Formation of kidney tissue requires the generation of kidney precursor cells and their subsequent differentiation into nephrons, the functional filtration unit of the kidney. Here we report that the gene odd-skipped related 1 (Odd1) plays an important role in both these processes. Odd1 is the earliest known marker of the intermediate mesoderm, the precursor to all kidney tissue. It is localized to mesenchymal precursors within the mesonephric and metanephric kidney and is subsequently downregulated upon tubule differentiation. Mice lacking Odd1 do not form metanephric mesenchyme, and do not express several other factors required for metanephric kidney formation, including Eya1, Six2, Pax2, Sall1and Gdnf. In transient ectopic expression experiments in the chick embryo, Odd1 can promote expression of the mesonephric precursor markers Pax2 and Lim1. Finally, persistent expression of Odd1 in chick mesonephric precursor cells inhibits differentiation of these precursors into kidney tubules. These data indicate that Odd1plays an important role in establishing kidney precursor cells, and in regulating their differentiation into kidney tubular tissue.
Vertebrate kidney tissue is derived from the intermediate mesoderm (IM), a strip of tissue located adjacent to the somites in the developing embryo(Saxen, 1987). In amniotes,the IM gives rise to three types of kidney tissue, called (from anterior to posterior) the pronephros (a transient embryonic structure), the mesonephros(the functional embryonic kidney, which also contributes to the male reproductive system) and the metanephros (the definitive adult kidney). While all kidney types are comprised of the same fundamental functional unit (the nephron), the size and organization of these kidney types is very different:the mouse metanephros can contain up to 11,000 nephrons(Yuan et al., 2002), while the mesonephros has merely twelve (Sainio,2003).
Kidney formation can be conceptualized as consisting of two stages:establishment of nephrogenic mesenchyme within the IM, and differentiation of that mesenchyme into functional nephrons. Through numerous studies over many years, much has been learned concerning the second stage: differentiation of nephrogenic mesenchyme. In the mouse metanephros, which is the best studied example, kidney tissue differentiates as the result of interaction between the metanephric mesenchyme, which is derived from the most posterior region of the IM, and the nephric duct, another IM derivative that migrates into the metanephric zone from a more anterior region of the embryo(Sariola and Sainio, 1997; Schultheiss et al., 2003; Vainio and Muller, 1997). Reciprocal signaling between the metanephric mesenchyme and a derivative of the nephric duct known as the ureteric bud results in branching of the ureteric bud and condensation of metanephric mesenchyme at its tips. The condensed mesenchyme is thought to form a precursor cell population, which both maintains itself at the tips of the ureteric bud (via proliferation and/or addition from the surrounding non-condensed mesenchyme) and gives off cells that differentiate into pretubular aggregates and renal vesicles, the precursors of the kidney tubules (Cho and Dressler, 2003).
The mechanisms that regulate the earlier phase of kidney development -formation of nephrogenic mesenchyme - are less understood. Multiple genes have been identified that are expressed specifically in the undifferentiated metanephric mesenchyme; many of these are required for proper differentiation of the metanephric kidney, including Pax2(Torres et al., 1995), Wt1 (Kreidberg et al.,1993), Eya1 (Xu et al., 1999), Six1 (Xu et al., 2003), the Hox11 paralogous group(Wellik et al., 2002) and N-myc (Mycn - Mouse Genome Informatics)(Bates et al., 2000). However,with the exception of Eya1, all of the above genes are dispensable for the initial formation of the metanephric mesenchyme, and are instead required for its subsequent differentiation and/or survival.
The current study characterizes the role during kidney formation of odd-skipped related 1 (Odd1)(Osr1 - Mouse Genome Informatics), a zinc finger-containing transcription factor related to the Drosophila pair rule gene odd skipped (Odd). Odd1 is expressed before all previously described kidney regulatory genes, and its expression is confined to undifferentiated kidney precursor tissue. Loss-of-function studies in the mouse revealed a requirement for Odd1 in the establishment of the metanephric mesenchyme and in the activation of a set of the earliest known genes expressed during metanephric kidney formation. Gain-of-function studies in the chick embryo revealed that Odd1 can activate early markers of kidney precursor cells but inhibits the production of differentiated kidney tubules. On the basis of these studies, we propose that Odd1 plays an important role in the establishment and maintenance of the nephrogenic mesenchyme, the precursor population from which the kidney is derived.
