Tailbud-derived mesenchyme promotes urinary tract segmentation via BMP4 signaling

The urinary tract is composed of two distinct domains: the kidneys and a conduit system, which includes the ureters, bladder and urethra. The kidneys filter and modify blood in order to rid the body of excess fluids, solutes and metabolic wastes. These byproducts are then propelled through the ureters to the bladder for storage prior to excretion via the urethra. As would be expected based on their specialized functions, the renal tubules and the conduit system tissues exhibit vastly different structures. The renal tubules are comprised of a selectively permeable epithelial monolayer that is intimately associated with the vasculature to facilitate filtration and the exchange of solutes and ions between urinary space and systemic circulation. By contrast, the conduit system tissues are composed of an impermeable,multilayered or stratified epithelium invested with a well-developed smooth muscle coat that provides the peristaltic movements required for urine excretion.

The junction between the kidneys and the conduit system is a common site for congenital defects in humans, including obstructions that compromise renal function (Becker and Baum,2006). This border lies between the intra-renal collecting tubule system and ureter, both of which derive from an epithelial primordium called the ureteric bud (UB). Cells at the proximal UB tip are induced to form the intra-renal collecting system via their interactions with the mesenchymal progenitors of the nephrons (Fig. 1), which secrete factors such as glial cell-derived neurotrophic factor (GDNF) (for a review, see Costantini and Shakya, 2006). The distal or tubular domain of the UB differentiates into the ureter and is surrounded by a mesenchymal cell population that can be distinguished from nephron progenitors by mRNA expression patterns. Several genes, including the transcription factor Tbx18, which are selectively expressed by this mesenchyme, are required for ureter morphogenesis (Airik et al., 2006; Miyazaki et al.,2000; Oshima et al.,2001). Thus, the different mesenchymal populations surrounding the proximal and distal UB are essential for collecting system and ureter morphogenesis, respectively.

The developmental process that establishes these distinct mesenchymal tissues has yet to be determined and can occur by at least two different mechanisms (Fig. 1C,D). The distal and proximal domains of the UB may be inherently different and direct a homogeneous population of intermediate mesoderm to either maintain a nephrogenic fate or differentiate into the unique population of mesenchyme surrounding the distal UB (Fig. 1C). Alternatively, the entire UB epithelium may be a homogeneous population of cells competent to differentiate into either collecting tubule or the ureter epithelium (Fig. 1D). In this model, distinct mesenchymal populations surrounding the proximal and distal UB are specified at a stage in development prior to their association with the UB; cell migration or tissue movements are required to establish the distinct mesenchymal populations around the UB. Discrimination between these two models is dependent on determining the origin of the mesenchymal cell populations surrounding the UB.

In this report we use fate mapping techniques to show that the tailbud, a population of undifferentiated cells that is located at the most caudal aspect of the embryo after the completion of gastrulation, contributes cells to the developing urinary tract. Specifically, our data show that tailbud-derived mesenchyme associates with the distal UB but not its proximal domains, which form the intra-renal collecting system. We found that tailbud-derived mesenchyme, including those derivatives that invest the distal UB or nascent ureter, expresses abundant levels of Bmp4 mRNA and our functional experiments show that the BMP4 secreted by this mesenchyme is required for ureter differentiation. Furthermore, our data demonstrate that ectopic BMP4 signaling can induce ureter differentiation in domains of the UB normally fated to form the intra-renal collecting system. Thus, cells located at the proximal UB tips are multipotent and competent to differentiate into either ureter or collecting tubule epithelia.

Collectively, these results indicate that the ureter epithelial fate is controlled by the association of UB epithelia with tailbud-derived mesenchyme. These findings suggest that the complex morphogenetic processes required to bring tailbud-derived mesenchyme into close contact with the developing urinary tract may contribute to the high incidence of human congenital defects localized to the border between the kidneys and the ureters.

