A Wnt7b-dependent pathway regulates the orientation of epithelial cell division and establishes the cortico-medullary axis of the mammalian kidney

The functional renal circuitry comprises a long tubular epithelial network with distinct origins. The main body of the nephron derives from the metanephric mesenchyme, which epitheliarizes in response to Wnt inductive signals (Merkel et al., 2007). The ureteric epithelium arises from branching growth of the ureteric bud,which establishes the collecting duct network. The collecting duct system and nephrons of the mammalian kidney are organized along a renal cortico-medullary axis, such that different segments of nephrons and collecting ducts are spatially restricted within distinct cortical or medullary domains. The renal corpuscles and the convoluted tubules (proximal and distal) of the nephrons reside in the cortex (Little et al.,2007). The loop of Henle, which connects proximal and distal tubules of the nephron, extends into the medullary zone of the kidney,adjacent to the elongated tubular network of the collecting duct system(Little et al., 2007). The collecting duct epithelium also contains distinct cell types in cortical and medullary domains (Schuster,1993). The cortico-medullary organization of the kidney is crucial for renal function; for example, the concentration of urine within the medullary compartment. However, the mechanisms that regulate cortico-medullary axis formation are unclear.

Our previous studies documented several Wnt family members that are expressed within subdomains of the developing mammalian kidney, and several of these regulate distinct renal developmental events(Carroll et al., 2005; Kispert et al., 1998; Kispert et al., 1996; Majumdar et al., 2003; Stark et al., 1994). Wnt7b encodes a Wnt ligand whose expression is restricted to the non-branching ureteric trunk component of the collecting duct system. Several studies have documented the actions of Wnt7b in multiple aspects of mammalian development, but its role in kidney development has not been addressed. Wnt7b signaling is crucial for placental, lung, eye,dendrite and bone formation (Lobov et al.,2005; Parr et al.,2001; Rajagopal et al.,2008; Rosso et al.,2005; Shu et al.,2002; Tu et al.,2007). In different tissues, Wnt7b functions via different branches of the Wnt signaling pathway, including the canonical Lef/β-catenin pathway (Lobov et al.,2005; Wang et al.,2005), the non-canonical planar cell polarity (PCP) pathway(Rosso et al., 2005), and the newly described G protein-linked PKC delta pathway(Tu et al., 2007). We demonstrate here that Wnt7b is essential for the establishment of the cortico-medullary axis of the mammalian kidney through the regulation of cell cleavage planes within the collecting duct epithelium. Mechanistic analyses suggest that Wnt7b acts indirectly by activating a canonical Wnt signaling pathway in an interstitial mesenchyme cell intermediate that coordinates the elongation of epithelial tubular networks forming the medullary zone.

For BrdU labeling, pregnant female mice were injected with 0.5 mg BrdU/10 g body weight 2 hours before the harvesting of embryonic kidneys. For histological staining, immunofluorescent staining of paraffin sections, and in situ hybridization of frozen sections, tissues were fixed in 4%paraformaldehyde (PFA) at 4°C for 24 hours. For immunofluorescent analysis of frozen sections, tissues were fixed in 4% PFA or 2% PFA for 1 hour at 4°C. For vibratome section immunofluorescent staining, freshly dissected kidneys were fixed in 4% PFA at 4°C for 30 minutes, and then embedded in 15% gelatin/PBS (phosphate-buffered saline). After solidifying for 30 minutes at 4°C, gelatin blocks were fixed in 4% PFA at 4°C for 2 hours.

