Apoptosis and survival of osteoblast-like cells are regulated by surface attachment.

We tested the hypothesis that RGDS peptides regulate osteoblast survival in culture. Osteoblast-like MC3T3-E1 cells were allowed to attach to RGDS peptides that had been tethered to a silicone surface utilizing a previously described grafting technique. The RGDS-modified surface caused up-regulation of alpha(v)beta(3) integrin. We noted that there was an increase in expression of activated focal adhesion kinase and activated Akt. There was no change in the expression level of the anti-apoptotic protein Bcl-2, the pro-apoptotic protein Bad, or the inactivated form of Bad, pBad. Attachment to the RGDS-treated membrane completely abolished apoptosis induced by staurosporine, the Ca(2+).P(i) ion pair, and sodium nitroprusside. However, the surface modification did not interfere with apoptosis mediated by the free RGDS peptide or serum-free medium. When the activity of the phosphatidylinositol 3-kinase pathway was inhibited, RGDS-dependent resistance to apoptosis was eliminated. These results indicated that the binding of cells to RGDS abrogated apoptosis via the mitochondrial pathway and that the suppression of apoptosis was dependent on the activity of phosphatidylinositol 3-kinase.

Osteoblast survival requires attachment to specific extracellular matrix proteins, a phenomenon termed anchorage dependence (1). The survival signal is mediated by transmembrane integrin receptors that provide a mechanical link between the cell and the hydrophilic peptide sequence Arg-Gly-Asp (RGD) on extracellular macromolecules (2,3). This amino acid motif enhances initial osteoblast attachment and spreading (4,5). In addition to influencing downstream maturation events (6), RGD signaling (7) induces mineralization of the extracellular matrix (8,9). Likewise, exposure of osteoblasts to immobilized RGD peptides increases osteoblast maturation (9 -12); at millimolar concentrations, free RGD-containing peptides induce skeletal cell death (13).
The interaction between the RGD motif and integrin receptors causes phosphorylation of the non-receptor protein-tyro-sine kinase and activation of focal adhesion kinase (FAK) 1 (14,15). Once phosphorylated, FAK provides binding sites for Src homology-2 domains of the survival signaling pathway molecules Grb2-Sos and phosphatidylinositol 3-kinase (PI3K) (16,17). Of the downstream targets of PI3K, Akt plays a critical role in the regulation of the balance between apoptosis and survival. Once the Akt pathway is activated, there is stimulation of a number of downstream events that include: phosphorylation and inactivation of caspase-9, resulting in inhibition of apoptosis and enhancement of cell survival (18,19); maintenance of the mitochondrial membrane potential and suppression of mitochondrion-initiated apoptosis (20); and phosphorylation of Bad and prevention of mitochondrial dysfunction (18,21). In addition, activated Akt phosphorylates a number of cell survival-related transcription factors, including Forkhead 1 (22), NF-B (23,24), and Creb (25).
Despite overwhelming evidence that relates integrin-mediated adhesion with cell survival (26), the relationship between adhesion and the regulation of bone cell apoptosis is poorly understood. In this study, we address this issue by examining signaling events that accompany the binding of osteoblast-like cells to an RGD-modified surface. We demonstrate that the surface increases both integrin and FAK expression and upregulates the PI3K survival pathway. In addition, we show that the RGD-modified surface inhibits mitochondria-dependent apoptosis.

Study Design
The goal of this investigation was to evaluate the mechanism by which RGDS-mediated integrin attachment promotes the survival of osteoblast-like cells. We chemically bonded an RGDS peptide to a functionalized silicone membrane. A similar RGES-grafted surface was used as a control substrate. MC3T3-E1 cells were then grown on the RGDS and control surfaces. We chose to use this osteoblast-like cell line, as our previous studies have shown that these cells exhibit an apoptotic response identical to that of primary osteoblasts (13,27). Other workers have demonstrated that this cell line mimics the major phenotypic characteristics of osteoblasts (28). Changes in intracellular signaling molecules, FAK, phospho-FAK, Akt, phospho-Akt, Bad, phospho-Bad, and Bcl-2, were evaluated by Western blot analysis. PCR analysis was also used to evaluate changes in expression of Bad and Bcl-2. Rather than evaluate a wide range of integrins, we focused on ␣ v ␤ 3 by Western blot analysis. These integrins are known to be expressed by MC3T3-E1 cells as well as other osteoblast-like cells and primary osteoblasts derived from bovine, rat, and human bone samples. Moreover, they are expressed at all stages of development. To evaluate the impact of the peptide on susceptibility to apoptosis, cells were cultured on the RGDS-grafted surface for 3 days and then treated by serum starvation or with low doses of the following agents: staurosporine, the Ca 2ϩ ⅐P i ion pair, sodium nitroprusside or free RGDS peptide. These agents were selected for two reasons: they are effective osteoblast apoptogens, and they activate the extrinsic, as well as the intrinsic, apoptotic pathway. Cell death was determined by the MTT assay. Finally, the importance of PI3K pathway-activated survival signals was assessed through the use of the specific kinase inhibitor, LY294002.

