Expansion of the renal capsular stroma, ureteric bud branching defects and cryptorchidism in mice with Wilms tumor 1 gene deletion in the stromal compartment of the developing kidney

Development of the mammalian kidney is orchestrated by reciprocal interactions of stromal and nephrogenic mesenchymal cells with the ureteric bud epithelium. Previous work showed that the transcription factor Wilms tumor 1 (WT1) acts in the nephrogenic lineage to maintain precursor cells, to drive the epithelial transition of aggregating precursors into a renal vesicle and to specify and maintain the podocyte fate. However, WT1 is expressed not only in the nephrogenic lineage but also transiently in stromal progenitors in the renal cortex. Here we report that specific deletion of Wt1 in the stromal lineage using the Foxd1cre driver line results at birth in cryptorchidism and hypoplastic kidneys that harbour fewer and enlarged ureteric bud tips and display an expansion of capsular stroma into the cortical region. In vivo and ex vivo analysis at earlier stages revealed that stromal loss of Wt1 reduces stromal proliferation, and delays and alters branching morphogenesis, resulting in a variant architecture of the collecting duct tree with an increase of single at the expense of bifurcated ureteric bud tips. Molecular analysis identified a transient reduction of Aldh1a2 expression and of retinoic acid signalling activity in stromal progenitors, and of Ret in ureteric bud tips. Administration of retinoic acid partly rescued the branching defects of mutant kidneys in culture. We propose that WT1 maintains retinoic acid signalling in the cortical stroma, which, in turn, assures proper levels and dynamics of Ret expression in the ureteric bud tips, and thus normal ramification of the ureteric tree. © 2020 The Authors. The Journal of Pathology published by John Wiley & Sons, Ltd. on behalf of The Pathological Society of Great Britain and Ireland.


Introduction
In the permanent kidney of mammals, the metanephros, nephrons, and collecting ducts present as a contiguous network of epithelial tubes, yet they develop by distinct tubulogenic processes from distinct progenitor tissues [1,2]. In the mouse, the primordium of the collecting duct system, the ureteric bud (UB), arises at embryonic day (E) 10.5 as an epithelial diverticulum of the nephric duct at the posterior pole of the intermediate mesoderm.
The proximal tip of the UB invades the adjacent metanephric mesenchyme (MM) and engages in numerous rounds of branching and elongation. Nephrons derive from the cap mesenchyme (CM), a SIX2 + subpopulation of the MM that surrounds the UB tips [3]. During each branching step, some cells of the CM aggregate and transit to an epithelial vesicle which develops into the glomerulus and the tubular segments of a nephron [4]. UB and CM development is coupled by reciprocal signalling systems. GDNF secreted from the CM binds to the receptor tyrosine kinase RET in the UB tips and triggers directed cell movements [5,6]. In turn, WNT9B from the UB tips induces the proliferative expansion of the CM and the formation of pretubular aggregates (PTAs) [7]. Development of both CM and UB tips is modulated by signals from a third renal lineage, the stroma [8]. Stromal progenitors are first identified at E11.0 as loosely organised FOXD1-expressing fibrocyte-like cells that ensheath the CM [9,10]. These progenitors are maintained in the cortical region until nephrogenesis ceases. From E12.5, they start to populate the interstitium between the epithelial tubes of the forming nephrons and collecting ducts and differentiate into fibrocytes of the capsular tissue, the cortical and medullary interstitium, and vascular supporting cells [10]. Ablation of FOXD1 + progenitors compromises the stromal endowment of renal epithelial structures, and leads to reduced branching of the UB, expansion of nephron progenitors, abnormalities of the renal capsule, and altered vascular patterns [11]. A number of genetic studies indicate that PTA formation and UB branching are controlled by distinct stromal signals [8].
The Wilms tumor 1 (Wt1) gene encodes a Zn-fingertype transcription factor that is expressed in the CM, in PTAs, and in prospective and definitive podocytes within the nephrogenic lineage [12]. Systemic loss of Wt1 leads to renal aplasia due to apoptosis of the MM, defining a role of WT1 in the maintenance and expansion of the CM [13]. Additional in vitro studies and conditional gene targeting approaches correlated the other expression domains with functions of WT1 in the epithelial transition of the PTA and podocyte development [14][15][16][17][18][19].
Germline mutations of WT1 have been associated with a number of human diseases including Wilms' tumour, WAGR syndrome, Denys-Drash syndrome, Frasier syndrome, and the steroid-resistant nephrotic syndrome [20][21][22][23][24][25][26][27][28]. While most of the phenotypic changes described in these syndromes reflect the aforementioned requirements in the nephrogenic lineage, others including the lack of gonadal descent and mesangial defects may indicate additional functions of WT1 in other renal and/or genital lineages. Here, we provide evidence for an independent requirement of Wt1 in the stromal lineage during kidney development.

