A highly conserved retinoic acid responsive element controls wt1a expression in the zebrafish pronephros

The Wilms' tumor suppressor gene Wt1 encodes a zinc-finger transcription factor, which is a key regulator of kidney development. Inactivation of WT1 in humans contributes to Wilms' tumor, a pediatric kidney cancer (Call et al.,1990; Gessler et al.,1990), as well as to other genitourinary malformations (reviewed by Rivera and Haber, 2005). A fundamental role for WT1 in the development of the urogenital system, as well as other organs, has also been demonstrated in mice. In this case,inactivation of Wt1 causes organ anomalies affecting the kidneys,gonads, heart, spleen and retina, as well as the olfactory system(Herzer et al., 1999; Kreidberg et al., 1993; Wagner et al., 2002; Wagner et al., 2005).

In both its temporal and spatial pattern, Wt1 expression is exquisitely regulated. In the mouse (and in humans), the highest expression levels are found in the developing kidney and the stromal cells of the gonads and spleen, as well as in the mesothelial cells lining the heart, diaphragm and peritoneum (Armstrong et al.,1993; Pritchard-Jones et al.,1990; Rackley et al.,1993). In the developing kidney, Wt1 mRNA levels are low in the uninduced metanephrogenic mesenchyme and are absent in its reciprocally induced tissue, the ureter epithelium. Wt1 expression rises significantly as the mesenchyme condenses. In the mature nephron, Wt1expression is confined to the podocytes, a highly specialized layer of epithelial cells that form the basis for blood filtration in the glomerulus.

In contrast to mammals, reptiles and amphibians(Kent et al., 1995), fish possess two wt1 paralogs. In the case of zebrafish they are named wt1a and wt1b, and during development they are expressed in an overlapping, but not identical, spatial and temporal pattern(Bollig et al., 2006). With wt1a being expressed earlier than wt1b, the activity of both genes can be detected in the intermediate mesoderm and becomes gradually restricted to the developing glomeruli. Here, wt1b is expressed more laterally than wt1a. After the completion of pronephros differentiation, the expression of wt1a and wt1b is limited to the podocytes.

wt1a and wt1b also differ in terms of their function. Whereas the inactivation of wt1a leads to the failure of glomerular differentiation and morphogenesis, resulting in a rapidly expanding general body edema, the knockdown of wt1b is compatible with early glomerular development (Perner et al.,2007). After 48 hours, however, wt1b morphant embryos develop cysts in the region of the glomeruli and tubules and develop subsequent pericardial edema at 4 days post fertilization. Thus, wt1aplays an early role in pronephros development and is essential for the formation of glomerular structures, whereas wt1b functions at later stages of nephrogenesis.

A significant amount of work has been directed towards the elucidation of the cis and trans elements that regulate Wt1 expression. The Wt1 locus, however, is remarkably complex and our knowledge of its regulation is still fairly limited. In renal cells, for example, 24 kb of 5′ flanking sequence was not sufficient to drive Wt1expression. However, a yeast artificial chromosome (YAC) spanning 620 kb,harboring the murine Wt1 gene flanked by 240 kb and 300 kb of 5′ sequence and 3′ sequence, respectively, was able to drive Wt1 expression in the human renal 293 cell line to a level that is comparable to that of the endogenous gene(Scholz et al., 1997). In the course of this work, an element located ∼15 kb upstream of the transcription start site was identified that included phylogenetically conserved sequence and was required, although not sufficient, for Wt1expression in 293 cells. In a study using transgenic mice, a 280 kb YAC construct containing the human WT1 locus correctly reproduced the endogenous Wt1 expression pattern(Moore et al., 1998).

Two members of the paired-box transcription factors, Pax8 and Pax2, the zinc-finger protein Sp1 and members of the NF-κB/Rel family have been implicated in the regulation of Wt1 expression(Cohen et al., 1997; Dehbi et al., 1996; Dehbi et al., 1998). Pax2 itself is negatively regulated by WT1(Ryan et al., 1995). However,all of these findings are based on transient overexpression and reporter gene assays in cell culture and their in vivo relevance has not been clarified.

