CD44 Antibodies Inhibit Osteoclast Formation

Authors

  • Janice R. Kania,

    1. Department of Pediatrics, Division of Endocrinology and Metabolism, Washington University in St. Louis, St. Louis, Missouri, U.S.A.
    Search for more papers by this author
  • Talia Kehat-Stadler,

    1. Department of Pediatrics, Division of Endocrinology and Metabolism, Washington University in St. Louis, St. Louis, Missouri, U.S.A.
    Search for more papers by this author
  • Stuart R. Kupfer

    Corresponding author
    1. Department of Pediatrics, Division of Endocrinology and Metabolism, Washington University in St. Louis, St. Louis, Missouri, U.S.A.
    • Stuart R. Kupfer Department of Pediatrics, Box 8116 Division of Endocrinology and Metabolism Washington University in St. Louis St. Louis Children's Hospital, Room 1012 One Children's Place St. Louis, MO 63110 U.S.A.
    Search for more papers by this author

Abstract

Osteoclast differentiation is a complex process requiring multiple factors and sequential regulation. We have determined that CD44, a cell surface glycoprotein that is known to function as an adhesion receptor, is involved in this process. By immunocytochemistry, we show that CD44 is expressed in mouse osteoclasts that develop in primary cultures of bone marrow cells treated with 1α,25-dihydroxyvitamin D3. Monoclonal antibodies to CD44 inhibit osteoclast formation in bone marrow cultures in a dose- and time-dependent manner. In contrast, CD44 Fab monomer antibodies have no effect on osteoclast development, suggesting that the inhibition of differentiation by the whole antibodies is facilitated by cross-linking of CD44 molecules. Cocultures of spleen cells and ST2 bone marrow stromal cells indicate that hematopoietic cells mediate the CD44 antibody inhibitory effect. CD44 antibodies do not inhibit osteoclast resorption of calcified matrix, indicating that CD44 is not absolutely required for resorption activity. These observations demonstrate that CD44 may play a role in osteoclast formation and suggest mechanisms by which CD44 antibody effects are mediated.

INTRODUCTION

Osteoclasts are the cells responsible for resorbing bone matrix and thus play an integral role in bone development and remodeling.1,2 Most evidence indicates that osteoclasts differentiate from hematopoietic progenitor cells, specifically from the monocyte-macrophage lineage.3,4 As these osteoclast progenitors differentiate, they acquire characteristic osteoclast markers, including tartrate-resistant acid phosphatase (TRAP), matrix metalloproteinase-9, calcitonin receptors, and vitronectin receptors.5–8 In the mouse, the differentiation process is dependent on the direct interaction of bone marrow stromal cells with osteoclast progenitor cells.9,10 In the human, stromal cells are not absolutely required for osteoclastogenesis in vitro but are capable of enhancing this process.11–13 The committed osteoclast precursors eventually fuse to form multinucleated cells that are capable of resorbing and degrading mineralized bone matrix. Defects in osteoclast differentiation, resulting from deficiencies in macrophage colony-stimulating factor (M-CSF) or c-fos, cause osteopetrosis, a disease due to inadequate bone resorption.14,15

CD44 is a widely expressed family of cell surface glycoproteins involved in cell–cell and cell–matrix adhesion.16–18 CD44 is expressed in various cell types: hematopoietic cells, fibroblasts, bone marrow stromal cells, and epithelial cells.19,20 The involvement of CD44 activity has been implicated in multiple cellular processes, such as lymphocyte and monocyte activation, hematopoiesis, and tumor metastasis.21–23 Functional diversity of CD44 is generated, in part, by cell type-specific glycosylation and differential usage of 10 exons encoding a portion of the extracellular domain.24,25 The most common isoform detected in hematopoietic cells is CD44H, which does not contain any of the differentially spliced exons.19,26 Although a number of ligands have been shown to bind CD44, including hyaluronate, fibronectin, and collagen type I, the mechanism of CD44-mediated activity is unknown.27–29

