Transactivation of an Intronic Hematopoietic-specific Enhancer of the Human Wilms’ Tumor 1 Gene by GATA-1 and c-Myb*

The Wilms’ tumor 1 gene (WT1) encodes a zinc-finger transcription factor which is expressed in a tissue-specific manner. Our studies indicate that in addition to the promoter, other regulatory elements are required for tissue-specific expression of this gene. A 258-base pair hematopoietic specific enhancer in intron 3 of the WT1 gene increased the transcriptional activity of the WT1 promoter by 8–10-fold in K562 and HL60 cells. Sequence analysis revealed both a GATA and a c-Myb motif in the enhancer fragment. Mutation of the GATA motif decreased the enhancer activity by 60% in K562 cells. Electrophoretic mobility shift assays showed that the GATA-1 protein in K562 nuclear extracts binds to this motif. Cotransfection of the enhancer containing reporter construct with a GATA-1 expression vector showed that GATA-1 transactivated this enhancer, increasing the CAT reporter activity 10–15-fold. Similar analysis of the c-Myb motif by cotransfection with the enhancer CAT reporter construct and a c-Myb expression vector showed that c-Myb transactivated the enhancer by 5-fold. A DNase I-hypersensitive site has also been mapped in the 258-base pair enhancer region. These data suggest that GATA-1 and c-Myb are responsible for the activity of this enhancer in hematopoietic cells and may bind to the enhancer in vivo.

Wilms' tumor is a pediatric nephroblastoma that is one of the most frequent solid tumors in children (1). Mutation or deletion of both copies of the Wilms' tumor 1 (WT1) 1 gene is associated with Wilms' tumor, implying that WT1 is a tumor suppressor gene (2,3). Mice with homozygous WT1 mutations fail to develop kidneys and gonads (4), suggesting that WT1 has a crucial role in early urogenital development. However, WT1 mutations have also been associated with other tumors (3), including mesotheliomas (5), juvenile granulosa cell tumor of the ovary (6), and secondary acute myelogenous leukemia (7). Recently, five WT1 mutations were found in four of 36 patients with sporadic acute leukemia (8), which is a mutation rate comparable to that found in Wilms' tumors.
WT1 has been mapped to human chromosome 11p13, cloned, and shown to encode a zinc-finger transcription factor (9,10). Recent studies have shown that WT1 is expressed in normal hematopoietic cells, at higher levels in immature hematopoietic cells (e.g. in CD34 ϩ bone marrow and fetal liver cells) (8,11,12) than in differentiated mature blood cells (12). It has been reported that in normal human bone marrow, fluorescenceactivated cell sorted CD34 ϩ /CD33 Ϫ /Lin Ϫ cells have levels of WT1 gene expression a hundred times greater than those of fluorescence-activated cell sorted CD34 ϩ CD33 ϩ Lin Ϫ cells (11). WT1 is strongly expressed in the peripheral blood of many patients with acute leukemia and in the blast-crisis (BC) phase of chronic leukemias but is absent in the chronic phase of chronic leukemias (13). The average levels of WT1 expression were more than 20 times higher for CD19 ϩ /CD20 Ϫ pro-B-cell acute lymphoblastic leukemia than for CD19 ϩ /CD20 ϩ pre-Bintermediate B-cell acute lymphoblastic leukemia, indicating that WT1 expression is associated with immature B phenotypes of acute lymphoblastic leukemia cells (11). In chronic myelogenous leukemia (CML), WT1 expression levels are clearly associated with the clinical phase, and the levels increase as the clinical phase progresses. This indicates that in CML, WT1 gene expression is also associated with immature leukemic cells. Thus, WT1 may be a new prognostic factor and a new marker for the detection of minimal residual disease in acute leukemia (11). It has also been reported that the WT1 gene is down-regulated during terminal differentiation of both K562 cells (a human erythroleukemia cell line derived from a patient with BC-CML) and HL60 cells (a human myelocytic leukemia cell line) (14,15). It appears, therefore, that WT1 gene expression is associated with the immature cells from which leukemic cells originate.