In recent years, transgenesis in zebrafish has become a powerful tool to identify enhancers in the control regions of developmentally important genes(Muller et al., 1999; Woolfe et al., 2005). We have used this approach to identify elements that are responsible for the kidney-specific expression of zebrafish wt1a and wt1b. First, we generated stable transgenic GFP lines under the control of ∼30 kb genomic fragments from the upstream regions of wt1a and wt1b. Notably, these lines recapitulated the endogenous expression of both wt1 genes. By a combination of deletion analysis and stable, as well as transient, transgenesis we have identified a distal 336 bp upstream element in the wt1b regulatory region that is necessary and sufficient for GFP expression in pronephric glomeruli. In the upstream regulatory region of wt1a, we could also identify a 299 bp sequence, which was necessary for the proper expression in the area of the developing pronephros. This element, however, did not give rise to glomerular expression but was required for early expression in the intermediate mesoderm. Interestingly, this sequence was bound by members of the retinoic acid receptor family of transcription factors and mediated responsiveness to retinoic acid in vivo and in cell culture experiments.

Zebrafish embryos were obtained from matings of wild-type fish of the TüAB strain that has been kept in laboratory stocks in Würzburg and Jena for many generations. Embryos were raised at 29°C and staged according to Kimmel et al. (Kimmel et al.,1995).

Generation of the construct pBS-wt1b-GFP containing the 25.9 kb XhoI/ApaI fragment from the genomic region of zebrafish wt1b has been described (Perner et al., 2007). For the construction of pBS-wt1a-GFP, a 35.1 kb ApaI/ApaI fragment lying 5′ of the wt1a gene was excised from PAC clone BUSMP706E04239Q9 of the zebrafish genomic P1 artificial chromosome (PAC) library(Amemiya and Zon, 1999) and ligated into the ApaI site of the vector pBS-eGFP(Perner et al., 2007).

The vector pBS-wt1a-6.8kb-GFP was constructed by the insertion of the 6.8 kb XbaI/ApaI fragment lying upstream of the wt1agene into the vector pBS-eGFP. Subsequently, a 100 bp region was deleted using the QuikChange XL Site-Directed Mutagenesis Kit (Stratagene) and primers 5′-CAGTGACTTCTGCTGACATGACAGGTAGGACCAGATCTCAGCTCATC-3′ and 5′-GATGAGCTGAGATCTGGTCCTACCTGTCATGTCAGCAGAAGTCACTG-3′, resulting in the vector pBS-wt1a-6.8kb-delta-GFP.

For transient transgenesis we employed several wt1b-GFP deletion vectors(vectors 1-14), which were all generated using standard cloning procedures. Deletion vectors 2-4 were constructed by the digestion of pBS-wt1b-GFP with the restriction enzymes indicated in the corresponding figures and the subsequent re-ligation of the remaining vector. For deletion vectors 5-8,genomic fragments were amplified by PCR and inserted between the NdeI and ClaI sites of vector 4. Deletions in vectors 9-12 were introduced by means of the QuikChange XL Site-Directed Mutagenesis Kit using vector 4 as a template. The vector pBS-wt1b-coreProm (vector 14) was generated by the excision of a 23.6 kb fragment from the vector pBS-wt1b-GFP using XbaI and the re-ligation of the remaining DNA fragment. An NdeI site in the vector pBS-wt1b-coreProm, which lies 1.5 kb upstream of eGFP, was used for the insertion of the 336 bp wt1b enhancer region, resulting in vector 13 (pBS-wt1b-coreProm-wt1bEnh). The sequences of the used primers and of all cloned fragments are available upon request.

The luciferase reporter vector pGL3-WT1-Enh was constructed by the insertion of the 287 bp human WT1 enhancer region into the vector pGL3-Promoter (Promega) via KpnI and SacI sites. The enhancer region was amplified with primers 5′-CTAGGGTACCTCTCTCGTGGTTTTTCTTTTCC-3′ and 5′-CTAGGAGCTCGGCACTGCCTCTATTATTATACCG-3′ using human genomic DNA as a template. Deletions of the conserved elements 1-3 were introduced by means of the QuikChange XL Site-Directed Mutagenesis Kit.

