WT1 regulates the development of the posterior taste field

The sense of taste is mediated by taste receptor cells that detect chemicals in the oral cavity and translate that information into an output signal that is sent to the brain. Taste receptor cells are housed in taste buds, which each comprise 50-150 taste cells. In mammals, most taste buds are located within specialized areas called papillae that are set within discrete areas of the tongue (Mistretta, 1991). There are three types of taste papillae on the tongue: (1) fungiform papillae that are dispersed on the anterior portion of the tongue and house one or two taste buds each; (2) circumvallate (CV) papillae that are located on the back of the tongue; and (3) foliate papillae, which are found on both sides of the tongue. Both the foliate and CV papillae house hundreds of taste buds. Although most taste buds are located within these papillae on the tongue, there also are some on the palate and throat (Kinnamon, 1987). Taste papillae comprise a mesenchymal core that is surrounded by an invaginated epithelium, where the taste buds are located (Kapsimali and Barlow, 2013).

Several signaling proteins, including members of the sonic hedgehog (SHH), bone morphogenetic protein 4 (BMP4) and Wnt/β-catenin signaling pathways are required for normal peripheral taste development (Beites et al., 2009; Bitgood and McMahon, 1995; Iwatsuki et al., 2007; Jung et al., 1999; Liu et al., 2007), but the factors controlling the expression of these pathways in the taste system have not been identified. In this study, we show that expression of the transcription factor Wilms' tumor 1 (WT1) is coincident with the emergence of CV taste placodes, the first discernible feature of the developing CV papillae. We found that Wt1 knockout mice exhibit significant defects in the anatomical specializations associated with CV papillae development. We show that WT1 regulates the expression of BMP4 as well as genes from the SHH and Wnt/β-catenin signaling pathways. Our data demonstrate for the first time that WT1 is required for the normal development of the posterior taste field.

Previous PCR analysis of mRNA isolated from mouse tissues detected Wt1 mRNA in the embryonic and adult tongue (Armstrong et al., 1993). Considering that WT1 affects the development of other sensory tissues (Wagner et al., 2002, 2005), we initiated a more thorough analysis of its expression in the developing mouse taste system. We first determined where WT1 was localized within the developing tongue using immunocytochemical analysis to characterize expression in the developing CV papillae (Fig. 1).

At embryonic day (E) 12.5, the peripheral taste system begins to form as a thickening of epithelial cells, termed placodes (Barlow, 2000; Mistretta, 1972). Analysis of E12.5 tongues where the placode is developing demonstrated that WT1 (Fig. 1A and Fig. 2A) is expressed at the same time as cytokeratin 8 (detected with anti-Troma1; Fig. 1B), an embryonic taste precursor cell marker (Mbiene and Roberts, 2003), and GAP43 (Fig. 1C), which labels the gustatory nerve that innervates the taste buds (Mbiene and Mistretta, 1997). We were unable to detect WT1 expression in E11.5 embryos (data not shown). By E14.5, the invaginations that will become the CV crypts had formed (Fig. 1A and Fig. 2A); WT1 expression became less diffuse and was present in the epithelia of the developing papillae and the surrounding tissue.

By birth [postnatal day (P) 0], WT1 expression was largely confined to the epithelium of the taste bud (Fig. 1A and Fig. 2A). Higher magnification revealed that WT1 expression is both cytoplasmic and nuclear in the cells that are forming the taste papillae (Fig. 2A, bottom panels). This is consistent with previous observations that WT1 can be present in both the nucleus and cytoplasm (Hohenstein and Hastie, 2006; Niksic et al., 2004; Vajjhala et al., 2003). WT1 was also detected in adult CV papillae (supplementary material Fig. S1A) but was primarily restricted to the taste buds. Immunoblotting of extracts from isolated adult taste cells confirmed the expression of WT1 (supplementary material Fig. S1B). Thus, WT1 is expressed from the early placode stage of taste development and continues to be expressed into adulthood. WT1 expression in the developing taste placode was independently confirmed using a different WT1 antibody (supplementary material Fig. S2).

Although the mechanisms controlling the development of the peripheral taste system are not well defined, several early regulators have been identified, including SHH, BMP4 and multiple members of the Wnt/β-catenin signaling pathway. Lef1 (which encodes a major effector of canonical Wnt signaling), Ptch1 (SHH receptor) and Bmp4 (a TGFβ ligand) play crucial roles in taste papillae formation (Iwatsuki et al., 2007; Zhou et al., 2006) and have been identified and established as WT1 target genes in the embryonic kidney (Hartwig et al., 2010). We therefore considered the possibility that WT1 regulates the expression of these genes during taste development and examined the embryonic expression of Lef1, Ptch1 and Bmp4 in the developing CV papillae. We were able to detect expression of all three proteins during mid-gestation, when the initial anatomical specializations that comprise the CV papillae are forming (Fig. 3A).

