The Osteopontin-CD44 Survival Signal Involves Activation of the Phosphatidylinositol 3-Kinase/Akt Signaling Pathway*

We have recently demonstrated that the gene encoding the osteopontin (OPN) protein is activated both by interleukin-3 and granulocyte-macrophage colony-stim-ulating factor signaling pathways and that, through binding to the cell surface receptor CD44, OPN contributes to the survival activities of interleukin (IL)-3 and GM-CSF (Lin, Y.-H., Huang, C.-J., Chao, J.-R., Chen, S.-T., Lee, S.-F., Yen, J. J.-Y., and Yang-Yen, H.-F. (2000) Mol. Cell. Biol . 20, 2734–2742). In this report, we demonstrate that the CD44-binding domain of OPN involves a region containing amino acid residues from 121 to 140 and that both threonine and serine at positions 137 and 147, respectively, are essential for the survival stimulatory effect of OPN. Substitution of either residue with alanine results into a dominant negative mutant that overrides the survival effect of IL-3. Upon binding to the CD44 receptor, the wild-type OPN but not the inactive mutant induces activation of phosphatidylinositol 3-kinase and Akt. Last, we demonstrate that two waves of Akt activation are detected in IL-3-treated cells and that the survival promoting effect of OPN is mediated predominantly through the phosphatidylinositol 3-kinase/Akt signaling pathway. Together, our results suggest that a positive autoregulatory loop is involved in the survival pathway of IL-3. m TNE m M Tris, m M NaCl, and 1 m M EDTA). The complex was then mixed with (cid:3) l of 2 mg/ml PI (phosphati- dylinositol, in m M HEPES, 7.4, and incubated on ice for 10 min. Kinase reaction was initiated by the addition of 40 (cid:3) l of reaction buffer containing 10 (cid:3) Ci of [ (cid:4) - 32 P]ATP, 20 m M HEPES, pH 7.4, 20 (cid:3) M ATP, and 5 m M MgCl 2 . After incubation at 30 °C for 30 min, the reaction was stopped by the addition of 100 (cid:3) l of 1 N HCl and the mixture was extracted with 200 (cid:3) l of chloroform:methanol (1:1). The lipid was spot- ted onto potassium oxalate (1%)-treated TLC plate and the plate devel-oped by chromatography in CHCl 3 , MeOH, 4 M NH 4 OH (9:7:2). The radiolabeled lipid was visualized by autoradiography.

Osteopontin (OPN) is an acidic phosphoprotein that is secreted by osteoblasts, macrophages, cardiac fibroblasts, activated T cells, and many other cell types (16,17). OPN binds to a variety of cell surface receptors including integrins ␣ v ␤ 1 , ␣ v ␤ 3 , and ␣ v ␤ 5 (18 -20) and CD44 (21). Many cellular processes, including cell attachment, spreading and migration, homing of lymphocytes, and other hematopoietic cells and vascular remodeling are thought to involve OPN binding to its cell surface receptors (16,22). Using OPN-null mutant mice as a model system, OPN was demonstrated to have a role in the growth or survival of metastatic cells (23), tissue remodeling (24), and the type-1 immune response (25). However, the molecular mechanisms that underlie these activities of OPN remain largely unclear. By use of a PCR-based subtraction cloning approach, we have previously identified osteopontin to be a gene induced by the membrane-distal region of the ␤ subunit between amino acids 573 and 755 of the human GM-CSF receptor (26). We have also demonstrated that, in response to stimulation with IL-3 or GM-CSF, OPN is induced and released into the medium of cultured cells and that, through binding to the cell surface receptor CD44, it contributes to the survival activities of these two cytokines (26).
To gain further insight into OPN's mechanism of action in cytokine-triggered cell survival response, in this report, we have undertaken a structure-function study of the OPN molecule. We demonstrate that both threonine and serine at positions 137 and 147, respectively, are essential for the survival promoting effect of OPN and that substitution of either amino acid residue with alanine results in a dominant negative mutant that overrides the survival effect of IL-3 on IL-3-dependent cells. We further show that, via binding to the CD44 receptor, the wild-type OPN but not the inactive mutant activates the PI3K/Akt (PKB) kinase cascade and activation of the latter signaling pathway plays a major role in the survival promoting effect of OPN in IL-3-dependent cells.