Generation of transgenic zebrafish was performed essentially as described(Perner et al., 2007; Thermes et al., 2002). Briefly, the plasmids were co-injected with I-SceI meganuclease in 1-cell stage zebrafish embryos. The injection solution contained 150 ng/μl plasmid DNA, 0.3-0.4 U/μl I-SceI meganuclease (Roche), 5 mM MgCl₂, 0.1% Phenol Red (Sigma) and 4 mM Tris-HCl pH 7.5.

For the isolation of transgenic founder fish, 50-100 embryos injected with a given plasmid were raised to maturity and intercrossed. Offspring were analyzed for GFP expression using a fluorescence stereomicroscope(SteREO Lumar.V12, Zeiss). GFP-positive embryos of individual founder fish were used to raise at least three stable transgenic lines per construct.

Zebrafish embryos treated with 0.003% 1-phenyl-2-thiourea (PTU) and 0.016%tricaine were embedded in 1% low melting point agarose with their dorsal side facing the bottom of a μ-Dish (ibidi GmbH, Munich, Germany) or in 3% methyl cellulose. Subsequently, embryos were imaged using an inverted fluorescence microscope (Axiovert 135 TV, Zeiss) or a fluorescence stereomicroscope (SteREO Lumar.V12, Zeiss). Overlays of transmission and fluorescence images were generated using the AxioVision software (Zeiss).

Alignments of long genomic sequences (4-150 kb) were performed with the mVISTA program (Brudno et al.,2003; Frazer et al.,2004). FASTA sequences lying upstream of Wt1 in different vertebrate genomes (wt1a and wt1b in fish genomes) were retrieved using either the Ensembl (Flicek et al., 2008) or the UCSC(Kent et al., 2002) genome browser. The sequences correspond to the following regions in the UCSC genome browser (August 2008): human chromosome 11: 32,412,822-32,420,821; mouse chromosome 2: 104,959,600-104,967,599; Xenopus scaffold 399:895,538-907,235 (for wt1 exon 1 this sequence was complemented by cDNA clone, accession number BC168476); stickleback chromosome 2:12,809,391-12,813,390; zebrafish chromosome 25: 15,039,784-15,044,783. For an alignment of the conserved region upstream of Wt1 at the nucleotide level the ClustalW program (Larkin et al.,2007) was used. For the prediction of transcription factor binding sites the programs MatInspector and PROMO were employed.

In situ hybridization on paraffin sections of adult zebrafish kidney was performed essentially as described(Leimeister et al., 1998). The probe for wt1a is described in Bollig et al.(Bollig et al., 2006). The wt1b probe was transcribed from a 944 bp cDNA fragment including exons 1-5 of wt1b, which was amplified with primers 5′-CATTCTGCTTCAGCAGACGCCTTC-3′ and 5′-CTCACTGCTGTCTGATGATGATT-3′. For GFP immunohistochemistry,paraffin sections were first incubated with rabbit anti-GFP IgG fraction(Invitrogen) and, after PBS washing, with biotinylated goat anti-rabbit IgG antibody (Jackson ImmunoResearch Laboratories). Antigen detection was carried out using the VECTASTAIN ABC Kit and the DAB Substrate Kit for peroxidase(Vector Laboratories). Finally, sections were counterstained with Hematoxylin.

For zebrafish embryos, whole-mount immunohistochemistry was performed before sectioning and counterstaining with DAPI. Embryos were fixed in 4%paraformaldehyde and permeabilized in acetone. After blocking, embryos were incubated with anti-GFP antibody followed by secondary antibody conjugated with Alexa Fluor 488 (Invitrogen).

