observed in the UHMWPE-implanted mouse (Fig. 2e, right panel, merged image of 2d and 2e shown in 2f). FACS quantification of the GFPinjected DCs recruited to the calvaria indicated that in the mice implanted with UHMWPE, 13% of the cells retrieved at the site of osteolysis were GFP+ (Fig. 2c). These results demonstrate an active recruitment of DCs to the site of osteolysis.

DCs AND BONE RESORPTION Taken together, these data indicate that cathepsin K+ GFP+DCs are readily recruited to the site of osteolysis. DCs recruited to the site of bone inflammatory osteolysis fuse into mgcs Active osteoclasts, the primary cells involved in bone erosion, are organized as mgcs that are closely connected to the erosion pits observed during inflammatory bone resorption. To determine whether DCs are recruited into the mgcs observed around the UHMWPE particles (Fig. 1b), bone marrow DCs were prepared from a C57BL/6-Tg (CAG-EGFP) female donor. These cells were then injected into male recipients that were previously surgically implanted with UHMWPE or had undergone a sham procedure. DC recruitment and their osteolytic potential were monitored by tracking GFP+ cathepsin K+ cells as described above (Fig. 4). In Fig. 4b, a GFP+ cell can be observed in the inflammatory infiltrates, which, in a serial section, did not stain for the Y chromosome (Fig. 4c, white arrow). Thus, this cell was clearly derived from the transplanted bone marrow DCs. A similar group of inflammatory cells adjacent to the bone are shown in Fig. 4d–f. Transplanted bone-marrow-derived recruitment into mgcs was demonstrated by detection of nuclei missing the Y chromosome in mgcs around the UHMWPE particles (Fig. 4g). These cells contained multiple nuclei, the majority of which were positive for the Y chromosome (recipient cells), but a small number were missing the Y chromosome, indicating that they were derived from the transplanted female donor DCs (Fig. 4g, white arrows, second and fourth panels). These data support the proposition that

DCs recruited to sites of osteolysis express cathepsin K In the next series of experiments, we sought to determine whether conventional DCs that migrate to the site of osteolysis express cathepsin K. Among the possible markers associated with osteolytic activity, we chose cathepsin K because it is a well-characterized osteoclast marker, and its activity in bone resorption is indicated by the bone macromegalia observed in cathepsin K knockout mice (18). More importantly, CD11c+ conventional DCs express low levels of cathepsin K, which can be readily upregulated in vitro following culture of DCs with M-CSF and RANK-L, indicating, at least in vitro, the bone resorption potential of DCs (Fig. 3a). To this goal, CD11c+ DCs were purified from the bone marrow of C57BL/6-Tg (CAG-EGFP) mice (Fig. 3b) and injected into syngenic non–GFP-recipient mice, which were either sham operated or implanted with UHMWPE. Confocal analysis determined that in the calvaria of sham-operated mice, essentially no recruited CD11c+ GFP+ cells could be observed (Fig. 3c), and the only cathepsin K+ cells (red) were the resident osteoclasts (Fig. 3c). In contrast, the calvaria of mice that had been implanted with UHMWPE particles exhibited a rich infiltrate of recruited GFP+ DCs (green), some of which stained positive for cathepsin K (Fig. 3d, 3e, red). In addition, some cells that were observed in contact with the bone (Fig. 3f, left panel) were positive for both GFP and cathepsin K (Fig. 3f, right panel, arrow), indicating the possibility that DCs may form boneresorbing pits in the UHMWPE-implanted mice (Fig. 3f). To better quantify the number of CD11c+GFP+ cathepsin K+ cells recruited to the site of osteolysis, the calvaria of sham- or UHMWPE-implanted mice was removed. Cells retrieved from the site of osteolysis were stained for cathepsin K and analyzed by flow cytometry. UHMWPEimplanted mice showed an increased calvarial recruitment of cathepsin K+ GFP+ DCs as compared with sham controls (Fig. 3g).

FIGURE 3. Cathepsin K-positive DCs are recruited to the site of osteolysis. a, Western blot analysis for cathepsin S, L, F, and K and TRAP of bone marrow-derived immature DCs differentiated for 8 d in GM-CSF (DC) or further differentiated with RANK-L and M-CSF for an additional week (osteoclasts). b, FACS analysis of CD11c expression of purified GFP+ DCs prior to injection into GFP2 recipient mice (solid line) or isotype control (dotted line). c–f, Confocal light (left panel) and fluorescent (right panel) images of calvarial sections analyzed for expression of cathepsin K+ (red) and GFP+ (green) DCs recruited to the site of surgery in sham (c) or UHMWPE-implanted mice (d–f). c and d, Original magnification 363; e and f, original magnification 3100. Arrow indicates cathepsin K+ GFP+ DCs recruited to areas of osteolysis. g, FACS profile of GFP+-cathepsin K+ DCs retrieved from the calvaria of sham or UHMWPE-implanted mice. One of three experiments is shown. In each experiment, three mice for each experimental condition (sham or UHMWPE) were used.