EXPERIMENTAL PROCEDURES
Cells and Cell Lines-Ba/F3 is a murine IL-3-dependent pro-B-cell line and was maintained in RPMI 1640 supplemented with 10% fetal bovine serum and 1% conditioned medium (CM) from WEHI-3B cells as a source of IL-3. The ␣␤755 and ␣␤573 derivatives of Ba/F3 cells have been described previously (13) and stably overexpress the human GM-CSF (hGM-CSF) receptor ␣ chain in combination with a COOH-terminal truncation mutant of the ␤ chain that terminates at residue 755 and 573, respectively. The ␣␤755 cells behave like Ba/F3 cells overexpress-ing the wild-type hGM-CSF receptors (␣␤wt) and are capable of synthesizing OPN and fully resistant to apoptosis in medium containing either mIL-3 or hGM-CSF (13,26), whereas the ␣␤573 cells in growth medium containing hGM-CSF have lost the ability to synthesize OPN and undergo apoptosis (13,26). DNAkt-9h and -27h are two clones of Ba/F3 derivatives stably overexpressing the dominant negative mutant of Akt, AktK179M (27). Neo is a mixture of Ba/F3 cells transfected with an empty expression vector (pCDNA3, Invitrogen). Both DNAkt-9h and -27h cells are more susceptible to IL-3 removal-induced apoptosis than the parental cells (Ref. 28 and Fig. 8 below). Ba/F3Akt*-A1 and Ba/ F3Akt*-D2 are two clones of Ba/F3 derivatives which stably overexpress the constitutively active form of the Akt protein as previously described (28). Primary IL-3-dependent cells were isolated essentially as described by Rodriguez-Tarduchy et al. (29). Briefly, bone marrow was flushed from the femurs of BALB/c mice with RPMI 1640 and cultured for 48 h in the same medium containing 10% fetal bovine serum and 10% CM from WEHI-3B cells. Then, the cells remaining in suspension were separated from the adherent population and maintained in medium containing 20 units/ml murine IL-3 (mIL-3) (R & D Systems, Minneapolis, MN) for 10 to 14 days before analysis. For all experiments described in the text, unless otherwise indicated, the recombinant mIL-3 and hGM-CSF (Sandoz Pharma Ltd., Basel, Switzerland) were used at concentrations of 10 units/ml and 1 ng/ml, respectively. CHOP is a Chinese hamster ovary cell line stably expressing the polyoma virus large T antigen (30) and was kindly provided by James W. Dennis (Mt. Sinai Hospital, Toronto, Canada). CHOP cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum.
CM Containing OPN-CHOP cells were transiently transfected with various OPN expression vectors by liposome-mediated gene transfer as previously described (26). In brief, the pre-formed DNA-lipid complex (12 g of DNA mixed with 25 l of LipofectAMINE (Life Technologies, Inc.)) was added to 10 6 cells cultured in a volume of 10 ml that had been seeded 1 day earlier. After incubation for 4 h in serum-free medium, the transfected cells were incubated for 24 h in regular growth medium. The latter was then removed, filtered through a 0.2-m pore size filter, and used as CM for the various assays as described. Immunoblotting assay with OPN-specific antibody (R & D Systems) indicated that under the same transfection conditions, except for the S156A mutant, the wild-type OPN and various other mutants present in the CM were all expressed at a very similar level (see Fig. 1B, below). With antibody immunodepletion assay, we have previously demonstrated that the survival promoting activity of CM from cells transfected with the wild type vector was entirely due to the presence of OPN in the CM (26).
Survival Promoting Assay of OPN-The survival stimulatory effect of OPN was analyzed essentially as previously described (26). Briefly, ␣␤573 cells which lost the ability to synthesize OPN in growth medium containing hGM-CSF were cultured in the presence or absence of CM (30%, v/v) from CHOP cells transiently transfected with wild type or mutant OPN expression vectors. At various times after culturing, the number of viable or apoptotic cells was determined by the trypan blue exclusion method or with an ELISA cell death detection kit (Roche Molecular Biochemicals), respectively. To examine the dominant negative effect of the mutant protein, CM containing wild-type OPN was first mixed with CM (1:1) from CHOP cells transfected with mutant OPN or GFP expression vector prior to being added to the culture medium to a final concentration of 50% (v/v).
