The Signaling Adapter Protein DAP12 Regulates Multinucleation During Osteoclast Development


  • The authors have no conflict of interest.


Deficiency of the signaling adapter protein DAP12 is associated with bony abnormalities in both mice and humans. We identify specific DAP12-associated receptors expressed by osteoclasts and examine function of DAP12 in murine osteoclasts in vivo and in vitro. These data show a new role for DAP12 signaling in regulating formation of multinucleated osteoclasts.

Introduction: Osteoclasts are bone-resorbing cells derived from hematopoietic precursors in the myeloid lineage. In other myeloid cell types, the signaling adapter protein DAP12 transmits activating signals on ligation of a DAP12-associated receptor (DAR). The aim of this study was to clarify the role of DAP12 signaling during osteoclast development.

Materials and Methods: Osteoclasts from DAP12−/− or control mice were analyzed in vitro for morphology, function, and for osteoclast markers. DARs were identified in osteoclast cultures through reverse transcriptase-polymerase chain reaction (RT-PCR). Bone density of DAP12−/− and control mice were analyzed by microcomputed tomography. DAP12−/− osteoclasts were retrovirally reconstituted with DAP12. RAW264.7 cells were transfected with FLAG-tagged DAP12 or TREM2 and stimulated by anti-FLAG antibody during in vitro osteoclastogenesis.

Results: C57BL/6 DAP12-deficient mice have higher bone mass than C57BL/6 wildtype controls. We verified the presence of DAP12 in pre-osteoclasts and osteoclasts derived from C57BL/6 or the pre-osteoclast line RAW 264.7 and identified the DARs expressed. DAP12−/− osteoclasts developed in vitro with macrophage colony-stimulating factor (M-CSF) and RANKL formed only intensely TRACP+ mononuclear cells and failed to generate multinuclear osteoclasts. These mononuclear cells are functional osteoclast-like cells because, by RT-PCR, they express other osteoclast markers and generate resorption pits on dentine slices, although quantitative assessment of bone resorption shows decreased resorption by DAP12−/− osteoclasts compared with C57BL/6 osteoclasts. Restoration of DAP12 expression by retroviral transduction of DAP12−/− osteoclast precursors rescued in vitro osteoclast multinucleation. Direct stimulation of DAP12 expressed in RAW264.7 during in vitro osteoclastogenesis led to a marked increase in the number of TRACP+ multinucleated osteoclast-like cells formed.

Conclusion: Our studies indicate that stimulation of the DAP12 adapter protein plays a significant role in formation of multinuclear osteoclasts and that DAP12 and DARs likely participate in the regulation of bony remodeling.


RESORPTION OF BONE is the unique function of the osteoclast, a cell type derived from hematopoietic precursor cells in the monocyte/macrophage lineage.(1) Osteoclast development requires the activation of specific cell surface receptors on preosteoclasts, with an absolute requirement for stimulation by RANKL (also known as TRANCE, ODF, OPGL, TNFRSF11) and macrophage colony-stimulating factor (M-CSF).(1, 2) Other immunomodulatory receptors that influence osteoclast development and function are less well understood. We have examined pre-osteoclasts and osteoclasts for the expression and function of a recently described group of receptors in myeloid cells, the DAP12-associated receptors (DARs). DARs are defined by their functional association with the signaling adapter protein DAP12 and can mediate cellular activation and maturation in myeloid lineage cells.(3–5) DAP12 deficiency has recently been associated with bony abnormalities in both mice and humans, suggesting a role for this signaling pathway in regulation of osteoclast development and/or function.(6-9)

DAP12 is a transmembrane protein found in myeloid cells, including macrophages, granulocytes, monocytes, and dendritic cells, as well as in natural killer (NK) cells and some T-cells. DAP12 is expressed as a homodimer at the cell surface and is always paired with a DAR through complementary charged amino acid residues in the transmembrane domains of each protein. In mouse myeloid cells, DARs include triggering receptor expressed on myeloid cells (TREM) 1, 2, and 3; myeloid DAP12 associated lectin (MDL)-1; and NKG2D.(10–15) DAP12 has an extremely short extracellular domain; thus, it is not likely to bind ligand directly.(3) The cytoplasmic domain of DAP12 contains an immunoreceptor tyrosine-based activation motif (ITAM) that is phosphorylated on ligation and/or ligand binding of its DAR.(3, 16) DAP12 phosphorylation leads to recruitment of the Syk and ZAP70 tyrosine kinases and initiation of an intracellular signaling cascade.(4, 12, 16)

A rare human disease, polycystic lipomembranous osteodysplasia with sclerosing leukoencephalopathy (PLOSL; also called Nasu-Hakola disease), was recently found to be associated with deletions or loss of function inactivating mutations in either the DAP12 or TREM2 gene.(6–8) Multiple bony cysts develop in the spongy bones of these patients, leading to pathological fractures and severe arthritis by 30 years of age. Patients also suffer from severe dementia and a fatal neurodegenerative disorder.(17) The finding that functional defects in either DAP12 or TREM2, a DAR, are correlated with the same disease phenotype strongly suggests that absence of the TREM2/DAP12 signaling complex in osteoclasts contributes to the development of bony disease.(7) Recently, Kaifu et al.(9) examined the phenotype of DAP12-deficient mice with a mixed 129/SvJ and C57BL/6 background and observed osteopetrosis and thalamic hypomyelinosis, showing abnormalities in the brain and bones of these mice. We have further explored the potential role of DAP12 and its associated receptors in osteoclast development in normal and DAP12-deficient mice.