MATERIALS AND METHODS
In situ hybridization
RNA probes were generated for chicken Odd1(James and Schultheiss, 2005), Pax2 (Burrill et al.,1997; Herbrand et al.,1998), Lim1 (Tsuchida et al., 1994) and mouse Pax2(Dressler et al., 1990), Lim1 (Fujii et al.,1994), Eya1 (Xu et al., 1999), Six2 (Xu et al., 1999), Gdnf(Pichel et al., 1996), Sall1 (Nishinakamura et al.,2001) and Wt1(Armstrong et al., 1993) using standard methods described previously(Schultheiss et al., 1995). Whole-mount in situ hybridization was performed as described previously(Schultheiss et al., 1995). For in situ hybridization on sections, paraformaldehyde-fixed cryostat sections were post-fixed, treated with 10 μg/ml proteinase K (Roche), and hybridized overnight with labeled RNA probe (1 μg/ml) at 70°C. The sections were then washed in 2× SSC (70°C), incubated with RNAse (1μg/ml, 2× SSC, 37°C) and washed in 0.2× SSC (65°C). After blocking in PBT (PBS with 2 mg/ml BSA, 0.1% Triton X-100), the sections were incubated in alkaline phosphatase antibody solution (Roche, 1:2000, in PBT + 10% sheep serum) overnight. Lastly, sections were washed and the color was visualized using NBT and BCIP.
Immunohistochemistry and lacZ histochemistry
Immunofluorescence microscopy on paraformaldehyde-fixed cryostat sections was performed as previously described(James and Schultheiss, 2003). The following primary antibodies were used: polyclonal guinea pig anti-Odd1[1:750 (James and Schultheiss,2005)], rabbit anti-Pax2 (BabCo, 1:250), mouse monoclonal and rabbit polyclonal anti-Gfp (1:100, Molecular Probes) and mouse monoclonal anti-Wt1 (1:100, Dako). Apoptotic cells were detected by TUNEL assay (Sigma)as per the manufacturer's instructions. Staining for lacZ expression was adapted from Lobe et al. (Lobe et al.,1999).
Odd1 knockout mice
Generation of mice carrying a targeted in-frame fusion of a lacZgene into the first coding exon of the Odd1 gene has been described(Wang et al., 2005). Embryos were harvested at the indicated times and processed for in situ hybridization,immunofluorescence or lacZ histochemistry as described above.
Plasmid and retroviral expression vectors
For electroporation studies, full-length chicken Odd1(James and Schultheiss, 2005)was cloned into the pMES expression vector(Swartz et al., 2001), which drives expression from a CMV/chicken β-actin promoter/enhancer, and which also expresses Gfp from an IRES element. For retroviral expression studies, full-length Odd1 was cloned into the ClaI site of the RCAS expression vector(Fekete and Cepko, 1993). The resulting RCAS-Odd1 vector was transfected into chick embryo fibroblasts, and retroviral particles were harvested from culture supernatant,concentrated and titered using previously described methods(Fekete and Cepko, 1993). RCAS-Gfp retroviral particles for control experiments were produced in a similar manner (RCAS-Gfp plasmid was a kind gift from Cliff Tabin).
Electroporation and infection of chicken embryos
Electroporation and culture of chicken embryos was performed as previously described (James and Schultheiss,2003; Wilm et al.,2004). Briefly, embryos were collected at Stage 5 or younger(Hamburger and Hamilton, 1951)and attached to a doughnut-shaped paper ring (P5, Fisher). The embryo was suspended in Tyrode's saline, ventral side down, above a 1 mm gauge platinum wire (positive electrode). Through a small hole in the vitelline membrane, 1μl of DNA solution (0.6 μg/μl) was injected into the space between the embryo and the membrane. A 20 μm gauge tungsten wire (negative electrode) was lowered above the embryo until it was submerged in Tyrode's and the embryos were pulsed three times for 25 ms at 12 V using an electro-square porator BTX-830 (BTX). Subsequently, embryos were incubated endoderm-side up on albumin-agar culture dishes [50% Albumin, 1.5% Glucose, 0.9% NaCl(Sundin and Eichele, 1992)] at 38°C for 24-48 hours.
