Signal adaptor DAP10 associates with MDL-1 and triggers osteoclastogenesis in cooperation with DAP12

DAP10-Deficient Mice Exhibit Mild Osteopetrosis Caused by a Reduction in the Osteoclast Number in Bone Tissues.

To clarify whether DAP10 could be an activating adaptor for osteoclast development, we first performed bone morphometric analyses of DAP10-deficient (DAP10^−/−) mice (see SI Text). Tibial bone sections from 6-month-old DAP10^−/− mice were found to exhibit slight increases in the metaphyses and trabecular bones, indicating mild osteopetrosis (Fig. 1A). Osteopetrosis was not evident in mice at 1.5 and 3 months of age, but it was significant in 6-month-old DAP10^−/− mice (Fig. 1B). This observation was in sharp contrast to that for DAP12^−/− mice, in which mild osteopetrosis was evident even in 1.5-month-old animals and was significant afterward at least up to 12 months of age (12). This difference is probably because the contribution of DAP10 to osteoclast development depends on DAP12 (see below). A small number of tartrate-resistant acid phosphatase (TRAP)⁺ osteo clasts was seen in bone tissues from FcRγ/DAP12/DAP10 triple knockout (FcRγ^−/−DAP12^−/−DAP10^−/−) mice and FcRγ^−/−DAP12^−/− mice (Fig. S1), supporting the above notion.

Bone morphometric analyses of the tibia of these mice at 6 months of age revealed significant reductions in osteoclast-related parameters (Fig. 1C Top), indicating reductions in the number and function of osteoclasts in DAP10^−/− mice, while keeping comparable values for osteoblastic parameters (Fig. 1C Middle and Bottom). These parameters for 1.5- and 3- month-old DAP10^−/− mice were comparable with those for WT mice (Fig. S2). These results indicate that the increased bone mass in DAP10^−/− mice was caused by attenuated osteoclastogenesis, although the overall magnitude of the defect was apparently smaller than that in DAP12^−/− mice, as judged from the occurrence of osteopetrosis in DAP12^−/− animals at as early as 1.5 months of age (12). We found that DAP10 protein is expressed in immature, prefusion osteoclasts (pOCs), mature, multinucleated osteoclasts (mOCs), and bone marrow-derived macrophages in vitro (Fig. 1D Upper), but not in osteoblasts (Fig. S3) on immunoblot analysis with DAP10-specific antibodies (Fig. S4), suggesting a role of DAP10 in the development of osteoclasts.

In Vitro Osteoclastogenesis Was Impaired in the Absence of DAP10.

With the RANKL/M-CSF system, culturing of bone marrow-derived monocyte/macrophage lineage cells (BMMs) as osteoclast precursor cells from DAP10^−/− mice resulted in a significant reduction in the number of TRAP⁺ mOCs, especially with suboptimal concentrations of RANKL and M-CSF (5–20 and 1–10 ng/mL, respectively) (Fig. 2 A and B). Concomitant with the reduction in mOCs, the resorptive pit area formed by DAP10^−/− mOCs was also reduced (Fig. 2C). In the presence of suboptimal concentrations of RANKL, an excess concentration (100 ng/mL) of M-CSF did not rescue the reduced osteoclastogenesis (Fig. 2D). Reduced osteoclastogenesis was also observed with the osteoblast coculture system (Fig. 2E). Flow cytometric analyses (Fig. S5) verified that the cell-surface expression levels of the M-CSF receptor c-fms and the RANKL receptor RANK on pOCs were comparable among B6, DAP10^−/−, DAP12^−/−, and DAP12^−/−DAP10^−/− (Fig. S6) cells, indicating that the reduced development of DAP10^−/− osteoclasts is not caused by reduced expression levels of c-fms and RANK but a deficit in DAP10. These observations suggest that the RANKL/M-CSF-mediated signal is insufficient in DAP10^−/− osteoclast precursor cells.

Identification of MDL-1 as a Distinct Cell-Surface Receptor That Associates with DAP10 in Osteoclasts.

