Ets-1 and Runx2 Regulate Transcription of a Metastatic Gene, Osteopontin, in Murine Colorectal Cancer Cells*

Osteopontin (OPN) is a sialic acid-rich phosphoprotein secreted by a wide variety of cancers. We have shown previously that OPN is necessary for mediating hepatic metastasis in CT26 colorectal cancer cells. Although a variety of stimuli can induce OPN, the molecular mechanisms that regulate OPN gene transcription in colorectal cancer are unknown. We hypothesized that cis- and trans-regulatory elements determine OPN transcription in CT26 cells. OPN transcription was analyzed in CT26 cancer cells and compared with YAMC (young adult mouse colon) epithelial cells. Clonal deletion analysis of OPN promoter-luciferase constructs identified cis-regulatory regions. A specific promoter region, nucleotide (nt) –107 to –174, demonstrated a >8.0-fold increase in luciferase activity in CT26 compared with YAMC. Gel-shift assays sublocalized two cis-regulatory regions, nt –101 to –123 and nt –121 to –145, which specifically bind CT26 nuclear proteins. Competition with unlabeled mutant oligonucleotides revealed that the regions nt –115 to –118 and nt –129 to –134 were essential for protein binding. Subsequent supershift and chromatin immunoprecipitation assays confirmed the corresponding nuclear proteins to be Ets-1 and Runx2. Functional relevance was demonstrated through mutations in the Ets-1 and Runx2 consensus binding sites resulting in >60% decrease in OPN transcription. Ets-1 and Runx2 protein expression in CT26 was ablated using antisense oligonucleotides and resulted in a >7-fold decrease in OPN protein expression. Ets-1 and Runx2 are critical transcriptional regulators of OPN expression in CT26 colorectal cancer cells. Suppression of these transcription factors results in significant down-regulation of the OPN metastasis protein.

OPN, an extracellular matrix protein secreted by a wide variety of cancers, functionally enables tumor progression (3)(4)(5)(6)(7)(8). The secreted phosphoprotein binds the ␣ v ␤ integrin and CD44 families of receptors to propagate cellular signals (8 -14). In colorectal cancer, gene profiling studies have identified a positive correlation between advanced or metastatic colon tumors and abundant OPN expression (15,16). We have previously shown that stable, RNAi-mediated down-regulation of OPN expression in CT26 colon adenocarcinoma cells results in suppression of matrix metalloproteinase-2 (MMP-2) expression, decreased in vitro tumor cell motility and invasiveness, and attenuation of functional in vivo hepatic metastasis (17). Together, these data suggest that OPN is necessary for CT26 colorectal metastasis. Subsequent studies in our laboratory have confirmed that OPN protein expression is highly up-regulated in metastatic CT26 but absent in nonmetastatic cell lines such as YAMC.
In this study, we investigated the molecular determinants that regulate OPN expression in CT26 colorectal tumor cells. CT26, an undifferentiated murine colon adenocarcinoma cell line, produces aggressive pulmonary and hepatic metastases in murine models (18,19), and constitutively expresses OPN at high levels. We hypothesized that specific cis-regulatory domains and trans-factors control OPN expression in CT26 colorectal cancer. Using transient transfection and clonal deletion analysis of OPN promoter-luciferase constructs, we identified cis-regulatory regions within the OPN promoter in CT26 cells versus YAMC control. Gel-shift assays sublocalized 20 -30-base pair regions in this cis-domain that could form DNA-nuclear protein complexes. Competition with unlabeled mutant probes characterized the nucleotide sequence of the essential nuclear factor binding sites. A TRANSFAC data base query and antibody-supershift assays identified the corresponding trans-regulatory nuclear proteins as Ets-1 and Runx2. Chromatin immunoprecipitation (ChIP) assays evaluated the in vivo binding capacity of these targets to the OPN promoter. Consensus binding site mutations in subcloned OPN promoter-luciferase constructs demonstrated a significant decrease in transcriptional activity. Antisense oligonucleotides to Ets-1 and Runx2 determined the functional effect of decreased expression of these trans-regulatory nuclear proteins in CT26 cells. Together, these data suggest that expression of the tumor metastasis protein, OPN, requires Ets-1 and Runx2.