Homozygous DAP12-deficient mice (previously described) were backcrossed to C57BL/6 mice for nine generations.(18) Heterozygous DAP12+/− mice were generated by mating DAP12−/− mice to wildtype C57BL/6. C57BL/6 mice were purchased from Simonsen. All mice were maintained under specific pathogen-free conditions in the animal facility of the University of California, San Francisco (UCSF) or the San Francisco Veterans Affairs (VA) Medical Center. All experiments were performed in accordance with AAALC guidelines under UCSF- and VA-approved protocols. In vitro studies used cells isolated from mice at 8–14 weeks of age.


RAW264.7, 293T, and 3T3 cells (ATCC, Manassus, VA, USA) were maintained in RPMI media (Mediatech, Herndon, VA, USA) supplemented with 10% fetal bovine serum (Atlantic Biologics, Atlanta, GA, USA) and 1% glutamine Pen-Strep (Irvine Scientific, Santa Ana, CA, USA). Cells were transfected with pMX-FLAG DAP12 or pMX-FLAG TREM2B(19) using Fugene 6 (Roche, Indianapolis, IN, USA) using the manufacturer's protocol. Transfectants were selected in G418 (Mediatech, Herndon, VA, USA) at 1 mg/ml. Stable G418-resistant clones were screened for anti-FLAG M2 antibody (Sigma, St Louis, MO, USA) binding by flow cytometry, subcloned, and maintained in G418-selection media.

Osteoclast differentiation

Bone marrow monocyte/macrophage precursor cells were isolated from femurs by flushing with PBS using a 25-G needle. Cells were washed and lysed with RBC lysis buffer (0.16 M NH4Cl, 0.17 M Tris; pH 7.65) for 2 minutes at room temperature. Cells were cultured in complete α-MEM (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum, 1% glutamine Pen-Strep, and 10–20 ng/ml M-CSF (Sigma). After 2 days of M-CSF stimulation, nonadherent osteoclast precursors were transferred to a new plate and cultured in complete α-MEM with RANKL 75-100 ng/ml (R & D Systems, Minneapolis, MN, USA) with 10 ng/ml M-CSF for an additional 5-7 days, with addition of fresh media every 3 days. RAW264.7 or RAW transfectants were cultured in 96-well plates with 3000 cells/well in α-MEM supplemented with 10% fetal bovine serum, 100 ng/ml RANKL, and 2 ng/ml TGFβ (R & D Systems). Media were changed every 3 days. For stimulation of FLAG-DAP12 or TREM2, plates were coated with goat F(ab)′2 fragments to mouse IgG (ICN Pharmaceuticals, Costa Mesa, CA, USA) at 10 μg/ml, and anti-FLAG M2 antibody (Sigma), or isotype-matched control antibody (WES 6.1, anti-human SIRPα (not cross reactive with murine SIRPα), was added at a final concentration of 20 μg/ml and present throughout the duration of culture.

RNA isolation and reverse transcriptase-polymerase chain reaction

After differentiation, cell cultures were washed extensively to remove the nonadherent cells, and RNA was recovered from the adherent subset, which corresponds with the TRACP+ strongly adherent cells representing osteoclasts. Total RNA was isolated from cultured cells using TRIzol reagent (Invitrogen), as directed by the manufacturer. The isolated RNA pellet was air-dried and resuspended in 1 mM EDTA. Reverse transcription (RT) reactions used 1 μg of oligo-dT (Amersham Pharmacia, Piscataway, NJ, USA), 5 μg of RNA, and H2O up to 20 μl, were heated briefly to 95°C, and incubated at 42–55°C. RT-Mix (30 μl) was added and incubated at 42°C for 1 h. RT-Mix (Roche) contains 1 μl RNase-inhibitor, 10 μl 5× RT buffer, 2 μl 10 mM dNTPs, 1 μl AMV-RT, and 16 μl H2O. RNA was denatured with 0.5 μl 0.5 M EDTA and 2 μl 1 M NaOH and boiled for 3 minutes. Two microliters 1 M HCl, 5.5 μl 3 M NaOAc, and 145 μl 100% ethanol were added, and cDNA was precipitated, washed with 70% ethanol, air-dried, and resuspended in 50 μl TE. Each 30-μl PCR reaction used 2 μl of cDNA, 3 μl 10× buffer, 0.5 μl TAQ polymerase (Roche), and 1 μl 10 mM dNTPs (Roche). A Perkin Elmer GeneAmp Cycler was used with the following program: 10 minutes at 94°C, 30 s at 50°C, 30 s at 72°C, and 30 s at 94°C for 30 cycles, except for GAPDH (26 cycles) and NKG2D (35 cycles). Murine primer pairs (each spanning >1 intron) used were GAPDH 5′-ACCACAGTCCATGCCATCAC, 3′-TCCACCACCCTGTTGCTGTA; RANK 5′-AAGATGGTTCCAGAAGACGGT, 3′-CATAGAGTCAGTTCTGCTCGGA; calcitonin receptor 5′-TTTCAAGAACCTTAGCTGCCAGAG, 3′-CAAGGCACGGACAATGTTGAGAG; integrin β3 5′-CTGGTAAAACGCGTGAAT, 3′-CGGTCATGAATGGTGATGAG; cathespin K 5′-ACGGAGGCATCGACTCTGAA, 3′-GATGCCAAGCTTGCGTCGAT; OSCAR 5′-TCATCTGCTTGGGCATCATA, 3′-ACAAGCCTGACAGTGTGGTG; DAP12 5′-CTCTGGAGCCCTCCTGGT, 3′-TGCCTCTGTGTGTTGAGGTC; MDL-1 5′-AAGGACATTACCGAGCAGGA, 3′-CTTGGTCAGAAAGGGAGTTTT; TREM2 5′-ATGGGACCTCTCCACCAGTT, 3′-GGGTCCAGTGAGGATCTGAA; TREM3 5′-ACTTGCCTTGGGGCCATT, 3′-CATTCATTTCGCAGATCCAG; NKG2Dshort 5′-TCCCTTCTCTGCTCAGAG, 3′-TTACACCGCCCTTTTCATGCAGATG; and NKG2Dlong 5′-CAGGAAGCAGAGGCAGATTATCTC, 3′-TTACACCGCCCTTTTCATGCAGATG.