Reagents and Antibodies
Anti-␣ v and anti-␤ 3 (polyclonal) integrin antibodies were obtained from Chemicon International Inc. (Temecula, CA). Anti-phospho-FAK (Tyr397) rabbit anti-serum and anti-FAK rabbit polyclonal antibodies were provided by Upstate Biotechnology (Lake Placid, NY). Anti-Akt and anti-phospho Akt antibodies were purchased from Cell Signaling Technology (Beverly, MA). Anti-Bad polyclonal, anti-phospho-Bad, and anti-Bcl-2 antibodies were purchased from Sigma-Aldrich. Anti-GAPDH antibody was obtained from Research Diagnostics Inc. (Flanders, NJ). RGDS and RGES peptides were obtained from Sigma. The PI3K inhibitor LY294002, was purchased from Cell Signaling Technology. Protein A-Sepharose beads were from Amersham, and gel/ blotting material was obtained from Invitrogen.

Preparation of RGD-treated Silicone Membranes
RGD peptides were grafted to silicone surfaces utilizing a technique described previously (9) based on a method originally reported by Dee et al. (7,29,30). Briefly, 0.005-inch silicone sheets (Silastic Q7-4840, Specialty Manufacturing Inc., Saginaw, MI) were exposed for 10 min to UV light/ozone (UVO Cleaner, Jelight Company Inc., Irvine, CA) to functionalize and oxidize the surface. The functionalized silicone surface was then modified by treatment with 0.2 mM 3-aminopropyltriethoxysilane (A3648, Sigma) in hexane for 45 min. The 3-aminopropyltriethoxysilane molecule reacted with the OH groups, generating an aminated surface. The silicone membranes were then sonicated (Branson Ultrasonics Corp., Danbury, CT) in a hexane bath to remove excess reactants. The aminated membranes were incubated with 0.2 mM RGDS or RGES peptides in 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide and N-ethyl morpholine (Sigma). The carboxyl terminus of the peptide reacted with the amino group of the silane molecule, thereby forming a covalent link to the biomaterial surface. The membranes modified with the bound peptides were rinsed with N,N-dimethylformamide and distilled water. Non-bound peptides were removed by sonication in N,N-dimethylformamide for 15 min. Finally, the membranes with covalently attached peptides were sterilized in 75% ethanol. A control surface was engineered using RGES peptides using an identical procedure as described for RGDS. It has been shown that when Glu is substituted for Asp there is a profound loss of biological activity (31).

Cell Culture
Osteoblast-like MC3T3-E1 cells were used in this study. Our previous work has demonstrated that the apoptotic response of MC3T3-E1 cells is identical to that of primary human osteoblasts isolated from bone fragments (27). The cells were maintained in 10 ml of complete medium consisting of Dulbecco's minimum essential medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and 50 g/ml penicillin/streptomycin, pH 7.4. After the cells had reached confluence, they were released by treatment with 5 ml of 0.1% collagenase in Ca 2ϩand Mg ϩ -free Hanks' buffered saline solution. The cells were then replated on the modified silicone membranes in 100-mm culture dishes and 12-or 24-well plates (Corning Glass). Cultures were fed every other day with complete medium supplemented with 50 g/ml ascorbic acid and 5 mM ␤-glycerophosphate. To evaluate cell death, osteoblast-like cells were plated on experimental (RGDS) and control (RGES) substrates in 12-or 24-well plates at a density of 50,000/well or 25,000/well, respectively. After 3 days in culture in complete media, the osteoblasts were incubated overnight with sublethal doses of staurosporine (0.1 and 0.5 M) (32,33), the Ca 2ϩ (2.4 and 2.9 mM) and P i (3 mM) ion pair (27), sodium nitroprusside (0.5 and 0.1 mM) (34), or RGDS (1 mM and 5 mM) (13) or were serum-starved (35). The ion pair, sodium nitroprusside, and staurosporine all induce cell death through the intrinsic pathway; RGDS and serum starvation probably kill osteoblasts by a mitochondria-independent pathway.

