INTRODUCTION

The Rous sarcoma virus (RSV)¹ is a potent tumor inducer in host animals. A 60-kDa tyrosine kinase, v-Src, is essential for the transforming activity of RSV (1). Transformation by v-Src and/or transient expression of v-Src induces a number of genes, such as 9E3/CEF-4, collagenase, transforming growth factor , c-fos, junB, and matrix metalloproteinase-9, proposed to play a role in tumor growth and metastasis (2, 3, 4, 5, 6, 7).

Osteopontin is a multifunctional extracellular matrix protein thought to be involved in cell adhesion and signaling (8). Osteopontin expression is increased by cell transformation (9, 10, 11, 12, 13, 14) and is stimulated in mouse epidermis by tumor promoters (15, 16). Osteopontin gene expression is also stimulated by growth factors, such as transforming growth factor  and epidermal growth factor (17, 18). Moreover, suppression of osteopontin expression by antisense RNA prevented malignant transformation, suggesting that osteopontin plays a critical role in tumorigenesis (19, 20, 21). Recently, it was reported that osteopontin expression is diminished in various tissues of c-src-deficient mice (22), suggesting that osteopontin expression is stimulated by c-Src. Osteopontin is one of the genes highly expressed in osteoclasts (23, 24), and c-src deficiency causes inactivation of osteoclasts and osteopetrosis (25, 26). Interestingly, epidermal growth factor and 12-O-tetradecanoylphorbol-13-acetate stimulation of osteopontin expression in c-src-deficient cells is similar to that in wild type cells, suggesting that c-Src stimulates osteopontin expression via a different pathway, which seems to be physiologically important (22).

In this report, we investigated the effect of v-Src on osteopontin promoter activity and tried to elucidate the mechanism of its stimulatory effects by identifying the transcription factor(s) involved.

MATERIALS AND METHODS

The HT1080 human fibrosarcoma cell line, which was used in the previous study to investigate v-Src stimulation of matrix metalloproteinase-9 promoter (7), was obtained from American Type Culture Collection (ATCC, Rockville, MD) and cultured in Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.) supplemented with 10% fetal bovine serum (JRH Biochemicals, Lenexa, KS). NIH 3T3 cells transformed by v-src (27) and their parental NIH 3T3 cell line were kindly provided by Drs. N. E. Kohl and J. B. Gibbs (Merck Research Laboratories) and maintained in DMEM supplemented with 10% fetal bovine serum. Chemicals were obtained from Sigma. Oligonucleotides were obtained from Life Technologies, Inc.

Plasmid pEcoRIB containing RSV src region (28, 29) was obtained from ATCC. An EcoRI-HindIII fragment containing the v-src-coding region was excised from pEcoRIB and cloned into pcDNA3 vector (Invitrogen, San Diego, CA) under the control of cytomegalovirus promoter. Deletion mutants of osteopontin promoter were constructed, using polymerase chain reaction (PCR) with a PstI fragment containing the 910 to +90 region of the mouse osteopontin gene as a template (30), and cloned upstream of the firefly luciferase gene of the pXP2 vector (31), generating op-910, op-79, op-72, op-65, op-55, and op-37 which contain the 910, 79, 72, 65, 55, and 37 to +79 region of the mouse osteopontin promoter, respectively. Mutations were introduced by PCR using primers containing mutations as shown in Fig. 1, and sequences were confirmed by sequencing using a Sequenase version 2.0 sequencing kit (U. S. Biochemical Corp.). A fragment containing 37 to +52 of herpes simplex virus (HSV) tk minimal promoter was prepared from pBLCAT2 vector (32) by PCR and inserted between the BglII and HindIII sites of pGL2-Basic vector (Promega, Madison, WI), generating tk-LUC. Two double strand synthetic oligonucleotides corresponding to 72 to 56 and 72 to 38 of mouse osteopontin promoter were prepared and inserted upstream of HSV tk minimal promoter of tk-LUC plasmid in forward orientations, generating reporter plasmids as follows; REx3-tk and REx1-tk contain three and one copies of 72 to 38 fragments, respectively, upstream of tk minimal promoter. OP6/7-tk contains one copy of the 72 to 56 fragment upstream of the tk minimal promoter.

Cells were transfected with calcium phosphate-DNA coprecipitation method as described previously (33). HT1080 cells were plated onto six-well multiwell dishes (Costar, Cambridge, MA) at a density of 3 × 10⁵ cells/well. Eighteen hours later, cells were transfected with reporter plasmids (2 µg/well) together with the v-Src expression plasmid or pcDNA3 plasmid (2 µg/well). To control for transfection efficiency and nonspecific effects of v-Src coexpression, the promoter activity of each promoter-luciferase reporter construct with or without v-Src coexpression was calculated and expressed relative to tk-LUC promoter activity, transfected into parallel separate cells with or without v-Src coexpression in the same experiments. HT1080 cells were harvested and used for luciferase assays 48 h after transfection. For NIH 3T3 cells, medium was changed to serum-free DMEM 5 h post transfection, and a cell lysate was prepared 43 h later. Luciferase assays were performed as described previously (34).