Paraffin blocks were sectioned at 6 μm for Hematoxylin and Eosin staining as described by Yu et al. (Yu et al., 2002). Frozen sections were sectioned at 12 μm. Gelatin blocks for vibratome sectioning were sectioned at 75 μm. Sections were blocked in blocking buffer (3% BSA, 1% normal donkey serum, PBS/0.1% Triton X-100) for 1 hour at room temperature and then incubated overnight at 4°C in primary antibodies diluted in blocking buffer. After washing in PBS/0.1%Triton X-100, sections were incubated in secondary antibodies diluted in blocking buffer at room temperature for 2 hours for frozen sections or at 4°C overnight for vibratome sections. Sections were then stained with Hoechst 33342 for 5 minutes, post-fixed with 4% PFA for 20 minutes at room temperature, and mounted with Vectashield mounting media (Vector Laboratories)with Hoechst 33342. TUNEL staining was performed with the ApopTag Red In situ Apoptosis Detection Kit (Chemicon International), following the manufacturer's instructions. Primary cilia immunostaining was visualized with a Personal DeltaVision microscope (Applied Precision). Three-dimensional projections were generated from stacks of optical sections. Other images were collected with a Zeiss LSM510 Axioplan 2 confocal microscope. For quantification of cell proliferation and apoptosis, 300-1600 cells from two to four sections of each kidney were counted. Primary antibodies used in this study were as follows:anti-phospho-histone H3 (Upstate Cell Signaling), anti-pan cytokeratin(Sigma), anti-BrdU (BD Pharmingen), anti-acetylated α-Tubulin (Sigma),anti-Polaris (gift of Dr B. K. Yoder, University of Alabama at Birmingham),anti-Cdh6 (gift of Dr G. Dressler, University of Michigan, Ann Arbor),anti-Umod (Biomedical Technologies), DBA-biotin (Sigma), anti-p57Kip2(Neomarkers), anti-Lef1 (Santa Cruz), anti-β-galactosidase (Cappel),anti-Integrin α3 (gift of Dr J. Kreidberg, Children's Hospital, Boston),and anti-E-Cad (Zymed).

Freshly dissected kidneys were fixed in 4% PFA at 4°C for 1 hour. Kidneys were stained in lacZ staining solution at 4°C overnight. After post-fixing in 4% PFA, kidneys were embedded in 15% gelatin/PBS to make vibratome blocks. Three hundred micrometer (E13.5) and 150 μm (other stages) vibratome sections were dehydrated through a graded methanol series and cleared in benzyl alcohol:benzyl benzoate (1:1) for photography.

Frozen blocks were sectioned at a thickness of 16 μm. In situ hybridization was performed as described by Little et al.(Little et al., 2007). Briefly, sections were treated with 10 μg/ml proteinase K for 10 minutes,and hybridized with 500 ng/ml digoxigenin-labeled riboprobes overnight at 68°C. After the first post-hybridization wash, sections were treated with 2 μg/ml RNase for 15 minutes at 37°C. Sections were incubated in anti-DIG-AP antibody (1:4000, Roche) at 4°C overnight. After incubation with BM purple to visualize signals, sections were fixed in 4% PFA/0.2%glutaraldehyde and mounted in glycergel mounting media (DAKO).

Fifty micrometer frozen sections of E15.5 kidneys were immunostained with anti-phospho-histone H3 and anti-pan cytokeratin antibodies. Stacks of confocal optical sections were collected, from which three-dimensional reconstructions were generated with Imaris software. Mitotic angles were measured according to Fischer et al.(Fischer et al., 2006). Briefly, in three-dimensional reconstructions, the poles of each of the two anaphase chromosome clusters were marked as measurement points P1 and P2 with Imaris software. The x, y and z coordinates of the two points were then acquired with the Imaris software for generation of the vector of the mitotic division axis. Two points (P3, P4) on the collecting ducts harboring the anaphase cell were taken to generate the vector of the longitudinal axis of the collecting duct epithelium. The coordinates of the four points were used to calculate the angles between the vector of the mitotic division axis and the vector of the longitudinal axis of the collecting duct epithelium, the mitotic angle. Mann-Whitney U tests were performed for statistical significance of the distribution of mitotic angles between control and Wnt7b mutant samples.

A number of Wnt genes are expressed with distinct spatial and temporal patterns within the mammalian metanephric kidney and regulate diverse aspects of kidney development (Merkel et al.,2007). Wnt7b expression is restricted to the Wolffian duct-derived ureteric trunk epithelium that gives rise to the collecting duct network and ureter of the urinary system(Fig. 1A-C), but is excluded from the ureteric tips. Wnt7b null mice die at embryonic day 9.5(E9.5), before the onset of major organogenesis, as Wnt7b is essential for placental development (Parr et al., 2001). To genetically dissect Wnt7b functions in organ formation, we placed loxP sites on either side of exon3 of the Wnt7b gene (Wnt7b^c3)(Rajagopal et al., 2008),generating a Cre-dependent conditional null allele. As expected, when Wnt7b function was removed from both extra-embryonic and embryonic tissues by Cre-mediated recombination, embryos homozygous for this allele were phenotypically identical to those harboring the previously described Wnt7b null allele (Parr et al.,2001).