To understand the regulation of WT1 in hematopoiesis and leukemia, we examined the elements involved in the transcriptional control of the WT1 gene (16). We have identified three transcription start sites and an essential promoter region of the WT1 gene. The WT1 promoter is a member of the GC-rich, TATA-less, CCAAT-less class of polymerase II promoters (16). Whereas the WT1 promoter is similar to other GC-rich tumor suppressor gene promoters, the WT1 expression pattern is tissue restricted, unlike RB and p53 patterns. However, the WT1 GC-rich promoter functions in all cell lines tested, independent of WT1 expression. This finding suggests that the tissue-specific expression of WT1 is modulated by additional regulatory elements. We previously identified a transcriptional enhancer at the 3Ј end of the gene, more than 50 kb downstream of the promoter (16). This 3Ј-enhancer increases the basal transcription rate of the WT1 promoter in the human erythroleukemia cell line K562 but not in the non-hematopoietic cell lines tested. The hematopoietic transcription factor GATA-1 binds and transactivates the 3Ј-enhancer (17).
Here we report the identification of a hematopoietic-specific enhancer in the third intron of the WT1 gene. While both GATA-1 and c-Myb could transactivate this intronic enhancer, the previously identified 3Ј-enhancer lacks c-Myb binding motifs (17). Consequently, the two WT1 enhancers functioned differently in hematopoietic cells of various lineages. The intronic enhancer was very active in myelocytic leukemia HL60 cells but only weakly active in erythroleukemia HEL cells. Conversely, the 3Ј-enhancer is very active in HEL cells but only weakly active in HL60 cells.
To assess the importance of the GATA sites, the most 5Ј-GATA motif (GATA-A) (Fig. 2) in the intronic enhancer CAT reporter construct was mutagenized. pCB.7mGATA-Ae258, was generated by two-step PCR amplification using pCATe258 as the template with two pairs of primers: the 5Ј-WT1e258A primer and the 3Ј-GATA mutant primer (5Ј-AGGCAACCCTAAGGCAGACGCGGGCGGCCG-3Ј) (mutant bases are underlined); and the 5Ј-GATA mutant primer (5Ј-CGGCCGC-CGCGTCTGCCTTAGGTTGCCT-3Ј) and 3Ј-vector primer CPB. The two PCR products generated by the two pairs of primers overlapped, and both contained the mutant GATA-A motif (mGATA-A). We annealed the products and extended them by mutually primed synthesis. This product was then PCR amplified with primers WT1e258A and CPB to generate the full-length intronic enhancer containing mGATA-A, which we then cut with BamHI and inserted into the BamHI site of pCB.7PH, to create pCB.7mGATA-Ae258. The sequence of each PCR clone was confirmed by automated sequencing (Applied Biosystems, Foster City, CA).
Transfections and CAT Assays-K562 cells were transfected by electroporation using a modification of the protocol of Chu et al. (19) adapted for leukemia cells (20). Cells (5 ϫ 10 6 ) were electroporated with the Gene Pulser (Bio-Rad) at 230 V and 960 microfarads in 200 l of serum-free medium containing 2.5-5.0 g of plasmid DNA and seeded into T-25 flasks. As an internal control for transfection efficiency, K562 cells were also cotransfected with 2.5 g of the expression vector pSV 2 ␤-Gal (Promega, Corp., Madison, WI). The cells were harvested 48 h after transfection, and cytoplasmic extracts were prepared by three cycles of freeze-thawing (21). The ␤-galactosidase activity in K562 cells was determined by standard methods (22). After thin-layer chromatography, the acetylated [ 14 C]chloramphenicol was quantitated by measuring the radioactivity with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). Aliquots of protein with equivalent amounts of ␤galactosidase activity were assayed for CAT activity (21). The relative activities from at least three different experiments were averaged.
The enhancer constructs were tested for CAT activity in K562 cells as HeLa and Saos-2 cells were cotransfected as described above, except that increasing amounts of the mouse GATA-1 cDNA expression construct (23) were added to each transfection of 5.0 g of CAT reporter DNA. The CAT reporter construct pCB.1e258 contains three potential GATA-binding sites and one c-Myb binding site in the 258-bp intronic enhancer ( Fig. 2), but no potential GATA-binding sites in the 104-bp WT1 minimal promoter (16). The DNA concentrations were held constant by the addition of the empty expression vector DNA (23). For the c-Myb transactivation experiments, 8.0 g of either a c-Myb cDNA expression construct or mutant c-Myb cDNA construct (lacking the DNA-binding domain) (24) or the empty vector were added to each transfection with 5.0 g of CAT reporter DNA.