TRACP staining

After 5–7 days in culture, cells were fixed with 3.7% formaldehyde in PBS for 10 minutes. Plates were washed twice in PBS, incubated for 30 s in 50% acetone/50% ethanol, and washed with PBS. Cells were stained for TRACP using a kit (product 435; Sigma) according to manufacturer's instructions. Multinucleated (>2 nuclei), TRACP+ cells were counted by light microscopy.

Actin rings

Osteoclasts were fixed as above for TRACP staining. Plates were incubated with 0.2–0.5 ml of 0.1% Triton X-100/PBS for 20 minutes to permeabilize the cells. Twenty microliters per well of Alexa Fluor 546 phalloidin (Molecular Probes, Eugene, OR, USA) was added, and plates were incubated 30 minutes in the dark at room temperature. Plates were washed with PBS and analyzed by fluorescent microscopy.

Resorption assays

Bone marrow macrophages were harvested as described above. Macrophages (1 × 105) were plated on dentine discs (IDS Ltd., Tyne and Wear, UK) or BD BioCoat Osteologic Discs MultiTest Slides (BD Biosciences, Bedford, MA, USA) and treated with RANKL 100 ng/ml and M-CSF 10 ng/ml for 5–10 days. Media were changed every 3 days. Dentine discs were fixed with 3.7% formaldehyde for 30 minutes, and cells were gently removed with a tissue before staining with 1% toluidine blue in 0.5% sodium tetraborate solution. Resorption pits were visualized by light microscopy. BD BioCoat discs were treated with bleach and agitated for 5 minutes to remove cells, followed by five washes with dH2O and air-dried. Resorption pits were analyzed by light microscopy, and digital images were recorded. The resorption area was determined by maximizing brightness and contrast controls in Adobe Photoshop (Adobe Systems, Inc., San Jose, CA, USA) and inverting the image. Using the histogram function, the percentile of pixels in the top 25% of the gray scale was determined as a representation of the total resorbed area.

Microcomputed tomography

The right proximal tibial metaphyses and distal femur were fixed serially in 10% formaldehyde and 70% ethanol and measured without further sample preparation with a desktop microcomputed tomography (μCT) scanner (μCT-20: Scanco Medical, Bassersdorf, Switzerland).(20) For image acquisition, the bone was placed in a 9-mm holder and scanned. The image consisted of 160 slices at nine voxel size in all three axes. Evaluation was done on 120 slices, initiating at a distance of 1 mm from the lower end of the growth plate(21) and encompassing a volume of 1. The trabecular and the cortical regions were separated with semi-automatically drawn contours.(20, 21) The complete secondary spongiosa of the proximal tibia was evaluated, thereby completely avoiding sampling errors incurred by random deviations of a single section. The resulting grayscale images were segmented using a low-pass filter to remove noise and a fixed threshold to extract the mineralized bone phase.(21, 22) The same threshold setting was used for all samples. From the binarized images, structural indices were assessed with three-dimensional techniques for trabecular bone. Relative bone volume (BV/TV), trabecular number (Tb.N), and thickness (Tb.Th) were calculated by measuring three-dimensional distances directly in the trabecular network(22) and taking the mean over all voxels. The diameter of spheres filling the structure was taken as Tb.Th. and the inverse of the mean distances of the skeletonized structure was calculated as Tb.N. The complete structure is distance-transformed, thickness and separation for every point was calculated, and the histograms of these indices were determined. Using a tetrahedron meshing technique generated with the marching cube method,(23) bone surface (BS) was calculated. By displacing the surface of the structure in infinitesimal amounts (dr), the structure model index (SMI) was calculated as SMI = 6 × (B × dBS/dr)/BS2.(24) The SMI quantifies the plate versus rod characteristics of trabecular bone, in which an SMI of 0 pertains to a purely plate-shaped bone, an SMI of 3 designates a purely rod-like bone, and values in between stand for mixtures of plates and rods.(24) Furthermore, connectivity density based on the Euler number was determined.(21, 25) In addition, a three-dimensional cubical voxel model of bone was built, and cortical thickness (Ct. Th) was measured.(25) Groups were compared using the non-parametric Kruskal-Wallis test. When Kruskal-Wallis testing showed overall significant differences among groups, we applied Ryan's post hoc test to identify groups that were significantly different (SPSS Version 10; SPSS Inc., Chicago, IL, USA). Differences were considered significant at p < 0.05.

Generation of retrovirus

FLAG-tagged DAP12 cDNA was inserted into the pMX-pie vector, upstream of an IRES, which is followed by an eGFP cDNA sequence.(19) Retrovirus was generated in RPMI (supplemented with 10% FBS, glutamine-pen-strep) by cotransfecting equal amounts of the pMX-pie vector and PCL-Ecotropic packaging plasmid(26) into 293T cells using lipofectamine (Invitrogen) or Fugene (Roche) according to manufacturer's instructions. Retroviral-containing supernatants were harvested after 48 h and used to infect bone marrow osteoclast precursors or 3T3 cells.