To bypass the embryonic lethality that results from the placental function of Wnt7b, Wnt7b activity was specifically removed from the mouse epiblast lineage by using the Sox2Cre driver line(Hayashi et al., 2002). The resulting Wnt7b^c3/-; Sox2Cre progeny (hereafter referred to as Wnt7b mutants) displayed a hypoplastic lung phenotype, were unable to breathe and died shortly after birth(Rajagopal et al., 2008). Histological examination of E18.5 kidneys revealed a striking phenotype: a complete absence of the renal medulla (Fig. 1J,K). In a wild-type kidney at this stage, the renal corpuscles are separated from the renal pelvis by the elongated epithelial network of the medullary region (Fig. 1J,arrow). By contrast, renal corpuscles abut the renal pelvis in Wnt7bmutants (Fig. 1K, arrow). Despite this pronounced medullary phenotype, the mutant kidneys were of a similar size to those of wild-type littermates: the average length of wild-type kidneys was 3.0±0.1 mm (n=3) and those of mutants 3.0±0.4 mm (n=4; P=0.90). Thus, the circumferential growth of the cortex, which is driven by branching morphogenesis of ureteric tips, did not appear to be compromised. Consistent with this interpretation,the mutant cortical region appeared histologically normal(Fig. 1K) and was similar in thickness to that of wild-type control kidneys when measured at the central-most plane (0.61±0.01 mm, n=4, versus 0.53±0.08 mm, n=3, respectively; P<0.17). Furthermore, the expression of an extensive set of markers of cortical development (Wnt11,Wnt4, Foxd1, Gsh1, Slc34a1 and Slc12a3) that assessed ureteric branching, nephron induction and nephron patterning were represented normally in the cortical region of medullary-deficient Wnt7b mutant kidneys at E15.5 (Fig. 2). That the size of the kidney and the thickness of the renal cortex are similar between wild-type and mutant kidneys at E18.5 in most cases demonstrated that the mutant renal pelvis occupied a similar volume to the control renal pelvis with a renal medulla in it. At E18.5, whereas all Wnt7b mutants showed a complete absence of the medullary region, less than one-third of mutants exhibited a hydroureter-like swelling of the pelvic region and ureter, and the expressivity of this phenotype varied considerably (data not shown). Thus, the enlarged-looking renal pelvic space in Wnt7b mutants does not appear to reflect an enlarged pelvis per se but rather the absence of the tubular epithelial network that demarcates the renal medulla.

To determine whether the observed renal phenotypes reflected a specific function for Wnt7b in the kidney and not a secondary consequence of Wnt7b signaling elsewhere in the embryo, a Hoxb7Cre driver line (Yu et al., 2002) was used to remove Wnt7b activity specifically within the nascent collecting duct epithelium. Wnt7b^c3/-; Hoxb7Cre embryos phenocopied the renal defects in Wnt7b^c3/-; Sox2Creembryos at E18.5 (data not shown). Pups with a kidney-specific Wnt7bremoval survived up to 11-12 days postpartum (P11-P12); however, the kidneys at this time were grossly abnormal and physiologically incompetent, which led to the postnatal death. Specifically, the urine osmolality of P10 Wnt7b^c3/-; Hoxb7Cre pups was only 56% that of wild-type littermates (see Fig. S2 in the supplementary material), which is consistent with a failure of medullary functions. Thus, kidney-specific removal of Wnt7b signaling demonstrates a direct action for the Wnt7bpathway in morphogenesis of the renal medulla and, as a consequence, the normal concentration of urine in this kidney compartment.

To determine whether the absence of a renal medullary compartment at E18.5 results from a failure of or a delay in renal medulla formation, or from a loss of the structure after the initiation of renal medulla formation, we analyzed a developmental series of wild-type and Wnt7b mutant embryos. No medullary compartment was evident in wild-type E14.5 embryos(Fig. 1D), but a medullary domain emerged between E15.5 and E16.5(Fig. 1). The first evidence of an elongating tubular epithelium separating the renal cortex from the renal pelvis was evident at E15.5 (bracketed in Fig. 1F). By E16.5, a converging elongated tubular network of the renal medulla was a prominent feature of kidney organization (Fig. 1H). Wild-type and Wnt7b mutants were indistinguishable at E14.5 (Fig. 1D,E); however,the nascent renal medulla of the wild-type was absent from Wnt7bmutant kidneys at E15.5 (Fig. 1F,G). This phenotype became markedly more pronounced by E16.5(Fig. 1H,I). No renal medulla was evident either in Wnt7b^c3/-; Hoxb7Cre mutant kidneys,even at P10 (Fig. 1L,M). In summary, Wnt7b signaling is essential for establishing the renal medulla of the mouse kidney. In this, Wnt7b appears to play a primary role in the establishment of a cortico-medullary axis of epithelial organization.

