Permissive ureter specification by TBX18-mediated repression of metanephric gene expression Free

The organs of the upper urinary system, the (metanephric) kidneys and ureters, are structurally and functionally diverse, yet they arise from a common mesodermal rudiment, the epithelial ureteric bud (UB), and its surrounding mesenchyme. In the mouse, the UB emerges at embryonic day (E) 10.5 from the nephric (Wolffian) duct (ND) and grows towards the metanephric mesenchyme (MM), a mesenchymal cell condensation at the posterior end of the intermediate mesoderm. The proximal ‘tip’ of the UB invades the MM and engages in repetitive rounds of branching and elongation to generate the collecting duct system of the kidney. The distal ‘stalk’ that remains outside the MM elongates before stratifying and differentiating into the urothelium – the specialized epithelium of the urinary drainage system. Parallel to this epithelial dichotomy, mesenchymal cells covering either region of the UB take different developmental routes. The MM gives rise to both the filtering units of the kidney, the nephrons, as well as the renal stroma, whereas loosely organized mesenchyme ensheathing the distal stalk differentiates into contractile smooth muscle cells (SMCs) and fibrocytes of the ureteric wall (Bohnenpoll and Kispert, 2014; Little and McMahon, 2012; McMahon, 2017).

A large number of studies have uncovered that early kidney development depends on the establishment of the MM and subsequent reciprocal interactions of renal mesenchymal sublineages with the proximal UB (Costantini and Shakya, 2006; Grobstein, 1955; Saxen, 1987). The MM is established around E10.25 by a network of transcription factors, including PAX2, SALL1, WT1, EYA1, SIX1, SIX2, OSR1 and HOX11 (Davidson, 2008). These factors direct the survival and proliferative expansion of the MM and induce the expression of glial-derived neurotrophic factor (GDNF). GDNF binds to the tyrosine kinase receptor RET on the ND and triggers UB formation, outgrowth and subsequent branching of the proximal tip region (Costantini and Shakya, 2006). Between E10.5 and E11.5, the MM separates into the nephrogenic mesenchyme (NM) and the stromal mesenchyme (SM). The NM is a self-renewing population of nephron progenitors that forms a cap of condensed SIX2⁺/CITED⁺ cells around the UB tips. The NM is surrounded by FOXD1⁺ mesenchymal cells, which give rise to all stromal cell types, mesangial and vascular SMCs of the kidney (Kobayashi et al., 2014, 2008; Mugford et al., 2008). WNT9B signals from the UB tips maintain the proliferation of the NM and induce, together with stroma-derived signals, expression of WNT4, which in turn drives nephron formation from pretubular aggregates (PTAs) (Bagherie-Lachidan et al., 2015; Karner et al., 2011; Kispert et al., 1998).

Similar to the kidney, ureter development relies on the initial specification of the mesenchymal progenitors, which then dictate the fate of the adjacent (distal) UB epithelium (Bohnenpoll and Kispert, 2014; Mills et al., 2017). Expression analyses in the mouse characterized the ureteric mesenchyme (UM) as a molecularly distinct cell population that is established before UB emergence at E10.25 (Bohnenpoll et al., 2013). As a response to epithelial SHH and WNT7B signals, the UM expresses the signaling molecule BMP4 (Bohnenpoll et al., 2017; Trowe et al., 2012; Yu et al., 2002), which controls epithelial and mesenchymal proliferation and differentiation, and, hence, the coordinated development of both tissue compartments of the ureter (Mamo et al., 2017; Wang et al., 2009). BMP4 is sufficient to impose a ureter-like morphology onto the proximal UB region and to partially induce urothelial differentiation (Brenner-Anantharam et al., 2007; Mills et al., 2017). However, loss of Bmp4 does not disturb the formation of the kidney and ureter per se (Mamo et al., 2017; Wang et al., 2009), suggesting that additional mesenchymal factors control UB regionalization and ureter specification.

The T-box transcription factor gene Tbx18 is expressed in undifferentiated UM from E10.5 to E14.5 (Airik et al., 2006; Bohnenpoll et al., 2013). In Tbx18-deficient mice, expression of Bmp4 is lost, UM cells partly mislocalize to the kidneys and testes, and differentiate into fibrocyte-like cells. As a consequence, the testes are tethered to the kidneys, the renal pelvis becomes dramatically enlarged at the expense of the ureter and hydronephrosis develops before birth (Airik et al., 2006; Bohnenpoll et al., 2013). Heterozygous loss of TBX18 is associated with ureteropelvic junction obstruction (UPJO) in humans (Vivante et al., 2015). Although these studies revealed an essential role for TBX18 in early ureter development, the molecular function of this transcription factor has remained enigmatic. Here, we have used genetic misexpression and molecular profiling experiments to address this open question. We provide evidence that TBX18 mediates ureter development by repressing metanephric gene programs.

Tbx18 expression is confined to the undifferentiated UM in the developing urinary system (Airik et al., 2006; Bohnenpoll et al., 2013). To address the significance of this highly localized expression and to explore the potential of this transcriptional regulator to impose a ureteric fate, we used a conditional Cre/loxP-based transgenic approach to misexpress Tbx18 in all surrounding mesenchymal lineages (Fig. 1A,B). To achieve this, we employed a mouse line harboring a bicistronic transgene-cassette containing the mouse Tbx18 open reading frame followed by IRES-GFP integrated into the ubiquitously expressed X-chromosomal Hprt locus (Hprt^Tbx18) (Greulich et al., 2012; Trowe et al., 2012), and a Pax3-cre line that mediates recombination in the mesenchyme of the posterior trunk, including the intermediate mesoderm from E9.5 onwards (Airik et al., 2010; Li et al., 2000). Owing to random X-chromosome inactivation in females, we used only male (Pax3-cre/+;Hprt^Tbx18/y) embryos that expressed the transgene uniformly in all recombined cells and at levels comparable with endogenous Tbx18 expression in the UM (Fig. 1B).