Materials and methods
Additional details are provided in supplementary material, Supplementary materials and methods.

Organ culture and GFP epifluorescence analysis
Kidney rudiments were explanted onto a 0.4 μm pore size polyester membrane Transwell (#3450; Corning Inc, Lowell, MA, USA) and cultured at the air-liquid interface in FCS-containing medium [32]. Retinoic acid (#0695; Tocris BioScience, Minneapolis, MN, USA) was added at a concentration of 1 μM and the medium was changed every second day of the culture period.

Histological analysis
Urogenital systems were paraffin-embedded and sectioned at 5 μm. Haematoxylin and eosin staining was performed according to standard procedures.

In situ hybridisation analysis
Non-radioactive in situ hybridisation analysis of gene expression was performed on 10-μm-thick midsagittal kidney sections with digoxigenin-labelled antisense riboprobes [33].

Immunofluorescence for detection of antigens
Immunofluorescence analysis was performed on 5-μmthick paraffin sections as previously described [32].

RT-PCR analysis
Total RNA of pooled E14.5 kidneys was reversetranscribed and analysed (see Supplementary materials and methods).

Image analysis
Sections were photographed using a Leica DM5000 microscope with a Leica DFC300FX digital camera or a Leica DM6000 microscope with a Leica DFC350FX digital camera and afterwards processed in Adobe Photoshop CS4.