The epithelial components of the renal medulla comprise medullary collecting ducts derived from the ureteric trunks and extensions of the loop of Henle, renal tubular components of the main body of the nephron. The major structure driving renal medulla formation appears to be the collecting duct epithelium, because compartmentalization of the renal cortex and renal medulla is evident in the absence of the loop of Henle in severe hypomorphs of Fgf8 signaling (Grieshammer et al., 2005; Perantoni et al.,2005) and following nephron-specific removal of the transcriptional regulator Lim1(Kobayashi et al., 2005). Therefore, we focused our initial analysis on the ureteric bud-derived collecting duct epithelium to understand the mechanisms of normal renal medulla formation and the role of Wnt7b signaling in this process.

New ureteric epithelium is generated in part through the branching growth at the cortical tips of the ureteric network. The cortical tip-specific expression of Wnt11 mRNA provides a general means of assessing both the normal regulation of the genetic circuitry governing tip growth and the number of branch tips within the kidney at any given stage. At E14.5,expression of Wnt11 was not noticeably different in Wnt7bmutant kidneys when compared with wild-type littermates; similar numbers of branching events were observed and Wnt11 was appropriately restricted to the tip region (Fig. 2; see also Fig. S3 in the supplementary material). A further analysis of cortical-medullary stratification of gene expression within the ureteric epithelium also demonstrated appropriate expression of two other Wnt family members: Wnt9b throughout much of the collecting duct epithelium except the tips and the distal-most region and ureter; and Wnt7bitself in more distal regions of the collecting ducts and the pelvic and ureteral transitional epithelium (Fig. 2). Foxa1, a novel marker we recently identified(Little et al., 2007) of more distal epithelia in the presumptive medullary collecting duct region and the pelvic and ureteral transitional epithelium, was expressed as expected in the mutant (Fig. 2). Finally, Foxi1 demarcates intercalated cells, a differentiated cell type specific to the collecting duct epithelium; its normal expression in mutant kidneys demonstrates that the epithelium also differentiated as expected(Fig. 2). Thus, the failure of renal medullary formation was unlikely to result from either a mis-patterning or a mis-specification of the epithelial network of the developing collecting duct system.

To develop a better understanding of the dynamic organization of the renal medullary collecting ducts, we genetically labeled the collecting ducts combining the Cre-dependent β-gal-producing R26R mouse line(Soriano, 1999) with the ureteric epithelium-specific Hoxb7Cre strain(Fig. 3). Reorganization was observed in both the organization of the proximal ureter and the distal collecting duct network in conjunction with renal medulla development. The appearance of the future renal pelvis was preceded at E14.5 by a swelling at the proximal end of the ureter, at the intersection with the presumptive collecting duct epithelium (Fig. 3A, arrow). Within the latter, branches were approximately evenly spaced along the future cortico-medullary axis at E13.5, but appeared to elongate in the interbranch component of epithelium in the distal region close to the ureter by E14.5. This feature is consistent with the reported longitudinal growth of the ureteric trunk(Cebrian et al., 2004). This distended region continues to expand from E14.5 to E15.5 as the renal pelvic space is established. By E15.5, the ureteric trunks generated from the first few branching events orientate and extend in a spoke-like arrangement into,and centered on, the pelvic space, marking the initiation of renal medulla formation (Fig. 3A). A high rate of proliferation within this region of the ureteric epithelium correlates with these events (Fig. 4A),suggesting that regional cell proliferation may play some roles in the formation of renal medulla. By E16.5, an elaborate network of ureteric branches funnels into the renal pelvic space, the typical architecture of the mature kidney structure.

The gross renal architecture was examined at E15.5 in wild-type and Wnt7b mutant littermates. In freshly prepared specimens (avoiding potential fixation artifacts), the ureter of mutants appeared normal and undilated [Fig. 3B; diameter of the wild-type ureter, 158.0±16.5 μm (n=6); diameter of the mutant ureter, 165.9±37.9 μm (n=6); P=0.63]. Histological analysis demonstrated that the renal pelvis was also comparable between wild-type and mutant kidneys (Fig. 1F,G, insets). However, whereas the collecting ducts in the nascent medullary region of wild-type embryos were long, thin epithelial tubes, those of Wnt7b mutants were considerably dilated and the branch points were less evident, most likely obscured by the epithelial dilation (Fig. 3C). These morphological disparities were observed even though the epithelium preserved a normal columnar, epithelial structure and cell proliferation was unaltered(Fig. 4A). Thus, neither rearrangement of cell shapes nor changes in the proliferative rate can readily account for the failure to establish a nascent medullary organization in Wnt7b mutants.

