PPARG contributes to urothelial integrity in the murine ureter by activating the expression of Shh and superficial cell-specific genes Open Access

In the main organs of the mammalian urinary drainage system, the ureters and the bladder, a highly specialized stratified epithelium, the urothelium, lines the luminal surfaces. The urothelium acts primarily as a physical barrier, preventing the uncontrolled exchange of substances between the urine and interstitial fluids. Central to this function is the luminal layer of umbrella or superficial (S) cells, which are interconnected by tight junctions. S cells are large and highly expandable post-mitotic cells with two or more nuclei. They are characterized by the synthesis of specialized glycoproteins, uroplakins (UPKs), which assemble into semi-crystalline plaques on the apical surface. S cells are underlain by small intermediate (I) cells that form layers with one to several tiers. A layer of cuboidal basal (B) cells connects the urothelium to the basement membrane and the fibromuscular wall (Arrighi, 2015; Castillo-Martin et al., 2010; Dalghi et al., 2020). In addition to shape and location, urothelial cells can be molecularly distinguished by the combinatorial expression of cytokeratin 5 (KRT5), an isoform of the transcription factor p63 (ΔNP63) and UPKs. B cells are KRT5⁺ΔNP63⁺UPK1B⁻, I cells are KRT5⁻ΔNP63⁺UPK1B⁺ and S cells are KRT5⁻ΔNP63⁻UPK1B⁺, with levels of UPK1B being significantly lower in I cells than in S cells (Bohnenpoll et al., 2017a; Gandhi et al., 2013).

The urothelial cytoarchitecture results from highly coordinated proliferation, patterning and differentiation processes acting on epithelial progenitors (the cloaca in the case of the bladder, or the stalk region of the ureteric bud in the case of the ureter), starting in the mouse around embryonic day (E) 10.5. After proliferative expansion of these mono-layered epithelial primordia, stratification and I cell differentiation occur. Two days later, luminal cells express UPKs, and after another 2 days, B cell differentiation begins. I cells continue to proliferate between these two layers. Under homeostatic conditions, the urothelium is largely quiescent with a limited replenishment of B cells and S cells by I cells (Bohnenpoll et al., 2017a; Gandhi et al., 2013). In response to injury or infection, S cells are shed and replaced by I cells (Gandhi et al., 2013). When S cells and I cells are depleted, a small subset of B cells, characterized by expression of KRT14 and possibly KRT15, may act as progenitors for all urothelial cells (Papafotiou et al., 2016; Tai et al., 2013).

The regulation of urothelial differentiation is poorly understood but several studies have implicated the ligand-dependent transcription factor peroxisome proliferator activated receptor gamma (PPARG) in this program. PPARG was originally characterized as a member of a small subset of the nuclear receptor superfamily with important roles in lipid metabolism. Peroxisome proliferator activated receptors (PPARs) form heterodimers with retinoid X receptors (RXRs) and bind to specific peroxisome proliferator response elements (PPREs) in the promoters of target genes. Upon binding of specific ligands, these heterodimeric complexes bind to coactivator complexes to activate gene transcription (for reviews, see Blanquart et al., 2003; Desvergne and Wahli, 1999; Varga et al., 2011).

Expression of PPARG protein was first described in the presumptive urothelium of the mouse urogenital sinus and in the mature urothelium of mice, rabbits and humans (Guan et al., 1997; Jain et al., 1998; Kawakami et al., 2002). Activation of PPARG by high-affinity agonists in cultures of normal urothelial cells suppresses the growth of these cells (Nakashiro et al., 2001), and induces gene expression changes associated with S-cell differentiation (Varley et al., 2006, 2004a). Conditional deletion of Pparg in progenitors of the urothelium of the mouse ureter results in increased proliferation of B cells and a lack of UPK and KRT20 expression in S cells at postnatal stages (Weiss et al., 2013). Conditional deletion of Pparg in bladder urothelial progenitors leads to squamous differentiation of B cells, loss of I cells and a failure of S-cell differentiation in adult mice. The latter phenotype is associated with mitochondrial defects and alterations in lipid metabolism (Liu et al., 2019).

While these studies support a role for PPARG in differentiation of S cells and maintenance of B cells in the developing ureter and bladder, it is unclear whether these functions are linked and what the targets of PPARG transcriptional activity are in the different urothelial cell types. Understanding the precise molecular function of this transcription factor is important because PPARG variants are frequently associated with urothelial cancer in humans (Rochel et al., 2019; Tate et al., 2021). Here, we used conditional gene targeting in combination with transcriptional profiling, data mining and pharmacological rescue experiments to address Pparg function in the mouse ureter. Our results show that PPARG activates the expression of Shh in B cells and I cells, as well as structural S cell genes, and that both gene programs cooperate to maintain urothelial integrity.

To obtain a detailed profile of Pparg expression during the development of the murine ureter, we performed RNA in situ hybridization analysis on transverse ureter sections from wild-type mice at different prenatal and postnatal stages. Pparg expression started at E14.5 in the ureteric epithelium (UE), which was still mono-layered at this stage. At E16.5, when the UE was two-layered, and from E18.5 to postnatal day (P) 40, when the urothelium was three-layered, expression of Pparg increased and was found in all epithelial cells (Fig. 1, first column). Immunofluorescence analysis detected nuclear PPARG protein in approximately half of the cells of the UE at E14.5. At later stages, PPARG protein was found in all urothelial cells. Expression appeared to be enhanced in the luminal cell layer (Fig. 1, second column). Since differences in expression levels can be masked by saturation of the amplification procedure in the immunofluorescence assay, we additionally used immunohistochemical detection. This method confirmed that PPARG expression in the luminal layer is increased compared to the basal and intermediate layer from E16.5 onwards (Fig. 1, third column).

