Gene Expression Profiling Leads to Identification of GLI1-binding Elements in Target Genes and a Role for Multiple Downstream Pathways in GLI1-induced Cell Transformation*

The zinc finger transcription factor GLI1, which mediates Sonic hedgehog signaling during development, is expressed in several human cancers, including basal cell carcinoma, medulloblastoma, and sarcomas. We identified 147 genes whose levels of expression were significantly altered in RNA obtained from cells demonstrating a transformed phenotype with stable GLI1 expression or stableHa-ras expression. Comparison of expression profiles fromGLI1- and Ha-ras-expressing cells established a set of genes unique to GLI1-induced cell transformation. Thirty genes were altered by stable GLI1 expression, and 124 genes were changed by stable Ha-ras expression. Seven genes had altered expression levels in both GLI1- andHa-ras-expressing cells. Genes whose expression was altered by GLI1 included cell cycle genes, cell adhesion genes, signal transduction genes, and genes regulating apoptosis. GLI1 consensus DNA-binding sequences were identified in the 5′ regions of cyclin D2, IGFBP-6, osteopontin, and plakoglobin, suggesting that these genes represent immediate downstream targets. Gel shift analysis confirmed the ability of the GLI1 protein to bind these sequences. Up-regulation of cyclin D2 and down-regulation of plakoglobin were demonstrated in GLI1-amplified compared with non-amplified human rhabdomyosarcoma cells. Many of theGLI1 targets with known function identified in this study increase cell proliferation, indicating that GLI1-induced cell transformation occurs through multiple downstream pathways.

Important gene hierarchies, in part coding for components of signal transduction pathways, regulate growth and differentiation during development. One such pathway is the Sonic hedgehog-Patched-Gli pathway (1). SHH 1 signaling is critical to the genetic specification of fate of many tissues during early organogenesis including the central nervous system (2,3), lung (4), prostate (5), bone (6 -8), and muscle (9). SHH signaling is mediated by the GLI family of transcription factors (10). One of these genes, GLI1, has been shown to be a transcriptional activator operating through a C-terminal VP-16-like acidic helical domain (11). GLI1 transforms cells in culture, and its expression is associated with significant human cancers including basal cell carcinoma (12), medulloblastoma (13), and sarcomas (14). Few downstream targets of GLI1 are known, which precludes a clear understanding of its action in carcinogenesis. Genetic evidence suggests that PTCH and Wnt genes are downstream targets of GLI1 (15), and biochemical evidence has established HNF-3␤ (Hepatocyte Nuclear Factor-3␤) as a target of GLI1 during development (16).
Microarray technology has provided a methodology to study the expression of thousands of genes simultaneously and has been used in many important settings (17). Among these is the dissection of signal transduction pathways. To identify unique downstream targets of GLI1, we have utilized a cell transformation phenotype as a selection system for the stable integration and expression of either GLI1 or Ha-ras in RK3E cells. To identify genes specific to the GLI1 transformation process, the expression profiles of cells transformed with GLI1 were compared with those from cells transformed with Ha-ras. Untransformed and transformed cells following transfection without drug selection were cloned, and gene expression profiles were established. We found 147 genes with altered expression levels from 4,608 UniGene clones, 30 as a result of stable GLI1 expression and 124 as a result of stable Ha-ras expression. Six genes were down-regulated by both GLI1 and Ha-ras expression. One gene was up-regulated by GLI1 and down-regulated by Ha-ras.
Genes whose expression was altered by GLI1 included cell cycle genes, cell adhesion genes, signal transduction genes, and genes regulating apoptosis. GLI1 consensus DNA-binding sequences were identified in the 5Ј regions of cyclin D2, IGFBP-6 (IGF-binding protein 6), osteopontin, and plakoglobin. Gel shift analysis confirmed the ability of GLI1 protein to bind these sequences. Up-regulation of cyclin D2 and down-regulation of plakoglobin were also demonstrated in GLI1-amplified human tumor cells. Thus, cellular transformation by the human oncogene GLI1 proceeds via multiple downstream pathways.