Preparation of nuclear extracts and gel shift assays were performed as described previously (34). Oligonucleotide competitors for AP-1, AP-2, CTF/NF-1, NF-B, and Sp1 were obtained from Stratagene (La Jolla, CA). An oligonucleotide competitor for C/EBP was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-CBF-A antibody (35) was kindly provided by Drs. S. Maity and B. de Crombrugghe (University of Texas). For the antibody shift experiments, 0.5-1 µl of anti-CBF antibody or rabbit preimmune serum was incubated with the binding reaction mixture for 15 min before the probes were added. The reaction mixture was loaded onto a 4-20% gradient polyacrylamide precast gel (NOVEX, San Diego, CA).

Methylation interference assay was performed as described previously (36). A double strand oligonucleotide corresponding to the 72 to -38 sequence of the osteopontin promoter was labeled by T4 polynucleotide kinase (New England Biolabs, Beverly, MA) and [-³²P]ATP (Amersham Corp.) at either end. Labeled oligonucleotides were partially methylated by dimethyl sulfate, incubated with nuclear extract of HT1080 cells, and subjected to gel mobility shift assays. Retarded and free probes were purified from the gel and treated with piperidine. The interrupted residues were visualized by gel electrophoresis on an 8% sequence gel.

When v-src-transformed NIH 3T3 cells and parental NIH 3T3 cells became semiconfluent, medium was changed to serum-free DMEM, and cultured for additional 48 h. Total RNA was prepared as described previously (37). Ten micrograms of total RNA was separated by agarose gel electrophoresis and blotted onto nylon membrane (Hybond-N, Amersham). The radioactive probe was prepared from mouse osteopontin cDNA and human -actin gene (38), and Northern blotting was performed as described previously (39).

RESULTS AND DISCUSSION

v-Src Stimulates Mouse Osteopontin Promoter Activity via a Response Element Located between 65 and 38

Cotransfection of op-910 and op-79 with a v-Src expression plasmid into HT1080 cells increased the expression of both constructs by approximately 5-6-fold (Fig. 2). Deletion up to 66 had no effect on the v-Src stimulation (Fig. 2). This suggests that the distal AP-1-like sequence in this fragment does not play a major role in v-Src stimulation. Removal of the 65 to 56 segment significantly decreased both v-Src stimulation and basal promoter activity. It should be noted that constructs op-55 and op-37 lost v-Src stimulation but still retained significant promoter activity as high as that of the tk-minimal promoter used as a control. Next, heterologous promoter constructs were made, in which the 72 to 56 or 72 to 38 fragment of the osteopontin promoter was inserted upstream of the HSV tk minimal promoter of tk-LUC. As shown in Fig. 3, the 72 to 38 fragment increased basal activity and conferred v-Src response to a tk minimal promoter, whereas the 72 to 56 fragment moderately increased basal activity but did not confer v-Src response. These observations suggested that the v-Src response element (v-SrcRE) is located between 65 and 38.

To characterize the nuclear factor(s) which bind to v-SrcRE, gel shift assays were performed using the double strand oligonucleotide corresponding to 72 to 38 of the mouse osteopontin promoter (Fig. 4). One major retarded band was observed, and its intensity was decreased by the addition of increasing amounts of cold oligonucleotide (Fig. 4, lanes 1-4). A 200-400-fold excess of cold oligonucleotide which contains the known consensus sequence for one of the following, AP-1, AP-2, Sp-1, NF-B, CTF/NF-1, or C/EBP, did not compete with the binding of the 72 to 38 fragment (data not shown). To localize the protein binding site, a methylation interference assay was performed (Fig. 5). Methylation of residues 57, 54, 53, 50, or 49 interfered with protein binding to the 72 to 38 fragment. This result indicates that the 57 to 49 segment, which contains an inverted CCAAT box, is important for binding. This was further confirmed by the fact that a fragment corresponding to the 72 to 38 sequence in which the CCAAT box was mutated (57 CCTGATTGG 49 to CTGATT) did not compete with the wild type 72 to 38 fragment for protein binding (Fig. 4, lane 8).

Specific protein binding to the 72 to 38 fragment of mouse osteopontin promoter.

To confirm that the CCAAT box plays a role in v-Src stimulation of the osteopontin promoter, mutations were introduced into the op-72 and op-79 constructs, as shown in Fig. 1. Mutation of AP-1-like sequence of op-79 had no effect on v-Src stimulation (data not shown). That is consistent with the fact that op-65 which lacks the AP-1-like sequence still retained v-Src stimulation (Fig. 2). Mutation of the CCAAT-box of op-72 significantly diminished both basal promoter activity and v-Src stimulation, whereas mutation of the adjacent GC-rich sequence had little effect on v-Src stimulation (Fig. 6).