Oriented cell division has been postulated to regulate the appropriate growth of collecting duct epithelium in the postnatal kidney(Fischer et al., 2006). We explored the possibility that a Wnt7b-dependent orientated cell division event may normally regulate medullary morphogenesis. If collecting duct epithelial cells divide along the longitudinal axis of the duct (mitotic angle = 0 degrees), this orientated cell division would be expected to increase the length of the duct but not the circumference. If, on the contrary, cells divide along the radial axis of collecting duct (mitotic angle= 90 degrees), the duct is expected to increase in circumference, dilating rather than elongating (Fig. 4B). To determine the orientation of cell division in wild-type and mutant embryos, we measured mitotic angles of prospective medullary collecting duct cells as renal medulla formation initiates (E15.5). Phospho-histone H3 immunostaining was performed to visualize the mitotic configuration in collecting duct epithelium delineated by pan-cytokeratin staining (Fig. 4B). For accurate measurement, only the anaphase configuration was considered. As shown in Fig. 4C, the majority of wild-type cells (65.7%) divided along the longitudinal axis of the collecting duct epithelium (mitotic angles <45 degrees), whereas the majority of Wnt7b mutant cells (67.5%) divided along the radial axis of the collecting ducts (mitotic angles >45 degrees). The statistically significant difference in the bias of the plane of cell division may contribute to the shorter, wider ducts of Wnt7b mutant kidneys.

A re-orientation of cell division has recently been linked to cyst formation in the collecting ducts of mouse models of polycystic kidney disease(Fischer et al., 2006; Saburi et al., 2008), where the structure and/or action of the primary cilium is defective. By contrast,the primary cilium within the developing collecting duct network of Wnt7b mutants appeared grossly normal(Fig. 4D). Thus, the reorientation of the plane of cell division in Wnt7b mutants was not obviously linked to a cilia structural defect. Furthermore, proliferation and apoptosis, which were both elevated in most cystic kidneys, were not affected in collecting duct precursors in Wnt7b mutants at the onset of the renal medullary defect (E15.5), although apoptosis in the distal collect epithelium showed a fourfold increase in the mutants at E17.5(Fig. 4A,E; see also Fig. S4 in the supplementary material). Thus, the only cellular property that we can identify that clearly correlates with the onset of the defects in epithelial organization in collecting ducts of Wnt7b mutant kidneys was the polarity of cell division.

Another major epithelial component of the renal medulla is the loop of Henle, an intermediate segment of the nephron. At E18.5, the loop of Henle spans the entire length of the renal medulla(Fig. 5A). By contrast, the loop of Henle in Wnt7b mutant kidneys was truncated(Fig. 5B), resembling an earlier stage of development before it elongates to reach the junction between the renal cortex and renal medulla (Nakai et al., 2003). An analysis of segment-specific markers(Slc12a1, Barttin) showed that the arrest of loop of Henle elongation in the nephron primordium was not due to defects in loop of Henle specification (data not shown); rather, we observed a large and statistically significant reduction (84%) in cell proliferation in this structure in Wnt7b mutants (Fig. 5). Thus, Wnt7b activity is essential for coordinated growth of the loop of Henle, thereby establishing an appropriate medullary organization for the nephron.

To understand the molecular mechanisms by which Wnt7b regulates renal medulla formation, we examined the medullary region for markers of Wnt signaling. Lef1 and Axin2, two Wnt pathway components, are also common targets of canonical Wnt signaling. Interestingly, both genes displayed Wnt7b-dependent expression in medullary interstitial cells adjacent to the Wnt7b-expressing collecting duct epithelium(Fig. 6B), whereas their expression in other renal populations was unaltered(Fig. 6B; data not shown). These data suggest that at least one action of Wnt7b is to engage interstitial mesenchymal cells through a paracrine, canonical Wnt signaling pathway, a conclusion strengthened by the co-expression of Lef1 with a BAT-gal canonical Wnt reporter transgene (Maretto et al., 2003) in this mesenchyme population (see Fig. S5B in the supplementary material).

