PAX 8 Regulates Human WT1 Transcription through a Novel DNA Binding Site*

The Wilms’ tumor gene (WT1) is an essential gene for kidney and gonadal development, although how WT1 expression is induced in these tissues is not known. One kidney transcription factor likely to play a role in this regulation is PAX 8. The co-expression of WT1 and PAX 8 during kidney development and in Wilms’ tumors with an epithelium predominant histology suggested a possible interaction, and indeed, we identified potential core PAX-binding sites in the WT1 promoter. Endogenous PAX 8 plays an important role in the activation of the WT1promoter, since promoter activity is much stronger in cells with PAX 8 than without. Using binding assays, we searched for evidence of PAX 8-DNA interactions throughout the 652-base pair human WT1promoter and found only one functional PAX 8 site with DNA binding activity, located 250 base pairs 5′ of the minimal promoter. The responsiveness of the PAX 8 site was confirmed by assessing its ability to function as an enhancer significantly activating the minimal promoter in a position- and orientation-independent manner. Using transfection assays, we demonstrated that either endogenous or exogenously added PAX 8 activated the WT1 promoter and that this promoter up-regulation depended upon the presence of an intact PAX 8-binding site. In contrast, the previously reported core PAX 8-binding sites identified by computer analysis of the WT1 promoter failed to specifically bind in vitro translated PAX 8 protein or activate the minimal promoter. Thus, we identified a novel functional binding site for the transcription factor PAX 8, suggesting that part of its role in kidney development may be as a modulator of WT1 expression in the kidney.

Differential WT1 expression during development suggests that WT1 expression is tightly regulated by tissue-specific transcription factors. PAX 8 encodes a developmentally important paired box transcription factor that is expressed in the developing kidney, among other tissues (26). The expression of PAX 8 in the developing kidney precedes WT1 expression. Its expression initiates in the induced condensing mesenchyme, peaks in the S-shaped bodies, and finally declines in the epithelial cells of the glomerulus (27). The WT1 expression pattern is similar to the PAX 8 pattern, with its peak occurring after PAX 8 expression begins to decline and WT1 expression persists in the mature podocytes of the glomerulii (27,28). Like WT1 expression, PAX 8 expression has also been found in the Wilms' tumors with an epithelial predominant histology 2 (27,29,30). The presence of PAX 8 mRNA in these tumors with high WT1 expression is consistent with its postulated role as an activator of WT1 expression. PAX 8 paralogues include PAX 2 and PAX 5. Unlike PAX 2 and PAX 8, PAX 5 is expressed in developing hematopoietic cells (pre-B cells). While in vivo downstream target genes have not been identified for these PAX genes, several genes whose expression is regulated by PAX 5 and PAX 8 in vitro have been identified, and consensus binding sites have been defined by oligonucleotide selection methods. PAX 5 can regulate expression of the B cell-specific surface protein CD19 (31), and PAX 8 can regulate expression of thyroperoxidase and thyroglobin (32). Recently, engrailed-2 was shown to be a target gene for Pax 2, Pax 5, and Pax 8 (which are co-expressed with Engrailed-2) in mouse embryos during mid-hindbrain development (33). PAX 2, PAX 5, and PAX 8 each have only a partial homeodomain, so DNA binding is mediated by the paired domain, which is composed of three ␣-helices. The NH 2 -terminal subdomain of the paired domain is highly conserved and binds a core recognition sequence of the PAX-binding site. However, the COOH-terminal subdomain also influences site-specific binding (34).
While the WT1 promoter coupled to the SV40 enhancer strongly activates transcription in all cell lines tested (35), its basal activity varies among cell lines, suggesting differential transactivation by tissue-specific factors. In constructs lacking the SV40 enhancer, the transcriptional activity of the WT1 promoter is much lower, as is typical of a TATA-less, CCAATless, GC-rich promoter. The ubiquitous transcription factor Sp1 has been shown by footprint analysis to bind at many positions throughout the WT1 promoter and to transactivate the promoter region in cotransfection assays (36,37). WT1 protein also binds at many positions throughout the WT1 promoter, and the least abundant isoforms lacking the KTS insertion strongly repress the WT1 promoter in cotransfection assays (38,39). The autoregulatory sites in the WT1 promoter may be essential for the down-regulation of WT1 expression observed during the differentiation of leukemic cells (40,41). Using deletion analysis, we previously identified the minimal promoter region necessary and sufficient for promoter activity in K562 and HeLa cells (35). This 104-bp minimal promoter is GC-rich (79% G ϩ C); contains seven overlapping potential transcription factor binding sites within a core of 30 bp, including Sp1, AP4, WT1, and CACCC binding sites; and has half the transcriptional activity of the full-length WT1 promoter.
To better understand the mechanism for differential promoter activity in various cell lines, derived from different tissues, we examined the 5Ј-flanking region of the full-length human WT1 promoter. Sequence analysis identified potential binding sites for the zinc finger GATA factors (42) and two paired box transcription factors, PAX 8 (32) and PAX 2 (43). The latter two are co-expressed with WT1 in the kidney and may contribute to kidney expression of WT1 in vivo. Recently, the murine WT1 promoter was shown to be transactivated by both Pax 2 and Pax 8 (44,45). We have identified a novel consensus site 250 bp 5Ј of the minimal promoter that strongly bound PAX 8 and mediated potent PAX 8 transactivation of the human WT1 promoter. This is in contrast to the potential PAX 8-binding sites previously identified in the human WT1 promoter (35), which were unable to bind PAX 8 or mediate transactivation of WT1.
To determine the role of the PAX core binding sites, we subcloned the 5Ј-flanking region containing two potential PAX 8 core sites (CTGCCC) and the distal potential PAX 2 core site (GTTCCC) but lacking the most 5Ј novel PAX site, WT1 PAX CON. The 652-bp WT1 promoter was used as a template for PCR amplification of the 130-bp fragment with primers altered to insert BamHI sites (altered bases are in boldface type) as described elsewhere (35). The forward PCR primer (5Ј-CTGCGGGATC-CTGAAGTTCC-3Ј) (bp 20 -39) and the reverse PCR primer (5Ј-TTCAG-GTAAGCAGTGGATCCG-3Ј) (bp 128 -149) were derived from the 5Јflanking region of the promoter sequence (beginning 20 bp 3Ј of the HindIII site). The PCR product was digested with BamHI, and the resulting 104-bp BamHI fragment was purified on a Qiagen column (Qiagen, Chatsworth, CA) and cloned in the reverse orientation into the BamHI site of the WT1 minimal promoter construct pcb.1 to create the 5Ј-flanking construct, pcb.1e.1. The orientation and sequence of the 5Ј-flanking region were confirmed by sequence analysis.