Methods for infecting unincubated chicken eggs with retroviral vectors were adapted from previous work (Andacht et al.,2004; Sang, 2004). Unincubated fertilized white-leghorn chicken eggs, obtained from Hy-Line International (Elizabethtown, PA), were placed on their sides at room temperature for 2 hours before microinjection. A small hole (approximately 5 mm in diameter) was ground into the shell using a blunt needle (cat. # 08-965A Fisher), while leaving the underlying membrane intact. A drop of Tyrode's saline was placed on top of the shell membrane, which was subsequently removed using a microscalpel. In order to prevent air from entering the egg, Tyrode's was continuously added to the hole; any eggs that developed air bubbles were discarded. After locating the embryo, a pulled glass capillary (100-200 μm in diameter), Picospritzer II injector and Leitz micromanipulator were used to inject 1-2 μl of retrovirus (107-108 infectious particles/ml) through the center of the embryo into the subgerminal space. The hole was then sealed with hot glue from a Superbonder glue gun (FPC). The glue was allowed to harden for 5 minutes, and the eggs were incubated glue-side down at 38°C until they were harvested for analysis.
Quantification of mesonephros, tubule and duct size
For each embryo, two images were collected using a Zeiss Axiophot microscope with a SPOT camera: (1) 10× DIC image of the entire section;and (2) a 20× fluorescent image of the urogenital ridge stained with anti-Pax2 antibody. For standardization, the ImageJ software package was used to calculate the two-dimensional area of the notochord from the first image. The freehand drawing tool was used to trace a line that encircled the notochord borders, and the measure command was executed to calculate its area in pixels. To calculate the total area of all tubular tissue from the second image, we calculated the area of the individual Pax2-positive epithelial condensates (groups of cells containing polarized nuclei) and lumenized tubules and summed them. Duct area was determined in the same manner. Finally,to control for variability of embryo size, the size of the notochord was used to standardize the tubular tissue and duct area measurements. The graph in Fig. 7 shows the mean of all embryos analyzed and the error bar represents one standard error. Differences between the mean values were tested for statistical significance at the 0.05 level by the two-sample Student's t-test.
Odd1 is expressed specifically in kidney precursor cells
Odd1 expression in the developing nephric system was examined in chicken and mouse embryos. Examination of developing chicken embryos revealed that Odd1 is expressed in nephrogenic tissue before previously characterized kidney genes. Transcription of Odd1 is initiated in the intermediate mesoderm immediately after gastrulation at Hamburger Hamilton(HH) Stage 5 (Hamburger and Hamilton,1951) just lateral to Henson's node(Fig. 1A). Shortly thereafter,its expression pattern broadens to include the intermediate mesoderm and the medial lateral plate (Fig. 1B,E). By contrast, other markers of kidney differentiation, such as Pax2, are initially expressed later (HH9, Fig. 1C), and in a more spatially restricted pattern (Fig. 1D) (see James and Schultheiss, 2003; Mauch et al., 2000; Obara-Ishihara et al., 1999).
Cellular morphogenesis begins shortly after kidney-specific gene expression is initiated. Analysis of sections during the early stages of kidney morphogenesis revealed that Odd1 is expressed only in the mesenchymal components. This is true at several different stages during the development of the urogenital region: initially at the onset of nephric duct differentiation(Fig. 1E), and later during mesonephric tubule differentiation (Fig. 1F,G). Pax2 and Odd1 have complementary expression patterns in the mesonephros: while Pax2 is detectable in the differentiated epithelium of the nephric duct and tubules(Mauch et al., 2000), Odd1 is limited to small regions of mesenchyme adjacent to the tubules (Fig. 1G).
That Odd1 is present in kidney mesenchyme, but absent in differentiated epithelium at several stages of development, suggested that Odd1 is expressed in kidney precursors and subsequently downregulated as they undergo epithelial morphogenesis. To investigate this idea further,Odd1 expression during kidney tissue differentiation was examined at single cell resolution. Immediately before nephric duct formation, Pax2 expression is initiated in the medial part of the urogenital region(Fig. 1J). Subsequently, the nephric duct rudiment is formed, as evidenced by a subset of the Pax2-expressing cells that migrate toward the ectodermal layer(Fig. 1M). Odd1 protein is detectable only in Pax2-expressing cells that have not migrated dorsally(compare Fig. 1I,J with 1L,M),indicating that Odd1 is downregulated in cells that have begun to differentiate. Similarly, in the mesonephros Odd1 expression is detectable in Pax2-expressing nephrogenic mesenchyme, but is absent in cells that have organized into tubules (Fig. 1O,P).