Proteins for electrophoretic mobility shift assay (EMSA) analysis were prepared by a coupled in vitro transcription/translation reaction using the TNT system (Promega). Constructs pSG5mRARβ and pSG5hRXRα were used as templates, each harboring the full-length cDNA of the respective gene and an upstream T7 promoter. Both plasmids were kindly provided by A. Baniahmad(Jena University Hospital, Jena, Germany). Plasmid Rc/CMV (Invitrogen) was used as a negative control. For probe generation, primers 5′-TTCCCCATTTTTTTCCCCTGTTG-3′ and 5′-TGGCCAGAGCAGATACGTAG-3′ and the template plasmid pGL3-WT1-Enh were used to amplify a 152 bp fragment from the human WT1 enhancer. The mutant competitor fragment was generated using the same primers and pGL3-WT1-Enh-Δ1. For the zebrafish probe (144 bp), primers 5′-TTTCCTCGCTCTCCTCTTCCC-3′ and 5′-TGGCTAAGCAGAGGCAGCTC-3′ and the template plasmid pBS-wt1a-6.8kb-GFP were employed. For generation of the mutant zebrafish fragment, a template harboring the Δ1 deletion in the background of the zebrafish enhancer was employed. Labeling of the probes, sample preparation and gel electrophoresis using a 12% polyacrylamide gel were performed as described (Wilhelm and Englert,2002). For competition experiments, a 500-fold molar excess of unlabeled fragment was used. For supershift experiments, 1 μg of polyclonal anti-RARβ (C-19, Santa Cruz) or a control antiserum (anti-HA, Santa Cruz)were pre-incubated for 10 minutes at room temperature before addition of the probe.

Nuclear extracts from P19 cells were prepared as described(Schreiber et al., 1989). Supershift experiments were performed with 5 μg of nuclear extract and an aliquot of monoclonal anti-RARα (9α9A6; a gift from H. Gronemeyer,IGBMC, Illkirch, France), control monoclonal antibody (anti-GST, Dianova), 0.5μg of polyclonal anti-RARβ or control antiserum.

Murine P19 cells (1×10⁶) were transfected using the Nucleofector Kit V (Amaxa) according to the manufacturer's instructions, with 1 μg of the respective luciferase reporter vector and 50 ng phRL-Tk(Renilla) control plasmid (Promega). Subsequently, cells were incubated for 4 days with 1 μM all-trans retinoic acid or 0.1% ethanol. Firefly and Renilla luciferase activity were measured using the Dual-Luciferase Assay Kit(Promega).

The similarity in Wt1 expression patterns in mammals, frogs and fish during early kidney development suggests that this expression is based on similar regulatory mechanisms that have been conserved through evolution. In a study employing transgenic mice generated using human genomic fragments(Moore et al., 1998), it was shown that regulatory elements driving WT1 expression during kidney development are located 5′ of the WT1 gene. We have therefore assumed that these elements lie in the intergenic region between WT1and the upstream neighboring gene EIF3M(Fig. 1A).

When compared to the WT1 region in humans, both zebrafish regions show partial synteny. Interestingly, only ccdc73 can be found 5′ of wt1a (on chromosome 25) and only eif3m (also known as ga17 or pcid1) can be found 5′ of wt1b (on chromosome 18), although the respective mammalian counterparts (CCDC73 and EIF3M) lie close together and even overlap in their 3′ regions (Fig. 1A). The distance of these genes to wt1 is much smaller in zebrafish (∼30 kb) compared with humans (∼50 kb), which facilitates the search for regulatory elements upstream of wt1a/b.

We cloned the intergenic regions between the zebrafish wt1paralogs and the respective 5′ neighboring genes and fused both fragments to a GFP reporter gene, yielding the constructs wt1a::GFP and wt1b::GFP. In the case of wt1a, this fragment includes the last two exons of the ccdc73 gene, a short putative gene with unknown function, and part of wt1a exon 1 (Fig. 1B). Regarding wt1b, the genomic fragment encompasses almost the entire region between eif3m and wt1b, including part of exon 1 of the latter. The generation of wt1b::GFP transgenic fish has already been published (Perner et al.,2007).