EXPERIMENTAL PROCEDURES
Preparation of Transformed Cell Lines-Rat kidney epithelial cells stably transfected with E1A (RK3E cells, American Type Culture Collection, CRL 1895) were maintained in Dulbecco's modified Eagles's medium (Invitrogen) supplemented with 10% fetal bovine serum, pen-icillin (100 units/ml), and streptomycin (100 g/ml). Both the GLI1 and Ha-ras genes have been shown to transform RK3E cells (18). The human GLI1 (pLTR-GLI) or Ha-ras (pOH6T1) (19) expression constructs were introduced into rat kidney cells (RK3E) by liposome-mediated transfection and transformed foci formed in 1-2 weeks. Cell lines were established by trypsinization of foci. GLI1-transformed RK3E cells did not show significant morphological differences compared with the parental RK3E cells at low density, however, and formed foci when grown to confluence. Ha-ras-transformed cells showed greater morphological changes and grew in a more anchorage-independent manner (data not shown) compared with the parental RK3E cells and even at low density formed foci. Both GLI1-and Ha-ras-transformed RK3E cells showed increased growth rates compared with RK3E cells. Western blot analysis using RK3E cells transformed with GLI1 revealed a 150-kDa GLI1 protein band. GLI1 protein was not detected in RK3E cells or Ha-ras-transformed RK3E cells establishing the presence of stable protein product from the integrated GLI1 gene (data not shown).
PCR Amplification for Microarray-Rat UniGene clones (4,608) supplied by Research Genetics (Huntsville, AL) were PCR-amplified with a primer set including T7/T3 promoters as follows: 5Ј-TTACGAATTTAA-TACGACTCACTATA-3Ј; 5Ј-AAGCTAAAATTAACCCTCACTAAAGGG-3Ј. GAPDH and ␤-actin served as control genes and were amplified with a primer set including M13/T7 promoters as follows: 5Ј-CAGGAAACA-GCTATGAC-3Ј; 5Ј-GTAATACGACTCACTATAGGGC-3Ј. PCRs were carried out in 384-well plates with a reaction volume of 10 l using a Peltier Thermal Cycler PTC-225 (MJ Research, Watertown, MA) including 1 l of 10ϫ PCR buffer, 0.5 units of Taq DNA polymerase purified using a Centri-Sep Column (Princeton Separation, Adelphia, NJ), 250 nM each primer, 200 M each dNTP, and 0.4 l of each cDNA clone. The first cycle was 95°C for 11 min, 55°C for 1 min, and 72°C for 1.5 min. The 2nd to 36th cycles were 95°C for 1 min, 55°C for 1 min, and 72°C for 1.5 min. The final extension was 72°C for 7 min. After the reactions, 7 l of water and 17 l of Me 2 SO were added to the reaction mixtures. All PCRs were analyzed with 2% agarose gel electrophoresis using EAGLE EYE II (Stratagene, La Jolla, CA).
Preparation of cDNA Microarray-The microarrays were produced on poly-L-lysine-coated slides with a spacing of 180 m in the area of 18 ϫ 36 mm using a 4-pin arrayer (Genomic Solutions, Ann Arbor, MI). Each cDNA was stamped from the 384-well plates directly without purification of PCR products. The printed arrays were incubated in a humid chamber (Sigma) to allow rehydration with 1ϫ SSC for 2 h and heated on hot plates at 85°C for 2 h. The arrays were UV cross-linked for 65 min with a Stratalinker (Stratagene), stabilized in 10% formalin for 1 min, and then rinsed with distilled water 2 times. They were immersed in succinic anhydride (5.5 g/325 ml of 1-methyl-2-pyrrolidinone) and then in 25 ml of 1 M sodium borate, pH 8.0, for 20 min, submerged in distilled water for 2 min at 95°C, quickly transferred into 95% EtOH, dried by centrifugation, and stored until used in the dark at room temperature.
RNA Isolation and Northern Blots-Poly(A) ϩ RNA was prepared using the Poly(A) Pure mRNA isolation kit (Ambion, Austin, TX). For Northern blot analysis, 3 g of poly(A) ϩ RNA from each cell line was electrophoresed in 1% agarose gels containing formaldehyde and then transferred to nylon membranes (BrightStar Plus, Ambion). The membrane was baked at 80°C for 2 h and then prehybridized at 42°C for 30 min in a hybridization buffer (Ultrahyb, Ambion). The hybridization was carried out at 42°C for 4 h with appropriate 32 P-labeled DNA probes prepared by random priming (Rediprime II, Amersham Biosciences). The RNA bands were detected by autoradiography. The probes were obtained from Research Genetics.