WT1 is expressed in stromal progenitors in the developing kidney
Previous work described renal expression of Wt1 in the nephrogenic lineage [12] and mentioned weak expression in the cortical stroma at E15.5 [34]. To more carefully characterise tissue-specific expression of WT1 during kidney development, we performed (co-)immunofluorescence analysis of WT1, the CM marker SIX2, and ALDH1A2, a marker of the cortical stroma, on midsagittal sections of E12.5, E14.5, and E18.5 wild-type kidneys. WT1 was expressed in the CM (overlapping SIX2) at all stages analysed. Expression was detected in proximal regions of the renal vesicles and of comma-and S-shaped bodies from E12.5 onwards, and in mature podocytes of the glomerulus at E14.5 and E18.5, as previously reported [12]. From E11.5 to Wt1 in the renal stroma 291 E14.5, WT1 expression was additionally found in the capsular mesenchyme and overlapping with ALDH1A2 in the cortical stroma ( Figure 1A). At E12.5, some WT1 + cells were detected in the medullary stroma (arrows in Figure 1A).
Deletion of Wt1 in the stromal lineage results in hypoplastic kidneys and cryptorchidism To investigate the specific role of Wt1 in the renal stroma, we used a conditional gene inactivation approach with a floxed allele of Wt1 and a Foxd1 cre line which mediates recombination in stromal progenitors from E11.0 onwards [18,30]. Absence of WT1 protein in the stromal compartment of E12.5 Foxd1 cre/+ ;Wt1 fl/fl (Wt1cKO) kidneys confirmed the validity of this approach ( Figure 1B). Wt1cKO mice died shortly after birth with reasons unknown. At E18.5, all components of the urogenital system were present in the mutant. However, the kidneys were smaller; the gonads and the sex ducts were stretched due to lateral ligamentous attachment to the kidney; and excessive fibrous tissue covered the upper urinary system ( Figure 2A). Histological analysis confirmed a normal cortico-medullary subdivision and the presence of papilla and pelvis in the mutant kidney ( Figure 2B, columns 1, 2). In the cortical region, the UB tips presented mostly as strongly dilated ampullae predominantly connected to long unbranched stems compared with the control, where paired small ampullae connected to short stems ( Figure 2B, columns 3, 4). The CM in the mutant was of normal thickness and followed the enlarged UB tips. In contrast, the capsular mesenchyme consisted of several layers of fibrocyte-like cells and extended between the UB tips and their CM into the cortex ( Figure 2B, columns 3, 4). We next screened expression of molecular markers indicating regionalisation and differentiation within the nephrogenic, stromal, and UB lineages ( Figure 2C and supplementary material, Figure S1 for overviews). In the control, Sfrp1 marked the capsular and medullary stroma; Foxd1 the capsular and cortical stroma; Aldh1a2 the cortical stroma; and Pdgfra the cortical, medullary, and papillary stroma. In Wt1cKO kidneys, Sfrp1 and Foxd1 expression extended from the thickened capsule into the cortical region. All other expression sites of these and other stromal markers were unchanged. The CM markers Eya1, Six2, and Wt1 were expressed at normal levels and surrounded the enlarged UB tips in Wt1cKO kidneys. Expression of Wt1 in renal vesicles and prospective and definitive podocytes in developing and mature glomeruli was unchanged. The PTA marker, Wnt4, was irregularly spatially expressed around the enlarged UB tips including occasional ectopic cortical positions. Expression of Ret and Wnt11 at the widened UB tips was patchy and reduced ( Figure 2C and supplementary material, Figure S1). Markers of nephron segments and of the collecting duct epithelium were unchanged, indicating normal nephrogenesis and UB differentiation in Wt1cKO kidneys (supplementary material, Figure S2).
Hence, loss of stromal Wt1 leads to an expansion of the capsular stroma into the cortical region. Disturbed morphology of UB tips indicates a non-cell autonomous requirement of stromal Wt1 for UB branching.
Phenotypic changes in Wt1cKO kidneys occur shortly after deletion of stromal Wt1 expression To define the onset of these phenotypic changes, we analysed Wt1cKO embryos at E12.5, i.e. shortly after conditional deletion of Wt1 in the stroma. On histological sections, the mutant kidney appeared smaller and was poorly detached from the body wall mesenchyme (supplementary material, Figure S3A, columns 1, 2). Histological sections as well as analysis of UB tip markers (Ret, Wnt11) and CM markers (Wt1, Eya1, and Six2) indicated fewer UB tips that were partly 'encircled' rather than 'capped' by the CM in the mutant (supplementary material, Figure S3A,B). This impression may result from skewed section planes due to the smaller size of the mutant kidney or from an altered architecture of the ureteric tree. Wnt4 + PTAs were present, but their number was reduced. Expression of Sfrp1, Foxd1, Aldh1a2, and Pdgfra was unchanged (supplementary material, Figure S3B).
Analysis of kidneys at E14.5 recapitulated the findings at E12.5. The mutant kidneys appeared smaller and were poorly detached from the body wall mesenchyme; UB tips were reduced in number. At this stage, Sfrp1 expression expanded from the capsular into the cortical region; Aldh1a2 expression in the cortical domain was clearly reduced. Expression of Ret at the UB tips appeared reduced (supplementary material, Figure S4).

Branching morphogenesis is altered in Wt1cKO kidneys
Our histological and molecular analyses suggested that loss of stromal Wt1 leads to a delay in branching morphogenesis, nephrogenesis, and growth during kidney development. To better appreciate these changes, we determined the number of UB tips and PTAs by counting Ret + and Wnt4 + expression domains, respectively, and estimated kidney size by the area in histological sections at the previously analysed stages, E12.5, E14.5, and E18.5 ( Figure 3A). In the control, the number of UB tips rose from 9 to 17 to 50, whereas in Wt1cKO embryos, 5, 10, and 33 UB tips were counted, presenting a significant reduction of 42%, 40%, and 34% at these stages. The number of PTAs was similarly reduced, which resulted in an unchanged ratio of nephrons to UB tips. The reduction in kidney size followed the pattern observed for UB tips and PTAs.