We started our phenotypic analysis at E18.5, i.e. shortly before birth. At this stage, the urogenital system (UGS) of Pax3-cre/+;Hprt^Tbx18/y embryos was grossly abnormal. Kidneys and ureters were absent. The testes had an elongated, irregular shape and were in contact with the adrenal glands. Excessive fibrous tissue covered the UGS (Fig. 1C). As both the ureter and the kidney arise from the UB, we analyzed the kidney and ureter rudiment at E11.5, when the UB has emerged from the ND, invaded the MM and branched once. GFP epifluorescence from a Hoxb7-GFP transgene (Srinivas et al., 1999) visualized the ND and the UB in control embryos. In Pax3-cre/+;Hprt^Tbx18/y;Hoxb7-GFP/+ embryos, the UB was absent (Fig. 1D). RNA in situ hybridization confirmed the absence of the UB marker Pax2, but also of Wt1, Six2, Gdnf and Eya1, which mark the MM in the control at this stage (Fig. 1E-I). Expression of UM markers (Bmp4, Sox9 and Gata2) was not changed in mutant embryos (Fig. 1J-L). To address the possibility that the MM is initially established but subsequently lost due to apoptosis, we analyzed Pax3-cre/+;Hprt^Tbx18/y embryos at E10.5, when transgene expression occurred in the entire trunk mesenchyme (Fig. 1M). Neither histological nor marker (Wt1 and Pax2) analysis detected a MM condensate next to the UB in these embryos (Fig. 1N-P). Importantly, the ND was unaffected, and the surrounding mesenchymal cells expressed the intermediate mesoderm marker Osr1 (So and Danielian, 1999) (Fig. 1P,Q). The TUNEL assay did not detect apoptotic cells in the metanephric region (Fig. 1R). We conclude that ectopic expression of Tbx18 in the posterior trunk mesenchyme impedes MM development, while the specification of the intermediate mesoderm and lower urinary tract development remain undisturbed.

Shortly after its emergence, the MM segregates into two sublineages, the NM and the SM (Kobayashi et al., 2014, 2008). To investigate whether misexpression of Tbx18 differentially affects these lineages, we combined the Hprt^Tbx18 allele with well-established sublineage-specific cre lines. For the NM, we used the Six2-cre line (Kobayashi et al., 2008) (Fig. 2A, Fig. S1A). In E11.5 Six2-cre/+;Hprt^Tbx18/y embryos, Tbx18 was ectopically expressed in the NM at a level similar to that in the UM, proving the suitability of this approach (Fig. 2B, Fig. S1B). At E18.5, the UGS of Six2-cre/+;Hprt^Tbx18/y embryos presented with severe kidney hypoplasia (Fig. 2C, Fig. S1C). Histological staining revealed the presence of the renal papilla with collecting ducts and a regular cortico-medullary subdivision. However, proximal and distal tubules, glomeruli, UB tips and stems, and the NM were strongly reduced (Fig. 2D, Fig. S1D). To validate these changes, we screened the expression of molecular markers indicating regionalization and differentiation within the NM, SM and UB lineages on sagittal kidney sections (Fig. 2E-O, Fig. S1E-O). Markers of the NM (Wt1, Eya1, Six2, Cited1 and Gdnf) showed a variably reduced and patchy expression in the cortex of Six2-cre/+;Hprt^Tbx18/y kidneys. Gdnf was hardly affected; Wt1, Eya1 and Six2 were strongly reduced, whereas Cited1 was almost completely absent (Fig. 2E-I, Fig. S1E-I). In the control, Wt1 was additionally found in renal vesicles and prospective and definitive podocytes. In Six2-cre/+;Hprt^Tbx18/y kidneys, this expression site was almost absent. The PTA marker Wnt4 was irregularly localized and found at very few sites (Fig. 2J, Fig. S1J). Expression of UB tip markers (Ret and Sox9) was similarly affected (Fig. 2K,L; Fig. S1K,L). In contrast, markers for stromal subdomains (Foxd1 and Aldh1a2 for the cortical stroma, Sfrp1 for the capsular and medullary stroma, and Wnt4 for the papillary stroma) appeared unaffected (Fig. 2J,M-O; Fig. S1J,M-O). Expression of Sox9, Bmp4, Gata2 and Sfrp2, which mark the undifferentiated UM, was not expanded into the NM in Six2-cre/+;Hprt^Tbx18/y kidneys (Fig. 2K,P-R; Fig. S1K,P-R). These changes were also found in mutant kidneys at E12.5 (Fig. S2). Hence, misexpression of Tbx18 in the NM interferes with the development of this lineage but does not impose a ureteric fate onto these cells.

We next used the Foxd1^cre line (Kobayashi et al., 2014) to recombine the Hprt^Tbx18 allele in the SM lineage (Fig. 3A,B; Fig. S3A,B). At E18.5, all components of the UGS were present in Foxd1^cre/+;Hprt^Tbx18/y embryos, but the adrenals were reduced in size, the epididymis was tethered to the kidney (cryptorchidism) and the kidneys had a rough surface with a cauliflower-like appearance (Fig. 3C, Fig. S3C). Additional features of renal dysplasia were apparent on histological stainings. The pelvis was reduced. The papilla and papillary collecting ducts were absent. Proximal and distal tubules and glomeruli were not located at the cortico-medullary border, as in the control, but were widely distributed in the medullary region, often exhibiting an irregular shape (Fig. S3D). The kidney capsule was thickened and extended fibrous strands into the cortex and medulla, separating dilated UB tips (Fig. 3D, Fig. S3D).