For competition assays, the unlabeled double-stranded oligonucleotides GATA-A, mGATA-A, and GATA consensus (GATA con) (5Ј-CACT-TGATAACAGAAAGTGATAACTCT-3Ј) (Santa Cruz Biotechnology, Santa Cruz, CA) were added to each reaction mixture first. After incubation with the K562 nuclear extracts for 15 min at room temperature, the end labeled probe was added and the mixture was incubated an additional 15 min at room temperature and analyzed as described above. For antibody ablation experiments, 0.5 and 1.0 g of anti-c-Myb rabbit polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA), anti-GATA-1 rat monoclonal antibody (Santa Cruz Biotechnology), anti-GATA-2 goat polyclonal antibody (Santa Cruz Biotechnology), or anti-GATA-3 mouse monoclonal antibody (Santa Cruz Biotechnology), were added before the addition of labeled probe and incubated for 15 min at room temperature. After the addition of labeled probe, the reaction mixture was incubated for an additional 15 min at room temperature and analyzed as described above.
Isolation of Nuclei and DNase I Digestion-Nuclei were isolated essentially as described by Levy-Wilson et al. (27). Suspension cells and confluent monolayers from the K562 and HeLa cells were harvested by centrifugation at 2,000 rpm. The cell pellet was washed in 20 ml of reticulocyte standard buffer (RSB)/sucrose (10 mM Tris, pH 7.5, 10 mM NaCl, 1 mM MgCl 2 , 0.25 M sucrose) containing 0.1 mM phenylmethylsulfonyl fluoride by gentle homogenization with a Dounce homogenizer, followed by a low-speed spin as described above. Cells were suspended in 3 volumes of RSB/sucrose/phenylmethylsulfonyl fluoride and lysed by the addition of Nonidet P-40 at a final concentration of 0.5% for 5 min on ice. The nuclear pellet was recovered by centrifugation at 4,000 rpm. Nuclei were washed once with RSB/sucrose/phenylmethylsulfonyl fluoride, and their integrity and purity were checked by light microscopy after they had been stained with trypan blue (0.4%).
Intact nuclei were suspended at a DNA concentration of 1 mg/ml in RSB/sucrose/phenylmethylsulfonyl fluoride. DNase I (10 units/l, Boehringer Mannheim) were added to various final concentrations, and the mixtures were incubated for 10 min at 37°C. The reactions were terminated by the addition of an equal volume of 2 ϫ lysis buffer (0.6 M NaCl, 20 mM EDTA, 20 mM Tris hydrochloride, pH 7.5, 1% sodium dodecyl sulfate). RNase A (U. S. Biochemical Corp.) was added at 40 g/ml for 30 min at 37°C, and then proteinase K was added at 100 g/ml for 4 -16 h at 37°C. DNA was purified by extraction with an equal volume of phenol, phenol/chloroform (1:1, v/v), and chloroform, followed by precipitation with isopropyl alcohol at room temperature. The DNA was recovered by centrifugation, washed once with 70% ethanol, and suspended in 100 l of H 2 O.
Gel Electrophoresis and Southern Blotting-Aliquots containing 30 g of DNA digested as above were digested with both EcoRI and XbaI. DNA was purified by extraction with an equal volume of phenol/chloroform and chloroform followed by precipitation with ethanol at Ϫ20°C for 2 h. The DNA was recovered by centrifugation, washed once with 70% ethanol, and suspended in 30 l of TE. Purified DNA were electrophoresed in 1.2% agarose gels, and transferred by vacuum blotter (Bio-Rad) onto Zeta-Probe ® GT blotting membranes (Bio-Rad) according to the manufacture's manual. All hybridizations were done according to the manufacture's instruction manual. The probe (EcoRI/SphI fragment, Fig. 9B) was 32 P-labeled using the Megaprime DNA labeling system kit (Amersham). The filters were exposed to x-ray film for 1-3 days. The sizes of the hybridizing bands appearing in the autoradiogram were determined by using HindIII/EcoRI-digested marker DNA fragments as standards in each gel.

Identification of an Intronic
Hematopoietic-specific Enhancer-Analysis of distant regulatory elements and intragenic enhancers for WT1 have been hampered by the size of the WT1 gene. We therefore focused on two regions often reported to contain regulatory elements: the 3Ј-and 5Ј-flanking regions and the first few introns (Fig. 1A).