292
A-C Weiss et al 4-day culture period ( Figure 3B). In the mutant, the number of UB tips was significantly reduced to 68% of the control both at day 2 and at day 3 of the culture. After 3 days, the numbers of terminal bifurcations and trifid (a stem with three terminal tips) and bifid (a stem with two terminal tips) branches were reduced, whereas the with the CM marker SIX2 on midsagittal sections of control and Wt1cKO kidneys at E12.5. White arrows in the Wt1cKO images indicate that WT1 expression is lost in the cortical and capsular stroma. In the higher magnification images of WT1 expression, the CM is 'stitched out' in white to distinguish it from cortical stroma. n ≥ 3, all staining. cb, comma-shaped body; cm, cap mesenchyme; g, glomerulus; ra, renal aggregate; rv, renal vesicle; sb, S-shaped body; st, stroma; ubt, ureteric bud tip.
Wt1 in the renal stroma 293  tip branching and stem elongation was shifted towards the latter ( Figure 3C). We also measured the length of the stems generated in each generation of branches in Figure 3 Legend on next page.
Wt1 in the renal stroma 295 this ex vivo culture system. At all the time-points analysed, the parental branch (P1) that arose from the ureter was (significantly) longer in Wt1cKO kidneys compared with controls. All daughter generations (filiae F1, F2, and F3) were shorter in the mutant, a finding that reached significance at day 3 of culture ( Figure 3D). We conclude that stromal Wt1 is required to control the branching behaviour of the UB epithelium.

Altered proliferation and lineage contribution in Wt1cKO kidneys
We next asked whether the observed changes are preceded and/or accompanied by alterations in apoptosis, proliferation, and/or cell fate. The TUNEL assay invariably detected apoptotic bodies in the medullary stroma of both control and Wt1cKO kidneys at E12.5 and E14.5 ( Figure 4A). We next performed co-staining for the Ki-67 antigen with ALDH1A2, SIX2, and CDH1 to mark proliferating cells in the cortical stroma, CM, and UB, respectively ( Figure 4B). Stromal cell division was significantly reduced in the mutant kidney at both analysed stages. The proliferative index of the CM was significantly elevated in the mutant, whereas proliferation was robustly reduced in the UB tips at E12.5.
To investigate whether the stromal fate is altered upon loss of Wt1, we crossed the R26 mTmG reporter line [29] into the mutant background and performed co-staining of the lineage marker GFP with the regionalisation markers ALDH1A2 (cortical stroma), SIX2 (CM), and SOX9/CDH1 (UB tips) at E14.5 and E18.5 ( Figure 4C). At E14.5, the mutant stromal lineage exclusively contributed to the capsular tissue, the cortical interstitium, and the entire medullary stroma as in the control (supplementary material, Figure S5). At E18.5, GFP + cells were ectopically and variably found in UB tips and stalks (~4.5 GFP + cells per UB tip, see white arrowheads in Figure 4C) in Wt1cKO kidneys. Hence, Wt1 is required early to increase proliferation in the cortical stroma; at later stages, it may prevent an epithelial transition and a fate shift of stromal cells.

Altered Spry expression and RA signalling in the cortical stroma of Wt1cKO kidneys
In order to determine the molecular changes that may cause the branching defects in Wt1cKO kidneys, we analysed the expression of genes encoding factors previously implicated in the development and crosstalk of the different renal lineages [8] on sections of E12.5, E13.5, and E14.5 kidneys. We did not find changes in the expression of Axin2target of WNT signalling [35,36]; of Ptch1target of sonic hedgehog (SHH) signalling [37,38]; of the bone morphogenetic protein (BMP) ligand genes, Bmp4 and Bmp7, of the BMP antagonist Bmper; of Id2 and Id4targets of BMP signalling [39,40]; of the fibrocyte growth factor (FGF) ligand genes Fgf7 and Fgf10; or of the targets of FGF signalling, Etv4 and Etv5 [41][42][43] (supplementary material, Figures S6 and S7). Other stromal genes and regulators such as Tcf21, Akap12, Decorin, and Fat4 [8] were not changed at E14.5 either (supplementary material, Figure S8).
In contrast, expression of Spry1 and Spry2, encoding targets and antagonists of FGF/mitogen-activated protein kinase (MAPK) signalling [44], was increased in the cortical stroma at E13.5 and E14.5. Expression of the gene encoding the RA synthesising enzyme, Aldh1a2, as well as of the targets of RA signalling, Ecm1 and Rarb [45,46], was reduced in this domain at these stages ( Figure 5). We concluded that loss of Wt1 affects RA and FGF/MAPK signalling in the cortical stroma at E13.5 and E14.5.

