Mouse metanephric kidney development begins with the induction of the ureteric bud (UB) from the caudal portion of the Wolffian duct by metanephric mesenchymal signals. While the UB undergoes branching morphogenesis to generate the entire urinary collecting system and the ureter, factors secreted by the UB tips induce surrounding mesenchymal cells to convert into epithelia and form the nephrons, the functional units of the kidney. Epithelial branching morphogenesis and nephrogenesis are therefore tightly orchestrated; defects in either of these processes lead to severe kidney phenotypes ranging from hypoplasia to complete aplasia. However, the underlying regulatory networks have been only partially elucidated. Here, we identify the transcription factor vHNF1 (HNF1β) as a crucial regulator of these early developmental events. Initially involved in timing outgrowth of the UB and subsequent branching, vHNF1 is also required for nephric duct epithelial maintenance, Müllerian duct formation and early nephrogenesis. Mosaic analyses further suggest a cell-autonomous requirement for vHNF1 in the acquisition of a specialized tip domain and branching morphogenesis. vHNF1 exerts these intricate functions at least in part through the direct control of key regulatory molecules involved in different aspects of early kidney development. Notably, vHNF1 acting directly upstream of Wnt9b appears to orchestrate Wnt signaling action in the mesenchymal-epithelial transitions underlying the initiation of nephrogenesis. These results demonstrate that vHNF1 is an essential transcriptional regulator that, in addition to the known later functions in normal duct morphogenesis, plays a crucial role during the earliest stages of urogenital development and provide novel insights into the regulatory circuits controlling events.
The urogenital system derives from the intermediate mesoderm and gives rise to the three major kidneys: the pronephros, mesonephros and metanephros (the definitive kidney) (Saxen and Sariola, 1987). The development of the definitive kidney begins at approximately E10.5 with the induction of the ureteric bud (UB) at the posterior end of the nephric duct (ND) [also known as the Wolffian duct (WD)] by signals from the adjacent metanephric mesenchyme (MM). The UB invades the MM, elongates and branches to form the renal collecting system. As each branch is formed, a subset of mesenchymal cells aggregate and condense near the UB tips to form the pretubular aggregates. These structures will further differentiate into comma- and S-shaped bodies, and subsequently into the Bowman's capsule of the glomerulus and the nephron tubules that will connect with the collecting duct (reviewed by Dressler, 2006; Vainio and Lin, 2002). This process of branching and differentiation is reiterated until nephrogenesis is completed shortly after birth in mice. Disruption of any of these events results in congenital anomalies of the kidneys and the lower urinary tract (CAKUT); these represent 20-30% of all anomalies identified in the prenatal period (Pope et al., 1999; Schedl, 2007). In addition to its crucial role in meso- and meta-nephric kidney formation, the WD gives rise to the male reproductive tract, including the epididymis, vas deferens and seminal vesicles. The WD is also required for elongation and maintenance of the Müllerian duct (MD), the precursor of the oviduct, upper vagina and uterus of the female genital tract.
Recent work has identified several signaling molecules and transcription factors that regulate renal branching morphogenesis and nephron formation (reviewed by Bouchard, 2004; Bridgewater et al., 2008; Dressler, 2006). The transcription factors Pax2, acting redundantly with Pax8, and Lim1 have essential functions in multiple steps of renal epithelial tubular morphogenesis and are all required for ND formation (Bouchard, 2004; Dressler, 2006). The UB outgrowth step is regulated primarily by the crucial regulatory molecule glial cell-line derived neurotrophic factor (Gdnf), which is expressed in the presumptive MM, and its receptor tyrosine kinase, Ret, which is localized in the UB (Costantini and Shakya, 2006). Subsequently, Ret and Gdnf play a role in UB branching through the establishment of a positive regulatory loop with Wnt11, a Wnt protein expressed in the tip domain (Majumdar et al., 2003). An essential role for the Wnt9b signaling pathway has recently been demonstrated for the induction of pretubular mesenchymal aggregates and early nephrogenesis (Carroll et al., 2005). However, the specific interactions of these key regulatory molecules with the genomic template have only been partly elucidated.
The vHnf1 (also known as Hnf1β or Tcf2) gene encodes a POU homeodomain transcription factor expressed in the developing liver, pancreas and the entire urogenital system. Heterozygous mutations or large deletions of this gene cause, in humans, a complex syndrome known as renal cysts and diabetes (RCAD), characterized by multiple abnormalities of the kidney and of the male and female genital tract, as well as by early-onset diabetes, pancreas hypoplasia and liver dysfunctions (Barbacci et al., 2004; Bellanne-Chantelot et al., 2004; Edghill et al., 2006; Haumaitre et al., 2006; Lindner et al., 1999). In the mouse, inactivation of the vHnf1 gene leads to embryonic lethality before gastrulation due to defective extra-embryonic visceral endoderm formation (Barbacci et al., 1999). By rescuing this early lethality through the use of tetraploid embryo complementation, we have shown that vHNF1 is required for normal pancreas morphogenesis and liver specification (Haumaitre et al., 2005; Lokmane et al., 2008). Moreover, vHnf1 tissue-specific inactivation in medullar kidney tubules or transgenic expression of dominant negative mutants causes the formation of medullar renal cysts (Gresh et al., 2004; Hiesberger et al., 2004). In addition, overexpression of human vHNF1 mutations in Xenopus results in morphogenetic alterations of pronephros development (Bohn et al., 2003). These studies, together with the observation that zebrafish hypomorphic vhnf1 mutants exhibit cystic kidneys (Sun and Hopkins, 2001), strongly suggest a conserved role for this transcription factor during vertebrate kidney development. However, the precise role of vHNF1 during the induction of the mammalian kidney and the reproductive system remains essentially unknown. Here, we investigate the role of this factor during early kidney development through the use of tetraploid and diploid embryo complementation (Haumaitre et al., 2005; Lokmane et al., 2008). Our results indicate that vHNF1 is a crucial component of the regulatory circuits involved in WD integrity, UB branching and early nephrogenesis.