NM markers exhibited increased expression in enlarged domains surrounding the dilated UB tips in the mutant kidney. The expression of Wnt4 was highly irregular. Expression of UB tip markers appeared reduced (Fig. 3E-L, Fig. S3E-L). Stromal markers were variably affected: Foxd1 appeared slightly reduced. The level of Aldh1a2 expression was normal but the domain of expression was irregular, indicating a blurred cortico-medullary border. Expression of Ntn1, Fibin and Smoc2, which mark a stromal subpopulation in the outermost cortex in the control (England et al., 2020), was lost in the mutant kidney, whereas Sfrp1 extended from the capsular stroma along the ectopic fibrous strands into the cortical and medullary region (Fig. 3M-R, Fig. S3M-R). Markers of the UM were not ectopically expressed in the SM (Fig. 3S-U, Fig. S3S-U). Molecular changes in the SM were already detectable in Foxd1^cre/+;Hprt^Tbx18/y kidneys at E12.5 (Fig. S4). Wt1 expression was absent in the SM; most other markers (Foxd1, Aldh1a2, Ntn1, Fibin and Smoc2) exhibited reduced expression, whereas Sfrp1 expression was expanded (Fig. 4C,K-P; arrows in C indicate missing Wt1 expression). In summary, misexpression of Tbx18 in the SM disturbs patterning and/or differentiation of this cell population. This may secondarily affect branching morphogenesis of the UB tips and nephrogenesis.

TBX18 binds to transcriptional co-repressors and represses transcription of reporter genes upon binding to conserved DNA elements in transactivation assays in vitro (Farin et al., 2007; Lüdtke et al., 2022; Rivera-Reyes et al., 2018). Whether TBX18 represses gene transcription in the UM has not yet been investigated. We addressed this question by analyzing the phenotypic consequences of misexpression in the UM of a variant of TBX18 that confers strong transcriptional activation. For this purpose, we combined a cre knock-in allele of Tbx18 (Tbx18^cre) with a variant of the Hprt^Tbx18 allele (Hprt^Tbx18VP16) in which Tbx18 was replaced by a cDNA encoding a fusion protein of TBX18 with the strong activation domain of the Herpes simplex virus VP16 protein (Greulich et al., 2012). In Tbx18^cre/+;Hprt^Tbx18VP16/y;Rosa26^mTmG/+ embryos, GFP expression from the Rosa26^mTmG reporter allele (Muzumdar et al., 2007) was restricted to the prospective UM, proving the specificity of the approach (Fig. S5A,B). As Tbx18^cre/+;Hprt^Tbx18VP16/y embryos did not survive beyond E11.5, we analyzed female Tbx18^cre/+;Hprt^Tbx18VP16/+ embryos for phenotypical changes at E18.5. The mutant UGS displayed shortened dilated ureters with accompanying hydronephrosis that was reminiscent of the lesions in Tbx18-deficient embryos (Fig. S5C-F). This is compatible with the notion that TBX18 acts as a transcriptional repressor in the UM.

Our misexpression experiments showed that ectopic TBX18 disrupts the development of the MM. Together with the findings that TBX18 acts as a transcriptional repressor in vitro and most likely in vivo, it seemed conceivable that loss of Tbx18 leads to ectopic activation of MM gene programs in the UM. To identify such programs, we profiled global transcriptional changes in the Tbx18-deficient UM. Taking advantage of a GFP knock-in allele of Tbx18 (Tbx18^GFP) (Christoffels et al., 2006), we FACS-isolated pools of ∼100,000 GFP⁺ cells from dissected posterior urogenital ridges of E11.5 Tbx18^GFP/+ and Tbx18^GFP/GFP embryos (Fig. S6A,B), isolated mRNA and subjected it to microarray analysis.

Using an intensity threshold of 100 and a fold change of ≥2.0 in each of the three individual comparisons as filters, we identified 98 genes with decreased expression (Fig. 4A; Table S1, GSE198129). This pool included Tbx18 itself, as well as genes previously shown to depend on Tbx18, such as Sfrp2, Sox9 and Epha7 (Airik et al., 2006; Bohnenpoll et al., 2013), demonstrating the validity of our approach. As TBX18 likely acts as a transcriptional repressor, we focused on the pool of 100 genes with increased expression (Fig. 4A; Table S2). Functional annotation by DAVID webware found an enrichment of terms related to ‘kidney development’ (Table S3). Comparison of the underlying genes with anchor genes for renal sub-populations (Brunskill et al., 2014; Magella et al., 2018) identified well-known players in the NM (Rspo1, Meox1, Six2, Wt1, Itga8 and Etv4) and in PTA formation (Wnt4) (Table S2).

Given the large pool of genes with increased expression in the Tbx18-deficient UM, we wished to identify those that might be directly transcriptionally regulated by TBX18. In the absence of suitable tools for chromatin immunoprecipitation (ChIP)-Seq analysis, we took an alternative approach. We assumed that direct target genes should be strongly activated when the above-mentioned activating variant of TBX18 (TBX18VP16) is misexpressed in the UM. We, therefore, profiled by microarray the transcriptome of GFP⁺ cells obtained from dissected urogenital ridges of E11.5 Tbx18^cre/+;Hprt^Tbx18VP16;Rosa26^mTmG/+ embryos and compared it with that of Tbx18^GFP/+ (control) embryos. Using the same filters as above, we identified 573 genes with decreased and 588 with increased expression (Fig. 4B; Fig. S6A,C; Tables S4,S5). When we compared the lists of genes with increased expression in the two genetic conditions, we found that 53 genes were in common (Fig. 4C,D; Table S6). Functional annotation by DAVID found ‘metanephros development’ as the top enriched term for this gene set (Fig. 4E; Table S7). Comparison with tables of genes with enriched expression in kidney compartments at E14.5 obtained from scRNA-Seq (Magella et al., 2018) revealed that 20 candidates can be annotated to the NM and developing nephrons (Table S6). We validated these findings by RNA in situ hybridization analysis on kidney/ureter sections of E12.5 wild type, Tbx18^GFP/GFP and Tbx18^cre/+;Hprt^Tbx18VP16/y;Rosa26^mTmG/+ embryos. We focused on genes that showed robust expression in the UM (average intensity mutant >1000) in both Tbx18^GFP/GFP and Tbx18^cre/+;Hprt^Tbx18VP16 microarrays, plus selected other candidate genes (Cited1, Eya1, Fzd10, Gdnf and Six2). Expression of Dach1, Eya1, Eya2, Fzd10, Gdnf, Mybpc1, Pla2g7, Rcsd1, Smoc2, Wnt4 and Wt1 was increased in the UM of Tbx18^GFP/GFP embryos – Dach1, Eya1 and Wt1 particularly strongly. Wnt4 showed strongly increased expression in cell clusters surrounding the ureteric epithelium (UE). Dach1, Eya2, Fzd10, Gdnf, Mybpc1, Pla2g7, Rcsd1, Smoc2, Wnt4 and Wt1 were upregulated in the UM of Tbx18^cre/+;HprtT^bx18VP16/y embryos, while Eya1 was unchanged in this context (Fig. 4F). All other candidates appeared unchanged in the UM of both genotypes (Fig. S7). We conclude that key regulators of the MM/NM, including Gdnf, Eya1, Wt1 and Wnt4, are ectopically expressed in the Tbx18-deficient UM.