We then used chromatin immunoprecipitation (ChIP) to determine whether the WT1 binding sites upstream of the Lef1, Ptch1 and Bmp4 genes are bound by WT1 in the developing peripheral taste system. Tongues were isolated from embryonic mice of a BALB/c background at E15.5 and subjected to ChIP analysis with either control IgG or WT1 antibodies (Fig. 3B). We found that WT1 was bound at the promoter regions of Lef1, Ptch1 and Bmp4 in the developing tongue and not at a control region 3′ of each gene. Similar results were obtained for ChIP assays performed using E15.5 tongues isolated from the Wt1 knockout strain that was purchased from Jackson Labs and not backcrossed into the BALB/c background (supplementary material Fig. S3). Follow up experiments confirmed that WT1 expression overlaps with that of all three of these proteins during development (Fig. 3C).

To determine if WT1 is required for normal taste development, we obtained Wt1 knockout (KO) mice that contain a mixed genetic background. These WT1-KO embryos do not survive beyond mid-gestation but backcrossing the WT1-KO mutation into a BALB/c background for one generation allows some embryos to survive until birth (Herzer et al., 1999). Herzer et al. (1999) concluded that the WT1-KO phenotype is less severe in the BALB/c background, which suggests some epigenetic control of the penetrance of the Wt1 mutation. We found that most WT1-KO embryos in this background still die mid-gestation (only 8% of surviving embryos were WT1-KO, n=263 embryos). Analysis of P0 BALB/c WT1-KO mice revealed that the cells forming the CV papillae failed to stain with WT1 antibodies (Fig. 4A). Analysis of E12.5 WT1-KO embryos revealed that the developing placode that will eventually become the CV papillae was significantly impaired (Fig. 4B-E). Using the Troma1 antibody, we labeled cells in the placodes at E12.5 (Fig. 4B). All sections of the KO and wild-type embryos within the vicinity of the CV express Troma1, but the KO embryos lacked the typical epithelial thickening that indicates the onset of CV papillae development. Thus, WT1 is required for normal CV placode formation.

Immunocytochemical analysis to detect other important markers in peripheral taste development further established the vital role of WT1. SOX2 (Fig. 4C), SHH (Fig. 4D) and GAP43 (Fig. 4E) labeling confirmed that, at E12.5, there is a lack of epithelium specialization and loss of organization in the surrounding tissue of WT1-KO embryos. SOX2 labeling was essentially lost, whereas SHH labeling was no longer specific to the developing placode. GAP43 labeling was used to identify the gustatory nerve that normally innervates the papillae during development. In WT1-KO embryos from the BALB/c background, we were unable to detect any GAP43 labeling at E12.5 (Fig. 4E). We therefore analyzed GAP43 expression at E13.5 and E14.5 in WT1-KO embryos compared with wild type (supplementary material Fig. S4). Consistent with the impaired CV development, the WT1-KO embryos showed a delay in innervation of the taste bud. Of the WT1-KO embryos from the mixed genetic background, all died mid-gestation (∼E15), but we were able to analyze some embryos at E12.5 when the placodes are beginning to form. In this more severe WT1-KO phenotype, the CV placode was completely disrupted (supplementary material Fig. S5).

Our data so far suggest that WT1 is required for normal development of the CV papillae. Furthermore, we demonstrated that WT1 is coexpressed with the products of the known WT1 target genes Lef1, Ptch1 and Bmp4, and that WT1 localizes to the promoter regions of these genes in the developing tongue. Analysis of LEF1, PTCH1 and BMP4 expression in E14.5 wild-type and KO embryos from the BALB/c background revealed reduced expression of these proteins in the KO embryos (Fig. 5A-C). LEF1expression at this time is low in the wild-type embryo, but is further reduced in the KO. PTCH1 expression is relatively widespread at this point in development, but in the KO embryo, is reduced in the epithelia that are forming the papillae. BMP4 expression appears to be largely absent in the KO embryo, in contrast to the wild type. Thus, the loss of WT1 is associated with significant disruption of LEF1, PTCH1 and BMP4 expression.

We next prepared mRNA from the tongues of wild-type or WT1-KO E13.5 embryos and analyzed the expression of Lef1, Ptch1 and Bmp4 by qPCR (Fig. 5, right panels). Lef1, Ptch1 and Bmp4 mRNA levels were significantly reduced in tongues derived from WT1-KO embryos as compared with wild type. We conclude that WT1 is essential for the normal development of the peripheral taste system and that it is first required during initial placode formation. Although these data demonstrate that WT1 is regulating the expression of Lef1, Ptch1 and Bmp4, we cannot conclude that these are the only genes affected by the absence of WT1. Further studies are needed to determine if other factors that are crucial in the development of the CV papillae are also regulated by WT1.