MATERIALS AND METHODS
Diploid and tetraploid chimera generation, β-galactosidase staining and kidney explant culture
Tetraploid (4n) chimeric embryos and diploid (2n) mosaics were generated as previously described (Haumaitre et al., 2005; Lokmane et al., 2008). The genotype of 4n embryos was always confirmed by PCR analysis of genomic DNA: only vHnf1–/– embryonic stem (ES)-cell-derived embryos in which no wild-type (WT) allele was detected were analyzed. Control embryos were obtained by parallel matings. Heterozygous knock-in mice expressing the lacZ gene into the vHnf1 locus were used either in organ culture or in expression analysis via X-Gal staining (Barbacci et al., 1999).
Urogenital tracts, microdissected metanephric rudiments or isolated MMs prepared as described (Lin et al., 2003) were cultured on Transwell filters (Millipore, 0.8 μm) in 5% CO2 at 37°C in DMEM/F12 (1:1) (Invitrogen) with 10% fetal calf serum, 50 μg/ml streptomycin/50 U/ml penicillin. When indicated, cultures were performed on a monolayer of cells expressing Wnt9b (L-Wnt9b cells) or Wnt4 (NIH3T3-Wnt4 cells), kindly provided by A. McMahon (Carroll et al., 2005); the medium was additionally supplemented with half of the conditioned media recovered from a four-day-old culture at the confluence of the corresponding cell line.
Immunohistochemistry, in situ hybridization and RT-PCR analysis
Mouse embryos were fixed, embedded in paraffin, sectioned and used for either in situ hybridization or immunostaining analyses as described (Lokmane et al., 2008). Serial sections from control and mutant embryos were analyzed in parallel and treated under identical conditions. At least three independently prepared embryos were examined. We used rabbit anti-HNF1β at 1:50 (H-85, Santa Cruz Biotechnology), rabbit anti-PAX2 at 1:500 (Covance) and rabbit anti-phosphohistone H3 at 1:250 (Upstate Biotechnology) as primary antibodies. We used biotinylated goat anti-rabbit (Vector) as the secondary antibody, followed by streptavidin-AlexaFluor 488 at 1:500 (Molecular Probes) or the Vectastain ABC kit (Vector laboratories). TUNEL staining was performed using the cell death detection kit (Roche). The following cRNA probes were used: vHnf1, Pax2 (G. Dressler), Lim1 (R. R. Behringer), Gdnf, Ret (V. Pachnis), Eya1 (P. Xu), Bmp7, Bmp4 (J. Collignon), Wnt11 (A. Kispert), Fgf8 (G. Martin), Pax8, Wnt4 (S. Vainio). Wnt9b, Wt1, Gata3 and Emx2 probes were generated by PCR.
RNA from microdissected metanephroi was extracted and subjected to semiquantitative RT-PCR as described (Lokmane et al., 2008). Primer sequences used were: Gapdh (For: 5′-TCCAGTATGACTCCACTCAC-3′; Rev: 5′-ACCTTGCCCACAGCCTTG-3′); Pax2 (For: 5′-CAGCCTTTCCACCCAACG-3′; Rev: 5′-GTGGCGGTCATAGGCAGC-3′); Lim1 (For: 5′-CGGGAAGGCAAGCTCTACT-3′; Rev: 5′-AACCAGATCGCTTGGAGAGA-3′); Wnt9b (For: 5′-CGAGGAGATGCGAGAGTGC-3′; Rev: 5′-GGAAGGGTGTCAGGACCTC-3′). PCR products were analyzed on agarose gels and quantified by QuantityOne 4.3.1 software (BioRad).
Cell culture, transfection, electrophoretic mobility shift assays and ChIP
Human embryonic kidney (HEK) 293 cells and human epithelial carcinoma C33 cells were maintained and transiently transfected as described (Barbacci et al., 2004). Gel shift assays were performed using protein extracts overexpressing vHNF1 and the double-stranded oligonucleotides indicated in Table S1 in the supplementary material (Barbacci et al., 2004). The genomic sequences –7946 to –7529, –7946 to –7809 and +13979 to 14601 of Wnt9b, –7310 to –6884 and –5762 to –5572 of Pax2, and –7934 to –7711 of Lim1 were PCR amplified and subcloned upstream to the thymidine kinase (TK) promoter (–51 to +10) in the ΔTKCAT6 vector. Sites were mutated using the QuikChange site-directed mutagenesis kit using the primers listed in Table S2 in the supplementary material.