Expression of Odd1 during mouse kidney development was examined in embryos heterozygous for a lacZ knock-in at the Odd1 locus(Wang et al., 2005). At embryonic day (E) 9.5, lacZ is localized to the intermediate and lateral plate mesoderm (Fig. 2A), similar to the localization of the Odd1 transcript described in chicken embryos (Fig. 1). LacZ expression is much higher in undifferentiated mesonephric mesenchyme than in the tubules of the mesonephric kidney(Fig. 2B). Starting at E10.5,the mouse metanephric kidney develops via reciprocal interaction between the ureter and metanephric mesenchyme (reviewed by Bard, 2003; Saxen, 1987). Initially, the ureteric bud forms as a thickening of the nephric duct, which bulges into the metanephric mesenchyme primordia (Fig. 2C, left). At this stage, Odd1 is expressed in the metanephric mesenchyme and is absent in the ureteric bud(Fig. 2D). Shortly thereafter,signals from the mesenchyme induce branching within the ureter(Fig. 2C middle, and lacZ localization Fig. 2E,F). Lastly, tubular condensates begin to epithelialize adjacent to the branching ureteric bud trunks (yellow circles, Fig. 2C, right). At all stages of ureteric bud branching, lacZ is detectable only in the condensing mesenchyme that surrounds the branching ureteric bud and not in the tubular tissue that differentiates from this mesenchyme(Fig. 2E-I). As in the mesonephros, Odd1 expression is downregulated in cells that have formed pre-tubular aggregates (arrow, Fig. 2H) and comma/S-shaped bodies (arrow, Fig. 2I). In summary, in both mouse and chick embryos, and during nephric duct, mesonephros and metanephros formation, Odd1 is expressed in kidney precursor cells and downregulated upon the initiation of epithelial differentiation.
Odd1 is required for metanephric mesenchyme formation
In order to determine whether Odd1 is required for development of the kidney, mice were examined that carry a targeted disruption of the Odd1 locus (Wang et al.,2005). The majority of mice that are homozygous for loss of Odd1 function die at E11.5 due to cardiac abnormalities(Wang et al., 2005). At E11.5 the ureteric bud has already invaded the condensed metanephric mesenchyme in wild-type and Odd1 heterozygous embryos, but the ureteric bud and metanephric mesenchyme condensation are completely absent in the homozygous mutant embryos (Fig. 3A,B).
In wild-type embryos, several factors required for kidney development are expressed in the metanephric mesenchyme before ureteric bud outgrowth,including: Six2 (Fig. 3C)(Xu et al., 1999); Eya1(Fig. 3E)(Kalatzis et al., 1998; Xu et al., 1999); Gdnf(Moore et al., 1996; Pichel et al., 1996; Sanchez et al., 1996); Pax2(Fig. 3H)(Dressler et al., 1990; Torres et al., 1995); and Sall1 (Nishinakamura et al.,2001). Embryos mutant for Odd1 completely lack expression of Six2,Eya1, Gdnf, Pax2 and Sall1 in the metanephric region(Fig. 3 and data not shown). Previous reports have shown that Eya1 mutants have similar defects in metanephric mesenchyme formation(Sajithlal et al., 2005). The data reported here implicate Odd1 as epistatic to Eya1 and as the earliest expressed factor known to be required specifically for metanephric mesenchyme formation.