Western Blot Analysis for Integrins
After 0.5, 3, 24, and 72 h in culture on the prepared RGDS or RGES surfaces, cells were solubilized in 1% Nonidet P-40 lysis buffer. Protein concentration was determined using a BCA protein assay kit (Pierce). 100 g of protein was immunocomplexed with 1:50 dilution of anti-␤ 3 integrin polyclonal IgG. After gentle mixing and incubation for 1 h on ice, 20 g of protein A-Sepharose beads was added to the lysate. The mixture was centrifuged at 10,000 ϫ g for 15 s at 4°C; the pellet was washed five times with lysis buffer and resuspended in Laemmli sample buffer. The proteins were separated in 3-8% Tris acetate gel (SDS-PAGE) and transferred to a nitrocellulose membrane. The membrane was blocked in Tris-buffered saline with 5% skim milk powder. After washing in Tris-buffered saline, the blots were incubated with ␣ v antibody (1:1000) overnight at room temperature. The blots were washed and incubated with horseradish peroxide-conjugated antibody for 1 h, and positive bands were detected using the ECL chemiluminescence kit (Amersham Biosciences) and Kodak X-Omat blue film (Eastman Kodak Co.). GAPDH was used as a control to estimate protein loading on the gel. The concentration of the anti-GAPDH antibody was 2.4 g/ml.

Evaluation of Signaling Pathways
Western Blot Analysis-The protocol described above was followed for assessing the expression of FAK and phosphorylated FAK (antibody concentration 1:500), Akt and phospho-Akt (antibody concentration 1:500), Bad, phosphorylated Bad, and Bcl-2 (antibody concentration 1:300). GAPDH was used as a loading control.

Assessment of Apoptotic Sensitivity
After 3 days in culture on the prepared surfaces, cells were incubated overnight with sublethal doses of apoptogens. Untreated cells and cells plated on the RGES surface were used as controls. Cell death was measured using the MTT analysis (36,37). This assay is based on the ability of mitochondrial dehydrogenases to oxidize thiazolyl blue (MTT) (Molecular Probes, Eugene, OR), a tetrazolium salt (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenylterazolium bromide), to an insoluble blue formazan product. The cells were incubated with the MTT reagent (120 g/ml) at 37°C for 2 h. After the supernatant was removed, 400 l of 0.04 mol/liter HCl in isopropanol was added to each well, and the optical density of the solution was read at 590 nm in an enzyme-linked immunosorbent assay plate reader. As the generation of the blue product is proportional to the dehydrogenase activity, a decrease in the absorbance at 590 nm provides a direct measurement of the number of viable cells. To determine the contribution of the PI3K pathway to inhibition of apoptosis, some cell populations were pretreated with 50 M LY294002, a PI3K inhibitor (38,39). Following this pretreatment, cell death was determined as described above.

Statistical Analysis
Experiments were repeated three to five times. Similar results were seen with each of the replicates. Data were analyzed using a one-way analysis of variance; the Student-Newman-Keuls post hoc test for a comparison of individual means was used. p Ͻ 0.05 was considered to be statistically significant.