EXPERIMENTAL PROCEDURES
Cell Culture-CT26 murine colon carcinoma cells were grown as monolayer cultures in DMEM-10% fetal bovine serum (FBS) (Invitrogen) supplemented with 100 IU/ml penicillin and 100 g/ml streptomycin. Cells were maintained in a 37°C incubator with 5% CO 2 -humidified air. YAMC cells were provided as a gift from Dr. Robert Whitehead (Vanderbilt University Medical Center, Nashville, TN). YAMC is a conditionally immortalized murine colonic epithelial cell line that expresses SV40-large-T-antigen when stimulated by interferon-␥ at 33°C. SV40-large-T-antigen is heat labile and degraded at 37°C. YAMC cells were grown in monolayer in RPMI 1640 (Invitrogen) supplemented with 5% FBS, 1 g/ml insulin, 5 IU/ml mouse interferon-, 100 IU/ml penicillin, and 100 g/ml streptomycin in a 33°C incubator with 5% CO 2 in humidified air. The utilization by established investigators of the metastatic and nonmetastatic companion cell lines CT26 and YAMC in studying colon cancer is well described in the literature (20,21).
Transient Transfection and Activity Assay-Preliminary experiments using pGL3-␤-galactosidase were performed to determine that transfection efficiency was comparable between the two cell lines, CT26 and YAMC. DNA transfections of CT26 and YAMC cells were carried out in 12-well plates using Lipofectamine 2000 (Invitrogen). 24 h prior to transfection, cells were harvested using trypsin and plated at a density of 2 ϫ 10 5 cells/well in 12-well plates in DMEM-10% FBS without antibiotics. 2 g of plasmid DNA diluted in Opti-MEM I (Invitrogen) and 2 l of Lipofectamine 2000 diluted in Opti-MEM I were combined and incubated for 20 min at room temperature. These reagents were incubated with the previously plated cells for 4 h at 37°C in a CO 2 incubator. After 4 h, the transfection medium was then replaced with DMEM-10% FBS. 0.1 g of pRL-SV40-Renilla, which contains a full-length Renilla reniformi luciferase gene under the control of a constitutive promoter was added to each well as a control for transfection efficiency between different wells. 24 h after transfection, the cells were harvested in 150 l of reporter lysis buffer (Promega), and dual luciferase reporter assays were performed by following the manufacturer's directions. 20 l of lysate was used for measurement in a luminometer (Turner Designs TD-20/20, Sunnyvale, CA). The results were reported as relative luciferase activity, which represents a ratio of luciferase firefly activity/Renilla activity. The data represent the mean Ϯ S.D. of three different experiments repeated in triplicate.
Nuclear Extract Preparation-Monolayers of CT26 cells were washed with phosphate-buffered saline and harvested by scraping into cold phosphate-buffered saline. The cell pellet obtained by centrifugation was resuspended in buffer containing 10 mM HEPES, pH 7.9, 10 mM KCL, 0.1 mM EDTA, 0.1 mM EGTA, 1.0 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride; then 10% Nonidet P-40 was added and vortexed briefly. The nuclei were pelleted by centrifugation. The nuclear proteins were extracted with buffer containing 20 mM HEPES, pH 7.9, 0.4 mM NaCl, 1.0 mM EDTA, 1.0 mM EGTA, 1.0 mM dithiothreitol, and 1.0 mM phenylmethylsulfonyl fluoride. Insoluble material was removed by centrifugation at 14,000 rpm, and the supernatant containing the nuclear proteins was stored at Ϫ80 C until use.
Transcription Factor Data Base Analysis-The nucleotide sequence corresponding to the cis-regulatory domain in the OPN promoter was cross-referenced with the TRANSFAC transcription factor data base to identify consensus binding sites of known transcription factors. Candidate transcription factors were screened for using antibody-supershift analysis in the gel-shift assays, as described above, to confirm the presence of the factor in our oligonucleotide-protein complexes.
Chromatin Immunoprecipitation Assay-Chromatin isolated from CT26 and YAMC cells were fixed and immunoprecipitated using the ChIP assay kit (Upstate Biotechnology, Lake Placid, NY) according to the manufacturer's instructions. The purified chromatin was immunoprecipitated using 10 g of anti-AML-3, anti-PEBP2␣, anti-Ets-1, or IgG isotype. After DNA purification, the presence of the selected DNA sequence was assessed by PCR. The PCR product was 265 bp in length. The PCR program was: 94°C for 4 min followed by 94°C for 45 s, 55°C for 45 s, and 72°C for 45 s, for a total of 28 cycles, and then 72°C for 7 min. PCR products were resolved in 10% acrylamide gels. The average size of the sonicated DNA fragments subjected to immunoprecipitation was 300 bp as determined by ethidium bromide gel electrophoresis. The ChIP assay utilized PCR primers 5Ј-CCTTTCATCCCCACTGATGT-3Ј and 5Ј-TGAGGTTTTTGCCACTACCC-3Ј.