As GDNF induces UB formation and maintains outgrowth of the collecting duct tree (Costantini and Shakya, 2006), we wondered whether ectopic expression of Gdnf and/or other MM factors affects the cellular and molecular program of the adjacent UE in Tbx18-deficient kidney/ureter rudiments. In fact, we detected upregulation of the gene encoding the GDNF receptor, Ret, in an epithelial outgrowth along the ureter stalk, next to ectopic Gdnf expression (Fig. 5A). To visualize epithelial morphogenesis during early ureter development, we crossed a Hoxb7-GFP transgene (Srinivas et al., 1999) into the Tbx18^GFP background, explanted E11.5 kidney/ureter rudiments and documented for 4 days GFP epifluorescence from the Hoxb7-GFP allele, which marked all ND-derived epithelial structures, including the ureter and collecting ducts. GFP expression from the Tbx18 allele was hardly detected under the chosen settings (Fig. 5B). In control (Tbx18^GFP/+;Hoxb7-GFP/+) cultures, the epithelium of the UB stalk steadily elongated and widened over the culture period to form the UE. The UB tips showed a highly symmetrical branching pattern. In the mutant (Tbx18^GFP/GFP;Hoxb7-GFP/+), an epithelial protrusion was detectable on the caudal side of the proximal UB stalk at culture day 0 (white arrow in Fig. 5B). 24 h later the protrusion stalled and the epithelium developed a kink, leading to a rotation of the kidney to lie parallel to the ND at day 2. An additional UB tip developed in this epithelial region and the kink resolved. The pattern of the collecting duct tree was highly asymmetrical, probably due to a failure of the first caudal UB tip to branch properly (Fig. 5B). We conclude that the proximal region of UB stalk epithelium acquired, at least partly, the character of the collecting duct epithelium and became incorporated into the kidney while the remaining distal region remained short and undifferentiated.

We also monitored UM progenitors during development by introducing the R26^mTmG reporter into the Tbx18-deficient background and recording GFP epifluorescence of E11.5 explants over a 4-day culture period. In the control situation (Tbx18^cre/+;R26^mTmG/+), GFP⁺ cells surrounded the ureter, which progressively elongated. In the mutant situation (Tbx18^cre/GFP;R26^mTmG/+), the GFP⁺ cells localized around the medullary renal collecting ducts, indicating that these cells now support renal branching morphogenesis (Fig. 5C; Fig. S8).

Given that components (Rspo1) as well as some targets (Gdnf, Itga8, Pla2g7 and Wnt4) of WNT9B signaling activity in the NM (Karner et al., 2011) were ectopically expressed in the Tbx18-deficient UM, we wished to evaluate whether the canonical (CTNNB1-dependent) sub-branch of WNT signaling (Barker, 2008), through which WNT9B but also WNT4 act (Park et al., 2007; Karner et al., 2011), is upregulated in the Tbx18-deficient UM at E12.5. For this, we analyzed the expression of Axin2, Ccnd1, Myc and Mycn, which are bona fide direct transcriptional targets of this pathway (Jho et al., 2002; Shtutman et al., 1999; ten Berge et al., 2008). In the control, these genes were weakly expressed in the inner ring of the UM, probably due to WNT7B signaling from the adjacent UE (Trowe et al., 2012). In the Tbx18-deficient UM, this ring of expression was absent but positive cell clusters were found in the UM, and the UE exhibited increased expression. Uncx, a WNT9B-dependent gene in the NM (Karner et al., 2011), was not detected into the mutant UM, whereas C1qtnf12, a WNT9B-dependent gene restricted to PTAs (Karner et al., 2011), was again found in clusters. These complex expression changes argue for a shift of WNT7B signaling from the UM to the UE, and activation of WNT4 signaling in PTAs. A broad activation of WNT9B signaling in the Tbx18-deficient UM seems unlikely.

To address the consequence of widespread and premature activation of canonical WNT signaling in the early UM, we used a conditional (Tbx18^cre-mediated) misexpression approach with a stabilized form of the key component of this branch of WNT signaling: Ctnnb1 [Ctnnb1^(ex3)fl] (Harada et al., 1999). We have previously shown that at E12.5, Tbx18^cre/+;Ctnnb1^(ex3)fl/+ embryos exhibit a strongly condensed and enlarged UM that expresses Axin2, a transcriptional target of the canonical WNT pathway (Jho et al., 2002; Trowe et al., 2012) (Fig. 6A,B). RNA in situ hybridization on sections of the kidney/ureter region of E12.5 Tbx18^cre/+;Ctnnb1^(ex3)fl/+ embryos revealed strong induction of Dach1, Fzd10, Pla2g7 and Wnt4, whereas expression of Wt1, Eya1, Eya2, Rcsd1 and Smoc2 was absent in the ‘UM’ of these mutants (Fig. 6C). Expression of markers and regulators of the undifferentiated UM, i.e. Bmp4, Sox9 and Ptch1 [a direct transcriptional target of SHH signaling (Ingham and McMahon, 2001)], was lost (Fig. 6D). This indicates that precocious and exaggerated activation of canonical WNT signaling in the UM interferes with the specification of this tissue, and leads to the activation of NM genes.