Osteoblast Binding to RGD-grafted Surfaces Up-regulates
Integrin Expression-Integrin expression levels were evaluated in MC3T3-E1 osteoblast-like cells cultured on RGDtreated membranes after 0.5, 3, 24, and 72 h. Fig. 1 shows that there was an increase in ␣ v ␤ 3 heterodimer expression on the RGDS surface when compared with the RGES controls. Expression levels were raised after 24 h and maintained for 72 h.
Osteoblast Binding to RGD Peptides Up-regulates Survival Pathways-Integrin signaling events that are triggered by osteoblastic attachment to the RGDS-grafted surface were evaluated by measuring Bcl-2, Bad, and FAK expression. FAK ( Fig.  2A) and phosphorylated FAK (Fig. 2B) were constitutively expressed by cells adherent to both the RGD-and RGE-grafted surfaces. A modest increase in phosphorylated FAK was evident in cells adherent to the RGD-grafted surface (Fig. 2B).
We investigated downstream signaling events mediated by activation of FAK. Western blot analysis showed that for the first 24 h, there was little difference in levels of Akt expressed by osteoblasts on the two surfaces (Fig. 3). However, at 72 h, there was a considerable increase in Akt protein expression in cells cultured on the RGDS surface. Interestingly, there was little difference in activated phospho-Akt on either surface at 0.5 or 3 h. However, by 24 h, there was a considerable increase in activated Akt in relationship to GAPDH. The increase in phospho-Akt was maintained for 72 h. At this time, the increase in phospho-Akt was directly proportional to the increase in total Akt protein.
To further explore downstream signaling events, we examined the expression of the pro-apoptotic protein Bad (Fig. 4A) and the inactive form of the protein, pBad (Fig. 4B). Both proteins were constitutively expressed by the osteoblast-like cells that were adherent to the RGDS-and RGES-grafted surface. No significant changes in Bad or pBad were observed. mRNA analysis confirmed this observation, as no changes in message levels were seen for Bad at any time point (Fig. 4C). Finally, we examined the Bcl-2 protein (Fig. 5A) and mRNA (Fig. 5B) expression. No difference was observed in either Bcl-2 protein or mRNA levels on the RGDS or RGES surface.
Osteoblast Binding to RGD-grafted Surfaces Suppresses Apoptotic Sensitivity-How does integrin-mediated adhesion of osteoblasts to the biomimetic surface influence apoptotic sensitivity? To address this question, cells were grown on the RGDS substrate and treated with 0.1 and 0.5 M staurosporine; no significant cell killing was observed (Fig. 6A). In contrast, staurosporine killed osteoblasts maintained on the RGES surface in a dose-dependent manner. Fig. 6B demonstrates the effect of sur-