To determine if the novel WT1 PAX CON site was sufficient for PAX 8 transactivation of WT1, we cloned the 30-bp double-stranded oligonucleotide used in EMSA (see below) 3Ј of the chloramphenicol acetyltransferase (CAT) gene in pcb.l, the minimal WT1 promoter construct. Initially, BamHI linkers were ligated to the 30-mer, and then both the double-stranded oligonucleotide insert and promoter construct were digested with BamHI. The free linkers were removed from the insert by column chromatography, and the insert was ligated to phosphatasetreated pcb.1, the minimal WT1 promoter construct, to create the PAX 8-enhancer construct, pcb.le.05.
While the WT1 promoter construct pcb.7PH contains several potential PAX 8-binding sites, the only site able to bind PAX 8 in EMSA is at the most 5Ј-end of the 652-bp promoter. The mutant transactivation reporter target construct was created by PCR amplification of the 652-bp WT1 promoter in pcb.7PH by using a forward PCR primer (5Ј-TATGACCAAGCTTACGCCAAGATTGTCTGAGTTCTTTCTG-3Ј) containing the WT1 promoter 5Ј-flanking sequence (underlined) with a mutant PAX 8 site (as discussed below) and pCAT®-Basic 5Ј-flanking sequence containing a 2-bp mismatch to create a HindIII site (altered bases are in boldface type). The reverse PCR primer S3 (5Ј-CTCCT-GAAAATCTCGCCAAGC-3Ј) was derived from pCAT®-Basic and is adjacent to the PstI cloning site. The PCR-amplified fragment was digested with HindIII and PstI, and the resulting 661-bp promoter fragment was cloned into the HindIII and PstI-digested pCAT®-Basic vector to create mpcb.7PH (replacing the wild-type promoter in pcb.7PH). The mutagenized binding site and its orientation in the clone were verified by DNA sequence analysis.
Transfections and CAT Assays-K562, 293, and HeLa cells were transfected with pcb.1, pcb.1e.1, pcb.le.05, pcb.7PH, and mutant pcb.7PH as described previously (35,46). TM4 cells were electroporated at 250 V in 200 l of serum-free medium containing 10 g of plasmid DNA. As an internal control for transfection efficiency, the cells were cotransfected with 5 g of the ␤-galactosidase control DNA, pSV40␤-Gal (Promega Corp., Madison, WI). The cells were harvested, cytoplasmic extracts were prepared, and ␤-galactosidase activity was determined as described previously (35). Extracts containing equivalent amounts of ␤-galactosidase activity were assayed for CAT activity (47). The protein content of the cell extracts was also determined by the Bradford assay (Bio-Rad), and the CAT assays were performed with extracts containing 25-50 g of protein (for K562 cells) or 50 -100 g of protein (for the other cell lines). After thin layer chromatography, the acetylated [ 14 C]chloramphenicol was quantitated by measuring the radioactivity with a Betascope 603 blot analyzer (Betagen Corp., Waltham, MA). The relative activities shown are the averages from at least three different experiments.
For transactivation assays PAX 8 was co-transfected into HeLa cells as described above except that increasing amounts of the human PAX 8 cDNA expression construct (48), pSVK3PAX8, were added to each transfection of 5 g of reporter pcb.7PH and 2 g of pSV40␤-Gal. The PAX 8 expression construct contains a 1.4-kilobase pair PAX8 cDNA including the entire open reading frame of the predominant isoform PAX 8a (32). This cDNA was initially obtained by screening a human fetal kidney cDNA library (CLONTECH, Palo Alto, CA) and was cloned into the expression vector pSVK3 (Pharmacia Biotech Inc.) (48). The mutant clone containing a mutagenized PAX 8-binding site mut.pcb.7pH, the empty pCAT®-Basic vector (Promega), the minimal promoter pcb.1e.1, and the PAX 8 enhancer construct pcb.1e.05 were also cotransfected with 5 or 10 g of the PAX 8 expression construct and 2 g of pSV40␤-Gal.
RT-PCR-Total cellular RNA was extracted from 2-day-old mouse kidneys and all cell lines described above by the method of Chomczynski and Sacchi (49) with STAT-60 (Tel Test, Friendswood, TX) according to the manufacturer's recommendations. First strand cDNA synthesis and PCR amplification were performed as described previously (50). PCR amplification of cDNA encoding the Pro-Ser-Thr transactivation domain and carboxyl terminus of the PAX 8 gene in human samples was performed by using the primer pairs: forward PAX 8 (5Ј-TCCACCCCT-TCCTCTTTATCT-3Ј) and reverse PAX 8 (5Ј-AGTCCTCCTGTTGCT-CAGTCG-3Ј). As a control, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) cDNA was amplified by using forward GAPDH (5Ј-TTGC-CATCAATGACCCCTTCA-3Ј) and reverse GAPDH (5Ј-CCAGTGAGCT-TCCCGTTCAGC-3Ј). The cDNA encoding the octapeptide, homeodomain, and 5Ј-end of the Pro-Ser-Thr transactivation domain of the PAX 8 gene in non-human samples was amplified by PCR with the murine Pax 8 primer pairs forward murine Pax 8 (5Ј-AAGTCTCTGAGCCCAG-GACA-3Ј) and reverse murine Pax 8 (5Ј-GGATCTGCAGCAAGTCG-GCT-3Ј). PCR amplification of cDNA encoding the WT1 zinc finger region and ␤-actin gene in human and non-human samples was carried out using the human and murine primer pairs as described (23,46). The conditions for PAX 8 and GAPDH co-amplification were 3 min of denaturation at 95°C; 35 cycles of 94°C denaturation for 1 min, 58°C (human) or 64°C (murine) annealing for 2 min, and 72°C extension for 2 min; and a final 5-min extension at 72°C. To compensate for the overabundance of GAPDH and ␤-actin mRNA, we used 1 ⁄10 the concentration of GAPDH and ␤-actin primers as PAX 8 primers (0.2 M). The PCR products were analyzed on an ethidium bromide-stained 1.5% agarose gel. The size marker used was a 100-bp DNA ladder (Life Technologies, Inc.). The duplex PCR amplification resulted in PCR products of the expected sizes: 600 bp for GAPDH, 441 bp for human PAX 8, 558 bp for murine Pax 8, 330 bp for murine WT1, and 540 bp for ␤-actin.