Embryonated chick eggs were purchased from CBT Farms (Chestertown, MD),incubated at 38°C for the given times and staged according to Hamburger and Hamilton (Hamburger and Hamilton,1992). All mice were housed in the Weill Medical College of Cornell University Animal Facility and treated according to the Research Animal Resource Center guidelines. Timed pregnant Swiss Webster female mice were purchased from Taconic Farms (Germantown, NY). Mice heterozygous for a Bmp4^lacZ reporter allele(Lawson et al., 1999), and mice homozygous for the Bmp4 conditional allele created by the introduction of Cre recombinase (loxP) sites into the BMP4 locus(Kulessa and Hogan, 2002),were provided by Brigid Hogan (Duke University, Durham, NC). The lacZtransgene was detected by using the following primer pair: forward,5′-ACCCAACTTAATCGCCTTGC-3′; reverse,5′-AACAAACGGCGGCGGATTGACC-3′. The following primers were used to genotype animals carrying the loxP sites in the Bmp4 allele:#79, 5′-AGACTCTTTAGTGAGCATTTTCAAC-3′; #80,5′-AGCCCAATTTCCACAACTTC-3′; #24,5′-AGGTGAGCAGAGCTAAGATG-3′. The Bmp4^flox/+versus Bmp4^flox/flox genotype was detected using primer pair #79 and #80. The #79 and #24 primer pair was used to detect the occurrence of a recombination event. Bmp4^flox/flox females were mated with Cre/Esr1 (estrogen receptor 1) heterozygous male mice(strain name: B6.Cg-Tg(Cre/Esr1)5Amc/J; The Jackson Laboratory, Bar Harbor,ME) (Hayashi and McMahon,2002). To check for the presence of the Cre recombinase transgene the following primer pair was used: forward,5′-GTGAAACAGCATTGCTGTCACTT-3′; reverse,5′-GCGGTCTGGCAGTAAAAACTATC-3′. Bmp4^flox/+ male offspring carrying the Cre transgene were selected and mated with Bmp4^flox/flox females to generate Bmp4^flox/flox;Cre/Esr1 embryos. Timed pregnant dams and their subsequent litters were normal and did not appear to be affected by the presence of the Cre transgene. Bmp4^flox/floxembryos carrying the Cre transgene died in utero approximately 24 hours after timed pregnant dams were injected with 5 mg tamoxifen (Sigma, St Louis, MO) (Hayashi and McMahon,2002).

Chick embryos at given stages were injected in ovo with ≈1 nl of either CMDiI (Molecular Probes) or 0.5×10⁷-10⁸ml^-1 virions of concentrated SNTZ retrovirus(Hyer and Mikawa, 1997; Obara-Ishihara et al., 1999). Retroviral stocks were routinely assayed for replication competence as described (Obara-Ishihara et al.,1999). Eggs were resealed and incubated for given times before fixation with 4% paraformaldehyde. Transverse vibratome sections or whole mounts of DiI-injected embryos were examined using epifluorescence optics. SNTZ-injected embryos were processed for β-galactosidase activity(Lobe et al., 1999), examined as whole mounts, and then embedded in paraffin. Serial paraffin sections were prepared and then stained with Hematoxylin and Eosin. Unless otherwise stated,all fate mapping data are representative of a minimum of four embryos analyzed for each experimental condition described.

Kidney rudiments were isolated from wild-type mouse embryos at the gestational stages noted; the day of vaginal plug was designated as E0.5. Rudiments were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomyocin on polycarbonate membrane filters (0.4 μm pore size, Transwell®; Corning, Wilkes Barre,PA). Recombinant BMP4, BMP7, Noggin and Gremlin were purchased from R&D Systems (Minneapolis, MN) and added to the culture medium daily. Timed pregnant Bmp4^flox/flox females that were mated with Bmp4^flox/+;Cre-Esr1 males were injected with tamoxifen at E12 and sacrificed 6 hours later. Kidney rudiments were isolated from embryos and cultured with 50 pM 4-hydroxytamoxifen (Sigma) to ensure complete Bmp4 excision. The heads of isolated embryos were used for genotyping.