Preparation and Hybridization of Fluorescence-labeled Probes-Cy3or Cy5-dUTP (Amersham Biosciences)-incorporated cDNA probes were prepared at 39°C from 4 g of mRNA in a reaction volume of 90 l including 9 l of 10ϫ PCR buffer, 10 mM dithiothreitol, 2.5 mM MgCl 2 , 4 g of (dT) [12][13][14][15][16][17][18] primer, 1800 units of SuperScriptII, 500 M each of dATP, dCTP, and dGTP, 40 M dTTP, 40 M Cy3-or Cy5-dUTP. After 120 min 8 units of RNase H (Invitrogen) was added, and the reaction mixture was incubated for 30 min at 37°C. After preparation, Cy3-and Cy5-labeled probes were mixed and purified with a Centri-Sep Column and a PCR purification kit (Qiagen, Valencia, CA). The probes were concentrated with Microcon YM-30 columns (Millipore, Bedford, MA). Ten g of poly(dA) (Sigma) and 20 g of yeast total RNA (Invitrogen) were added to the concentrated probe. The hybridization solution (11 l; 3.5ϫ SSC and 0.35% SDS) was applied to the microarray under a 22 ϫ 22-mm coverslip for 17 h at 65°C. The slide was rinsed in 2ϫ SSC, 0.1% SDS for 2 min, 0.2 ϫ SSC for 2 min and then 0.05 ϫ SSC for 2 min at room temperature. The slide was scanned with a ScanArray 5000 (General Scanning, Watertown, MA) to detect the two-color fluores-cence hybridization signals. Because of potential bias produced by different incorporation rates between Cy3-and Cy5-dUTP, two slides were used for each of four independent RNA preparations. For one slide, fluorescent cDNA probes were prepared from mRNA from RK3E cells (Cy3-labeled) and GLI1-or Ha-ras-transformed RK3E cells (Cy5-labeled); for the other slide fluorescent cDNA probes were prepared from mRNA of GLI1-or Ha-ras-transfected RK3E cells (Cy3-labeled) and RK3E cells (Cy5-labeled). The procedure was performed using two independent clones obtained from two independent transfections for both GLI1 and Ha-ras transformed cells.
Analysis-Data reduction was done with the program Gleams 2.0 (NuTec Services, Stafford, TX). Each spot was defined by manual positioning of a grid of circles over the array image. Signal intensity was determined by subtraction of local background from the mean intensity. Normalization between the dyes was accomplished by normalizing each dye to mean intensities of all genes. The threshold was set at 50% of the mean (Cy3 intensity ϩ Cy5 intensity) of all genes for each array. In the plot of log (Cy3 ϩ Cy5) versus log (Cy3/Cy5) (20,21), genes with a signal intensity below 50% of the mean of (Cy3 ϩ Cy5) showed large variances of the Cy3/Cy5 ratio. By using this threshold, 1778 and 1823 genes were selected in GLI1-RK3E versus RK3E and Ha-ras-RK3E versus RK3E microarrays, respectively. Following the intensity normalization, the log values of GLI1-RK3E or Ha-ras-RK3E versus RK3E ratios were counted for each experiment, and the mean ratios of four slides in each group was calculated for each gene. The means (ϮS.D.) of the log of the ratios, calculated as described above, for all genes that met the 50% threshold were 0.003 (Ϯ0.133) for GLI1-RK3E versus RK3E and Ϫ0.019 (Ϯ0.199) for Ha-ras-RK3E versus RK3E cells. Genes were identified as having significantly different expression levels by comparison with the mean Ϯ 2 S.D. of GLI1-RK3E versus RK3E (0.003 Ϯ 0.266) to evaluate differential gene expression with a common criterion between GLI1-RK3E versus RK3E and Ha-ras-RK3E versus RK3E. The distribution of gene expression of Ha-ras-RK3E versus RK3E cells was wider than that of GLI1-RK3E versus RK3E cells indicating that Ha-ras induces expression changes in more genes than does GLI1. We evaluated the variation induced by differences in dye incorporation and variation by cell batch and slide. The log of ratios of all genes for each slide was compared with the above mean Ϯ 2 S.D. (0.003 Ϯ 0.266) and scored as follows. Ratios Ͻ mean Ϫ2 S.D. were scored Ϫ1; ratios Ͼ mean ϩ 2 S.D. were scored ϩ1, and others were scored 0. The score was summed for all slides, and genes with scores of 4, 3, Ϫ3, or Ϫ4 were selected. The ratio of mRNA level is expressed as GLI1-RK3E/RK3E or Ha-ras-RK3E/RK3E. Positive values mean that the mRNA level in GLI1-or Ha-ras-transformed cells is larger than that in RK3E cells; negative values mean that the mRNA level in RK3E cells is larger than that in GLI1-or Ha-rastransformed cells. Sequence homology searches of genes identified in this way were done using GenBank TM data bases and the BLAST search program (NCBI). The criterion of homology was 85% sequence match without gaps over 100 base pairs of sequence (22).
