The IGF-I receptor gene promoter is a molecular target for the Ewing's sarcoma-Wilms' tumor 1 fusion protein.

Desmoplastic small round cell tumor (DSRCT) is an abdominal malignancy in children which is characterized by a recurrent chromosomal translocation, t(11;22)(p13;q12). This rearrangement results in the fusion of the ubiquitously expressed EWS1 gene to the Wilms' tumor suppressor (WT1) gene. The chimeric protein contains the N-terminal domain of EWS1 fused to the DNA-binding domain of WT1, including zinc fingers 2-4. Because WT1 has been shown previously to bind and repress the insulin-like growth factor I (IGF-I-R) promoter, we investigated whether this promoter is, in addition, a target for the aberrant EWS/WT1 transcription factor. EWS/WT1 activated the IGF-I-R promoter ~340%, whereas a fusion protein containing a three-amino acid insert (KTS) between zinc fingers 3 and 4 had no effect. On the other hand, expression vectors encoding either WT1 or EWS1 reduced the activity of the promoter to 46 and 58% of control values, respectively. Results of gel shift assays indicate that the binding affinity of EWS/WT1 to a fragment of the 5′-flanking region of the receptor promoter was higher than the affinity of WT1 itself. Consistent with the results of functional assays, the binding of EWS/WT1(+KTS) was significantly reduced. Due to the central role of the IGF-I-R in tumorigenesis, activation of the receptor promoter by EWS/WT1 may constitute a potential mechanism for the etiology and/or progression of DSRCT.

The insulin-like growth factor I receptor (IGF-I-R) 1 is a tyrosine kinase membrane-bound receptor that mediates the growth and differentiation actions of the IGFs (IGF-I and IGF-II). It is constitutively expressed by most tissues, where it is basically required for progression through the cell cycle (1)(2)(3)(4). Thus, deletion of the IGF-I-R gene in mice by homologous recombination is incompatible with postnatal life (5,6). The IGF-I-R is also implicated in malignant transformation, as shown by its high level of expression in many tumors and cancer cell lines (7) and by the inability of a number of oncogenes to transform cell lines lacking IGF-I-R (8 -10).
Some human malignancies are characterized by recurrent chromosomal translocations, frequently resulting in the fusion of genes (11). One of the best characterized rearrangements is the translocation of the c-myc gene in Burkitt's lymphoma (12). In this type of cancer, the c-myc protooncogene is juxtaposed to an immunoglobulin gene by chromosomal fusion, thereby activating the oncogene. A second example of a chromosomal translocation in hematopoietic tumors is the fusion of the bcr and c-abl genes on the Philadelphia chromosome in chronic myelogenous leukemia, with generation of an oncogenic fusion protein (13). More recently, a number of pediatric malignant tumors, such as soft-tissue sarcomas, were shown to be characterized by recurrent chromosomal translocations, frequently resulting in the fusion of genes (11). These fusion gene products often comprise potential transcription factors and nucleic acid-binding proteins. An example of this growing family of disrupted transcription factors is the (11;22) chromosomal translocation found in nearly 90% of Ewing's sarcomas (EWS) and primitive neuroectodermal tumor of childhood (14 -17). This translocation results in the fusion of the 5Ј region of the ubiquitously expressed EWS1 gene to the 3Ј region of the FLI-1 gene. This chimeric gene product (EWS/FLI-1) is an aberrant transcription factor which contains the transcriptional domain of EWS1, which is usually involved in protein-protein interactions, and the DNA-binding domain of FLI-1. Although capable of promoting tumorigenesis (18,19), the target gene(s) of EWS/FLI-1 are not yet fully known.