We determined the BamHI and BglII sites in the WT1 genomic sequence in the 5Ј-flanking region cosmid cB5-2 ( Fig.  1A) and subcloned small BamHI fragments from the cosmid into pCB.7PH, in which the CAT gene is under control of the WT1 promoter. Initially, we identified an enhancer element in the region between the 3Ј end of exon 2 and the 5Ј end of intron 3 (Fig. 1A). This 1.5-kb enhancer increased the basal transcription from the WT1 promoter by 5.84-fold in the human BC-CML cell line K562 but not in 293 (a human embryonic kidney cell line) or II-14 (a rat mesothelioma cell line), which also expressed WT1 (Fig. 1A). This suggested that the activity of the 1.5-kb fragment was hematopoietic specific. We also screened about 10 kb of the 3Ј-flanking region of WT1 in 293 and II-14 cells but found no enhancer elements (data not shown).
Deletion Analysis of the Intronic Enhancer-To delineate the enhancer region, the 1.5-kb fragment was subcloned into pCAT ® -Promoter to generate pCATp1.5, and the deletion constructs derived from it were transfected into K562 cells (Fig.  1B). We found that the 1.5-kb fragment could enhance the activity of the WT1 promoter, but not the SV40 promoter. This lack of activity can be partially explained by the presence of an SV40 repressor in the 5Ј-enhancer region in pCATp678 (Fig.  1B). Therefore, we examined the 3Ј-region (in, pCATp833), which had strong enhancer activity (4.54 times to that of the SV40 promoter). Further deletion of the 833-bp fragment showed that the most 3Ј 258-bp fragment was sufficient for enhancer activity in K562 cells. When the fragment was transfected into K562 cells, it increased the basal transcription level of the SV40 promoter by 3.29 times and had enhancer activity nearly equivalent to that of the SV40 enhancer (data not shown). The 258-bp enhancer was located in WT1 intron 3, approximately 11 kb downstream of the promoter (Fig. 1A). Sequence analysis showed that this 258-bp fragment contains many potential binding sites for transcription factors, including Ets-1, GATA, c-Myb, and AP-2 (Fig. 2). It is possible that one or more of these transcription factors facilitate the expression of WT1 in hematopoietic cells.
Lineage Specificity of the Intronic and 3Ј-WT1 Enhancers-To determine the tissue specificity of the intronic enhancer, the 258-bp enhancer was cloned into pCB.7PH 3Ј of the WT1 promoter to generate the construct pCB.7e258. The 258-bp enhancer increased basal transcription of the WT1 promoter in K562, HL60, and HEL cells but not in 293, HeLa, and CEM cells (Table I). This indicates that the 258-bp enhancer was hematopoietic specific. Although the 258-bp fragment functioned in three hematopoietic cell lines (K562, HL60, and HEL), the enhancer had different degrees of activity in each cell line: it increased the basal transcription levels of the WT1 promoter by 8 -9-fold in K562 cells, 5-6-fold in HL60 cells, and only 2-3-fold in HEL cells. This suggested the intronic enhancer had the strongest activity in myeloid-lineage uncommitted progenitor cells.
To compare the tissue specificity of the intronic enhancer and the 3Ј-enhancer previously characterized by Wu et al. (17), the 3Ј-enhancer was transfected into HL60 and HEL cells. The 3Ј-enhancer increased the basal transcription levels of the WT1 promoter 6-fold in HEL cells but only 2-fold in HL60 cells (Table I). This indicates that unlike the intronic enhancer, the 3Ј-enhancer functioned better in cells with erythroid characteristics (K562 and HEL) than in cells with only myelocytic characteristics (HL60).
Identification of the Minimal WT1 Intronic Enhancer-To determine whether the intronic enhancer can function in an orientation-independent manner, the 258-bp fragment was cloned into pCB.7PH in both the normal and reverse orientations to generate pCB.7e258ϩ and pCB.7e258Ϫ. Both constructs had strong activity (greater than 8-fold activation) in K562 cells (Fig. 3). We also examined the effect of the intronic enhancer on the minimal promoter, independent of the multiple regulatory elements located within the full-length promoter. The 258-bp fragment was subcloned into pCB.1 (17) to generate pCB.1e258, which contains a 104-bp minimal promoter. The intronic enhancer increased the activity of the minimal promoter by 12.1 times (Fig. 3). To define the minimal enhancer region of the 258-bp fragment, we dissected it into four overlapping pieces: the 127-bp 5Ј-portion, 129-bp middle portion, 154-bp 3Ј-portion, and 62-bp most 3Ј-portion. These fragments were subcloned into pCB.7PH to generate pCB.7e127, pCB.7e129, pCB.7e154, and pCB.7e62, respectively. None of these fragments had strong enhancer activity in K562 and HL60 cells (Fig. 3), indicating either that the 258-bp fragment cannot be further subdivided without total loss of activity or that all the cleavage sites are at positions essential for activity in K562 and HL60 cells.