Electrophoretic Mobility Shift Assays (EMSAs)-EMSAs were performed by using Caki-1 renal cell nuclear extracts and in vitro translated PAX 8 protein. The in vitro PAX 8 expression construct pBSPAX8 was derived from the PAX 8 expression construct pSVK3PAX8. The 1.4-kilobase pair PAX 8 coding sequence was released by EcoRI digestion, gel-purified, and subcloned into the EcoRI site in pBlueScript II (Stratagene). The adjacent T7 polymerase binding site was used for in vitro transcription according to the manufacturer's recommendations (Promega). PAX 8 protein was translated in vitro by coupled transcription and translation of 1 g of the PAX 8 expression construct pBSPAX8 in a wheat germ extract system according to the manufacturer's recommendations (Promega). The molecular size of the PAX 8 protein product was confirmed by SDS-polyacrylamide gel electrophoresis of [ 35 S]methionine-labeled cell extracts and compared with prestained molecular weight markers (Amersham Corp.). Nuclear miniextracts were prepared from exponentially growing K562 and Caki-1 cells by the method of Dignam et al. (51) as modified by Lee et al. (52).
The EMSAs were performed essentially as described by Singh et al. (53). The 30-bp double-stranded oligonucleotides were labeled with [␥-32 P]ATP by using T4 polynucleotide kinase. The WT1 promoter PAX consensus probe (WT1 PAX CON), 5Ј-CGCTTCTTTGAAGCTTGACT-GAGTTCTTTC-3Ј, was used to identify the PAX 8-binding site (predicted consensus PAX 8-binding sites are underlined). Radiolabeled control PAX 8 oligonucleotide CT (5Ј-TGATGCCCACTCAAGCTTAGACAGG-3Ј), which was derived from the thyroperoxidase gene (32), was used as a positive control for comparison with the WT1 promoter probe WT1 PAX CON. One to two hundred fmol of labeled oligonucleotide (1-4 ϫ 10 5 cpm) was incubated with either 6 g of the nuclear protein extracts or 1 l of a 50-l reaction of in vitro translated PAX 8 protein lysate and 2 g of poly(dI-dC) in 15 mM Tris-HCl (pH 7.5), 6.5% glycerol, 90 mM KCl, 0.7 mM EDTA, 0.2 mM dithiothreitol, and 0.1% bovine serum albumin. After a 30-min incubation at room temperature, the reaction mixture was analyzed by electrophoresis at 300 V on a 5% polyacrylamide/bisacrylamide gel (39:1) in 0.5 ϫ TBE (45 mM Tris borate, and 1 mM EDTA, pH 8.0), dried, and visualized by autoradiography with Hyperfilm-AP film (Amersham). For competition assays, the unlabeled double-stranded PAX 8 oligonucleotide CT and consensus GATA oligonucleotide (Santa Cruz Biotechnology, Santa Cruz, CA) were used as positive and negative controls, respectively. To verify the importance of two invariant bases within the WT1 PAX CON site, the mutant WT1 promoter oligonucleotide MWT containing the mutated PAX CON site 5Ј-CGCTTCTTTGAAGATTGTCTGAGTTCTTTC-3Ј was used (altered bases in boldface type). Competitor oligonucleotides were preincubated for 15 min at room temperature with the extracts described above, and then the labeled WT1 PAX CON oligonucleotide probe was added to the binding reaction, incubated for an additional 15 min, and analyzed as described above. For PAX antibody supershift assays, 1 l of crossreacting anti-PAX antiserum (prepared by inoculating a rabbit with PAX 6 protein (54)) was added either before or after the addition of labeled probe and incubated for an additional 15 min at room temperature. To compare binding of WT1 PAX CON to binding with oligonucleotides containing either the even-skipped PAX 2 site GTTCC, or the thyroglobin PAX 8 site CTGCCC we obtained oligonucleotides from the human WT1 promoter containing these sites and referred to as Distal and Proximal, respectively. Distal is 5Ј-TCGAGTTCCCGCCCTCTTG-GAGCC-3Ј and contains overlapping PAX 2 and Sp1 sites along with a TCGA tag. Proximal is 5Ј-TCGATACCTGCCCCCCCTCCA-3Ј and contains overlapping PAX 8 and AP2/Sp1 sites along with a TCGA tag. The human Proximal probe differs from the murine probe by a single base change (C to T) at the fifth position (in boldface type) (44).

Identification of Potential PAX 8-binding Sites-
The initial sequence analysis of the WT1 promoter identified three potential PAX 8-binding sites containing the paired box core binding site CTGCCC. Two of these sites were located in the 5Ј-flanking region, 59 and 125 bp 3Ј of the HindIII site in the full-length WT1 promoter (Fig. 1). The third site was located in the 5Јuntranslated region of the WT1 gene, 445 bp 3Ј of the HindIII site (86 bp 3Ј of the minimal promoter). However, functional assays revealed no PAX 8 binding at these sites and no transcriptional activation of the minimal promoter by the two 5Јmost sites (see "PAX 8 Transactivation of the WT1 Promoter").
Because the initially identified PAX 8 core binding sites lacked functional activity, we examined additional PAX-binding sites in other genes. An alignment of consensus binding sites for PAX 8, derived from the thyroperoxidase and thyroglobin gene promoters (32); PAX 5, derived from the CD19 gene promoter (55); and PAX 2, derived from oligonucleotide selection (43) revealed a consensus site independent of the CTGCCC core site. Our comparison of all of these paralogous PAX-binding sites indicated that a novel consensus site, CAST-SANGCNK (where S represents G or C and K represents T or G), may mediate PAX 8 activation as well. Two of 11 bases in this consensus site are conserved within all of the binding sites analyzed and are referred to as invariant. This novel consensus site was then identified in the 5Ј-flanking region of the WT1 promoter ( Fig. 1) using the GCG sequence analysis program. This site is found in only one location within both the WT1 promoter and the 2-kilobase pair region 5Ј of the WT1 promoter. Functional assays demonstrated that this novel consensus site mediated both PAX binding and transcriptional activation (see below).