Embryos were fixed with 4% paraformaldehyde and processed for wholemount in situ hybridization according to standard protocols(Henrique et al., 1995). Briefly, samples were treated with 10 μg/ml of proteinase K for 30 minutes at 25°C. Hybridization with digoxigenin-labeled antisense RNA probes was performed at 68°C. Reactions were developed with NBT/BCIP (Roche,Indianapolis, IN). Bmp4 plasmids were kindly provided by Brigid Hogan's laboratory. Detection of Bmp4 and Pax2 mRNAs in the same sample was performed according to the double in situ hybridization techniques of Hurtado and Mikawa (Hurtado and Mikawa, 2006).

Frozen sections, prepared according to standard protocol(Stern, 1993), or whole explants were blocked in 1% normal donkey serum (NDS)/PBS for 1 hour and primary antibodies diluted to appropriate concentrations in the same solution. Primary antibodies were incubated with samples for up to 3 hours at 37°C and secondary antibodies at described dilutions were applied after samples were thoroughly washed. Samples were examined with a Hamamatsu C4742-95 digital camera using a Metamorph Image capturing system (Universal Imaging of Molecular Devices, Downingtown, PA). Primary and secondary antibodies and their dilutions were as follows: uroplakin (rabbit anti-serum against total bovine uroplakins, a gift from the Tung-Tien Sun laboratory, New York University, 1:3000 dilution; secondary is donkey anti-rabbit, Molecular Probes, Carlsbad, CA, 1:200 dilution); smooth muscle actin (SMA)(Cy3-conjugated mouse monoclonal, Sigma, 1:400 dilution); E-cadherin (R&D Systems, 1:200 dilution; secondary is donkey anti-goat, Molecular Probes,1:200); DAPI (Sigma, 1:1000 dilution).

The early chick embryo was used to analyze the lineage of mesenchyme surrounding the UB because it is easily accessible to direct fate mapping techniques. We first demonstrated that the avian ureter and intra-renal collecting system both derive from the UB. SNTZ, a replication-defective retrovirus derived from spleen necrosis virus, was used to transfer the lacZ gene into cells for lineage tracing(Hyer and Mikawa, 1997). Replication-defective viral stocks were injected into the intermediate mesoderm of Hamburger-Hamilton (HH) st10 embryos at the axial level of somites 6-10 (Fig. 2A-C); we have previously shown that this domain of intermediate mesoderm gives rise to the nephric duct. After 24 hours of incubation, the only lineage-tagged,β-galactosdiase-positive cells detected in the caudal aspect of the embryo are localized to the nephric duct (data not shown)(Obara-Ishihara et al., 1999). After 6 days of incubation, lineage-tagged cells were detected within the nephric duct, short tubules extending into the mesonephros, and the UB(Fig. 2B). Epithelial cells of the cloaca, which gives rise to the terminal portion of the conduit system,were unlabeled (Fig. 2B). After 14 days of incubation, lineage-tagged cells were localized to the nephric duct, ureter and intra-renal collecting tubules. These data demonstrate that the intrarenal collecting tubule system and the ureters of the avian embryo,as in the mouse, derive from the UB. By contrast, the epithelia of the cloaca derive from another tissue present in the early chick embryo(Fig. 2C).

We next determined whether the mesenchymal cells surrounding the proximal and distal UB derive from a common embryonic tissue. Classical embryological studies indicate that nephron progenitors surrounding the proximal UB derive from the intermediate mesoderm (Saxen,1987). Therefore, we transferred lacZ into this tissue to test whether it gives rise to both mesenchymal nephron progenitors surrounding the proximal UB as well as the mesenchyme surrounding the distal UB. Intermediate mesoderm from the axial levels of somites 25-27 gave rise to mesonephric nephrons (n=6; data not shown). Metanephric nephron progenitors were efficiently labeled when SNTZ was injected into the unsegmented mesoderm caudal to somite 28, approximating the position of the intermediate mesoderm (n=3). Targeted lacZ transfer into this site resulted in the presence of large populations of lineagetagged,β-galactosidase-positive nephron epithelia at later stages of development(Fig. 2D). However,lineage-tagged cells were not observed in tissues surrounding the ureter or cloaca in any of the embryos analyzed. These data suggest that the proximal and distal UB are associated with mesenchyme derived from different embryonic sources.