Similarly, desmoplastic small round cell tumor (DSRCT), a distinctive primitive tumor in children, is associated with a recurrent translocation, t(11;22)(p13;q12) (20 -22). Recently, a genomic DNA fragment containing an EWS1 and Wilms' tumor (WT1) fusion gene has been isolated from these tumors (23)(24)(25). Analyses of chimeric transcripts showed fusion of RNAs encoding the N-terminal domain of EWS1 to both alternatively spliced forms of the last three zinc fingers of the DNA-binding domain of WT1. As for the EWS/FLI-1 chimeric protein, the functional target gene(s) of EWS/WT1 remain unknown. WT1 itself is a tumor suppressor gene product that functions as a transcription factor and whose deletion or mutation has been implicated in the etiology of a subset of Wilms' tumors (26 -29). WT1 interacts with target promoters containing the consensus sequence GCGGGGGCG by means of four zinc finger motifs located at its C terminus, and usually suppresses their activity by means of its N-terminal activation domain (30 -32). We have previously shown that the IGF-I-R gene is overex-pressed in Wilms' tumor, consistent with the IGF-I-R gene promoter being a target for the inhibitory action of WT1 (33). Accordingly, WT1 expression in Chinese hamster ovary (CHO) cells results in a dose-dependent decrease in the activity of a co-transfected IGF-I-R promoter-luciferase reporter construct (34,35). This effect of WT1 involves the interaction of its zinc finger domain with multiple consensus binding sites in both the 5Ј-flanking and 5Ј-untranslated regions of the IGF-I-R gene.
Since the EWS/WT1 chimeric product obtains the three Cterminal zinc fingers of the DNA binding domain of WT1 it is anticipated that this fusion protein may modulate transcription of target genes containing WT1 binding motifs, such as the IGF-I-R gene. Due to the pivotal role of IGF-I-R in transformation events we have investigated the potential molecular mechanisms for the regulation of the IGF-I-R gene promoter by the EWS/WT1 fusion protein at the transcriptional level.

EXPERIMENTAL PROCEDURES
Cell Cultures and Plasmids-Saos-2 and G401 cells were obtained from the American Type Culture Collection (Rockville, MD). Saos-2 is a human osteogenic sarcoma-derived cell line and G401, initially considered a Wilms' tumor-derived cell line, is now thought to be a sarcomaderived cell line (36). Saos-2 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 5% calf serum and G401 were grown in McCoy's 5A medium, containing 10% fetal bovine serum. Media were supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, 2.5 g/ml Fungizone, and 2 mM L-glutamine.
For transient co-transfection experiments, a genomic IGF-I-R fragment containing 476 base pairs of 5Ј-flanking region and 640 base pairs of 5Ј-untranslated region was subcloned upstream of a promoterless firefly luciferase reporter gene, pOLUC. The promoter activity of this construct, p(Ϫ476/ϩ640)LUC, which includes most of the proximal promoter region, has been previously described (34). In some transient transfection experiments, the following fragments of the IGF-I-R gene promoter were used: Ϫ455/ϩ30, Ϫ118/ϩ640 ,and Ϫ40/ϩ640 (nucleotide no. 1 corresponds to the transcription start site of the IGF-I-R gene).
The WT1 expression vector (pCMV-hWT) was constructed by inserting a 2.1-kilobase pair human WT1 cDNA downstream of the CMV promoter in the pCB6ϩ vector, as described previously (32). The EWS1 expression vector (pCMV-EWS1) was constructed by subcloning a 2.2kilobase pair NotI-ClaI human EWS1 cDNA (a gift from Dr. Cristopher Denny, Department of Pediatrics, University of California, Los Angeles) into the NotI and ClaI sites of pCB6ϩ. A ␤-galactosidase expression vector (pCMV ␤-galactosidase, Clontech) was used as a control for transfection efficiency.
DNA Transfections-Saos-2 cells were transfected using a calcium phosphate transfection kit (5 Prime 3 3 Prime Inc., Boulder, CO). Each 100-mm dish received 10 g of reporter plasmid, 2.5 g of pCMV-␤-gal, and variable amounts of pCMV-EWS/WT1 (without or with KTS), or of pCMV-hWT or pCMV-EWS1. In each case the total amount of DNA transfected (15 g) was kept constant using pCB6ϩ DNA. Four hours after transfection, DNA-containing medium was changed to complete medium, and the plates were incubated for an additional 40 h at 37°C. G401 cells were transfected using 50 g of Lipofectin® reagent (Life Technologies, Inc.) in Opti-minimal Eagle's medium. Each dish received 1 g of reporter plasmid, 20 g of expression vector, and 4 g of pCMV-␤-galactosidase. After 16 h the Lipofectin reagent-containing medium was changed to McCoy's 5A medium, and the plates were incubated for an additional 48 h at 37°C. At the indicated times, cells were harvested, lysed, and luciferase and ␤-galactosidase activities were measured as described previously (37).
