Rho-dependent Rho Kinase Activation Increases CD44 Surface Expression and Bone Resorption in Osteoclasts*

Osteoclasts from osteopontin-efficient mice exhibit decreased CD44 surface expression. Osteopontin (OPN)/ (cid:1) v (cid:2) 3 generated Rho signaling pathway is required for the surface expression of CD44. In this work we show the Rho effector, Rho kinase (ROK- (cid:1) ), to be a potent activator of CD44 surface expression. ROK- (cid:1) activation was associated with autophosphorylation, leading to its translocation to the plasma membrane, as well as its association with CD44. ROK- (cid:1) promoted CD44 surface expression through phosphorylation of CD44 and ezrin-radixin-moesin (ERM) proteins and CD44 (cid:1) ERM (cid:1) actin complex formation. Osteoclasts from OPN (cid:3) / (cid:3) mice exhibited an (cid:1) 55–60% decrease in basal level ROK- (cid:1) phosphorylation as compared with wild type osteoclasts. Furthermore, Rho Val-14 transduction was only partially effective in stimulating ROK- (cid:1) /CD44 phosphorylation, as well as CD44 surface expression, in phosphate-buffered saline; TAT, transactivator with transforming properties; HSV-TK, herpes simplex virus-thymi- dine kinase; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-pro-panesulfonic acid; FACS, fluorescence-activated cell sorter; NI, non- immune serum; PI 4P-5kinase, phosphatidylinositol 4P-5kinase; immunoprecipitation; ab,

osteoclasts with exogenous OPN partially rescued the phenotype by stimulating motility and increasing the expression of CD44 on the basolateral plasma membrane surface. Both antibodies to CD44 and ␤ 3 blocked the stimulatory actions of OPN on motility and bone resorption. These results demonstrated that the surface expression of CD44 was a key function of OPN and hence results in the hypomotility of OPN-deficient osteoclasts (15).
We have shown that Rho activation is a critical step in the stimulation of motility by exogenous OPN (16). OPN/␣ v ␤ 3 generated outside-in the Rho signaling pathway is required for the surface expression of CD44 and motility of osteoclasts (15). One downstream effect of Rho signaling pathway is targeting of the CD44⅐ERM complex to the osteoclast plasma membrane. OPNdeficient osteoclasts fail to efficiently move CD44 to the surface of the plasma membrane, and exogenous addition of OPN or Rho is sufficient to partially rescue the phenotype. Rescue of the phenotype by exogenous OPN is blocked by antibodies to the ␤ 3 integrin (15).
Here, we analyze the mechanism by which OPN-stimulated Rho activation mediates movement of CD44 to the plasma membrane. The Rho downstream effector, Rho kinase, has been implicated in cell migration (18,19). Both the Rho A GTPase and Rho kinase (ROK-␣) are physically associated with CD44 and its associated cytoskeletal proteins (19). The CD44 cytoplasmic tail interacts with cytoskeletal-related components including actin, ankyrin, and the ezrin-radixin-moesin (ERM) family (20). ROK plays a pivotal role in CD44 phosphorylation and its interaction with ankyrin and the ERM complex (21). ERM⅐actin complexes appear to be essential prerequisites for Rho-induced cytoskeletal changes and cell migration (22). The ERM family and Rho are localized at specialized plasma membrane areas, including membrane ruffling area and cell-cell contact sites in Madin-Darby canine kidney epithelial cells (23). The translocation of ERM proteins to the plasma membrane is thought to be regulated by Rho signaling (20,23,24). ERM proteins cross-link actin filaments with plasma membrane. The carboxyl terminus of these proteins is associated with actin, and the amino terminus binds plasma membrane using a binding partner, CD44 (25,26).
Rho A (complexed with CD44 V3, 8 -10 ) regulates the phosphorylation of CD44, as well as the proteins associated with CD44 through ROK activation (19). Because the transduction of constitutively active Rho Val-14 into osteoclasts increased the cell surface expression of CD44, as well as osteoclast motility (15), and Rho effector, ROK, has a predominant role in the phosphorylation of CD44/ERM proteins during actin filament and plasma membrane interaction (28 -30), we hypothesized that phosphorylations of CD44 and ERM proteins require the activation of Rho effector(s).
We report here that OPN stimulated Rho-dependent activation and phosphorylation of Rho kinase (ROK-␣ or ROCK II), its association with CD44, and CD44 surface expression in osteoclasts generated from WT mice. In osteoclasts generated from OPNϪ/Ϫ mice, CD44 surface expression was decreased, even though there were no changes in actin levels, podosome organization, or morphology. This was because of diminished levels of ROK-␣ associated with CD44. Rho Val-14 transduction and OPN treatment increased ROK-␣ phosphorylation and CD44 surface expression in OPNϪ/Ϫ osteoclasts but not to levels of full stimulation as in wild type osteoclasts. Furthermore, ROK-␣ activation, dependent on OPN stimulation of ␣ v ␤ 3 signaling, was required for CD44 surface expression.
[␥-32 P]ATP and rainbow molecular weight markers for proteins were purchased from Amersham Biosciences.
Osteopontin-deficient Mice-The OPN-deficient mouse colony originally established at Rutgers University by homologous recombination in ES cells (32) was re-derived at Washington University by caesarian section. The analyses described were performed using wild type and null animals on a 129 ϫ C57BL/6 hybrid background.
Purification of Osteopontin Protein-Mouse OPN cDNA was cloned into the BamHI/XbaI site of pQE 30 vector (Qiagen, Inc. Valencia, CA). OPN was expressed as a His 6 -tagged protein. OPN was purified from the bacterial lysate using nickel-nitrilotriacetic acid affinity chromatography following the manufacturer's instructions (Qiagen, Inc. Valencia, CA).