FIG. 4. Bad and pBad expression by RGDS-attached osteoblast-like cells by Western blot and RT-PCR analysis.
Cells were attached to RGDS and RGES surfaces for 0.5, 3, 24, and 72 h. Bad (A) and pBad (B) expression was determined by Western blot analysis. Bad transcript levels were evaluated by RT-PCR (C); GAPDH was used as a loading control. face chemistry on the apoptogenic activity of the Ca 2ϩ ⅐P i ion pair. Although 3 mM P i and 2.4 mM Ca 2ϩ did not kill osteoblasts grown on the RGDS surface, there was a significant increase in cell death on the RGES surface. A further increase in cell death was apparent when osteoblasts were treated with 3 mM P i and 2.9 mM Ca 2ϩ . Again, there was no significant increase in death among osteoblasts maintained on the RGDS surface. The effect of the NO donor sodium nitroprusside on osteoblast apoptosis was similar to that of the Ca 2ϩ ⅐P i ion pair and staurosporine. Fig. 6C shows that at concentrations of 0.1 and 0.5 mM, sodium nitroprusside killed 45 and 65% of osteoblasts cultured on the RGES surface, respectively. These concentrations of sodium nitroprusside failed to kill osteoblasts on the RGDS surface. In contrast to these agents, free RGDS (Fig. 6D) and serum-free medium (Fig.  6E) killed osteoblasts on both surfaces. The tetrapeptide RGDS killed osteoblasts that were anchored to both the RGDS-and RGES-grafted surface (Fig. 6D). A similar effect was seen when there was serum withdrawal (Fig. 6E). Thus, in the absence of serum, 45 and 60% of osteoblasts cultured on RGES and RGDS were killed, respectively.
Resistance to Apoptosis by RGD-anchored Osteoblasts Is Mediated by PI3K-To test the hypothesis that resistance to apoptosis was mediated by the PI3K pathway, RGDS-anchored cells were cultured in media supplemented with LY294002. When cells adherent to the RGDS surface were then treated with staurosporine, a profound increase in cell death was observed (Fig. 7A), comparable with the response of cells adherent to the RGES surface. A similar response was seen when the MC3T3-E1 cells were treated with the Ca 2ϩ ⅐P i ion pair (Fig.  7B). It was evident that when the PI3K survival pathway was blocked, the protection provided by the RGD surface was lost, and the cells became susceptible to apoptosis. In contrast, when osteoblast-like cells were attached to RGES and treated with the Ca 2ϩ ⅐P i ion pair, their apoptotic sensitivity was not affected by the PI3K inhibitor. DISCUSSION Results of the investigation indicated that attachment of osteoblasts to an RGDS-grafted surface significantly inhibited the activation of apoptosis. The RGDS-grafted surface caused up-regulation of ␣ v ␤ 3 integrin receptors as well as raised expression of activated FAK. Furthermore, this study revealed that there was a demonstrable increase in activated Akt fol-lowing culture on RGDS. Using the inhibitor LY294002, we confirmed that activation of the PI3K pathway was required for protection against apoptosis. These results suggest that activation of the PI3K pathway and the subsequent phosphorylation of FAK and Akt enhance the survival of osteoblasts challenged with exogenous apoptogens. Surprisingly, survival on the RGDS-grafted surface was evident when osteoblasts were treated with a subgroup of apoptogens. Thus, although serum starvation and free RGDS peptides cause a dramatic fall in the number of live osteoblasts, those agents that acted through the mitochondrial pathway (the Ca 2ϩ ⅐P i ion pair, staurosporine, and sodium nitroprusside) failed to induce cell death on the RGDS surface. Based on these findings, we conclude that RGDS-integrin binding enhanced PI3K activity and protected cells from mitochondria-dependent apoptosis.
To engage the osteoblast integrin receptor, we chemically engineered the silicone surface with RGDS peptides (9). We have reported recently that this technique introduces a very high (saturating) surface concentration of RGDS peptides (40). As the same chemistry was used to tether the control peptides, a similar high concentration of RGES peptides would be expected to be present on those membranes, and, as might be expected, the tethered RGDS and control peptides exhibited similar physical characteristics. Despite these chemical and physical similarities and in line with the studies reported by Hynes (2), the RGDS-engineered surface enhanced the immediate expression of the ␣ v ␤ 3 integrin. In contrast, there was delayed expression of these integrin receptors on cells attached to the RGES-grafted surface. Late receptor expression was not surprising, as adherent cells would be expected to secrete extracellular proteins that contain RGDS peptide motifs. Indeed, we found that by 72 h, osteoblasts attached to both the RGDS and RGES surfaces expressed the same integrin levels.
At early time periods, prior to synthesis of an extensive extracellular matrix, we noted that integrin-mediated attachment to RGDS influenced cell function. Thus, there was an immediate increase in FAK expression and a time-dependent increase in phosphorylated FAK. Although the role of the phosphorylated form of FAK is not known with certainty, it is likely that it activates PI3K, possibly through the Src homology domain 2, and promotes the transduction of survival signals (42)(43)(44). Indeed, at 3 days, we noted up-regulation and activation of Akt (phosphorylated Akt) in bone cells cultured on the RGDS-modified surfaces. However, in the same time period, there was also an increase in total Akt protein. This result suggests that although the effective concentration of activated Akt is elevated on the RGDS surface, the increase reflects a raised level of total protein expression rather than an increase in Akt phosphorylation activity. Irrespective of whether there is raised phosphorylation of Akt or just an increase in the total Akt level, the end result should be the same, enhanced osteoblast survival and protection from exogenous apoptogens.