Chromatin immunoprecipitation (ChIP) assays were performed using isolated E14.5 metanephros essentially as described (Wiebe et al., 2007). Chromatin was sheared to a range of 0.3 to 0.7 kb, and assays were performed using the ChIP assay kit from Upstate (ref 17-295). The equivalent of 50 μg of chromatin was used in each ChIP experiment and immunoprecipitated with rabbit anti-HNF1β (H85, Santa Cruz), anti-acetylated histone H4 (Upstate) or rabbit pre-immune serum. Purified immunoprecipitated DNA was resuspended into 30 μl of distilled water. The equivalent of 10% of input chromatin was DNA purified in parallel and resuspended in an equal volume. 3 μl were subjected to semiquantitative PCR. Each assay was repeated three to five times using different pools of E14 metanephros chromatin. Pair primers are provided in Table S3 in the supplementary material.
vHnf1 is expressed in the ND epithelium and its derivatives
We performed a detailed analysis of the expression pattern of vHnf1 at both the transcript and protein levels from the first steps of kidney development. Consistent with previous X-gal staining of vHnf1 heterozygous embryos (Barbacci et al., 1999), at E9.5, vHnf1 transcripts were detected in the WD epithelium and cranial mesonephric tubules (Fig. 1A). This expression pattern is maintained at later stages, with high expression in the entire ND and the emerging UB (Fig. 1B). At E11.5, when the UB has undergone the first branching event to generate the T-stage, vHnf1 was expressed homogeneously in the epithelium of the UB and the WD (Fig. 1C). In situ hybridization and immunostaining analysis revealed that vHnf1 transcripts (not shown) and protein were present in the epithelium of the UB branches and the pretubular aggregate derivatives, the comma- and S-shaped bodies (Fig. 1D,E). At E17.5, the expression pattern was similar to that of the adult (Fig. 1E,F), with expression in the entire collecting system and all segments of the nephron. Expression was absent in the glomeruli. In addition to the WD, vHnf1 was expressed in the MD and genital tract derivative tissues (not shown, see below). In adults, vHnf1 is expressed in the epididymis, vas deferens, seminal vesicle, prostate, uterus and oviduct (Haumaitre et al., 2006; Reber and Cereghini, 2001). Thus, the expression pattern of vHnf1 is consistent with the function of this factor during the earliest stages of UB branching and nephrogenesis.
Lack of vHnf1 leads to defective urogenital development
To directly address the role of vHNF1 during early kidney development, we generated vHnf1–/– ES-cell-derived embryos (further denoted as vHnf1–/–) by tetraploid aggregation (Lokmane et al., 2008). We initially compared X-Gal-stained urogenital tracts from mutant and heterozygous embryos microdissected at different stages (Fig. 2A-H). At E10.5, the WD duct was apparently normal and reached the caudal end of mutant embryos (not shown) (Haumaitre et al., 2005). At E11.5, when the UB reached the T-stage, the UB was not detectable in mutant embryos (Fig. 2A,B). Only a rudimentary UB was observed at around E12.5 (Fig. 2C,D). Furthermore, subsequent branching after the T-stage was disrupted, and the UB had a Y shape rather than a T shape, indicating abnormal branching (Fig. 2F, arrowhead). Histological analysis of E13.5 vHnf1–/– embryos confirmed severe kidney hypoplasia (approximately 70% smaller than WT), the defective UB branching and a paucity of MM condensations, whereas the adrenal glands were normal (Fig. 2I,J). We also noted premature degeneration of the WD in different regions and the complete absence of cranial mesonephric tubules (Fig. 2C,D). In the mutants, the MD was not observed at any stage, whereas it was clearly observed in heterozygous embryos (Fig. 2E-H).
Thus, vHNF1 is required for different aspects of early urogenital development, including the maintenance of WD integrity, formation of mesonephric tubules and MDs, as well as timed emergence and subsequent branching of the UB.
ND and UB growth defects in vHnf1 mutants
We subsequently examined a number of regulatory molecules expressed by either the UB or the MM cells, and implicated in UB branching morphogenesis and ND formation (Fig. 3A-C).
The initial step of kidney development is the outgrowth of the UB from the WD in the adjacent MM. Gdnf, which is expressed in the MM, and its receptor, Ret, which is restricted to the WD epithelium, are both required for UB formation and the branching process (Costantini and Shakya, 2006; Shakya et al., 2005). At E11.5, Ret and Gdnf were apparently normally induced in mutant UB rudiments (Fig. 3B), despite the important delay in their formation. However, analysis of tissue sections revealed that they were both expressed at relatively lower levels in vHnf1–/– embryos compared to expression in the WT (Fig. 3B). At E12.5, the expression of Ret and Gdnf was severely reduced in mutant metanephros (Fig. 3C). Interestingly, Wnt11, a downstream target of GDNF/Ret signaling (Majumdar et al., 2003) that is normally expressed by the tips of the branched UB and in the ND, was specifically absent in the vHnf1–/– UB tips, but remained unaffected in the mutant ND (Fig. 3C; see Fig. S1 in the supplementary material).