Comparison of WT1 Promoter Activity and Endogenous PAX 8 Expression-The embryonic expression pattern of PAX 8 and WT1 combined with the potential PAX 8-binding sites within the WT1 promoter suggested that PAX 8 may modulate WT1 expression. To examine the correlation between PAX 8 and WT1 expression, we first obtained RT-PCR evidence of coexpression in various cell lines. Eight of nine cell lines were either positive or negative for both PAX 8 and WT1 expression (Fig.  2), further strengthening the correlation of expression. In one of six kidney cell lines, we observed discordant expression, suggesting that PAX 8 expression is not necessary for WT1 expression in 293, adenovirus 5 transformed human embryonic kidney cells. In 293 cells, which express high levels of E1A and E1B, transcription factors other than PAX 8 are probably responsible for WT1 expression.
To further assess the role of PAX 8 in the regulation of WT1 expression, we compared the activity of the WT1 promoter in vitro with PAX 8 expression. Fig. 3 shows the differential activity of the basal WT1 promoter in different cell types that expressed WT1, PAX 8, or both. Ten micrograms of the basal WT1 promoter construct lacking an enhancer (pcb.7PH) was transfected with 2 g of pSV40␤-Gal into several different cell lines. The basal promoter was only 3-4-fold more active than the empty vector pCAT®-Basic in HeLa and 293 cells, which expressed very little or no PAX 8, respectively. In contrast, the basal WT1 promoter construct (pcb.7PH) was 8 -10-fold more active than the empty vector CAT®-Basic in TM4 and K562 cells, which both expressed PAX 8 (Fig. 2).
PAX8 Transactivation of the WT1 Promoter-To identify the transcription factors that modulate WT1 expression in the kidney, we examined the promoter regulatory regions by functional and binding assays using CAT cotransfections of HeLa cells and EMSAs. The basal WT1 promoter functioned weakly in HeLa cells (Fig. 3), but the addition of the PAX 8 expression construct greatly increased WT1 promoter activity (Fig. 4). The PAX 8 responsiveness of the WT1 promoter was assessed by cotransfection of 5 g of pcb.7PH (the WT1 reporter construct) and 2 g of pSV40␤-Gal and increasing amounts (0 -20 g) of human PAX 8 cDNA expression construct pSVK3PAX8. We observed dose-dependent activation of the WT1 promoter by PAX 8 cotransfection: 5 g of the PAX 8 expression construct increased WT1 promoter activity 17-fold relative to activation by the pSVK3 empty expression vector control; and 10 g of PAX 8 increased promoter activity by 30-fold. HeLa cells expressed a very low level of PAX 8 mRNA, as detected by RT-PCR (Fig. 2), and this correlated with their weak basal WT1 promoter activity (Fig. 3). However, the expression of high levels of exogenous PAX 8 greatly increased WT1 promoter activity (Fig. 4). Overall, these data showed that WT1 promoter activity was correlated with PAX 8 expression and that the basal promoter could be strongly transactivated by the addition of a transcription factor (PAX 8) that is normally expressed with WT1 during the development of the kidney (28).
Localization of PAX 8-responsive Sites in the WT1 Promoter-While transactivation of the full-length WT1 promoter demonstrated that PAX 8 up-regulated WT1 expression, it did not determine which site(s) mediated transactivation. To address this question, we compared the activity of two different minimal WT1 promoter constructs containing different potential PAX-binding sites cloned 3Ј of the CAT gene (Fig. 5). Initially, we determined whether the WT1 PAX CON site could enhance transcription of the WT1 minimal promoter (Fig. 5A). In the construct pcb.1e.05, one copy of the WT1 PAX CON oligonucleotide containing the novel PAX 8-binding site was inserted 3Ј of the CAT gene. In K562 cells that express endogenous PAX 8, the CAT activity of the PAX 8 enhancer construct, pcb.1e.05, containing one copy of the novel WT1 PAX CON site was 6-fold greater than that of the minimal promoter pcb.1 alone. In contrast, in 293 and HeLa cells lacking endog- This histogram shows the differential activity of the basal WT1 promoter in cell types that differentially express PAX 8. Ten micrograms of the basal WT1 promoter construct pcb.7PH were transfected with 2 g of pSV40␤-Gal either into cells that express PAX 8 protein (K562 and TM4) or into cells that express little or no PAX 8 protein (293 and HeLa). CAT activity is depicted relative to the activity of the empty vector pCATBasic; the error bars depict S.E. enous PAX 8, the enhancer element failed to activate the minimal promoter. Thus, the novel WT1 PAX CON site can function as an enhancer, increasing the activity of the minimal promoter (which lacks PAX-binding sites) in a position-and orientation-independent manner.
To determine whether additional PAX 8 sites were functional, we examined a second construct, pcb.1e.1, containing the 100-bp 5Ј-flanking region (bp 20 -140) of the WT1 fulllength promoter cloned 3Ј of the CAT gene (Fig. 5B). The 3Ј-insert in this construct lacks the WT1 PAX CON binding site but includes the two PAX core sites identified by sequence analysis and the Distal and Proximal sites thought to mediate PAX 8 transactivation of the murine wt1 promoter (Fig. 2). In contrast to the significant increase in promoter activity associated with a single copy of the WT1 PAX CON site in the PAX 8 enhancer construct, the construct containing two other potential PAX core binding sites, pcb.1e.1, had no increased activity over the minimal promoter construct, pcb.1 (Fig. 5B). This suggests that in K562 cells with endogenous PAX8, the 100-bp region cannot function independently as an enhancer, i.e. these core sites are unable to act in a position-and orientationindependent manner.