Ets-1 and Runx2 siRNA and Assays-We utilized an RNAimediated approach to develop cell lines that could be utilized in an in vivo model of metastasis. Ets-1 and Runx2 siRNA (sc-35346, sc-37146; Santa Cruz Biotechnology) were used at a con-centration of 10 M for transient transfection of CT26 cells as described above. CT26 were transfected using mismatch siRNA or Ets-1/Runx2 siRNA. After 24 h, cells were transfected again using the corresponding siRNA. After 72 h from the initial transfection, cell lysates were collected and analyzed using Western blotting to confirm the extent of Ets-1, Runx2, and OPN protein expression. Subsequently, cells were assayed for viability by trypan blue exclusion. Stably down-regulated cell lines would then be developed for use in a previously established model of in vivo experimental metastasis (17).
Statistical Analysis-Data are expressed as mean Ϯ S.D. Statistical analysis was performed using SigmaStat, version 3 (Systat Software, Point Richmond, CA). Individual comparisons were made with Student's t test. The criterion for significance was p Ͻ 0.05 for all comparisons.

Transient Transfection Analysis of OPN Promoter Deletion
Constructs-To localize potential cis-acting elements in the OPN promoter, deletion constructs were analyzed in CT26 colon cancer cells using transient transfection (Fig. 1). Serial  JULY 14, 2006 • VOLUME 281 • NUMBER 28 deletion constructs demonstrated a significant 3-fold increase in luciferase activity between nt Ϫ88 and Ϫ258 in CT26 cells (Fig. 1A). Transient transfection of these deletion constructs in YAMC, a nonmetastatic colonic epithelial cell line, did not demonstrate any increased luciferase activity (Fig. 1A). CT26 demonstrated an ϳ8-fold increased luciferase activity in comparison with YAMC. Deletion fragments containing nucleotides upstream from Ϫ258 did not confer any additional, significant increase in luciferase activity in CT26 cells (Fig. 1A). Using further serial deletion constructs, this area of increased OPN promoter activity was further localized to the length of the promoter from nt Ϫ107 to Ϫ160 (Fig. 1B). Fragment Ϫ120 and Ϫ134 demonstrated an incremental 3.8-fold and a 6.4-fold increase in luciferase activity in comparison with control. These results suggest that more than one enhancer may reside in the nt Ϫ107 to Ϫ160 cis-regulatory domain of the OPN promoter.

Transcriptional Regulation of OPN in Colorectal Cancer
Gel-shift Analysis-To determine whether trans-activating nuclear factors may regulate the nt Ϫ107 to Ϫ160 region of the OPN promoter, gel-shift analysis was performed using [ 32 P]ATP-labeled fragments (ϳ20 to 30 nt) that span this region of interest (Fig. 2). Nuclear proteins were isolated from CT26 cells and incubated with three overlapping, labeled oligonucleotide probes: probe A (nt Ϫ101 to Ϫ123); probe B (nt Ϫ121 to Ϫ145); and probe C (nt Ϫ140 to Ϫ165), as described under "Experimental Procedures." Nuclear proteins bound to probe A (Fig. 2, lane 4) and probe B (Fig. 2, lane 5) but not to probe C. Three gel-shift complexes (E1, E2, and E3) were present with probe A, whereas two distinct complexes (R1 and R2) were seen with probe B. Gel-shift complexes E1, E2, and E3 and R1 were extinguished in the presence of 50-fold excess unlabeled probe A or B, respectively (Fig. 2, lanes 7 and 8), but persisted in the presence of a 50-fold excess of nonspecific unlabeled competitor (Fig. 2, lanes 10 and 11). These gel-shift complexes corresponding to E1, E2, E3, and R1 suggest that potential trans-regulatory nuclear proteins are bound to the OPN promoter in the region of nt Ϫ101 to Ϫ145 in CT26 cells.
