A novel protein kinase C α-dependent signal to ERK1/2 activated by α_Vβ₃ integrin in osteoclasts and in Chinese hamster ovary (CHO) cells

Osteoclasts are multinucleated cells, whose bone resorption function is dependent on tight, but dynamic, adhesion to the bone matrix (Boyle et al., 2003; Tanaka et al., 2003; Teitelbaum, 2000). Osteoclast adhesion structures, termed podosomes (Marchisio et al., 1984), organize the so-called actin ring, a paramarginal area that provides a tight seal to the underneath resorbing lacuna (Teti et al., 1991). Podosomes are also formed in invasive cells, such as macrophages and metastatic cancer cells, where they are called invadopodia (Chen, 1990; Kelly et al., 1994; Linder and Aepfelbacher, 2003; McNiven et al., 2004).

A substantial body of evidence suggests that the main molecular matrix-recognition mechanism of podosomes is the integrin α_Vβ₃ (Pfaff and Jurdic, 2001; Duong et al., 2000). This integrin is predominantly expressed in cancer cells, activated endothelia and osteoclasts (Varner and Cheresh, 1996; Duong et al., 2000; Naik et al., 2003; Jin and Varner, 2004; Guo and Giancotti, 2004), and has been regarded as a potential target for controlling the abnormal behaviour of these cells in pathological conditions (Teti et al., 2002; Kumar, 2003; Nemeth et al., 2003; Jin and Varner, 2004). The integrin α_Vβ₃ in osteoclasts recognises specific bone matrix proteins (i.e. osteopontin and bone sialoprotein II) and is activated by these ligands to provide adhesion, cytoskeletal re-organization and intracellular signalling mandatory for cell survival and bone resorbing activity (Nakamura et al., 1999; Ross et al., 1993). This receptor is expressed at high level in mature osteoclasts along with the α₂β₁ receptor for type I collagen (Helfrich et al., 1996; Villanova et al., 1999).

Podosomes assemble and disassemble within minutes (Destaing et al., 2003; Saltel et al., 2004), are Ca²⁺ and pH sensitive (Teti et al., 1989; Miyauchi et al., 1990; Teti et al., 1991), and are altered by bone resorption regulating factors, which favour their assembly when resorption is stimulated. It is now well recognized that α_Vβ₃ integrin actively participates in podosome regulation, triggering not only signal transduction pathways leading to osteoclast activation but also negative feed-back signals inducing proteasome-dependent cytoskeleton disassembly and remodelling (Sanjay et al., 2001), largely considered key events for podosome dynamisms.

It has been established that α_Vβ₃-mediated adhesion to bone induces osteoclast extracellular signal-regulated kinase (ERK)1/2 phosphorylation, an occurrence thought to be central to osteoclast differentiation and survival (Kim et al., 2003; Nakamura et al., 2003). The link between matrix recognition and ERK1/2 activation has not yet been fully elucidated and is likely to depend upon complex pathways. One such signal could be mediated by the serine-threonine protein kinase C (PKC), since some members of this family are known to participate in the signal cascade to ERK1/2, at least in certain circumstances (Grammer and Blenis, 1997; Aplin et al., 1998; Ni et al., 2003; Pintus et al., 2003). In osteoclasts, ERK1/2 is phosphorylated in response to calcitonin (Zhang et al., 2002), a bone resorption-inhibiting hormone that largely affects the cytoskeletal arrangement and cell motility. Interestingly, PKCs are downstream of calcitonin response (Naro et al., 1998) and regulate the phosphorylation status of proline-rich tyrosine kinase (PYK2) and focal adhesion kinase (FAK) associated with the Ca²⁺ signalling (Zhang et al., 2002). These tyrosine kinases are important integrin effectors and ERK1/2 have been reported to be activated downstream of FAK and PYK2 following integrin ligation (Chen et al., 1994). Furthermore, many of the pathways involved in cytoskeletal remodelling in osteoclasts are regulated by cytosolic Ca²⁺. [Ca²⁺]_i increases in response to α_Vβ₃ engagement (Duong et al., 2000), but it also causes the loss of the actin ring in a manner similar to the effect of calcitonin (Malgaroli et al., 1989; McNiven et al., 2004; Zhang et al., 2002). Cytosolic Ca²⁺ is also an effector of calcitonin response (Teti et al., 1995) and among the many roles played within the cell, it activates the Ca²⁺-dependent members of the PKC family (Hofmann, 1997).

Taken together, these observations suggest a complex interaction among several signalling mechanisms regulating osteoclast adhesion and cytoskeletal remodelling, which could lead to assembly/disassembly cycles of podosomes and actin rings in line with the phases of bone resorption (Teti et al., 1991). We report here that PKCα isoenzyme is recruited by α_Vβ₃ upon activation of the signal transduction pathway associated with α_Vβ₃ and contributes to adhesion-dependent ERK1/2 activation with a novel mechanism not involving the only recognized direct ERK1/2 activator, MAP/ERK Kinase (MEK)1/2.