We hypothesized that CD44 might be important for osteoclast differentiation and resorption activity for several reasons. First, CD44 appears to have a role in cell–cell adhesion, a process that is critical for fusion of osteoclast precursors and for interaction of differentiating osteoclast progenitor cells with bone marrow stromal cells.10,30 Second, CD44 expression has been observed in osteoclasts in human bone and in rat tibia by immunohistochemistry.31,32 Third, cell adhesion and cell migration, processes in which CD44 participates in some cell types, are functions that osteoclasts must possess to resorb calcified matrices normally.33,34 Therefore, we utilized mouse primary bone marrow cultures and CD44 monoclonal antibodies to address these hypotheses. Our results demonstrate that a subset of CD44 antibodies blocks osteoclast formation by targeting hematopoietic cells, suggesting the importance of CD44 in osteoclast differentiation.

MATERIALS AND METHODS

Primary bone marrow cultures

Mouse bone marrow cells were isolated by a modification of previously published methods.35,36 Five- to 10-week-old C57BL/6 male mice (The Jackson Laboratory, Bar Harbor, ME, U.S.A.) were sacrificed by cervical dislocation. Femora and tibiae were isolated by aseptic techniques and dissected free of adherent tissue. The ends of bones were cut with scissors, and bone marrow cells were obtained by flushing bones with α minimum essential medium (α-MEM) containing 10% heat-inactivated fetal calf serum (FCS; GIBCO-BRL, Gaithersburg, MD, U.S.A.) utilizing a 25G needle. Cells were washed twice in α-MEM with 10% FCS and plated in 0.5 ml of the same medium at a density of 4 × 106 nucleated cells per well in 24-well plates. Cells were treated continuously with 10 nM 1α,25-dihydroxyvitamin D3 (1α,25(OH)2D3; BIOMOL Research Laboratories, Plymouth Meeting, MA, U.S.A.), 1 mM prostaglandin E2 (PGE2; Sigma, St. Louis, MO, U.S.A.), or 100 ng/ml rat parathyroid hormone, PTH(1–34) (Bachem, Torrance, CA, U.S.A.) beginning on day 1 of the culture. Medium, hormones, and antibodies were replaced every 3 days. Cultures were maintained at 37°C in humidified air containing 5% CO2.

Spleen cell cocultures

Spleens were isolated from C57BL/6 male mice, and spleen cells were dispersed by passing through a 100 μm pore screen. ST2 bone marrow stromal cells, which support osteoclast formation in spleen hematopoietic cells,9 were obtained from Riken Gene Bank (Japan). In some experiments spleen and ST2 cells were pretreated for 3 days with various antibodies, as indicated, and then washed extensively with α-MEM. (The pretreatment of spleen cells for 3 days, even in the absence of antibodies, reduced the ultimate yield of osteoclasts in coculture by 50–70%.) Spleen cells (1 × 106) and ST2 cells (1 × 105) were cocultured for 8 days in α-MEM and 10% FCS in the presence of 10 nM 1α,25(OH)2D3 and 100 nM dexamethasone. Medium and hormones were replaced every 3 days.

TRAP histochemical staining

Mouse bone marrow cultures were analyzed for TRAP activity after 8 days of culture. Cells were fixed in 2.5% glutaraldehyde in phosphate-buffered saline (PBS), pH 7.2, for 15 minutes and permeabilized in 65% acetone and 10 mM sodium citrate, pH 3.6, for 1 minute. Enzyme histochemistry for TRAP was performed using an acid phosphatase kit (Sigma). The total area of each 2 cm2 well was analyzed by light microscopy for quantitation of osteoclasts (TRAP-positive multinucleated cells with ≥3 nuclei) and osteoclast progenitor cells (TRAP-positive mononuclear cells).