To determine whether this paracrine pathway might underpin Wnt7bfunction in renal medulla formation, we used a Foxd1-Cre line(Foxd1GC; A.K. and A.P.M., unpublished) to specifically removeβ-catenin activity throughout the renal interstitium, including the sub-population of medullary interstitial cells (Ctnnb1^c/-;Foxd1GC, referred to as β-catenin interstitium mutants hereafter;see Fig. S6 in the supplementary material). Although β-catenin has a well-documented role in both canonical Wnt signaling and cell-cell adherens junctions in epithelia, in non-epithelial populations Wnt signaling appears to be its major activity. Remarkably, the removal of β-catenin from interstitial mesenchyme led to a failure in renal medulla development, and a loss of Lef1 and Axin2 expression(Fig. 6A,B). By contrast,branching growth and nephron induction in the renal cortex was relatively normal, although the cortical region appeared to be somewhat thicker by histological analysis (0.52±0.02 mm, wild type versus 0.59±0.02 mm, mutant; P<0.002; n=4; Fig. 6A), a phenotype that may reflect additional roles for Wnt-dependent signaling within a non-medullary interstitial compartment. Together, these data lend strong support to the conclusion that canonical signaling within the interstitial mesenchyme plays a central role in mediating Wnt7b action in regulating renal medulla formation.

To date, only two interstitial regulators have been specifically linked to medullary development. Pod1 (Tcf21) encodes a transcription factor expressed in the renal interstitium along with a number of other renal cell types (Quaggin et al.,1999). Pod1 knockout kidneys have no renal medulla, and Pod1 mutant cells fail to contribute to medullary interstitium in chimeras with wild-type cells (Cui et al.,2003). However, expression of Pod1 was not markedly altered in the renal interstitium of Wnt7b mutants (see Fig. S7 in the supplementary material). p57Kip2 encodes a cyclin-dependent kinase (Cdk) inhibitor implicated in Beckwith-Wiedemann syndrome, whose expression in the renal interstitial cell compartment is restricted to a subset of interstitial mesenchyme in the renal medulla. The renal medulla,although present, is markedly reduced in p57Kip2 mutants(Zhang et al., 1997). p57Kip2⁺ cells co-localize with Bat-gal reporter-expressing cells indicating that p57Kip2⁺ cells overlap with the mesenchymal canonical Wnt-target population (see Fig. S5C in the supplementary material). Interestingly, p57Kip2 expression was lost in this mesenchymal component in both Wnt7b and β-catenin interstitial cell mutants(Fig. 7), while Hoxa11, a more general marker of all renal interstitium, was still detected(Fig. 7; data not shown). Thus,p57Kip2 regulation may provide one direct link with Wnt7b function in elaboration of the renal medulla formation.

Wnt7b is expressed in the ureter epithelium and hydroureter was observed in some Wnt7b mutants at later stages, raising the possibility of a Wnt7b function in the ureter. However, we could find no evidence of a specific role of Wnt7b in ureter development. The histology of the mutant ureter appeared normal (data not shown). Furthermore,when dyes were injected into the mutant renal pelvic space at E18.5, the dyes accumulated in the bladder, demonstrating that the mutant ureter was not constricted and inserted into the bladder (see Fig. S8 in the supplementary material). In addition, when dyes were injected into the bladder, no vesico-ureteral reflux occurred in Wnt7b mutants (Fig. S8 in the supplementary material). Finally, sonic hedgehog signaling and smooth muscle differentiation were both normal (Fig. S8). Collectively, the data suggest that Wnt7b is not required for ureter differentiation. The hydroureter defect observed in a subset of Wnt7b mutants is likely to be a secondary complication of the medullary deficiencies.

Our work provides the first evidence that signaling by the ureteric epithelium establishes the cortico-medullary axis of the mammalian kidney. We identify Wnt7b as a key paracrine signaling factor in this process. Wnt7b is expressed exclusively in the non-branching collecting duct epithelium. Removal of Wnt7b activity leads to a failure of medullary development while other aspects of kidney development, notably ureteric branching and nephrogenesis, are unaffected. Canonical Wnt signaling targets are expressed in the subjacent interstitial mesenchyme, suggesting a paracrine signaling axis. Consistent with this view, expression of these targets is Wnt7b dependent. Additional support for canonical Wnt signaling in this interstitial population comes from the genetic removal of β-catenin in the renal interstitium. The resulting phenotype, an absence of the medullary area, closely resembles that caused by the loss of Wnt7bactivity.