Preparation of Mouse Osteoclast Precursors-Mouse osteoclasts were generated in vitro using mouse bone marrow cells as described previously (33). Mouse osteoclasts were generated in vitro using the mouse bone marrow cells. Cells isolated from five mice were cultured into 100-mm dishes with 20 ml of ␣-MEM medium supplemented with 10% fetal bovine serum (␣-10). After culture for 24 h, non-adhered cells were layered on histopaque-1077 (Sigma) and centrifuged at 300 ϫ g for 15 min at room temperature. The cell layer between the histopaque and the media was removed and washed with ␣Ϫ10 media at 2000 rpm for 7 min at room temperature. Cells were resuspended in ␣-10 media and cultured with the appropriate concentrations of macrophage colonystimulating factor-1 (10 ng/ml) and OPGL (55-75 ng/ml). After 3 days in culture, media were replaced with fresh cytokines. The multinucleated osteoclasts were seen from day 4. About 90 -95% TRAP-positive osteoclasts were observed from day 5 onwards. To remove the osteoclasts for in vitro bone resorption or motility studies, cells were washed with PBS and kept in a cell stripper solution (Cellgro by Media Tech, Inc., Hemdon, VA) for 15-30 min. Cell stripper is a non-enzymatic cell dissociation solution designed to gently dislodge adherent cells in tissue culture. After incubation with the cell stripper solution, osteoclasts were removed from the plates by gentle scraping. Some of the removed cells were re-plated and stained either with trypan blue or for TRAP. Cells excluded trypan blue, and they were 99% TRAP-positive. These TRAPpositive cells were used for migration and bone resorption assays as described below.
Protein Transduction into Mouse Osteoclasts-HA-TAT expression vector containing cDNAs for RhoGTPases (Rho, Rac, and Cdc42 constitutively active or dominant negative form) were cloned into a bacterial expression vector to produce TAT fusion proteins. Herpes simplex virusthymidine kinase (HSV-TK) protein (42 kDa) was used as a nonspecific protein control. The purification of the protein was done as described previously (16,34). After osteoclasts were kept in serum-free ␣-MEM for 2 h, TAT proteins were added to cells to a final concentration of 100 nM in serum-free media. Dose-and time-dependent uptake of proteins were determined. Maximum uptake was observed between 20 and 45 min. We chose a 30-min time period for the incubation with TAT proteins. GST-C3 exoenzyme (16) and ROK inhibitor Y-27632 (35,36) were used at concentrations of 100 -150 ng/ml and 10 M (21), respectively prior to OPN treatment or Rho Val-14 transduction. Rho transduction was done as described (16).
Lysate Preparation and Western Analysis-Osteoclasts were kept in serum-free medium for 2 h and treated with OPN to a final concentration of 25 g per ml for 15 min at 37°C as described previously (37). Some cultures were treated with one of the following antibodies: anti-␣ v , anti-␤ 3 (25-30 g/ml for antibodies from Santa Cruz Biotechnology, Inc. or 50 -65 g/ml for antibodies from Chemicon), anti-CD44 antibody (20 -30 g/ml for anti-CD44 antibody developed by Dr. Jayne Lesley (31) or 50 g/ml monoclonal CD44 antibody (AR4401) purchased from BioSource International Inc. (Camarillo, CA)); 50 g/ml OPN antibody developed in the laboratory from the His 6 -tagged protein was used. Following treatments with OPN or TAT proteins, as described above, osteoclasts were washed three times with ice-cold PBS, and cells were lysed with radioimmune precipitation assay lysis buffer (10 mM Tris-HCl, pH 7.2, 150 mM NaCl, 1% deoxycholate, 1% Triton X-100, 0.1% SDS, 1% aprotinin, and 2 mM phenylmethylsulfonyl fluoride) (23). Protein contents were measured using the Bio-Rad protein assay reagent kit. Equal amounts of lysate proteins were used for immunoprecipitation. Western analysis and kinase assays were performed as described (37).
In Vitro ROK Assay-Equal amounts of lysate proteins were subjected to immunoprecipitation with either CD44 or ROK antibodies. The immunoprecipitates were washed three times with the lysis buffer (23) and two times with the kinase assay buffer (50 mM Tris-HCL, pH 7.5, 1 mM EDTA, 5 mM MgCl 2 , 0.06% CHAPS). The kinase assay was performed in 50 l of kinase assay buffer containing 5 M ATP, 5-10 ng of histone H1, and [␥-32 P]ATP (5 Ci/assay). Assay was carried out with or without exogenous substrate, histone H1. After incubation for 30 min at 30°C, the reaction mixture was boiled in SDS containing sample buffer and resolved by SDS-PAGE (10%). The radiolabeled bands were visualized by autoradiography (36,38).
Bone Resorption Assay-Whale dentine slices (1.5 cm 2 in size and 0.75 mm in thickness) were cut from previously prepared rectangular sticks on a low speed saw and were stored in 70% ethanol until required. Prior to being used in a resorption assay, the slices were sonicated for 3-5 min in distilled water and washed for 2 h in two changes of fresh distilled water. The slices were re-sterilized in fresh 70% ethanol and placed into wells of a 48-well tissue culture plate. Following three washes with ␣-MEM serum-free medium, the slices were placed overnight in a 37°C incubator in the same medium. Osteoclasts were gently scraped as described above, and the osteoclast suspension (2 ϫ 10 4 cells) was added to each well. After 2 h of adherence, the culture medium was replaced with ␣-MEM containing either mouse OPN (10 g/ml) or Tat proteins as described above. Some wells were treated with GST-C3 exoenzyme (16) or ROK inhibitor Y-27632 (35,36) at concentrations of 100 -150 ng/ml and 10 M (21), respectively, in addition to OPN treatment or Rho Val-14 transduction. Rho transduction was done as described (16). After 24 h, the medium was replaced with fresh medium containing the respective proteins, antibodies, or inhibitors at concentrations as described above. Following incubation for 48 h, cells were scrapped from the slices, and the slices were washed seven to ten times with water. Pits were stained with acid hematoxylin (Sigma) for 6 min, washed well with water, and counted. The pits were imaged under a ϫ40 objective in a Zeiss inverted phase contrast microscope fitted with a CCD camera. The images were processed by the Adobe Photoshop software program (Adobe Systems, Inc., Mountain View, CA) (16).