Cell culture media, reagents and fetal bovine serum (FBS) were from Hyclone (Röntegenstraat, The Netherlands). Culture dishes and sterile plastic ware were from Falcon Becton-Dickinson (Lincoln Park, NJ) and from Costar Co. (Cambridge, MA). The Enhanced ChemiLuminescence (ECL) kit and Hybond nitrocellulose were from Amersham Pharmacia Biotech (Little Chalfont, Bucks, UK). The anti-phospho-p44/42 ERK1/2 (cat. no. 9106), -phospho stress-activated protein kinase/c-Jun N-terminal kinase (SAPK/JNK) (cat. no. 9251), -phospho-p38 (cat. no. 9211), -MEK-1/2 (cat. no. 9122), -phospho-Src(Y416) (cat. no. 2101) were from New England Biolabs Inc. (Beverly, MA). The monoclonal anti-PKCα (cat. no. 610107), -PKCδ (cat. no. 610397) and -integrin β3 (cat. no. 61114U) were from Transduction Laboratories (Lexington, KY). The polyclonal anti-PKCα (cat. no. sc-208), -PKCβ1 (cat. no. sc-209), -PKCϵ (cat. no. sc-214), -PKCζ (cat. no. sc-216-G), -integrin β3 (cat. no. sc-6626), -phospho-MEK-1/2 (cat. no. sc-7995), -ERK2 (cat. no. sc-154), -JNK2 (cat. no. sc-572), -p38 (cat. no. sc-728), -PY99 (cat. no. sc-7020), -actin (cat. no. sc-1616), -focal adhesion kinase (FAK; cat. no. sc-558), -Raf-1 (cat. no. sc-133), -phospho-Raf-1 (cat. no. sc-12358), -Src-homology collagen (Shc) (cat. no. sc-1695), the monoclonal anti-Shc (cat. no. sc-967), the horseradish peroxidase (HRP)-conjugated secondary antibodies and the protein G-plus agarose (cat. no sc-2002) were from Santa Cruz Biotechnology Inc. (Heidelberg, Germany). The anti-α_Vβ₃ (LM609), -β1 subunit (clone P5D2), -α_Vβ₅ (clone P1F6), -α_Vβ₆ (clone E7P6) and fluorescein isothiocyanate (FITC)-conjugated anti-mouse IgGs were from Chemicon (Temecula, CA). The anti-v-Src antibody (cat. no OP-07) was from Oncogene Research Laboratories (Boston, MA). The anti-growth factor receptor-bound protein (Grb2) antibody (cat. no 05-372), the Ras activation assay kit (cat. no 17-218), the Raf-1 kinase cascade assay kit (cat. no 17-357) and the MEK1/2 immunoprecipitation kinase assay kit (cat. no 17-159) were from Upstate (Lake Placid, NY). 1,2-bis(o-amino-phenoxy)ethane-N,N,N′,N′-tetraaceticacidtetra (acetoxymethyl) ester (BAPTA-AM) (cat. no.196419), Gö6976 (cat. no. 365250), PP2 (cat. no. 529573), 5-Iodo-3-[(3,5-dibromo-4-hydroxyphenyl)methylene]-2-indolinone (cat. no. 553008), PD98059 (cat. no. 513000) and U0126 (cat. no. 662005) were from Calbiochem (San Diego, CA). 1,25(OH)₂vitamin D₃ was kindly provided by Domenico Criscuolo and Mauro Piatti (Roche SPA, Milan, Italy). All other chemicals, of the purest grade, were from Sigma Aldrich Chemical Co. (St Louis, MO).

Differentiated primary osteoclast-like cells were obtained from the bone marrow of newborn CD1 mice by a modification of the method described by David et al. (David et al., 1998). Procedures involving animals and their care were conducted in conformity with the NIH Guide for the Care and Use of Laboratory Animals, and with our Institutional guidelines and Ethical Board provisions, in compliance of the Italian Government Regulation no. 116, 27/1/1992.

Mice were sacrificed by cervical dislocation when 5- to 7-days old, and long bones were dissected free from soft tissues and cut into small fragments. Bone marrow cells were released by gently pipetting in Dulbecco's modified minimal essential medium (DMEM) supplemented with 100 IU/ml penicillin, 100 μg/ml streptomycin, 2 mM L-glutamine and 10% FBS. Cells were plated in culture dishes and allowed to attach for 24 hours, then non-adherent cells were removed by aspiration. The total adherent cell fraction was cultured up to 7 days in the presence of 10^-8 M 1,25(OH)₂vitamin D₃. Osteoclast phenotype was evaluated by their multinucleate morphology and tartrate-resistant acid phosphatase (TRAcP) positivity, routinely used in this study for osteoclast recognition, as well as by resorption pit formation ability, appearance of actin rings, expression of α_Vβ₃ integrin receptor, metalloproteinase-9 and cathepsin K and cell retraction in response to salmon calcitonin.

The CHO parental cells (CHOα_V) and the CHO cells stably transfected with the human β3 integrin subunit (CHOα_Vβ₃) were kindly donated by M. H. Ginsberg and Y. Takada, Department of Vascular Biology, Scripps Research Institute, La Jolla, CA. CHOα_V cells were cultured in DMEM supplemented with antibiotics, 1% non essential amino acids and 10% FBS, at 37°C in a 95% air and 5% CO₂ incubator. CHOα_Vβ₃ cells were cultured in the same conditions except for the culture medium which contained geneticin (700 μg/ml) for selection of transfected cells.

Tissue culture plates were coated with 10 μg/ml of LM609 in serum-free DMEM, overnight at 4°C, and blocked with 1% bovine serum albumin (BSA) in phosphate-buffer saline (PBS) for 1 hour at 37°C. Plates were fixed with 60% methanol at 4°C, washed in PBS and maintained in Tris-HCl buffer (Tris-HCl 50 mM pH 7.8, NaCl 110 mM, CaCl₂ 5 mM, 1% BSA, PMSF 0.1 mM) prior to use. Near confluent CHOα_Vβ₃ cells and differentiated osteoclasts were starved overnight in DMEM containing 0.2% BSA or 1% FBS, respectively. Cells were washed in PBS and detached in 0.02% ethylene diamine tetraacetic acid (EDTA), then collected, washed in PBS, re-suspended in serum-free DMEM and held in suspension at 37°C for 1 hour. Cells were either left in suspension and collected as baseline, or placed on LM609-coated plates and allowed to adhere for various times. Cell adhesion to FBS was performed in the same conditions, with serum concentration for coating of 20% in DMEM.