How then does Wnt7b action regulate this key morphogenetic process? The cellular and molecular basis of renal cortico-medullary axis formation is poorly understood. One early event that accompanies renal medulla formation is a significant longitudinal elongation of the renal medullary collecting ducts (Cebrian et al.,2004). One model implicates pelvic contractile forces in this process. In this, the smooth muscle-regulated peristaltic contractions and downward pulling forces from the renal pelvis may influence the longitudinal growth of the renal medulla. In this scenario, Wnt7b would act in some way to promote pelvic contraction, and in the absence of these contractions the pelvic region enlarges, which results in the failure of renal medulla formation. Several observations argue against this model. First, no pelvic enlargement was observed in Wnt7b mutants prior to the defects in renal medulla formation at E15.5. An enlarged renal pelvis was evident at later stages; however, whereas the failure of medullary development was fully penetrant, only 30% of Wnt7b mutants exhibited an expanded pelvic region. Furthermore, the removal of calcineurin (Ppp3r1 - Mouse Genome Informatics) from the pelvic and ureteral mesenchyme, or Angiotensin type 1 receptor (Agtr1a, Agtr1b - Mouse Genome Informatics) disrupts effective peristalsis, but a renal medulla forms and is normal at birth(Chang et al., 2004; Miyazaki et al., 1998). Both mutants display defective smooth muscle. Smooth muscle formation initiates at E15.5 in the kidney and ureter (Yu et al.,2002), and peristaltic contractions of the ureter and renal pelvis at a later stage. That the smooth muscle-based activity does not initiate medullary development is evident from sonic hedgehog mutants, where no smooth muscle forms at E15.5 but renal medullary development initiates normally(Yu et al., 2002).

Our study indicates that the emergence of the medullary region correlates with an elongation of the ureteric epithelium caused by a non-random plane of cell division. In this, new cells are added predominantly to a longitudinal axis of growth, as would be expected if the epithelium extends in a cortico-medullary direction. Strikingly, the failure of medulla formation in Wnt7b mutants correlates with a re-orientation of the cleavage plane,such that cells divide predominantly along the radial axis of the collecting duct. This shift in the orientation of cell division is expected to increase the circumferential growth and decrease longitudinal growth. Consistent with this view, a dilation of collecting duct epithelium was the only cellular defect we found associated with the onset of the failure of renal medulla formation in Wnt7b mutants (E15.5), although apoptosis was elevated at a later stage (E17.5). This suggests that oriented cell division is one crucial parameter initiating reorganization of the collecting duct epithelium to elongate the tubular network and establish the renal medulla. Cell death may play a later contributory role in the failure of renal medulla formation in Wnt7b mutants, but by this stage it is less clear whether increased apoptosis is a direct consequence of a failure of a Wnt7b-dependent regulatory function or a secondary consequence of phenotype-correlated defects; for example, a potential increase in intrapelvic pressure due to a higher volume of urine flow.

Oriented cell division has recently been demonstrated during postnatal renal collecting duct elongation (Fischer et al., 2006). In that process, cells also divide predominantly along the longitudinal axis of the duct. Thus, common mechanisms are employed throughout the extended period of kidney growth and development, although the control of orientated cell division is tighter postnatally. In polycystic kidneys, this longitudinal cell division is disrupted, such that cells divide randomly. A causal link is suggested of orientated cell division with polycystic kidney disease (PKD) etiology(Fischer et al., 2006; Saburi et al., 2008). In Wnt7b mutants, orientated cell division is also disrupted; however,unlike in PKD, the orientation of cell division appears to be such that cells tend to re-orientate to the plane opposite to that of wild-type cells. This difference may reflect additional regulatory inputs operating at the early stages; for example, a radial cleavage-promoting cue whose activity is masked by a dominant Wnt7b regulatory input. Alternatively, the PKD phenotype may result from a complete failure to sense any position-orientating cues, randomizing the cleavage planes within the collecting duct epithelium. Clearly, other tissues have specific pathways controlling the polarity of epithelial divisions, although these appear to be distinct from those reported here. For example, the removal of α-catenin or integrin β1 activity in skin leads to random cell division, whereas p63 mutants selectively disrupt asymmetric cell divisions (Lechler and Fuchs, 2005).

The primary cilium is a central structure in the etiology of PKD. Mutants that lack a primary cilium or primary cilium-associated proteins that mediate its signaling activity exhibit PKD (Lina and Satlinb, 2004; Siroky and Guay-Woodford, 2006; Yoder,2007). Interestingly, unlike in polycystic kidneys, the primary cilium appears structurally normal in Wnt7b mutant kidneys, and is thus unlikely to play a similar role in the action of Wnt7b. Thus,differences in the structure or function of the primary cilium might be associated with the distinct defects in the orientation of cell division between PKD and Wnt7b mutants.