To investigate the role of Pparg in the UE, we used a conditional gene inactivation approach with an allele of Pparg in which exons 2 and 3 are flanked by loxP sites, and a Pax2-cre line that mediates recombination in the nephric duct, the UE and the renal collecting duct system (Bohnenpoll et al., 2017a; He et al., 2003; Trowe et al., 2011).

Using the exon2/3 region and the 3′-UTR of a Pparg cDNA (reference transcript: NM011146.3) as probes in RNA in situ hybridization analysis, we detected expression of a Pparg mRNA lacking the exon2/3 region in the UE of E16.5 and E18.5 Pax2-cre/+;Pparg^fl/fl (Pparg-cKO) embryos. At E18.5, expression of the mutant transcript was increased compared to the control, indicating a compensatory upregulation (Fig. S1A). Importantly, immunohistochemical analysis using a monoclonal antibody raised against the N-terminal PPARG region detected a protein at E16.5 and E18.5 in the control but not in the mutant UE (Fig. S1B). This is consistent with the prediction that cre-mediated deletion of exon2/3 from the floxed Pparg allele results in a truncated protein lacking the N-terminal region, including parts of the DNA-binding domain, and thus a non-functional protein (He et al., 2003).

Pparg-cKO embryos and postnatal animals were found at the expected Mendelian ratio at all stages analyzed (Table S1). The morphological appearance of E18.5 isolated urogenital systems and of P40 kidneys with ureters of mutant mice was normal (Fig. S2). However, histological analysis of Pparg-cKO ureters at P40 revealed a greatly enlarged tubular lumen with an irregular outline (Fig. 2A,B, Fig. S3, Table S2). The urothelium had a highly eosinophilic appearance (Fig. 2A). Expression of CDH1, a marker for basolateral membranes in epithelial cells, was irregular with a partial apical localization. In addition, the mutant urothelium was two-layered and not three-layered as in the control at this stage (Fig. 2C). Cells in the luminal layer of the mutant urothelium were reduced in size compared to control luminal cells (Fig. 2C, Fig. S4, Table S3). Expression of UPK1B and UPK2, which weakly labeled I cells and strongly labeled S cells in control tissue, was absent (UPK1B) or restricted to a few luminal cells (UPK2) in the mutant. ΔNP63, a marker for B cells and I cells, was restricted to the basal layer in the mutant urothelium, which was also positive for the B cell marker KRT5. KRT14 and KRT15, which have been associated with a regenerative response in the bladder (Papafotiou et al., 2016; Tai et al., 2013), were ectopically expressed in B cells of Pparg-cKO ureters (Fig. 2D). KRT6 and KRT10, markers for squamous differentiation (Liu et al., 2019), were detected in neither control nor mutant urothelium (Fig. S5A). The expression of smooth muscle cell (SMC) markers (ACTA2, TAGLN, CKM) appeared irregular and slightly reduced in the mutant ureter, the expression of the lamina propria marker ALDH1A2 was normal (Fig. 2D, Fig. S5B). In conclusion, P40 Pparg-cKO ureters have a two-layered urothelium with undifferentiated luminal cells and differentiated B cells. The latter ectopically express KRT14 and KRT15 (Fig. 2E). The SMC layer also appears to be affected, possibly related to the dilation of the ureteral lumen.

To define the onset and progression of these cellular defects, we compared Pparg-cKO and control ureters at fetal stages (E14.5, E16.5, E18.5) and at P7. Histological staining revealed a thinning of the urothelium in Pparg-cKO mice at P7, whereas the ureteral lumen and the mesenchymal compartment appeared unaffected at all stages analyzed (Fig. 3A,B, Fig. S3, Table S2). CDH1 expression in combination with nuclear counterstaining showed that the urothelium of Pparg-cKO ureters remained two-layered from E18.5 onwards (Fig. 3C). The cells in the luminal cell layer were significantly smaller than in the control at E18.5 and P7 (Fig. 3C, Fig. S4, Table S3). In the control ureter, UPK1B and UPK2 expression was activated in luminal cells at E16.5, strongly labeling S cells and more weakly labeling I cells at later stages. In the Pparg-cKO ureter, only a few luminal cells were positive for these two markers at P7. Expression of ΔNP63 started in the control ureter at E14.5 and continued at later stages with high levels in B cells and at lower levels in I cells. In Pparg-cKO mice, expression of ΔNP63 was normally activated at E14.5, but was restricted to the B cell layer from E16.5. KRT5 expression was not affected in the urothelium of the Pparg-cKO ureter: expression started in a few basal cells at E16.5, labeled about half of the cells in this layer at E18.5, and was found in all B cells at P7 as in the control (Fig. 3C). Expression of squamous epithelial markers (KRT6, KRT10) was not found in either control or mutant UE (Fig. S5C). Expression of both KRT14 and KRT15 was ectopically activated in B cells of the Pparg-cKO ureter at P7. In the control, ALDH1A2 expression was activated in suburothelial mesenchymal cells at E18.5 and homogeneously labeled the lamina propria at P7. In the Pparg-cKO ureter, ALDH1A2 expression was detected in only a few suburothelial mesenchymal cells at P7. Expression of SMC markers started at E16.5 and continued indistinguishably in the mesenchymal wall of control and Pparg-cKO ureters at E18.5 and P7 (Figs 3C, Fig. 5D). Pparg-cKO ureters explanted at P0 and cultured for 6 days showed no defects in the onset, frequency and intensity of contractions, (further) excluding structural and functional alterations in the peristaltic machinery in Pparg-cKO ureters (Fig. S6, Table S4). (Note that in this and some later experiments we used P0 ureters instead of E18.5 ureters due to a change in the German animal protection legislation.)