All three wt1a::GFP lines showed a similar and consistent GFP expression pattern, which was restricted to the pronephric region(see below) and the dorsal hindbrain (see Fig. S1 in the supplementary material). The wt1b::GFP lines also showed expression in the pronephros and in several other embryonic organs [e.g. the pancreas, gut, eyes and epicardium, described in detail by Perner et al.(Perner et al., 2007)].

The pronephric expression of wt1a and wt1b in zebrafish is well documented (Bollig et al.,2006; Drummond et al.,1998; Serluca and Fishman,2001; Wingert et al.,2007). wt1a starts to be expressed as bilateral stripes in the intermediate mesoderm 11 hours post fertilization (hpf) at the 3-somite stage, whereas wt1b expression in the pronephric field appears later(14 hpf) and is restricted to a smaller area (see schematic representation in Fig. 2). GFP fluorescence in the transgenic lines recapitulated this pattern with a delay of 1 to 2 hours(Fig. 2). This is probably caused by the translation of GFP mRNA into protein and the accumulation of GFP protein to detectable levels. At later stages (>24 hpf), wt1a and wt1b expression is restricted to a small area ventral of the third somite where the glomeruli are developing. In the functional pronephric glomeruli (>40 hpf), wt1a expression fields are fused in the midline, whereas wt1b is expressed more laterally in the glomeruli. In line with this pattern, a fusion of GFP fluorescence in the glomerular region could be seen only in wt1a::GFP embryos. The only difference between the pronephric expression of the wt1paralogs and the GFP reporter gene in the respective lines was the appearance of GFP expression in the pronephric tubules of wt1b::GFP embryos.

In adult zebrafish the clearance function is accomplished by the mesonephros. From serial histological sections, we estimated the number of nephrons in the mesonephros to be 50-100 (data not shown). We compared the expression of the wt1 paralogs (by in situ hybridization) with the expression of GFP (by immunostaining) in the respective transgenic lines (Fig. 3). The expression of wt1a and GFP in wt1a::GFP transgenic zebrafish was found throughout entire glomeruli.

The wt1b expression profile in the mesonephros is highly variable. Some glomeruli were stained completely(Fig. 3B) and others possessed wt1b mRNA only in a subset of cells. This variability was reflected by the GFP expression pattern, with complete but also with partial staining of the glomeruli. Only the appearance of GFP expression in the neck region of the tubule differed from the wt1b expression pattern. This is highly reminiscent to the situation in the embryonic kidney of wt1b::GFP fish (Fig. 2) where GFP, but not wt1b, is expressed in the pronephric tubule.

In conclusion, both in the embryonic pronephric kidney and in the mesonephros, transgene-driven GFP expression recapitulates the endogenous expression of the wt1 genes.

Using different bioinformatics tools we could not find any significant conservation of the region 5′ of wt1b compared to regions upstream of Wt1 in other vertebrates. We therefore decided to systematically narrow down the region(s) upstream of wt1b that is/are important for expression in the pronephros by generating several wt1b-GFP deletion constructs and using transient transgenesis(Fig. 4A). Using this strategy,we could localize two adjacent regions of 170 bp that were indispensable for glomerular expression (construct numbers 10 and 11) and that together represent a 336 bp region necessary for reporter gene expression in the pronephros. We then wanted to test whether the identified element was also sufficient to drive glomerular expression. To this end, we inserted the 336 bp element close to the wt1b transcription start site. Using the resulting construct (construct number 13), we did indeed find expression in the glomerulus, but not when the 336 bp enhancer region was absent (construct number 14).

To confirm the results from transient transgenesis we generated stable transgenic lines with the two latter constructs. Two lines harboring the wt1b-coreProm transgene (construct number 14) displayed only weak and diffuse GFP expression (Fig. 4B, left panel). By contrast, injection of the wt1b-coreProm-wt1bEnh construct(construct number 13) gave rise to five transgenic lines all showing a strong GFP signal anterior and ventral to the third somite(Fig. 4B, right panel),indicating expression in the pronephric glomeruli.