Gel Retardation Assays-A fragment of the IGF-I-R 5Ј-flanking re-gion extending from Ϫ331 to Ϫ40 was isolated by digestion of a genomic IGF-I-R clone with XmaI and PmlI. Following purification from agarose gels, the fragment was end-labeled with [␥-32 P]ATP, using T4 polynucleotide kinase, and separated from unincorporated nucleotide using Elu-Tip columns (Schleicher & Schuell). The following recombinant proteins were produced as histidine-fusion proteins in Escherichia coli and employed in gel retardation assays: EWS/WT1, EWS/WT1(ϩKTS), and the zinc finger domain of WT1, or WTZF (23). The EWS/WT1 fusion proteins contain 21 EWS1-encoded amino acids N-terminal of the fusion point and zinc fingers 2, 3, and 4 of WT1. They either lack or contain the three-amino acid insert between zinc fingers 3 and 4. Recombinant protein WTZF contains all four zinc fingers of WT1; the binding activity of this protein was described previously (30).
Gel retardation assays were performed by preincubating 0, 5, 20, and 50 ng of the purified proteins in 9 l of 20 mM Hepes, pH 7.5, 70 mM KCl, 12% glycerol, 0.05% Nonidet P-40, 100 M ZnSO 4 , 0.05 mM dithiothreitol, 1 mg/ml bovine serum albumin, and 0.1 mg/ml poly(dI⅐dC), with or without the indicated unlabeled competitor DNA, on ice. After 15 min, 75,000 dpm (0.2-1 ng) of the labeled fragment was added, and the reaction was incubated for an additional 10 min. Shifts in mobility were assessed by electrophoresis through a 5% polyacrylamide gel that was run at 250 V for 2 h at 4°C.
DNase I Footprinting-DNase I footprinting reactions were performed as described previously (34) using a labeled fragment of the IGF-I-R promoter extending from Ϫ331 to ϩ115. Maxam-Gilbert A and G sequencing reactions were run as markers.

Regulation of IGF-I-R Promoter Activity by the EWS/WT1
Fusion Protein-In previous studies, we have identified the promoter region of the IGF-I-R gene as a molecular target for tumor suppressor WT1 (33)(34)(35). We showed that WT1 binds both upstream and downstream of the IGF-I-R gene transcription start site and suppresses transcription of transfected IGF-I-R promoter constructs, as well as of the endogenous gene. Since the chimeric EWS/WT1 protein retains most of the DNA binding domain of WT1, we began to analyze the transcriptional regulation of the IGF-I-R promoter by EWS/WT1. For this purpose, we cotransfected the human osteosarcoma cell line Saos-2 with a reporter construct containing most of the proximal region of the IGF-I-R promoter upstream of a luciferase gene. This fragment, extending from nucleotide Ϫ476 in the 5Ј-flanking region to ϩ640 in the 5Ј-untranslated region, contains 12 WT1 binding sites. As shown in Fig. 1B, cotransfection of an EWS/WT1 expression vector lacking the 3-amino acid insert (pCMV-EWS/WT1) increased the activity of the IGF-I-R promoter ϳ340%. The effect of EWS/WT1 was dose-dependent, with maximal activity seen already at 2.5 g of expression vector (Fig. 1C). On the other hand, an EWS/WT1 expression vector including the 3-amino acid insert (pCMV-EWS/ WT1(ϩKTS)) was unable to stimulate promoter activity (Fig.  1B). For control purposes we cotransfected cells with a fulllength WT1 expression vector. As shown previously in other cell lines (34,35), WT1 suppressed promoter activity (46% of control levels) in Saos-2 cells. Interestingly, full-length EWS1 similarly reduced the activity of the IGF-I-R promoter to 58% of control values (Fig. 1B).
Cotransfections were also performed in G401, a Wilms' tumor-derived cell line in which we previously studied the tumorsuppressing effect of WT1. Similarly to Saos-2 cells, EWS/WT1 stimulated the activity of the IGF-I-R promoter, although the effect was much less pronounced. EWS/WT1(ϩKTS) had no effect, whereas both native WT1 and EWS reduced promoter activity (Fig. 2).