To measure adhesion, CHOα_V and CHOα_Vβ₃ cells were fixed in 80% methanol for 30 minutes, then stained with 0.5% crystal violet in 20% methanol for 5 minutes. Crystal violet was dissolved with 0.1 N sodium citrate in 50% ethanol, and absorbance, linearly proportional to the number of attached cells, was evaluated at 595 nm in an ELISA plate reader. For osteoclasts, cultures were fixed with 3.7% paraformaldehyde in 0.1 M sodium cacodylate buffer for 15 minutes and subjected to histochemical staining for TRAcP activity, using the Sigma-Aldrich kit no. 386 according to the manufacturer's instructions. The number of TRAcP-positive multinucleated cells was then counted.

CHOα_V and CHOα_Vβ₃ cells were detached at confluence by treatment with 0.02% EDTA, washed in PBS and maintained in suspension 1 hour in DMEM plus 10% FBS and 1% nonessential amino acids. Cells were then incubated with 10 μg/ml of primary antibody for 1 hour at 4°C, washed in PBS and incubated with FITC-labelled secondary antibody for 1 hour at 4°C. After washing in PBS, cell surface immunofluorescence was analysed by a flow cytometer fluorescence-activated cell sorting (FACS)scan (Becton Dickinson, Mountain View, CA).

Cells adherent to LM609 or in suspension were washed in PBS and lysed in RIPA buffer (10 mM Tris-HCl pH 7.2, 1% Nonidet P-40, 158 mM NaCl, 1 mM EDTA, 50 mM NaF, 1 mM PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 mM sodium orthovanadate, 10 mM sodium pyrophosphate). Protein content was measured by the Bradford method. For immunoprecipitations, 5 μg of specific antibodies with or without preimmune serum were incubated for 2 hours at 4°C with protein G-conjugated agarose beads. After five washes in RIPA buffer, 1 mg of each sample was added and incubated with beads overnight at 4°C. Samples were then washed five times with RIPA buffer, re-suspended in 2× reducing or nonreducing Laemmli sample buffer and boiled prior to SDS-PAGE.

Cells adherent to LM609 or maintained in suspension were lysed in hypotonic homogenization buffer (10 mM Tris-HCl pH 7.4, 0.5 mM EDTA, 1 mM sodium orthovanadate, 1 mM PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin) for 20 minutes on ice after washing with PBS and centrifuged at 30,000 g for 30 minutes at 4°C. The soluble cytosolic fraction was recovered while the pellets were solubilized in homogenization buffer containing 1% Triton X-100. The soluble membrane fraction was then recovered after centrifugation at 30,000 g for 30 minutes at 4°C. The Triton X-100-insoluble pellets were re-suspended in RIPA buffer and supernatants (Triton X-insoluble fraction) were collected after centrifugation at 10,000 g for 30 minutes at 4°C.

For cytosol and nuclear protein extraction, cells were homogenized in Yoshi A buffer (10 mM Tris-HCl pH 8, 1.5 mM MgCl₂, 10 mM KCl) plus protease inhibitors and centrifuged at 720 g for 10 minutes at 4°C. Supernatants (cytosol fraction) were collected, centrifuged at 11,600 g for 20 minutes a 4°C. Pellets of the first centrifugation were resuspended in Yoshi B buffer (20 mM Tris-HCl pH 8, 25% glycerol, 0.42 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA) plus protease inhibitors, and supernatants (nuclear fraction) were collected after centrifugation at 11,600 g for 20 minutes at 4°C.

Protein content was determined in each fraction by the Bradford method.

Immunoprecipitation samples, 30-60 μg of total proteins or 30 μg of fractionated proteins, were subjected to 10% SDS-PAGE, then electro-transferred to nitrocellulose membranes. After blocking of blots with 5% non-fat milk in TBS-T buffer (20 mM Tris-HCl, pH 7.6, 137 mM NaCl, 0.2% Tween 20), specific primary antibodies, diluted in 1% non-fat milk in TBS-T as indicated in figure legends, were incubated with the blots at 4°C, overnight. Species-specific HRP-conjugated secondary antibodies were diluted 1:5,000 in 1% non-fat milk in TBS-T and incubated for 1 hour at room temperature prior to ECL detection of immuno-complexes, according to the manufacturer's instructions.

GTP-Ras was detected by the Upstate Ras activation assay kit according to the manufacturer's instructions. Briefly, cells were collected and homogenized in 1× MLB Buffer (Mg²⁺ lysis/wash buffer). Cell lysates were centrifuged (5 minutes, 10,000 g, 4°C), then supernatants containing the protein fraction were recovered and incubated with Raf-1 RBD (Ras assay reagent) agarose for 45 minutes at 4°C under gentle agitation. Samples were then centrifuged and the agarose beads washed three times with 1× MLB and resuspended in 2× Laemmli reducing sample buffer. After boiling, samples were subjected to 10% SDS-PAGE, then electro-transferred to nitrocellulose membranes. After blocking with 5% non-fat milk in TBS-T buffer, filters were incubated overnight at 4°C with anti-Ras antibody (1 μg/ml) in TBS-T plus 1% non-fat milk, washed three times and incubated with HRP-conjugated secondary antibody (dilution 1:5,000). The immunoreaction was detected by ECL.

Raf-1 and MEK1/2 kinase assays were performed using the Upstate Raf-1 kinase cascade assay kit and MEK1/2 immunoprecipitation kinase assay kit according to the manufacturer's instructions. Briefly, cell lysates were incubated with inactive MEK1/2 substrate (Raf-1 kinase assay) or inactive MAP kinase 2/Erk 2 substrate (MEK1/2 kinase assay), and with the magnesium/ATP cocktail for 30 minutes at 30°C under gentle agitation. Samples were then centrifuged and incubated with myelin basic protein and [γ-³²P]ATP for 10 minutes at 30°C under agitation. At the end of the incubation, samples were transferred to phosphocellulose squares, washed three times with 0.75% phosphoric acid and once with acetone, then the incorporated radioactivity was counted using a β-counter.