Immunostaining-After 5 days in culture, osteoclasts were kept in serum-free medium for 2 h. Then the culture medium was replaced with ␣-MEM (serum-free) medium containing either mouse OPN (10 g/ml) or Tat proteins as described above. Osteoclasts were washed with PBS and incubated with a monoclonal antibody to CD44 (1:100 dilution) for 2 h. Cells were then rinsed two to three times with cold PBS, fixed with paraformaldehyde, and treated with ethanol for permeabilization of osteoclasts. After rinsing twice with cold PBS, cells were incubated with an anti-goat ROK-␣ antibody (1:100) for 2 h. Cells were then rinsed a few times with PBS and counter-stained with CY3-conjugated antimouse or CY2-conjugated anti-goat secondary antibodies to anti-mouse CD44 or anti-goat ROK-␣ antibodies for 2 h. The cells were then rinsed three to four times with PBS and mounted on a slide in a mounting solution (Vector laboratories, Inc., Burlingham, CA) (16). The cells were viewed on a Bio-Rad (radiance 2100) confocal laser-scanning microscope (Bio-Rad) using a ϫ60 oil immersion lens. Confocal images were processed by the Adobe Photoshop software program (Adobe Systems, Inc., Mountain View, CA).
Fluorescence-activated Cell Sorting (FACS) Analysis-After 4 days in culture, osteoclasts were gently scraped as described above, and cells were resuspended in ␣-10 medium containing appropriate concentrations of macrophage colony-stimulating factor-1 (10 ng/ml) and OPGL (55-75 ng/ml). The osteoclast suspension (2 ϫ 10 5 to 1 ϫ 10 6 cells) was added to 35-mm tissue culture dishes and incubated at 37°C overnight. After 24 h, the cells were kept in serum-free medium for 2 h. Subsequently, the culture medium was replaced with serum-free ␣-MEM medium containing either mouse OPN (10 g/ml) or Tat proteins for 30 min as described above (16). Cells were washed with PBS, and the potential sites for nonspecific antibody binding were blocked by a 60-min incubation with a blocking solution containing PBS and 8% bovine serum albumin. After blocking, cells were incubated with a monoclonal antibody to CD44 (1:100 dilution) in the blocking solution for 2 h on ice. Nonspecific IgG (mouse) was included and used to set FACS gates for analysis. Cells were washed with cold serum-free medium and incubated with CY2-conjugated anti-mouse secondary antibody (1:100 dilution) for 2 h on ice. Additional washes (at least twice) and final suspension were done with PBS. To measure the total intracellular levels of CD44 expression, osteoclasts were treated with 2% paraformaldehyde for 15 min and permeabilized with 0.1% Triton X-100 for 5 min. Cells were washed twice with cold serum-free medium and incubated with antibodies as described above. The fluorescence intensity was measured using a BD Biosciences FACS scan flow cytometer (39). The experiment was repeated with three different osteoclast preparations. In a typical experiment, about four to six 35-mm dishes were used for each treatment.
Data Analysis-All comparisons were made as % control, which refers to vehicle-treated cells. The other treatment groups in each experiment were normalized to their respective control value. Data presented are mean Ϯ S.E. of experiments done at different times normalized to intraexperimental control values. For statistical comparisons, analysis of variance was used with the Bonferroni corrections (Instat for IBM, version 2.0; GraphPad software).

Analysis of Rho Kinase (ROK-␣) Association with CD44
Rho kinase has been identified as one of the downstream targets for CD44-bound Rho A GTPase (19). Because OPN stimulates podosome assembly, and osteoclast motility is dependent on Rho activation (16,40), we analyzed the effects of OPN on ROK association with CD44. Two closely related isoforms of Rho-associated kinase, ROK-␣ (ROCK II) and ROK-␤ (ROCK I), have been identified as Rho effectors containing serinethreonine kinase domains with 90% identity (38,41,42). To identify the isoform of ROK that is associated with CD44, specific antibodies against the carboxyl terminus of ROK-␣ (C20) (Fig. 1A) and ROK-␤ (C19) (Fig. 1B) (both from Santa Cruz) were used for immunoprecipitation and Western analyses. To have an internal control, an antibody to actin was also added to each immunoprecipitate ( Fig. 1, A, B, and D) including the non-immune serum immunoprecipitate (blot A, lane 4). Immunoprecipitation with actin alone (blots C and D, lane 1) or actin antibody added with non-immune IgG (blot A, lane 4) was used as identification control. Blots shown in Fig. 1, A, B, and D were subjected to Western analysis with two antibodies as indicated in the figures. In addition to the antibody of interest, as shown under each blot, an antibody to actin was also added. Immunoprecipitations of osteoclast lysates made from WT mice with anti-CD44 (Fig. 1A, lane 1) or Rho (lane 2) antibodies demonstrated coprecipitation of ROK-␣. Coprecipitation of ROK-␤ was observed with immunoprecipitates made with an antibody to Rho (Fig. 1B, lane 3) and not with CD44 antibody (Fig. 1B, lane 2). These observations demonstrate that CD44 is associated with ROK-␣ and not with ROK-␤. To determine what fraction of the total cellular CD44 associates with ROK-␣, immunoprecipitates made with ROK-␣ antibody were Western analyzed with anti-CD44 antibody (Fig. 1C). An increase in the levels of CD44 in OPN-treated osteoclasts indicates that ROK-␣ association with CD44 is increased in response to OPN  Fig. 1D. Lane 4 represents the fraction of ROK-␣ associated with CD44. The percentage of ROK-␣ associated with CD44 was quantitated by densitometric scans of the three blots. Immunoprecipitation with ROK-␣ was used as 100% control. Quantitative analysis of the CD44 immunoprecipitates made from PBS-and OPN-treated wild type osteoclasts re-vealed that about 9.2 Ϯ 2.6 and 38.6 Ϯ 13.8% (p Ͻ 0.001; OPN versus PBS; n ϭ 3) of total cellular ROK-␣ was associated with CD44, respectively. Even though there are changes in the levels of proteins of interest in blots A-D, actin protein, which was used as internal control, did not demonstrate significant changes in the levels indicating that equal amount of lysate proteins were used for immunoprecipitation.