Reduced RA synthesis contributes to branching defects in Wt1cKO kidneys
Since stromal RA acts via Ret to maintain branching morphogenesis of the ureteric tree [46][47][48][49], we wished to further validate alteration of Aldh1a2 and Ret expression in Wt1cKO kidneys. In fact, RT-PCR analysis confirmed the reduction of Aldh1a2 and Ret expression in mutant kidneys at E14.5. Expression of the CM gene Six2 was unchanged and expression of Spry1 and Spry2 was increased in this assay, confirming our in situ hybridisation data ( Figure 6A and supplementary material, Figure S9).  Table S1.

296
A-C Weiss et al To explore whether reduced RA synthesis contributes to the observed branching defects, we explanted Wt1cKO kidneys at E11.5 and treated them with 1 μM RA. After 3 days in culture, we detected a significant increase in UB tips, terminal bifurcations, and stem and bifid branches, and a decrease of trifid and monofid branches in Wt1cKO kidneys treated with RA compared with mutant kidneys treated with solvent. Importantly, monofid branches were reduced to the level of the untreated control ( Figure 6B and supplementary material, Figure S10). The length of the parental branch that arose from the ureter was reduced to the untreated control level in Wt1cKO kidneys treated with RA ( Figure 6C). We conclude that stromal Wt1 maintains branching morphogenesis of the ureteric tree at least partly by maintaining Aldh1a2 expression and RA signalling.  Table S1. (C) Lineage tracing of Foxd1 + stromal progenitors in control and Wt1cKO kidneys at E18.5. Co-immunofluorescence analyses on midsagittal kidney sections of the lineage marker GFP and the marker for cortical stroma ALDH1A2, the CM marker SIX2, the UB tip marker SOX9, and the epithelial marker CDH1 of control (Foxd1 cre/+ ;R26 mTmG/+ ) and mutant (Foxd1 cre/+ ; Wt1 fl/fl ;R26 mTmG/+ ) embryos are shown. Nuclei are counterstained with DAPI. White arrows point to GFP + cells in UB tips. n ≥ 3, each staining. csb, comma-shaped body; cm, cap mesenchyme; cos, cortical stroma; g, glomerulus; ms, medullary stroma; u, ureter; ubt, ureteric bud tip.
Wt1 in the renal stroma 297 Figure 5. RA signalling is transiently reduced in Wt1cKO kidneys. In situ hybridisation analysis of expression of the genes encoding the FGF/MAPK antagonists, Spry1 and Spry2; the genes encoding the RA synthesising enzyme, Aldh1a2; and the targets of RA signalling Ecm1 and Rarb on midsagittal sections of control and Wt1cKO kidneys at E12.5, E13.5, and E14.5. Expression of Spry1 and Spry2 is enhanced in the cortical stroma; expression of RA pathway genes is reduced in Wt1cKO kidneys at E13.5 and E14.5. At least three embryos of each genotype were used for each analysis.

298
A-C Weiss et al

Discussion
Previous work described that WT1 controls multiple cellular programs in the nephrogenic lineage. Here, we revealed an independent requirement of Wt1 in the FOXD1 + stromal lineage to maintain the proliferative expansion of cortical stromal progenitors; to restrict differentiation of capsular stromal cells; and to control branching morphogenesis of the UB tips. The latter function may at least partly be mediated by maintaining Aldh1a2 expression, hence RA signalling.  Table S1.
Wt1 in the renal stroma 299