Immunoblot Analysis-Cell lysates or CM from CHOP cells transiently transfected with various OPN expression vectors were subjected to immunoblot analysis as previously described (13). In brief, 100 g of proteins were resolved by SDS-polyacrylamide gel electrophoresis, transferred to a polyvinylidene difluoride membrane (Millipore), and probed with antibodies to OPN (R & D Systems) or to HA tag (Roche Molecular Biochemicals, Mannheim, Germany). Immune complexes were detected with horseradish peroxidase-conjugated goat antibodies to mouse IgG and an ECL (enhanced chemiluminescence) kit (Amersham Pharmacia Biotech). For detection of the activation status of the AKT (PKB) protein, cells to be analyzed were lysed in the lysis buffer for the PI 3-kinase assay (see below) and equal amounts of cell lysates were analyzed by immunoblotting using antibody specific to the active form of AKT (phospho-Akt Ser 473 ) or to all forms of Akt (both from Cell Signaling Technology).
PI 3-Kinase Assay-Cells (10 7 ) to be analyzed were washed with ice-cold phosphate-buffered saline twice and lysed on ice with 1 ml of lysis buffer (137 mM NaCl, 2.7 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 1% Nonidet P-40, 10% glycerol, 1 mg/ml bovine serum albumin, 20 mM Tris, pH 8.0, and 2 mM orthovanadate). Cell lysates were immunoprecipitated with 1 g of anti-phosphotyrosine antibody (UBI, clone 4G10) overnight at 4°C. The immune complex after brought down with 50 l of protein A-Sepharose was washed three times with lysis buffer, twice with LiCl buffer (0.5% LiCl, 100 mM Tris, pH 7.6), and twice with TNE buffer (10 mM Tris, pH 7.6, 100 mM NaCl, and 1 mM EDTA). The immune complex was then mixed with 10 l of 2 mg/ml PI (phosphatidylinositol, Sigma) in 20 mM HEPES, pH 7.4, and incubated on ice for 10 min. Kinase reaction was initiated by the addition of 40 l of reaction buffer containing 10 Ci of [␥-32 P]ATP, 20 mM HEPES, pH 7.4, 20 M ATP, and 5 mM MgCl 2 . After incubation at 30°C for 30 min, the reaction was stopped by the addition of 100 l of 1 N HCl and the mixture was extracted with 200 l of chloroform:methanol (1:1). The lipid was spotted onto potassium oxalate (1%)-treated TLC plate and the plate developed by chromatography in CHCl 3 , MeOH, 4 M NH 4 OH (9:7:2). The radiolabeled lipid was visualized by autoradiography.

RESULTS
Thr 137 and Ser 147 Are Both Essential for the Survival Promoting Activity of OPN-To gain further insight into the mechanism of action of the OPN-CD44 signaling in the survival pathway of IL-3, we first undertook a structure-function analysis of the OPN molecule. We have previously reported that mutant OPN lacking the internal region between amino acid residues 79 and 140 (BOPN ⌬79 -140) is completely devoid of the survival stimulation activity (26), suggesting that this region of the OPN molecule may contain a domain(s) important for activation of the CD44 receptor. To examine this possibility, the peptide competition experiment was first carried out. In this experiment, the standard assay for demonstrating OPNs survival promoting activity as described under "Experimental Procedures" was performed in the presence of various amounts of synthetic peptides covering amino acid residues 81-100, 101-120, 121-140, or 143-147 (GRGDS) of the mouse OPN molecule. As illustrated in Table I, among different peptides tested, the peptide covering residues 121 to 140 inhibited wildtype OPNs activity most effectively (ID 50 , 20 M), suggesting that this region of OPN is involved in binding to the CD44 receptor.