Production of Biotinylated GLI1 Fusion Protein-Biotinylated GLI1 fusion protein was produced in Escherichia coli using the PinPoint TM Xa system (Promega, Madison, WI). The pinpoint-GLI1 aa-211-1106 fusion construct was prepared by inserting the BamHI fragment of the human GLI1 cDNA (pK12 GLI) into the BamHI site of the PinPoint Xa2 plasmid (Promega). For protein production, bacterial cultures were induced with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside and incubated for 3 h. Bacteria were then harvested, sonicated, and cleared by centrifugation. The GLI1 aa 211-1106 and control proteins (protein from Promega PinPoint plasmid or biotinylated GLI1 aa 879 -1106, which lacks the zinc fingers) in the clear lysate were purified using SoftLink TM Avidin resin (Promega).

Gene Expression Profiles Revealed That GLI1 and Ha-ras
Regulate Distinct Sets of Genes-We performed microarray assays using the RK3E cell line transformed independently with two oncogenes (Figs. 1 and 2) and examined 4,608 rat UniGene sequences. Comparison of expression profiles from GLI1-and Ha-ras-expressing cells established a set of genes unique to GLI1-induced cell transformation. This method identified 147 genes whose levels of expression were significantly altered in the RNA obtained from RK3E cells with stable GLI1 expression or stable Ha-ras expression. Thirty genes were altered by stable GLI1 expression as follows: 11 were highly up-regulated (Ͼ3 S.D.), 4 were up-regulated (Ͼ2 S.D.), 5 genes were highly down-regulated (Ͼ3 S.D.), and 10 genes were down-regulated (Ͼ2 S.D.). A total of 124 genes were changed by stable Ha-ras expression as follows: 30 genes were highly up-regulated (Ͼ3 S.D.), 19 genes were up-regulated (Ͼ2 S.D.), 56 genes were highly down-regulated (Ͼ3 S.D.), and 19 genes were downregulated (Ͼ2 S.D.). Only 7 genes had altered expression levels in both RK3E cells with stable GLI1 expression and RK3E cells with stable Ha-ras expression. Six of these were down-regulated by both GLI1 and Ha-ras expression, whereas one of these, serine dehydratase-2, was up-regulated by stable GLI1 expression and down-regulated by stable Ha-ras expression. This observation clearly suggests that the oncogenes GLI1 and Ha-ras depend on distinct sets of gene activities to transform the same RK3E cells.
Annotation-The gene expression profiles of known genes are summarized in Tables I and II. The altered genes are distributed among 8 functional categories including cell cycle genes, cell adhesion genes, signal transduction genes, genes regulating apoptosis, and a small group of gene clones of unknown function. PTCH, a member of SHH-Patched signaling pathway, which is believed to be a downstream target of GLI1 based on genetic analysis (23), was induced by GLI1 as predicted. We also observed induction of cyclin D2, implicating GLI1 in the regulation of the cell cycle.
RK3EϩGLI1 cells also expressed a number of genes involved in cell structure, movement, and adhesion. Osteopontin and lysyl hydroxylase genes were up-regulated, whereas embigin and plakoglobin genes were down-regulated. A gene involved in IGF signaling (IGFBP-6) was up-regulated. A number of transcription factor genes showed altered expression in the GLI1transformed cells. Npdc1 (neural proliferation differentiation and control gene 1) and TSC-22 (TGF-␤-stimulated clone 22) are in this category. Known genes down-regulated by both GLI1 and Ha-ras were TSC-22, which is a target of the tumor suppressor TGF-␤, and Tom1 (target of myb1).
Northern Blot Analysis Verifies the Microarray Assays-We performed Northern analysis to confirm the microarray data using probes to a number of the target genes identified by the microarray analysis (Fig. 2B). By using cDNA probes to the GLI1-regulated genes, significant increases in signal were seen for osteopontin, cyclin D2, H19, IGFBP-6, and PTCH, whereas significant decreases in signal were seen for TSC-22 and Tom1 as predicted by the microarray analysis.