Nephric duct and mesonephros defects in Odd1 mutant embryos
The ureteric bud forms as an outcropping of the nephric duct(Saxen, 1987). As ureteric bud defects were observed in Odd1 mutant embryos(Fig. 3B), the earlier development of the nephric duct was analyzed. At E9.5, Lim1, which is required for formation of the nephric duct(Kobayashi et al., 2005), is a specific marker for the nephric duct primordia(Fig. 4A,B). In Odd1mutant embryos (Fig. 4A,B), Lim1 is expressed, but at significantly lower levels than in heterozygous control embryos (Fig. 4C,D). Analysis of serial sections of Odd1 mutants demonstrated that fewer cells express Lim1 at E9.5 relative to the number seen in control embryos (Fig. 4B,D), and whole-mount in situ hybridization for Lim1revealed significant gaps in the Odd1 mutant nephric ducts(Fig. 4C,D). Decreased diameter of the nephric duct as assessed by Lim1 and Pax2 expression was also seen at E8.5 (data not shown), indicating that defects in the nephric duct are present from close to the initiation of nephric duct formation. Visualization of embryos stained for Pax2 at E10.5 also showed that the nephric duct does not migrate toward the cloaca and metanephric mesenchyme in Odd1 mutants (Fig. 4F) as it normally does in wild-type embryos(Fig. 4E, arrow points to cloaca).
A small number of mesonephric tubules differentiate in Odd1mutants (Fig. 4F, arrowhead). Normally, two types of mesonephric tubule form in mouse embryos: anterior tubules, which are fused to the nephric duct, and posterior tubules, which remain separate from the duct (Sainio et al., 1997). Only anterior mesonephric tubules form in Odd1 mutant mice (Fig. 4F), supporting the suggestion that anterior and posterior mesonephric tubules form via different mechanisms(Kobayashi et al., 2005; Sainio et al., 1997). In order to determine whether the mesonephric tubules in Odd1 null embryos exhibit normal patterning, immunofluorescence was used to localize the expression of Wt1 and Pax2. Normally, the duct and the distal tubule express only Pax2 (Fig. 4G), while the proximal tubule coexpresses Pax2 and Wt1 (arrow Fig. 4G). Odd1 mutant embryos exhibited normal regionalization of Pax2 and Wt1 expression within the duct and tubule (Fig. 4H),although the number and size of tubules was significantly smaller than in control embryos.
Gene expression defects precede apoptosis in Odd1mutants
In order to determine whether apoptosis was contributing to cell loss in the developing meso- and metanephros, TUNEL assays were performed on sectioned mouse embryos. At E8.5, the nephric duct of Odd1 mutant embryos, as assessed by Lim1 and Pax2 expression, is already thinner than in control embryos, while the levels of apoptosis in the mesonephric region of mutant embryos are similar to those observed in controls (data not shown). At E9.5,there is some normal apoptosis at the border of the proximal tubule and the mesonephric mesenchyme (Fig. 5A, arrowhead). Odd1 mutants exhibit significant inappropriate apoptosis in the mesenchyme surrounding the mesonephric tubules(compare Fig. 5A,B), indicating that increased programmed cell death is a probable contributor to the observed mesonephric size decreases in Odd1 mutant embryos.
Expression of the metanephric mesenchyme genes Pax2(Bouchard et al., 2002; Dressler et al., 1990),Pax8 (Bouchard et al.,2002), Eya1(Sajithlal et al., 2005) and Wt1 (Armstrong et al.,1993) is initiated by E9.5 (see also Fig. 5C,E). The levels of apoptosis in metanephric mesenchyme precursor cells at E9.5 are similarly low in Odd1 mutant and control embryos (compare Fig. 5C,D). Therefore,apoptosis begins in Odd1 mutant mice after metanephric mesenchyme gene expression is normally initiated (see absence of mesenchymal Pax2 in Fig. 5D and Eya1 in Fig. 5F). This implies that cell death cannot explain the initial metanephric mesenchyme gene expression defects observed in Odd1 mutant embryos, and that Odd1 is required autonomously for normal gene expression in the metanephric mesenchyme. Apoptosis does increase in Odd1 mutant mesenchyme by E10.5 (Fig. 5G,H), and this may contribute to the metanephric size decreases observed in Odd1mutants. Even after the levels of apoptosis increase in the metanephric mesenchyme at E10.5, there are still many mesenchymal cells in the meso/metanephric region of Odd1 mutant embryos that express nephrogenic genes, as evidenced by expression of Wt1(Fig. 5I,J) and the detection of many lacZ-positive cells exhibiting intact Odd1 promoter activity (Fig. 5K,L).