For migration assay, CHOα_Vβ₃ cells were pretreated with 6 nM Gö6976 for 30 minutes, then plated in the upper compartment of a transwell chamber on 12 μm polycarbonate filters pre-coated with 10 μg/ml LM609. After 4 hours of incubation in the presence of NIH3T3-cell-conditioned medium, added to the lower compartment as chemoattractant, filters were stained with Hematoxylin/Eosin. Cell migration ability was evaluated by counting the cells migrated to the lower side of the filters in five randomly chosen fields.

Invasion assay was performed with the same protocol, except that filters were pre-coated with 15 μg/cm² Matrigel.

Osteoclasts were detached in 0.02% EDTA, pretreated with 2 μM Gö6976 for 30 minutes, then plated in the upper compartment of a transwell chamber on 12 μm polycarbonate filters pre-coated with 10 μg/ml LM609 (Chellaiah et al., 2000). After 12 hours in the presence of 25 μg/ml osteopontin added to the lower compartment as chemoattractant, filters were stained for TRAcP and osteoclast migration evaluated by counting the cells migrated to the lower side of the filters in five randomly chosen fields.

Osteoclasts were detached in 0.02% EDTA, pre-treated with 2 μM Gö6976 for 30 minutes and re-plated on bone slices in the presence of the inhibitor for 48 hours. Bone slices were then fixed in 3% paraformaldehyde in 0.1 M cacodylate buffer, cells were removed by ultrasonication in 1% sodium hypochlorite, and slices were stained with 0.1% Toluidine Blue. Pits were counted and the pit index computed according to the method of Caselli et al. (Caselli et al., 1997).

Quantitative data are expressed as the mean ± s.e.m. of at least three independent experiments. Statistical analysis was performed using one-way analysis of variance (ANOVA), followed by the unpaired Student's t-test. A P value <0.05 was considered statistically significant.

The aim of this study being to delineate the role of PKC family members in α_Vβ₃ integrin-mediated signalling, we used as a substrate immobilized LM609, a monoclonal antibody specifically recognizing this integrin. Fig. 1A shows the integrin profile of parental CHO cells, expressing the α_V but not the β₃ subunit (CHOαV), and CHO cells stably transfected with the β₃ subunit (CHOα_Vβ₃). When expressed (Fig. 1Aa), the exogenous β₃ subunit functionally associated with the α_V to form the α_Vβ₃ receptor (Fig. 1Ab lower left panel). CHOα_Vβ₃ cells adhered to LM609 more slowly than to FBS, but after 60 minutes the number of attached cells was indistinguishable between the two substrates (Fig. 1B, lower panels). In contrast, the β3-negative CHOα_V cells, which adhered to FBS in a manner similar to CHOα_Vβ₃ cells, failed to recognise the immobilized LM609 and their adhesion to this substrate was negligible at any time tested (Fig. 1B, upper panels).

In similar assays, prompt adhesion of multinucleated osteoclasts and their putative TRAcP-positive mononuclear precursors to immobilized LM609 was observed, with a kinetics analogous to that of adhesion to FBS-coated substrate, but slower than that of CHOα_Vβ₃ cells (Fig. 1C).

To test whether specific PKC isoforms were engaged in the signalling pathways triggered by α_Vβ₃ integrin, we performed western blot analysis of fractionated cell lysates, which were then probed with antibodies against members of all classes of PKCs. We noted that in CHOα_Vβ₃ cells, adhesion to LM609 induced time-dependent translocation of the classical PKCα from the cytosol to the membrane and to the Triton X-100-insoluble compartments (Fig. 2A, first row). Translocation was apparent after 5 minutes of adhesion, peaking at 20 minutes and remaining stable thereafter. In the same circumstance, translocation of classical PKCβ1 was also noticed. Already after 5 minutes, translocated PKCβ1 appeared in the Triton X-100-insoluble fraction followed by its appearance in the membrane fraction after 10 minutes. It then increased with time only in the former compartment disappearing from the latter at 30 minutes (Fig. 2A, second row). This cellular redistribution of classical PKCs was assumed to be consistent with activation of their catalytic function as it was also induced by short-term (5 minutes) treatment with 12-O-tetradecanoylphorbol-13-acetate (TPA) (Fig. 2A, first and second rows), a reagent known to activate classical and novel PKCs within minutes of exposure (Housey et al., 1988). In contrast, members of the novel (PKCδ, PKCϵ) and atypical (PKCζ) subfamilies were not affected by adhesion of CHOα_Vβ₃ cells to LM609. As expected, the former PKCs were then again activated by TPA within 5 minutes of treatment (Fig. 2A, third and fourth rows).

Similar to CHOα_Vβ₃ cells, osteoclasts attached to LM609 also showed cellular redistribution of PKCα unlike osteoclasts kept in suspension. In these cells, however, the PKCα isoenzyme was already abundant in the membrane in suspended cells and was then mainly recruited to the Triton X-100-insoluble fraction upon adhesion to substrate (Fig. 2B, first row). In contrast, PKCβ1 did not appear to be affected by α_Vβ₃ integrin-mediated adhesion of osteoclasts, as it remained in the cytosolic and membrane compartments (Fig. 2B, second row). Similar to CHOα_Vβ₃ cells, members of the novel (PKCδ and PKCϵ) and atypical (PKCζ) classes remained unaffected upon adhesion to LM609 (Fig. 2B, third to fifth rows). Thus, engagement of α_Vβ₃ integrin in CHOα_Vβ₃ cells and osteoclasts lead to the translocation and activation of classical, but not novel and atypical, PKCs.