A sequence alignment of OPN from mouse, rat, and human a Dose required to inhibit 50% of OPNs growth promoting activity in the standard assay as described under "Experimental Procedures." b Synthetic peptides covering amino acid residues 81-100, 101-120, or 121-140 of the mouse OPN. species (31)(32)(33) revealed that 7 amino acid residues in the region covering positions 121-140 are highly conserved (Fig.  1A, underlined residues). Of these 7 residues, three threonines at positions 123, 132, and 137 were each mutagenized into alanine by the standard site-directed mutagenesis approach and the survival promoting activities of these mutants were compared with that of the wild-type protein. Fig. 1B shows that mutation at position 123 or 132 did not significantly affect OPNs activity, whereas mutation at position 137 completely abolished the survival promoting activity of OPN. As a comparison, mutants with mutations that fall outside of the putative receptor-binding domain were also generated. The serine residues at positions 147 and 156 were randomly selected and were each mutagenized into alanine. Unexpectedly, the S147A mutant completely lost activity, whereas the S156A still manifested full activity even though its expression level was much lower than that of the wild-type protein (Fig. 1B, lower panel). Taken together, these results indicate that Thr 137 and Ser 147 are both essential for the survival promoting activity of OPN. S147A and T137A Are Both Dominant Negative Mutants of OPN-As shown in Table I, the GRGDS peptide did not significantly inhibit OPNs activity until a much higher dose (ID 50 , 180 M) was applied in the assay. This result suggests that the GRGDS residues at positions 143-147 may not be involved in direct contact with the CD44 receptor. This result further suggests that the inactive mutant S147A may still retain a full receptor binding ability and may even exert a dominant negative effect on the wild-type molecule. To test this possibility, conditioned medium from CHOP cells transiently transfected with the S147A mutant was added to the standard OPN activity assay as described in Fig. 1. Fig. 2 shows that indeed the S147A mutant completely abolished wild-type OPNs activity (panel A). Unexpectedly, under the same conditions, the other inactive mutant T137A whose mutation falls in the putative receptor-binding domain also exerted a dominant negative effect on the wild-type molecule, albeit its inhibitory effect was weaker than that of S147A (Fig. 2, panel B; see "Discussion").
Blocking the OPN-CD44 pathway by neutralizing antibodies to OPN or CD44 inhibits the survival promoting activity of IL-3 (26). We next examined whether these two dominant negative mutants could achieve a similar blocking effect. Fig. 3 shows that the S147A mutant effectively overrode the survival activity of IL-3 both in ␣␤573 cell line (panel A) and in primary IL-3-dependent cells (panel B). Consistent with results shown above, the T137A mutant that had a weaker dominant negative effect on the wild-type protein was less effective in blocking the survival activity of IL-3 (Fig. 3, panels A and B).
Involvement of the PI3K/Akt Pathway in the Survival Promoting Activity of OPN-We have previously demonstrated that the PI3K/Akt (PKB) pathway plays an important role in the survival activity of IL-3 in Ba/F3 cells (28). We next tested whether OPN also mediated its survival promoting activity through activation of this survival pathway. Out of five independent experiments, we consistently observed that, under our experimental conditions (see "Experimental Procedures"), the wild-type OPN but not the S147A mutant stimulated activation of PI3K (Fig. 4, panel A) and Akt (Fig. 4, panel B) with a strength that was ϳ40% of that achieved by IL-3. Furthermore, OPN induced activation of the PI3K/Akt pathway in Ba/F3 cells was mediated through the CD44 receptor, because this activation effect was inhibited by pretreatment of cells with neutralizing antibody to CD44 (clone KM114, Fig. 5) but not to integrin ␣ V subunit (␣ V , Fig. 5).
Given the fact that the OPN gene is activated within 3 h treatment of cells with IL-3 (26) and that IL-3 and OPN both activate the PI3K/Akt pathway (Fig. 4), we were curious to test whether two waves of Akt activation could be detected in cells stimulated with IL-3. In this experiment, both ␣␤755 and ␣␤573 cells were tested, as they all produce OPN in response to IL-3 stimulation (26). As shown in Fig. 6A, following IL-3 treatment, in both cell lines Akt activation was first observed at the 5-30-min time point and the second wave of activation appeared around the 3-h time point where the OPN gene was markedly activated by the IL-3 signaling pathway (26). In contrast, upon stimulation with hGM-CSF, the two-wave activation kinetics of Akt was only observed in OPN-producing ␣␤755 cells, but not in ␣␤573 cells that, under such conditions, do not produce OPN (Fig. 6B).