ROK-␣ Phosphorylation and Activity Is Dependent on Rho Activity
In Vitro Rho Kinase Assay-To determine whether ROK-␣ phosphorylation is mediated through Rho activity in osteoclasts, we transduced constitutively active Rho Val-14 and dominant negative Rho Asn-19 into osteoclasts as described previously (16). We compared the effect of Rho Val-14 transduction on ROK-␣ autophosphorylation with that of OPN treatment ( Fig.  2A). Immunoprecipitates from osteoclasts variously treated were made using an antibody to CD44 as indicated in Fig. 2A. These were subjected to an in vitro Rho kinase assay (38). ROK-␣ was coprecipitated with CD44, and ROK-␣ autophos ROK has been identified as the Rho effector regulating CD44 interaction with actin and the plasma membrane (28). Our studies demonstrated coprecipitation of ROK-␣ with CD44 and OPN stimulated the autophosphorylation of ROK-␣. To examine whether the ROK-␣ associated with CD44 is active, in vitro kinase assay was performed in the presence of an exogenous substrate, histone. Anti-CD44 immunoprecipitates made from osteoclasts subjected to various treatments as shown below the respective lanes in Fig. 2B were subjected to in vitro protein kinase assays. ROK-␣ was coprecipitated with CD44, and ROK-␣ phosphorylation was stimulated by OPN treatment (Fig. 2B, lane 3) or Rho Val-14 transduction (lane 4) as compared with PBS-treated (lane 2) osteoclasts. In addition to the phosphorylation of ROK-␣, an increase in phosphorylation of CD44 (*) and ezrin with molecular mass of 80 kDa (**) was observed in OPN-treated (lane 3) and Rho Val-14 -transduced (lane 4) osteoclasts. The increased phosphorylation of ROK-␣ is required for the phosphorylation of CD44, ezrin, and the exogenous substrate, histone. Osteopontin-stimulated phosphorylation of ROK-␣, CD44, and ezrin, as well as the exogenous substrate, histone, was blocked by an antibody to ␤ 3 (lane 5) or OPN (lane 6). Y-27632 has been shown to suppress Rho-or ROK-␣-induced stress fiber assembly (43). The inhibition of OPN-induced autophosphorylation of ROK-␣, as well as phosphorylation of CD44 and ezrin by Y-27632 (lane 7), suggests that phosphorylation of these proteins is ROK-␣ dependent.
The effects of OPN, as well as inhibitors of Rho and ROK on total cellular ROK-␣ phosphorylation and activity, were also analyzed (Fig. 2C). Lysates made from osteoclasts after various treatments as indicated in Fig. 2C were subjected to immunoprecipitation using an antibody to ROK-␣, and in vitro kinase assays were performed in the presence of histone, as an exogenous substrate. The basal level kinase activity in PBS-treated osteoclasts is shown in lane 1. Treatments with OPN (lane 5) or Rho Val-14 (lane 7) increased the ROK-␣ phosphorylation, as well as its activity measured by phosphorylation of the exogenous substrate, histone. Osteopontin-induced phosphorylation and activity of ROK-␣ was blocked by an antibody to ␤ 3 (lane 2) and OPN (lane 3) but not by CD44 (lane 4). Also, OPN-or Rhoinduced ROK-␣ activation was blocked by C3 transferase (lanes 6 and 8) or Y-27632 (lane 9). Even though phosphorylation of few other proteins (indicated by arrows) was observed in OPNand Rho-treated osteoclasts, the levels of these proteins coprecipitated with ROK-␣ were very low. Therefore, weak phosphorylation of these proteins was observed.
Western Analyses-We have further investigated the effects of OPN and Rho protein transductions on ROK-␣ and CD44 phosphorylations by Western analysis of the anti-CD44 immunoprecipitates with phosphothreonine (Fig. 3A) or phosphoserine antibody (Fig. 3C). One-half of the CD44 immunoprecipitates were used for Western analysis with phosphothreonine antibody (Fig. 3A), and subsequently, the blot was stripped and immunoblotted with ROK-␣ antibody (Fig. 3B, top  panel); the second half were used for Western analysis with anti-CD44 antibody to detect the levels of CD44 protein immunoprecipitated (Fig. 3B, bottom panel). Immunoblotting with anti-phosphothreonine antibody (Fig. 3A) confirmed that ROK phosphorylation was low in immunoprecipitates prepared from PBS-treated osteoclasts (lane 1). OPN treatment (lane 2) or Rho Val-14 (lane 3) transduction stimulated CD44-associated ROK-␣ phosphorylation. In addition to ROK-␣, phosphorylation of two other proteins with the molecular masses of 80 and 78 kDa was observed in OPN-treated or Rho Val-14 -transduced osteoclasts. These proteins were identified as ezrin (*) and moesin (**) by Western analysis (data not shown). We did not detect phosphorylation of CD44 with the anti-phosphothreonine antibody. Osteopontin-induced ROK-␣ phosphorylation was decreased by Rho inhibition with C3 exoenzyme (lane 5). Similarly, transduction of Rho Asn-19 (lane 4) or C3-treated Rho Val-14 (lane 6) decreased ROK-␣, as well as ezrin and moesin, phosphorylation on threonine residues. These results confirm that ROK-␣ associates with CD44 and that Rho regulates the phosphorylation of ROK-␣. Y-27632 blocked OPN-or Rho Val-14 -induced ROK-␣ or ERM phosphorylation (lanes 7 and  8). Immunoprecipitation with non-immune serum (NI) is shown in lane 9.  Fig. 3B, respectively. There are no changes in the levels of CD44, and only a single band of 85-kDa CD44 protein was detected (Fig. 3B, bottom panel). Significantly increased amount of ROK-␣ was recruited in the CD44-associated complex in response to OPN (Fig. 3B, top panel, lane 2) and Rho Val-14 (lane 3) treatments. The associated ROK-␣ is inducibly phosphorylated under these conditions. ROK-␣ association with CD44 is blocked by C3 or Y-27632 treatments (lanes 5-8).