The binding sites were further characterized by serial mutations of the OPN promoter between nt Ϫ101 to Ϫ123 in probe A (Fig. 3A) and between nt Ϫ121 to Ϫ145 in probe B (Fig. 3C ). These mutated sequences were then used as excess unlabeled competitors in gel-shift assays (Fig. 3, B and D). Mutation of nucleotides Ϫ115 to Ϫ118 in probe A resulted in persistence of complexes E1, E2, and E3 (Fig. 3B, lanes 8 and 9), whereas mutations of nt Ϫ101 to Ϫ114 and nt Ϫ119 to Ϫ123 resulted in ablation of complexes E1, E2, and E3 in competition gel-shift assays (Fig. 3B, lanes 1-7, 10, and 11). These data suggest that nt Ϫ115 to Ϫ118 (AAGG) represents an essential binding site for the nuclear proteins corresponding to the E1, E2, and E3 complexes. Mutation of nt Ϫ129 to Ϫ134 in probe B resulted in persistence of complexes R1 and R2 (Fig. 3D, lanes 5-7), whereas mutations of nt Ϫ121 to Ϫ128 and Ϫ135 to Ϫ145 resulted in ablation of R1 in competition gel-shift assays (Fig.  3D, lanes 1-4 and 8 -12). These data suggest that nt Ϫ129 to Ϫ134 (AACACC) represents an essential binding site for the nuclear proteins corresponding to the R1 complex. Together, these data from the probe A and B serial deletions indicate that two potential enhancer binding sites (nt Ϫ115 to Ϫ118 and nt Ϫ129 to Ϫ134) reside within the cis-regulatory region nt Ϫ101 to Ϫ145.
Identification of trans-Regulatory Proteins and Confirmation of DNA-Protein Binding-To identify the nuclear factors that bind our cis-element of interest, we searched the TRANSFAC data base for consensus binding sites of known transcription factors within the nt Ϫ101 to Ϫ145 segment (Fig. 4). The region nt Ϫ115 to Ϫ118 matched with the known consensus binding site of Ets (Fig. 4A), whereas the region nt Ϫ129 to Ϫ134 matched with Runx2, SBF-1 (spliceosome-binding factor-1), and C/EBP (Fig. 4B). We selected these nuclear proteins as candidates for specific binding in supershift and ChIP assays. Gelshift assays were repeated in the presence of Ets-1 antibody (Fig.  5A, lanes 7 and 8) and Runx2 antibodies (Fig. 5B, lanes 7-8 and  10 -11) with the previously described DNA probes A and B. In the presence of antibody against Ets-1, complexes E1, E2, and E3 were diminished, and a specific supershift band was formed (Fig. 5A, lanes 7 and 8). In the presence of antibody against Runx2 (S-19 X), complex R1 was diminished with formation of a supershift band (Fig. 5B, lanes 7 and 8). This effect was more pronounced with the use of ␣-aml-3, a different antibody raised against Runx2, with the formation of an intense supershift band  4 -6). Three complexes (E1, E2, and E3) were formed in the presence of probe A. Two complexes (R1 and R2) were formed with probe B. Unlabeled specific competitor oligonucleotides (SC ) corresponding to probes A, B, and C were used in Ͼ50-fold excess to evaluate the specificity of binding to the nuclear proteins (lanes 7-9). A nonspecific unlabeled competitor oligonucleotide (NC ) corresponding to nt Ϫ174 to Ϫ202 was used as control (lanes 10 -12). Free probe in the absence of nuclear extract is also shown (lanes 1-3). The blot is representative of three separate experiments. Comp, competitor.
( Fig. 5B, lanes 10 and 11). No shift was detected using IgG isotype, and no complexes were formed between the antibodies and DNA probe in the absence of nuclear extract (Fig. 5, A, lane  5, and B, lanes 6 and 9). No supershift complexes were formed using antibodies to SBF-1 and C/EBP (data not shown). ChIP assays were then performed to confirm the in vivo binding of Ets-1 and Runx2 to this portion of the OPN promoter (Fig. 5C ). Control reactions using no antibody or IgG isotype did not exhibit specific immunoprecipitation, whereas reactions containing antibodies to Ets-1 or Runx2 demonstrated specific binding to this region of the OPN promoter. Chromatin from YAMC cells did not result in specific Ets-1 or Runx2 antibodymediated immunoprecipitation (Fig. 5D). Together, these data suggest that Ets-1 and Runx2 bind specifically to enhancer domains in the cis-regulatory region nt Ϫ101 to Ϫ145 of the OPN promoter in CT26 cells.