Pax2, expressed in the ND, the UB and the induced MM, plays a crucial role during early kidney development: in Pax2 null embryos, the ND initially forms but degenerates prematurely and therefore the kidneys and genital tracts are never formed (Dressler, 2006; Torres et al., 1995). At E11.5, Pax2 was expressed in the mutant MM and a reduced number of caudal mesonephric tubules, but was specifically downregulated in the ND (Fig. 3B). Pax2 downregulation in the ND (particularly in its caudal portion) was stronger at E11.5 (Fig. 3B) than at E10 (Fig. 3A). In addition, the discontinuous expression of Pax2 reflected probably premature WD degeneration. Analysis of serial sections confirmed strong downregulation of Pax2 transcripts in the mutant WD and UB; the Pax2 MM expression domain was also reduced compared to that in WT embryos (Fig. 3B). By contrast, the weak expression of Pax8 in the ND appeared unchanged in the mutants (not shown). Interestingly, the expression of Lim1, a transcription factor involved in ND elongation (Pedersen et al., 2005) and normally expressed in the ND, mesonephric tubules and UB, was barely detected in the mutant ND epithelium by E10.5 (Fig. 3A,B) and was lost at later stages (Fig. 3C). However, the strong downregulation of Pax2 and Lim1 was not due to a generalized differentiation defect of the mutant ND, as indicated by the analysis of additional epithelial markers. The expression of Emx2, a transcription factor required for normal UB formation and growth that displays a strikingly similar phenotype to that of the vHnf1–/– embryo (Miyamoto et al., 1997), was, however, only modestly reduced (Fig. 3C). Likewise, the expression of Gata3, another transcription factor expressed in the ND and UB epithelium and required for normal ND morphogenesis (Grote et al., 2008; Lim et al., 2000), was not altered in mutant embryos (Fig. 3C).
At E12.5, the expression of several essential mesenchymal genes, including Eya1, Wt1 (Donovan et al., 1999; Xu et al., 1999) and Bmp7 (Dudley et al., 1995), was unchanged in the mutant MM, although their expression domains were reduced compared that in the WT (Fig. 3C). By contrast, the expression of Bmp4, normally restricted to a subpopulation of mesenchymal cells surrounding the ND and UB branches (Miyazaki et al., 2000), was strongly reduced in the mutant mesenchyme (Fig. 3C).
These data confirmed the requirement for vHNF1 in multiple steps of early kidney development. First, vHNF1 is involved in ND differentiation, very probably by controlling the expression of both Lim1 and Pax2 in this structure. Although vHNF1 is not required for initial outgrowth of the UB or initial condensation of the MM, it appears essential for UB branching from the T-stage, probably through the control of Ret and Gdnf signaling (see also below, Fig. 6).
We next examined whether decreased proliferation or increased apoptosis affected the UB and the surrounding mesenchyme. Remarkably, using the mitosis marker phosphorylated histone H3, we found comparable numbers of proliferating cells in the entire UB epithelium of E12.5 mutant and control embryos (see Fig. S2A,B in the supplementary material) and only a moderate reduction in the percentage of proliferating cells in the mutant mesenchymal population (see Fig. S2 in the supplementary material). By contrast, TUNEL experiments performed at the same stage indicated a large increase in the number of apoptotic cells in both the mutant UB epithelium and mesenchymal cells (see Fig. S2 in the supplementary material). Thus, the precise balance between proliferation and apoptosis that characterizes normal kidney morphogenesis appears disturbed in mutant kidneys, with a global increase in apoptosis. As vHnf1 expression is initially restricted to the UB epithelium, these findings suggest that vHNF1 plays a role in sustaining the survival of epithelial UB cells, whereas the increased apoptosis of mesenchymal cells might reflect defective signaling from the UB to these cells.
vHNF1 is required for early nephrogenesis
Early nephrogenesis begins with the formation of mesenchymal pretubular aggregates close to the ureteric branches. Fgf8, Pax8, Wnt4 and Lim1 are all expressed in the pretubular aggregates and their derivatives. Loss of either Fgf8 or Wnt4 signaling results in complete failure of renal vesicle formation (Grieshammer et al., 2005; Perantoni et al., 2005; Stark et al., 1994). By contrast, Lim1 is not required for the initial formation of renal vesicles, but is necessary for their proper patterning (Kobayashi et al., 2005). In E12.5 vHnf1-deficient embryos, the expression of Fgf8, Pax8 and Wnt4 was strongly reduced (Fig. 4A). Furthermore, mesenchymal expression of Lim1 was undetectable (Fig. 3C). Thus, in mutant embryos, the expression of the characteristic markers of the early nephron structures is severely impaired. Remarkably, at E12.5, the expression of Wnt9b, a factor that acts as a paracrine signal to induce nephrogenesis via Wnt4 (Carroll et al., 2005), was completely lost in the vHnf1–/– embryos (Fig. 4A). Furthermore, whole-mount in situ hybridization at earlier stages showed that, in our mutants, Wnt9b was not detected in either the ND or the UB (Fig. 4B), indicating that vHNF1 activity was required for Wnt9b induction. Thus, lack of vHnf1 leads to the absence of early nephrogenesis, very probably due to disruption of the regulatory pathway involved in this process, which is mediated by Wnt9b and its targets, Fgf8, Pax8 and Wnt4.
Consistent with the results shown in Figs 3 and 4, semiquantitative RT-PCR analysis from control and vHnf1–/– microdissected metanephroi confirmed the strong reduction of Wnt9b in E12.5 mutant metanephroi, as well as of Pax2 and Lim transcript levels, compared to either E11.5 or E12 control metanephroi (Fig. 4C).
