The Wilms’ Tumor Gene Product, WT1, Represses Transcription of the Platelet-derived Growth Factor A-chain Gene*

The Wilms’ tumor locus on chromosome llp13 con- tains a tumor suppressor gene, wtl, which encodes a DNA binding protein (WTl) with four zinc fingers and a glutamine-proline-rich N terminus and which functions as a repressor of transcription. The platelet-de- rived growth factor (PDGF) A-chain gene encodes a potent growth factor, which is expressed in high levels in a number of tumor cell lines. We initiated a search for WT1 target genes and now report that WT1 strikingly represses transcription of the PDGF A-chain gene in transient transfection assays and that the WT1 protein interacts directly with a highly G+C-rich re- gion of the PDGF A-chain promoter in gel mobility shift assays. The results suggest that WT1 may function to repress expression of the PDGF A-chain gene and that loss of this or related repressor activities may contribute to the abnormal growth of Wilms’ tumors.

The Wilms' tumor locus on chromosome l l p 1 3 contains a tumor suppressor gene, w t l , which encodes a DNA binding protein (WTl) with four zinc fingers and a glutamine-proline-rich N terminus and which functions as a repressor of transcription. The platelet-derived growth factor (PDGF) A-chain gene encodes a potent growth factor, which is expressed in high levels in a number of tumor cell lines. We initiated a search for WT1 target genes and now report that WT1 strikingly represses transcription of the PDGF A-chain gene in transient transfection assays and that the WT1 protein interacts directly with a highly G+C-rich region of the PDGF A-chain promoter in gel mobility shift assays. The results suggest that WT1 may function to repress expression of the PDGF A-chain gene and that loss of this or related repressor activities may contribute to the abnormal growth of Wilms' tumors.
Wilms' tumor (WT) is an embryonal cell malignancy of kidney, which occurs both in familial and in sporadic forms (1). The association of Wilms' tumor with aniridia, mental retardation, and urogenital abnormalities (the WAGR syndrome) (2,3) allowed mapping of a potential tumor suppressor gene to chromosome llp13-15. Recently, a candidate Wilms' tumor gene, w t l , was isolated from the llp13 locus and characterized (4,5). w t l encodes a protein (WT1) with four zinc finger domains and a characteristic glutamine-and proline-rich N-terminus. These structural motifs are associated with sequence-specific DNA binding transcriptional regulation and respectively (4,5). The WT1 protein recognizes DNA sequences containing the core element, 5'-GCGGGGGCG-3' (6), a sequence similar to the consensus sequence recognized by the egr-1 gene product, EGR-1, also called NGFI-A, Zif-268, Krox-24, and T1S8 (8) (reviewed in Ref. 9). The WTl protein functions as a transcriptional repressor when bound to the EGR-1 consensus sequence (7).
The PDGF' A-chain gene encodes a potent mitogen and chemoattractant for cells of mesenchymal origins. It is selectively expressed in high levels in a number of transformed cell lines and may function to promote the growth of transformed cells (reviewed in Ref. 10). In order to pursue mechanisms by which expression of the PDGF A-chain gene may be regulated and a possible role of the PDGF A-chain in transformed cell growth, we isolated, sequenced, and analyzed the human PDGF A-chain gene promoter region (11). A DNA sequence upstream of the TATA box is extremely G+C-rich and is required for efficient expression of the PDGF A-chain gene (12). We now show that the wtl gene product binds directly to this highly G+C-rich region and that expression of the PDGF A-chain gene is dramatically reduced by coexpression of the w t l gene.

MATERIALS AND METHODS
Plasmid Constructions-Restriction endonucleases and DNA ligases were obtained from Bethesda Research Laboratories.
The expression vectors used contained the full-length protein coding regions of the human wtl or murine egr-l genes, the cytomegalovirus immediate-early promoter, and the SV-40 polyadenylation signal (7). The reporter vector 840PDGF contains the DNA sequences -446 to +388 relative to the transcription start site of the PDGF A-chain gene promoter fused upstream of Basic CAT (Promega) (Fig. 1). Clone A34PDGF contains a 34-bp deletion from -34 to -72 ( Fig. 1) and was prepared as described elsewhere (12).
Cell Cultures, DNA Transfections, and CAT Assays-NIH3T3 fibroblasts or human embryonic kidney 293 cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum. The cells were plated a t a density of 1 X lo6 cells/lOO-mm dish 24 h prior to transfection and then transfected by calcium phosphate co-precipitation (13) with 840PDGF (2.5 pg), A34PDGF (2.5 pg), or expression plasmid (5 or 15 pg) and a 0-galactosidase expression plasmid (1 pg) to establish transfection efficiency. The expression vector added in each transfection was kept constant by addition of CMV vector alone to a total of 20 pg. Forty-eight hours after transfection, cell extracts were prepared, aliquots were normalized for transfection efficiency by assay of 0-galactosidase activity (14), and CAT activity was determined (15).
The running buffer consisted of 45 mM Tris borate (pH 8.3), 45 mM boric acid, and 1 mM EDTA. The gels were dried and analyzed by autoradiography. The oligonucleotides of the PDGF A-chain promoter and their competitors are illustrated in the legend to Fig. 3. The oligonucleotides were labeled by filling the recessed 3' ends with avian myeloblastosis virus reverse transcriptase and [ C X -~~P I~G T P (Du Pont-New England Nuclear); specific activities ranged from 0.2 to 1.0 X lo6 cpm/pmol DNA. Unlabeled competitor DNAs, when appropriate, were added at the same time as the labeled DNAs.