Functional Analysis of ETS-1 and Runx2 on DNA Binding, OPN Transcription, and Protein Expression-To determine whether trans-activation of Ets-1 and Runx2 is critical to the transcription of OPN in CT26 cells, we constructed OPN promoter sequences containing mutant binding sites for Ets-1 and Runx2 and analyzed them in gel-shift and luciferase-promoter assays. Mutation of nt Ϫ115 to Ϫ118 from AAGG to TTTT resulted in ablation of the E1 and E3 complexes with significant FIGURE 4. Consensus binding sites of known transcription factors in the cis-regulatory region nt ؊101 to ؊145. The TRANSFAC data base was queried for consensus binding sites of known transcription factors for the sequence nt Ϫ101 to Ϫ145. A, binding sites are shown for Ets, upstream stimulating factor (USF), and alcohol dehydrogenase regulator-1 (ADR1) in the domain Ϫ101 to Ϫ123. B, binding sites are shown for C/EBP, Runx2, and spliceosome-binding factor-1 (SBF-1). FIGURE 3. Characterization of the nuclear binding sites using competitive gel-shift analysis. A, to sublocalize binding sites within nt Ϫ101 to Ϫ123, 11 mutant oligonucleotides based on the probe A template were synthesized to contain sequential 2-3-bp mutations. WT, wild type. B, these 11 mutant unlabeled oligonucleotides were used in Ͼ50-fold excess in competitive gel-shift assays against 32 P-labeled probe A (nt Ϫ101 to Ϫ123). Differential ablation of nuclear protein-probe A binding is shown for mutants 1-11. Control reactions showing NE ϩ probe A, NE ϩ probe A ϩ specific competitor (SC ), and NE ϩ probe A ϩ nonspecific competitor (NC ) are shown for comparison. Three specific complexes (E1, E2, and E3) are distinguishable in the blot. The blot is representative of three separate experiments. C, to sublocalize binding sites within nt Ϫ120 to Ϫ145, 12 mutant oligonucleotides based on the probe B template were synthesized to contain sequential 2-3-bp mutations. D, these 12 mutant unlabeled oligonucleotides were used in Ͼ50-fold excess in competitive gel-shift assays against 32 P-labeled probe B (nt Ϫ120 to nt Ϫ145). Differential ablation of nuclear protein-probe B binding is shown for mutants 1-12. Control reactions showing NE ϩ probe B, NE ϩ probe B ϩ specific competitor, and NE ϩ probe B ϩ nonspecific competitor are shown for comparison. Two specific complexes (R1 and R2) are distinguishable in the blot. The blot is representative of three separate experiments. JULY 14, 2006 • VOLUME 281 • NUMBER 28 reduction in the binding of E2 in gel-shift assays (Fig. 6, lane 2). This effect was duplicated by mutating the consensus binding site for Ets-1 with complete ablation of complexes E1, E2 and E3 (Fig. 6, lane 3). Both mutation of nt Ϫ129 to Ϫ134 from AACACC to TTTACA and mutation of the consensus binding site for Runx2 resulted in ablation of complexes R1 and R2 (Fig. 6, lanes 5  and 6).

Transcriptional Regulation of OPN in Colorectal Cancer
The OPN promoter fragment Ϫ512 demonstrated promoter activity similar to full-length OPN promoter in CT26 cells (data not shown) and was used as a template for constructing consensus mutants. OPN Ϫ512 wild type, OPN Ϫ512⌬Ets, OPN Ϫ⌬Runx, and OPN Ϫ⌬Ets/ Runx double mutant were subsequently transfected into CT26 cells (Fig. 7A). Luciferase activity was reduced by ϳ40 and ϳ52% in OPN Ϫ512⌬Ets and OPN Ϫ512⌬Runx, respectively, in comparison with OPN Ϫ512 wild type (Fig. 7A). In OPN Ϫ⌬Ets/Runx double mutant, luciferase activity was reduced by ϳ63% in comparison to OPN Ϫ512 wild type.