Because we did not detect CD44 protein phosphorylation in the Western analysis with phosphothreonine antibody (Fig.  3A), and in vivo phosphorylation of CD44 has been shown to occur on Ser-323 and Ser-325 residues (44), Western analysis of CD44 immunoprecipitates with anti-phosphoserine antibody (Fig. 3C) was performed. Our results demonstrate that serine phosphorylation of not only ROK-␣ and ezrin proteins but also CD44 is increased in response to OPN (Fig. 3C, lane 2).

Neutralizing Antibodies to ␣ v and ␤ 3 Blocked OPN-induced ROK-␣ Phosphorylation
In Vitro Rho Kinase Assay-Consistent with the findings shown in Fig. 2, B and C, OPN-induced phosphorylation of ROK-␣ was decreased by antibodies to ␣ v and ␤ 3 (Fig. 4). The ability of neutralizing antibodies to block OPN-induced phosphorylation of ROK-␣ was studied with antibodies from different sources as described under "Experimental Procedures." Antibodies to ␤ 3 (lanes 4 and 5) and ␣ v (lane 6) but not to CD44 (lanes 7 and 8) blocked the effect of OPN on ROK phosphorylation. Even though antibodies to CD44 did not block OPN effect on ROK-␣ phosphorylation, a blocking effect of these antibodies has been observed in the bone resorption assay (15). Inhibition of OPN-induced ROK-␣ phosphorylation by antibodies to ␤ 3 and ␣ v strongly suggests that the signal from OPN to Rho activation is mediated by ␣ v ␤ 3 integrin receptor.
Taken together, these studies (see Figs. 2-4) demonstrated that CD44 coprecipitates ROK-␣, ezrin, and moesin. The findings that Rho inhibition by C3 transferase or ROK inhibition by Y-27632 (see Figs. 2 and 3) affect the levels of ROK-␣ and ERM proteins associated with CD44 suggest that ROK-␣ phosphorylation, as well as its activation, are required for the CD44⅐ROK-␣⅐ERM complex formation. Moreover, ROK-␣ phosphorylation is dependent on ␣ v ␤ 3 -mediated Rho signaling.

Immunolocalization Analysis of CD44 and ROK-␣
Having established the roles of Rho in CD44 surface expression (15) and ROK-␣ in the phosphorylation of CD44 (see Figs. 2 and 3), we attempted to analyze the localization of ROK-␣ and CD44 in osteoclasts subjected to various treatments (Fig. 5). Osteoclasts were washed with PBS after various treatments and incubated with a monoclonal antibody to CD44 for 2 h. To label the CD44 expressed at the surface, immunostaining with an antibody to CD44 was performed before fixation and permeabilization of osteoclasts. Cells were then rinsed with PBS, fixed with paraformaldehyde, permeabilized with ethanol as described under "Experimental Procedures," and incubated with an anti-goat ROK-␣ antibody for 2 h. After rinsing a few times with cold PBS, osteoclasts were incubated with respective CY2-or CY3-conjugated secondary antibodies (Fig. 5). CD44 (red) and ROK-␣ (green) colocalization (yellow) was found in discrete areas in the PBS (Fig. 5A)-or TAT-TK-treated (Fig.  5D) osteoclasts. OPN (Fig. 5B) and TAT-Rho Val-14 (Fig. 5E) induced the colocalization of CD44 and ROK-␣, as well as CD44 surface expression (see Fig. 6 for XZ scan). ROK-␣ was distributed evenly in the cytoplasm of cells irrespective of treatments (Fig. 5, A-F). Similar to the observations shown in Fig. 3, OPNor Rho-induced interaction of CD44 and ROK-␣ is reduced in osteoclasts treated with Y-27632 (Fig. 5, C and F) or C3 transferase (data not shown). Colocalization of ROK-␣ and CD44 was detected in sparsely dispersed patches in PBS (Fig. 5A)-, Tat-TK (Fig. 5D)-, and Y-27632 (Figs. 5, C and F)-treated osteoclasts. These results demonstrate OPN and Rho stimulation of CD44-ROK-␣ interaction in osteoclasts.
The cellular distribution of CD44 and ROK-␣ was further analyzed by XZ scanning (lateral confocal view) of an OPNtreated osteoclast (Fig. 6). Osteoclast immunostained for CD44 (red) and ROK-␣ (green) is shown in Fig. 6A. The white line in Fig. 6A indicates the location of XZ scanning. Superimposed image (merged) of Fig. 6, B and C is shown in Fig. 6D. Punctate CD44 staining was observed on the surface (indicated by arrowheads in A and arrows in D), and in some areas it is colocalized (yellow) with ROK-␣ on the surface. Colocalization of ROK-␣ and CD44 was also observed at the perimembranous region (yellow; D). Diffuse ROK-␣ staining (green) was observed throughout the cytoplasm.

Surface Expression of CD44 Is Dependent on ␣ v ␤ 3mediated Rho and ROK-␣ Activation
Biotinylation Experiments-Next, we used the Y-27632 inhibitor to determine whether ROK-␣ association with CD44 increased the surface expression of CD44. Y-27632 indeed blocked both Rho Val-14 (Fig. 7A, lane 1)-and OPN (Fig. 7A, lane 3)-induced CD44 surface expression similar to C3 transferase (lanes 2 and 4). The blot shown in Fig. 7A was stripped and reprobed with an antibody to CD44 to demonstrate the total cellular levels of CD44 immunoprecipitated in each lane (Fig.  7B). Only minor changes were observed in the total cellular levels of CD44.
In addition, we have analyzed the effects of neutralizing antibodies to ␣ v ␤ 3 and CD44 receptors on OPN-induced CD44 surface expression (Fig. 7C). Consistent with the observations shown in Fig. 4, OPN-induced surface expression of CD44 is blocked by neutralizing antibody to ␤ 3 (Fig. 7C, lane 4) and not by an antibody to CD44 (Fig. 7C, lane 6). The blot shown in Fig.  7D demonstrates the cellular levels of CD44 in the immunoprecipitates shown in Fig. 7C. Even though there are differences in the total cellular levels of CD44 in the treatments shown in Fig. 7D, it still demonstrates that OPN-induced CD44 expression is blocked significantly by a neutralizing ␤ 3 antibody (lane 4) and not by non-immune IgG (lanes 3 and 5) or a neutralizing CD44 antibody (lane 6).