Tissue planar cell polarity (PCP) has been linked to the regulation of orientated cell division in the kidney and other tissue contexts(Baena-Lopez et al., 2005; Gong et al., 2004; Saburi et al., 2008). Thus, a Wnt/PCP pathway, where the relevant ligands might be regulated by Wnt7b, may be involved in the regulation of oriented cell division in renal collecting duct elongation. Although we cannot rule out an autocrine role for Wnt7b in the medullary collecting duct epithelium, our data indicate that the interstitial mesenchyme is a primary target of Wnt7b signaling, and removing the ability of these cells to respond to a canonical Wnt input replicates the Wnt7b mutant medullary phenotype. These interstitial cells lie in close proximity to the Wnt7b-secreting ureteric epithelium and express at least three Wnt ligands, Wnt5a, Wnt4 and Wnt11. The expression of each of these is either absent or downregulated in a Wnt7b/β-catenin canonical Wnt signaling-dependent manner (see Fig. S7 in the supplementary material). Two of these Wnt ligands, Wnt5a and Wnt11, are predominantly associated with non-canonical planar cell polarity signaling, whereas Wnt4 may act as both a canonical or a non-canonical ligand in a context-dependent manner. Thus, these Wnts are well placed to signal directly to the collecting duct epithelium. The medullary region develops in single mutants of Wnt5a and Wnt11. In Wnt4 mutants, any medullary role would be obscured by the requirement for Wnt4 in renal vesicle induction: loss of Wnt4 leads to an early arrest in kidney development. Thus, compound mutants and novel genetic strategies will need to be generated to determine whether Wnt7b initiates a reciprocal Wnt/PCP signaling pathway from subjacent mesenchyme to stimulate longitudinal cell division in the overlying collecting duct epithelium. Furthermore,alternative models such as a non-PCP pathway of regulation or a more complex interplay between indirect mesenchymal and direct epithelial Wnt7bsignaling cannot be ruled out. Genetic, cellular and biochemical strategies will be required to unravel this novel, critical regulatory interaction in organ biogenesis.

Pod1, p57Kip2 and integrin α3 (Itga3) are three factors that have previously been shown to be involved in renal medulla morphogenesis. Pod1 knockout kidneys have no renal medulla, whereas p57Kip2 and Itga3 knockout kidneys have a reduced renal medulla (Kreidberg et al.,1996; Quaggin et al.,1999; Zhang et al.,1997). Of these, only p57Kip2 expression is lost in Wnt7b and β-catenin interstitium mutants(Fig. 7; see also Fig. S7 in the supplementary material), suggesting that p57Kip2 may be a specific downstream target of Wnt7b signaling, whereas Pod1and Itga3 may act in a parallel pathway or upstream of Wnt7b. Although p57Kip2 is mainly known as a Cdk inhibitor,the kidney defects of p57Kip2 null mice do not appear to directly associate with changes in cell proliferation(Zhang et al., 1997). Furthermore, several reports indicate Cdk inhibitor-independent activities for p57Kip2 (Chang et al.,2003; Joseph et al.,2003; Yokoo et al.,2003). Importantly, a small renal medulla forms in p57Kip2 mutants. This phenotype is less severe than that of Wnt7b mutants,indicating that there are likely to be additional p57Kip2-independent targets of Wnt7b signaling.

Interestingly, the organizing function of Wnt7b also extends beyond the collecting duct epithelium to the nephron itself. Elongated growth of the loop of Henle is Wnt7b dependent; in the absence of Wnt7b action, little proliferative expansion of the loop of Henle anlage is observed. These data provide evidence for an ongoing role for ureteric epithelial signaling in nephron development downstream of Wnt9b-mediated induction of the nephron precursor(Carroll et al., 2005). In the loop of Henle anlagen, cell proliferation appears to be the crucial cellular response to Wnt7b functions. The different responses of loop of Henle and prospective medullary collecting duct epithelium to Wnt7bactivities may result from tissue context-specific effects of a common signal downstream of Wnt7b, or from distinct downstream signals acting on the two epithelial tissues.

In summary, our data point to an integrated control of distinct epithelial networks through diverse cellular processes to generate a functional medullary compartment. In this, Wnt7b plays a pivotal role. Its activity is essential for orientating cell cleavage in the collecting duct epithelium and for normal mitogenic activity in the loop of Henle. Furthermore, the data indicate that some, and possibly all, of these actions may be mediated through a hitherto neglected population of cells in our understanding of kidney development, the interstitial mesenchyme, highlighting the importance of this population in governing the spatiotemporal development of a vital component of functional kidney architecture, the renal medulla. Our findings provide a starting point to understanding how axial polarity in the mammalian kidney contributes to the establishment of a crucial axis of renal structure and function.