To examine further the potential involvement of vHNF1 in nephrogenesis, we performed in vitro cultures of mutant urogenital tracts using vHnf1+/– urogenital tracts as controls (Fig. 5A). Given the considerable morphogenesis observed upon in vitro culture of entire urogenital tracts and the fact that vHnf1–/– embryos die around E14.5, these experiments enabled us to also examine the potential rescue of UB branching and nephrogenesis defects upon further culture under different conditions. In particular, because previous studies have shown that cells expressing Wnt ligands, including Wnt9b and its direct target Wnt4, can induce isolated MM to undergo tubulogenesis (Carroll et al., 2005; Kispert et al., 1998), we initially tested these ligands. As shown in Fig. 5B, culture of vHnf1–/– urogenital tracts with either Wnt9b or Wnt4 had no effect on either the UB branching defect or the defective formation of mesonephric tubules. However, it did result in the formation of several mesenchymal aggregates, as suggested by whole-mount X-gal (Fig. 5B) and E-cadherin stainings (not shown). Additional studies involving cultures, in the presence of Wnt4, of either mutant metanephric rudiments (Fig. 5C) or isolated mutant MM, derived from epiblast vHnf1-specific inactivation using the Sox2-cre driver line (our unpublished data) (Hayashi et al., 2002), confirmed the formation of pretubular aggregates positive for Pax8 (Fig. 5C and data not shown). In these experiments, we used cells expressing Wnt4, because these cells induce pretubular aggregates both more efficiently and reproducibly than Wnt9b-expressing cells (not shown).
These results show that mutant MM appears to be competent to condense and express early nephron markers, further suggesting that defective nephrogenesis in mutant embryos reflects impaired Wnt9b signaling from the UB epithelium.
vHnf1 acts cell autonomously in UB branching
The observation that, in vHnf1–/– embryos, Wnt11, normally confined to the UB tips, was not detected (Fig. 3C and see Fig. S1 in the supplementary material), together with the observed downregulation of Ret in the tip domain by E12.5 (Fig. 3C), suggested defective formation and/or maintenance of a specialized tip domain. A number of markers, including the cell adhesion proteins E-cadherin and β-catenin and the tight junction protein ZO-1, were, however, correctly localized in the mutant epithelium (Fig. 6A and data not shown). Similarly, Sox9, a transcription factor expressed by E11.5 to E12.5 throughout the entire UB branches and subsequently enriched in the UB tips, was also correctly induced in the mutant UB (Fig. 6A). By contrast, the expression of calbindin-D28K, an intracellular protein with a high affinity for calcium and expressed at this stage by UB cells, was strongly reduced (Fig. 6A). These data suggest that defective UB branching is apparently not caused by abnormal cell polarity of the mutant epithelium. However, the strong reduction in calbindin-D28K expression might imply that the UB mutant epithelium is not fully differentiated.
To obtain further information on vHNF1 function during UB branching and ND formation, we performed diploid mosaic chimeras. WT diploid embryos were aggregated with vHnf1-deficient ES cells; in such chimeras, the embryonic tissues are derived from a variable mix of WT and mutant cells (Fig. 6B), thus enabling the definition of a cell-autonomous function for vHNF1. Diploid chimeras using ES vHnf1+/– were normal and exhibited an apparently random distribution of WT and vHnf1+/– (not shown). Mosaic urogenital tracts derived from embryos with variable contributions of mutant cells were initially dissected at E13.5 (n=27) and E14.5 (n=9), and X-gal stained. A representative set of E14.5 urogenital tracts exhibiting an increasing number of mutant cells (β-gal positives) is shown in Fig. 6B. Chimeric embryos were indistinguishable from WT embryos when the percentage of mutant cells ranged between 40% and 60%. However, when this percentage was higher than 70-80%, chimeric metanephroi were hypoplastic and exhibited reduced and abnormal branching, and even, in one case, a duplex kidney (Fig. 6B). In these mosaics, mutant cells were able to contribute extensively to the WD, to the primary UB and to early nephron structures. There was, however, a lower proportion of mutant cells in the UB branches, ureter and mesonephric tubules compared with early nephron structures or the WD. Note also that mutant cells remained in clusters rather than being intermingled.
To obtain a better definition of the fate of mutant cells, we then focused our analysis on E12.5, because at this stage the individual UB tips can easily be examined and the branching pattern is well defined (Majumdar et al., 2003). The fate of WT versus mutant cells was next examined in different sections using an anti-vHNF1 antibody; adjacent sections were analyzed for the expression of Pax2. In E12.5 WT embryos, vHNF1 is normally present in the epithelium of UB branches, whereas PAX2 is expressed in the MM as well as in the epithelium of the UB. A fraction of E12.5 2n↔ES vHnf1–/– chimeras exhibiting a significantly higher contribution of mutant cells relative to WT displayed a partial phenotype: UB branching was significantly reduced when compared to WT (i.e. 3 tips instead of the 6-8 normally observed in WT embryos; Fig. 6C). Remarkably, in these mosaics, vHNF1-positive cells, and therefore WT, co-expressed with PAX2 and were strictly localized to the tip domain of UB branches (Fig. 6C, black and white arrows). Moreover, mutant cells (vHNF1 and PAX2 negatives) were able to contribute to the epithelium of the stalks (Fig. 6C). Interestingly, ES cells lacking vHNF1 behave like Ret–/– cells (Shakya et al., 2005), both being excluded from the UB tips. This further suggests that vHNF1 might be required for the branching process through its action at the tip domain. These studies suggest a cell-autonomous function for vHNF1 in UB branching morphogenesis and further confirm that its activity is required for Pax2 expression in the UB.