As Tbx18^cre/+;Ctnnb1^(ex3)fl/+ embryos die shortly after E12.5, we in-crossed the R26^mTmG reporter to monitor the further development of the UM in culture. GFP epifluorescence revealed that the UM aggregated around the short ureter throughout a 4-day culture period in explants of E11.5 Tbx18^cre/+;R26^mTmG+/;Ctnnb1^(ex3)fl/+ embryos (Fig. 6E). The adjacent UE (visualized by a Hoxb7-GFP transgene) failed to elongate and thicken but developed several kinks (Fig. 6F). We conclude that enhanced and precocious canonical WNT signaling in the UM recapitulates some of the molecular and cellular changes of ureters lacking Tbx18.

Our analyses showed that Wt1 expression correlated inversely with that of Tbx18 in our different genetic conditions. Given the essential role of Wt1 as a master regulator of the MM and its SM and NM sublineages (Motamedi et al., 2014; Weiss et al., 2020), we wished to investigate its contribution to the phenotypic changes in Tbx18-deficient ureters. To do this, we introduced a floxed allele of Wt1 (Gebeshuber et al., 2013) in the Tbx18-mutant background (Tbx18^cre/GFP) and analyzed the phenotypic consequences (Fig. 7). As we did not recover embryos double mutant for Tbx18 and Wt1 at E18.5, we compared Tbx18^cre/GFP with Tbx18^cre/GFP;Wt1^fl/+ embryos at this stage. The UGS of Tbx18^cre/GFP embryos displayed malformations as previously reported (Airik et al., 2006; Vivante et al., 2015). Ureters were dramatically shortened and connected to the posterior end of the kidney, which was rotated by 90°. Moreover, the ureters were predominantly undilated due to UPJO. The sex ducts and gonads were invariably tethered to the anterior aspect of the kidneys by ectopic fibrous tissue. In Tbx18^cre/GFP;Wt1^fl/+ embryos, hydroureter dominated over UPJO. The ureter length was increased and the majority of the kidneys were positioned correctly. The male gonads were no longer tethered to the kidneys but were descended (Fig. 7A-C, Table S8A).

In explant cultures of E11.5 kidney/ureter rudiments, Tbx18^cre/GFP;Wt1^fl/+ and Tbx18^cre/GFP;Wt1^fl/fl ureters grew significantly longer than those of Tbx18^cre/GFP embryos, and the UM (visualized by GFP epifluorescence from the GFP allele) appeared less dispersed (Fig. 7D,E; Fig. S9; Table S8B). Expression of Wnt4 was still found in the UM of E12.5 Tbx18^cre/GFP;Wt1^fl/fl embryos but appeared uniform rather than clustered. Eya2, Dach1, Pla2g7 and Fzd10 expression exhibited the low levels of the control. Ptch1 expression was reconstituted, while Bmp4 was still absent. We conclude that the derepression of Wt1 contributes to the morphological, histological and molecular changes of the Tbx18-deficient UGS.

We have previously shown that in Tbx18-deficient embryos, mesenchymal cells that surround the UB stalk fail to proliferate and differentiate into SMCs, while UM located more laterally survives, delocalizes to the kidney surface and gains a fibrocytic character (Airik et al., 2006). Although these changes are compatible with individual functions of TBX18 in cohesion, apoptosis, proliferation and/or differentiation of the UM, it seemed conceivable that TBX18 acts upstream of these individual programs by specifying the UM lineage. This notion is supported by the finding that the Tbx18-deficient UM loses expression of Bmp4, which encodes the key signal for proliferation and differentiation in the early ureter (Bohnenpoll et al., 2013; Mamo et al., 2017).

To distinguish these possibilities, we misexpressed Tbx18 in all adjacent mesenchymal primordia. Our phenotypic characterization of Pax3cre/+;Hprt^Tbx18/y embryos unambiguously showed that Tbx18 is insufficient to impose the development of an ectopic ureter or to activate expression of key regulators of the early UM in adjacent mesenchymal tissues, excluding an instructive role for TBX18 in UM specification. Our experiments also exclude functions for TBX18 in the control of apoptosis, proliferation and differentiation, because the early and broad misexpression of Tbx18 did not affect trunk development, the specification of the intermediate mesoderm or the development of the lower urinary tract. However, kidney aplasia in combination with severely reduced expression of key regulators of the MM upon misexpression of Tbx18 in the intermediate mesoderm, kidney hypoplasia upon misexpression of Tbx18 the NM, and kidney dysplasia upon misexpression of Tbx18 in the SM, together show that ectopic Tbx18 specifically disrupts the development of the MM and its sublineages.

Our transcriptional profiling revealed that the Tbx18-deficient UM did not only lose expression of factors that regulate the development of this tissue but also gained expression of genes that are normally confined to the MM. Moreover, in the proximal region, the UB stalk grew tips and became incorporated into the kidney, strongly indicating that it had acquired the character of collecting ducts. This leads, as previously reported, to an enlarged hydronephrotic kidney and to a severely shortened dilated ureter at birth (Airik et al., 2006). Together, these genetic manipulations confirm that Tbx18 is both sufficient and essential for inhibiting metanephric development.