Given a previous report that loss of Pparg in the bladder urothelium affects mitochondrial structure in the luminal layer, we performed an ultrastructural analysis at E18.5. In Pparg-cKO ureters, the mitochondrial endowment appeared unaffected. However, consistent with impaired S cell differentiation, hinges, known to separate urothelial plaques, were absent from the surface, which was instead covered with a large number of small vesicular bodies. In addition, fusiform vesicles, a compartment for intracellular transport of urothelial plaques, were small and oval-shaped in the luminal cells of Pparg-cKO ureters (Fig. S7).

We conclude that in the Pparg-cKO ureter, the UE fails to activate S cell differentiation at E16.5 and generate I cells for further stratification from E18.5, but ectopically expresses KRT14 and KRT15 at postnatal stages (Fig. 3D, schematic). The delayed activation of ALDH1A2 suggests a non-cell-autonomous effect of epithelial loss of Pparg on lamina propria development.

We next examined whether the cellular changes in the mutant urothelium were preceded and/or accompanied by alterations in apoptosis and/or proliferation. The terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay did not detect any changes in programmed cell death in Pparg-cKO ureters at fetal stages (E14.5, E16.5, E18.5) and at P7 (Fig. 4A). Proliferation was detected by the Ki67 antigen (Mki67). In the control ureter, urothelial proliferation reached approximately 10% at E18.5 and decreased to 1% at P7. In the Pparg-cKO urothelium, proliferation was 15% at E18.5 and 8% at P7. Increased proliferation throughout the mutant urothelium was detected at E18.5. At P7, luminal cells were quiescent in both mutant and control urothelium, but proliferation of (KRT5⁺) basal cells was greatly increased to 12% in the mutant urothelium compared to 1% in the control (Fig. 4B-D, Table S5).

To identify, in an unbiased manner, the molecular changes that may cause the stratification and differentiation defects in Pparg-cKO ureters, we performed microarray-based gene expression profiling at E16.5 and E18.5. Using an intensity threshold of 100 to reduce expression noise and fold changes of at least 1.5 to detect robust expression changes in the two individual arrays, we detected 59 genes with increased expression and 87 genes with decreased expression in mutant ureters at E16.5, and 189 genes with increased expression and 239 genes with decreased expression at E18.5 (Fig. 5A, Tables S6-S9; GSE254236, GSE254237). Assuming that the expression of PPARG-regulated genes is altered at both stages, we overlapped the two arrays and found a common set of 33 genes with increased expression and 56 genes with decreased expression (Fig. 5B, Tables S10 and S11). Functional annotation using the DAVID software tool (Huang et al., 2009) revealed an enrichment of gene ontology (GO) terms related to altered biosynthetic pathways and metabolism in both pools (Tables S12 and S13). Manual inspection of the list of upregulated genes identified major urinary proteins and keratins (Krt4, Krt6a/b). Krt14 expression was upregulated (E16.5: +1.4; E18.5: +1.9) as well as Krt15 (E16.5: +1.4; E18.5: +1.5). Inspection of the list of commonly downregulated genes identified several genes previously assigned to S cells, including Krt20, Upk3b and Upk1a, but also Shh, which encodes a signal restricted to B cells and I cells (Harnden et al., 1995; Yu et al., 2002, 1990) (Fig. 5C). In a recent report, Sanchez and colleagues performed chromatin immunoprecipitation with sequencing (ChIP-Seq) experiments for PPARG in cultured urothelial cells (Sanchez et al., 2021). Using this resource, we found that 17 out of the 56 commonly downregulated genes, including Krt20 and Shh, were assigned peaks for PPARG binding (Fig. 5C, Fig. S8), indicating that our transcriptional profiling enriched direct targets of PPARG transcriptional activation.

To (further) dissect the cell-type specificity of PPARG function, we performed RNA in situ hybridization analysis for all putative direct targets and for many of the additional downregulated genes for which we were able to generate probes. Most of the candidates, including Krt20, Hsd17b2, Ly6d, Sptssb, Fam25c (also known as Fam25a), Adssl1, Rnf144b, Upk1a, Sh3bgrl2, Sdr42e1, Ly6k and Psca, were specifically expressed in S cells in control ureters at E16.5 and/or at E18.5 and were strongly reduced or absent in the mutant. Ivl, Upk3b, Fa2h, Paqr5 and Glb1l2 were strongly expressed in S cells and I cells in control ureters at these stages, and appeared at reduced levels in the mutant ureter. Shh expression in B cells and I cells was lost in the mutant. Pan-urothelial expression of Magea9 and of Dock8 was strongly reduced, while Aqp3 was weakly affected in the mutant (Fig. 5D). We did not detect specific signals for Cck, Ermp1, Jakmip1 and Slc47a1 in control and Pparg-cKO ureters at E16.5 and E18.5 (Fig. S9).