Search for the Immediate Downstream Targets of GLI1-We have identified 30 genes regulated by the GLI1 oncogene. To determine whether these were direct downstream targets, we exploited the fact that GLI1 is a sequence-specific DNA-binding protein that interacts with the motif GACCACCCA. Among the genes regulated by GLI1, we found five genes with 5Ј sequences suitable for analysis. Of these, four genes contained putative GLI1-binding elements (Fig. 3). The putative GLI1-binding elements each show 1-base difference from the 9-bp consensus. It was shown previously that the HNF-3␤ enhancer, which contains 8 bp of matching sequence (GAACACCCA), is a functional GLI1-binding site (16). A putative GLI1-binding motif (CACCACCCA) was found in the core promoter of human cyclin D2 (24). Osteopontin promoters of rat and mouse have a putative GLI1-binding site (GACCTCCCA). The motifs GTCCAC-CCA and GACCCCCCA were found in rat and human IGFBP-6 5Ј regions, respectively. Interestingly, although plakoglobin expression is reduced in cells transfected with GLI1, the human plakoglobin promoter also contains a putative GLI1-binding element (GACCACCAA).
Gel Shift Assays Confirmed Physical Interactions between GLI1 and the Putative GLI1-binding Elements-We performed gel mobility shift assays to determine whether the GLI1 protein physically binds to the putative GLI1-binding elements (Fig. 4). The putative GLI1-binding element from the human cyclin D2 core promoter region shifted in the presence of GLI1 protein (Fig. 4A). The same gel shift assay was performed with IGFBP-6 probe (Fig. 4B), osteopontin probe, and plakoglobin probe (Fig. 4C). All of the probes produced specific shifted bands.
Human Tumor Cells with GLI1 Show Up-regulation of Cyclin D2 and Down-regulation of Plakoglobin-By using probes for human cyclin D2 and human plakoglobin, Northern blots were performed on cells derived from two different human rhabdomyosarcomas (Fig. 5). The A673 rhabdomyosarcoma does not express GLI1, whereas RMS-13 is highly amplified for GLI1. The A673 cells do not express demonstrable levels of cyclin D2 by Northern blot but have abundant plakoglobin message. In the RMS-13 human rhabdomyosarcoma cells, there is abundant cyclin D2 and no plakoglobin RNA. Thus, cyclin D2 and plakoglobin Northern data from human tumor cells are consistent with the microarray results.

GLI1 Activates a Unique Set of Genes during Cellular Transformation-
The GLI family of proteins are key mediators of SHH signaling in mammalian development. GLI1 is an oncogene expressed at high levels in many human cancers. GLI1 is a transcriptional activator, but the downstream targets of this gene, other than HNF-3␤ in mouse (16) and wg in Drosophila (25), have not been identified at a biochemical level. There are no proven human targets. Without knowing the range of genes regulated by GLI1, it will be difficult to understand its role in development or in cancer.
Here we show that the expression of 30 genes was altered by GLI1 expression in transformed RK3E cells. Among them, 15 genes were up-regulated and 15 genes were down-regulated. This represents 0.65% of the genes examined, whereas 2.7% of the genes examined were altered by Ha-ras expression. Importantly the gene expression profiles of GLI1 and Ha-ras did not overlap with the exception of seven genes regulated by both oncogenes. Furthermore, the genes activated by GLI1 were not those regulated by other oncogenes such as c-myc (26,27). Thus, the expression profile for GLI1 transformation is specific and unique.
GLI1-binding Elements Were Identified in the Promoters of Target Genes-GLI1 has been proven to be a transcriptional activator, although convincing evidence of repressor activity is lacking. Nevertheless, we observed that a number of genes were down-regulated, although this may be the result of indirect effects of GLI1 expression. We examined the 5Ј sequences for GLI1-binding sites in five genes for which appropriate sequence data were available that had altered expression levels in response to GLI1 expression. Functional GLI1 binding domains were found in four of them, one of which one is downregulated in response to GLI1 expression.