WT1 acts downstream of FOXD1 to mediate some of the stromal functions
Our immunofluorescence analysis detected expression of WT1 in the cortical stroma at E12.5 and E14.5. Weak expression may persist until E18.5 according to recent scRNA-seq data [50]. Conditional deletion in the stromal lineage left WT1 expression in the nephrogenic lineage unchanged, indicating that both domains are independent of each other. Wt1cKO kidneys exhibited reduced proliferation of the cortical stroma and an expansion of the capsular stroma into the cortical region. Maintenance or expansion of capsular fibrocytes may also account for tethering of the gonads and the sex ducts to the kidneys, which we observed in Wt1cKO embryos. Interestingly, a significant number of patients with germline mutations in WT1 display cryptorchidism or a non-scrotal testis position. This phenotype was initially explained by genital defects including failed masculinisation [51]. More recently, a Rarb-cre line was used to conditionally delete Wt1 in the gubernacula, the caudal genital ligaments important for testicular descent. About 40% of adult males exhibited unilateral, always left-sided cryptorchidism which was explained by a failed differentiation of the gubernacula from mesenchymal precursors [52]. Given the complete penetrance of cryptorchidism in our conditional model and the fact that Rarb is expressed in stromal cells, it is likely that the ligaments important for gonadal descent are derivatives of FOXD1 + stromal cells, the correct differentiation of which depends on Wt1.
In any case, it shows that cryptorchidism is unrelated to a genital function but reflects a novel stromal requirement for Wt1. Besides these stromal differentiation defects, we observed delayed and altered branching morphogenesis of the ureteric tree in Wt1cKO kidneys. The total nephron number was reduced but correlated with the number of UB tips, indicating that the nephron-inducing capacity of the UB tips in the mutant (at least until E18.5) was not affected. Kidney hypoplasia is therefore likely to relate to the delay in branching morphogenesis. Occasional ectopic cortical expression of Wnt4 may be due to altered architecture of the ureteric tree rather than premature differentiation of the CM.
The phenotypic changes observed in Wt1cKO kidneys are much less severe than those observed in mice with loss of Foxd1 or the FOXD1 lineage [9,11], and do not comprise an expansion of SIX2 + CM observed in mice with conditional gene targeting of Pbx1, Tcf21, Sall1, and Fat4 in the stroma [53][54][55][56][57]. This confirms that stromal control of CM and UB tip development are molecularly distinct and shows that WT1 acts downstream of FOXD1 in a subprogram of stromal proliferation and differentiation, and in signal-mediated control of branching morphogenesis of the UB epithelium.