Transient transfection of antisense oligonucleotides against Ets-1 and Runx2 resulted in a significant Ͼ3-fold and Ͼ4-fold decrease in OPN protein expression, respectively (Fig. 7B). Co-transfection of Ets-1 and Runx2 antisense resulted in a Ͼ7-fold decrease in OPN protein expression in comparison with control (Fig. 7B). Together, these data suggest that ETS-1 and Runx2 trans-activation are critical for OPN expression in CT26 cells.

Functional Analysis of ETS-1 and Runx2 Down-regulation on in Vivo
Metastasis-We aimed to develop stable Ets-1/Runx2-down-regulated cell lines that could be used in in vivo assays of tumor metastasis. Initially, Ets-1 and Runx2 siRNA were used to evaluate for efficiency of OPN down-regulation. Co-transfection of Ets-1 and Runx2 siRNA resulted in decreased levels of OPN protein comparable with that achieved with antisense (data not shown). However, analysis of tumor  4) were used as controls. This blot is representative of three experiments. C, ChIP analysis of Ets-1 and Runx2 binding. Chromatin from CT26 cells was fixed and immunoprecipitated using the ChIP assay kit as recommended by the manufacturer (Upstate Biotechnology). The purified chromatin was immunoprecipitated using 10 g of anti-Ets-1, SC-19 X, anti-aml-3, corresponding IgG isotype, or no antibody (no Ab) as control. The input fraction corresponded to 0.1% of the chromatin solution before immunoprecipitation. After DNA purification, the presence of the selected DNA sequence was assessed by PCR. The blot is representative of three experiments. D, chromatin from YAMC cells was fixed and immunoprecipitated using the ChIP assay kit as recommended by the manufacturer (Upstate Biotechnology). The purified chromatin was immunoprecipitated using 10 g of ␣-Ets-1, SC-19 X, ␣-aml-3, the corresponding IgG isotype, or no antibody as control. The input fraction corresponded to 0.1% of the chromatin solution before immunoprecipitation. After DNA purification, the presence of the selected DNA sequence was assessed by PCR. The blot is representative of three experiments.
cell viability at 72 h after transfection demonstrated that Ͼ76% of cells were nonviable compared with control in triplicate experiments (Fig. 8). No viable cells were seen in culture for Ets-1/ Runx2 siRNA transfectants at Ͼ4 days in culture (data not shown). Cell lines that stably expressed down-regulated levels of Ets-1/ Runx2 could not be generated for use in in vivo murine metastasis models.

DISCUSSION
OPN is a ϳ298-amino acid secreted phosphoprotein that is expressed at high levels by T-lymphocytes, epidermal cells, bone cells, macrophages, endothelial cells, and tumor cells in remodeling processes such as inflammation, ischemia-reperfusion, bone resorption, arteriosclerosis, and tumor progression (3). Increased OPN expression is associated with tumor invasion or metastasis in cancers of the breast (23-25), stomach (26), lung (27,28), prostate (29), liver (30,31), and colon (15). In previous studies, we have shown that up-regulated expression of OPN is a critical determinant of the metastatic phenotype of CT26 colon cancer cells. Using RNA interference, stable downregulation of OPN protein levels by ϳ3.0-fold compared with CT26 wild type resulted in a Ͼ3.0-fold decrease in MMP-2 expression, a 3.6-fold decrease in tumor cell motility in vitro, a 4.1-fold reduction in tumor invasiveness in vitro, and a 2.0-fold attenuation of in vivo hepatic metastasis (17). This demonstration of OPN function in colorectal metastasis supports similar data from various cancer models (32)(33)(34). Although the target genes activated by OPN in metastasis are being actively investigated, the molecular mechanisms that regulate this increased OPN expression in colorectal cancer remain unknown. Analysis of the OPN promoter has uncovered multiple consensus FIGURE 7. Transient transfection analysis of Ets-1 and Runx2 consensus binding site mutations on OPN promoter activity. A, the histograms are representations of luciferase activity from OPN promoter constructs containing Ets-1 consensus binding site mutations (Ϫ512⌬Ets), Runx2 consensus binding site mutations (Ϫ512⌬Runx), or combined Ets-1/Runx2 binding site mutations (Ϫ512⌬Dm) subcloned into pGL-3 and normalized to Renilla-luciferase activity from co-transfected pRL-SV40-Renilla plasmids. Deletion fragments corresponding to OPN Ϫ69 to ϩ79, Ϫ120 to ϩ79, and Ϫ512 to ϩ79 were included as controls. The values are expressed as the mean Ϯ S.D. of three triplicate experiments. *, p Ͻ 0.05 versus Ϫ512 wild type (wt). B, transient transfection analysis of antisense oligonucleotides to Ets-1 and Runx2 and their effect on OPN protein expression. Sense and antisense oligonucleotides to Ets-1 (lanes 2 and 4), Runx2 (lanes 3 and 5), or Ets-1/Runx2 (lanes 6 and 7) were transiently transfected into CT26 cells. Cell lysates were collected after 48 h, and OPN protein levels were measured in Western blots. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels are shown as an internal control for loading. The blot is representative of three separate experiments.  1 and 4). The blot is representative of three experiments. WT, wild type.