To further test this hypothesis, we used two independent fate mapping techniques to determine the origin of the mesenchymal cells surrounding the distal UB or nascent ureter. DiI, a carbocyanine dye, was injected into HH st14-16 embryos in ovo (Fig. 3A-F). Dye injections were targeted to focal domains of mesoderm from the axial level of somite 23 through to the axial level of somite 27, and unsegmented mesoderm rostral to the posterior neuropore. In addition, DiI was injected into the medial aspect of the embryo caudal to the closed neuorpore,which can be defined as the tailbud (Catala et al., 1995; Krenn et al.,1990; Wilson and Beddington,1996). Embryos were analyzed 2-4 days post-injection and only the six embryos that were injected with DiI into the tailbud exhibited large populations of tagged cells immediately adjacent to the cloaca and the distal domain of the UB or the nascent ureter(Fig. 3E,F). Most importantly,DiItagged cells derived from the tailbud were not present around the proximal domain of the UB (Fig. 3D). Instead, the proximal UB tips or nascent collecting tubules were surrounded by unlabeled Pax2-expressing nephrogenic mesenchyme, which derives from the intermediate mesoderm (Fig. 3D) (Dressler et al.,1990). We also transferred lacZ into tailbud mesenchyme to mark cells derived from this compartment in a heritable fashion as DiI can be diluted by successive rounds of cell division(Fig. 3G-I). These fate mapping experiments, which were then analyzed at later stages of development, confirm and extend DiI fate mapping results. Cells deriving from the tailbud differentiate into sacral connective tissues, including muscle, and are present in the connective tissue surrounding the cloaca and the distal-most domain of the ureter (Fig. 3G-I). Tailbud-derived cells were not observed in the metanephros of any of the seven embryos analyzed (data not shown). Collectively, these data raise the possibility that signals secreted by tailbud-derived mesenchyme surrounding the distal UB may play a role in controlling ureter morphogenesis.

We used a candidate gene approach to test whether tailbud mesoderm expresses genes required for ureter morphogenesis. In both the developing chick and mouse embryo, several genes, including Fgf8, Tbx6, Wnt5aand Bmp4, are abundantly expressed in the tailbud(Gofflot et al., 1997; Knezevic et al., 1998). The spatial and temporal expression patterns of one of these genes, Bmp4,parallels the movement of tailbud-derived mesenchyme to a position surrounding the epithelial tissues that give rise to the conduit system. As can be seen in Fig. 4, Bmp4 is abundantly expressed in the avian tailbud at HH st14(Fig. 4A). Approximately 2 days later, abundant Bmp4 expression is detected around the caudal-most aspect of the developing urinary tract, including the cloaca and distal UB. By contrast, the proximal UB is surrounded by Pax2-expressing nephron progenitors (Fig. 4B-D).

Similarly, in the developing mouse Bmp4 is abundantly expressed in the tailbud (Gofflot et al.,1997) and later, from E11.5-E13.5 by mesenchyme surrounding the distal UB and urogenital sinus (Fig. 4E,F). Thus, the spatial and temporal expression patterns of Bmp4 mRNA are consistent with the movement of tailbud-derived cells to the developing conduit system and a possible role in ureter morphogenesis. Moreover, by E15.5, overt ureter differentiation, as determined by the upregulated expression of uroplakins in the UB epithelium and presence of an organized, SMA-positive connective tissue coat, occurs only in the domain of the UB invested with Bmp4-expressing mesenchyme(Fig. 5A-C). Uroplakins are not expressed by collecting tubule epithelia and although SMA-positive cells are interspersed between collecting tubules in the medullary interstitium at later stages of development, these SMA-positive cells are not organized into a thick, circumferentially oriented connective tissue investment. Finally, we show that ureter differentiation occurs in metanephric kidney explants(Fig. 5D-F). Neither uroplakins(Fig. 5D) nor SMA-positive mesenchyme (data not shown) can be detected in freshly isolated E12.5 metanephric kidney rudiments, whereas these markers of ureter differentiation are abundantly expressed after 4 days of culture(Fig. 5E,F). Importantly, the domain of the UB that displays the ureteral phenotype in vitro parallels the spatially restricted expression of Bmp4 in adjacent mesenchyme(Fig. 5G). Thus, this explant system successfully recapitulates ureter differentiation.