Since the activity of the single WT1 PAX CON site is significantly greater than that of the two core sites identified by sequence analysis, we asked whether this novel consensus site could also mediate transactivation by exogenous PAX 8. The responsiveness of the WT1 PAX CON site to endogenous PAX 8 was confirmed by co-transfection of the PAX 8 enhancer construct, pcb.1e.05, with a PAX 8 expression construct in HeLa cells lacking significant PAX 8 expression (Fig. 5C). Five micrograms of each of the reporter constructs, pcb.1, pcb.1e.05, and pcb.1e.1, were cotransfected with 10 g of the pSVK3PAX8 expression construct and 2 g of pSV40␤-Gal in HeLa cells. The CAT activity of the PAX 8 enhancer construct, pcb.1e.05, was 4-fold greater after cotransfection with pSVK3PAX8 than with pSVK3 vector alone. In contrast, the cotransfected PAX 8 expression construct was unable to transactivate pcb.1 (the minimal promoter) or pcb.1e.1 (minimal promoter/5Ј-flanking region), the construct that contained both the Distal and Proximal sites; i.e. the CAT activity of pcb.1e.1 cotransfected with pSVK3PAX8 was only 1.3-fold greater than with pSVK3 vector alone. Overall, a com-parison between the transcriptional activity of the WT1 PAX CON site located at the 5Ј-end of the human WT1 promoter and the potential PAX binding core sites located within the 100-bp 5Ј-flanking region reveals that the Distal and Proximal sites are nonfunctional but that the novel WT1 PAX CON site mediates PAX 8 activation of the WT1 promoter.
Having identified a PAX 8-responsive site that enhances WT1 promoter activity only in the presence of PAX 8 (either endogenously present or exogenously added PAX 8), we then confirmed that strong WT1 promoter activity in K562 cells depends upon an intact PAX 8-binding site. To assess the importance of two invariant bases (in boldface type) within the WT1 PAX CON site, CASTSANGCNK, we mutagenized the T to A and the G to T and investigated whether this alteration affected CAT activity of the WT1 promoter. The mutagenized binding site resulted in a substantial loss (57%) of WT1 promoter activity in K562 cells containing endogenous PAX 8 (Fig.  6A) and no significant loss of activity in HeLa and 293 cells lacking endogenous PAX 8. Taken together, the data demonstrate that 1) endogenous PAX 8 plays an important role in the activation of the WT1 promoter, since the promoter activity in K562 cells (Fig. 3) is much stronger than that in 293 and HeLa cells and 2) WT1 promoter activity depends upon the presence of an intact PAX 8-binding site. The presence of some residual activity in the mutated promoter, despite the complete absence of PAX 8 binding to the mutated site in EMSA (see below), suggests that additional factors present in K562 cells bind and activate the WT1 promoter. In fact, several functional Sp1-(36, 37) and WT1-(38, 39) binding sites and two GATA-binding sites 3 have been identified and shown to modulate WT1 promoter activity. Thus, while PAX 8 is a significant activator of WT1 expression in vivo, it is not the only factor that can activate the WT1 promoter.
Having identified the PAX8 site sufficient for WT1 promoter activity and two invariant bases essential for strong WT1 promoter activity in K562 cells containing endogenous PAX 8, we then confirmed the necessity of an intact PAX 8 site in HeLa cells co-transfected with the PAX 8 expression construct (Fig. 6B). In HeLa cell co-transfection assays, the PAX 8 expression construct strongly activated the WT1 promoter containing an intact PAX 8 site, but the 2-bp mutation resulted in a substantial loss (50%) of PAX 8 transactivation. Overall, the activities of both the PAX 8 enhancer construct (pcb.1e.05) and the full-length WT1 promoter in K562 cells containing endogenous PAX 8 closely resemble the activities seen in HeLa cells cotransfected with exogenous PAX 8. This demonstrates that the PAX 8 site is functional in vivo and that either endogenous or exogenous PAX 8 regulates WT1 expression. The presence of some residual activity in the mutated promoter despite the complete absence of PAX 8 binding to the mutated site in EMSA suggests that either additional factors present in HeLa cells stabilize PAX 8 binding to the mutated site, weakly activating the WT1 promoter, or alternatively, some weak PAX 8 binding to other cryptic sites may occur in the absence of the preferred binding site. In fact, two alternative PAX-binding sites have been identified in the mouse promoter, which fail to bind in vitro translated PAX 8 protein in EMSA but do contribute to the murine wt1 promoter activity (44). Taken together, these data show that while PAX 8 is a significant activator of WT1 expression in vivo, it is not the only factor that can activate the WT1 promoter, and that while PAX 8 binding at its cognate site is sufficient to strongly activate the WT1 promoter, additional cryptic sites may also be able to partially activate the promoter. Promoter-To confirm that the activation of WT1 by PAX 8 correlates with DNA binding, we examined potential binding sites within the WT1 promoter by EMSA (Fig. 7). The WT1 PAX CON probe used in the EMSA contains the 11-bp consensus flanked by the corresponding genomic WT1 sequence. Since the consensus sequence is in the first 11 bp of the WT1 promoter cloned into the construct pcb.7PH, the oligonucleotide used for EMSA contained 20 bp of the 5Ј-end of the WT1 promoter cloned into pcb.7PH. To position the binding site in the middle of the 30-bp oligonucleotide used for EMSA, we used the adjacent 10-bp sequence from the distal promoter sequence (35), thus allowing analysis of the 30-bp site as it exists in genomic DNA.