RESULTS
The human PDGF A-chain gene promoter contains a highly G+C-rich sequence immediately 5' to the TATA box (Fig.  1B). This sequence includes a 13-bp oligo(dG):oligo(dC) motif and three continuous SP1 binding sites, forming a large "GC box." The GC box also contains two uninterrupted sequences (Fig. lB, underlined), which are recognized by the wtl gene product (6), sequences which also are similar to consensus sequences recognized by the EGR-1 protein (7) (Fig. 1). The mutant A34PDGF CAT, which retains 10 of the 13-bp dG:dC motif, one S p l binding site (5'-GGGCGG-3'), and no obvious core consensus for WT1 binding (Fig. l ) , retains less than 20% of the promoter activity of the PDGF A-chain gene (Fig.  2).
T o determine if the WT1 and EGR-1 regulated transcription of the PDGF A-chain gene, expression vectors (Fig. lA), which contained the full-length coding regions of wtl (CMV-WT1) and egr-1 (CMV-EGR-1) were co-transfected with 840 PDGF (wild type) CAT or with A34PDGF CAT reporter plasmids. CAT activity was then determined in lysates of NIH3T3 fibroblasts co-transfected with 840PDGF CAT plus CMV-ECR-1 or 840PDGF CAT plus CMV-WT1.
Co-transfection of the WT1 expression vector with the intact PDGF A-chain promoter repressed expression of the reporter CAT activity to very low levels whereas the egr-1 gene product reduced reporter CAT activity only slightly ( Fig.  2 A ) . A CMV vector containing only the zinc finger region of WT1 (CMV-WTZF) did not repress transcription of 840PDGF CAT (Fig. 2 A ) , consistent with earlier observations that the N-terminal glutamine-proline-rich repression domain is required for this activity (7). Thus, the structural requirements for repression of PDGF A-chain gene by W T l are similar to what has been previously described (7). That is, an intact zinc finger domain and N-terminal repression domain of WT1 are required as well as binding sites for WT1 in the promoter region of the target gene. The WT1 gene product does not act as a generalized repressor of all promoters since reporter plasmids lacking WT1 binding sites or containing mutated forms of the WT1 site are not repressed by W T l (7,18). Basal activity of A34PDGF CAT which lacks two S p l sites but which retains G+C-rich residues was -20% as active as 840 PDGF CAT. Expression of the A34PDGF CAT con-  (1 p g ) as an internal control for transfection efficiency. The total amount of CMV vector in each transfection mixture was kept constant at 20 mg by addition of CMV vector alone. Forty-eight hours after transfection, cell extracts were prepared and aliquots normalized for transfection efficiency via assay of b-galactosidase activity were used for determination of CAT activity. After autoradiovaphic exposure, the thin layer chromatography plates were scanned and percent conversion values were calculated. struct was also substantially inhibited by co-expression of the WT1 protein, despite the lack of an obvious core consensus site for WT1. However, the continued ability of WT1 to repress the A34PDGF CAT construct can be explained by the observation that the WTl protein still recognizes a truncated version of the GC box present in A34PDGF CAT (see Fig.  3C) but not when the remaining sequences within the GC box are deleted (data not shown).

FIG. 2. Regulation of transcription from the PDGF A-chain CAT reporter plasmids by ECR-1 and WTl. A, calcium phosphate-mediated transfections were performed in murine NIH3T3 fibroblasts and 29.1 cells. Each dish of cells was transfected with R40 PDGF (2.5 pg) or A34 PDGF (2.5 pg). the indicated expression plasmid (5 pg), and a &galactosidase expression vector
The influence of the CMV-WT1 and CMV-EGR-1 genes on the expression of 840PDGF CAT was also tested by cotransfection a t increasing concentrations of CMV-WT1 and CMV-EGR-1 expression vectors. Even at the lowest concentration of the WT1 expression vector used, 840 PDGF CAT activity was substantially repressed in both NIH3T3 cells and human embryonic kidney-derived 293 cells (Fig. 2R, Table I).
Interestingly, co-transfection of egr-1 expression vector with the 840 PDGF CAT plasmid consistently resulted in a small but positive increment of expression of CAT in 293 cells, while in NIH3T3 cells EGR-1 consistently demonstrated a repressor effect on 840PDGF-CAT expression (Table I). This result suggests that the influence of EGR-1 on the PDGF Achain promoter is cell type-specific and that EGR-1 has the  FIG. 3. Gel retardation and competition assays. A, the 30-bp oligonucleotides containing the GC box was end-labeled and mixed with 10 and 50 ng of purified proteins WTAZF (WT-ZF(delF3)), WTI, Spl, and EGR-1. The labeled DNA-protein mixture was fractionated on native polyacrylamide gel, dried, and exposed to x-ray film. R, the competition assay of WT1-DNA complex was done with the unlabeled probe and the competitor. Spl site, 5"ATTCGATCG-GGCGGGGCGAGC-3'; EGR-1 site, 5'-CGCCCTCGCCCCCGCGC-CGGG-3'. C, the 33-bp oligonucleotide including the deleted G+Crich region was end-labeled, and a gel retardation assay was performed. The protein-DNA complex is indicated by arrows. capability of being both a positive and a negative regulator of transcription depending on cell type.