How does DAP10 participate in osteoclastogenesis? Although DAP10 physiologically associates with the only known partner, NKG2D, in immune cells (22), we failed to detect a significant amount of NKG2D mRNA in osteoclast lineage cells after GeneChip (9) and RT-PCR analyses (Fig. S7). Therefore, the formation of a functional NKG2D–DAP10 complex in osteoclasts is unlikely. To find a candidate receptor that associates with DAP10, we prepared macrophages, pOCs, and mOCs from B6 and DAP10^−/− mice, which were surface-biotinylated, either glycosidase-treated or left untreated, and then immunoprecipitated with anti-DAP10 antibodies. Comparison of the profiles for surface-biotinylated proteins of DAP10-sufficient and -deficient cells enabled us to detect a discrete band of ≈27-kDa glycosylated protein species, whose deglycosylated form was ≈18 kDa (Fig. S8A). We excised a gel strip containing this protein and subjected it to TOF/MS analysis (see SI Text). Sixty-six tryptic peptide sequences obtained were searched for in the NCBInr database (www.protein.sdu.dk/gpmaw/GPMAW/Databases/NCBInr/ncbinr.html), and 6 murine proteins were shown to be significant hits, among which C-type lectin domain family 5 member A (i.e., MDL-1) (25) was the only notable one in the context of our investigation (Fig. S8B). The expression of mRNA for MDL-1 was confirmed by RT-PCR analysis of samples from pOCs, mOCs, and macrophages as described (26) (Fig. S8C). MDL-1 was identified originally (25) as a type II transmembrane glycoprotein of the C-type lectin superfamily (Fig. S8D), expressed as a DAP12-dependent manner as both long and short forms possibly arising from alternative splicing, but neither the physiological role of nor the ligand for MDL-1 has been elucidated. Based on the knowledge obtained in these preceding studies, our present observations indicate that MDL-1 could associate with DAP10 in addition to DAP12 in macrophages and osteoclasts and might function as an activating receptor in osteoclasts. Although C-type lectin superfamily molecules are often expressed on the cell surface as a disulfide-linked homodimer like NKG2D or Ly49C (22, 27), or a heterodimer such as CD94–NKG2A (28), the results from our immunoblot analyses indicated that the majority of the MDL-1 short and long forms exist as monomers (SI Text and Fig. S8E).

DAP10 Forms an MDL-1–DAP12/DAP10 Trimolecular Complex Depending Almost Solely on DAP12.

To determine whether MDL-1 expression solely depends on DAP12 or not, we performed flow cytometry of MDL-1 expression on macrophages and pOCs from mice deficient in either DAP12, DAP10, DAP12 and DAP10, or FcRγ (Fig. 3A). The MDL-1 expression levels on macrophages and pOCs were not changed by FcRγ deletion, but were reduced markedly in the absence of DAP10 and decreased to a minimum with either DAP12 or DAP12 and DAP10 double deletion. We verified that the MDL-1 mRNA levels in pOCs from these mutant mice were not reduced when compared with that in pOCs from B6 mice (Fig. S9). Thus, MDL-1 cell-surface expression primarily depends on DAP12, and partially, but substantially also depends on DAP10.

Given that the DAP12-associated receptor TREM-2 (18) has been implicated in osteoclastogenesis (29–31), it would be interesting to determine whether the expression of TREM-2 is also diminished in DAP10^−/− and DAP12^−/−DAP10^−/− osteoclasts like that of MDL-1. Our flow cytometric analysis (Fig. 3B) revealed that TREM-2 expression on pOCs did not change with the deficiency of DAP10 or DAP10 and DAP12, and in the deficiency of DAP12, as reported (29), indicating that the TREM-2 expression itself does not depend on DAP10 or DAP12, which is in sharp contrast to the case of MDL-1. However, this observation does not exclude the possibility that DAP10 could enhance TREM-2–DAP12-mediated signaling.

To confirm the DAP12- and DAP10-dependent expression of MDL-1, lysates of macrophages, pOCs, and mOCs derived from B6, DAP10^−/−, and DAP12^−/− mice were immunoprecipitated with anti-MDL-1 polyclonal antibodies and then subjected to immunoblotting with antibodies to MDL-1 or DAP10 and DAP12. As shown in Fig. 4A, top 2 panels), immunoprecipitation/immunoblotting analyses with MDL-1 antibodies and with or without N-glycosidase treatment revealed that macrophages, pOCs and mOCs from B6 mice (lanes 1–3) and macrophage-like RAW cells (lane 11) exhibited high expression of a diffuse, glycosylated 38- to 45-kDa protein species corresponding to the highly glycosylated, monomeric long form of MDL-1, and several discrete and a diffuse protein species ≈20–28 kDa corresponding to the differently glycosylated, monomeric short form of MDL-1. These protein species were confirmed to migrate at the positions consistent with the predicted core sizes for MDL-1 long and short forms after treatment with the combination of N- and O-glycosidases and neuraminidase (Fig. S10; see also Fig. S8E). We observed less abundant, long and short form MDL-1 bands for samples of macrophages, pOCs and mOCs from DAP10^−/− and DAP12^−/− mice (Fig. 4A, lanes 4–9).