GATA-1 Bound to the WT1 Intronic Enhancer-GATA-1 is a zinc-finger transcription factor believed to play an important role in gene regulation during the development of erythroid cells (28). A computer search for transcription factor-binding  E1, E2, and E3) and introns. Bg, BglII; N, NotI; and B, BamHI. One of the major transcription start sites is indicated by an arrow. The location and sizes of each of the fragments tested are marked. Enhancer activity was tested in hematopoietic (K562), kidney (293), and mesothelial (II-14) cells. CAT activity of each of the fragments is expressed relative to that of pCB.7PH, which contains the WT1 promoter but no enhancers. ND, not determined. B, deletion analysis of the 1.5-kb WT1 intronic enhancer. Exons 2 and 3 are indicated by solid boxes. The 1.5-kb BamHI fragment, 678-bp BamHI/SphI fragment, 833-bp SphI/BamHI fragment, and 258-bp HindIII/BamHI fragment were cloned into the pCAT ® -Promoter in the reverse orientation relative to the CAT gene. The enhancer activity (relative CAT activity) of each construct is expressed relative to that of the pCAT ® -Promoter, which contains the SV40 promoter but no enhancers.  sites revealed three non-consensus GATA sites at nucleotides 108, 207, and 254 in the intronic enhancer of WT1 (Fig. 2). The non-consensus GATA-1-binding sequences were previously identified as GATA-binding sites by oligonucleotide selection methods (29,30). To determine whether GATA-1 can bind the GATA site in the intronic enhancer, we performed EMSAs. K562 nuclear extracts were incubated with the end-labeled probe, GATA-A, which contains the most 5Ј-GATA motif in the intronic enhancer (Fig. 4A). Three major complexes were seen, complexes 1, 2, and 3 (Fig. 4B, lane 2). They were all diminished when 50 and 150 times molar excesses of unlabeled probes were added before the incubation of the radiolabeled probe (Fig. 4B, lanes 3 and 4). This suggested that these complexes were specific for the GATA-A probe; however, they might not be GATA specific. To determine which complex represents binding to the GATA-A motif, additional competition experiments were performed. Excess of the competitor mGATA-A (Fig. 4A), which contains the mutant 108-bp GATA motif, was added before incubation with the probe. We found that none of the complexes could be competed by mGATA-A (Fig. 4B, lanes 5 and 6), indicating that all of the complexes contained proteins which bound preferentially to the GATA-A probe. To determine whether the GATA-A motif binds GATA protein with the same affinity as the GATA consensus (GATA con) sequence (Fig. 4A), 50-and 150-fold molar excesses of double-stranded GATA consensus competitors were added before the incubation with the radioactive probe. We found that both complexes 2 and 3 were eliminated by the competition with GATA consensus oligonucleotide (Fig. 4B, lanes 7 and 8), suggesting that both of these complexes contained GATA-binding proteins but that complex 1 did not. To verify that complex 1 did not contain GATA-binding proteins, mGATA-A was radiolabeled and incubated with K562 nuclear extracts. The GATA-specific complexes 2 and 3 were not formed, only complex 1 was seen (Fig. 4C, lane 4). Overall, this indicates that complex 1 forms on a region of the oligonucletides distinct from the GATA sites, thus contains proteins which are not GATAbinding proteins. The inability of the mGATA to compete off complex 1 when it is bound to the GATA-A oligonucleotide indicates it has a higher affinity to GATA-A than to the mGATA-A, possibly signifying cooperative interaction between the GATA complexes and complex 1.