Resorption assay

Resorption activity of osteoclasts was assayed by previously described methods.37,38 Bone marrow cells were isolated as described above and plated at a density of 1.5 × 108 nucleated cells per 10 cm dish in α-MEM containing 10% FCS and 10 nM 1α,25(OH)2D3. After 8 days of culture, cells were scraped and washed once in medium. Dentin slices (0.2 mm thickness, 6 mm diameter) were prepared from sperm whale teeth (obtained from the United States Department of Fisheries) using a low speed diamond saw (Buehler, Lake Bluff, IL, U.S.A.). Dentin slices were preincubated in α-MEM containing 10% FCS 48 h prior to each experiment. Cells were replated on dentin slices in 24-well plates at a density of 1.0 × 106 cells per well. Cells were cultured for 5 days in medium containing 10 nM 1α,25(OH)2D3 with replacement of medium on day 3. On day 5, cells were stained for TRAP, as described above, and 10 fields per dentin slice at 20× magnification were analyzed to quantitate TRAP-positive multinucleated cells (MNCs). Cells were then removed from the dentin slices by brief treatment with 10 mM NH4OH and gentle scraping. Identical TRAP-stained fields were analyzed for pit number and pit area using dark field reflective microscopy and a Leica Quantimet (Leica UK Ltd., Milton Keynes, U.K.) image analysis program. Mean pit area, number of pits per TRAP-positive MNC, and total resorption area per TRAP-positive MNC were determined.

Antibodies

All of the CD44 antibodies used in these experiments were purified rat monoclonal antibodies raised against mouse CD44. IRAWB14.4 (IgG2a)39 and KM81 (IgG2b)21 were kindly provided by Dr. Paul Kincade (Oklahoma Medical Research Foundation, Oklahoma City, OK, U.S.A.). IM7.8.1 (IgG2b)40 was purchased from Endogen, Inc. (Boston, MA, U.S.A.), and KM114 (IgG1),21 and isotype controls were purchased from Pharmingen (San Diego, CA, U.S.A.). Fab monomer fragments of KM81 and IM7.8.1 were kindly provided by Dr. Jayne Lesley (Salk Institute, San Diego, CA, U.S.A.). Purified rat polyclonal IgG Fab monomer was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA, U.S.A.).

Immunocytochemistry

Mouse bone marrow cells were isolated as described above and plated in 24-well plates in α-MEM containing 10% FCS and 10 nM 1α,25(OH)2D3. After 8 days, cells were fixed in 1% paraformaldehyde in PBS for 20 minutes at RT and washed in PBS. Nonspecific binding sites were blocked with 5% normal goat serum (Sigma) for 30 minutes at 4°C. Cells were incubated with a pool of CD44 monoclonal antibodies (IM7.8.1, IRAWB14.4, KM81, and KM114; each at 1 μg/ml) for 1 h at 4°C. After washing in PBS, the cells were incubated with 1 μg/ml Cy3-conjugated goat anti-rat IgG (Jackson ImmunoResearch Laboratories) for 30 minutes at 4°C. After washing with PBS, cells were examined by fluorescence microscopy, and selected fields were photographed. Cells were then stained for TRAP as above, and identical fields were photographed by light microscopy.

Statistical analysis

Data are reported as the mean ± SEM of three samples. Statistical significance was analyzed by one-way analysis of variance (ANOVA) using the Dunnett's test for multiple comparisons. The data presented are representative examples of at least three experiments.

RESULTS

CD44 is expressed in osteoclasts in primary mouse bone marrow cultures

Previous reports have demonstrated that rat and human osteoclasts express the CD44 antigen.31,32 To determine if CD44 is expressed in osteoclasts that develop in the mouse whole bone marrow system, bone marrow cultures were treated for 8 days with 10 nM 1α,25(OH)2D3 and analyzed for CD44 and TRAP expression. As shown in Figs. 1A and 1C, CD44 monoclonal antibodies recognized both TRAP-positive multinucleated cells and TRAP-positive monocytes. CD44 appeared to be localized to the cell membrane, an observation consistent with reports in other cell types.41,42 Other mononuclear cells within the bone marrow culture also stained positively for CD44 protein. Immunocytochemistry using isotype control antibodies revealed no artifactual background staining of osteoclasts (Figs. 1B and 1D). These results indicate that CD44 is expressed in multiple 1α,25(OH)2D3-treated mouse bone marrow cells, including TRAP-positive mononuclear and multinucleated cells, and suggest that this system would be suitable for studying the role of CD44 in osteoclast differentiation and resorption activity.