Since MAPK are activated downstream of α_Vβ₃ and PKCs, we then assessed changes in the MAPK activation profile. Cells were allowed to attach to LM609 for 30 minutes and evaluated by western blot using antibodies recognizing the phosphorylated forms of ERK1/2, p38 and JNK. In both CHOα_Vβ₃ cells (Fig. 3A) and osteoclasts (Fig. 3B), increase in the phosphorylated species of ERK1/2 was observed upon adhesion to LM609 (Figs 3A,B left panels). In contrast, phospho-p38 remained unchanged and phospho-JNK was undetectable both in suspended cells and in cells attached to LM609 (Fig. 3A,B middle and right panels, respectively). Fig. 3A also includes a phospho-JNK-positive control obtained by treating CHOα_Vβ₃ cells with 10^-7 TPA for 5 minutes.

In CHOα_Vβ₃ cells, ERK1/2 phosphorylation was clearly appreciable after 10 minutes of adhesion, and increased with time up to 30 minutes (Fig. 3C). Analysis of fractionated CHOα_Vβ₃ cells (Fig. 3D) and osteoclasts (Fig. 3E) showed that total ERK1/2 was distributed in the cytosol, membrane and Triton X-100-insoluble fractions, and that adhesion to LM609 triggered its phosphorylation mostly in the cytosol and membrane compartments, with little, if any, activation in the Triton X-100-insoluble fraction.

Immunoprecipitations with preimmune serum (Fig. 4Aa) demonstrate absence of non-specific bands in the assays (shown here for PKCα, not shown for all other assays). To assess whether PKCα was recruited in a complex with the α_Vβ₃ integrin, lysates from suspended and adherent cells were immunoprecipitated with LM609 antibody and western blotted with anti-PKCα antibody. Adhesion to immobilized LM609 recruited PKCα in a complex with α_Vβ₃ integrin both in CHOα_Vβ₃ cells and osteoclasts (Fig. 4Ab,B, left panels). Reciprocal immunoprecipitation of the same lysates with anti-PKCα antibody confirmed co-immunoprecipitation in the same complex of both PKCα and α_Vβ₃ integrin (Fig. 4Ab,B, middle panels). This complex did not appear to recruit ERK, suggesting that this MAPK could be phosphorylated upon α_Vβ₃ engagement only downstream and not by direct association. The complex recruited FAK in CHOα_Vβ₃ cells and PYK2 in osteoclasts (Fig. 4Ab,B, left panels). This difference was not surprising as FAK was the only species expressed by CHOα_Vβ₃ cells (Fig. 4Ab, right panels), and PYK2 was the major species expressed by osteoclasts (Fig. 4B, right panels). However, immunoprecipitation with PKCα antibody failed to reveal co-precipitated FAK or PYK2 (Fig. 4Ab,B middle panels), suggesting that these adhesion tyrosine kinases could interact only with the α_Vβ₃ integrin and not with PKCα. Fig. 4C, shows immunofluorescence analysis of an osteoclast attached to LM609 for 30 minutes, where PKCα (left panels) and α_V integrin subunit (right panels) appear to localize in the same paramarginal area.

At variance with PKCα, activated PKCβ1 failed to associate with the α_Vβ₃ receptor in CHOα_Vβ₃ cells (not shown), suggesting that its recruitment to the Triton X-100-insoluble fraction observed in this cell type does not require direct interaction with the integrin.

ERK activation in response to α_Vβ₃ integrin engagement could be dependent upon activation of the Ras pathway. Indeed it has been proposed that integrin-dependent Ras activation is a consequence of the engagement of the Shc/Grb2 complex (Foschi et al., 1997). To address whether α_Vβ₃ integrin requires this molecular interaction to trigger ERK1/2 phosphorylation, we tested whether Shc was recruited in a complex with α_Vβ₃ upon adhesion to LM609. Our results failed to demonstrate an α_Vβ₃/Shc complex in either CHOα_Vβ₃ cells (Fig. 5A, left panels) or osteoclasts (Fig. 5B, left panels) attached to LM609. Consistently, in PKCα antibody immunoprecipitates from the same lysates, Shc was not apparent upon adhesion to LM609 (Fig. 5A,B, middle panels). In addition, no changes in Shc subcellular distribution (not shown) or in its phosphorylation status (Fig. 5A,B, right panels) were observed in both cell types attached to LM609, relative to suspended cells.

In contrast to Shc, Grb2, which is also known to be activated by direct binding to FAK or PYK2 through its SH2 domain independent of the Shc adaptor protein (Schlaepfer et al., 1994), was recruited by the integrin upon adhesion to LM609 (Fig. 5A,B left panels). It could not, however, be detected in PKCα immunoprecipitates either in CHOα_Vβ₃ cells (Fig. 5A middle panels), or in osteoclasts (Fig. 5B middle panels), suggesting no physical interaction with the kinase.