vHNF1 recruitment in vivo to the regulatory sequences of key factors involved in urogenital development
To obtain a more comprehensive picture of the putative vHNF1-mediated transcriptional programs, we searched for conserved core vHNF1-binding sites within the non-coding sequences of the regulatory genes whose expression was severely affected by vHnf1 deficiency. We identified several highly conserved sites, in either the 5′ or intronic sequences, for a subset of these genes, including Wnt9b, Pax2, Lim, Pax8 and Wnt4. Among these potential targets, Wnt9b, Pax2 and Lim1 attracted our attention because their defective regulation appeared to contribute to the vHnf1 mutant phenotype. Interestingly, we identified a cluster of four partially conserved binding sites in the upstream sequence of Wnt9b, and a single site in the Wnt9b first intron within a region that is highly conserved among mouse, rat and human orthologous sequences (Fig. 7A; see Fig. S3A in the supplementary material). We also found several vHNF1-binding sites within conserved regions of the already characterized Pax2 regulatory sequences (Kuschert et al., 2001) (Fig. 7A; see Fig. S3B in the supplementary material) and within a site in the 5′ upstream sequence of Lim (Fig. 7A; see Fig. S3C in the supplementary material). The ability of vHNF1 to bind to all of these sites was confirmed in vitro by electrophoretic gel mobility assays (see Fig. S3D in the supplementary material; data not shown). Interestingly, the identified vHNF1-binding sites are clustered with the consensus binding sites of other key transcription factors, notably Pax2/8, Sox-9, HNF4 and FoxA2.
To examine whether vHNF1 was recruited to these sites in vivo, we performed ChIP assays on freshly isolated E14.5 metanephros. Soluble cross-linked chromatin was immunoprecipitated with antibodies against vHNF1. The presence of sequences corresponding to the identified binding sites in the immunoprecipitated material was then assessed by semiquantitative PCR. As shown in Fig. 7B, for the samples immunoprecipitated with the anti-vHNF1 antibody, we were able to specifically amplify the regions encompassing the different consensus binding sites identified. Primers designed to amplify a sequence located 30 kb upstream of Wnt9b, which does not contain HNF1-binding motifs, did not yield any enrichment product relative to the control immunoprecipitates, further confirming the specificity of the ChIP assay. Among the four binding sites located within the Wnt9b upstream region, we found strong and significant enrichment compared to control immunoprecipitates, only with pair primers surrounding the sites at position –7800 and –7878. Because these sites lie immediately adjacent one another, it most likely represents enrichment from a single target site, probably that at –7800, as suggested by CAT reporter assays (see below). Strong enrichments compared to control immunoprecipitates were also observed for the intronic Wnt9b site, as well as for the regions containing the sites found in the Pax2 (–5742, –6929 and –7247) and Lim1 (–7847) promoter sequences (Fig. 7B). As expected, ChIP assays using an anti-acetylated histone H4 demonstrated that the genomic regions of Wnt9b, Pax2 or Lim1 to which vHNF1 was recruited in vivo exhibited H4 acetylation, a hallmark of active transcription and euchromatin. Thus, vHNF1 is recruited in vivo to the promoter-enhancer elements of key regulators involved in early kidney development and nephrogenesis, which very probably results in their direct transcriptional activation.
We then assessed the ability of vHNF1 to activate transcription through the sequences identified by ChIP analysis. Selected genomic fragments were cloned upstream of the ΔTK-CAT reporter. The effect of co-expressing increasing amounts of vHNF1 was assayed by transient expression assays. The results show that vHNF1 was able to activate these different promoter-enhancer constructs in a dose-dependent manner (Fig. 7C). Although vHNF1 often behaves as a weak transactivator (Barbacci et al., 2004), it is interesting to note that, among the various enhancer-promoter elements examined, the element located in the Wnt9b first intron was specifically activated by vHNF1 at one of the highest known levels. By contrast, the Lim1 enhancer region was modestly but reproducibly transactivated by vHNF1 (n=6). Further studies are required to define whether the genomic fragment examined lacked additional sequences containing an essential co-activator-binding site or encompassed an unidentified repressor-binding element. Site-directed mutagenesis or selective deletion suggests that vHNF1 activates the Wnt9b upstream region through the site located at –7800, whereas it appears to use the two close binding sites present at –6929 and –7247 in the Pax2 promoter (Fig. 7C).
Remarkably, the vHNF1-binding sites identified in the Pax2 promoter are all located within an 8.5 kb region that drives specific expression, in transgenic mice, in the WD and derivatives but not in the induced MM (Kuschert et al., 2001). This expression profile corresponds precisely to the expression pattern of vHnf1, further confirming the possibility that vHNF1 directly controls the expression of Pax2 in these structures. These observations provide strong support for the hypothesis that different regulatory circuits operate in the ND versus the induced MM, with vHNF1 being a central component of the ND core regulatory circuit, together with Pax2 and Lim1.