To determine which type of GATA proteins formed the com-plexes, GATA-1, GATA-2, and GATA-3 antibodies were used for antibody ablation experiments (Fig. 4D). c-Myb antibody was used as a nonspecific antibody control. We found that only formation of complex 2 was prevented with anti-GATA-1 antibody, formation of complex 3 was not. Both complexes 2 and 3 were prevented with anti-GATA-2 antibody but they were not prevented with c-Myb and GATA-3 antibodies, indicating that the ablation was antibody specific. Since complex 1 formation was not prevented with either c-Myb, GATA-1, -2, or -3 antibodies, this further demonstrated that complex 1 did not contain GATA-1, -2, or -3. In contrast, complex 2 contained both GATA-1 and GATA-2 proteins and complex 3 contained GATA-2 protein only. Thus, it is clear that both GATA-1 and GATA-2 can bind the GATA-A site in intronic WT1 enhancer in vitro.
To determine if a second GATA site in the intronic enhancer may also bind GATA-1 or GATA-2 proteins, we also performed EMSAs with the central GATA-B site at nucleotide 207. K562 nuclear extracts were incubated with radiolabeled doublestrand oligonucleotide which contains GATA-B site (Fig. 2) and analyzed as described above. No GATA-specific complexes were detected (data not shown). This suggested that GATA proteins in K562 nuclear extracts cannot bind to the GATA-B site in the intronic WT1 enhancer in vitro.
GATA-1 Transactivation of the WT1 Intronic Enhancer-To determine whether GATA-1 can also transactivate the intronic enhancer, we cotransfected the WT1 reporter construct containing the 258-bp WT1 intronic enhancer (with three GATA sites) and the minimal WT1 promoter (with no GATA-binding sites) with the mouse GATA-1 expression vector in HeLa cells (which do not express endogenous GATA-1 protein). In cotransfection assays using 5.0 g of the WT1 reporter construct, pCB.1e258 and increasing amounts of a mouse GATA-1 cDNA expression construct (23), a dose-dependent increase in enhancer activity was observed (Fig. 5). Four micrograms of the mouse GATA-1 expression construct produced a 15-fold increase in WT1 promoter activity in HeLa cells and also transactivated pCB.1e258 in Saos-2 and 293 cells by 10-fold (data not shown).
GATA-A Is Responsible for the Enhancer Activity-There are three non-consensus GATA motifs in the 258-bp enhancer located at positions 108, 207, and 254. Of them, the GATA-A motif is most similar to the GATA consensus motif. Therefore, FIG. 3. Deletion analysis of the 258-bp intronic enhancer. A series of deletion constructs was made to determine the essential region of the 258-bp intronic enhancer. The full-length WT1 promoter is shown by two open boxes flanking a shaded box (the 104-bp WT1 minimal promoter). GATA-binding sites are shown by filled circles. The 258-, 127-, 129-, 154-, and 62-bp fragments were generated by PCR amplification as described in the text and subcloned into pCB.7PH. The orientation of the fragments relative to the WT1 promoter is shown by arrows. The 258-bp fragment was also subcloned into pCB.1 containing the 104-bp minimal WT1 promoter alone (16), to generate pCB.1e258. The enhancer activity was tested in erythroid (K562) and myeloid (HL60) cells. CAT activity of each construct is expressed relative to that of the empty vector (pCB.7PH or pCB.1). ND, not determined. we mutated this site by altering 2 bp (GA to CT) within the binding site by PCR amplification. The CAT reporter construct containing the WT1 promoter and intronic enhancer with the mGATA-A motif, pCB.7mGATA-Ae258, was transfected into K562, HL60, and HEL cells. The mutagenized enhancer (pCB.7mGATA-Ae258) was 60% less active in K562 and HL60 cells and 50% less active in HEL cells than the wild type enhancer (pCB.7e258) (Fig. 6).
Additive Effects of the Two Enhancers-Because both enhancers function in K562 cells, we examined whether activation by both the intronic and the 3Ј-enhancers has an additive or synergistic effect on the WT1 promoter in K562 cells. The 3Ј-enhancer was linked to the 258-bp intronic enhancer to generate the dual enhancer construct pCB.7e3Јeint, which was then tested in K562 cells (Fig. 7). The construct containing both enhancers was much more active than the constructs that contained either enhancer alone and was slightly more active than the additive value of the two enhancers (Fig. 7). No synergistic or additive effects were observed in HL60 cells (data not shown), which is not surprising as the 3Ј-enhancer does not function in this cell line (Table I).