We next examined whether activation of the PI3K/Akt pathway played a role in the survival activity of OPN. To address FIG. 4. Wild-type but not mutant OPN stimulates activation of PI3-K and Akt. A, ␣␤573 cells were starved in hGM-CSF containing medium for 8 h prior to stimulation with IL-3, wild-type, or mutant OPN (S147A) for 5 or 10 min. After treatments, cells were lysed with lysis buffer and lysates were immunoprecipitated with anti-phosphotyrosine antibody. The PI 3-kinase activity in the immune complex was measured by in vitro kinase assay as described under "Experimental Procedures." The relative activity under each treatment was shown on the bottom of the figure. B, ␣␤573 cells were treated as described in A. At various times after each treatment, cells were lysed and equal amounts of cell lysates from each group were analyzed by immunoblotting using antibodies specific to active form (pAkt) or to all forms of Akt (Akt).

FIG. 3. Both S147A and T137A mutants override the survival activity of IL-3 in ␣␤573 (A) or in primary IL-3-dependent cells (B).
Cells were cultured for 24 h in the presence of the indicated reagents, after which the number of apoptotic cells was quantified with an ELISA cell death detection kit. The wild-type or mutant OPN proteins used here were CM (50%) from CHOP cells transiently transfected with respective expression vectors. For comparison, cells were also cultured in cytokine-free medium (Free) or in growth medium supplemented with 50% CM from CHOP cells transiently transfected with the green fluorescent protein expression vector (GFP). Data are mean Ϯ S.D. of duplicates from experiments that were repeated three times with similar results. The apoptotic effect elicited by S147A or T137A mutant is statistically significant as compared with that observed in cells treated with CM from CHOP cells transfected with the control vector (GFP). *, p Ͻ 0.0001; #, p Յ 0.001. this issue, the survival promoting assay of OPN was carried out in cells pretreated with PI3K inhibitors, wortmannin or LY294002, or in Ba/F3 cells stably overexpressing the dominant negative mutant of Akt (DNAkt). As shown in Fig. 7, both wortmannin and LY294002 effectively (Ͼ90%) blocked OPNs anti-apoptotic activity, indicating that a majority of OPNs survival stimulatory effect is mediated through activation of PI3K. On the other hand, the survival stimulatory effect of OPN was only partially compromised in Ba/F3 derivatives stably overexpressing the dominant negative mutant of Akt (Fig. 8A). Under our experimental conditions, whereas 60 -70% inhibition of cell death was achieved by OPN in control cells (Neo, Fig. 8B, right panel), ϳ30 -40% inhibition was observed in cells overexpressing DNAkt (clones 9h and 27h, Fig. 8B, right panel).
Considering the fact that the S147A mutant effectively (70 -90%) inhibited the anti-apoptotic activity of IL-3 (Fig. 3) and that the latter activity was mediated predominantly (Ͼ90%) through the PI3K/Akt pathway in the Ba/F3 cell system (28), the partially inhibitory effect (30 -40%) of DNAkt observed in the results shown in Fig. 8 was somewhat unexpected. To clarify this issue, we next examined whether overexpression of the constitutively active form of Akt (Akt*) would significantly reverse the inhibitory effect of the S147A mutant on the survival activity of IL-3. Fig. 9 shows that a majority of the inhibitory effect of S147A on IL-3 was indeed reversed in cells stably overexpressing Akt* (clones A1 and D2). These results suggest that the partially inhibitory effect of DNAkt on OPN in the stable lines tested (Fig. 8) is likely due to incomplete inhibition of the endogenous Akt activity. Taken together, our results demonstrate that activation of the PI3K/Akt signaling pathway plays a major role in OPNs survival promoting activity in IL-3-dependent cells. Whether there is any other minor pathway(s) involved in this case remains to be determined. DISCUSSION The PI3K/Akt (PKB) signaling pathway plays an important role in the survival response induced by a variety of growth factors, matrix adhesion, and oncogene transformation (27, 28, 34 -39). In this report, we demonstrate that the survival promoting effect of OPN in IL-3-dependent cells also involves activation of the PI3K/Akt signaling pathway. OPN protects endothelial cells from serum withdrawal-induced apoptosis via interaction with integrin ␣ V ␤ 3 and activation of nuclear fac- FIG. 5. OPN activates PI3K and Akt through CD44 but not integrin ␣ V -containing receptors. ␣␤573 cells were treated essentially as described in the legend to Fig. 4 except that specific antibodies as indicated on top of the figure were added to cells 1 h prior to stimulation with OPN for 10 min. After treatments, cells were lysed and the cell lysates were analyzed for PI3K activity (panel A) or for the activation status of the Akt kinase (panel B) as described in the legend to Fig. 4. KM114 and ␣ V are neutralizing antibodies to CD44 and integrin ␣ V subunit, respectively (both from Pharmingen). "r" and "h" refer to isotype-matched control IgG for KM114 and ␣ V , respectively. tor-B (NF-B) (40). In contrast, the anti-apoptotic activity of OPN in IL-3-dependent cells is mediated predominantly through interaction with the CD44 receptor (26) and activation of the PI3K/Akt signaling pathway. These results suggest that OPN exert its survival activity in a cell type-specific mechanism.