Figure FIG. 1.

Expression of CD44 and TRAP in 1α,25(OH)2D3-treated mouse bone marrow cells. Mouse bone marrow cells were cultured for 8 days in the presence of 10 nM 1α,25(OH)2D3. (A) Cells were fixed, washed, and incubated with a pool of rat monoclonal anti-mouse CD44 antibodies (IRAWB14.4, IM7.8.1, KM81, KM114). (B) Cells were incubated with a pool of rat anti-mouse isotype control antibodies (IgG1, IgG2a, and IgG2b). (C; D) Identical fields to (A) and (B), respectively, were stained for TRAP and viewed by light microscopy (20×). The large cell in the center of each photograph is a multinucleated osteoclast. The arrow indicates a CD44− and TRAP-positive mononuclear cell.

Specific CD44 antibodies inhibit the development of osteoclasts

To investigate a possible functional role for CD44 in osteoclast formation, primary mouse bone marrow cultures were treated with 10 nM 1α,25(OH)2D3 for 8 days in the presence of various purified rat monoclonal anti-mouse CD44 antibodies. Treatment with IM7.8.1 inhibited the formation of TRAP-MNCs by 85% at the maximum concentration of 10 ng/ml (Fig. 2A), and IRAWB14.4 completely inhibited TRAP-MNC formation (Fig. 2C). The KM114 antibody did not inhibit TRAP-MNC formation (Fig. 2D). The KM81 antibody caused a slight inhibition of TRAP-MNC formation (Fig. 2B), but this effect was apparently nonspecific since the IgG2b isotype control inhibited to the same degree (Figs. 2A and 2B). Treatment with IgG2a and IgG1 isotype control antibodies had no significant effect on the development of TRAP-MNCs (Figs. 2C and 2D). Interestingly, the KM81 and KM114 antibodies recognize the same CD44 epitopes and, unlike IM7.8.1 or IRAWB14.4, universally block hyaluronate binding to CD44.43,44 The range of responses that we observed with this group of CD44 antibodies could possibly be explained by differences in epitope recognition or ligand binding. This pattern of antibody responses is different from studies of B lymphopoiesis in long-term bone marrow cultures, which was inhibited by IM7.8.1, as well as KM81 and KM114 antibodies.21 These results indicate that a subset of CD44 antibodies blocks the development of osteoclasts in 1α,25(OH)2D3-treated whole bone marrow cultures.

Figure FIG. 2.

Specific CD44 whole antibodies inhibit osteoclast formation. Mouse bone marrow cells were cultured for 8 days in the presence of 10 nM 1α,25(OH)2D3 and treated with rat monoclonal anti-mouse antibodies. CD44 antibodies include: (A) IM7.8.1, (B) KM81, (C) IRAWB14.4, (D) KM114. Isotype control antibodies (CON) include: (A) and (B) IgG2b, (C) IgG2a, (D) IgG1. Osteoclasts were quantitated by counting the number of TRAP-positive multinucleated cells with ≥3 nuclei in each 2 cm2 well. An asterisk indicates p < 0.05 compared with untreated control.

We observed no generalized growth inhibition of bone marrow stromal cells in any of the CD44 antibody treatment groups. At the highest concentration of IRAWB14.4, the CD44 antibody that most potently inhibited 1α,25(OH)2D3-induced osteoclast formation, the morphological appearance of the bone marrow culture was identical to cultures that were not exposed to 1α,25(OH)2D3. The CD44 antibodies produced the same pattern of inhibition in PTH- and PGE2-induced cultures, indicating that the antibody effect is not specific to 1α,25(OH)2D3 treatment (data not shown).