Taken together, we identified a 336 bp enhancer region located 21 kb 5′ of wt1b exon 1, which is necessary and sufficient for wt1b expression in the pronephric glomeruli of zebrafish. We did not find any similarities between this sequence (see Fig. S2 in the supplementary material) and the zebrafish wt1a locus or wt1 loci in other fish genomes.

In contrast to the situation for wt1b, we found a conserved region 4.2 kb upstream of the wt1a gene when we aligned the genomic region 5′ of wt1a with regions lying upstream of Wt1 in human, mouse and frog, and upstream of wt1a in stickleback(Fig. 5A). The distance of the conserved region to the first exon of Wt1 varies from 2.1 kb(stickleback) to 10.9 kb (Xenopus). The conserved region could be detected also in several other vertebrates, such as rat, opossum, platypus,lizard, medaka and pufferfish (data not shown), and harbors three highly conserved elements (cons 1-3 in Fig. 5B) with 12, 16 and 10 identical nucleotides, respectively. The first conserved element (cons 1,5′-AGTTGA-(N)₅-GGGTCA-3′) is predicted as an RXR heterodimer binding site. For the other two elements (cons 2 and 3), several transcription factor binding sites were suggested, however, the significance of these predictions is low due to the shortness and the frequent occurrence of the sites.

To test the hypothesis that the conserved region upstream of wt1aacts as a regulatory element in vivo, we generated GFP reporter fish with two further constructs (6.8 kb and 6.8 kb delta, Fig. 6A). In the first construct, ∼28 kb is deleted from the 5′ end of the vector wt1a-GFP(which is 35 kb), and the conserved region is still present. In the second construct, a 100 bp region including the first two conserved elements is deleted within the 6.8 kb fragment (indicated in Fig. 5B). In the offspring of all founder fish harboring the 6.8 kb construct, GFP was expressed in bilateral stripes in the intermediate mesoderm(Fig. 6B), ventrolateral to the somites. Compared with wt1a::GFP embryos (35 kb fragment)this early expression in the intermediate mesoderm was expanded anteriorly, as well as posteriorly. At 42 hpf, GFP fluorescence could be detected outside of the glomerular region of wt1a-6.8kb::GFP embryos,whereas in the glomeruli (adjacent to the third somite) GFPexpression was low. This indicates that the deleted 5′ region harbors elements that are necessary for both restricted wt1a expression in the intermediate mesoderm and expression at later stages in the glomeruli.

Deletion of the conserved 100 bp fragment in the transgene, to give wt1a-6.8kb-delta::GFP embryos, completely abrogated early GFP expression in the intermediate mesoderm (see Fig. S3 in the supplementary material). Altogether, our data strongly indicate that the conserved region upstream of wt1a is indispensable for early wt1a expression in the intermediate mesoderm.

Two pieces of evidence suggested that the conserved region upstream of wt1a could be a target for retinoic acid (RA) signaling. First, a conserved element within this region is predicted to act as an RXR heterodimer binding site (Fig. 5B, cons 1). Second, it was shown recently that treatment of zebrafish embryos with all-trans retinoic acid (atRA) leads to a strong increase of wt1amRNA levels in the intermediate mesoderm(Wingert et al., 2007). In order to test whether the conserved element upstream of wt1a indeed mediates RA responsiveness, we treated zebrafish embryos from different transgenic lines with 1 μM atRA from 12 hpf to 24 hpf(Fig. 7). In atRA-treated wt1a::GFP embryos (35 kb fragment) a strong increase in and also an expansion of GFP expression was detectable, whereas in wt1b::GFP embryos reporter gene expression decreased slightly after RA treatment. This indicates the presence of RA responsive element(s) in the region upstream of wt1a. The strongest increase of GFP fluorescence was observed in atRA-treated embryos harboring the truncated wt1a transgene (wt1a-6.8kb::GFP). With this construct, ectopic GFP expression was found in the brain and in the tail, probably due to a loss of inhibitory elements. By contrast,responsiveness of the reporter gene to RA was completely lost in all three wt1a-6.8kb-delta::GFP lines that lack the 100 bp fragment within the conserved region. In conclusion, we provide evidence that the conserved region upstream of wt1a is responsive to RA in vivo.