Consistent with the marginal changes in the microarrays, RNA in situ hybridization did not detect expression changes of transcription factor genes that have previously been implicated in urothelial differentiation, such as Elf5, Foxa1, Klf5, Trp63 and Grhl3 in E16.5 and E18.5 Pparg-cKO ureters (Bell et al., 2011; Varley et al., 2009; Weiss et al., 2013; Wu et al., 2015; Yu et al., 2009) (Fig. S10A). GRHL3 expression was found in luminal cells of E16.5 and E18.5 Pparg-cKO ureters, as in the control (Fig. S10B). Fabp4, a direct target gene of PPARG in adipose tissue and the bladder urothelium (Liu et al., 2019; Rival et al., 2004), was not expressed in the UE of E16.5 and E18.5 control and Pparg-cKO embryos (Fig. S11). We conclude that Pparg is required for the expression of a unique set of genes in the developing urothelium of the ureter, namely Shh and a few other genes in the cells of the basal layer, and a large number of S cell-specific genes in the luminal cells.

To investigate whether the loss of Shh expression in Pparg-cKO ureters translates into a reduction of SHH signaling, we analyzed the expression of Ptch1, a bona fide direct target of this pathway, as well as the effector gene Aldh1a2 (Deuper et al., 2022; Ingham and McMahon, 2001; Straube et al., 2025). At E18.5 and P7, the expression of both genes was reduced in peri-urothelial mesenchymal cells, i.e. the lamina propria, in Pparg-cKO ureters, confirming reduction of SHH signaling (Fig. S12).

Given the important role of SHH signaling in mesenchymal and epithelial proliferation and differentiation programs in the embryonic and early fetal ureter, i.e. from E11.5 to E14.5 (Bohnenpoll et al., 2017b; Yu et al., 2002), we examined whether individual inhibition of SHH signaling would recapitulate some of the cellular changes in postnatal Pparg-cKO ureters. To this end, we explanted E18.5 wild-type ureters and cultured them for 6 days in the presence of the SHH signaling inhibitor cyclopamine at a concentration of 10 µM, which was previously determined to effectively inhibit SHH signaling (Bohnenpoll et al., 2017b; Incardona et al., 1998; Meuser et al., 2022; Straube et al., 2025). We observed a loss of Ptch1 and ALDH1A2 expression, recapitulating the situation in Pparg-cKO ureters at E18.5 and (partially) at P7 (Fig. S13A,B). However, neither the expression of cyto-differentiation markers (ΔNP63, KRT5, UPK1B, UPK2) and KRT15 (KRT14 expression was not reliably detected in the culture setting) nor the urothelial cell number or the ratio of luminal cells, intermediate cells and basal cells was affected at the culture endpoint (Fig. S13B-D, Table S14). Urothelial proliferation was slightly decreased (Fig. S13E, Table S14). This indicates that the individual loss of SHH signaling in wild-type ureters does not affect urothelial development and/or maintenance in a major fashion at late fetal and/or early postnatal stages.

We next assessed whether reconstitution of SHH signaling could rescue some of the observed cellular changes in Pparg-cKO ureters. To this end, we cultured E18.5 wild-type and Pparg-cKO ureters for 6 days in the presence of the SHH signaling agonist purmorphamine (Meuser et al., 2022; Wu et al., 2004), and performed histological and molecular analyses (Fig. 6).

Explant cultures of E18.5 Pparg-cKO ureters faithfully recapitulated the in vivo changes. Ptch1 expression was decreased (Fig. 6A); the luminal circumference and urothelial cell number were greatly increased. The urothelium was only two-layered, lacked S cell differentiation and showed KRT15⁺ cells in the basal layer (Fig. 6B,C). Proliferation in the luminal cell layer was slightly increased. Treatment of mutant ureters with 2 µM purmorphamine led to strong Ptch1 expression throughout the ureteric mesenchyme (Fig. 6A). Ureteral lumen, urothelial cell number and B cell proliferation were reduced to control levels. The number of I cells was increased, but the luminal (S) cell differentiation defect and KRT15 expression in the basal cell layer appeared unaltered in the mutant (Fig. 6B-F, Table S15). This indicates that loss of SHH signaling contributes to some of the cellular changes in postnatal Pparg-cKO ureters.

To better identify the molecular programs, which are rescued by purmorphamine treatment in cultures of mutant ureters, we first profiled the transcriptional changes in 8-day cultures of P0 Pparg-cKO ureters by microarray analysis. Using an intensity threshold of 100 and fold changes of at least 1.5 in the two individual arrays, we detected 399 genes with increased expression and 450 genes with decreased expression in Pparg-cKO ureters (Fig. 7A-C, Tables S16 and S17; GSE270635). Functional annotation using DAVID (Huang et al., 2009) revealed for the pool of upregulated genes an enrichment of GO terms related to cell junctions and cell adhesion and to cornification and keratinization. In the pool of downregulated genes, numerous terms related to lipid biosynthesis were enriched (Tables S18 and S19). Notably, both pools were significantly larger than the pools of deregulated genes found for E16.5 and E18.5 Pparg-cKO ureters, suggesting an accumulation of secondary changes related to compensatory mechanisms to maintain tissue integrity (upregulated genes) and to the loss of terminally differentiated S cells with their enormous biosynthetic activity (downregulated genes) over time.

Administration of purmorphamine resulted in profound transcriptional changes in (wild-type and) Pparg-cKO ureters (Fig. 7D-F, Tables S20 and S21; GSE270635). With respect to the 399 genes with upregulated expression in Pparg-cKO ureters, we found that 130 of these genes showed a decrease in expression, i.e. were partially ‘rescued’. Functional annotation of this gene set revealed enrichment of the terms ‘keratinization’ and ‘cornification’ related to decreased expression of keratins such as Krt6b, Krt16, Krt17 (Fig. 7E, Table S22).