The PI3K/Akt pathway is activated by IL-3 (39 and this study) as well as by OPN, a downstream effector gene product of IL-3 (26). Interestingly, the activation kinetics of PI3-K and Akt by OPN (within 5-10 min stimulation) is very similar to that stimulated by IL-3. The weaker activation effect of OPN (Fig. 4) could be either due to an intrinsic property of the OPN-CD44 interaction or due to the limitation of our experimental conditions where the maximal effect of OPN might not yet be achieved with the amounts of conditioned medium (50%) used in the assay. Given that IL-3 induction of OPN expression occurs at a much later time point (ϳ3 h, see Ref. 26), it is unlikely that IL-3-triggered early activation of the PI3K/Akt kinase cascade is mediated through induction of OPN expres-  9. Constitutively active Akt reversed the inhibitory effect of S147A on the anti-apoptotic activity of IL-3. Control (Neo) or Ba/F3 cells stably overexpressing the constitutively active form of Akt (clones Akt*-A1 and -D2) were cultured in the presence of the indicated reagents as described in the legend to Fig. 3. Twenty-four hours after each treatment, the number of apoptotic cells in each case was quantified with an ELISA cell death detection kit. Data are mean Ϯ S.D. of duplicates from experiments that were repeated three times with similar results. The inhibitory effect of Akt* on S147A is statistically significant as compared with that of vector control. **, p Ͻ 0.001; #, p Ͻ 0.00001. sion. In fact, following cytokine treatment of cells, two waves of Akt activation were observed in cells under conditions where OPN was produced (e.g. ␣␤755 cells stimulated with IL-3 or hGM-CSF, and ␣␤573 cells stimulated with IL-3), but not in the same cells when OPN expression was not induced (e.g. ␣␤573 cells stimulated with hGM-CSF). Our results strongly suggest that the early wave of Akt activation is mediated through the interaction between IL-3 and its receptor, whereas the second wave of Akt activation is mediated mainly through the OPN-CD44 pathway. Activation of the PI3K/Akt pathway is crucial to the survival activity of IL-3 (28,39). Our results suggest that IL-3-dependent cells (at least in the Ba/F3 cell system) may have evolved to adopt a positive autoregulatory mechanism by which the PI3K/Akt survival pathway can be first activated from the IL-3 receptor itself and later from its downstream effector gene product such as OPN (Fig. 10). In this way, the crucial survival signal can thus be effectively activated.
Mutational studies revealed that both Thr 137 and Ser 147 residues are critical for OPN to assume a conformation that can activate the CD44 receptor. However, it is not clear why S147A has a prominent dominant negative effect, whereas the T137A mutant only exerts a partially inhibitory effect on OPN. One possibility is that these two mutants may differ in their ability to compete with the wild-type protein for binding to the CD44 receptor. However, more experiments including a detailed structural analysis of the wild-type OPN molecule are required to reveal the exact mechanism(s) responsible for differential properties of these two mutants. On the other hand, several post-translational variants of OPN have been identified (17) and the phosphorylation status affects OPNs ability to interact with the integrin receptor (25). It would be interesting to determine whether post-translational modification of either Thr 137 or Ser 147 residue (or both) is required for OPNs antiapoptotic activity.
OPN expression is increased in the blood of patients with metastatic disease (41). Whereas overexpression of OPN in benign cells leads to increased metastasis (42), reduced production of OPN by an antisense approach inhibits the tumorigenicity of transformed cell lines (43)(44)(45). Our finding that OPN has a survival promoting activity (26) further helps us interpret why a large variety of malignant cells have evolved to produce an increased level of OPN and have a growth advantage in vitro and in vivo. Moreover, the identification of S147A to be an effective dominant negative mutant of OPN implies that S147A may have a useful application in the therapy of certain types of metastatic tumor that involve deregulated expression of OPN.