We next searched for the possible type of CCAAT-binding factor involved. Oligonucleotides corresponding to consensus sequences for CTF/NF-1 or C/EBP did not compete with the 72 to 38 fragment for specific protein binding (data not shown). However, there is another class of CCAAT box-binding factors, called NF-Y or CBF, originally identified as factors binding to the major histocompatibilty complex class II and to the 2 chain of type I collagen promoters, respectively (40, 41). Biochemical characterization of the protein complex binding to the two CCAAT boxes of RSV LTR suggested that it is identical or closely related to CBF or NF-Y (42). A fragment corresponding to the 155 to 122 sequence of RSV LTR, containing the distal CCAAT box, partially competed with the 72 to 38 fragment of mouse osteopontin promoter in the gel shift assay (Fig. 4, lanes 5-7). A fragment corresponding to the 72 to 56 sequence of the osteopontin promoter did not compete by itself with the 72 to 38 fragment; however, the combination of the RSV CCAAT fragment and the osteopontin 72 to 56 fragment competed with the 72 to 38 fragment as effectively as the 72 to 38 fragment itself (data not shown). On the other hand, excess cold osteopontin fragment 72 to 38, as well as the RSV CCAAT fragment itself, displaced to the same extent binding activity to the RSV CCAAT fragment (Fig. 4, lanes 9-15). These results indicate that there are two factors binding to the osteopontin 72 to 38 fragment, and they migrate to similar positions in the gel shift assays. One seems to bind to the inverted CCAAT box, and the other to the 72 to 56 segment. The binding activity to the 72 to 56 fragment in HT1080 nuclear extract was also confirmed by gel shift assay (data not shown). Interestingly, in Fig. 5, methylation of residues 65, 63, 62, 60, or 59 increased the binding, suggesting that these residues may also affect the protein-DNA interaction. However, heterologous promoter analysis (Fig. 3) and mutation analysis (Fig. 6) strongly suggested that the inverted CCAAT box but not the 72 to 56 segment plays a critical role in v-Src stimulation of the osteopontin promoter. Therefore, we concentrated on the CCAAT box-binding factor.

To further characterize this CCAAT box-binding factor, we used anti-CBF-A antibody which can supershift the CBF-complex in gel shift assays (35). As shown in Fig. 7, the anti-CBF antibody supershifted both the complex formed with the osteopontin 72 to 38 fragment and with the RSV CCAAT fragment. These findings suggest that the factor binding to the osteopontin 72 to 38 fragment is either the CBF, which binds to the CCAAT box of RSV LTR, or is closely related to it.

Dutta et al. (43) reported that two CCAAT boxes in RSV LTR play an important role in serum-dependent transcription. They also found that v-Src stimulates binding activity to the CCAAT box by using a temperature-sensitive mutant of v-Src. Our findings suggest that a similar mechanism is involved in the stimulation of the osteopontin promoter by v-Src. This mechanism may play a role in the transformation of cells by RSV and the participation of osteopontin in this process.

To examine if v-Src stimulation of the osteopontin promoter occurs in situ, we compared osteopontin mRNA levels in v-src-transformed NIH 3T3 and in parental cells. In v-src-transformed NIH 3T3, osteopontin mRNA levels were significantly higher than in the parental NIH 3T3 (Fig. 8). The relative promoter activity of op-72 to op-37 was also significantly higher in v-src-transformed cells than in the parental cells (Fig. 9). Moreover, the stimulation was not observed when the CCAAT box was mutated (Fig. 9). These results suggest that, upon transformation by v-src, osteopontin expression increases, and at least part of the stimulation is due to the increased promoter activity mediated by the v-SrcRE that we identified. In a gel shift assay, we could not see any difference in the protein-DNA complex formed with the 72 to 38 fragment using nuclear extract prepared from v-src transformed and wild type NIH 3T3 cells (data not shown), suggesting that the effect may not be due to the abundance of CBF.

It has been reported that a dominant negative Ras mutant or a farnesyltransferase inhibitor, which interferes with Ras activity, blocked oncogenic action of v-Src, suggesting that at least part of v-Src action is mediated by Ras (27, 44). Recently, a ras-activated enhancer was identified in the mouse osteopontin promoter, which interacts with an ETS-related transcription factor (45). The response element, GGAGGCAGG, was located at 725 to 717, which is far upstream of the v-SrcRE identified here. However, that study also suggested that the 88 to +79 region contains regulatory elements which contribute to the increased expression in ras-transformed cells. Our findings are consistent with those results and suggest that several enhancer elements may contribute to the increased expression of osteopontin upon transformation.

The protooncogene of v-src, c-src, plays an important role in cell signaling (46). Gene knockout of c-src resulted in osteopetrosis in mice (25). In c-src-deficient mice, osteoclasts are inactivated and do not resorb bone (26). Recently, it was reported that osteopontin expression was diminished in c-src-deficient cells (22). Our results complement these findings, although c-Src and v-Src may have different modes of action. Osteopontin is highly expressed in osteoblasts and osteoclasts in bone tissue (23, 24). The mechanism by which c-src deficiency causes osteoclast inactivation has not been elucidated. A reduction in osteopontin expression in osteoblasts and/or osteoclasts of c-src-deficient mice could play a role in osteoclast inactivation. Thus, stimulation of the osteopontin promoter by Src may have a function not only in tumor cells but also in the physiological function of normal tissues.