GLI1 Regulates Cell Cycle Genes-Recent evidence demonstrates that SHH opposes cell cycle arrest in epithelial cells (28). Because the GLI transcription factors mediate SHH sig-  cell movement, and apoptosis. For example, we found up-regulation of the osteopontin gene a secreted glycoprotein expressed in bone but not detected in the normal matrix of most tissues. Osteopontin promotes cell attachment and migration through its recognition by integrins (29). It is highly expressed in a wide variety of cancers including breast (30) and malig-nant gliomas (31)(32)(33). GLI1 down-regulates the expression of plakoglobin (␥-catenin), a cytoplasmic protein associated with cadherins. Cadherins are transmembrane proteins that are linked to cytoplasmic actin bundles via catenins. Increased expression of plakoglobin is associated with stunted hair growth, reduced epidermal proliferation, and apoptosis (34). Another example of negative regulation by GLI1 is reduction of embigin expression. Embigin is a cell adhesion molecule that is known to promote integrin-mediated cell-substratum adhesion (35,36). Lysyl hydroxylase, which cross-links collagen through intermolecular hydroxylation and plays a role in bone matrix formation, is up-regulated (37).
GLI1 Regulates Signal Transduction Genes-The gene expression profiles showed that GLI1 expression regulates the activity of genes in several signal transduction pathways. Patched (PTCH), the transmembrane receptor of SHH, is upregulated. Because PTCH inhibits smoothened, another transmembrane receptor, which activates GLI1, this establishes a negative feedback loop in the presence of SHH signaling. GLI1 down-regulates TGF-␤-stimulated clone TSC-22 (a leucine zipper transcription factor). The pattern of expression during murine embryonic development of TSC-22 (38) is similar to that of GLI1. TSC-22 mediates TGF-␤ tumor suppressor function by activating apoptosis and repressing growth (39,40). Because GLI1 represses TSC-22, loss of its tumor suppressor function may be an important consequence of GLI1 expression in malignant transformation.
Up-regulation of growth factor regulators was also discovered with this analysis. H19 and IGF2 genes are imprinted and closely linked on the chromosome (41,42). They can be upregulated together in a variety of tumors (43). We observed increased H19 expression in GLI1-transformed RK3E cells. IGFBP-6 is an IGF2-binding protein whose function is not clearly understood. IGFBP-6 expression is stimulated by retinoic acid (44) and is inhibited by TGF-␤1 (45). Its overexpression is also associated with several tumors. The up-regulation of H19 and IGFBP-6 by GLI1 expression may be involved in maintaining the proliferative state of cells and preventing cellular differentiation.
Combined Transcriptional Regulation by GLI1 Leads to Cellular Proliferation-In aggregate these results show that GLI1 provides strong cellular proliferation signals by transcriptional activation of cell cycle regulators, cell adhesion genes, growth factor regulators, and regulators of apoptosis (Fig. 6). For example, activation of cyclin D2 forces cells through the cell cycle. Repression of embigin and plakoglobin with induction of osteopontin reduces cell adhesion. Signaling changes in the SHH pathway itself by induction of PTCH, reduction of TSC-22, and changes in IGF signaling through increases in IGFBP-6 increase cellular proliferation and decrease apoptosis. In addition   FIG. 3. GLI1-binding motifs in the 5 region of GLI1 downstream targets. Potential GLI1-binding sites are underlined. The name of the gene is shown on the left (IGFBP-6, insulin-like growth factor-binding protein 6; HNF-3␤, hepatocyte nuclear factor 3␤). h, human; r, rat; m, mouse. HNF-3␤ sequence was from Ref. 16.

FIG. 4. Gel shift assays demonstrate GLI1 binds the 5 regions.
A, electrophoretic mobility shift assays demonstrated two shifted bands (arrows) with competition from the addition of 25-100 molar excess of non-radiolabeled human cyclin D2-specific oligonucleotide DNA. Control protein (PinPoint protein or PinPoint fused to GLI1 aa 879 -1106, lacking zinc fingers) or a mutant oligonucleotide DNA at 25-100-fold molar excess did not affect the mobility shift, indicating the specificity of the GLI1-cyclin D2 oligonucleotide interaction. B, IGFBP-6 specific oligonucleotide probe showed strong mobility shifts. 100-Fold molar excess of non-radiolabeled IGFBP-6 oligonucleotide DNA completely abrogated the shifted bands. Mutant oligonucleotide (16) at 100-fold molar excess did not compete the shifted bands. C, similarly, osteopontin and plakoglobin probes resulted in specific shifted bands which were abrogated by 200-fold molar excess of non-radiolabeled oligonucleotide or mutant competitor.