WT1 Represses Expression of PDGF A-chain Gene
To further characterize the interaction of the WT1 protein with the GC box of the PDGF A-chain promoter, a 30-hp oligonucleotide (5'-GGCGGCGGGGCCCGCGGCGGCGG-AGGGGCG-3') containing the GC box was synthesized, endlabeled, incubated with purified WT1, SP1, and EGR-1 proteins, and analyzed in gel mohility shift assays (Fig. 3A ). A single distinct protein-DNA complex was identified when purified WT1 protein was mixed with the end-laheled prohe (Fig. 3 A ) . In a control experiment, a mutated WTl protein, WT-ZF(delF3), which lacks the third zinc finger domain, did not bind to the GC box (Fig. 3 A ) . EGR-1 protein DNA complexes also were observed.
To test the specificity of W T l / G C hox interactions, competition binding assays were done with unlaheled DNA fragments, including the unlabeled prohe itself, a fragment with a typical consensus Spl binding site (Fj'-GGGCG(;-3'), and a fragment with a typical consensus EGR-I hinding site (5'-GCGGGGGCG-3'). The unlaheled homologous prohe and the EGR-1 consensus sequence were effective competitors with the WT1 protein/GC box interaction whereas the SPl sequence did not compete (Fig. 3 R ) . Since WTI repressed the activity of the A34PDGF CAT plasmid (Fig. 21, we synthesized the oligonucleotide .5'-AGCTTCCG(;GGAGGCGGG-GGGGGGGCGGCGGCA-3', which contains the mutated GC box, and performed gel shift assays (Fig. X ) . Importantly, the WT1 and EGR-1 proteins hound to this sequence to a similar extent as the wild-type GC box used in Fig. 3A. These results suggest that WT1 and EGR-1 recognize the DNA sequence 5'-GGGGGGGCG-3' as well as the canonical core consensus 5'-GCGGGGGCG-8'. Thus, hoth EGR-1 and WTI proteins interact with the GC box of the PDCF A-chain promoter in a sequence-specific manner.

DISCUSSION
T h e wtl gene is a potential tumor suppressor gene whose homozygous inactivation is associated with development of Wilms' tumor (3,4). The u!tl gene encodes a protein with four zinc fingers and other domains characteristic of transcription factors (4). However, hoth the normal function(s) of WT1 and the loss of function associated with development of embryonal cell tumors are unknown. The PDGF A-chain gene encodes a potent mitogen that is expressed in very high levels in a number of transformed cells and is highly temporally and spatially regulated during development. The mechanisms which regulate transcription of the PDGF A-chain gene are only beginning to be understood, and its in oivo functions remain to be fully established. The data presented here strongly suggest that the WTI protein binds to the GC box of the PDGF A-chain gene and functions as a potent repressor of expression of the PDGF Achain gene in transient transfection assays. The ECR-1 protein also recognizes the GC box of the PDGF A-chain gene and functions as both an activator or repressor of transcription depending on the cell line used for assays. These results thus suggest that regulation of function of the PDGF A-chain gene promoter is highly complex and that WTl is one of the transcription factors that has the potential to strikingly repress its expression in oioo. Since the w t l gene product functions as a potent repressor of transcription, w t l expression or its function may he significantly reduced or ahsent in tumors that contain high levels of PDGF A-chain. One function of tumor suppressor genes may he to directly repress the expression of important growth factor-encoding genes which may act in an autocrine manner during normal development by guest on March 24, 2020 http://www.jbc.org/ Downloaded from WT1 Represses Expression of PDGF A-chain Gene (10,16,17). In support of this, we have recently shown that the insulin-like growth factor I1 gene (IGF-II), a gene which is overexpressed in Wilms' tumors, is also a target for repression by the WT1 protein (18).
It is important to note that the WT1 and EGR-1 proteins are only two of the many transcription factors that may regulate PDGF A-chain gene expression during growth and development. The PDGF A-chain gene is temporally and spatially regulated during growth and development (16). The highly restricted pattern of WT1 expression in developing kidney and urogenital systems (19,20) suggests that WT1 plays a major role in PDGF A regulation in these organs, yet it also suggests the presence of other negative regulators of PDGF A expression in other tissues and at other times of embryonic development. These factors may be additional WT1-like or EGR-1-like molecules, which recognize the GC box. In this context, it is interesting to note that the EGR-1 protein (which is expressed in many different tissues and cell types (9)) can function as both an activator and repressor of transcription depending on cell type. Identification of additional transcription factors, which negatively regulate PDGF A-chain expression, may lead to the isolation of new tumor suppressor genes which, like WT1, function in a highly tissuespecific manner.