Antibody-mediated oligomerization or cross-linking of CD44 molecules on the cell surface has been demonstrated to enhance hyaluronate binding to CD44.45,46 Therefore, we hypothesized that the CD44 antibody effect on osteoclast formation could be due to cross-linking of CD44 molecules on the surface of target cells. To address this question, osteoclast formation in mouse bone marrow cultures was evaluated in the presence of Fab monomer fragments of IM7.8.1 and KM81. Since the IRAWB14.4 Fab has a 20− to 50-fold lower binding affinity than the intact antibody,45 it was excluded from this analysis. The IM7.8.1 Fab had no significant inhibitory effect on osteoclast formation compared with the whole IM7.8.1 antibody (Fig. 3). Although the IM7.8.1 Fab monomer has 2− to 3-fold lower avidity for CD44 than the whole antibody (J. Lesley, personal communication), this difference is not sufficient to account for the lack of inhibition observed with the Fab monomer. The degree of osteoclast formation in the presence of IM7.8.1 Fab was comparable to the control Fab fragments at the highest concentration tested. As expected, the KM81 Fab monomer did not inhibit osteoclast formation, consistent with the lack of inhibitory effect of the whole KM81 antibody (Figs. 2B and 3). These results suggest that the effect of whole IM7.8.1 antibody on osteoclast formation is mediated, in part, by cross-linking of CD44 molecules. The possibility that the whole IM7.8.1 antibody is acting by competitive inhibition of a ligand(s) is less likely since the Fab monomer, which presumably binds the same epitope as the whole antibody, did not inhibit osteoclast formation.

Figure FIG. 3.

CD44 Fab monomers do not inhibit osteoclast formation. Mouse bone marrow cells were cultured for 8 days in the presence of 10 nM 1α,25(OH)2D3 and treated with rat monoclonal anti-mouse CD44 Fab monomers of IM7.8.1 and KM81. As controls, parallel cultures were treated with purified rat Fab monomers and IM7.8.1 whole antibody. TRAP-MNC indicates TRAP-positive multinucleated cells per 2 cm2 well. None of the treatment group means was significantly different (at p < 0.05) from the untreated control group.

CD44 antibody blocks osteoclast formation in early differentiation

In primary bone marrow cultures, osteoclastogenesis progresses through several stages, including proliferation of osteoclast progenitors, differentiation of osteoclast-like monocytic cells, and fusion of monocytic osteoclast precursors to form multinucleated osteoclasts.47 To determine the stage at which CD44 antibodies are disrupting osteoclast formation, bone marrow cultures were treated at different time points with IRAWB14.4. In our bone marrow culture system, the maximum number of multinucleated osteoclasts that develop in response to 1α,25(OH)2D3 occurs at 8 days. As shown in Fig. 4, antibody treatment on days 1–8 of the 8-day culture period resulted in complete inhibition of TRAP-MN cell formation. Treatment on days 1–3 or days 4–6 resulted in 94% and 84% inhibition, respectively, whereas treatment on days 7–8 caused no significant inhibition. To further assess the process of osteoclast formation in CD44 antibody-treated cultures, we quantitated the number of TRAP-positive mononuclear cells. A dose-response study with IRAWB14.4 revealed decreasing TRAP-positive mononuclear cells with increasing antibody concentration (Fig. 5). These results indicate that CD44 antibodies are blocking osteoclast development by preventing the proliferation and/or early differentiation of osteoclast progenitor cells. An additional possibility is that CD44 antibodies are inducing apoptosis of osteoclast precursors.

Figure FIG. 4.

CD44 antibody blocks osteoclast formation in early differentiation. Mouse bone marrow cells were cultured in the presence or absence of 10 nM 1α,25(OH)2D3 and treated with the CD44 antibody IRAWB14.4 (1 μg/ml) at different time points. TRAP-MNC indicates TRAP-positive multinucleated cells per 2 cm2 well. An asterisk indicates p < 0.05 compared with 1α,25(OH)2D3-treated control.

Figure FIG. 5.

CD44 antibody inhibits the formation of mononuclear osteoclast precursors. Mouse bone marrow cells were cultured for 8 days in the presence of 10 nM 1α,25(OH)2D3 and treated with CD44 antibody (IRAWB14.4) or IgG2a isotype control antibody (CON). Mononuclear preosteoclasts were quantitated by counting the number of TRAP-positive mononuclear cells in each 2 cm2 well. An asterisk indicates p < 0.05 compared with untreated control.