To address whether Ras was activated by adhesion to LM609, CHOα_Vβ₃ cells and osteoclasts were allowed to attach for 30 minutes and Ras activity evaluated by the appearance of GTP-Ras. Consistent with the previous results, GTP-Ras was not found to be increased by adhesion to LM609 in CHOα_Vβ₃ cells (Fig. 6A). An analogous negative result was observed when Raf-1 activation was evaluated in immobilized LM609-attached cells. In fact, immunoprecipitated Raf-1 was not serine- (Fig. 6B) or tyrosine- (Fig. 6C) phosphorylated upon substrate recognition. Consistent with these results, immunoprecipitated Raf-1 failed to recruit PKCα (Fig. 6B) as well as c-Src (Fig. 6C), which is implicated in the tyrosine phosphorylation required for its full activation. Positive controls, performed by treatment of serum-starved cells with 20% FBS, showed integrity of these pathways, with expected PKCα recruitment and Raf-1 serine phosphorylation (Fig. 6B) as well as c-Src recruitment and Raf-1 tyrosine phosphorylation (Fig. 6C) in Raf-1 immunoprecipitates. Furthermore, treatment with 15 μM Raf-1 inhibitor, 5-iodo-3-[(3,5-dibromo-4-hydroxyphenyl)methylene]-2-indolinone, was without effect on CHOα_Vβ₃ cells attached to LM609 (Fig. 6D, right panels), while blocking ERK1/2 activation in FBS-treated cells (Fig. 6D, left panels). Finally, Raf-1 kinase assay demonstrated no activity of the kinase in response to adhesion to LM609, as opposed to treatment with FBS (Fig. 6F).

Similar independence of this pathway was observed in osteoclasts, as suggested by lack of both appearance of GTP-Ras, and Raf-1 serine and tyrosine phosphorylation (not shown). Furthermore, no changes in ERK1/2 activation in the presence of the Raf-1 inhibitor (Fig. 6E), or Raf-1 kinase activity (Fig. 6G) were observed in osteoclasts attached to LM609 as opposed to FBS-treated osteoclasts.

We then assessed the involvement of MEK1/2 in α_Vβ₃-mediated intracellular signalling. MEK1/2 kinase assays showed intense activity in response to treatment with FBS, but failed to demonstrate any detectable change over baseline in response to adhesion to LM609 both in CHOα_Vβ₃ cells (Fig. 6H) and in osteoclasts (Fig. 6I). Furthermore, CHOα_Vβ₃ cells (Fig. 7A,B) and osteoclasts (Fig. 7C) attached to LM609 failed to show increased MEK1/2 phosphorylation. Treatments with the MEK1/2-specific inhibitors PD98059 and U0126 proved efficacious in inhibiting MEK1/2 phosphorylation in both CHOα_Vβ₃ cells (Fig. 7B, upper panels) and osteoclasts (Fig. 7C, upper panels). However, in this circumstance, consistent with the previous results, no effect on adhesion-induced ERK1/2 phosphorylation was observed (Fig. 7B,C, third panels from above), indicating that this event is independent of MEK1/2 activity.

ERK1/2 is believed to be constitutively bound to MEK1/2 in the cytoplasm and activation of MEK1/2 not only phosphorylates ERK1/2 but also releases ERK1/2 from MEK1/2 so it can translocate to the nucleus. This could imply that MEK1/2-independent ERK1/2 phosphorylation may activate ERK1/2 but could prevent its nuclear translocation. Therefore, to address this issue, we performed cytosol/nuclear fractionation of CHOα_Vβ₃ cell (Fig. 7Da) and osteoclast (Fig. 7Db) lysates, and observed phosphorylated ERK1/2 nuclear translocation in cells attached to LM609 both in control conditions and in the presence of PD98059 or U0126, thus ruling out that α_Vβ₃-mediated signalling could result only in partial ERK1/2 activation. To assess the efficacy of the inhibitors and the specificity of the MEK1/2-independent pathway, we treated CHOα_Vβ₃ cells (not shown) and osteoclasts (Fig. 7E) with 20% FBS, which contains a variety of mitogens inducing the canonical MEK1/2-dependent ERK1/2 activation, and consistently observed MEK1/2 phosphorylation (Fig. 7E,) as well as ERK1/2 phosphorylation (Fig. 7E,) and nuclear translocation (Fig. 7E), which were all inhibited by the PD98059 and U0126.

Taken together, these results suggest that ERK1/2 activation by α_Vβ₃ engagement is not downstream of the canonical Ras/Raf-1/MEK1/2-dependent signals.

As the PKCα is a Ca²⁺-dependent PKC family member, we next assessed the role of Ca²⁺ in PKCα redistribution and recruitment by α_Vβ₃, and in ERK1/2 activation. Addition of 2 mM O,O′-bis(2-aminoethyl)ethyleneglycol-N,N,N′,N′-tetraacetic acid (EGTA) to chelate extracellular Ca²⁺ was without effect (not shown). In contrast, chelation of intracellular Ca²⁺ by incubation with 2 mM EGTA/50 μM BAPTA-AM prevented adhesion-dependent PKCα translocation from cytosol to membrane and/or Triton X-100-insoluble fraction (Fig. 8A,B upper panels), PKCα recruitment by engaged α_Vβ₃ (Fig. 8A,B, lower left panels) and ERK1/2 phosphorylation (Fig. 8A,B, lower right panel) both in CHOα_Vβ₃ cells (Fig. 8A) and in osteoclasts (Fig. 8B). It is therefore clear that all the effects of α_Vβ₃ on PKCα and on ERK1/2 activation are Ras/Raf-1/MEK1/2 independent but Ca²⁺ sensitive.