Retroviral transduction of osteoclast precursors

Osteoclast precursor cells were isolated from bone marrow as described above and cultured 24 h in complete MEM containing 20 ng/ml M-CSF. Retrovirus encoding FLAG-DAP12 and GFP or GFP alone (pMX-pie control) was used to transduce bone marrow osteoclast precursors from C57BL/6 or DAP12−/− mice. After a 24-h culture in M-CSF, cells were replated in 24-well plates at 1–2 × 105 cells/well in 0.25 ml of enriched α-MEM containing 20 ng/ml M-CSF, 20% FBS, glutamine-pen-step, and 4 mM HEPES (pH 7.4). Retroviral supernatant (0.75 ml) containing 4 μg/ml polybrene was added to each well, and plates were centrifuged for 90 minutes at 1141g at room temperature. Two hours after addition of the retroviral supernatant, it was removed and replaced with 1 ml of α-MEM containing 20 ng/ml M-CSF, 10% FBS, and glutamine-pen-step. The transduction procedure was repeated the next day, after which cells rested in supplemented MEM overnight. Cells were then placed in fresh α-MEM containing 10% FBS, glutamine-pen-strep, 10 ng/ml M-CSF, and 75 ng/ml RANKL for 7-10 days. Control 3T3 cells were transduced with retrovirus in parallel using the same centrifugation conditions, but without addition of enriched media to the cells. Efficiency of transduction was assessed by expression of GFP. Bone marrow monocyte/macrophage precursor cells retrovirally transduced in parallel, but grown in M-CSF alone for 7-10 days, were examined by flow cytometry analysis for expression of GFP using a FACScan (BD Biosciences).


Osteoclasts derived from C57BL-6 mice express TREM-2, TREM-3, MDL-1, and NKG2D mRNA

Osteoclasts were derived in vitro from bone marrow macrophages by 7 days of treatment with M-CSF and RANKL. RNA isolated from these osteoclasts was examined for expression of a panel of DARs by RT-PCR analysis and compared with expression of these transcripts in bone marrow-derived macrophages treated with M-CSF alone for 2 days. As shown in Fig. 1, macrophages in M-CSF alone expressed DAP12, MDL-1, and low levels of NKG2D and TREM-2. In contrast, osteoclasts derived in vitro from C57BL/6 mice expressed mRNA for DAP12, TREM-2, TREM-3, and MDL-1, but not TREM-1. The short form of NKG2D (NKG2Ds), which was recently demonstrated to associate with DAP12 and DAP10, was also expressed along with low levels of the long splice variant of NKG2D (NKG2Dl), which pairs only with the DAP10 adapter protein.(15) Thus, osteoclasts derived in vitro from bone marrow precursor cells express at least four DARs. Cell surface expression of DARs on murine osteoclasts and preosteoclasts cannot be confirmed at this time because antibodies are not currently available.

Figure FIG. 1..

Expression of DAP12 and DARs in C57BL/6 osteoclasts. The expression of DAP12 and several DARs was assessed by semiquantitative RT-PCR. Bone marrow monocyte/macrophage precursors were isolated from C57BL/6 mice and treated with M-CSF (10 ng/ml) alone (left) or M-CSF (10 ng/ml) and RANKL (75 ng/ml) to stimulate osteoclast differentiation (right). In vitro derived osteoclasts express DAP12, TREM-2, TREM-3, MDL-1, and splice variants of NKG2D, whereas M-CSF-treated bone marrow macrophages only express DAP12, MDL-1, and low levels of TREM-2 and NKGD2. GAPDH expression was used as the internal control. These data are representative of three experiments.

Tibias from DAP12-deficient mice show increased bone mass by μCT

To examine the role of DAP12 in bone remodeling in vivo, μCT analysis was performed on the right proximal tibia metaphyses and distal femur isolated from DAP12−/−, DAP12+/−, and C57BL/6 wildtype mice at 12 weeks of age. As shown in Fig. 2, the femur of DAP12−/− mice have increased bone mass compared with C57BL/6 wildtype controls. In the tibia, the bone volume fraction of DAP12−/− mice was 16 ± 6% compared with C57BL/6 mice, which have a bone volume fraction of 8 ± 2% (p < 0.01, Fig. 2B). Heterozygous DAP12 mice have a bone volume fraction that is the same as that in observed in C57BL/6 mice; thus, the changes observed in DAP12−/− mice are not because of a contribution of the original 129/SvJ background. The three-dimensional trabecular structural parameters are detailed in Figs. 2B-2G. Notably, 129/SvJ mice are known to have high trabecular bone volume compared with C57BL/6 mice,(27) and their trabecular bone volume fraction is 24 ± 4%. The difference in tibial bone density was also reflected in the distal femurs of the same mice, but was not as prominent. The femoral trabecular bone volume fraction measured 12 ± 5% in DAP12−/− mice and 8 ± 1% in C57BL/6 mice, a difference that did not reach statistical significance (Fig. 2C). Further analysis of the tibias demonstrated that DAP12−/− mice showed greater trabecular thickness than C57BL/6 mice (Fig. 2F, p < 0.05) and tended toward greater trabecular number (Fig. 2D) and connectivity density (Fig. 2G) than C57BL/6 or DAP12+/− mice. Cortical thickness was the same as in C57BL/6 mice (Fig. 2F). These findings show that C57BL/6 DAP12−/− mice have increased bone mass, and our observations suggest that DAP12 influences may differ in distinct local bony environments.

Figure FIG. 2..