c-Myb Transactivation of the Intronic Enhancer-Because the intronic enhancer was very active in HL60 cells which lack GATA-1, we asked whether transcription factors other than GATA-1 are responsible for enhancer activity in HL60 cells. By computer analysis, we found a c-Myb-binding site in the intronic enhancer fragment (Fig. 2). High-level expression of c-myb in normal cells occurs primarily in immature hematopoietic progenitor cells of various lineages (31). The c-myb proto-oncogene is also expressed in both acute and chronic leukemias, and its expression overlaps the expression of WT1. Therefore, we tested c-Myb for its ability to transactivate the 258-bp intronic enhancer containing one c-Myb site. We transfected Saos-2 cells with 5.0 g of CAT reporter pCB.1e258 (which contains the WT1 minimal promoter and the intronic enhancer) and 8.0 g of the empty vector, wild type c-Myb expression vector, or a mutant c-Myb (lacking the DNA-binding domain) expression vector. We found that the wild-type c-Myb construct transactivated the reporter construct pCB.1 e258 nearly 5-fold more than the empty vector, while the mutant c-Myb did not significantly transactivate pCB.1 e258 (Fig. 8).
DNase I Hypersensitivity in the 258-bp WT1 Intronic Enhancer Region-To determine whether there are DNase I-hypersensitive sites in the 258-bp intronic enhancer region. Nuclei from either K562 cells or HeLa cells were treated with increasing amounts of DNase I, and then the purified DNA was digested with both EcoRI and XbaI (Fig. 9). The blots were probed with the 32 P-labeled EcoRI-SphI fragment. Fig. 9A shows the representative results. A genomic map of the partial intron 3 region of the WT1 gene from BamHI site to XbaI site is shown at Fig. 9B. When EcoRI-XbaI fragment was analyzed with the labeled probe in K562 cells, it was observed that the parental 1.7-kb fragment was digested with increasing amounts of DNase I. A 1.3-kb fragment was shown by an arrow. When nuclei from HeLa cells were treated in an identical manner with DNase I, the EcoRI-XbaI fragment was resistant to degradation and no hypersensitive sites were detected.

DISCUSSION
Here we described the identification of a second hematopoietic-specific enhancer of the WT1 gene. This enhancer was both position-and orientation-independent and was capable of increasing basal transcription levels from both the SV40 and WT1 promoters in K562 cells. Although both this enhancer and the previously isolated 3Ј-enhancer were hematopoietic-specific, activating transcription in K562 cells, they differed in lineage specificity. The 3Ј-enhancer was very active in HEL cells with erythroid characteristics, whereas the intronic enhancer was very active in HL60 cells with myeloid characteristics. K562 cells are multipotent leukemic cells that have both erythroid and myeloid characteristics and can differentiate into erythroid and megakaryocytic cells in vitro (14). HEL cells are erythroleukemic cells that can also be induced to differentiate along erythroid and megakaryocytic pathways (32). In contrast, HL60 cells are promyelocytic leukemia cells and can be induced to differentiate into monocytes and neutrophils (15). Interestingly, WT1 expression decreases after treatment of K562 and HL60 cells with differentiation inducers (14,15). While both T-and B-cell leukemias express WT1 (11,13), neither the 3Ј nor the intronic enhancer was active in CEM cells, a T-cell leukemia cell line. We have not yet determined whether these enhancers function in cell lines of B-cell origin, but our results suggest that these two enhancers lack T-cellspecific WT1 regulatory enhancer elements.
The fact that GATA-1 is present in almost all erythroidspecific gene promoters suggests that in K562 and HEL cells, GATA-1 plays an important role in the activation of the WT1 promoter, possibly by directly contacting the basal transcriptional machinery and influencing the frequency of initiation. Two GATA-binding sites have also been found in the WT1 promoter, and GATA-1 transactivation increased the WT1 promoter activity 10-fold. 2 We hypothesize that interaction between GATA-binding proteins may help establish contact between the WT1 promoter and the two enhancers.  Fig. 3. The activities of pCB.7PH (which contains the 652-bp WT1 promoter alone), pCB.7e258 (which contains the 652-bp WT1 promoter and wild-type intronic enhancer), and pCB.7mGATA-Ae258 (which contains the WT1 promoter and mGATA-A intronic enhancer) were tested by transfection into K562 (dark hatched bars), HL60 (shaded bars), and HEL cells (light hatched bars). The activities of pCB.7e258 and pCB.7mGATA-Ae258 are relative to that of pCB.7PH.