The role of BMP signaling in controlling ureter morphogeneis was analyzed in vitro using metanephric rudiments isolated from mice in which Bmp4alleles had been deleted by Cre recombinase-mediated excision(Fig. 6A). Female Bmp4^flox/flox mice were bred with male Bmp4^flox/+ mice harboring a transgene containing sequence encoding a fusion protein of Cre recombinase and estrogen receptor 1 under the control of a chicken beta-actin promoter/enhancer coupled with the cytomegalovirus (CMV) immediate-early enhancer(Hayashi and McMahon, 2002). The Cre/ESR1 fusion protein can enter the nucleus only when bound to an estrogen receptor 1 ligand. Therefore, Cremediated excision of the floxed Bmp4 alleles is inducible by maternal administration of the synthetic estrogen, tamoxifen. Timed pregnant mice were injected with tamoxifen at E12 and embryos harvested 6 hours post-injection because Bmp4^flox/flox embryos carrying the Crerecombinase transgene begin to die in utero 24 hours after tamoxifeninduced Bmp4 excision. Metanephric kidney rudiments were isolated and cultured in vitro in the presence of tamoxifen to ensure complete gene excision and Bmp4 expression levels were assessed after 4 days of growth by in situ hybridization. Abundant Bmp4 mRNA expression was seen in rudiments from Bmp4^flox/flox embryos lacking the Cre transgene (Fig. 6B), whereas Bmp4 mRNA expression was not detected in rudiments isolated from Bmp4^flox/flox, Cre/ESR1 embryos(Fig. 6C). After 4 days of culture, uroplakin and SMA expression were detected in the distal-most domain of Bmp4^flox/flox rudiments lacking the Cretransgene (Fig. 6D,F,H,J),whereas the expression of these markers indicative of ureter differentiation was dramatically reduced when Bmp4 alleles were excised at E12(n=4; Fig. 6E,G,I,K). Excision of Bmp4 alleles at developmental stages later than E12.5 did not perturb ureter differentiation, as determined by upregulated uroplakin expression and the formation of a SMA-positive connective tissue coat(n=4; data not shown). Collectively, these data demonstrate that BMP4 signaling prior to E12.5 is required for ureter morphogenesis.

Because Bmp4 expression is restricted to the distal UB at the time that this tissue is sensitive to BMP-dependent ureter induction, we tested whether ectopic BMP4 signaling can induce ureter differentiation in proximal domains of the UB normally fated to differentiate into the collecting tubules. E12.5 metanephric kidney rudiments were cultured in the absence or presence of increasing concentrations of recombinant, soluble BMP4 for 4-5 days(Fig. 7A-F). Explants were assayed for ureter differentiation by immunofluorescent techniques.