Retinoic acid acts by binding to heterodimers of one of the family members of the retinoic acid receptor (RAR) α/β/γ transcription factors with the partner protein retinoic X receptor (RXR). We therefore examined binding of RAR/RXR proteins to the Wt1 enhancer by EMSA analysis. The zebrafish and the human Wt1 enhancers encompassing all three conserved elements were bound by complexes of RARβ and RXRαbut not by either of the proteins alone(Fig. 8A). The binding could be competed with an excess of unlabeled wild-type probe but not with a fragment that harbored a deletion of the conserved region 1. No gel shift could be observed when the latter was used as labeled probe (data not shown). This shows that members of the RAR/RXR family preferentially bind to conserved region 1. Pre-incubation of the extracts with RARβ antiserum, but not with a control antiserum, resulted in immunodepletion of the DNA-protein complexes. Of note, the zebrafish wt1a enhancer was efficiently bound by a complex of the mammalian RARβ and RXRα proteins.

We then used extracts from P19 mouse embryonic carcinoma cells and performed supershift experiments using the human WT1 enhancer and antibodies against RARβ and RARα. We chose P19 cells because it has been shown that Wt1 expression is strongly induced in these cells after differentiation with RA (Scharnhorst et al., 1997). After 24 hours of treatment with RA, the extracts showed a supershift with the RARβ as well as with the RARα antibody(Fig. 8B). In the absence of RA, a supershift was only obtained with the RARα antibody. This could be explained by the presence of different levels of the two transcription factors in P19 cells. Moreover, RA treatment of P19 cells seemed to enhance RARβand RARα levels and/or increased their binding affinity. These data show that the endogenous proteins of the RAR family can also bind to the WT1 enhancer.

In order to test the function of the conserved region upstream of human WT1, we performed luciferase reporter assays in P19 cells. The human conserved region (287 bp) was cloned in front of the SV40 promoter(pGL3-WT1-Enh, Fig. 8C). In addition, constructs were generated in which the elements that show the highest conservation among vertebrates were deleted (Δ1-3). No significant difference in luciferase activity was seen in cells transfected with the different constructs after treatment with the vehicle alone(Fig. 8D). However, in cells treated with RA the WT1 enhancer region mediated a threefold increase in luciferase activity compared with pGL3-transfected cells. In line with the EMSA analysis, deletion of the first conserved element led to a total abrogation of this effect, whereas the enhancer regions harboring either the second or third deletion were still partially responsive to RA. This demonstrates that the conserved enhancer upstream of Wt1 is involved in the induction of Wt1 expression by RA both in vivo in zebrafish and in mammalian cells, thus underscoring the function of this element.

In this study, we took advantage of the more compact wt1 loci(compared with human and mouse) and efficient transgenesis in zebrafish. We identified two genomic regions that are important for the expression of the wt1 paralogs in the developing zebrafish pronephros. The first 336 bp element lies ∼21 kb upstream of the wt1b gene and drives glomerular expression, whereas the second 299 bp element lies 4.2 kb upstream of the wt1a gene, is responsive to retinoic acid (RA) and is necessary for early expression in the intermediate mesoderm.

Based on the similarities of the Wt1 expression pattern in different vertebrates, including mammals, reptiles, amphibians and fish(Bollig et al., 2006; Carroll and Vize, 1996; Kent et al., 1995), we assumed that the mechanisms driving the glomerular expression of Wt1 are conserved between mammals and fish, and involve a set of orthologous transcription factors. Surprisingly, we could not find any sequence conservation of the identified 336 bp wt1b enhancer region in mammals and not even in other fish species such as medaka, fugu and tetraodon.