Although Bad is a pro-apoptotic protein, its phosphorylation by Akt serves to decrease the impact of this key signaling molecule. Accordingly, we determined both Bad and pBad expression levels in the RGDS-tethered cells. We observed no differences in Bad expression between bone cells cultured on the RGDS-grafted and control surfaces, nor was there a significant increase in pBad. In concert with Bad, we evaluated the expression of the anti-apoptotic protein, Bcl-2. This protein has been shown to be up-regulated in Chinese hamster ovary cells by the ␣ 5 ␤ 1 integrin (45). However, in osteoblast-like cells, we failed to detect significant changes in the expression of Bcl-2 at either the message or protein levels. Although steady state values of pBad, Bad, and Bcl-2 were not determined, the result suggests that integrin binding and FAK and Akt activation did not influence the activities of key pro-and anti-apoptotic regulatory proteins concerned with survival of osteoblast-like cells adherent to the RGDS surface.
We used a number of different apoptogens to explore the relationship between cell survival and integrin attachment to tethered RGDS peptides. The apoptogen concentrations that were used in this study were shown previously to induce apoptosis in a number of cell types (13,27,(32)(33)(34)(35). By allowing the osteoblast-like cells to attach to the RGDS-modified surfaces over 72 h, we mitigated the immediate signaling effects of integrin binding and observed how attachment modulates the apoptotic response of the tethered differentiating osteoblasts. With respect to the Ca 2ϩ ⅐P i ion pair, we were able to obtain almost complete protection from apoptosis by allowing the cells to bind to the RGDS peptide. This finding was of considerable interest, as we had reported previously that when cells are FIG. 6. Apoptotic sensitivity of osteoblasts adherent to the RGDS-grafted surface. MC3T3-E1 cells were maintained on the RGDS and RGES surfaces for 72 h. Following incubation with apoptogens for 24 h, cell death was analyzed by the MTT assay. Cells were challenged with the following apoptogens: staurosporine (A), Ca 2ϩ ⅐P i ion pair (B), sodium nitroprusside (SNP) (C), free RGDS peptides (D), and serum starvation (E). Each bar represents the mean and standard error of the mean (n ϭ 3). *, p Ͻ 0.05 when compared with control (both untreated and RGES surface).
attached to tissue culture plastic, the ion pair promotes a mitochondrial membrane permeability transition and a rapid induction of osteoblast apoptosis (27). To confirm that the attachment peptide blocked apoptosis, we treated the osteoblasts with sodium nitroprusside. We had shown previously that this agent induces death via the mitochondrial pathway, and inhibitors of NO synthase block Ca 2ϩ ⅐Pi-mediated cell death (27). In concert with the ion pair, sodium nitroprusside failed to promote apoptosis of osteoblast-like cells bound to the RGDSgrafted surface. On the basis of these two studies, it is concluded that the surface maintains cell viability by blocking the mitochondrial membrane permeability transition and thereby preserving mitochondrial function.
Of the other apoptogen studied, it was noted that staurosporine failed to kill osteoblasts attached to the RGDS membrane. Although staurosporine is a nonspecific protein kinase inhibitor, it does cause a mitochondrial membrane permeability transition (47), and hence its mode of killing is similar to the ion pair (27) and sodium nitroprusside (41). Conversely, serum starvation or treatment with RGDS peptides promoted death of cells attached to both the RGDS-and RGES-grafted surfaces. Although the death pathway activated by free RGDS peptides is not known, it is likely that the peptides directly engage pro-caspase-3, and there is minimal mitochondrial involvement (46). The effects of serum starvation are more complex and probably encompass changes in both mitochondrial function and mitogenic signals (13). This dual effect would explain why some cells responded to this type of apoptogenic challenge.
To test the hypothesis that inhibition of apoptosis was dependent on the activity of PI3K, we treated adherent osteoblast-like cells with the inhibitor LY294002 and then challenged the cells with a number of agents discussed above. We noted that in the presence of the inhibitor, osteoblasts were no longer resistant to the Ca 2ϩ ⅐P i ion pair, and more than 60% of the cells were killed. Likewise, when treated with staurosporine and sodium nitroprusside, the tethered cells underwent apoptosis. The results of all of these studies using multiple apoptogens strongly indicate that when osteoblasts are attached to RGDS-tethered surfaces, there is inhibition of apoptotic responses transduced by mitochondria and enhanced survival activated by the PI3K pathway.
Finally, it is important to acknowledge that the extracellular matrix is a very complex, heterogeneous, and dynamic structure, and the number of receptors and receptor subtypes in bone cells is enormous. Nevertheless, the use of a single tethered peptide provides a unique approach to delineating which pathways are activated and permits dissection of the sequence of events that characterize outside-inside signal transduction. Because many of the attachment motifs are now known and linkage chemistries are available, it should be possible to evaluate the contribution of each of the proteins of the extracellular matrix to osteoblast differentiation, maturation, and function.
The use of such a system should enhance elucidation of crosstalk between ligands and growth factors and provide access to downstream effector pathways.