1489

FIGURE 4. Female donor DCs are recruited into male recipient mgcs at sites of osteolysis. Confocal light (a, d) and fluorescent (b, c, e, f) images of sections from the calvarium of an UHMWPE-injected mouse. b and e show sections stained for cathepsin K (red) and GFP (green). c and f represent serial sections of b and e, respectively, subjected to chromosomal painting for the Y chromosome (pink). In c, note the absence of the Y chromosome (white arrow) in a GFP+ cell shown in b. d–f show an inflammatory infiltrate adjacent to the bone surface that contains many GFP+ cells (e), some of which lack the Y chromosome (f, white arrows). g, H&E staining of mgcs around UHMWPE particles (asterisks) containing a mixture of Y chromosome-positive and -negative nuclei (second and fourth panels, white arrows) within the same cell. One of two experiments is shown. a–g, Original magnification 3100.

recruited DCs can participate in the formation of giant cells found at inflammatory sites adjacent to the bone. DCs recruited to the inflammatory site possess bone resorption activity To determine in a more definitive manner whether DCs recruited to sites of bone inflammation possess osteolytic potential, we performed the UHMWPE calvarial model experiments in mice devoid of osteoclasts. To do this, we used Csf1r2/2 mice, which are severely depleted of macrophages and osteoclasts and which phenotypically present with extensive osteopetrosis (9). Also, because these mice lack the CSF-1R, monocytes that might have been recruited to the inflammatory site cannot differentiate into osteoclasts. The Csf1r2/2 mice are difficult to breed because the majority of newborn mice die within 1 mo. The few mice that survive are very frail, and several attempts to perform surgical experiments on these mice failed, and so chimeras were created. To create an endogenous population of GFP+ DCs, Csf1r2/+ mice were lethally irradiated and transplanted with Csf1r2/2 bone marrow cells that had been previously transduced with a lentiviral vector expressing GFP under the control of the CD11c promoter. The tissue specificity of the CD11c promoter was compared with the non– tissue-specific PGK promoter. As expected, transplantation of bone marrow cells with the non–tissue-specific PGK promoter controlling GFP expression gave rise to green fluorescent cells in both the lymphoid (T and B cells) and myeloid populations (macrophages, DCs, and granulocytes) (data for spleen shown in Fig. 5a). In contrast, in mice transplanted with total bone marrow cells transduced with the tissue-specific CD11c promoter, GFP expression was restricted to the myelomonocytic population in the spleen (Fig. 5a). Two months after irradiation, chimeric (Csf1r+/2 recipient and Csf1r2/2 bone marrow donor) mice were either surgically implanted with UHMWPE on their calvaria or were sham operated. Two weeks following surgery, the calvaria were removed and analyzed

by CT scan for bone erosion and resorption. One hundred sixty-two scan images were overlapped as a three-dimensional image (Fig. 5b). Sham-operated chimeras or control Csf1r2/+ did not display any sign of osteolysis (Fig. 5b, first and third panels). In contrast, UHMWPE-injected mice exhibited extensive osteolysis (Fig. 5b, second and fourth panels). Mice competent to produce osteoclasts, macrophages, and DCs (Fig. 5b, fourth panel), as well as mice repopulated with DCs but lacking osteoclasts and macrophages (Fig. 5b, second panel), displayed bone surface erosion (Fig. 5b), indicating the in vivo bone resorption activity of recruited DCs.

Discussion Under physiological conditions, maintenance of bone mass is achieved through a delicate balance between the activity of boneforming osteoblasts and bone-resorbing osteoclasts (19). This constant bone remodeling is necessary to support calcium homeostasis and to meet the structural needs of the musculoskeletal system. However, in certain inflammatory diseases, such as rheumatoid arthritis, this balance is lost, and osteolytic activity predominates, resulting in diffuse periarticular osteopenia and localized bone erosion (20–22). Understanding the mechanisms that determine the relative activity of osteoblasts and osteoclasts is important for the development of strategies for minimizing skeletal injury in these inflammatory diseases. Osteoclastogenesis depends upon progenitor cells of the monocytic hematopoietic lineage. These cells differentiate into osteoclasts under the influence of M-CSF and RANK-L, derived at least in part from bone stromal cells (23, 24). DC-specific transmembrane protein and avb3 integrin are also part of the osteoclast differentiation pathway. DC-specific transmembrane protein controls cell fusion, which leads to mgcs, whereas avb3 integrin is important for providing giant cells with the ability to adhere to the bone surface and to form an acidified extracellular microenvironment. Within this latter space, cathepsin K and TRAP are released to promote