We next evaluated pharmacologically the involvement of PKCα in α_Vβ₃ integrin-mediated ERK1/2 phosphorylation. In Fig. 9A,B, the effect of the PKCα catalytic activity inhibitor, Gö6976, as well as that of the PKCα downregulation by long-term treatment with TPA (Fig. 9A left panels) are shown. Both inhibitors prevented ERK1/2 phosphorylation in CHOα_Vβ₃ cells (Fig. 9A, right panels) and osteoclasts (Fig. 9B) upon adhesion to LM609.

c-Src is an important mediator of the α_Vβ₃ signals (Sanjay et al., 2001). Therefore we investigated whether any interaction between this tyrosine kinase and PKCα was exhibited downstream of the integrin engagement. Similar lack of ERK1/2 activation was observed in CHOα_Vβ₃ cells (Fig. 9C) and osteoclasts (Fig. 9D) treated with the c-Src inhibitor PP2. However, this effect appeared to be independent of physical interaction with PKCα. In fact, a complex with c-Src was not observed in cell lysates immunoprecipitated with PKCα antibody (Fig. 9E,F) in both cell types, which was, however, noticed in serum-starved cells treated with 20% FBS used as positive control. Moreover, in the presence of PP2, α_Vβ₃ recruited PKCα upon adhesion to LM609 in a manner similar to untreated cells (Fig. 9G,H). Interestingly, as previously shown in Fig. 6B,C, Raf-1 activation was not downstream of c-Src, ruling out the involvement of the Ras/Raf-1/MEK1/2 pathway also in α_Vβ₃ integrin-mediated, c-Src-dependent ERK1/2 phosphorylation.

Finally, to assess the functional role of PKCα signalling induced by α_Vβ₃ engagement in CHOα_Vβ₃ cells and osteoclasts, we performed adhesion experiments in the presence of PKCα inhibitor, Gö6976. We did not observe any change in the adhesion rate of CHOα_Vβ₃ cells (Fig. 10A) or in the number of osteoclasts (Fig. 10B) attached to LM609, suggesting that activation of PKCα in our experimental conditions may affect events downstream of adhesion. Consistent with this hypothesis, we noted that migration of CHOα_Vβ₃ cells (Fig. 10C) and osteoclasts (Fig. 10D), in Boyden chambers whose membranes were coated with LM609, was significantly reduced by the PKCα inhibitor. In contrast, the ability of CHOα_Vβ₃ cells to penetrate through Matrigel, a reconstituted basement membrane recognized by several integrins, was not affected by Gö6976 (Fig. 10E). Notably, the osteoclast resorbing ability, evaluated by the pit assay, was significantly reduced (Fig. 10F) suggesting an important functional role in osteoclast activity. Finally, no effect of PKCα inhibitors were observed on cell survival (not shown).

Despite the fact that the α_Vβ₃ function is well known in osteoclasts and much information is available on its signalling mechanisms and interaction with membrane receptors relevant to the osteoclast activity (McHugh et al., 2000; Nakamura et al., 1999), the molecular machinery associated with its downstream effectors is still poorly understood (Duong et al., 2000). In this study, a role for the so-called classical (Ca²⁺- and diacylglycerol-dependent) PKC isoform α has been elucidated. This serine/threonine kinase was found to be involved in α_Vβ₃-mediated ERK1/2 activation in an intracellular Ca²⁺-dependent manner, but independently of the adaptor protein Shc, of the Ras/Raf-1/MEK complex, and of the c-Src non-receptor tyrosine kinase signals.

In previous reports, the use of promiscuous substrates recognized by more than one integrin may have made it difficult to dissect the mechanisms associated with each specific matrix receptor (Arnaut et al., 2002). In our study, we used a highly specific monoclonal antibody, largely employed to activate the α_Vβ₃ receptor (Bhattacharya et al., 2000; Murphy et al., 2003), and detected the downstream signals associated exclusively with this integrin. This is indicated by the fact that in CHO cell lines, only the clonal cells carrying the α_Vβ₃ receptor and not the parental cells, expressing the α_V but not the β₃ subunit, could adhere to the LM609 antibody. Adhesion to this substrate triggered PKCα translocation, i.e. activation, and subsequent ERK1/2 phosphorylation in CHOα_Vβ₃ cells and also in osteoclasts, which are known to express high copy numbers of the α_Vβ₃ receptor (Duong et al., 2000; Villanova et al., 1999; Helfrich et al., 1996).

α_Vβ₃ ligation-dependent ERK1/2 phosphorylation was prevented by the PKCα-specific inhibitor Gö6976, and by downregulation of PKCα by long-term treatment with TPA, demonstrating that it occurs as a result of activation of this PKC isoform. PKCα inhibition also had functional consequences, causing reduction in migration ability and, in the case of osteoclasts, also reduction in bone resorption. These observations suggest that an important role of PKCα activation downstream of α_Vβ₃ engagement could be with cell motility, the impairment of which could also affect the ability of osteoclasts to resorb bone. No changes in survival rate were observed when cells were treated with Gö6976 (not shown), and this inhibitor did not alter the ability of cells to interact with the LM609 substrate, clearly indicating that the PKCα involvement occurs mainly in post-adhesion events.

It is interesting to note that α_Vβ₃ integrin-mediated PKCα translocation occurred to the Triton-X-insoluble fraction, particularly in osteoclasts, where similar levels of the enzyme were instead found in the Triton-X-soluble, membrane fraction, irrespective of α_Vβ₃ engagement. This compartment is believed to consist of membrane microdomains, denominated lipid rafts, that are enriched in cholesterol and glycosphingolipids, implicated in several processes, including signal transduction, endocytosis, membrane trafficking, cytoskeletal reorganization and pathogen entry (Munro, 2003; Pike, 2004). α_Vβ₃ integrin may be accumulated in rafts (Triantafilou and Triantafilou, 2003), and its ligand-induced activation is enhanced by glycosyl phosphatidylinositol anchor (McDonald et al., 2004). Lipid raft integrity is required for receptor-mediated actin cytoskeleton reorganization and PKCα translocation (Becart et al., 2003) and the raft fraction is assumed to be important for the Ca²⁺ signalling and the Ca²⁺-dependent association of classical PKC to these microdomains (Orito et al., 2001). It is, therefore, possible that this fraction may play an important role in the Ca²⁺-dependent PKCα signalling induced by ligand-dependent α_Vβ₃ activation.