From the set of 450 genes with decreased expression in Pparg-cKO ureters, we found that purmorphamine led to increased expression of 91 of them. Functional annotation for these 91 genes revealed ‘apical plasma membrane urothelial plaques’ as a term related to increased expression of Upk genes. Notably, other S cell genes, including Hsd17b2, Krt20, Sptssb, Ttr, Fam25c and Rab27b also regained some expression, suggesting that S cell-specific gene expression is controlled by PPARG both directly and in a non-cell-autonomous manner via SHH signaling (Fig. 7F, Table S23).

Because (small) expression changes are difficult to detect by RNA in situ hybridization on sections of cultured ureters, we validated and quantified the expression (changes) of candidate genes by reverse transcription quantitative polymerase chain reaction (RT-qPCR). This assay confirmed the expression changes of Hhip, a direct target of SHH signaling (Chuang et al., 2003) (Fig. 7G, Table S24A), as well as of Krt14 and Krt15. For other keratins, we could not detect expression (Krt6b) or confirm the expression changes, possibly due to very low overall expression levels (Fig. 7H, Table S24B). In contrast, RT-qPCR confirmed the expression changes of S cell structural genes (Upk genes, Krt20, Hsd17b2) as well as of Grhl3, including the upregulation by purmorphamine treatment in both control and Pparg-cKO ureters (Fig. 7I, Table S24C).

Previous in vitro studies described that Upk gene expression depends on nuclear activation of PPARG and that PPARG is a direct upstream regulator of Upk genes (Salma et al., 2004; Varley et al., 2009, 2004b). Furthermore, in mice with a conditional (Hoxb7-cre mediated) loss of Pparg in the UE, Upk gene and Krt20 expression was found to be significantly reduced at P1 and in adult stages (Weiss et al., 2013).

Our independent conditional approach not only confirms that Pparg is required for the expression of these classical S cell markers, but also shows that the entire S cell differentiation program in the mouse ureter depends on Pparg. In Pparg-cKO ureters, a luminal cell layer was formed at E16.5, but the cells remained small and did not express UPKs at fetal and postnatal stages. Our transcriptional profiling of fetal ureters revealed a large set of genes, expression of which was not activated in the luminal layer of Pparg-cKO ureters. Comparison with a ChIP-Seq dataset of PPARG in a bladder urothelial cancer cell line showed that a large fraction of these genes contain PPARG-binding sites (Sanchez et al., 2021). Together with the strong expression of PPARG in the luminal cell layer from E16.5 onwards, this argues that PPARG drives differentiation of S cells by directly activating the transcription of Upk genes as well as of many other genes that collectively transform the small luminal precursors into large biosynthetically highly active cells.

The expression of some S cell-specific genes, including Sptssb, Upk1b, Upk3b and Fa2h, was reduced but not absent in the Pparg-cKO ureter, suggesting alternative activating inputs in expression of these genes. A strong candidate is the transcription factor GRHL3. In mice with a systemic loss of Grhl3, the cells of the luminal layer of the bladder epithelium (the ureter was not analyzed) were small and did not express UPKs and KRT20 (Yu et al., 2009). Transcriptional profiling of the mutant bladder revealed strongly reduced expression of Upk2 and Snx31 (a gene involved in endosomal recycling), whereas expression of other Upk genes and S cell markers was less affected (Yu et al., 2009). In Pparg-cKO ureters, the expression of Upk2 and Snx31 was only slightly affected at E16.5 and E18.5. Importantly, the expression of Grhl3/GRHL3 was normal in the Pparg-cKO urothelium at these stages. These results suggest that PPARG and GRHL3 are independently required for S cell differentiation in the developing ureter and bladder. Some S cell-specific genes (Upk2, Snx31) may be preferentially activated by GRHL3, some by PPARG (Krt20, Hsd17b2), while others (e.g. Upk3b) may be co-regulated by both transcriptional activators.

In mice with a conditional loss of Pparg in the bladder epithelium, Grhl3 expression has been shown to be reduced in an RNA-sequencing analysis of adult mutant bladders (Liu et al., 2019). As Grhl3 expression has not been validated at mRNA or protein levels in adults, or at fetal stages when secondary changes can be excluded, it remains unclear whether PPARG acts upstream of Grhl3 in the bladder urothelium.

Work in normal human urothelial (NHU) cells has identified the transcription factor genes FOXA1 and IRF1 as direct targets and mediators of PPARG function (Varley et al., 2009). The expression of both transcription factor genes was unchanged in Pparg-cKO ureters, making it unlikely that they act as functional mediators of PPARG in the mouse ureter.

Analysis of mice with a conditional loss of Pparg in the bladder epithelium confirmed a requirement for Pparg in S cell differentiation, and suggested a primary function in the transcriptional control of mitochondrial biogenesis and fatty acid transport in this process (Liu et al., 2019). In our transcriptional profiling of fetal Pparg-cKO ureters, we did not find genes related to mitochondrial functions. Together with our finding that Fabp4, a direct transcriptional target of PPARG in adipose tissue and the bladder urothelium, is not expressed in the UE, this may indicate that PPARG regulates common but also distinct targets in S cell differentiation in the bladder and ureter. However, it should be noted that the profiling of the mutant bladders was performed in adults, and not at the fetal onset of PPARG function as in our study. Therefore, the transcriptional changes in the mutant bladder may not represent primary PPARG targets, but rather the long-term alterations associated with the absence of functional S cells. Circumstantial evidence for such a notion is provided by the ChIP-Seq data set for PPARG in urothelial carcinoma cells, which did not identify PPARG-binding sites in genes related to fatty acid metabolism and mitochondrial function (Sanchez et al., 2021).