Recombinant BMP4 markedly alters the growth of metanephric kidney rudiments(Bush et al., 2004; Cain et al., 2005; Raatikainen-Ahokas et al.,2000). E12.5 rudiments cultured with increasing concentrations of recombinant BMP4 exhibit a dose-dependent decrease in UB tip and nephron number as compared with control cultures(Fig. 7C-F). Most importantly,in this report we show that exogenous BMP4 converted the proximal UB tips into tubular structures that exhibit a ureteral phenotype, in a dose-dependent manner (Fig. 7C-F). The proximal UB tips exhibit upregulated uroplakin expression(Fig. 7C,E) and become invested with a condensed, SMA-positive smooth muscle coat(Fig. 7D,F). Developmental plasticity persists in the UB, as similar results were obtained when E14.5 rudiments were cultured in vitro with high concentrations of recombinant BMP4(Fig. 7G-I). As expected, in both control and BMP4-treated cultures, the distal domain of the UB differentiated into the ureter (Fig. 7G,H). Strikingly, the proximal UB tips also acquired a ureteral phenotype when older rudiments were cultured with recombinant BMP4, whereas large portions of the UB network between the definitive ureter and the UB tips did not develop ureteral characteristics(Fig. 7I,J). These data demonstrate that the proximal UB tips, normally fated to differentiate into the collecting system, acquire the ureteral phenotype in the presence of ectopic BMP4 signaling.

By contrast, high concentrations of BMP7, a TGFβ family member expressed by the UB and by nephrogenic mesenchyme around its proximal tips(Dudley and Robertson, 1997),were unable to induce ureter morphogenesis in the proximal domains of the UB normally destined to form the collecting system in vitro (data not shown). Together, these data demonstrate that ectoptic signaling by BMP4, a paracrine factor normally present in mesenchyme surrounding the distal UB, is capable of inducing ureter morphogenesis in domains of the UB fated to differentiate into the intra-renal collecting system. Thus, spatially restricted Bmp4expression plays a role in determining the location at which the UB is subdivided into the ureter and the collecting system.

The patterning of the UB into the intra-renal collecting system and ureter has significant functional consequences, as these segments of the urinary tract are essential for the regulation of body fluid volume and the excretion of wastes, respectively. Moreover, the junction between the collecting system and ureter is highly prone to congenital defects in humans. Here, we have shown that this patterning event is dependent on the association of the UB with two distinct mesenchymal cell populations: tailbud and intermediate mesenchyme. Tailbud-derived mesenchyme selectively associates with the distal UB and induces ureter morphogenesis, in part, via expression of BMP4.

BMP signaling is likely to play a role in multiple processes required for ureter morphogenesis. Our results, combined with previous studies, implicate BMP signaling in the control of ureteral smooth muscle formation(Airik et al., 2006; Miyazaki et al., 2003).

Moreover, targeted deletion of Shh, which is expressed in the nascent ureter epithelium, results in a deficit of Bmp4-expressing peri-ureteral mesenchyme and an abnormally thin ureteral smooth muscle coat(Yu et al., 2002). Collectively, these data indicate that formation of the ureteral smooth muscle coat is dependent on signals from the nascent ureter epithelium, similar to what has been shown by Baskin in the developing bladder, and signaling within the mesenchymal population surrounding the nascent ureter(Baskin et al., 1996).

In this study we show that Bmp4 is also required for the upregulated expression of uroplakins in the differentiating ureter. Uroplakins are a family of transmembrane proteins selectively expressed by the ureter epithelium and required for its water-impermeable properties(Hu et al., 2001). BMP4 secreted by periureteral mesenchyme may signal directly to the underlying epithelium, as UBs isolated from the surrounding mesenchyme cease branching when exposed to recombinant BMP4 in vitro(Bush et al., 2004). However,targeted deletion of SMAD4 in the UB epithelium does not grossly affect ureter morphogenesis in vivo (Oxburgh et al.,2004). Because SMAD4 is an obligatory component of SMAD-dependent BMP signaling, if BMP4 directly signals to the UB epithelium, it is likely to do so via SMAD-independent mechanisms such as activation of LIM1 kinase(Lee-Hoeflich et al., 2004). Alternatively or additionally, Bmp4 may upregulate the expression of unknown mesenchymal signaling factors that then directly induce ureter epithelial differentiation. Future experiments examining ureter differentiation in animal models lacking BMP receptors within the UB will be required to further dissect the mechanisms mediating BMP-dependent ureter epithelial differentiation.