Misexpression experiments in Xenopus have recently shown that Tbx18 can disrupt pronephric development (Naert et al., 2022). This argues that the function of Tbx18 in repressing kidney development pre-dates the evolutionary emergence of a metanephros and indicates that Tbx18 was co-opted in the evolution of the metanephros to subdivide the mesenchymal progenitor populations at the posterior intermediate mesoderm. Restriction of the MM program to the proximal UB tip allowed ureter formation and, hence, a more anterior localization of the kidneys. This role of Tbx18 as a patterning factor resembles that in the somitic mesoderm, where expression of Tbx18 in the anterior somite compartment is required to prevent the expansion of posterior somite half identity (Bussen et al., 2004).

The Tbx18-deficient UM gained the character of MM, whereas misexpression of Tbx18 in the MM and the NM abolished or strongly reduced this cell population, respectively. In combination with earlier reports that TBX18 acts as a transcriptional repressor in vitro (Farin et al., 2007; Lüdtke et al., 2022; Rivera-Reyes et al., 2018) and our finding that misexpression of an activating version of TBX18 (TBX18VP16) in the UM recapitulates the phenotypic changes of the Tbx18-mutant UGS, we assume that TBX18 directly represses the crucial regulators of MM/NM establishment, maintenance and expansion.

Our transcriptional profiling of Tbx18-deficient UM identified several transcription factor genes, namely Hoxb members Osr2, Six2 and Wt1, that may present functional targets of TBX18. In the chick, ectopic expression of Hoxb4 in anterior non-kidney intermediate mesoderm resulted in ectopic kidney gene expression (Preger-Ben Noon et al., 2009), indicating that members of the HOXB subgroup may aid in gaining a MM program. This may copy the activity of members of the HOX11 paralogous group that have been implicated in the expression of Gdnf and other features of the early MM in the mouse (Wellik et al., 2002). Osr2 is not required for mammalian kidney development but, similar to its orthologue Osr1, regulates pronephros development in Xenopus (Tena et al., 2007). SIX2 is not only a marker for the NM but is also a crucial factor in maintaining the nephron progenitor state (Self et al., 2006). It functionally interacts with EYA and DACH family members that showed increased expression in Tbx18-deficient UM (Li et al., 2004; Xu et al., 2014).

Wt1 is expressed in the early MM, the NM and the proximal part of the forming nephron (Armstrong et al., 1993). It is required for the survival of the MM, for epithelial-to-mesenchymal transition and for podocyte development in the NM lineage (Essafi et al., 2011; Guo et al., 2002; Kreidberg et al., 1993). Wt1 expression was strictly inverse to that of Tbx18: Wt1 was strongly activated in UM deficient for Tbx18 or overexpressing TBX18VP16, it was abolished in the MM of Pax3-cre/+;Hprt^Tbx18/y embryos and was strongly reduced in the Six2-cre/+;Hprt^Tbx18/y NM. Moreover, Wt1 expression was lost in the SM of Foxd1^cre/+;Hprt^Tbx18 kidneys. We have previously reported phenotypic changes of kidneys with conditional loss of Wt1 in this lineage that are virtually identical to those observed in Foxd1^cre/+;Hprt^Tbx18 mice (Weiss et al., 2020), arguing that Wt1 is the major target of TBX18 in the SM.

Deleting Wt1 function in the Tbx18-deficient UM ameliorated the phenotypic changes in ureteral and UGS development, indicating that Wt1 is a functional target of TBX18 repressive activity in the UM. As WT1 regulates NM progenitor gene programs together with some of the other upregulated transcription factors (O'Brien et al., 2018), we assume that the concerted upregulation of these factors accounts for the conversion of the Tbx18-deficient UM into MM.

Previous work showed that FGFR1/FGFR2 signaling is essential for MM survival and expansion (Poladia et al., 2006). We detected increased expression of Etv4 (7.6-fold higher), a bona fide target of this signaling pathway (Roehl and Nüsslein-Volhard, 2001) as well as of Fgf20 (12.3-fold higher) in the Tbx18-deficient UM. (Fgf20 is not presented in Table S2 as one of the intensities in the individual arrays did not reach the threshold.) Given that Fgf20 and other Fgf genes are direct and functional targets of WT1 in the MM (Motamedi et al., 2014), we strongly assume that upregulation of these ligand genes and of FGF signaling contributes to the WT1-supported fate switch of the Tbx18-deficient UM.

Of course, it is highly desirable to learn whether TBX18 directly represses Wt1 and other MM genes by binding to a consensus DNA-binding site in the promoter and/or enhancer regions of these genes. A common approach to investigating this is to perform ChIP-Seq experiments in cells in which the transcription factor is expressed. The success of this method relies on a sufficient amount of chromatin and a suitable antibody for immunoprecipitation of the chromatin-bound transcription factor. As UM cells are extremely scarce at E11.5, and neither a suitable antibody for TBX18 nor an allele of TBX18 with a fused peptide tag are currently available, we used TBX18VP16 misexpression as a surrogate strategy. Fusion of the VP16 peptide confers a strong transcriptional activation activity to any DNA-binding protein such that, upon binding to specific DNA sites, nearby genes should be invariably activated. Our transcriptional profiling experiment of E11.5 Tbx18VP16-misexpressing UM detected over 500 genes that were strongly activated. The fact that this number largely exceeds that of genes derepressed in Tbx18-deficient UM may not come as a surprise because only a fraction of genes that harbor specific DNA-binding sites are actually transcriptionally regulated by the corresponding factor. Importantly, more than half of the genes that were derepressed in Tbx18-deficient UM, including Wt1, Wnt4 and other MM factors, were activated by ectopic Tbx18VP16, providing a strong hint that they represent direct targets of TBX18 binding and repressive activity.