Hematopoietic cells mediate the CD44 antibody inhibitory effect

To determine the cell population mediating the CD44 inhibition of osteoclast formation, we utilized the spleen cell coculture system. When ST2 bone marrow stromal cells were cocultured with spleen cells in the presence of 1α,25(OH)2D3 and dexamethasone, the CD44 antibody IRAWB14.4 inhibited osteoclast differentiation (561 ± 22 TRAP-MNC for control vs. 142 ± 19 TRAP-MNC for IRAWB14.4 antibody at 1 μg/ml). Since the effect of the IRAWB14.4 antibody in the spleen coculture system recapitulated the effect in the whole bone marrow system, we used the spleen cell coculture system to determine which cell population, hematopoietic cells (spleen cells) or osteoclast-supporting stromal cells (ST2 cells), is mediating the CD44 antibody effect. For 3 days prior to coculture, spleen cells and ST2 cells were cultured separately and treated with the CD44 antibody IRAWB14.4, isotype control antibody, or no antibody. After this pretreatment phase, ST2 and spleen cell treatment groups were cocultured in various combinations and treated with 1α,25(OH)2D3 and dexamethasone for 8 days. Pretreatment of spleen cells with IRAWB14.4 caused 50–70% inhibition of TRAP-MNC formation compared with pretreatment with no antibody or the isotype control (Fig. 6). In contrast, pretreatment of ST2 stromal cells with IRAWB14.4 or isotype control did not inhibit TRAP-MNC formation. Furthermore, no additive effect was observed when both spleen cells and ST2 cells were pretreated with IRAWB14.4. These results indicate that the cell type mediating the CD44 antibody inhibition of osteoclast differentiation is in the hematopoietic cell population. Although this observation tends to implicate the osteoclast progenitor cell as the target cell, these experiments do not definitively prove this due to the heterogeneity of the spleen cell population.

Figure FIG. 6.

Hematopoietic cells mediate the CD44 antibody inhibitory effect. Spleen cells and ST2 cells were pretreated for 3 days with CD44 antibody (IRAWB14.4), isotype control (CON), or no antibody (none), as indicated. Various combinations of spleen and ST2 cells were then cocultured for 8 days in the presence of 10 nM 1α,25(OH)2D3 and 100 nM dexamethasone. TRAP-MNC indicates TRAP-positive multinucleated cells per 2 cm2 well. An asterisk indicates p < 0.05 compared with controls without antibody.

CD44 antibodies do not alter osteoclast resorption activity

CD44 is important for cell motility and adhesion of certain cell types.16,17 Since these characteristics are essential for maximal resorption of calcified matrices by osteoclasts,33,34 we analyzed the possible involvement of CD44 in osteoclast resorption activity. Osteoclast-containing bone marrow cells were cultured on dentin slices. Cultures were treated with 10 nM 1α,25(OH)2D3 and the CD44 antibody IRAWB14.4. CD44 antibody treatment did not inhibit the number of pits per osteoclast or total resorption area per osteoclast compared with untreated controls (Figs. 7A and 7C). CD44 antibody caused a slight increase in the mean pit area (Fig. 7B). An isotype control antibody had no significant effect on resorption activity compared with untreated controls (Figs. 7A, 7B, 7C). In agreement with the IRAWB14.4 results, the CD44 antibody IM7.8.1 also did not inhibit osteoclast resorption activity (data not shown). These results suggest that CD44 is not an essential factor in osteoclast resorption activity.

Figure FIG. 7.

Osteoclast resorption activity is not affected by CD44 antibody. Osteoclasts were cultured on dentin slices in the presence of 10 nM 1α,25(OH)2D3 and treated with CD44 antibody (IRAWB14.4), or isotype control (CON) at 1 μg/ml. Osteoclasts were quantitated by counting the number of TRAP-positive multinucleated cells (TRAP-MNC). Resorption pit area was analyzed using dark field reflective microscopy and an image analysis program. (A) Number of pits per TRAP-MNC. (B) Mean pit area. (C) Total resorption area per TRAP-MNC. An asterisk indicates p < 0.05 compared with untreated control.