We have also shown that ectopic BMP4 signaling induces ureter differentiation. Raatikainen-Ahokas et al. demonstrated that the large portions of the UB network acquire a dense connective tissue coat characteristic of the ureter when metanephric kidney explants are cultured with recombinant BMP4 (Raatikainen-Ahokas et al., 2000). Our experiments confirm and extend these observations. We show that ectopic BMP4 signaling at the early stages of renal development converts the entire UB network into tubules exhibiting a ureteral phenotype as determined by the upregulated expression of uroplakins and the acquisition of a thick connective tissue coat. This phenotypic conversion may be due to BMP-dependent migration or overproliferation of connective tissue and uroplakin-expressing cells from the nascent ureter to more proximal domains of the UB network. However, experiments analyzing UB differentiation in E14.5 kidney rudiments suggest that ectopic BMP4 signaling actually changes the fate of the UB from a collecting tubule to ureteral phenotype. High concentrations of recombinant BMP4 induce the most proximal UB tips to acquire a ureteral phenotype, whereas large portions of the UB network located between these abnormal UB tips and the definitive ureter do not develop ureteral characteristics. This result strongly suggests that ectopic BMP4 signaling changes the fate of the proximal UB from a collecting tubule to ureteral phenotype, because it is extremely unlikely that large populations of both uroplakin- and SMA-positive cells move from the distal-most domain of the UB to its most proximal tips without populating the intervening UB segments. These results are consistent with earlier studies demonstrating that the differentiated phenotype of epithelia derived from the UB is controlled, in part, by factors secreted by surrounding mesenchyme(Lipschutz et al., 1996).

Modulation of the timing and level of BMP signaling may be essential for the proper patterning of these tissues. Our results suggest that portions of the UB network become refractory to BMP-induced ureter differentiation or committed to a collecting tubule phenotype by E14.5. This early specification of the UB appears to allow BMP4 signaling to mediate additional developmental processes at later stages. For example, BMP4 expressed in the medullary interstitium and glomerulus may pattern the renal vasculature, as signaling by this TGFβ family member has been implicated in blood vessel formation,and antagonists of BMP signaling compromise the integrity of the glomerular capillary tuft (Miyazaki et al.,2006; Nimmagadda et al.,2005).

Sensitivity to both specific ligands and the total level of BMP signaling in the target tissue may also allow the BMPs to promote the formation of different cell types even in the absence of temporal restriction. Dose-response experiments included in this study demonstrate that high concentrations of recombinant BMP4 induce ureter morphogenesis in domains of the UB normally fated to differentiate into the renal collecting system whereas BMP7 does not. BMP7 is expressed by the UB and by nephrogenic mesenchyme surrounding its branching tips in vivo and is essential for maintaining the viability of nephron progenitors within the developing kidney(Dudley et al., 1995; Dudley and Robertson, 1997; Oxburgh et al., 2005). Since the receptor specificity of BMP4 and BMP7 differ, it is possible that the receptors mediating BMP4-dependent ureter morphogenesis do not respond to BMP7 signaling. However, despite the distinct receptor specificity and functions of BMP4 and BMP7 in the developing kidney, Bmp4 can substitute for Bmp7 in supporting nephron progenitor viability in vivo(Oxburgh et al., 2005). We hypothesize that the levels of ectopic Bmp4 mRNA expressed around the UB tips when the gene is knocked into the Bmp7 locus are not sufficient to induce ectopic ureter morphogenesis in vivo(Oxburgh et al., 2005). This hypothesis is consistent with data demonstrating that different levels of BMP signaling promote diverse responses in many embryonic cell types(James and Schultheiss, 2005; Oxburgh et al., 2005; Wilson et al., 1997).