Effect of Osteopontin Deficiency on the Interaction of CD44 and ROK-␣
In Vitro Rho Kinase Assay-We extended our analyses of ROK-␣ activity to osteoclasts isolated from OPNϪ/Ϫ mice. Although OPN stimulated phosphorylation of ROK-␣ in OPNϪ/Ϫ osteoclasts (Fig. 8A, lane 3 OPN (lanes 2-6). Some osteoclasts were treated with neutralizing antibodies to ␤ 3 (lane 4) and CD44 (lane 6), in addition to OPN treatment. Non-immune IgG treatment was used as nonspecific control (lanes 3 and 5) for treatments shown in lanes 4 and 6. After various treatments, osteoclasts were surface-labeled with biotin, and the lysates were immunoprecipitated with an antibody to sCD44. C, Western analysis of the CD44 immunoprecipitates with streptavidin-horseradish peroxidase to visualize the surface expression of CD44. D, the CD44 immunoblot shown in C was stripped and immunoblotted with an anti-CD44 antibody to demonstrate the cellular levels of mice. ROK-␣ immunoprecipitates made from OPNϪ/Ϫ (lane 1) and WT osteoclast lysates were utilized for identification control. Despite the lack of change in the phosphorylation of total cellular level ROK-␣ in both OPNϪ/Ϫ (lane 1) and WT (lane 10) osteoclasts, a decrease in the levels of ROK-␣ association with CD44 was observed in OPNϪ/Ϫ osteoclasts. Osteoclasts from OPNϪ/Ϫ mice exhibited about 55-60% decrease in basal level phosphorylation of CD44-associated ROK-␣ as compared with WT osteoclasts. These data demonstrate a role for OPN in directing CD44 traffic through Rho/ROK activation.
Even though Rho and Rac GTPases regulate distinct biological processes, a functional overlap exists between these two GTPases. Moreover, NIH3T3 cells transfected with constitutively active Cdc42 induced phosphorylation of ERM proteins through the activation of a Cdc42 binding kinase, MRCK (45). Therefore, we tested the effects of constitutively active form of Rac Val-12 (Fig. 8B, lane 7) and Cdc42 Val-12 (Fig. 8B, lane 6) on the CD44 associated ROK-␣ kinase activity in osteoclasts isolated from wild type mice. Constitutively active forms of

Analysis of CD44 Surface Expression in WT and OPN Null Osteoclasts
Analysis of CD44 Surface Expression by Flow Cytometry-ROK-␣ regulation of CD44 surface expression was also evaluated by FACS analysis (Fig. 9 and Table I). A representative flow cytometry plot for the effects of OPN (Fig. 9A) and Rho   (Fig. 9B) are shown in Fig. 9, and mean fluorescence intensity of osteoclasts subjected to various treatments is shown in Table I. Levels of CD44 expression were found to be higher in OPN-and Rho Val-14 -treated osteoclasts as compared with the respective PBS-or HSV-TK-treated control cells. The increase in the levels of CD44 surface expression can be visualized from the shift in the fluorescence histogram in OPN (Fig.  9A)-and Rho Val-14 (Fig. 9B)-treated WT osteoclasts. Despite OPN or Rho Val-14 -induced CD44 surface expression in OPNϪ/Ϫ osteoclasts, the surface level of CD44 was increased only to the basal level of CD44 observed in PBS-treated WT osteoclasts (Table I). A representative flow cytometry plot for the effect of OPN on the surface expression of CD44 in WT and OPNϪ/Ϫ osteoclast is shown in Fig. 9C. A shift in the fluorescence histogram can be observed for the WT osteoclasts (Fig. 9C). The ROK inhibitor, Y-27632, indeed blocked both Rho Val-14 -and OPN-induced CD44 surface expression in both WT and OPNϪ/Ϫ osteoclasts (Table I). To measure the total cellular levels of CD44, osteoclasts were treated with paraformaldehyde and Triton X-100. The mean fluorescence intensity of the total intracellular levels of CD44 was found to be 203 Ϯ 10.2 and 192 Ϯ 12.2 in WT and OPNϪ/Ϫ osteoclasts, respectively. No significant changes in the total cellular levels of CD44 between PBS-and OPN-treated osteoclasts from WT and OPNϪ/Ϫ mice were observed. These results again confirm the previous demonstration of OPN and Rho stimulation of CD44 surface expression in osteoclasts (see Fig. 7 and Ref. 15).

Immunolocalization of CD44 and ROK-␣ in Osteoclasts Isolated from WT and OPNϪրϪ Mice
Because of the decrease in ROK-␣ activity and CD44 surface expression in OPNϪ/Ϫ osteoclasts, the distribution of CD44 and ROK-␣ was analyzed in osteoclasts isolated from WT (Fig.  9D) and OPNϪ/Ϫ (Fig. 9E) mice. After 5 days in culture, osteoclasts were washed with PBS, and immunostaining was performed using the method as described in the legend for Fig.  5. In wild type osteoclasts, colocalization was observed in patches as shown in Fig. 5A. In OPNϪ/Ϫ osteoclasts, ROK-␣ association with CD44 (CD44/ROK-␣) is decreased because of decreased basal level surface expression of CD44 (red) despite there being no changes in the intracellular distribution of ROK-␣ (green) in both WT and OPNϪ/Ϫ mice.