Previous work has shown that WNT9B is required for the maintenance of the NM and the induction of Wnt4 and PTA formation (Karner et al., 2011). Intriguingly, Wnt9b expression is not confined to UB tips but also occurs in the UB stalk from E11.5 to E14.5 (Carroll et al., 2005). Our transcriptional profiling experiments revealed activation of Rspo1, which encodes an enhancer of WNT9B signaling (Vidal et al., 2020), as well as of targets of WNT9B signaling (Crym, Itga8, Pla2g7 and Wnt4) (Karner et al., 2011) in the Tbx18-deficient and/or the Tbx18VP16⁺ UM. Moreover, precocious and enhanced canonical WNT signaling led to cellular and molecular changes that mimicked the loss of Tbx18 in this domain: Dach1, Fzd10, Pla2g7 and Wnt4 were ectopically activated, while UM-specific genes (Sox9, Ptch1 and Bmp4) were lost; the UM was highly condensed and the ureter stalk developed kinks and remained short.

Although these findings suggest that canonical WNT9B signaling is a functional target of TBX18 repressive activity in the UM, our additional expression studies do not support this. First, we did not detect a broad activation of genes controlled by WNT9B in the NM, such as Uncx, Six2 and Pla2g7 in the Tbx18-deficient UM. Second, we did not detect a uniform activation of classical targets of canonical WNT signaling, including Axin2, Ccnd1, Myc and Mycn in this region. The normal expression of these genes in the inner UM was lost, while ectopic expression occurred in the UE. Third, we detected clusters of expression of Axin2 in the Tbx18-deficient UM that resembled the pattern of expression of the PTA markers Wnt4 and C1qtnf12. As WNT4 acts via the canonical pathway as an auto-inducer of nephrogenesis (Kispert et al., 1998), we assume that clustered expression of Axin2 and C1qtnf12 reflects WNT4 signaling activity and that forced expression of a stabilized form of CTNNB1 mimics the uniform activation of WNT4 rather than of WNT9B signaling.

Expression of Wnt4 was patchy in Tbx18-deficient UM, but became rather uniform in the UM upon additional deletion of Wt1. Although this excludes an activating function of WT1 for Wnt4 in this tissue, it suggests a role for WT1 in the condensation of Wnt4-expressing cells into PTAs in line with previous reports (Essafi et al., 2011; O'Brien et al., 2018).

Forced and premature activation of canonical WNT signaling in the UM did not only activate a MM gene program, it also abolished SHH signaling and Bmp4 expression in this region. This clearly shows that repression of canonical WNT signaling in the early UM is essential to allow activation of the SHH-BMP4 axis that drives the distinct development of the distal UB stalk and its surrounding mesenchyme (Bohnenpoll et al., 2017). Although we do not understand how WNT9B signaling is prevented in the UM, and how WNT7B signals from the UE are titrated to activate SMC differentiation after E12.5 (Trowe et al., 2012), repression of the PTA-inducer gene Wnt4 by TBX18 is one means to avoid deleterious precocious activation of the CTNNB1-dependent signaling branch.

Together, our findings suggest that TBX18 acts between E10.5 and E11.5 as a permissive factor for the specification of the UM. By repressing regulators of MM/NM development, including transcription factors such as Wt1 and signals such as Wnt4 and Gdnf, TBX18 confines the nephrogenic program to the mesenchyme surrounding the proximal UB. This generates a mesenchymal zone around the distal UB in which SHH and BMP4 signaling drives the proliferation and differentiation of the ureteric tissue compartments (Fig. 7F).

Transgenic mice harboring a knock-in of the GFP gene in the Tbx18 locus (Tbx18^tm2Akis; Tbx18^GFP for short) (Christoffels et al., 2006), mice with an insertion of the cre recombinase gene in the Tbx18 locus (Tbx18^tm4(cre)Akis; Tbx18^cre for short) (Airik et al., 2010), mice for conditional misexpression of Tbx18 (Hprt^{tm3(CAG-Tbx18,-Venus)Akis}; Hprt^Tbx18 for short) (Greulich et al., 2012), and of a gene encoding a fusion protein of TBX18 with VP16 (Hprt^{tm4(CAG-Tbx18*,-Venus)Akis}; Hprt^Tbx18VP16 for short) (Greulich et al., 2012) were all previously generated in our lab. Transgenic mice in which Six2 promoter/enhancer regions drive a EGFPcre fusion protein within a BAC transgene [Tg(Six2-EGFP/cre)1Amc; Six2-cre for short, JAX 009606] (Kobayashi et al., 2008), the Foxd1^{tm1(GFP/cre)Amc} line (Foxd1^cre for short; JAX 012463) with an integration of the cre recombinase gene in the Foxd1 locus (Kobayashi et al., 2014) and the reporter line Gt(ROSA)26Sor^{tm4(ACTB-tdTomato-EGFP)Luo};R26^mTmG for short; JAX 007576) (Muzumdar et al., 2007) were purchased from the Jackson Laboratory. Transgenic mice expressing cre recombinase under the proximal 1.6 kbp Pax3 promoter fragment [Tg(Pax3-cre)1Joe; Pax3-cre for short) (Li et al., 2000) were obtained from Jonathan Epstein (University of Pennsylvania, Philadelphia, USA) and Feng Chen (Washington University, St. Louis, USA). Mice expressing GFP under the control of a Hoxb7 promoter fragment [Tg(Hoxb7-EGFP)33Cos; Hoxb7-GFP for short) (Srinivas et al., 1999) were obtained from Rolf Zeller (University of Basel, Switzerland) after permission from Frank Costantini (Columbia University, New York, USA). Mice with conditional misexpression of an activated form of CTNNB1 (Ctnnb1^tm1Mmt; Ctnnb1^(ex3)fl for short) (Harada et al., 1999) were provided by Makoto Mark Taketo. Mice with conditional deletion of Wt1 (Wt1^tm1.1Ceng; Wt1^fl for short) (Gebeshuber et al., 2013) were obtained from Christoph Englert (Leibniz Institute on Aging, Jena, Germany).