1490

DCs AND BONE RESORPTION

FIGURE 5. DCs induce inflammatory osteolysis in absence of endogenous macrophages and osteoclasts. a, FACS profile of T cells (CD3), B cells (CD19), granulocytes (Gr-1), macrophages (CD11b), DCs (CD11c), and NKT cells (NK1.1) from the spleen of mice previously transduced with lentiviral vector expressing GFP under the control of a general promoter (PGK) or a tissue-specific promoter (CD11c). Mice were analyzed 2 mo after receiving the bone marrow transduced with the lentiviral vectors. Ten days prior to the FACS analysis, mice were injected with the B16–GM-CSF cells that produce endogenous GM-CSF, thus increasing the number of splenic DCs and facilitating the analysis of the CD11c promoter tissue specificity. b, Csf1r+/2 recipient mice were lethally irradiated and transplanted with the bone marrow of either Csf1r2/2 or Csf1r+/2 mice. The procedure generates chimeras that only express CD11c+ DCs (first and second panels) or DCs, macrophages, and osteoclasts (third and fourth panels), respectively. CT scan of the calvaria of chimeras mice implanted with UHMWPE or sham controls. One of two experiments is shown (in each experiment, three mice for each experimental condition [sham or UHMWPE] were used).

solubilization of calcium and phosphate and to digest matrix collagen and elastin within the resorbing lacunae (25). A still-unresolved question is the source of osteoclast precursors. At the present time, we do not know how many cells are derived from the bone pool of physiological preosteoclasts versus the recruitment of peripheral circulating precursors. This question is particularly important for understanding bone resorption in inflammatory conditions. Monocytes and circulating DCs are normally recruited to inflamed tissue (24, 26, 27), where monocytes may directly differentiate into osteoclasts (26). However, DCs could play either a direct or indirect role in osteoclastogenesis (28, 29). An indirect role would involve the ability of these cells to produce cytokines, such as IL-1, IL-6, and TNF-a, that augment expression and release of TRAP and cathepsin K by resident osteoclasts (30). DCs could also promote osteoclastogenesis indirectly by stimulating T cells to express RANK-L, a major differentiation factor for osteoclast precursors (31). In contrast, DCs could potentially directly participate in bone resorption following cathepsin K release. Indeed, several in vitro studies support the notion that immature DCs can differentiate into osteoclast-like cells (14, 32, 33). Culture of DCs with naive T cells and human periodontal pathogens indicated the in vitro ability of DCs to fuse into multinucleated cells with bone resorptive activity (14). In the current study, we addressed the osteolytic potential of DCs using an in vivo model that permits us to track recruitment of DCs to sites of bone inflammation, as well as testing knockout mice that lack macrophages and osteoclasts (9). The re-

sults demonstrated that exogenous myeloid DCs migrate to an inflammatory site of osteolysis, where they are active participants in bone resorption. A still-open question is whether the same growth factors that control physiological osteoclastogenesis are also involved in de novo inflammatory-induced osteoclastogenesis. In vitro analysis indicates that, similarly to bona fide osteoclast precursors, immature DCs are also dependent on RANK-L and M-CSF for differentiation into mature osteoclasts (14). These data support the notion that, at least in vitro, a similar growth factor-mediated differentiation pathway is used by both precursors. However, it is also possible that additional cytokines or growth factors play a role during the in vivo process of DCs’ differentiation into osteoclasts. In fact, Langerhans cells have been shown to fuse into mgcs in the rare disease of histiocytosis, which is associated with extensive focal bone erosions (34). Fusion of this subtype of DCs does not require RANK-L or M-CSF; instead it is dependent on IL-17 (34). Similarly, fusion of macrophages during inflammatory diseases does not rely on RANK-L but mostly on high levels of IL-4 (35). Thus, it is possible that conventional DC fusion into multinucleated cells with bone-resorption properties is also dependent on other cytokines and growth factors. It would be of great interest to determine whether the biological function of mature osteoclasts derived from bona fide osteoclasts precursors or derived from immature DCs differs. This point is of extreme interest because the fusion of DCs and their role in bone resorption is primarily observed during inflammatory

The Journal of Immunology conditions. Intervening in the recruitment of DCs to the bone matrix or their fusion into giant cells could be part of a therapeutic approach for limiting joint injury in diseases, such as osteoarthritis and rheumatoid arthritis (36). This model could be a useful tool for the development and testing of such agents.