Odd1 promotes kidney precursor gene expression
As described above, Odd1 is expressed in non-epithelial kidney precursor cells and is required for expression of many metanephric mesenchyme-specific factors. To test if Odd1 could function to promote expression of kidney-specific genes, an Odd1-IRES-Gfp plasmid was electroporated into HH4 (gastrula) chick embryos, specifically targeting kidney and somite precursors. Morphologically, ectopic Odd1 often disrupted somite formation (8/16 embryos; see Fig. 6B,C,E). Ectopic Odd1 promoted expression of the kidney transcription factors Pax2(11/14 embryos; Fig. 6A,B) and Lim1 (4/5 embryos; Fig. 6D,E) in somitic cells within 24 hours of electroporation. Upregulation of Pax2 and Lim1 did not occur in all cells that were electroporated. In particular, Gfp-labeled cells present in the lateral plate did not express Pax2 (arrows in Fig. 6B,C). This is consistent with the absence of kidney gene expression in the medial part of the lateral plate where Odd1 is normally expressed (Fig. 1E,I,L), and indicates that the ability of Odd1 to promote kidney gene expression is context-dependent.
Odd1 represses kidney tubule differentiation
We have found that Odd1 is normally downregulated as intermediate mesoderm derivatives differentiate into epithelial structures (Figs 1, 2). In order to determine whether downregulation of Odd1 is required for normal kidney tubule differentiation, chicken mesonephroi were infected with RCAS retrovirus expressing Odd1, which allows for the stable ectopic expression of Odd1. Unincubated (pre-gastrula) chicken embryos were infected with the RCAS-Odd1 retrovirus and, after incubation for 4-5 days, adjacent sections were stained for Odd1 by in situ hybridization or for Pax2 by immunofluorescence. For analysis, `epithelialized tubules' were defined as Pax2-positive cells arranged around a distinct lumen, while the broader term`tubular tissue' was defined as Pax2-positive cells either in epithelialized tubules or in cellular aggregates with polarized nuclei. RCAS-Gfp-infected embryos served as controls.
Several observations emerged from these experiments. First, cells that express ectopic Odd1 did not differentiate into nephric duct or epithelialized tubules (Fig. 7D), while duct and tubules typically formed from adjacent non-Odd1-expressing cells (Fig. 7D, outline). As RCAS-Gfp efficiently infected the mesonephric duct and epithelial tubules(Fig. 7A), the lack of Odd1-infected epithelial cells cannot be due to infection artifact and probably reflects a reduced ability of Odd1-expressing cells to form duct and tubules. Second, visualization of urogenital ridges by DIC or Pax2 immunostaining showed that there was less tubular tissue overall in Odd1-infected embryos than in controls (compare Fig. 7E,F with 7B,C). The average total `tubular tissue' area per section in RCAS-Odd1-infected embryos was only 49% of that of control embryos(Fig. 7G, P<0.005). Interestingly, the size of the nephric duct epithelium was similar in control and RCAS-Odd1 infected embryos(Fig. 7H). Increased levels of TUNEL staining were not seen in the mesonephric regions of RCAS-Odd1infected embryos, indicating that the decrease in the area of tubular epithelium in RCAS-Odd1-infected embryos was not caused by apoptosis(data not shown). Together, these data indicate that ectopic expression of Odd1 can promote kidney precursor gene expression, but inhibits organization of precursors into tubular tissue and epithelialized tubules.
Odd1 is required for metanephric mesenchyme formation
The metanephric kidney develops from interactions between the ureteric bud and the metanephric mesenchyme. In the absence of Odd1, many of the genes that mark the early metanephric mesenchyme rudiment are never expressed,including Eya1, Six2 and Pax2(Fig. 3). Absence of Odd1 results in a more profound defect in the development of the metanephric mesenchyme than has been previously described in other loss-of-function studies, such as in mutations of Six1(Li et al., 2003; Xu et al., 2003), Pax2 (Torres et al.,1995), Gdnf (Moore et al., 1996; Pichel et al.,1996; Sanchez et al.,1996), Sall1(Nishinakamura et al., 2001), Wt1 (Kreidberg et al.,1993) and Wnt4(Kispert et al., 1998; Stark et al., 1994), all of which show some evidence of metanephric mesenchyme formation and expression of at least a subset of metanephric genes. Additionally, we have shown here that Odd1 is epistatic to Eya1, which has also been shown to be crucial for the formation of the metanephric mesenchyme(Sajithlal et al., 2005; Xu et al., 1999). These data implicate Odd1 as one of the earliest regulators of formation of the metanephric mesenchyme. These findings are consistent with the report that rare Odd1 null embryos surviving beyond E11.5 have no morphologically detectable metanephric kidneys (Wang et al., 2005).