DAP12-deficient mice have increased bone mass. Tibias and femurs from 12-week-old DAP12−/−, DAP12+/−, C57BL/6, and 129Sv/J mice were examined by three-dimensional μCT. (A) Three-dimensional reconstruction of distal femurs reveals increased bone mass and trabecular thickness of DAP12−/− mice (right) compared with C57BL/6 mice (left). Figure shown is representative of n = 9. (B-G) Three-dimensional trabecular structural parameters in the secondary spongiosa of the distal tibia, including bone volume fraction (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular connectivity density (ConnDens). DAP12−/− (n = 9), C57BL/6 (n = 12), DAP12+/− (n = 3), 129Sv/J (n = 5). Data are expressed as mean ± SD. Differences in BV/TV were statistically significant between DAP12−/− and C57BL/6 (*p < 0.01). Tibial trabecular thickness (Tb.Th) was significantly different between DAP12−/− and C57BL/6 (**p < 0.05). There were no statistically significant differences between DAP12+/− and C57BL/6 mice. Differences between 129Sv/J and C57BL/6 mice were significantly different at all parameters (*p < 0.01, **p < 0.05). Differences between groups was assessed using nonparametric Kruskal-Wallis testing with Ryan's post hoc test to identify groups with significant differences defined at p < 0.05.

DAP12-deficient mice form mononuclear osteoclasts in vitro

We examined the in vitro differentiation of osteoclasts isolated from the bone marrow of DAP12−/− mice. Surprisingly, only mononuclear, TRACP+ osteoclasts could be derived from DAP12−/− macrophages (Fig. 3B) using the same osteoclast differentiation conditions that lead to multinucleated TRACP+ osteoclasts from cells from C57BL/6 mice (Fig. 3A). Hundreds of multinucleated (>3 nuclei), TRACP+ osteoclasts routinely formed in cultures from C57BL/6 mice and typically comprised 50% of the TRACP+ cells, whereas no osteoclasts with >2 nuclei formed from DAP12−/− mice. Staining with phalloidin to evaluate for actin rings showed that some of these mononuclear TRACP+ cells formed small actin rings (Fig. 3D), although significantly different from the large rings formed by multinucleated osteoclasts (Fig. 3C). Despite being mononuclear, these osteoclast-like cells from DAP12−/− mice retained their ability to resorb dentine and form pits during in vitro culture on dentine slices in the presence of M-CSF and RANKL (Fig. 3F). This was consistent with the finding that, although DAP12−/− mice have increased bone mass, they do not exhibit severe osteopetrosis causing abnormalities in bony development or formation of teeth. Although osteoclasts are generally defined as multinucleated cells, we suggest that the mononuclear TRACP+ cells derived from DAP12−/− bone marrow precursors are indeed osteoclasts, given their ability to resorb bone and their expression of markers of osteoclast differentiation as shown in Fig. 4A. By RT-PCR, these cells expressed the following markers consistent with differentiation to osteoclasts: integrin β3, OSCAR receptor, and cathepsin K. Transcripts for calcitonin receptor seemed to be expressed only at low levels in our semiquantitative analysis. Given their mononuclear phenotype, we were interested in the functional capacity of these cells to resorb bone. We analyzed in vitro differentiated osteoclasts developed on calcium phosphate discs to quantify resorption area by individual osteoclasts. As shown in Fig. 4B, DAP12-deficient osteoclasts resorbed approximately 50% less calcium phosphate substrate in vitro than osteoclasts derived from C57BL/6 mice. While in vitro DAP12−/− cells only formed mononuclear osteoclasts, we were able to show the presence of multinucleated osteoclasts in histological sections of bone from the tibias of C57BL/6 DAP12−/− mice (data not shown). This is similar to the findings of Kaifu et al.(9) in DAP12−/− mice of 129/C57 mixed background. These findings suggests that DAP12 associated receptors may be of particular importance in vitro with culture conditions containing only osteoclast precursors, whereas in the bony microenvironment, other receptors may be able to partially compensate for the absence of DAP12.

Figure FIG. 3..

Osteoclast-like cells derived in vitro from DAP12-deficient mice are mononuclear and resorb dentine. (A and B) TRACP staining of osteoclast cultures derived in vitro from bone marrow under the influence of M-CSF and RANKL. Only mononuclear, intensely TRACP+ cells are derived from DAP12−/− mice (B, arrow) compared with large, multinucleated TRACP+ cells from C57BL/6 mice (A, arrow). (C and D) A and B are stained with phalloidin to evaluate actin ring formation. Osteoclast-like cells from both (C) C57BL/6 and (D) DAP12−/− (D) mice exhibit actin ring formation (arrows). (E and F) Osteoclast-like cells from (E) C57BL/6 and (F) DAP12−/− mice form pits (arrows) on dentine, indicating that these osteoclasts are functional. Data are representative of five experiments.

Figure FIG. 4..

Osteoclast-like cells derived in vitro from DAP12−/− mice express osteoclast markers, but resorb poorly compared with those from C57BL/6 mice. (A) Markers expressed by mature osteoclasts are evaluated by RT-PCR of in vitro derived osteoclasts form DAP12−/− mice. Calcitonin receptor (CTR), cathepsin K (CATH K), integrin β3 (INT B3), OSCAR, and RANK were present. GAPDH was the internal control. These data were representative of more than five experiments. (B) In vitro derived osteoclast-like cells from C57BL/6 or DAP12−/− mice plated on submicron calcium-phosphate discs were treated with RANKL and MCSF for 5–10 days. The area resorbed per osteoclast was calculated (see the Materials and Methods section). DAP12−/− osteoclasts resorb 50% less substrate than control C57BL/6 osteoclasts (*p < 0.0008, mean ± SEM). Data are representative of more than three experiments with triplicate samples.