Disclosures The authors have no financial conflicts of interest.

References 1. Wu, Y., M. B. Humphrey, and M. C. Nakamura. 2008. Osteoclasts - the innate immune cells of the bone. Autoimmunity 41: 183–194. 2. Abu-Amer, Y., I. Darwech, and J. Otero. 2008. Role of the NF-kappaB axis in immune modulation of osteoclasts and bone loss. Autoimmunity 41: 204–211. 3. Horowitz, M. C., and J. A. Lorenzo. 2007. Immunologic regulation of bone development. Adv. Exp. Med. Biol. 602: 47–56. 4. Faccio, R., S. Takeshita, G. Colaianni, J. Chappel, A. Zallone, S. L. Teitelbaum, and F. P. Ross. 2007. M-CSF regulates the cytoskeleton via recruitment of a multimeric signaling complex to c-Fms Tyr-559/697/721. J. Biol. Chem. 282: 18991–18999. 5. Hayman, A. R. 2008. Tartrate-resistant acid phosphatase (TRAP) and the osteoclast/immune cell dichotomy. Autoimmunity 41: 218–223. 6. Blair, H. C., S. L. Teitelbaum, R. Ghiselli, and S. Gluck. 1989. Osteoclastic bone resorption by a polarized vacuolar proton pump. Science 245: 855–857. 7. Teitelbaum, S. L. 2007. Osteoclasts: what do they do and how do they do it? Am. J. Pathol. 170: 427–435. 8. Ryan, G. R., X. M. Dai, M. G. Dominguez, W. Tong, F. Chuan, O. Chisholm, R. G. Russell, J. W. Pollard, and E. R. Stanley. 2001. Rescue of the colonystimulating factor 1 (CSF-1)-nullizygous mouse (Csf1(op)/Csf1(op)) phenotype with a CSF-1 transgene and identification of sites of local CSF-1 synthesis. Blood 98: 74–84. 9. Dai, X. M., X. H. Zong, V. Sylvestre, and E. R. Stanley. 2004. Incomplete restoration of colony-stimulating factor 1 (CSF-1) function in CSF-1-deficient Csf1op/Csf1op mice by transgenic expression of cell surface CSF-1. Blood 103: 1114–1123. 10. Ginhoux, F., F. Tacke, V. Angeli, M. Bogunovic, M. Loubeau, X. M. Dai, E. R. Stanley, G. J. Randolph, and M. Merad. 2006. Langerhans cells arise from monocytes in vivo. Nat. Immunol. 7: 265–273. 11. Auffray, C., Y. Emre, and F. Geissmann. 2008. Homeostasis of dendritic cell pool in lymphoid organs. Nat. Immunol. 9: 584–586. 12. Boyce, B. F., E. M. Schwarz, and L. Xing. 2006. Osteoclast precursors: cytokinestimulated immunomodulators of inflammatory bone disease. Curr. Opin. Rheumatol. 18: 427–432. 13. Sørensen, M. G., K. Henriksen, S. Schaller, D. B. Henriksen, F. C. Nielsen, M. H. Dziegiel, and M. A. Karsdal. 2007. Characterization of osteoclasts derived from CD14+ monocytes isolated from peripheral blood. J. Bone Miner. Metab. 25: 36–45. 14. Alnaeeli, M., J. M. Penninger, and Y. T. Teng. 2006. Immune interactions with CD4+ T cells promote the development of functional osteoclasts from murine CD11c+ dendritic cells. J. Immunol. 177: 3314–3326. 15. Rivollier, A., M. Mazzorana, J. Tebib, M. Piperno, T. Aitsiselmi, C. RabourdinCombe, P. Jurdic, and C. Servet-Delprat. 2004. Immature dendritic cells transdifferentiation into osteoclasts: a novel pathway sustained by the rheumatoid arthritis microenvironment. Blood 104: 4029–4037.