All these mouse lines were maintained on an NMRI outbred background. For misexpression of Tbx18 and Tbx18VP16, we mated males heterozygous for the cre driver line with females homozygous for Hprt^Tbx18 and Hprt^Tbx18VP16 alleles, respectively. In some cases, we combined the cre allele with a Hoxb7-GFP transgene in the male or the reporter R26^mTmG in the females. Cre-negative littermates were used as controls. Kidney/ureter rudiments for FACS-mediated harvest of the UM were obtained from matings of mice heterozygous for the Tbx18^GFP allele. For timed pregnancies, noon on the day of vaginal plug detection was designated as embryonic day (E) 0.5. Organ rudiments and embryos were dissected in PBS, fixed in 4% paraformaldehyde (PFA) in PBS and stored in methanol at −20°C. For genotyping by PCR, genomic DNA prepared from liver or yolk sac biopsies was used.

Mice were housed with ad libitum access to food and water under conditions of regulated temperature (22°C) and humidity (50%) and a 12 h light/dark cycle in the central animal facility (ZTL) of Hannover Medical School. All experiments were carried out in accordance with the German Animal Welfare Legislation and the ARRIVE guidelines, were approved by the local Institutional Animal Care and Research Advisory Committee of the Medizinische Hochschule Hannover, and were permitted by the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (LAVES, Lower Saxony State Office for Consumer Protection and Food Safety; AZ 33.12-42502-04-13/1356, AZ42500/1H).

Kidney and ureter rudiments for explant cultures were dissected from embryos in CO₂-independent L-15 Leibovitz medium (F1315, Biochrom), explanted on 0.4 µm polyester membrane Transwell supports (3450, Corning) and cultured at the air-liquid interface with DMEM/F12 (21331020, Gibco) supplemented with 10% FCS (Biochrom), 1× penicillin/streptomycin (15140122, Gibco), 1× pyruvate (11360070, Gibco) and 1×glutamax (35050038, Gibco) in a humidified incubator with 5% CO₂ at 37°C. The medium was refreshed every second day. Branching patterns were evaluated by GFP epifluorescence.

Embryos, urogenital system and ureter explants were fixed in 4% PFA, embedded in paraffin wax and sectioned at 5 µm. Sections were stained with Hematoxylin and Eosin according to standard procedures (Fischer et al., 2008).

Immunofluorescence staining for GFP was performed on 5 µm paraffin wax-embedded sections using polyclonal rabbit-anti-GFP (1:200, sc-8334, Santa Cruz Biotechnology) as a primary antibody and Alexa-488-conjugated goat-anti-rabbit IgG (1:500; A-11034, Invitrogen) as a secondary antibody.

For antigen retrieval, paraffin wax-embedded sections were deparaffinized, pressure-cooked for 15 min in citrate-based antigen unmasking solution (H3300, Vector Laboratories), washed in TBST (0.05% Tween-20 in TBS) and incubated in TNB blocking buffer (NEL702001KT, Perkin Elmer). Sections were then incubated with primary antibodies at 4°C overnight. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI, 6335.1, Carl Roth).

Apoptosis in tissues was assessed by a TdT-mediated dUTP-biotin nick end labeling (TUNEL) assay using ApopTag Plus Fluorescein In Situ Apoptosis Detection Kit (NC9815837, Millipore) on 5 µm paraffin wax-embedded sections.

We performed whole-mount RNA in situ hybridizations following a standard procedure with digoxigenin-labeled antisense riboprobes (Wilkinson and Nieto, 1993). Stained specimens were transferred in 80% glycerol for documentation. RNA in situ hybridization on 10 µm paraffin wax-embedded sections was carried out essentially as described previously (Moorman et al., 2001).

Microarrays were performed on FACS-isolated pools of UM cells obtained from Tbx18^GFP/+, Tbx18^GFP/GFP and Tbx18^cre/+;Hprt^Tbx18VP16 embryos. Embryos were isolated at E11.5 and the metanephric rudiments were dissected and pooled for either GFP or Venus epifluorescence. Tissue rudiments were homogenized by shaking and additional pipetting in 0.25% Trypsin in EDTA/saline with 5 µl DNAseI (4716728001, Roche). Trypsinization was stopped by adding FCS (5% final concentration) and cells were subjected to FACS (FACSAria Fusion or Ilu, Becton-Dickinson; MoFlo XDP, Beckman-Coulter; 100 µm nozzle). GFP⁺ or Venus⁺ cells were directly collected in RLT lysis buffer; total RNA extraction was performed using the RNeasy Mini Kit (74104, Qiagen). We obtained three independent pools of 100,000 cells each from Tbx18^GFP/+ and Tbx18^GFP/GFP embryos and two individual pools from Tbx18^cre/+;Hprt^Tbx18VP16 embryos. The RNA pools were subsequently processed by the Research Core Unit Transcriptomics of Hannover Medical School. Agilent whole Mouse Genome Oligo v2 (4x44K) Microarrays (G4846A) were used for transcriptome analysis. Normalized expression data were filtered using Microsoft Excel. Functional enrichment analysis for up- and downregulated genes was performed with DAVID 6.8 web software (david.ncifcrf.gov), and terms were selected based on the P-value.

Statistical analysis was performed using a two-tailed Student's t-test. Values are indicated as mean±s.d. P<0.05 was considered significant.

Whole-mount specimens were photographed on a Leica M420 with the Fujix digital camera HC-300Z; sections were photographed on a Leica DM5000B with the Leica digital camera DFC300FX. GFP or Venus epifluorescence in living tissues was documented with the Leica DMI 6000 microscope. Images were acquired with Leica Fire Cam or Leica Application Suite X (LAS X) software and subsequently processed and composed into figures using Adobe Photoshop CS4.

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. The Cell Sorting Core Facility of Hannover Medical School advised on the gating strategy and performed FACS isolation. We thank Jonathan Epstein, Feng Chen, Rolf Zeller, Frank Costantini, Christoph Englert and Marie-Christine Chaboissier for mice, Imke Peters for technical help, and Martina Mühlenhoff for support.