We also examined the CV papillae in the BALB/c wild-type and WT1-KO pups that survived to P0. At birth in the wild type, the CV papillae are formed but the taste buds are not yet mature (Kapsimali and Barlow, 2013). Analysis of Troma1-expressing cells revealed defective expression in CV papillae from the KO mice. The Troma1-expressing cells were not confined to the epithelial layer and were aberrantly localized to the papillae crypt (Fig. 6A): whereas 75% of the Troma-expressing cells were found in the CV epithelium of wild-type mice, only 25% of these cells were localized in the epithelium of WT1-KO mice.

We have shown above (supplementary material Fig. S4) that GAP43 expression is significantly delayed in the developing CV of the WT1-KO embryo. When we analyzed GAP43 expression at P0, the gustatory nerve was branched around the taste bud areas but failed to innervate the individual taste buds forming in the WT1-KO embryos as compared with wild type (Fig. 6B). Thus, even though the basic papillae structure ultimately developed in these KO embryos, the CV papillae did not develop normally.

ChIP analysis revealed that WT1 still binds to the promoters of Lef1, Ptch1 and Bmp4 at P0 in the BALB/c mice (supplementary material Fig. S6A), and analysis of mRNA from wild-type and WT1-KO P0 taste tissue found that Lef1, Ptch1 and Bmp4 expression was significantly reduced in the WT1-KO specimens compared with wild type (supplementary material Fig. S6B).

Our data suggest that WT1 affects CV papillae development by regulating the expression of target genes involved in signaling pathways that have known roles in taste development. To better understand the role of WT1 in the peripheral taste system, we used primary taste cells in culture as a model to study WT1 function. WT1 expression is robust in the cultured mouse taste cells, in particular the type II taste cells that express TRPM5 (Fig. 7A). Cultured taste cells were transfected with either control or WT1 siRNA. After 48 h, immunocytochemical analysis found that WT1 was significantly reduced after transfection with WT1 siRNA but not control siRNA (Fig. 7B). Real-time PCR analysis also revealed a significant reduction in Wt1 mRNA levels after transfection with WT1 siRNA, as well as a significant decrease in Lef1 and Ptch1 expression (Fig. 7C). Since WT1 does not bind to the Bmp4 promoter in adult mice, we did not analyze BMP4 expression in the cultured taste cells. We conclude that WT1 actively regulates both Lef1 and Ptch1 in taste cells.

Multiple factors have been shown to play a role in peripheral taste development, the best characterized of which are the Wnt/β-catenin and SHH pathways. The Wnt/β-catenin pathway is required for fungiform taste placode formation, whereas the SHH pathway is an important negative regulator of placode formation (Bitgood and McMahon, 1995; DasGupta and Fuchs, 1999; Iwatsuki et al., 2007; Jung et al., 1999; Liu et al., 2007, 2004). Although both of these signaling pathways are crucial for normal development of the peripheral taste system (Kapsimali and Barlow, 2013), the initial induction factor for these pathways has not been identified. Our results suggest that WT1 is required for the normal induction of Ptch1 and Lef1 expression during CV papillae development. These factors are crucial components of the SHH and Wnt pathways that regulate peripheral taste development, although their function is better defined in fungiform than in CV papillae development (Kapsimali and Barlow, 2013). WT1 was also required to regulate BMP4 expression in embryonic taste cells. We note that WT1 expression is sustained in adult taste buds, suggesting its continued requirement for the formation of new taste cells and/or their maintenance.

Previous studies found that SOX2 is required for the formation of the placode in the fungiform papillae (Okubo et al., 2006). Our results suggest that the level and expression pattern of SOX2 is disturbed in the development of the CV of WT1-KO mice. Although Sox2 has not yet been reported as a WT1 target gene, it is possible that there is some level of connection between these factors in the developing CV papillae. CV papillae formation depends on a balance of expression between Sprouty and FGF genes, which regulate receptor tyrosine kinase signaling to control the number of CV papillae formed (Petersen et al., 2011). Whether there is any relationship between these genes and WT1 expression is currently unknown, but since loss of WT1 expression causes aberrant CV formation, it would be interesting to investigate the potential relationship between these signaling molecules.