1491 16. Maitra, R., C. C. Clement, G. M. Crisi, N. Cobelli, and L. Santambrogio. 2008. Immunogenecity of modified alkane polymers is mediated through TLR1/2 activation. PLoS ONE 3: e2438. 17. Follenzi, A., L. E. Ailles, S. Bakovic, M. Geuna, and L. Naldini. 2000. Gene transfer by lentiviral vectors is limited by nuclear translocation and rescued by HIV-1 pol sequences. Nat. Genet. 25: 217–222. 18. Asagiri, M., T. Hirai, T. Kunigami, S. Kamano, H.-J. Gober, K. Okamoto, K. Nishikawa, E. Latz, D. T. Golenbock, K. Aoki, et al. 2008. Cathepsin K-dependent toll-like receptor 9 signaling revealed in experimental arthritis. Science 319: 624–627. 19. Cui, W., E. Cuartas, J. Ke, Q. Zhang, H. B. Einarsson, J. D. Sedgwick, J. Li, and A. Vignery. 2007. CD200 and its receptor, CD200R, modulate bone mass via the differentiation of osteoclasts. Proc. Natl. Acad. Sci. USA 104: 14436–14441. 20. Bar-Shavit, Z. 2007. The osteoclast: a multinucleated, hematopoietic-origin, bone-resorbing osteoimmune cell. J. Cell. Biochem. 102: 1130–1139. 21. David, J. P. 2007. Osteoimmunology: a view from the bone. Adv. Immunol. 95: 149–165. 22. Teitelbaum, S. L. 2000. Osteoclasts, integrins, and osteoporosis. J. Bone Miner. Metab. 18: 344–349. 23. Koulouvaris, P., K. Ly, L. B. Ivashkiv, M. P. Bostrom, B. J. Nestor, T. P. Sculco, and P. E. Purdue. 2008. Expression profiling reveals alternative macrophage activation and impaired osteogenesis in periprosthetic osteolysis. J. Orthop. Res. 26: 106–116. 24. Lorenzo, J. 2007. Osteoclast precursor cells. Adv. Exp. Med. Biol. 602: 77–82. 25. Faccio, R., S. L. Teitelbaum, K. Fujikawa, J. Chappel, A. Zallone, V. L. Tybulewicz, F. P. Ross, and W. Swat. 2005. Vav3 regulates osteoclast function and bone mass. Nat. Med. 11: 284–290. 26. Kahn, A. J., S. L. Teitelbaum, J. D. Malone, and M. Krukowski. 1982. The relationship of monocytic cells to the differentiation and resorption of bone. Prog. Clin. Biol. Res. 110(Pt B): 239–248. 27. Schett, G. 2008. Review: Immune cells and mediators of inflammatory arthritis. Autoimmunity 41: 224–229. 28. Santiago-Schwarz, F., P. Anand, S. Liu, and S. E. Carsons. 2001. Dendritic cells (DCs) in rheumatoid arthritis (RA): progenitor cells and soluble factors contained in RA synovial fluid yield a subset of myeloid DCs that preferentially activate Th1 inflammatory-type responses. J. Immunol. 167: 1758–1768. 29. Alnaeeli, M., J. Park, D. Mahamed, J. M. Penninger, and Y. T. Teng. 2007. Dendritic cells at the osteo-immune interface: implications for inflammationinduced bone loss. J. Bone Miner. Res. 22: 775–780. 30. Horwood, N. 2008. Lymphocyte-derived cytokines in inflammatory arthritis. Autoimmunity 41: 230–238. 31. Gillespie, M. T. 2007. Impact of cytokines and T lymphocytes upon osteoclast differentiation and function. Arthritis Res. Ther. 9: 103–107. 32. Gallois, A., J. Lachuer, G. Yvert, A. Wierinckx, F. Brunet, C. Rabourdin-Combe, C. Delprat, P. Jurdic, and M. Mazzorana. 2010. Genome-wide expression analyses establish dendritic cells as a new osteoclast precursor able to generate bone-resorbing cells more efficiently than monocytes. J. Bone Miner. Res. 25: 661–672. 33. Alnaeeli, M., and Y. T. Teng. 2009. Dendritic cells: a new player in osteoimmunology. Curr. Mol. Med. 9: 893–910. 34. Coury, F., N. Annels, A. Rivollier, S. Olsson, A. Santoro, C. Speziani, O. Azocar, M. Flacher, S. Djebali, J. Tebib, et al. 2008. Langerhans cell histiocytosis reveals a new IL-17A-dependent pathway of dendritic cell fusion. Nat. Med. 14: 81–87. 35. Helming, L., and S. Gordon. 2007. The molecular basis of macrophage fusion. Immunobiology 212: 785–793. 36. Stoch, S. A., and J. A. Wagner. 2008. Cathepsin K inhibitors: a novel target for osteoporosis therapy. Clin. Pharmacol. Ther. 83: 172–176.