During kidney organogenesis, the epithelium of the UB undergoes repetitive dichotomous branching events, leading to the formation of novel UB tips. Reciprocal inductive interactions between the branched ureteric tips and the MM lead to mesenchymal epithelial transition (MET) and tubular morphogenesis to form the nephrons. UB branching morphogenesis is therefore a fundamental process that defines the normal architecture of the kidney and the number of nephrons.
In this study, we uncovered novel developmental roles for the transcription factor vHNF1 using tetraploid and diploid chimera analysis. Initially involved in timing outgrowth of the UB and subsequent branching, it is also required for ND epithelial maintenance and MD duct formation, as well as for early nephrogenesis. Molecular analysis suggests that vHNF1 exerts these functions, at least in part, through the direct control of several key regulatory molecules. This enables us to propose a model in which vHNF1 plays a predominant role in the distinct regulatory networks controlling ND morphogenesis, UB branching and early nephrogenesis (Fig. 8).
Nephric lineage specification and maintenance
The initial step of kidney development, the conversion of mesenchymal cells in the intermediate mesoderm to epithelial cells of the ND, is mediated primarily by the transcription factors Pax2, acting redundantly with Pax8 (Bouchard et al., 2002; Torres et al., 1995), and Lim1 (Kobayashi et al., 2005; Pedersen et al., 2005). Unlike Pax2/8 and Lim1, vHnf1 is not expressed in the intermediate mesoderm, but in the epithelium of the entire ND as it differentiates from the intermediate mesoderm and in cranial mesonephric tubules. In vHnf1–/– embryos, the ND forms normally, but subsequently looses its integrity; cranial mesonephric tubules are absent and only rare caudal mesonephric tubules are formed. Furthermore, formation of the MD is disrupted. This phenotype is highly reminiscent of that reported when Lim1 is inactivated in the ND (Kobayashi et al., 2005; Pedersen et al., 2005) or in Pax2 mutants (Torres et al., 1995). Although not involved in the initial induction of Pax2 and Lim1 in the intermediate mesoderm, we show that vHNF1 is required for maintenance of their expression specifically in the ND, but not in the MM. These observations, together with the results of our ChIP analysis, led us to propose that vHNF1 is a central component of the ND core regulatory circuit, along with Pax2/8 and Lim1 (Fig. 8).
During embryogenesis, the WD is essential for both elongation and maintenance of the MD, the anlagen of the female reproductive tract. In vHnf1–/– mutants, formation of the MD is disrupted. Cranial mesonephric tubules, which give rise to the epididymis, are also absent. Lack of Lim1, Wnt9b or Pax2 expression (or, alternatively, their combined strong downregulation) very probably explains this phenotype, because all these genes have been shown to be required for the extension of the MDs (Carroll et al., 2005; Kobayashi et al., 2005; Torres et al., 1995) and for mesonephric tubule formation (Carroll et al., 2005). Furthermore, at E11-11.5, the expected expression of Lim1, Wnt9b or Pax8 in an anterior epithelial structure corresponding to the initial coelomic invagination of the developing MD was not observed (data not shown). Thus, in vHnf1 mutants, there is either no MD formation or the anterior-most anlagen of the MD is formed, but it does not express these factors. Regardless of the potential requirement for this early event, our results indicate that vHNF1, in addition to being a crucial regulator of ND morphogenesis, is essential for the generation of a functional male and female reproductive tract, providing a mechanistic explanation for the reported genital tract abnormalities in humans carrying heterozygous mutations in vHNF1 (HNF1β (Edghill et al., 2006; Haumaitre et al., 2006; Lindner et al., 1999).
Early UB branching and acquisition of a specialized tip domain
During the process of UB branching, the tip cells exhibit a highly distinctive gene expression profile (Schmidt-Ott et al., 2005) and a strong proliferative activity (Michael and Davies, 2004; Bridgewater et al., 2008). UB tip cells, which are in close proximity to the induced mesenchymal cells, control multiple processes, including EMT and subsequent formation of the nephrons.
In vHnf1-deficient embryos, formation of the UB is delayed and branching of the UB is disrupted after the first branching event. Whereas Ret and Gdnf expression are apparently normally induced, their expression is strongly downregulated at later stages and Wnt11, a component of the Ret-Gdnf-Wnt11 feedback signaling loop required for UB outgrowth (Majumdar et al., 2003), is not detected at the mutant bud tips, suggesting an abnormal UB tip specification. Although these phenotypes are closely related to those observed upon conditional inactivation of Lim1 in the ND (Kobayashi et al., 2005; Pedersen et al., 2005) or in Wnt9b mutants (Carroll et al., 2005), there are some differences that suggest additional functions of vHNF1 other than just maintenance of the expression of Lim1 and/or induction of Wnt9b in the ND. As in vHnf1–/– embryos, in Lim1 mutants, formation of the UB is delayed. In contrast to vHnf1–/– embryos, Wnt11 is normally induced in the swollen UB tips of Lim1–/– embryos, albeit the strongly reduced levels of Ret (Kobayashi et al., 2005; Pedersen et al., 2005). Furthermore, although in Wnt9b–/– embryos subsequent branching after the T-stage is disrupted, neither the timing of UB outgrowth nor the size of the UB are affected (Carroll et al., 2005).