Reduced RA signalling accounts for defects in branching morphogenesis in Wt1cKO kidneys
A number of studies provided compelling evidence that stromal RA acts via Ret to maintain branching morphogenesis. Deletion of Aldh1a2 resulted in smaller kidneys, decreased branching morphogenesis, and reduced expression of Ret. Conversely, isolated UB tips treated with RA upregulated Ret expression [49]. Misexpression of a dominant-negative RARa in the UB lineage also led to reduced Ret expression and branching, as did the combined stromal loss of Rara and Rarb [47,49]. Induced expression of Ret in UB tips rescued UB branching in RA signalling mutants [48]. More recently, evidence was provided that RA-induced extracellular matrix 1 (ECM1) reduces RET in the UB cleft, where bifurcation normally occurs [46].
Our expression analysis showed that expression of the genes encoding the RA synthesising enzyme, Aldh1a2, and the targets of RA signalling, Rarb and Ecm1, was downregulated in the cortical stroma of Wt1cKO kidneys at E13.5 and E14.5. This correlated with reduced expression of Ret at the UB tips at this and later stages, and an irregular patchy expression at the widened UB tips at E18.5. Importantly, we found that administration of RA to explant cultures of Wt1cKO kidneys ameliorated the branching defects, mainly by favouring UB tip bifurcation at the expense of stem elongation. Together, these findings argue that stromal Wt1 is transiently required to maintain or increase Aldh1a2 expression and, hence, RA signalling, which in turn, possibly via ECM1, modulates Ret expression and RET dynamics at the UB tips. At later stages, when Wt1 expression is downregulated, Aldh1a2 expression becomes independent of WT1. It is important to note that the UB branching defects in WT1cKO kidneys are much less severe than those observed in kidneys with complete loss of Aldh1a2 or of stromal and epithelial RA signalling. This is in line with our observation that Aldh1a2 expression and RA signalling, as well as Ret expression, are transiently and weakly reduced in WT1cKO kidneys. Interestingly, dependence of Aldh1a2 expression on WT1 function has been observed in other developmental contexts as well [58][59][60][61][62].
Our analysis detected increased expression of the genes encoding the FGF/MAPK signalling inhibitors, Spry1 and Spry2 [44], in the cortical stroma of Wt1cKO kidneys. Although our in situ hybridisation analysis did not find changes of the FGF signalling targets Etv4 and Etv5 [41], it remains possible that enhanced SPRY expression interferes with FGF receptor activities and leads to transcription-independent changes in the activity of downstream kinases. Possibly, this contributes to reduced proliferation of the cortical stroma in WT1cKO kidneys.
A detailed molecular analysis previously reported that WT1 acts in the CM by maintaining FGF signalling and inhibiting BMP signalling via induction of Bmper [63]. We did not observe changes of these pathways in the mutant kidney, indicating that WT1 regulates distinct sets of target genes in the two tissue contexts.
Our lineage tracing analysis identified that the bulk of FOXD1 + stromal progenitors contributed to capsular tissue, and interstitial fibrocytes of cortex, medulla, and papilla in Wt1cKO kidneys. However, expression of the GFP reporter was also detected in some cells of the 300 A-C Weiss et al UB tips. This may be due to sporadic and ectopic activation of Foxd1 expression, hence cre activity, in this epithelial compartment. Alternatively, some mutant cortical stromal cells may have transitioned to an epithelial state and become incorporated in the UB tip. We cannot distinguish these possibilities experimentally, but it is interesting to note that WT1 is essential for repression of the epithelial phenotype in epicardial cells and during embryonic stem cell differentiation through direct transcriptional regulation of Snai1 and Cdh1, encoding two of the major mediators of EMT [64].

Stromal WT1 and Wilms' tumour
Wilms' tumour, the most frequent renal paediatric cancer, classically exhibits a triphasic histology with epithelial duct-like structures, blastemal elements of an undifferentiated, strongly proliferative character, and a stromal fibrous component that ensheaths the blastema [65]. Given the finding that WT1 is mutated in almost 15% of all presented Wilms' tumour cases [22], it is conceivable that a stromal loss of Wt1 may somehow contribute to tumour initiation or progression. However, a recent study that deleted Wt1 in the stromal lineage with or without Ctnnb1 or Igf2 stabilisation did not detect tumour development in these mice [66]. We did not detect tumours in Wt1cKO embryos either but observed Sfrp1-positive strands of capsular stroma invading the cortex of the mutant kidneys after E14.5. This is reminiscent of the stromal appearance in Wilms' tumours and may suggest a contribution of stromal Wt1 loss to the cytoarchitecture of this childhood tumour.

SUPPLEMENTARY MATERIAL ONLINE
Supplementary materials and methods Figure S1. Wt1cKO kidneys show a normal cortico-medullary subdivision but exhibit subtle changes in the organisation of the cortical region at E18.5 Figure S2. Nephron development and collecting duct differentiation are not affected in WT1cKO kidneys at E18.5 Figure S3. Phenotypic changes in Wt1cKO kidneys are present at E12.5 Figure S4. Phenotypic changes in Wt1cKO kidneys are present at E14.5 Figure S5. The fate of stromal cells is not changed in Wt1cKO kidneys at E14.5 Figure S6. Loss of Wt1 in the cortical stroma does not affect WNT, SHH, or BMP signalling in the developing kidney Figure S7. Loss of Wt1 in the cortical stroma does not affect FGF signalling in the developing kidney Figure S8. Loss of Wt1 in the cortical stroma does not affect expression of genes encoding regulators of stromal development in the E14.5 kidney Figure S9. Expression of Spry1 and Spry2 is increased in Wt1cKO at E14.5 Figure S10. Ectopic RA ameliorates branching defects in Wt1cKO kidneys in culture Table S1. Statistical analyses of all quantitative assays performed in this study Wt1 in the renal stroma 303