Our detailed phenotypic characterization of Pparg-cKO ureters at both fetal and postnatal stages revealed not only a lack of differentiated S cells, but also severe alterations in B cell and I cell development. Stratification of the UE into a three-layered architecture did not occur, resulting in a complete absence of I cells after birth. Cells of the basal layer continued to proliferate and ectopically expressed markers of a potential regenerative response, KRT14 (and KRT15) (Papafotiou et al., 2016; Tai et al., 2013), in postnatal life. These findings extend results from previous conditional Pparg knockout studies that demonstrated B cell proliferation in mutant ureters at P1, and described a lack of I cells and an expansion of KRT14⁺ cells in adult bladders, respectively (Liu et al., 2019; Weiss et al., 2013).

It is known that the complete absence of S cells and I cells after severe injury results in the activation of KRT14 expression in B cells and their subsequent proliferative expansion (Mysorekar et al., 2009; Papafotiou et al., 2016). Similar changes have been found in mice deficient for Upk1b, Upk2 and Upk3a, which fail to form urothelial plaques (Carpenter et al., 2016; Hu et al., 2000; Kong et al., 2004). Thus, the proliferative expansion of B cells in the Pparg-cKO urothelium is most likely secondary to the loss of differentiated, and thus functional, S cells.

Loss of S cells and I cells in an infection model leads to a rapid restoration of urothelial architecture by dedifferentiation of B cells into I cells, and subsequent differentiation of I cells into S cells (Wiessner et al., 2022). In Upk1b-deficient mice, the urothelium becomes hypercellular with multiple layers of B cells and I cells (Carpenter et al., 2016). In contrast, the Pparg-cKO urothelium failed to generate new I cells and remained two-layered into adulthood. Increased B cell proliferation likely led to the lateral expansion of the tissue.

Our experiments provide compelling evidence that this disparate response in the Pparg-cKO urothelium is caused by the loss of Shh expression. Activation of SHH signaling in cultures of perinatal Pparg-cKO ureters rescued the epithelial circumference, reduced urothelial cell number, decreased B cell proliferation and, importantly, increased I cell number, indicating a shift from horizontal expansion to vertical stratification. Support for such a role for SHH signaling comes from studies in urothelial carcinoma cells, in which SHH overexpression after PPARG inhibition resulted in a partial rescue of cell proliferation and migration/invasion (Sanchez et al., 2021).

Our transcriptional profiling provides a first indication of the programs that SHH signaling may affect in this condition. We found that reactivation of SHH signaling leads to a downregulation of some keratins, which are involved in the proliferative response (Krt14, Krt15) and/or squamous differentiation (Krt6, etc., although this needs further validation). It seems plausible that reduced levels of these intermediate filament proteins but also of cell adhesion and polarity regulators could restore the normal physical properties of B cells, and/or lead to reduced proliferation and an altered cell division plane. However, we also found that the expression of many S cell-specific genes (such as Krt20, Upk genes, Fam25c) was increased (independently of PPARG), suggesting that these genes may receive an independent activating input from SHH signaling, possibly via Grhl3 upregulation. These findings complement previous reports that active SHH signaling limits tumor progression in the bladder (Kim et al., 2019; Shin et al., 2011), and suggest that restoration of SHH signaling may be a therapeutic option in urothelial carcinomas with inactivated PPARG.

Because urothelial SHH exclusively acts in the surrounding mesenchyme (Bohnenpoll et al., 2017b; Yu et al., 2002), it is likely that a mesenchymal signal transduces its function back to the urothelium. Based on the loss of ALDH1A2 expression in E18.5 Pparg-cKO ureters and reports that mesenchymal BMP4 mediates SHH signaling in the early ureter (Bohnenpoll et al., 2017b), retinoic acid and BMP4 are interesting candidates for such a relay function.

Complete pharmacological inhibition of SHH signaling in perinatal ureter cultures affected lamina propria formation, as recently reported (Straube et al., 2025), but did not result in urothelial differentiation defects and slightly decreased rather than increased B cell proliferation. This suggests that SHH signaling is not individually required for perinatal urothelial development and homeostasis. In bladder injury and infections as well as in Upk1b-deficient bladder urothelium, Shh expression is strongly upregulated (Carpenter et al., 2016; Kim et al., 2019). This suggests that signals from differentiated S cells normally limit Shh expression in B cells. Such signals may also prevent the dedifferentiation of B cells into I cells in homeostasis. In the absence of such inhibitory signals, i.e. when S cells are absent or non-functional, SHH signaling is enhanced and participates in the regenerative response.

We conclude that PPARG affects urothelial development and integrity in at least two ways. PPARG induces S cell differentiation from E16.5 onward by directly activating a large number of genes in the luminal cell layer. PPARG also directly activates Shh expression in B cells. In development, SHH signaling is possibly required for formation of the intermediate cell layer after E16.5. After injury and infection, enhanced SHH signaling attenuates the proliferative expansion of B cells, supports the formation of I cells, and enhances the differentiation of S cells, i.e. is part of the regenerative response to restore urothelial integrity.

Similar to a previously reported conditional knockout model (Weiss et al., 2013), Pparg-cKO mice developed progressive bilateral ureteral dilatation with parenchymal thinning by P40. The hydroureter, as this condition is known, was not caused by SMC or pacemaker defects, as the expression of SMC markers was normal and peristalsis was unaffected in the perinatal stages. We also found no evidence of physical obstruction along the ureter and the vesicoureteric junction as a possible cause.