DISCUSSION

We have determined that CD44, a cell surface glycoprotein that is known to function as an adhesion receptor, is involved in osteoclast development. CD44 is expressed in mouse osteoclasts that develop in primary cultures of bone marrow cells treated with 1α,25(OH)2D3. Whole CD44 monoclonal antibodies inhibit osteoclast formation in bone marrow cultures in a dose- and time-dependent manner. The greatest inhibition occurs early in the culture, indicating that CD44 antibodies block proliferation and/or early differentiation of osteoclast progenitor cells. The inhibition of the formation of TRAP-positive mononuclear cells by CD44 antibody supports this observation. CD44 Fab monomer antibodies have no effect on osteoclast development, suggesting that the inhibition of differentiation by the whole CD44 antibodies is facilitated by cross-linking of CD44 molecules. Cocultures of spleen and ST2 cells indicate that hematopoietic cells are the target cells for the CD44 antibody inhibitory effect. CD44 antibodies have no effect on osteoclast resorption of calcified matrix indicating that CD44 is not absolutely essential for resorption activity. Our observations demonstrate that CD44 may play a role in osteoclast formation and suggest mechanisms by which CD44 antibody effects are mediated. These results are in agreement with Udagawa et al. who also observed that CD44 antibodies inhibit osteoclast development.48 In contrast, we did not observe CD44 antibody-mediated inhibition of osteoclast resorption activity reported by these authors.

The involvement of CD44 in the differentiation of osteoclasts, a cell derived from hematopoietic progenitors, is consistent with the association of CD44 in the differentiation of other hematopoietic cell types.40,49 CD44 is expressed in most human and murine hematopoietic cells, and levels of CD44 expression vary with the stage of differentiation. For example, during murine T cell differentiation, CD44 is expressed in a triphasic manner. High levels of CD44 are present in the earliest T cell progenitors, expression declines during early differentiation, and CD44 is expressed again in more mature T cells.50,51 Additional evidence linking CD44 with hematopoiesis is the observation that CD44-specific monoclonal antibodies can inhibit differentiation of some hematopoietic cell types. For example, CD44 antibodies block the hyaluronate-induced differentiation of human CD34+ progenitor cells into eosinophils.52 Likewise, in long-term bone marrow cultures, CD44 antibodies completely inhibit B cell lymphopoiesis.21 The mechanism of inhibition of lymphopoiesis is uncertain; however, it was shown that CD44 on the lymphoid cells was recognizing hyaluronate on the surface of stromal cells.53 Since murine osteoclast differentiation from hematopoietic precursors is stromal cell–dependent,9,10 CD44 may influence osteoclast differentiation by similar stromal cell and progenitor cell interactions. This hypothesis is supported by our observations that CD44 is expressed in osteoclasts and that CD44 on hematopoietic cells is the target of the CD44 antibody inhibitory effect on osteoclast differentiation.

CD44 antibodies have been widely utilized to study CD44 function. For example, the CD44 antibody IM7.8.1 inhibits infiltration of murine T cell lymphoma LB cells into lymph nodes.54 In other studies of human T cells, CD44 antibodies can activate specific functions, such as LFA-1–mediated cell aggregation.55,56 Whether these antibodies are activating or inhibiting a CD44 function is not clear and may vary with the cell type, the specific antibody utilized, and the function being analyzed. In our experiments, a subset of CD44 antibodies inhibited osteoclast development. Our data do not indicate whether these antibodies activate or inhibit a specific CD44 function. We also observed that, whereas the whole IM7.8.1 CD44 antibody inhibited osteoclast formation, the Fab monomer did not. These results imply that the inhibitory effect may be mediated by cross-linking of CD44 molecules rather than by blocking ligand binding. This finding is also consistent with other studies, demonstrating that hyaluronate binding affinity is enhanced when multimers of CD44 are formed.44–46 Antibody-mediated cross-linking could possibly induce the formation of a multivalent CD44 structure or facilitate cell–cell adhesion. Nonspecific cross-linking of CD44 alone, however, is clearly not sufficient to mediate the inhibitory effect on osteoclast formation since two of the four whole CD44 antibodies we investigated had no inhibitory effect on osteoclast formation.