There is a growing list of studies demonstrating that the regulatory function of enhancer elements is maintained over long evolutionary distances despite the remarkable turnover of transcription factor binding sites. For instance, it was shown that regulatory sequences from the human RETlocus are capable of driving ret-specific expression in zebrafish,even though no sequence similarity between human and zebrafish regulatory elements was detected (Fisher et al.,2006). Along the same line, enhancer elements from the even skipped locus, which are functionally conserved between scavenger flies(Sepsidae) and vinegar flies (Drosophilidae), do not show sequence conservation (Hare et al.,2008). Based on these studies, we assume that elements exist in other vertebrates, which are functionally equivalent to the wt1benhancer region that drives expression in the developing pronephric glomerulus. However, alterations within transcription factor binding sites such as base substitutions might obscure the identification of homologous elements by sequence comparison.

We have identified a 299 bp region ∼4.2 kb upstream of the wt1a gene in zebrafish that is conserved in other fish species,amphibians, reptiles and mammals. According to the criteria set by Bejerano et al. (Bejerano et al., 2004),the respective region upstream of human WT1 can almost be regarded to be an `ultraconserved' element, with an identity of 97.8% over 226 bp to the orthologous regions in mouse and rat. Interestingly, it has been reported that`ultraconserved' regions are often located around genes that are involved in the regulation of development, including a large set of transcription factor genes (Bejerano et al., 2004; Woolfe et al., 2005). Many of these conserved regions have been shown to function as tissue-specific enhancers in transgenic zebrafish (Woolfe et al., 2005) and in transgenic mice(Pennacchio et al., 2006).

It is notable that the genes wt1a and wt1b themselves show very high sequence homology (Bollig et al., 2006), whereas the respective regulatory regions are vastly different. This is in line with the notion that after gene duplications the control region sequences of the duplicates change much faster than the sequences within the genes, and that subfunctionalization occurs at the regulatory level rather than at the level of the respective coding region(Kleinjan et al., 2008).

Retinoic acid (RA) has been shown to act as a morphogen in embryonic development (Ross et al.,2000). In this process, its local concentration is regulated by both synthesizing (Raldh family) and metabolizing (cytochrome P450) enzymes(Ross et al., 2000). RA controls gene expression at the transcriptional level, functioning as a ligand for heterodimeric nuclear receptors (RXR and RAR) that typically bind to direct repeats of hexamer sequences separated by one to five nucleotides(Germain et al., 2006).

Within the wt1a enhancer we found three stretches of nucleotides that are conserved between mammals and fish. All investigated sequences harbored the motif AGTTGA-(N)₅-GGGTCA, predicted as a nuclear receptor binding site, within the first conserved element. Deletion of a region including the first two conserved elements caused a total loss of intermediate mesoderm expression and responsiveness to RA in transgenic fish in vivo. Consistently, an invitro-translated RAR/RXR complex bound to the human as well as the zebrafish Wt1 enhancer element in vitro. This binding was abrogated upon deletion of conserved region 1. In addition, we could show by supershift experiments that endogenous RARα and RARβfrom P19 cells also bind to the human WT1 enhancer. Finally, in reporter gene assays using P19 cells, deletion of either one of the three elements in the context of the human enhancer region led to a significant reduction of RA-mediated activation, whereby deletion of the first element had the most dramatic effect.

In zebrafish, the inhibition of RA signaling results in a loss of the glomeruli and proximal parts of the tubules(Wingert et al., 2007). It has been shown in this study by Wingert et al. that RA-deficient embryos failed to express wt1a in the intermediate mesoderm, whereas RA-treated embryos showed increased wt1a expression. This indicates that wt1aexpression in the intermediate mesoderm is driven by endogenous RA and that in RA-deficient embryos the lack of Wt1a contributes at least in part to the observed phenotype. The authors of this study proposed a model in which the adjacent paraxial mesoderm is the source for RA.

Our data suggest that the induction of wt1a by RA is mediated by the conserved enhancer that we have identified. It is tempting to speculate that this is a common mechanism in all vertebrates, as both the conserved region and early Wt1 expression in the intermediate mesoderm(Armstrong et al., 1993; Kent et al., 1995) can be found in several vertebrates.