Because c-Myb could also transactivate the intronic enhancer and the c-Myb and the GATA-A motifs are only 10 bp apart (Fig. 2), it is possible that GATA-1 and c-Myb can form a functional complex and facilitate transcription. It has been reported that both GATA-1 and c-Myb can form heterodimers with other transcription factors. For example, GATA-1 activity can be enhanced by forming dimers or heterodimers with either Sp1 or erythroid Krü pel-like factor (33,34). It has also been found that c-Myb and the core-binding factor act synergistically but bind independently to adjacent sites in the T-cell receptor ␥-enhancer (35).
Although we showed that both GATA-1 and c-Myb could transactivate the intronic enhancer, we cannot rule out the possibility that other transcription factors may also modulate the enhancer activity. In the EMSAs (Fig. 4C), we found that both complexes 2 and 3 could be eliminated by GATA-2 antibody, indicating that GATA-2 is probably involved. Consistent with these data, the activity of the intronic enhancer with the mutant GATA-A motif was decreased by 60% in HL60 cells, which have no endogenous GATA-1 (Table I) but have other GATA-binding proteins, for example, GATA-2. Recently, we found that GATA-2 also could transactivate this enhancer by 8-fold in 293 cells. 3 It is important to establish the physiological role of this enhancer. Previously, we showed that (a) mature leukocytes in normal blood expressed low levels of WT1 and GATA-1, whereas the normal CD34 ϩ bone marrow hematopoietic progenitors expressed a significant amount of WT1 and GATA-1 (36); (b) leukocytes from acute myelogenous leukemia patients are expressed a significant amount of both WT1 and GATA-1 (36). This suggests that the intronic enhancer may be responsible for the high expression of WT1 in normal bone marrow 3 X. Zhang and G. F. Saunders, unpublished results. FIG. 7. Additive effect of the two enhancers in K562 cells. The 258-bp intronic enhancer sequence was cloned into pCB.7e3Ј, which contains the WT1 promoter and the 258-bp 3Ј-enhancer (17), to generate pCB.7e3Јeint. The WT1 promoter and the GATA sites are depicted as in Fig. 3. The activities of the 3Ј-enhancer, the intronic enhancer, and both enhancers together were compared with that of pCB.7PH which contains just the WT1 promoter. One representative CAT assay is shown with the average percentage conversion of acetylated [ 14 C]chloramphenicol from three different experiments listed on the right.
FIG. 8. c-Myb transactivated the intronic enhancer. Five micrograms of reporter construct pCB.1e258, which contains the 104-bp WT1 minimal promoter and the 258-bp intronic enhancer, was transfected into Saos-2 cells with 8.0 g of empty expression vector, wild-type, or mutant c-Myb expression vector. The CAT activity is expressed relative to the activity of the reporter construct in the presence of 8.0 g of empty expression vector.
FIG. 9. Mapping of DNase I-hypersensitive sites in the 258-bp enhancer region. A, nuclei isolated from K562 and HeLa cell lines treated with 0 -10 units of DNase I/mg for 10 min at 30°C as described under "Experimental Procedures." The blot was hybridized with the 5Ј-EcoRI/SphI fragment probe, shown in panel B, which detects a 1.7-kb parental EcoRI/XbaI segment and a 1.3-kb sub-band which is generated by cleavage at hypersensitive sites. B, diagram of the region including hypersensitive sites. The black box represents the 258-bp intronic enhancer. The striped box represents the EcoRI/SphI probe. Horizontal lines below the map represent the positions of the parental 1.7-kb EcoRI/XbaI segment and the 1.3-kb sub-band generated by DNase I hypersensitivity. A vetical arrow on the map shows the position of hypersensitive site, and the numbers above them indicate the distance (in kilobases) from the WT1 gene transcription start site. and acute myelogenous leukemia patients.
It is essential to determine whether GATA-1, c-Myb, or other transcription factors bind to the 258-bp minimal enhancer region in its natural context. We examined the DNase I-hypersensitive site in the 1.7-kb EcoRI/XbaI region which contains the 258-bp intronic enhancer. One hypersensitive site was mapped to the 258-bp enhancer region in K562 cells, but not in HeLa cells. This result not only confirmed our previous transfection data but also indicated that the transcription factors which are responsible for the activity of the hematopoieticspecific enhancer are located in the 258-bp intronic region in vivo.