Collectively, our data indicate that spatially restricted BMP4 signaling is essential for normal urinary tract patterning. Restricted BMP4 expression in the developing urinary tract may be imposed by several different mechanisms. Fate mapping experiments combined with mRNA expression analyses presented in this study show that the mesenchyme surrounding the distal UB derives from the tailbud and expresses Bmp4 prior to associating with the developing urinary tract. By contrast, the mesenchyme surrounding the domain of the UB fated to form the renal collecting system does not express Bmp4 and derives from the intermediate mesoderm. Therefore, one of the mechanisms that confines BMP4 signaling to the domain of the UB fated to become the ureter is the association of tailbud-derived mesenchyme with the most caudal segments of the developing urinary tract, including the distal UB.

Tailbud-derived mesenchyme also associates with the cloaca. The cloaca derives from endoderm and its ventral wall forms the bladder, the structure into which the ureter drains (Carlson,1996). The bladder epithelium has different in vitro growth characteristics from the ureter epithelium, which derives from mesoderm(Liang et al., 2005). Despite these differences, the ureter and bladder epithelium share many similar features, including morphology and gene expression patterns(Liang et al., 2005). We suggest that the association of a common population of tailbud-derived,BMP4-expressing mesenchyme with both the distal UB and cloaca accounts for these similarities.

Our fate mapping results address the origin of the mesenchyme surrounding the distal UB in the developing chick embryo and several lines of evidence suggest that tailbud-derived mesenchyme plays a role in conduit system formation in a variety of species. Sustained BMP signaling within the tailbud is essential for the formation of the terminal portions of the zebrafish urinary tract (Pyati et al.,2006). Similarly, targeted deletion of the vitamin A degradation enzyme cyp26, which is localized to the murine tailbud, results in both massive tailbud apoptosis and gross ureter and bladder defects(Sakai et al., 2001).

Mesenchyme from the tailbud remains poorly understood. As this tissue does not arise from primary gastrulation, it is possible that it has a distinct gene expression profile and developmental potential compared with other embryonic mesenchyme (Catala et al.,1995; Gofflot et al.,1997; Knezevic et al.,1998; Wilson and Beddington,1996). Our data raise the possibility that this mesenchyme is required for the formation of multiple structures within the murine urogenital system. In support of this hypothesis, a recent study indicates that mesenchyme surrounding the cloaca, which we show derives from the tailbud,plays a fundamental role in the differentiation of the bladder, urethra and genitalia (Haraguchi et al.,2007). Thus, tailbud mesenchyme may be required for the formation of multiple structures, both in the urogenital system and throughout the lumbar and sacral regions of the embryo. We suggest that the fate and function of tailbud mesenchyme merits further investigation.

Our data shows that a sharp boundary between the ureter and the branched collection system can be detected in the developing murine urinary tract by E15.5. This initial boundary undergoes further differentiation to create the renal pelvis, which does not fully mature until birth and shares properties with both the renal collecting system and ureter(Miyazaki et al., 1998). The mechanisms regulating renal pelvis formation are currently unclear, but it is likely that a failure to properly form an initial boundary at this site will ultimately result in general defects at the urteral-pelvic junction.

In conclusion, our results indicate that formation of the ureteralpelvic junction, the border between the kidney and conduit system in the mature animal, is a complex, multi-stage process that requires the integration of cells from at least three different embryonic lineages into a single,composite structure. These aspects of ureteralpelvic junction formation are reminiscent of cardiac outflow tract morphogenesis, which requires the integration of cells derived from the neural crest, lateral and cardiac mesoderm into a single, functional structure(Carlson, 1996). Quite possibly, as a result of the complex tissue interactions involved, the outflow tract is a major site of congenital malformations in humans. Our experiments analyzing ureter morphogenesis suggest that the high incidence of congenital defects localized to the ureteral-pelvic junction may also be related to the complexities of integrating cells from a variety of lineages into a single composite site.

We thank Dr J. Timmer for his helpful comments and patience, Dr F. Stern for his support and Dr B. Hogan for the use of mice constructed in her laboratory. This work was supported by NIHR01DK45218 and NIHR01DK056365 awarded to D.H. and a Postdoctoral Fellowship awarded by the Spanish Ministerio de Educación y Ciencia to C.C.