There are several possible alternative explanations for the absence of metanephric mesenchyme gene expression in the Odd1 null mice. One possibility is that the defects in metanephric mesenchyme gene expression are secondary to defects in the formation of the nephric duct and its derivative,the ureteric bud, which would lead to problems in the inductive interactions between ureteric bud and metanephric mesenchyme(Saxen, 1987). However, the nephric duct is not required for initial expression of genes in the metanephric mesenchyme, as seen for example in the Gata3 knockout mouse, where Pax2-expressing metanephric mesenchyme cells are found in the metanephric zone despite failure of the nephric duct to migrate into this region of the embryo (Grote et al.,2006; Sainio,2003). Thus the effects of Odd1 on nephric duct development cannot explain the effects of Odd1 loss on early metanephric gene expression.
Another possibility is that, in the absence of Odd1, the tissue that would give rise to the metanephric mesenchyme undergoes apoptosis before the onset of metanephric mesenchyme gene expression, and that such apoptosis eliminates all metanephric mesenchyme precursor cells. Several lines of evidence point away from this explanation. While there is apoptosis in the metanephric region in Odd1 null mice, the apoptosis begins after E9.5, which is after the time that the early metanephric mesenchyme markers Pax2 and Eya1 begin to be expressed(Fig. 5). Also, abundant cells with Odd1 promoter activity are found in Odd1 null mice at E10.5, after the initiation of apoptosis(Fig. 5L), indicating that many cells that normally express Odd1 survive in the absence of Odd1. Many Wt1-expressing cells are present in the metanephric region of Odd1 null mice (Fig. 5J), further indicating that cells with nephrogenic potential exist in the absence of Odd1, even if such cells do not proceed to express other kidney markers. Metanephric mesenchyme apoptosis also occurs in mouse lines mutant for Pax2, Wt1, Six1 and Eya1(Kreidberg et al., 1993; Torres et al., 1995; Xu et al., 1999; Xu et al., 2003). The apoptosis seen in the Odd1 null metanephros after E9.5 could be due to a combination of the absence of Odd1 as well as of these other metanephric mesenchyme genes. Taken together, these data support an essential role for Odd1 in the initial formation of metanephric mesenchyme.
Defects in nephric duct and pro/mesonephric tubule formation in Odd1 null nice
By contrast to the complete absence of the metanephros, the nephric duct and some pro/mesonephric tubules do form in Odd1 null mice(Fig. 4). The nephric duct phenotype in Odd1 mutants is more severe than that seen in mutants for a number of other kidney genes, including Eya1, Six1, Wt1 and Sall-1, all of which have morphologically normal ducts(Kreidberg et al., 1993; Nishinakamura et al., 2001; Xu et al., 1999; Xu et al., 2003). However, Odd1 mutant nephric duct defects are less severe than those of double mutants for Pax2 and Pax8, which have no duct at all(Bouchard et al., 2002). Pax2 and Pax8 are expressed in the nephric duct itself,suggesting a cell-autonomous requirement for Pax2/8 activity in the nephric duct. By contrast, Odd1 is expressed in nephric duct precursors and in nephrogenic mesenchyme, but not in the nephric duct itself(Fig. 1). Thus, the nephric duct defects seen in Odd1 mutant mice can be attributed either to defects in duct precursor cells, or to defects in interactions between the duct and Odd1-expressing mesenchyme, which may be necessary for nephric duct maintenance (Obara-Ishihara et al., 1999). Wang et al.(Wang et al., 2005) have reported that the nephric duct defects are consistently more severe on the left side than on the right, suggesting an interaction of Odd1activity with laterality regulation pathways.