These observations, together with the results of our diploid chimera analysis, suggest a cell-autonomous requirement for vHNF1 to establish a specialized tip domain. This appears to be, at least in part, independent of Lim1 and Wnt9b activities. Interestingly, in these mosaic experiments, cells lacking vHnf1 behave similarly to Ret null cells (Costantini and Shakya, 2006). It is uncertain whether vHnf1 functions upstream of Ret or in parallel pathways; however, it is probable that the behavior of vHnf1 mutant cells is related to the observed decreased levels of Ret signaling (Fig. 3C) (Chi et al., 2009). It is currently unknown how the boundary between tip and stalk cells is established, but recent studies suggest that WD cells undergo extensive movement and the subsequent rearrangements appear to result from competition between cells based on the levels of Ret signaling (Chi et al., 2009). It is therefore tempting to speculate that the dynamic social interactions among epithelial cells reported to occur during Drosophila tracheal branching (Ghabrial and Krasnow, 2006) might also be operative in our chimera mosaic analysis (Fig. 6C). In this respect, vHnf1–/– cells may be unable both to compete with WT cells and to cooperate with themselves to shape a specialized tip domain.
vHNF1 and Wnt9b and the initiation of the tubulogenic program
The nephrons of the amniote kidney originate from induced mesenchymal cells that aggregate next to the invading UB epithelium and then undergo MET transition and tubulogenesis. Wnt9b secreted by the UB has been shown to provide a primary inductive signal to a subset of cells within the adjacent mesenchyme to epithelialize, establishing the renal vesicle and initiating the process of nephrogenesis through the canonical Wnt signaling pathway (Carroll et al., 2005; Park et al., 2007). Remarkably, vHnf1-deficient embryos present a similar phenotype to that of Wnt9b–/– embryos (Carroll et al., 2005), in regard to the defective initiation of the tubulogenic program. Our results show that vHNF1 acting directly on Wnt9b orchestrates Wnt signaling action in the MET process underlying early nephrogenesis. Consistent with this, culture of urogenital tracts in the presence of either Wnt9b or its direct target, Wnt4, induces the formation of pretubular aggregates by the mutant mesenchyme (Fig. 5B,C). Further studies involving specific inactivation of vHnf1 in the MM are, however, required to define whether vHNF1 has additional functions at later stages in the epithelialization of pretubular aggregates and their further differentiation into mature nephrons. The potential involvement of this factor during nephrogenesis and proximal tubule differentiation is suggested both by previous analysis of cystic dysplastic kidneys of human fetuses carrying mutations in the vHNF1 (HNF1β) gene (Haumaitre et al., 2006) and by recent detailed transcriptional profiling of the developing mouse kidney. These studies show, in particular, a strong statistical association between the expression of vHNF1 in the developing proximal tubules and the presence of well-conserved HNF1-binding sites in the promoters of many genes highly enriched in this nephron structure (Brunskill et al., 2008).
Together, our findings demonstrate that vHNF1 is an essential transcriptional regulator that, in addition to its later functions in normal renal tubule morphogenesis (Gresh et al., 2004; Hiesberger et al., 2004) and its involvement in human RCAD syndrome, plays a crucial role during the earliest stages of urogenital development. This has provided new insights into the regulatory circuits controlling these processes. At later stages, renal-specific inactivation of vHnf1 resulted in the formation of renal cysts accompanied by a reduced expression of genes involved in cystogenesis, including Pkd2, Pkhd1 and Umod (Gresh et al, 2004), thus potentially linking the vHNF1 transcriptional regulatory network to ciliogenes. The expression of the cystoproteins encoded by these genes is, however, not altered in the multicystic kidneys of two fetuses carrying heterozygous vHNF1 mutations (Haumaitre et al, 2006), suggesting that, in humans, impaired gene dosage causes cystic kidneys by additional mechanisms. Notably, and strikingly similar to vHNF1, recent studies show that, in addition to its initial role in nephrogenesis, Wnt9b plays a later role in renal tubule morphogenesis. Specifically, attenuation of Wnt9b signaling during kidney morphogenesis affects planar cell polarity (PCP) of the tubule epithelium and leads to cyst formation (Karner et al., 2009). Collectively, these results suggest that the vHNF1 to Wnt9b cascade may be a major regulatory pathway involved in the development of dysplastic/cystic kidneys in vHNF1 (HNF1β) mutant carriers.
We thank N. Hastie, A. Brandli and M.Paces-Fessy for critical reading of the manuscript, and A. McMahon, G. Dressler, R.R. Behringer, V. Pachnis, P. Xu, J. Collignon A. Kispert, G. Martin and S. Vainio for reagents. We also thank F. Nailllat and S. Vainio for advice on the isolation of MM from the UB epithelium, I. Talianidis, K. Zaret and L. Sach for advice on ChIP assays, and E. Declercq, E. Manzoni and F. Gellier for animal care. This work was supported by Association pour la Recherche sur le Cancer (ARC) Contract N3911, Ligue Nationale contre le Cancer (Comite de Paris, RS 06/75–48), EuReGene Contract LSHG-CT-2004 005 085, Agence National de la Recherche ANR Blan06-2_139420, CNRS, INSERM and Université Pierre et Marie Curie. L.L and C.H. are recipients of a PhD student fellowship from Ministère de la Recherche et de la Technologie. L.L. was also recipient a PhD fellowship from ARC.