Reconstitution of DAP12 expression in DAP12−/− precursors restores formation of multinuclear osteoclasts in vitro

To prove that the absence of multinucleated osteoclasts derived in vitro from DAP12-deficient mice was caused by the absence of the DAP12 molecule, we used retroviral transduction to express DAP12 in DAP12−/− osteoclast precursors. Retrovirus encoding FLAG-tagged DAP12 and green fluorescent protein (GFP; under control of an IRES segment distal to FLAG-DAP12) or GFP alone (control) was used to transduce bone marrow-derived macrophages isolated from C57BL/6 wildtype or DAP12-deficient mice. After transduction, cells were treated with 10 ng/ml M-CSF and 75 ng/ml RANKL for 7–10 days. Efficiency of transduction by expression of GFP was 30-50% at 7-10 days, as assessed by flow cytometric analysis of cells transduced in parallel, but maintained in M-CSF without RANKL (Fig. 5A). Infection of DAP12−/− osteoclast precursors with DAP12 retrovirus led to formation of 17 ± 2.4 multinucleated osteoclasts per well (p < 0.0001; Figs. 5C and 5D). Osteoclasts formed were clearly multinuclear and TRACP+ but of smaller size than those in wildtype osteoclast cultures. C57BL/6 bone marrow-derived macrophages transduced with DAP12 retrovirus rapidly formed large multinucleated osteoclasts by day 3-5 after addition of RANKL to the culture, whereas C57BL/6 cells transduced with control virus formed multinucleated osteoclasts at days 5-7. In both cases, multinucleated osteoclasts formed were similar in appearance and number to those that were sham-transduced (data not shown). DAP12−/− bone marrow-derived macrophages transduced with control virus or sham-transduced did not develop TRACP+ multinuclear osteoclast-like cells after incubation with M-CSF and RANKL, but instead developed only TRACP+ mononuclear osteoclasts (Fig. 5B). Thus, restoration of DAP12 expression in DAP12−/− osteoclast precursors rescued formation of multinucleated osteoclasts, which strongly supports a role for DAP12 itself in the development of osteoclast multinucleation in vitro.

Figure FIG. 5..

Expression of DAP12 in DAP12−/− pre-osteoclasts rescues multinucleation during in vitro osteoclastogenesis. (A) Approximately 30% of bone marrow-derived cells transduced with retrovirus encoding DAP12 and GFP expressed GFP at 7–10 days after transduction compared with 45% of control virus. Mock-transduced cells did not express GFP. (B) Mock or control retrovirus transduction failed to rescue the ability of DAP12−/− osteoclasts to become multinuclear (arrow). (C) Multinucleated TRACP+ osteoclasts (arrow) formed in response to RANKL and M-CSF only after transduction with DAP12 retrovirus. (D) DAP12 expression in DAP12−/− pre-osteoclasts led to 17 ± 2.4 multinuclear TRACP+ osteoclasts per well compared with control virus (*p < 0.0001, mean ± SEM, n = 3). Data shown are representative of more than three experiments with triplicate samples.

Both DAP12 and DARs are present in osteoclast-like cells differentiated in vitro from RAW264.7

To determine if DAP12/DARs are present in osteoclasts derived from the pre-osteoclastic tumor cell line RAW264.7, we examined their expression by using RT-PCR. RAW264.7 cells will differentiate in osteoclast-like cells under the influence of RANKL and TGFβ but do not require exogenous M-CSF as they exhibit autocrine production of mCSF. As shown in Fig. 6A, untreated RAW264.7 cells expressed mRNAs for DAP12, TREM-2, TREM-3, and MDL-1. The same receptors were found in osteoclast-like cells derived from this line after stimulation with RANKL and TGFβ. Message for the NKG2D receptor was not detected in this cell line, before or after differentiation, and there was no significant expression of TREM-1. As shown in Fig. 6B, osteoclast-like cells derived from RAW264.7 expressed markers of osteoclast differentiation, including calcitonin receptor, integrin β3, OSCAR receptor, and cathepsin K, similar to osteoclasts obtained from bone marrow-derived precursors.

Figure FIG. 6..

DAP12 and several DARs are expressed in RAW264.7 cells. (A) The expression of DAP12 and several DARs was investigated by means of semiquantitative RT-PCR. DAP12, TREM-2, −3, and MDL-1 were present in both untreated RAW264.7 cells and in osteoclast-like cells (OCL) derived from RAW264.7 cells under the influence of RANKL and TGFβ. (B) The strongly adherent, multinuclear TRACP+ osteoclast-like cells derived from RAW264.7 expressed appropriate markers of osteoclasts, including calcitonin receptor (CTR), cathepsin K (CATH K), integrin β3 (INT B3), OSCAR, and RANK. GAPDH was the internal control. Data are representative of more than five experiments.

Activation of constitutively expressed DAP12 or TREM-2 in RAW264.7 leads to increased formation of TRACP+ multinucleated osteoclast-like cells in vitro

To assess the role of DAP12 activation in osteoclast differentiation, we generated stable transfectants of RAW264.7 cells expressing FLAG-tagged mouse DAP12 or FLAG-tagged TREM-2. The cell surface expression of FLAG-DAP12 in RAW264.7 likely occurs through its association with the endogenous DARs, TREM-2, TREM-3, and MDL-1. Differentiation of the RAW264.7 DAP12-expressing clone (RAW.DAP12) or TREM-2-expressing clone (RAW.TREM-2) with RANKL and TGFβ for 5 days led to osteoclast differentiation, similar to that observed in untransfected cells. Furthermore, the osteoclast-like cells from RAW.DAP12 or RAW.TREM-2 were similar to those derived from parental RAW264.7 cells in their expression of osteoclast differentiation markers (data not shown). Thus, the expression of FLAG-tagged DAP12 or TREM-2, in addition to the endogenous DAP12 and DARs, did not alter RAW264.7 differentiation to osteoclasts. Thus, in these cells, it is unlikely that the level of DAP12 or DARs is limiting in the development of multinucleated osteoclasts. It is possible that the absence of sufficient ligand to stimulate DARs and DAP12 is of greater importance.