Nuclear extracts of Caki-1 cells containing abundant levels of the transcription factor PAX 8 were bound to radiolabeled WT1 PAX CON (WT) (Fig. 7A). The PAX 8-specific complex is indicated with an arrow. PAX 8 expression in these cells was verified by RT-PCR analysis of Caki-1 cell RNA (Fig. 2). We also prepared in vitro translated PAX 8 protein by using wheat germ extracts programmed with a PAX 8 cDNA expression vector clone (pBSPAX8) to determine whether the DNA-protein complex observed with Caki-1 extracts (Fig. 7A, lane 2) migrated similarly to the complex formed with PAX 8 protein alone (Fig. 7A, lane 3). In vitro translated PAX 8 proteins are able to form an identical appearing complex with the WT1 PAX CON oligonucleotide, suggesting that no additional proteins in Caki-1 extracts are required for PAX 8 complex formation. Negative control lysates from in vitro translated wheat germ extracts generated in the absence of the pBSPAX 8 expression vector were also tested to verify the specificity of PAX 8 binding. As expected, these unprogrammed wheat germ lysates failed to bind the radiolabeled WT1 PAX CON oligonucleotide (Fig. 7A, lane 4). The binding specificity of the WT1 PAX CON oligonucleotide was tested by competition with a control oligonucleotide, CT, containing a known PAX 8-binding site derived from the thyroperoxidase promoter, a target gene for PAX 8. The Caki-1 complex was specifically eliminated by competition with the 20-and 100-fold molar excesses of unlabeled CT (Fig.  7A, lanes 5 and 7). In contrast, the Caki-1 complex was not eliminated by competition with as much as a 100-fold molar excess of unlabeled GATA oligonucleotide (Fig. 7A, lanes 6 and  8). Because the PAX 8-binding site in the WT1 promoter was a FIG. 5. A novel PAX 8 site activates the minimal WT1 promoter in cells with endogenous or exogenously added PAX 8. A, this histogram shows the differential activity of the PAX 8 enhancer construct in cell types that differentially express PAX 8. Five micrograms of the promoter constructs shown below were transfected with 2 g of pSV40␤-Gal either into cells that express PAX 8 protein (K562) or into cells that express little or no PAX 8 (293 and HeLa). The basal promoter construct, pcb.1, contains the minimal WT1 promoter (black box) but lacks any potential PAX 8-binding sites, and the PAX 8 enhancer construct, pcb.1e.05, includes the minimal promoter and the 30-bp WT1 PAX CON oligonucleotide (shaded box) cloned 3Ј of the CAT gene. The CAT activity of the PAX 8 enhancer construct is depicted relative to the activity of the minimal promoter construct, pcb.1; the error bars depict S.E. B, this histogram shows the differential activation of the minimal WT1 promoter by either a single copy of the novel WT1 PAX CON site or by the two potential PAX core binding sites in the 100-bp 5Ј-flanking region. Five micrograms of either the minimal WT1 promoter (pcb.1) or the PAX 8 enhancer construct (pcb.1e.05) or the flanking region construct (pcb.1e.1) were transfected with 2 g of pSV40␤-Gal into K562 cells that express endogenous PAX 8 protein. The flanking region construct, pcb.1e.1, includes the minimal promoter (black box) and the 100-bp 5Ј-flanking region (hatched box) cloned 3Ј of the CAT gene. CAT activity is depicted relative to the activity of the minimal promoter construct, pcb.1; the error bars depict S.E. C, this histogram shows the differential response of the minimal WT1 promoter (pcb.1) or the PAX 8 enhancer construct (pcb.1e.05) or the flanking region construct (pcb.1e.1) to PAX 8 transactivation in HeLa cells. It demonstrates the ability of a single WT1 PAX CON element to serve as a target for PAX 8 transactivation of the WT1 promoter. Five micrograms of the promoter constructs shown above were cotransfected with 2 g of pSV40␤-Gal and 10 g of either pSVK3PAX8 or the empty expression construct pSVK3. CAT activity is depicted relative to the activity of the promoter constructs cotransfected with the pSVK3 vector control. The values shown are the average relative activities from three different experiments; the error bars depict S.E. novel consensus site, we compared the Caki-1 complex formed by binding to radiolabeled CT that formed with radiolabeled WT1 PAX CON (Fig. 7A, lane 10). The Caki-1 complex was specifically eliminated by competition with 20-fold and hundred-fold molar excesses of unlabeled WT1 PAX CON oligonucleotide (Fig. 7A, lanes 11 and 12). We also compared the in vitro translated PAX 8 protein complex formed by binding to radiolabeled CT with radiolabeled WT1 PAX CON (Fig. 6A,  lane 13). To confirm that the Caki-1 complex contained PAX 8 protein, we used a cross-reactive antibody prepared by inoculating rabbits with PAX 6 protein (55) to supershift the Caki-1 complex (Fig. 7A, lanes 16 and 17). The PAX antibody dramatically reduced Caki-1 complex formation (Fig. 7A, lane 17). The diminished complex is indicated by a thick arrow on the left, and a portion of the PAX-binding DNA complexes were supershifted (see complex indicated by thin arrow on the right). Although this PAX 6 antibody could also cross-react with PAX 2, Caki-1 cells do not express detectable amounts of PAX 2 (32).  6. Mutation of the novel PAX 8 site diminishes WT1 promoter activity in cells with endogenous or exogenously added PAX 8. A, this histogram shows the differential effect of mutagenesis of the WT1 PAX CON binding site on WT1 promoter activity in cell types that differentially express PAX 8. Five micrograms of the promoter constructs shown below were transfected with 2 g of pSV40␤-Gal either into cells that express PAX 8 protein (K562) or into cells that express little or no PAX 8 (293 and HeLa). The basal promoter construct, pcb.7PH contains the full-length WT1 promoter, and the mutated promoter construct, mut pcb.7PH, has a 2-bp substitution in the WT1 PAX CON binding site (asterisk). The CAT activity of the mutated promoter construct is depicted relative to the activity of the full-length WT1 promoter construct, pcb.7PH; the error bars depict S.E. B, this histogram assesses the effect of mutagenesis of the WT1 PAX CON binding site on PAX 8 transactivation in HeLa cells. It demonstrates the substantial decrease in PAX 8 transactivation of the WT1 promoter containing a 2-bp substitution in the WT1 PAX CON binding site. Five micrograms of the promoter constructs shown below were cotransfected with 2 g of pSV40␤-Gal and 10 g of either pSVK3PAX8 or the empty expression construct pSVK3. CAT activity is depicted relative to the activity of the promoter constructs cotransfected with the pSVK3 vector control. The values shown are the average relative activities from three different experiments; the error bars depict S.E.