Effects of Rho and ROK Inhibitors on in Vitro Bone Resorption Activity of WT Osteoclasts
To determine whether the inhibition of ROK-␣ activity would result in a decrease in osteoclast motility and bone resorption activity, osteoclasts plated on dentine slices were treated with Y-27632, in addition to OPN or Rho Val-14 protein (Fig. 10). Rho Val-14 transduction (Fig. 10F) had effects similar to OPN (Fig. 10B) in stimulating resorption pit formation. C3 exoenzyme blocked the OPN (Fig. 10C) and Rho Val-14 effects (Fig.  10G). Similarly, the ROK inhibitor, Y-27632, blocked the effects of OPN (Fig. 10D) and Rho Val-14 (Fig. 10H). Osteoclasts treated with PBS (Fig. 10A) or transduced with HA-TAT (Fig.  10E) demonstrated simple resorption pits. Osteopontin treatment (Fig. 10B) and Rho Val-14 (Fig. 10F) transduction stimulated the formation of multiple overlapping pits, which is produced by the simultaneous process of motility and resorption. The decrease in the pit area in osteoclasts treated with C3 (C and G) or Y-27632 (D and H) indicates that OPN-/Rho-stimulated ROK-mediated signaling regulates osteoclast motility, as well as bone resorption. Even though Y-27632-treated osteoclasts demonstrated formation of some overlapping pits, the resorption area was much smaller than that observed in OPN-or Rho Val-14 -treated osteoclasts. Thus, efficient and significant inhibition of bone resorption was observed in osteoclasts treated with C3 exoenzyme (C and G) or Y-27632 (D and H).

DISCUSSION
The observations shown in this paper demonstrated a direct possible role for ROK-␣ in the phosphorylation of CD44 and the proteins associated with it. The key findings of this paper are as follows: 1) OPN stimulated the phosphorylation of ROK-␣ and its association with CD44. 2) ROK-␣ activation and its association with CD44 increased the phosphorylation of CD44, as well as the ERM proteins associated with CD44. 3) Transduction of constitutively active Rho Val-14 mimicked the OPN effects indicating the role of Rho signaling in CD44 surface expression. 4) The stimulatory effects mediated by OPN or Rho Val-14 were blocked by the ROK inhibitor, Y-27632. 5) Osteoclasts from OPNϪ/Ϫ mice exhibited diminished CD44/ ROK-␣ association, and addition of soluble OPN to OPNϪ/Ϫ osteoclasts only partially restored this interaction. 6) Antibodies against ␣ v , ␤ 3 , or CD44 inhibited the migration and bone resorption of osteoclasts from wild type mice, whereas only anti-␣ v or -␤ 3 antibody blocked the OPN-induced phosphorylation of ROK-␣, CD44, and the ERM proteins.   CD44 is a multifunctional cell adhesion molecule implicated in a wide variety of biological processes including wound healing, cell proliferation, cell differentiation, cell migration, angiogenesis, tumor progression, and metastasis (46 -53). A widely expressed CD44 isoform is the standard or hematopoietic CD44 (sCD44) with molecular mass of 85-95 kDa (54), and we have demonstrated previously (15) the expression of sCD44 in osteoclasts. OPN stimulated the surface expression of CD44, which is required for osteoclast motility and bone resorption. Osteopontin and CD44 expression were increased in migrating fetal fibroblasts (12), and anti-CD44 or anti-OPN antibodies inhibited the migration of CD44-transfected fibroblasts toward OPN in Boyden chambers (6), suggesting CD44-OPN interaction and a role during chemotactic function.
In the present study we show that OPN stimulated ROK-␣ phosphorylation and its interaction with CD44. It is likely that the increase in the phosphorylation of CD44 by ROK-␣ is required for the association of ERM proteins and actin with CD44. This is consistent with other findings that the cytoplasmic domain of CD44 interacts with ERM proteins, and Rho regulates this interaction (24,56). Bourguignon et al. (19) have identified ROK-mediated phosphorylation of certain cellular proteins including the cytoplasmic domain of CD44 v3, 8 -10 (19). To further assess the role of ROK-␣ on the phosphorylation of CD44 and ERM proteins, we used the recently identified ROK inhibitor, Y-27632 (43,57). Y-27632 significantly inhibited OPN-or Rho Val-14 -induced phosphorylation of CD44 and the associated complex. Similarly, confocal immunofluorescence microscopy analysis demonstrated that OPN or constitutively active Rho Val-14 transduction increases colocalization of CD44 and ROK-␣. In control osteoclasts treated with PBS, colocalization of ROK-␣ and CD44 was detected in sparsely dispersed patches or plaques. Colocalization of ROK-␣ and CD44 was reduced or not seen in osteoclasts treated with C3 transferase, the Rho inhibitor, or Y-27632, the ROK inhibitor.
ERM proteins anchor actin to CD44 through Rho-and Racinduced cytoskeletal reorganization (58). In vitro kinase assay, as well as Western analysis, demonstrated coprecipitation of ezrin or moesin with CD44. Treatment of osteoclasts with OPN or HIV-TAT/Rho Val-14 enhanced the phosphorylation of these proteins associated with CD44 whereas transduction of constitutively active Rac or Cdc42 had no effect on the increase in phosphorylation of ROK-␣ or the other CD44 associated proteins. Therefore, these data underscore the distinct role of Rac, as well as Cdc42, and strongly support the critical role of Rho and its downstream effectors on CD44-associated complex formation. In the studies presented here, our approach was facilitated by the use of a ROK inhibitor, Y-27632, as well as by HIV-TAT-mediated delivery of constitutively active and dominant negative Rho proteins into osteoclasts. Our analyses demonstrated that ROK-␣ activation by Rho resulted in the phosphorylation of CD44 and the associated proteins. Inhibition of CD44 phosphorylation, as well as surface expression of CD44 by Y-27632, identified ROK-␣ as the primary candidate required for CD44 surface expression.