In this present study, we have identified a specific OPN promoter region, nt Ϫ107 to Ϫ160, which significantly increases transcriptional activity in CT26 colorectal cancer cells. Using gel-shift assays, two cis-regulatory enhancer domains were localized to nt Ϫ115 to Ϫ118 and nt Ϫ129 to Ϫ134. These regulatory domains were demonstrated to specifically bind the nuclear proteins Ets-1 and Runx2. Loss-of-trans-activation of these nuclear factors through mutation of the consensus binding sites or through antisense inhibition resulted in decreased OPN promoter activity and protein expression. These data suggest that Ets-1 and Runx2 are critical transcription factors that regulate OPN expression in CT26 colorectal cancer cells. A direct analysis of Ets-1/Runx2(Ϫ/Ϫ) CT26 mutants in an in vivo metastasis model was not feasible, but we have previously demonstrated that OPN expression is integral to the CT26 metastatic phenotype in a syngeneic BALB/c murine model (17). In the context of this previously published data, our current study demonstrating in vitro down-regulation of OPN expression by Ets-1/Runx2 antisense and in vivo binding of Ets-1/Runx2 to CT26 DNA in our ChIP assays strongly suggest that Ets-1 and Runx2 function as regulators of OPN-mediated metastasis in CT26 cells.
Ets-1 and Runx2 are critical to normal physiologic development, and their activity is also enhanced during tumorigenesis. Ets-1 is expressed in lymphoid cells, developing T and B cells, endothelial cells, and vascular smooth muscle cells (36 -39), and it belongs to a family of ϳ30 mammalian nuclear transcription factors. These proteins contain multiple conserved domains including a winged-helix-loop-helix domain that mediates binding to purine-rich DNA sequences with a central GGA(A/T) core, a Pointed domain that regulates proteinprotein interactions, a Ras-responsive phosphorylation site at threonine 38, an autoinhibitory module on exon VII that is activated by calcium-dependent phosphorylation on Ser-25, Ser-257, Ser-282, and Ser-285, and a C-terminal domain that mediates trans-activation (40 -43). The Ets nuclear proteins cooperate with other transcription factors to activate or repress transcription in a variety of processes including cell proliferation, apoptosis, development, differentiation, angiogenesis, and oncogenic transformation (41)(42)(43). Augmentation of Ets-mediated gene transcription results from coactivation with AP-1, AML-1, CBP (CREB-binding protein)/p300, Lef-1 Sp1, NF-B, Stat-5, and Maf-B, whereas suppression of Ets function occurs with Erg, Daxx, and Ubc9 (41)(42)(43). The Runx gene family encodes the DNA binding ␣-subunit of a heterodimeric transcription complex. The Runx genes have several aliases including acute myeloid leukemia (AML), core-binding factor-␣ (CBF␣), and polyomavirus enhancer-binding factor-2 (PEBP2) (44). The Runx transcription factors play a pivotal role during normal development and tumorigenesis (45)(46)(47). The mammalian Runx genes (Runx1, Runx2, Runx3) encode three different isoforms of the ␣-subunit and share a highly conserved 128-amino acid Runt homology domain that mediates DNA binding. Dimerization with the PEBP2␤/CBF␤-subunit enhances DNA binding of the ␣-subunit (47). Despite their structural similarity, the Runx genes have divergent biological roles in mammalian development with Runx1-, Runx2-, and Runx3null mice showing major defects in hematopoietic, osteoblastic and neuronal development, respectively (48).