To determine the region of the IGF-I-R promoter responsible for the response to EWS/WT1, cotransfection experiments were performed in Saos-2 cells using a number of IGF-I-R promoter/ reporter plasmids containing different portions of 5Ј-flanking and 5Ј-untranslated regions (Fig. 3). Construct p(Ϫ455/ ϩ30)LUC, in which all six potential WT1 sites in the 5Ј-un-  Fig. 3) were cotransfected into Saos-2 cells with 2.5 g of the indicated expression vector using the calcium phosphate method. After 40 h the cells were harvested, and the levels of luciferase and ␤-galactosidase activities were measured. The luciferase values, normalized for ␤-galactosidase, are expressed as percentage over the activity of the empty pCB6ϩ expression vector. Experiments were performed between three and seven times, each time in duplicate. Bars are mean Ϯ S.E. C, dose-dependence of the effect of EWS/WT1. Ten micrograms of the p(Ϫ476/ϩ640)LUC were cotransfected in Saos-2 cells with increasing amounts of EWS/WT1 expression vector. Cells were processed as indicated above. translated region were removed, was stimulated by EWS/WT1 to an extent that was not significantly different from the effect seen with p(Ϫ476/ϩ640)LUC. Likewise, reporter construct p(Ϫ188/ϩ640)LUC, in which WT1 sites at positions Ϫ262/ Ϫ254, Ϫ250/Ϫ242, Ϫ220/Ϫ212, and Ϫ196/Ϫ188 were removed, was stimulated by the chimeric protein. Removal of the WT1 site at position Ϫ163/Ϫ155 in the p(Ϫ40/ϩ640)LUC construct resulted in a drastic reduction in basal promoter activity to levels resembling those of the promoterless luciferase reporter p0LUC (data not shown). Reporter plasmid p(Ϫ40/ϩ640)LUC was not responsive to EWS/WT1 (Fig. 3).
Interaction of EWS/WT1 Fusion Protein with the IGF-I-R Promoter-To analyze the interactions between EWS/WT1 and the promoter region of the IGF-I-R gene, gel retardation assays were performed using a fragment of the 5Ј-flanking region extending from Ϫ331 to Ϫ40, together with the following recombinant proteins: EWS/WT1, EWS/WT1(ϩKTS), and WTZF. EWS/WT1 proteins contained the 21-amino acid fragment of EWS1 located N-terminal to the fusion point and most of the C-terminal domain of WT1, including zinc fingers 2-4 ( Fig.   4A). Incubation of the labeled Ϫ331/Ϫ40 fragment with increasing amounts of the fusion protein lacking the KTS insert resulted in the appearance of five retarded bands (at 50 ng of protein), consistent with the number of WT1 sites footprinted in this region (Fig. 4B). Incubation with WTZF generated four bands, which suggests that the chimeric protein had an increased affinity for WT1 binding sites in comparison to the native WT1 protein. EWS/WT1(ϩKTS) had a largely reduced affinity for WT1 sites, a fact that is consistent with: (a) the inability of an expression vector encoding EWS/WT1(ϩKTS) to stimulate promoter activity (Fig. 1B) and (b) the results of early studies which suggested that the KTS insert interferes with the binding of native WT1 protein to the IGF-I-R promoter region (34).
The formation of the DNA-protein complexes was greatly diminished when the binding reactions were performed in the presence of a 30 -300-fold molar excess of the Ϫ331/Ϫ40 unlabeled fragment, but not when an upstream DNA fragment (Ϫ455/Ϫ331) lacking WT1 sites was used as competitor (Fig.  5A). Conversely, incubation of the labeled Ϫ455/Ϫ331 fragment with EWS/WT1 or WTZF proteins generated a single retarded band which appears to be the result of a nonspecific interaction, since its formation was not abolished by excess of cold competitor. When combined, these results indicate that interaction between EWS/WT1 and the IGF-I-R promoter occurs only at specific WT1 sites (Fig. 5B).
Finally, to determine whether the fusion protein and WT1 bind to the same or to different sites on the IGF-I-R promoter, DNase I footprinting was employed using a 32 P-labeled fragment extending from Ϫ331 to ϩ115, together with purified EWS/WT1 and WTZF proteins. As shown in Fig. 6, both proteins generated the same footprints in this region of the receptor gene, indicating that the fusion protein binds to the same sites than the native protein.

DISCUSSION
Tumor-specific chromosomal translocations resulting in chimeric transcription factors emerged as a general theme in oncogenesis (11). The modular organization of transcription factors is disrupted by the chromosomal event, and gain-offunction mutant genes harboring motifs derived from unrelated genes are usually generated. Many of these fusion proteins have been shown to be highly oncogenic.