While the number of mesonephric tubules that form in Odd1 mutants is greatly reduced compared with wild type, it is nevertheless significant that a few anterior tubules form in Odd1 mutant mice. It has been reported that two types of mesonephric tubules develop in mice: anterior tubules that fuse with the nephric duct and posterior tubules that remain unfused with the duct (Sainio et al.,1997). In Odd1 mutants, as well as in Wt1mutants (Sainio et al., 1997),only the anterior, fused tubules differentiate. It has been suggested that the anterior, fused tubules differentiate simultaneously with the nephric duct from common precursor cells or by direct extension from the developing duct(Hiruma and Nakamura, 2003; Sainio et al., 1997). Our data indicate that Odd1 function is not required for the formation of this anterior group of pro/mesonephric tubules.
By contrast, posterior mesonephric tubules develop from a band of Pax2-expressing nephrogenic mesenchyme adjacent to the nephric duct. In Odd1 mutants, there is no evidence for mesenchymal Pax2 expression or formation of this nephrogenic mesenchyme(Fig. 4), which would explain the severe defects in posterior mesonephric tubule formation. Mutants for Eya1, Six1 and Sall1, which are required for metanephros development, have normal mesonephric tubule formation(Nishinakamura et al., 2001; Sajithlal et al., 2005; Xu et al., 1999; Xu et al., 2003). This suggests that Odd1 is required for common processes that underlie both mesonephric and metanephric tubule formation, while genes such as Eya1, Six1 and Sall1 are required specifically for formation of the metanephros.
The Odd1 mutant mesonephric tubule phenotype is less severe than that seen in Pax2 mutants, which have no mesonephric tubules(Brophy et al., 2001; Torres et al., 1995). However, Pax2, unlike Odd1, is expressed in tubules themselves and in the nephric duct (Dressler et al.,1990). Thus, the lack of all pro/mesonephric tubules in Pax2 mutants may reflect an autonomous requirement for Pax2at a later stage in tubule differentiation or in duct formation.
Odd1 may promote a nephric precursor state
Odd1 can activate the transcription of Pax2 and Lim1 when transiently expressed ectopically by electroporation in the paraxial mesoderm (Fig. 6). However, when Odd1 is stably expressed ectopically via a retroviral vector, and embryos are examined after 4 to 5 days, relatively little ectopic expression of Pax2 or Lim1 is seen. Rather, the predominant effect of long-term ectopic expression is the inhibition of tubule formation in the mesonephric region (Fig. 7). Odd1-expressing mesonephric cells express Pax2 but do not adopt an epithelial morphology. Indeed, Odd1-expressing cells are selectively excluded from epithelial tubules (Fig. 7). One way to reconcile the results of short-term versus longer-term ectopic expression of Odd1 is to postulate that Odd1 promotes the establishment of a Pax2-positive nephric precursor fate. Odd1 may be able to both produce such a state and maintain it, thereby blocking subsequent tubule differentiation. The low degree of ectopic Pax2 expression in the retrovirus-infected embryos compared with the electroporated embryos could be due to levels of expression (the RCAS retroviral vector integrates only one copy of the gene per cell, while electroporation can introduce many gene copies per cell) or to the absence of factors needed to maintain kidney gene expression in regions outside the intermediate mesoderm.
These data suggest that while Odd1 is necessary for the activation of early kidney gene expression, downregulation of Odd1 may be necessary for differentiation of nephric tissue into epithelial structures. Such a role for Odd1 is consistent with the Odd1 expression pattern, as Odd1 is expressed in all nephric precursors but is downregulated upon tubule formation (Figs 1, 2). In future work, it will be important to understand the molecular mechanisms by which Odd1 is downregulated during epithelium formation, as well as the mechanisms through which Odd1 regulates kidney gene expression and inhibits epithelial differentiation.
We gratefully acknowledge A. Bober, P. Danielian, M. Goulding, T. Jessell,J. Kreidberg, R. Maas, R. Nishinakamura and C. Tabin for sharing probes and expression plasmids, and D. Herzlinger for providing an improved lacZstaining protocol. We also thank Mozhgan Afrakhte, Iain Drummond, Doris Herzlinger and Sasha Petrova for critical reading of the manuscript and for many helpful discussions and suggestions. R.G.J. was supported by a Predoctoral Fellowship from the Howard Hughes Medical Institute, and C.N.K. is supported by a Predoctoral Fellowship award from the National Science Foundation. This work was supported by grants R01 DK59980 and R01 DK71041 from the NIH (NIDDK) to T.M.S., and R01 DE013681 (NIH/NIDCR) to R.J.