Our results are in agreement with several other studies wherein the association between CD44 and ERM proteins is regulated by Rho signaling (20, 58 -61) and that ROK has a predominant role in the phosphorylation of CD44/ERM proteins during actin filament and plasma membrane interaction (28 -30, 62). In various hematopoietic cells including neutrophils, lymphocytes, and platelets, the Rho/Rho kinase pathway mediates the regulation of the actin cytoskeleton (63). Several cellular functions such as cell shape changes and cell motility orchestrated by cytoskeletal reorganization require this regulatory mechanism (64 -66). Rho kinase negatively regulates bone formation, and it has been suggested that inhibition of Rho/Rho kinase may be useful to counteract bone loss in glucocorticoid-induced osteoporosis (67). Also, Rho kinase activity has been shown to be essential for the axonal growth cone dynamics and dendritic patterning (68,69). However, in epithelial cells, even though phosphorylation of CD44 was occurring in both Ser-323 and Ser-325 amino acid residues, mutation analyses demonstrated that phosphorylation of CD44 is not required for targeting of the protein to the basolateral membrane or cytoskeletal interaction (44). Instead, association of cytoskeletal elements involves the transmembrane domain of CD44 (70). Overall, several lines of evidences including our present observations suggest that changes in actin organization, cell shape changes, CD44⅐ERM complex formation, and phosphorylation underlie Rho-mediated activation of ROK-␣.
Although ROK has been identified as direct regulator of CD44⅐ERM complex formation, PI 4P-5kinase also appears to regulate CD44⅐ERM complex formation (29,59,71). Our most recent observations have shown that osteoclasts either transduced with Rho Val-14 or treated with OPN increased the phosphatidylinositol 4,5-bisphosphate levels associated with ezrin or moesin. This latter result is consistent with our previous observations that OPN stimulated gelsolin-associated phosphatidylinositol 4,5-bisphosphate levels indicating the activation of PI 4P-5kinase (37). Even though ROK-␣ increases CD44-associated complex formation, the underlying molecular mechanisms are not clearly understood. One plausible explanation of these observations was that both ROK-␣ and PI 4P-5kinase are required for the CD44-associated ERM complex formation. Further analyses are directed toward understanding the physiological roles ROK-␣ and PI 4P-5kinase in the CD44⅐ERM⅐actin complex formation.
We have shown previously (15) that the lack of autocrine stimulated signal transduction in OPN-deficient osteoclasts leading to decreased CD44 on the surface, which was the OPN null phenotype. These osteoclasts are hypomotile and less resorptive. OPN stimulated CD44 expression on the osteoclast surface, and CD44 was shown to be required for osteoclast motility. Immunocytochemical analyses of non-permeabilized osteoclasts have revealed the localization of CD44 and ␤ 3 on the surface (72). Colocalization of moesin and CD44 at the basolateral plasma membrane was also shown by electron microscopy analyses (30). In Madin-Darby canine kidney cells, endogenous and transfected CD44 are localized to the basolateral plasma membrane (73). Immunostaining of fibroblasts with an antibody to CD44 revealed a relatively even distribution of CD44 on the plasma membrane including the membranous projections. The distribution of CD44 was found to be 100 -200-fold less abundant in the clathrin-coated pits than on the reminder of the cell surface (70,74). Our current observations do not necessarily identify the intracellular distribution of CD44, because the immunostaining for CD44 was performed in non-permeabilized osteoclasts. However, we have demonstrated previously (15) dense CD44 staining at the perinuclear region, near the vacuoles, as well as at the periphery of the cell closer to the plasma membrane of WT and OPNϪ/Ϫ osteoclasts permeabilized with Triton X-100 (15). Our observations on the effects of OPN regulation of CD44 surface expression and the resulting role in osteoclast cell motility are consistent with our previous observations (15) and those of others (72). Also, confocal XZ scanning of OPN-treated osteoclast demonstrated that CD44 localization is limited to the cell surface and the region immediately adjacent to the perimembranous surface. Diffuse distribution of ROK-␣ was observed throughout the cell. CD44 at the perimembranous surface is colocalized with ROK-␣. It is pos-sible that colocalization of CD44 with ROK-␣ may occur at the cytoplasmic domain of CD44.
To further demonstrate the activation of the key step in the OPN-stimulated signal transduction, we used neutralizing antibodies to ␣ v or ␤ 3 and CD44. Osteoclasts were treated with OPN and antibodies to ␣ v or ␤ 3 and CD44 to analyze the role of ␣ v ␤ 3 and CD44 on the phosphorylation and surface expression of CD44. Neutralizing antibodies to ␣ v or ␤ 3 blocked the OPNinduced phosphorylation of ROK-␣ whereas anti-CD44 had no effect. The studies reported here and the previous studies (15) demonstrate that CD44 is a critical matrix receptor of the osteoclast regulating cell motility. This is in agreement with the studies demonstrating the role of CD44 in cell motility (19,76,77). Several studies have implicated the roles of ␣ v ␤ 3 and CD44 in cell motility (see Refs. 7, 55, and 78 -82; reviewed in Ref. 27). Fibroblasts from OPNϪ/Ϫ and CD44Ϫ/Ϫ mice displayed impaired migration and a reduced adhesion to hyaluronan beads (77). Cooperative activity of variant CD44 and ␤ 1 integrin in response to OPN has been observed to be an important step in the stimulation of cell motility and chemotaxis (75).
Rho signaling mediated by integrin ␣ v ␤ 3 is required for both ROK phosphorylation and CD44 surface expression. OPN-induced phosphorylation and surface expression of CD44 was decreased by neutralizing antibodies to ␣ v and ␤ 3 and not by an anti-CD44 antibody. CD44 surface expression is dependent on ␣ v ␤ 3 receptor whereas osteoclast motility and bone resorption are dependent on both ␣ v ␤ 3 and CD44 receptors. A cooperativity exists between ␣ v ␤ 3 and CD44 receptors for osteoclast motility and bone resorption. Still, the details of how CD44 is responsible for osteoclast motility and molecular mechanism involved in CD44 trafficking remain unclear. These questions are being addressed in the ongoing studies in our laboratory.
In summary, our studies demonstrated that OPN-stimulated ␣ v ␤ 3 signaling is required for full activation of CD44 surface expression, and surface expression of CD44 may influence multiple pathways that are critical for osteoclast motility. Moreover, OPN/␣ v ␤ 3 generated outside-in Rho-dependent signaling was required for the surface expression of CD44. ROK-␣ was identified as a critical Rho effector mediating the effects of Rho activation on CD44. Y-27632 elicited effects such as decreased CD44 surface expression and bone resorption in osteoclast culture, providing a means of manipulating ␣ v ␤ 3 -mediated signaling mechanism involved in osteoclast function.