In the developing CV, WT1 expression coincides with that of LEF1, PTCH1 and BMP4, which is consistent with a role for WT1 in their expression. We note, however, that LEF1 expression is not evident in all the regions of the placode that express WT1, suggesting that other factors are also important. Indeed, we and others have previously shown that the function of WT1 in transcriptional activation is modulated by its interaction with BASP1 and that this is both cell type- and gene promoter-specific (Essafi et al., 2011; Goodfellow et al., 2011; Green et al., 2009). It will therefore be important to study the potential role of BASP1 in modulating WT1 function during CV formation. Indeed, BASP1 is highly expressed in developing neuronal tissues (Carpenter et al., 2004) and might have a role in the developing taste system. Future studies are needed to address this question.

WT1 plays a crucial role in reciprocal mesenchyme-epithelial transitions, with the cell type direction specified by WT1 co-factors such as BASP1 (Essafi et al., 2011). It will be interesting to determine if WT1 engages its co-factors to regulate the mesenchymal-epithelial interactions that are important in the normal development of the taste buds (Kapsimali and Barlow, 2013). WT1 controls the development of many organ systems and has been shown to be important in the normal development of other peripheral sensory systems including the olfactory epithelium and retinal ganglia (Wagner et al., 2002, 2005). Our data have revealed that WT1 is required for the normal development of the peripheral taste system, in particular the posterior taste field that encompasses CV papillae formation. Taken together with the earlier findings, our data support an extensive role for WT1 in the formation of sensory tissues.

Two types of WT1-KO mice were used in these experiments. WT1-KO mice from Jackson Labs (stock number 010911) have a mixed genetic background containing C57BL/6, Swiss Webster and B6129SF1/J. These mice have an endogenous Wt1 coding region replaced with the gene for green fluorescent protein (GFP) and die between E12 and E15. Where indicated, Wt1 heterozygous mice were crossed with BALB/c mice for one generation and heterozygous offspring were mated to obtain WT1-KO mice. These WT1-KO embryos can survive longer in utero and in some litters survive until birth. Of the 263 embryos collected from the BALB/c backcrossed mice, 21 WT1-KO embryos were collected. Some of these embryos were breaking apart and were not usable. A total of seven KO embryos survived to birth.

Animals were cared for in compliance with the University at Buffalo Animal Care and Use Committee. Taste-enriched tissue was collected from tongues at E15.5 for the qPCR and ChIP experiments. In these experiments, the back half of the tongue was collected and homogenized for analysis. For the taste samples collected from P0 mice, the epithelium was peeled and the CV was closely cropped to minimize the amount of non-taste tissue included. Isolated taste receptor cells were harvested from the adult CV papillae as previously described (Hacker et al., 2008) for qPCR, western blot and ChIP experiments.

Immunocytochemical analysis was performed as previously described (Rebello et al., 2011; Starostik et al., 2010). Antibodies were: anti-WT1 [89 (Toska et al., 2012); 1:50], anti-WT1 (C-19; sc192; 1:50), anti-SHH (sc-9024; 1:50), anti-GAP43 (sc10786; 1:50) from Santa Cruz Biotechnology, anti-Troma1 from Developmental Studies Hybridoma Bank (Iowa City, IA, USA; 1:50), anti-LEF1 (C1285; 1:50; from Signal Transduction), anti-SOX2 (AB97959; 1:50), anti-BMP4 (AB155033; 1:50), anti-WT1 (AB89901; 1:50) and anti-PTCH1 (AB53715; 1:50) from Abcam. Secondary antibodies were purchased from Jackson ImmunoResearch. Some experiments were performed with anti-WT1 that had a fluorophore directly attached using the Mix-n-Stain CF Dye antibody labeling kit (Biotium). In all images, only the brightness and contrast were adjusted. All data are shown as a single 0.1 μm slice from a confocal image except where indicated in the figure legend. If comparisons were made between sections, the laser intensity and brightness/contrast were matched between the conditions being compared.

Total RNA was prepared using the Qiagen RNeasy kit, and cDNA was prepared using the Bio-Rad cDNA synthesis kit. ChIP assays were performed as described previously (Hartkamp et al., 2010). Primers for RNA analysis and ChIP analysis are listed in supplementary material Table S1. All experiments consisted of at least three biological repeats. Data for both the ChIPs and qPCR are reported as average value with standard deviation.

Isolated mouse taste receptor cells were cultured in Iscove's modified Dulbecco's medium on treated coverslips as previously described (Ozdener et al., 2006). After 4 days of culture, cells were transfected with i-Fect (Neuromics) containing control siRNA (Thermo Scientific) or WT1 siRNA (Qiagen). After 48 h, cells were harvested and immunocytochemical or mRNA analysis was performed.

We thank Nicholas Costable for technical assistance with genotyping the embryos for analysis.