To assess the effect of direct stimulation of DAP12 on osteoclast development, the FLAG-DAP12 on RAW.DAP12 cells was ligated with anti-FLAG antibody, which stimulates DAP12 signaling. Treatment of RAW.DAP12 cells with anti-FLAG antibody cross-linked with goat anti-mouse IgG F(ab′)2 led to a 3-fold increased number of multinucleated, TRACP+ osteoclast-like cells formed after 5 days of culture (Fig. 7D). Notably, the multinucleated TRACP+ cells that formed after DAP12 stimulation were not only increased in number, but also in size, with many nuclei, multiple pseudopod-like projections, and irregular morphology. This was not seen in parental RAW264.7 cells (Fig. 7B), indicating that the effect is mediated by stimulation through FLAG-DAP12, and not, for example, through Fc receptors on RAW264.7 cells. An isotype-matched control antibody caused no enhancement of osteoclastogenesis in either RAW264.7 cells (Fig. 7A) or RAW.DAP12 cells (Fig. 7C). No formation of multinuclear, TRACP+ cells was seen in RAW264.7 cells or RAW264.7 transfectants in the absence of RANKL and TGFβ. The number of multinucleated TRACP+ cells formed is quantified in Fig. 7E. As with RAW.DAP12 cells, stimulation of RAW.TREM-2 transfectants with anti-FLAG antibody also increased the formation of TRACP+ multinucleated osteoclast-like cells. These data indicate that direct stimulation of DAP12 or the TREM-2 DAR during in vitro osteoclastogenesis leads to increased formation of TRACP+ multinucleated osteoclast-like cells and supports our other findings suggesting a role for DAP12 and DARs in osteoclast development.

Figure FIG. 7..

Activation of constitutively expressed DAP12 or TREM-2 in RAW264.7 cells leads to increased formation of TRACP+ multinucleated osteoclast-like cells. RAW264.7 cells expressing mouse FLAG-tagged DAP12 (RAW.DAP12) or FLAG-tagged TREM-2 (RAW.TREM-2) were differentiated to osteoclasts-like cells by treatment with RANKL and TGFβ (see the Materials and Methods section). Expression of DAP12 or TREM-2 alone did not alter the number of osteoclast-like cells compared with control RAW264.7 cells (E; open bars). Treatment of transfectants with anti-FLAG antibody, to directly activate DAP12 or TREM-2, led to a 3-fold increased number in multinuclear TRACP+ osteoclasts-like cells in RAW.DAP12 (D and E, black bars) and RAW.TREM-2 (E, black bar). Anti-FLAG antibody treatment of RAW264.7 failed to stimulate multinucleation in response to RANKL and TGFβ (B and E, black bar). Isotype-matched control antibody treatment of RAW (A), RAW.DAP12 (C), or RAW.TREM-2 did not increase multinucleation (E, gray bars) (*p < 0.0002, mean ± SEM). Data representative of five experiments with triplicate samples.


We examined the role of the signaling adapter protein DAP12 in osteoclast development and function in the DAP12-deficient mouse and in osteoclast development and function in vitro. We show that C57BL/6 DAP12-deficient mice have increased bone density and that DAP12-deficient osteoclasts develop abnormally in vitro under the influence of M-CSF and RANKL and show decreased capacity for resorption. In these studies, we verify the presence of DAP12 in murine preosteoclasts and osteoclasts and show the spectrum of DARs expressed in these cells by mRNA expression. As antibodies for these recently described receptors become available, it will be important to show protein expression on preosteoclasts and osteoclasts. Here we show evidence for the involvement of DAP12 and DARs in differentiation to multinucleated osteoclasts.

Bone marrow monocyte/macrophage precursors from DAP12-deficient mice failed to form multinucleated osteoclasts in vitro in response to RANKL and M-CSF stimulation, suggesting an abnormality in osteoclast development and a role for DAP12 in the formation of multinucleated osteoclasts. The DAP12−/− mononuclear TRACP+ cells are functional “osteoclasts” in that they are able to resorb bone and express markers associated with mature osteoclasts. Interestingly, DAP12 was recently demonstrated to be one of three genes upregulated by the PU.1 transcription factor, which has previously been shown to be essential for osteoclast development.(28) Osteoclast function has been previously correlated with the degree of multinucleation(29); therefore, we were unable to determine if DAP12 plays a direct role in promoting the resorptive function of osteoclasts or whether it is the effect on development of multinuclear osteoclasts that indirectly affects the capacity for bone resorption.

Importantly, we demonstrated that reconstitution of DAP12 expression by retroviral transduction restores formation of multinucleated osteoclasts by DAP12−/− cells in vitro. In in vitro cultures, we did not observe formation of any multinucleated TRACP+ cells from DAP12−/− precursors. This result is in contrast to the observations of Kaifu et al.,(9) who observed the in vitro formation of a few small multinucleated osteoclasts from DAP12−/− cells, although the number was significantly reduced compared with wildtype mice. Potential explanations for these observed results include the differences in the background mouse strain or different in vitro culture conditions for osteoclast development. Genetic determinants leading to the marked differences in bone mass observed between C57BL/6 and 129/SvJ mouse strains have not been identified. Regardless, both of our studies show a defect in the in vitro formation of multinucleated osteoclasts by DAP12-deficient cells, and our study further shows that reconstitution of DAP12 expression can rescue this abnormality. In both strains of mice, multinucleated osteoclasts can be found in the bony microenvironment, Therefore, it is most likely that DAP12 signals are important but not the sole regulators of osteoclast multinucleation. Recent studies in humans deficient in DAP12 or TREM-2 have demonstrated similar findings, with significant defects in osteoclast differentiation in vitro.(30, 31) In the study by Paloneva et al.,(30) multinucleated osteoclasts are found in bone histology, while in vitro, osteoclast differentiation is defective, producing primarily mononuclear or abnormally shaped small multinucleated osteoclasts.