Interestingly, within the MAPK family, only ERK1/2 were found to be activated by signals triggered by α_Vβ₃ ligation. Neither p38 nor JNK were altered by adhesion to LM609, and ERK1/2 phosphorylation was triggered in response to α_Vβ₃-mediated adhesion not only in a PKCα-, but also in a Ca²⁺- and c-Src-dependent manner.

[Ca²⁺]_i appears to be central to the α_Vβ₃ signal transduction pathway. It has been previously demonstrated that adhesion of osteoclasts to α_Vβ₃ substrates induces intracellular Ca²⁺ mobilization, which is believed to be an important signal for cytoskeletal remodelling (Duong et al., 2000). Cytosolic Ca²⁺ acts at, at least, three levels: PKCα (this study), PYK2 (Lev et al., 1995) and gelsolin (Robinson et al., 2000). These three pathways are regulated in a Ca²⁺-dependent fashion and are implicated in different downstream events associated with α_Vβ₃-mediated signals. PYK2 appears to recruit c-Src to the adhesion sites and trigger c-Src-dependent c-Cbl phosphorylation which has been suggested to cause ubiquitination and proteasome degradation of the podosome molecular complex (Sanjay et al., 2001). These events may be highly relevant for podosome assembly/disassembly cycles which make these adhesions highly dynamic. Another Ca²⁺-sensitive component of podosomes is the microfilament-severing protein gelsolin (Akisaka et al., 2001). This protein, which is associated with phosphatidylinositol bisphosphate in the membrane, is released from it upon receptor-mediated lipid breakdown, and is activated in a Ca²⁺-dependent manner to severe pre-existing microfilaments and nucleate new microfilaments for cytoskeletal remodelling (Wang et al., 2003). Results from our study also include PKCα in this scenario. Its α_Vβ₃-dependent subcellular redistribution and recruitment in a complex with α_Vβ₃ integrin is, not surprisingly, largely dependent on cytosolic Ca²⁺ concentration and is prevented by the intracellular Ca²⁺ chelator BAPTA. PKCα is member of the classical, Ca²⁺ and diacylglycerol dependent, PKCs and its activation by cytosolic Ca²⁺ was expected. Miranti et al. (Miranti et al., 1999) had previously suggested that integrin-associated PKC signal was triggered by phospholipase C (PLC)γ1, an event found not to occur in our study (not shown). However, we found that phosphoinositide 3-kinase, which is also involved in integrin-induced signalling (Duong et al., 2000), is not implicated in PKCα recruitment by the α_Vβ₃ receptor, or in ERK1/2 activation (not shown); the upstream regulators of this signal, therefore, remain to be elucidated.

Cytosolic Ca²⁺ chelation by BAPTA inhibited ERK1/2 phosphorylation upon adhesion of cells to LM609. Consistent with this result, inhibition of PKCα by the specific inhibitor Gö6976, or by long-term exposure to TPA, also prevented ERK1/2 phosphorylation. Analogous inhibition was observed upon pre-treatment of cells with the c-Src kinase inhibitor, PP2, but this mechanism appeared to be independent of PKCα, as the two kinases were not found to be associated in a molecular complex upon engagement of α_Vβ₃ integrin, neither was the formation of the α_Vβ₃/PKCα complex prevented by PP2.

PKCα and c-Src downstream signals of ERK1/2 phosphorylation remain to be established. Surprisingly, we could not observe Shc engagement in a complex with α_Vβ₃ integrin, or its phosphorylation or subcellular redistribution. Therefore, at variance with previous results (Giancotti and Ruoslahti, 1999; Hood and Cheresh, 2002), in our experimental conditions, Shc appears not to be involved in the α_Vβ₃ activation of ERK1/2. However, we did observe recruitment of Grb2 into a complex with the integrin upon adhesion of cells to LM609. In other signalling pathways, Grb2 has been shown to establish direct molecular interactions with FAK or PYK2, independent of Shc (Schlaepfer and Hunter, 1996). Most intriguing, however, is the observation that downstream of Grb2, neither Ras, Raf-1 nor MEK1/2 were involved in ERK1/2 activation. There are only a few examples in the literature suggesting alternative pathways for ERK1/2 involving classical PKCs that, at variance with our findings, co-exist with the typical Ras/Raf-1/MEK1/2 cascade (Grammer and Blenis, 1997). These pathways are known to regulate the prolonged activation of ERK1/2, but their specific components remain to be elucidated.

Similar independence of MEK1/2 was observed in the c-Src pathway leading to ERK1/2 phosphorylation, suggested by the lack of c-Src-induced MEK1/2 tyrosine phosphorylation. Therefore, it appears that PKCα and c-Src converge on ERK1/2 in response to α_Vβ₃ ligation, although they do not seem to interact in the same complex. In T-cells, Niu et al. (Niu et al., 2003) have shown a link between a member of the c-Src family (Lck) and PKC in response to activation of a subset of integrins (β₁) requiring the CD45 receptor tyrosine kinase. However, this interaction leads to Shc activation, an event not observed in our experimental conditions.

In conclusion, our study suggests that PKCα is central to α_Vβ₃-mediated intracellular signalling inducing ERK1/2 activation by a novel pathway, and that it regulates events downstream of adhesion, including cell migration and, in the case of osteoclasts, bone resorption.

This work was supported by the Telethon grant E.0831, the Fondo per gli Investimenti per la Ricerca di Base (FIRB) grant RBAU01X3NH, the Associazione Italiana per la Ricerca sul Cancro (AIRC), the Agenzia Spaziale Italiana (ASI) grant I/R/108/00 and EC grants METABRE (contract no. LSHM-CT-2003-503049) and OSTEOGENE (contract no. LSHM-CT-2003-502941) to A.T. The excellent technical support of Luigi Pellegrino and Rita Di Massimo is gratefully acknowledged.