To determine whether the two invariant bases (see below, in boldface type) contained within the WT1 PAX CON consensus site CASTSANGCNK are essential for binding and activation by PAX 8, the binding site was mutagenized (T to A and G to T). A loss of binding activity was observed by competition EMSA (Fig.  7B). The binding specificity of the WT1 PAX CON oligonucleotide for the in vitro translated PAX 8 protein was tested by competition with the control oligonucleotide CT, which contains a known PAX 8-binding site. As previously shown for Caki-1 nuclear extracts (Fig. 7A), the PAX 8 protein complex was specifically eliminated by competition with 20-and 100-fold molar excesses of unlabeled CT oligonucleotide (Fig. 7B, lanes 4 and 5). In contrast, the PAX 8 complex was not eliminated by competition with as much as a 100-fold molar excess of unlabeled mutant WT1 PAX CON oligonucleotide. Identical results were obtained by using a radiolabeled CT oligonucleotide and an excess of unlabeled mutant WT1 PAX CON oligonucleotide. Radiolabeled mutant WT1 PAX CON oligonucleotide failed to bind either PAX 8 or the Caki-1 extracts (data not shown).
Using EMSA and cotransfection analyses, we examined the rest of the WT1 promoter region and found no additional functional PAX 8-binding sites. Although two potential core binding sites for PAX 8 are located within the 5Ј-flanking region (35) and one within the 5Ј-untranslated region of the WT1 gene, we observed DNA binding only at a novel PAX 8 site, termed WT1 PAX CON, located 250 bp 5Ј of the minimal promoter. Initially, we demonstrated that these three core PAX 8-binding sites previously identified were not responsible for PAX 8 transactivation, since no specific complexes were observed by using 30-bp oligonucleotides containing these potential binding sites. We confirmed this finding by demonstrating that the Distal and Proximal sites thought to mediate PAX 2 and PAX 8 activation of the murine WT1 promoter (44,45) were unable to bind specifically to PAX 8 protein (Fig. 8). Again, we observed only nonspecific binding mediated by the unprogrammed wheat germ lysate controls. These results can be explained by the presence of endogenous Sp1 or AP2-like factors in the unprogrammed lysates binding the Sp1 or AP2 sites within the Distal and Proximal site oligonucleotides. Thus, identical com- FIG. 7. PAX 8 specifically bound a novel site in the WT1 promoter. A, nuclear extracts from Caki-1 cells or in vitro translated PAX 8 proteins from either PAX 8 programmed or unprogrammed control (C) wheat germ lysates were bound to radiolabeled 30-bp oligonucleotide WT1 PAX CON, which contains the novel consensus site (WT sequence shown below). The PAX 8-specific complex is marked by a large arrow, and the complex formed by binding in vitro translated PAX 8 protein was identical to the complex formed by binding Caki-1 extracts. Specificity of the PAX 8 complex was demonstrated by competition with 20and 100-fold molar excesses of unlabeled control PAX 8 oligonucleotide CT and nonspecific control GATA oligonucleotide (G). The Caki-1 complex formed by binding to radiolabeled CT was compared with the complex formed by binding to WT1 PAX CON. Specificity of the complex formed by CT binding was demonstrated by competition with 20-and 100-fold excesses of unlabeled WT1 PAX CON oligonucleotide (WT). The WT1 PAX CON probe contains the 11-bp novel PAX 8 consensus sequence (underlined with arrow depicting reverse orientation) flanked by genomic WT1 sequence. The positive control probe CT (31) also contains the novel PAX 8 consensus binding site. PAX 8 complexes were supershifted (small arrow) by binding cross-reacting PAX antibody (Ab) to the Caki-1 extracts either before (*) or after (ϩ) binding them to the radiolabeled WT1 PAX CON oligonucleotide. B, the lanes are labeled as in panel A. In vitro translated PAX 8 protein or unprogrammed control wheat germ lysates (C) were bound to the radiolabeled WT1 PAX CON (as shown in panel A). The PAX 8 complex is marked with an arrow. Specificity was demonstrated by competition with 20-and 100-fold molar excesses of unlabeled control PAX 8 oligonucleotide CT and 20-or 100-fold molar excesses of unlabeled mutagenized WT1 PAX CON oligonucleotide (MWT). The PAX 8 complex formed by binding of in vitro translated PAX 8 protein to radiolabeled CT was compared with the complex formed by binding of WT1 PAX CON. Specificity was demonstrated by competition with 20-and 100-fold molar excesses of unlabeled WT1 PAX CON oligonucleotide (WT) but not by 20-or 100-fold molar excesses of unlabeled mutagenized WT1 PAX CON oligonucleotide (MWT). The mutant WT1 PAX CON probe (MWT) contains two base pair changes (doubly underlined) that alter the two bases conserved in most PAX 2, PAX 5, and PAX 8 consensus sites. plexes would be formed with both the PAX 8 programmed lysates and unprogrammed lysate controls. In contrast, the human WT1 PAX CON sequence, which contained PAX 8 sites but no Sp1 or AP2 sites, bound specifically to both Caki-1 nuclear extracts containing endogenous PAX 8 protein and in vitro translated PAX 8 protein. Overall, the binding data are consistent with the transactivation data; the WT1 PAX CON sequence, but not the core PAX sites, mediated both PAX binding and transcriptional activation of the WT1 promoter.
To determine whether any other regions within the WT1 promoter are involved in PAX 8 binding, we used four overlapping, PCR-amplified 100 -200-bp fragments that spanned the 652-bp promoter to screen the entire promoter region for evidence of PAX 8 binding. Not only were no additional WT1 PAX CON consensus sites identified by computer analysis, but no additional PAX 8-binding sites were identified by EMSA (data not shown). Thus, a novel functional PAX 8 site, located 250 bp 5Ј of the minimal promoter-bound in vitro translated PAX 8 protein, formed a specific complex with Caki-1 extracts and uniquely mediated potent PAX 8 transactivation of the WT1 promoter. DISCUSSION The development of the kidney requires WT1 expression (25), but how WT1 expression is induced in this tissue is unknown. The following lines of evidence suggest that two transcription factors likely to play a role in this induction are PAX 2 and PAX 8: 1) these genes are expressed with WT1 at the appropriate time during kidney development and in the Wilms' tumors with an epithelial predominant histology; 2) the WT1 promoter contains potential PAX-binding sites; 3) Both PAX 8 and PAX 2 can transactivate WT1 reporter constructs in vitro; and 4) PAX 8 can form DNA-protein complexes with WT1 promoter sequences. Using EMSA, we searched for evidence of PAX 8-DNA interactions throughout the 652-bp WT1 promoter and found only one PAX 8 site with DNA binding activity, located 250 bp 5Ј of the minimal promoter. Thus, we identified a novel PAX 8-binding site in the human WT1 promoter and confirmed that both endogenous and exogenous PAX 8 could up-regulate the transcription of WT1 reporter constructs in vitro. These results suggest that PAX 8 may also modulate WT1 expression in the kidney, and we observed coexpression of PAX 8 and WT1 in five of six kidney cell lines examined. Further, these results address the differential activity of the basal WT1 promoter in the different cell types that co-express WT1, showing that basal WT1 promoter activity in different cell lines correlates with their endogenous PAX 8 expression levels.