As shown in Fig. 4A (bottom 2 panels), we observed DAP12 and DAP10 coimmunoprecipitation with MDL-1 for all 3 cell populations from WT B6 mice (lanes 1–3) and in RAW cells (lane 11). For DAP10^−/− cells, we observed DAP12 coprecipitation with MDL-1 (lanes 4–6) as expected from the results of flow cytometric analysis. Rather curiously, however, we did not detect DAP10–MDL-1 coprecipitation for DAP12^−/− cells (lanes 7–9). To check the overall expression levels of DAP10 and DAP12, and their possible interaction, we performed immunoprecipitation and immunoblot analyses of these 2 adaptors in macrophage lysates from B6, DAP10^−/− and DAP12^−/− mice (Fig. 4B). It was evident that both DAP10 and DAP12 were expressed in total macrophage lysates (Fig. 4B Upper, lanes 7 and 8 and Fig. 4B Lower, lanes 1 and 2). Interestingly, DAP10 expression was markedly reduced in the absence of DAP12 (Fig. 4B Upper, lane 9 and Fig. 4B Lower, lane 3) but importantly not lost completely (Fig. 4B Upper, lane 6), suggesting that the DAP10 protein is unstable in the absence of DAP12. This finding is in sharp contrast to in the case of IL-2-activated killer cells (LAK) in which DAP10 is abundantly expressed in the absence of DAP12 (Fig. 4B Upper, lane 10 and Fig. 4B Lower, lane 5) and is functional because DAP10-dependent NKG2D expression is normal in the absence of DAP12 (Fig. S11). Also, it was verified that DAP10 was coimmunoprecipitated with DAP12 (Fig. 4B Upper, lanes 2 and 4), suggesting that DAP10 is a component of one or more DAP12-receptor complexes in macrophages and RAW cells. To examine directly the association among MDL-1, DAP10, and DAP12, we performed immunoprecipitation/immunoblot analysis of RAW cell extracts with polyclonal antibodies specific to each molecule (Fig. 4C). We found that DAP10 and DAP12 immunoprecipitation resulted in coprecipitation of DAP12 and DAP10, respectively, in addition to MDL-1. To examine the presence of DAP12–DAP10 heterodimers associating with MDL-1, total lysates of pOCs were immunoprecipitated with anti-MDL-1 mAb, separated by SDS/PAGE under nonreducing conditions, and then immunoblotted with anti-DAP12 or DAP10 antibodies (Fig. 4D). We found the 16-kDa protein species in B6 pOCs that was reactive with both DAP12 and DAP10 antibodies (Fig. 4D, lanes 1 and 3) and was lost in the DAP10 deficiency (Fig. 4D, lane 2), suggestive of the DAP12–DAP10 heterodimer. Collectively, these results suggest that the DAP12–MDL-1 complex is a prototype structure for the surface expression of MDL-1, and that DAP10 association with MDL-1 almost solely depends on the presence of DAP12–MDL-1, forming an MDL-1–DAP12/DAP10 trimolecular complex.

MDL-1 Functions as a Positive Regulator of Osteoclastogenesis.

To determine how MDL-1 functions in osteoclastogenesis, we stimulated BMMs from B6 mice with a plate-bound anti-MDL-1 mAb and found an increase in RANKL-induced mOCs (Fig. 5A). To further confirm the activating function of MDL-1, we examined the effect of suppression of mRNA for MDL-1 by the RNA silencing technique. Transfection of MDL-1 siRNA, but not nontargeting siRNA, reduced the level of MDL-1 mRNA (Fig. S12) and led to the reduced number of mOCs (Fig. 5B). These results led to our conclusion that MDL-1 functions as a positive regulator of osteoclastogenesis.