In ‘classical’ hydroureter, the accumulation of urine leads to a uniform expansion of the luminal circumference and tissue thinning. In Pparg-cKO ureters, the luminal circumference was uneven and folded, arguing against a fluid-mediated dilatation. Since the expression of numerous keratins and basal cell proliferation was greatly increased postnatally, it is possible that proliferative urothelial expansion is a cause and not a consequence of the luminal dilatation. However, lack of differentiated S cells may contribute to dilatation by disrupting urothelial integrity, widening of the ureteral orifice and causing vesicoureteral reflux, as observed in mice lacking Upk gene expression (Carpenter et al., 2016; Hu et al., 2000; Kong et al., 2004).

For conditional deletion of Pparg in the UE, we combined the transgenic cre driver line Tg(Pax2-cre)1AKis (synonym: Pax2-cre) (Trowe et al., 2011) with a floxed allele of Pparg (Pparg^tm2Rev ; synonym: Pparg^fl). In this allele, loxP sites flank exons 2 and 3 of the Pparg gene (reference transcript: NM011146.3), which results in the generation of an N-terminally truncated non-functional PPARG protein after cre-mediated deletion (He et al., 2003). The mice were maintained on an NMRI outbred background. Pax2-cre/+;Pparg^fl/fl (Pparg-cKO) mice were obtained from matings of Pax2-cre/+;Pparg^f/l+ males and Pparg^fl/fl females; cre-negative littermates served as controls. Embryos for Pparg expression analyses were obtained from matings of NMRI wild-type mice.

Pregnancies were timed as E0.5 by vaginal plugs in the morning after mating. Pregnant females were sacrificed by cervical dislocation. Embryos and urogenital systems were dissected in PBS. Specimens were fixed in 4% paraformaldehyde in PBS, transferred to methanol and stored at −20°C prior to further processing. PCR genotyping was performed on genomic DNA prepared from biopsies (Table S25A).

Mice were housed in light- and temperature-controlled rooms at the central animal facility of the Hannover Medical School. The experiments were performed in accordance with the German Animal Welfare Legislation (§4, TierSchG). They were approved by the local Institutional Animal Care and Research Advisory Committee and authorized by the Lower Saxony State Office for Consumer Protection and Food Safety (reference number 42500/1H).

Ureters for explant cultures were dissected in L-15 Leibovitz medium (F1315, Biochrom) and cultured on 0.4 µm PET membrane ThinCert inserts (657641, Greiner Bio-One) at 37°C in 5% CO₂ in organ culture medium (DMEM/F12) (21331-020, Thermo Fisher Scientific) supplemented with 100 units/ml penicillin, 100 μg/ml streptomycin (15140-122, Thermo Fisher Scientific), 1 mM sodium pyruvate (11360-039, Thermo Fisher Scientific), 1× MEM NEAA (11140-035, Thermo Fisher Scientific) and 1× GlutaMAX (35050-038, Thermo Fisher Scientific) at the air–liquid interface. Medium was changed every other day. For pharmacological manipulation of SHH signaling, we used cyclopamine (S1146, Selleck Chemicals GmbH) dissolved in DMF at a final concentration of 10 µM, and purmorphamine (540220, Merck) dissolved in DMSO at a final concentration of 2 µM. At the end of the culture period, the tissue was prefixed in methanol, 4% paraformaldehyde in PBS, transferred in methanol, and stored at −20°C prior to further processing.

Embryos, urogenital systems and explants were embedded in paraffin wax and sectioned at 5 µm. Hematoxylin/Eosin and Sirius Red staining were performed according to standard procedures. For antigen detection, primary antibody labeling was performed overnight at 4°C after antigen retrieval (5 min at 121°C; H-3300, Vector Laboratories), blocking of endogenous peroxidases with 6% H₂O₂ for 30 min, and incubation in blocking buffer provided from the kits (TNB) for 30 min. Primary antibodies were visualized with either biotinylated Fab fragments or fluorophore-conjugated secondary antibodies (Table S26). The TSA Tetramethylrhodamine Amplification Kit (1:100; NEL701001KT, PerkinElmer) was used for antibody amplification. Cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; 6335.1, Carl Roth) according to the manufacturer's instructions. DAB substrate solution (NEL938001EA, PerkinElmer) was used for immunohistochemical detection of PPARG.

Electron microscopy was performed on Epon-embedded sections as described previously (Rudat et al., 2014).

Cell proliferation rates in explant cultures (n=5 per condition and genotype) were assessed by the detection of incorporated 5-bromo-2'-deoxyuridine (BrdU) on 5 μm paraffin wax sections according to published protocols (Bussen et al., 2004). Explant cultures were grown for 3 h with 3.3 μg/ml BrdU in the culture medium. For each specimen, five sections of the proximal ureter were evaluated. The BrdU labeling index was defined as the number of BrdU-positive nuclei relative to the total number of nuclei as detected by DAPI counterstaining of the ureter. Data were expressed as mean±s.d.

Cell proliferation rates in vivo (n=5 per condition and genotype) were assessed by immunofluorescence detection of Ki67 on 5 μm paraffin wax sections. For each specimen, n=5 sections of the proximal ureter were evaluated. The Ki67 labeling index was defined as the number of Ki67-positive nuclei relative to the total number of nuclei as detected by DAPI counterstaining of the ureter. Data were expressed as mean±s.d.

Tissue apoptosis was assessed by TUNEL assay using the ApopTag Plus Fluorescein In Situ Apoptosis Detection Kit (S7111, Merck KGaA) on 5 μm paraffin wax sections from at least three specimens per genotype.

RNA in situ hybridization analysis on 10-µm-thick paraffin wax sections using digoxigenin-labeled anti-sense probes was performed as described (Moorman et al., 2001). Primers for PCR amplification of new DNA templates for in vitro transcription of RNA probes are listed in Table S25B.