A number of CD44 monoclonal antibodies have been characterized on the basis of their epitope recognition and ability to influence hyaluronate binding. Interestingly, the KM81 and KM114 antibodies, which had little or no effect on osteoclast formation in the present study, universally block hyaluronate binding and recognize the same or overlapping epitopes in the CD44 extracellular domain.21,43 However, IM7.8.1 and IRAWB14.4 antibodies, which inhibited osteoclast differentiation, recognize extracellular epitopes distinct from each other and from the KM81/KM114 epitope.43 IM7.8.1 blockade of hyaluronate binding is variable and dependent on the specific cell type being analyzed.39,43 IRAW14.4, which mediated the greatest inhibitory effect on osteoclast formation, has been demonstrated to markedly enhance hyaluronate binding in some circumstances.39,44 The variable osteoclastogenic responses and the differences in CD44 antibody characteristics suggest that osteoclast formation may depend upon a specific CD44 ligand or a discrete CD44 structure. An additional hypothesis is that CD44 antibodies may be inhibiting osteoclast formation by steric interference with other important cell surface molecules.

Other adhesion receptors, in particular the integrins, play an important role in osteoclast function and formation. The osteoclast integrins, αvβ3, α2β1, and αvβ1, mediate binding to the bone surface via bone matrix proteins and have been shown to be important for mediating osteoclast bone resorption.57 In addition, synthetic RGD peptides that block integrin binding inhibit formation of osteoclasts by blocking migration of mononuclear precursors.58 Function-blocking antibodies to E-cadherin inhibit osteoclast formation by disrupting fusion of osteoclast precursors in vitro.59 In the present study, we have identified CD44 as another adhesion receptor that is involved in osteoclast formation. Interestingly, a subset of the known ligands for CD44 include components of the bone matrix, such as hyaluronate, fibronectin, collagen type I, and osteopontin.27,29,60 Interaction of these bone matrix proteins with CD44 could be important for mediating osteoclast differentiation and function. Additional studies are required to determine which, if any, of these ligands mediate the CD44 effects on osteoclast formation.

A number of cytokines and growth factors have been shown to influence osteoclast differentiation. For example, M-CSF, interleukin-1 (IL-1), and tumor necrosis factor-α (TNF-α) are inducers of osteoclast formation, whereas interferon-γ (IFN-γ) and interleukin-4 (IL-4) are inhibitors of this process.1,2 Interestingly, activation of the CD44 molecule may induce production of cytokines, such as IFN-γ and IL-4 in memory T lymphocytes.61 In addition, a CD44-specific monoclonal antibody blocks the hyaluronate-induced expression of IL-1 and TNF-α in murine bone marrow–derived macrophages.62 Thus, specific networks of CD44 signals could potentially mediate production of cytokines that block osteoclast formation, suggesting another possible mechanism for the antibody effects we observed in this study.

The inhibition of osteoclast formation by CD44 antibodies suggests that CD44 could be a potential pharmacological target for the treatment of metabolic bone disease, such as osteoporosis or Paget's disease. Likewise, the CD44 variant isoforms expressed in breast carcinoma63,64 and multiple myeloma65 could play a role in the induction of osteoclast formation and subsequent osteolysis and could also be therapeutic targets. If CD44′s effects on osteoclast formation in these situations were mediated by particularly rare CD44 isoforms, perhaps relatively specific therapeutic agents could be designed without modulating other CD44 functions. Our results warrant further investigation into the role of CD44 in both normal and pathologically induced osteoclast formation.

Acknowledgements

We thank Jayne Lesley and Paul Kincade for providing CD44 antibodies. We also thank Shannon Holliday, Jeanne Erdmann, Jean Chappel, and Patrick Ross for their technical advice. We are especially grateful to Philip Osdoby and Jeffrey Milbrandt for their advice and support. This work was supported by a grant (S.R.K.) from the Edward Mallinckrodt, Jr. Foundation.

Ancillary