Although the common binding site for the PAX 2, PAX 5, and PAX 8 transcription factors has not been identified, we compared the binding sequences for each described target gene and derived a consensus element, which we have identified in the WT1 promoter. Analysis of paired binding sites demonstrated that a sequence containing a binding site nearly identical to WT1 PAX CON (except for one mismatch) can bind the paired domains of PAX 2, PAX 5, and PAX 8 (56). Additionally, one of the PAX-binding sites in the Engrailed-2 promoter (BSII) is also nearly identical to WT1 PAX CON (with one mismatch), and the BSII site has been shown to mediate PAX 2, PAX 5, and PAX 8 binding in vivo as well as in vitro.
Our EMSA results strongly argue against the role of the originally reported potential PAX 8-binding sites (CTGCCC) previously identified by sequence analysis, since they are unable to mediate PAX 8 binding and therefore cannot mediate transactivation. Rather, our results demonstrate that the novel PAX consensus site mediated PAX 8 activation of the human WT1 promoter. This site plays a significant role in PAX 8 activation of the WT1 promoter, as demonstrated by the significant increase in promoter activity observed following the addition of a single copy of this PAX 8 site to the minimal promoter construct. A comparison of the activity of the minimal promoter constructs containing either the active WT1 PAX CON or the inactive core PAX sites showed that only the WT1 PAX CON sequence activated the WT1 promoter and mediated PAX 8 transactivation. Additionally, mutagenesis of the two invariant bases in WT1 PAX CON not only eliminated binding in EMSA but also resulted in a substantial loss of promoter activity in cells containing PAX 8 and greatly reduced the transactivation of the WT1 promoter by exogenously added PAX 8. That mutagenesis did not completely eliminate transactivation suggests that additional proteins present in HeLa cells may help stabilize PAX 8 binding to the mutant site and that those stabilizing proteins are absent in Caki extracts. Alternatively, cryptic binding sites may function (in the absence of the predominant site) during overexpression studies.
The human WT1 PAX CON site was not conserved in the mouse wt1 promoter, although there was an analogous potential WT1 PAX CON binding site 350 bp 5Ј of the minimal promoter. That the WT1 PAX CON site was not conserved in the murine promoter suggests that PAX 8 modulates WT1 expression by using different regulatory elements in the human and mouse systems. While the murine wt1 promoter has recently been shown to be transactivated by both Pax 2 and Pax 8 (44,45), how this transactivation occurs is unclear. The Pax 2-binding site within the murine promoter was not identified (45). The Distal oligonucleotide containing a potential Pax 2-FIG. 8. PAX core binding sites in the human WT1 promoter did not bind PAX 8. In vitro translated PAX 8 protein or unprogrammed control wheat germ lysates (WGL control) alone were reacted with the radiolabeled oligonucleotides Distal (PAX 2), Proximal (PAX 8), and WT1 PAX CON. The PAX 8 complex is labeled and marked by an arrow, and the nonspecific complexes observed with wheat germ lysates (WGL control) alone are indicated by asterisks. The Proximal and Distal oligonucleotides contain the potential PAX 8 and PAX 2 core binding sites, respectively (42). Equivalent amounts of radioactively labeled probes were loaded, and the autoradiograph was exposed for 48 h.
binding site failed to form DNA-protein complexes with the in vitro translated Pax 2 protein, and no additional binding sites were identified. Additionally, in our hands, the human Distal and Proximal oligonucleotides containing potential PAX 8 sites both fail to specifically bind in vitro translated PAX 8 protein (Fig. 8). Possibly differences in EMSA conditions could explain the different results; for example, our binding buffer does not include pBlueScript DNA as a nonspecific competitor but relies solely on poly(dI-dC). In contrast to the lack of binding by the PAX core sites, the human WT1 PAX CON oligonucleotide specifically bound both Caki-1 extracts containing endogenous PAX 8 protein and in vitro translated PAX 8 protein (Fig. 7). Formation of the specific PAX 8-DNA complex was blocked by competition with the oligonucleotide CT, which contains a known PAX 8-binding site, and not by competition with the nonspecific oligonucleotide GATA or the mutant WT1 PAX CON oligonucleotide. Also, the formation of the majority of the PAX 8 complexes by the Caki-1 extracts was prevented, and a portion of the complexes were supershifted by PAX-reactive antibodies.
PAX 2 expression in the developing kidney precedes PAX 8 expression; PAX 2 is present in the ureteric bud, which induces mesenchymal cell condensation, and in the induced condensing mesenchyme. However, unlike PAX 8 and WT1, PAX 2 RNA is absent in the S-shaped bodies, which initially led to the hypothesis that WT1 can repress PAX 2 expression as demonstrated by in vitro reporter assays (15). Therefore, it was hypothesized that expression of PAX 8 enhances WT1 expression, which, in turn, represses PAX 2 expression. This repression of PAX 2 expression is essential for normal kidney differentiation (56).
These results suggest that the PAX-binding factors expressed in the kidney activate the WT1 promoter so that in kidney, strong positive acting transcription factors induce high levels of WT1 expression. The promoters of many kidney growth-promoting genes contain WT1 binding sites, and WT1 represses transcription from the promoters of many growth factor and growth factor receptor genes in vitro, which suggests that WT1 may be an essential part of the cascade of transcription factors controlling kidney development. We hypothesize that the initial lower levels of WT1 expressed in the kidney may be sufficient for its function as a repressor of growth factor and growth factor receptor genes, but after PAX induction, WT1 expression levels can become very high and are then sufficient to activate the WT1 autorepression mechanism, down-regulating further WT1 expression (38,39). This suggests that PAX 8 binding and activation of the WT1 promoter is essential for kidney development.