Although both DAP12-deficient mice and humans have bony abnormalities, the phenotypes are not identical. In humans with PLOSL, large bony cysts filled with lipid are present(7, 8, 17) in skeletal regions that have a high percentage of cancellous bones including the carpals, tarsals, wrists, and ankles. Additionally, there is a relative sparing of the skeletal regions with a high cortical bone content such as the mid-shaft of the long bones. Kaifu et al.(9) suggested that osteosclerosis, possibly resulting from osteoclast dysfunction, could lead to increased fractures, which have been described as part of the bony changes in PLOSL patients and might contribute to cyst formation.

We have found that our DAP12−/− mice have increased trabecular bone mass in the proximal tibia and metaphysis, and less prominently in the distal femur compared with control animals. These results suggest that DAP12 influences may vary in different local bony environments. Although we studied only female mice at 12 weeks of age, these data should be representative of the adult mouse skeleton. Kaifu et al.(9) has previously shown that DAP12-deficient mice of a mixed genetic background have increased bone mass at 6, 13, and 48 weeks. In both mice and humans, the trabecular-rich regions of bone that seem to be most sensitive to DAP12 deficiency have large surface areas that are in close proximity to nests of bone marrow cells that may include not only osteoblasts but a variety of immune cells. These bony microenvironments may lead to osteoclast differentiation and activation through local cell-cell interactions, cytokine, hormonal factors, or integrin activation. The relative sparing of cortical bone in DAP12 deficiency may indicate that osteoclast dysfunction within the haversian system in response to osteotropic hormones is less prominent. In humans with PLOSL, all bones affected are those that are formed by endochondral ossification; no lesions have been noted in the skull, clavicle, or mandible, bones formed by intramembranous ossification.(30) We suggest that DAP12 and DARs may be of particular importance in response to specific stimuli of bony remodeling and likely play an overlapping role with other immunoregulatory receptors and signaling pathways.

The DARs include a group of innate immune receptors described on other myeloid lineage cells that use DAP12 to initiate intracellular signaling cascades.(3) DAP12 contains an ITAM signaling motif that functions as a docking site for the syk and/or ZAP70 tyrosine kinases. The subsequent cascade involves recruitment and activation of PI3K, PLCγ1, and p44/p42 ERK activation,(3, 12, 16) which includes signals that are known to modulate osteoclast development and function. In a recent study examining the role of M-CSF signaling in β3 integrin−/− cells, the activation of ERK kinases was proposed to be a critical factor in upregulation of c-fos and rescue of in vitro osteoclast development.(32) Our findings suggest that DAP12 may similarly influence osteoclast development through modification or amplification of signals triggered through cytokine receptors or integrins.

The specific effect on multinucleation in vitro suggests that, under these conditions, DAP12 signals can trigger changes in actin polymerization and cytoskeletal organization essential for these processes. Under these in vitro conditions, it seems likely that a specific DAR interaction with its ligand may be critical for triggering the appropriate signaling cascade required for formation of multinucleated cells. Although specific ligands for the myeloid-associated DAR receptors TREM-2, −3, and MDL-1 are not yet identified, TREM-2 was recently shown to bind a wide variety of anionic ligands, most likely through pattern recognition similar to scavenger receptors and several Toll-like receptors.(33) Identifying the potential ligand(s) for TREM-2 in the bony environment will be necessary to decipher the role of TREM-2/DAP12 signaling in vivo. It is not known whether DAP12 signals regulate other cell surface receptors on osteoclasts that have been implicated in macrophage fusion, leading to formation of multinucleated osteoclasts.(34)

We also show that direct stimulation of DAP12 or TREM-2 in RAW264.7 cells leads to increased formation of multinucleated osteoclast cells. Notably, these cells are increased in both number and size, such that the morphology of the stimulated RAW.DAP12 osteoclast-like cells was larger and somewhat irregular. The unusual morphology of our osteoclast-like cells derived from RAW264.7 compared with bone marrow-derived osteoclasts is similar to other studies of RAW cell-derived osteoclasts. These findings indicate that direct stimulation of DAP12 signals may influence osteoclast development. The stimulated RAW.DAP12 osteoclast-like cells have an appearance reminiscent of highly multinucleated abnormal osteoclasts observed in Paget's disease.(35) This is of interest given the long controversial history of possible association of viral infection and Paget's disease,(36, 37) because DARs in NK cells have been recently demonstrated to interact directly with viral proteins and virally or stress-induced ligands expressed on virus-infected cells.(38–40) Clearly, identification of ligands for DARs in the osteoclast environment will be critical to understanding the situations in which DAP12 signals are of greatest importance.

In summary, our studies indicate that stimulation of DAP12 adapter protein plays a significant role in the formation of multinucleated osteoclasts and that DAP12 and DARs likely participate in the regulation of bone remodeling.


This work was supported by the Department of Veterans Affairs, American Cancer Society (MCN), the Rosalind Russell Arthritis Center (MBH, MCN, and NEL), and National Institutes of Health Grants CA89294 (LLL), AR43052, and AR48841 (NEL). MBH is an Abbott Scholar in Rheumatology Research. KO is supported by Human Frontier Science Program Long-term Fellowship. LLL is an American Cancer Society Research Professor. We thank Marina Abramova for assistance with gene-deficient mice, Neil Fishbach for assistance with retroviral methods, Bill Seaman and Bob Nissenson for review of the manuscript, and Gail Cassafer for assistance with histology.