Total RNA was extracted from pools of ureters dissected from E16.5 (n≥9) and E18.5 (n≥11) sex-sorted control and Pparg-cKO embryos using the peqGOLD RNAPure reagent (732-3312, Peqlab). Total RNA was extracted from pools of sex-sorted P0 control and Pparg-cKO ureters cultured for 8 days with or without 2 µM purmorphamine using the RNeasy kit (QIAGEN) according to the manufacturer's recommendations. All RNA pools were sent to the Research Core Unit Transcriptomics of Hannover Medical School where labeled cRNA was synthesized and hybridized to the ‘Whole Mouse Genome Oligo Microarray V2’ (AMADID 026655, Agilent Technologies). Differentially expressed genes for each stage were identified by filtering normalized expression data using Excel (Microsoft Corp.) based on an intensity threshold of 100 and a 1.5-fold change. Microarray data were submitted to Gene Expression Omnibus under accession numbers GSE254236, GSE254237 and GSE270635.

For prediction of direct target genes of PPARG, we used data available in the Gene Expression Omnibus database (GSE166803) for 5637 urothelial carcinoma cells analyzed by chromatin-immunoprecipitation followed by deep sequencing for PPARG and IgG controls to investigate genome-wide PPARG occupancy (Sanchez et al., 2021). Genes upregulated in our microarray were manually compared in the Integrative Genomics Viewer [version 2.3.37(45); Robinson et al., 2011] with ChIP sequencing data (in human, hg38) based on the gene symbol.

RNA extraction and RT-qPCR analysis for gene expression was performed on three independent pools of five ureters each of P0 control and Pparg-cKO ureter explants cultured for 8 days in presence of 0.1% DMSO (control) or 2 µM purmorphamine, as previously described (Bohnenpoll et al., 2017b; Incardona et al., 1998; Meuser et al., 2022; Straube et al., 2025). Primers are listed in Table S25C.

Sections were photographed on a Leica DM5000 microscope using a Leica DFC300FX digital camera. Whole-mounts were photographed on a Leica M420 with a Fujix HC-300Z digital camera. All images were assembled into figures using Adobe Photoshop CS4.

Quantification of specific cell types and proliferating cells was performed on transverse proximal ureter sections, and marker-positive cells were counted within the urothelium. Nuclei were counter-stained with DAPI. Measurements and cell counts were performed using Fiji/ImageJ 2.3.0 (NIH) (Schindelin et al., 2012) and Adobe Photoshop CS4 software.

The circumference of the ureteral lumen was measured on scaled pictures with the Fiji/ImageJ line tool. The luminal cell area was measured on scaled pictures with the Fiji/ImageJ freehand selection tool, based on DAPI and CDH1 immunofluorescence staining (basolateral borders plus autofluorescence of the cell soma).

To determine the onset of ureter peristalsis, the presence of a peristaltic wave along the ureter was judged within a 1-min time interval. For quantification of the peristaltic frequency, the number of peristaltic waves along the ureter was counted within a 1-min time interval. For the assessment of contraction intensity of cultured ureters, 60 s videos of the explants were recorded with five frames per second on a Leica DM6000 microscope and Leica K3M digital camera and analyzed using Fiji/ImageJ 2.3.0 (NIH) (Schindelin et al., 2012). Contraction intensity was analyzed at mid ureter length by calculating the relative width of the ureter during contraction and relaxation. GraphPad Prism7 and Numbers version 6.1 were used to plot the respective graphs.

Cell proliferation rates were assessed by immunofluorescence detection for BrdU (culture) or Ki67 (in vivo). Cells were manually counted using the counting tool in Adobe Photoshop CS4. The labeling index was defined as the number of BrdU/Ki67-positive nuclei relative to the total number of nuclei as detected by DAPI counterstaining of the urothelium (total urothelium), KRT5⁺ cells (basal cells), KRT5⁻P63⁺ cells (intermediate cells), unstained cells (luminal cell nuclei) or KRT15⁺ cells (KRT15⁺ cells). All data were expressed as mean±s.d.

Statistical analyses of cell counts were performed using GraphPad Prism7 and Numbers version 6.1. The normal distribution of the data was tested using the Shapiro–Wilk normality test (normal distribution if P≥0.05); equality of variances was tested using the F-test (equal variances if P≥0.05). To compare two groups with normal distribution and equal variances, an unpaired, two-sided t-test was applied. The two-sided Welch's t-test was used if two groups with normal distribution had unequal variances. The two-sided Mann–Whitney U-test was used to compare two groups with non-parametric distribution. Results were expressed as mean±s.d. All significant differences are indicated (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001).

RT-qPCR data were evaluated using GraphPad Prism 8.4.37. Normality and equality of variances were assessed using the Shapiro–Wilk test and two-way ANOVA followed by Tukey's post-hoc test. Values are presented as arithmetic mean±s.d. P<0.05 was considered statistically significant. For details, see figure legends.

The Research Core Unit Transcriptomics and Genomics of Hannover Medical School validated the quality of RNAs, generated labeled cRNAs, performed hybridizations on microarrays and provided normalized expression data. We thank Georg Hansmann (Hannover Medical School) for the Pparg^fl mouse line, Rolf Kemler (Max-Planck-Institute for Immunobiology and Epigenetics, Freiburg, Germany) for the CDH1 antibody. The antibody CK-JAC(6F11) (CKM) developed by G.E. Morris was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA 52242.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.204324.reviewer-comments.pdf