DSRCT is a very aggressive clinicopathological subcategory of small cell round tumors (SCRT) which occurs most frequently in adolescent males and is mainly circumscribed to the abdomen (38). Histopathological studies showed the presence of epithelial, mesenchymal, and neuronal elements, similarly to triphasic Wilms' tumors which also contain epithelial, mesenchymal, and stromal cells (39). Cytogenetic analyses revealed the presence of a recurrent chromosomal translocation, The chimeric EWS/WT1 is a member of a family of fusion proteins which involve the N-terminal domain of EWS1, and which includes EWS/FLI-1, EWS/ERG, and EWS/ATF1 (15,40,41). The EWS1 gene encodes an ubiquitous 656-amino acid protein of unknown function. EWS1 comprises an N-terminal domain homologous to eukaryotic RNA polymerase II and a C-terminal domain homologous to RNA-binding domains (42). The common feature of these oncogenic proteins, thus, is the presence of the highly potent transcriptional activation N-terminal domain of EWS1 fused to a novel DNA binding domain.
The results of this study demonstrate that, whereas EWS1 and WT1, individually, suppressed the activity of the IGF-I-R promoter, the EWS/WT1 fused gene sustained a gain-of-function mutation which conferred upon its product the ability to transactivate the IGF-I-R promoter. The mechanism for EWS1 suppression of IGF-I-R promoter is presently unknown. WT1 has been previously demonstrated to bind to specific sites in the 5Ј-flanking and 5Ј-untranslated regions of the IGF-I-R promoter and to suppress its activity in a manner which is directly dependent on the number of sites involved (34). In the case of EWS/WT1, however, removal of all the binding sites in the 5Ј-untranslated region does not decrease its stimulatory effect, thus suggesting that WT1 sites in the 5Ј-flanking region are mainly responsible for the effect of the fusion protein. Interestingly, construct p(Ϫ188/ϩ640), which lacks the four N-terminal WT1 sites in the 5Ј-flanking region was still stimulated by EWS/WT1, suggesting that the site at position Ϫ163/Ϫ155 had a preponderant effect in the context of this promoter. We may speculate that, from the standpoint of WT1, loss of its Nterminal domain in EWS/WT1 and replacement by the potent activation domain of EWS1, confers upon it a very strong oncogenic potential. From the standpoint of EWS1, loss of its C-terminal domain in EWS/WT1 abrogates its RNA-binding activity. Fusion to the zinc finger domain of WT1 converts EWS1 from an RNA-binding to a DNA-binding protein with a well defined set of target genes. Such gain-of-function changes are not unprecedented. WT1 itself, for example, has been shown to be an inhibitory factor in the presence of p53, whereas in the absence of this protein it stimulates the activity of target genes (43).
As previously shown using a synthetic early growth response gene-binding site oligonucleotide, the results of gel shift assays indicate that the affinity of the fusion protein for binding to the IGF-I-R promoter is higher than that of WTZF itself. This increased affinity of EWS/WT1 can be explained by the lack of zinc finger 1, which has been previously shown to destabilize the DNA binding activity of WT1 (44).
The IGF-I-R has a central role in transformation and proliferation events, and its presence is a fundamental prerequisite for the transforming ability of simian virus 40 large T antigen, Ras, and other oncogenes (1, 8 -10). Under most physiological conditions the expression of the potent IGF-I-R promoter is negatively controlled by a number of tumor suppressors, including p53 and WT1 (45). Inhibitory control of IGF-I-R gene expression by WT1 can be overcome by the pathologic fusion of EWS1 to WT1, an event which abrogates the tumor suppressor A, competition of binding with excess unlabeled probe. 50 ng of the purified EWS/WT1 protein were incubated with the 32 P-labeled Ϫ331/Ϫ40 probe, in the presence of increasing amounts of the unlabeled Ϫ331/Ϫ40 probe, or in the presence of excess of the upstream Ϫ455/ Ϫ331 fragment, which does not include any WT1 binding site. B, the region extending from Ϫ455 to Ϫ331 was end-labeled and used in binding reactions with the purified EWS/WT1 and WTZF, in the absence or presence of unlabeled Ϫ455/Ϫ331 competitor. effect of WT1 and which generates an oncogenic chimera. By means of the WT1-derived DNA binding domain, EWS/WT1 is able to recognize and activate the same set of target genes which were previously negatively regulated by WT1.
In conclusion, we have presented evidence that the IGF-I-R promoter is a molecular target for the EWS/WT1 fusion protein. Relief from negative regulation by WT1 and activation by EWS/WT1 constitute a novel paradigm in tumorigenesis.