Ets-1 and Runx2 protein expression are up-regulated in tumor progression, and their activation is intimately linked with cellular migration, invasion, and tumor metastasis. Ets-1 transforms NIH3T3 fibroblasts (41), regulates MMP-1, MMP-3, MMP-7, and MMP-9, and urokinase-type plasminogen activator during tumor invasion and metastasis, and is in turn up-regulated by VEGF, basic fibroblast growth factor (bFGF), and hypoxia-inducible factor-1 (HIF-1) during tumor angiogenesis (41)(42)(43)49). In the context of colorectal cancer, immunohistochemical studies have shown that Ets-1 expression is significantly increased in adenomas, carcinomas in situ, and colonic adenocarcinomas but is absent from normal or hyperplastic polyps (50). In other colon cancer models, Ets-1 and Ets-2 expression is directly linked to lymph node metastasis (50), depth of invasion (51), VEGF expression, and lowered survival rates (52). In contrast, there have been no published studies describing a similar role for Runx2 in colorectal carcinoma. However, the Runx genes have been shown to function in tumor suppression of gastric cancer (53), granulocytic differentiation of myeloid precursor cells and leukemias (54), and osteoblast differentiation and cleidocranial dysplasia (55)(56)(57). Both Runx enhancer and repressor activity have been demonstrated previously in molecular models (45,58).
Together, these previous studies provide a limited understanding of how Ets-1 and Runx2 mediate functional colorectal metastasis. In this context, our study represents the first report describing a mechanistic role for Ets-1 and Runx2 in the transactivation of OPN expression in a colon cancer cell line. There are previous reports using osteogeneic (59) and mammary cell lines (60) that describe a regulatory role for these factors in OPN expression. Runx2 was shown to induce OPN expression in NIH3T3 fibroblasts with Ets-1 acting as a synergistic coactivator (61). More recently, Runx2 activation of OPN expression has been reported in metastatic mammary epithelial cells (55,62). Inman and Shore (55) confirm that Runx2 functions as a transcriptional activator of the murine OPN promoter in HC11 mouse mammary epithelial cells and that targeted downregulation of Runx2 results in Ͼ50% reduction of OPN promoter activity in HC11 cells (55). In other studies, El-Tanani et al. (60) demonstrate that a cooperative interaction between Ets, PEA3, ␤-catenin/Lef-1, and AP-1 activates OPN transcription in RAMA 37 cells (60). In the context of these data, our results add support to the role of Ets-1 and Runx2 as significant activators of cancer metastasis.
Our attempts to develop stable Ets-1/Runx2 down-regulated cell lines for use in functional, in vivo metastatic assays were unsuccessful, as our Ets-1/Runx2-deficient cell lines were not viable. These results are consistent with data from other investigators who have demonstrated that Ets-1 knock-out results in increased T-cell apoptosis (63), abnormal development and differentiation in NK cells and B-cells (64,65), reduced cell adhesiveness in HeLa cells (66), and diminished cell proliferation in glioma cells (67). In a similar fashion, Runx2 gene knock-out in mice causes a lethal phenotype (57) and RNAi-mediated downregulation of the Runx homologue, run, in Caenorhabditis elegans results in an early larval lethal phenotype (69). Recent studies from Passaniti and colleagues (70) demonstrate that RNAi-mediated knock-out of Runx2 inhibits cell proliferation and delays cell cycle progression in human bone marrow endothelial cells. This inhibition of cell proliferation occurred over a similar time course in comparison with our data. Together, these data suggest that Ets-1 and Runx2 regulate essential, basic processes during mesenchymal cell differentiation and also during tumor progression.
An investigation into the molecular interactions between Ets-1 and Runx2 in our cell model will be the focus of future studies. We suspect that Ets-1 and Runx2 cooperate to activate OPN transcription in a synergistic fashion in CT26. Previous studies in other cell models have shown that Ets-1 can bind the Runx2-related AML-1. Ets-1 stimulates DNA binding activity of AML-1 by associating with its negative regulatory domain for DNA binding (NRDB) (71,72). In turn, AML-1 binds to the exon VII domain and blocks the inhibitory module of Ets-1 (71,68). This cooperative synergism is supported by the findings of Sato et al. (61), who demonstrate that Ets-1 and PEBP2␣ cooperate to regulate OPN transcription in skeletal tissues (61). The precise mechanisms that govern the regulated activity of Ets-1 and Runx2 are complex and unknown. Further insight into the study of these paired transcriptional activators of OPN may identify potential therapeutic targets in the regulation of cancer metastasis.