Distinct and sequential tissue-specific activities of the LIM-class homeobox gene Lim1 for tubular morphogenesis during kidney development

Development of the kidney is one of the best-studied processes of vertebrate organogenesis (Grobstein,1953; Grobstein,1956; Kuure et al.,2000; Saxen, 1987; Vainio and Lin, 2002). This developmental process includes extensive tubular morphogenesis, including mesenchyme-to-epithelium transition, tubular elongation, branching and differentiation (Dressler,2002; Hogan and Kolodziej,2002; Lubarsky and Krasnow,2003). During embryogenesis, the urogenital system derives from the intermediate mesoderm of the gastrula. Nephric ducts are paired epithelial tubes, formed in the intermediate mesoderm in an anteroposterior direction,that join the cloaca (Obara-Ishihara et al., 1999). The primary kidney, the pronephros, is formed at the anterior region of the nephric duct(Bouchard et al., 2002; Saxen, 1987). Subsequently,the secondary kidney, the mesonephros, is formed posterior to the pronephros(Sainio et al., 1997; Sainio and Raatikainen-Ahokas,1999; Saxen,1987). In amniotes (birds, reptiles, and mammals), the pronephros and mesonephros are transient embryonic kidneys that subsequently degenerate. The development of the amniote-specific definitive kidney, the metanephros, is initiated by induction of the ureteric bud at the posterior end of the nephric duct by signals from the adjacent metanephric mesenchyme that in mice occurs at approximately embryonic day 10.5 (E10.5). Subsequently, the ureteric bud elongates and branches to form the renal collecting duct system(Davies, 2001; Davies and Davey, 1999; Pohl et al., 2000; Shah et al., 2004). The metanephric mesenchyme, from which nephrons are derived, is reciprocally induced by the ureteric bud to form the condensed (induced) mesenchyme around the tip of the ureteric bud. Subsequently, a subset of cells in the condensed mesenchyme proliferates and forms the pretubular aggregate(Bard et al., 2001; Sariola, 2002; Stark et al., 1994). The renal vesicle is formed from the pretubular aggregate by a mesenchyme-to-epithelium transition and further differentiates into comma- and S-shaped bodies, and ultimately Bowman's capsule of the glomerulus and the tubules of the nephron(Saxen, 1987). How the nephron precursors are patterned to generate distinct cell types of the different physiological compartments of the nephron is not well understood.

To understand tissue-specific gene function in kidney organogenesis, in vitro tissue recombination assays have been used by culturing the ureteric bud and metanephric mesenchyme from wild-type and mutant animals(Grobstein, 1953; Grobstein, 1956; Miyamoto et al., 1997). However, this technique is limited if mutants die before metanephros induction occurs or fail to form the ureteric bud or metanephric mesenchyme. Moreover,some cell types in the metanephros (e.g. endothelial cells and neurons) are not generated in the metanephros cultured in vitro(Vainio and Lin, 2002). Recently, conditional knockout strategies including the Cre-loxPrecombination system in mice have been developed, which enables the study of tissue-specific gene function in vivo, avoiding embryonic lethality and/or early defects in precursor tissues (Kwan,2002; Nagy,2000).

The nephric duct also differentiates into the Wolffian duct, the primordium of the male reproductive tract that includes the epididymis, vas deferens and seminal vesicle. Subsequently, the Müllerian duct, the primordium of the oviduct, uterus and vagina of the female reproductive tract, is formed adjacent to the Wolffian duct in both male and female embryos before sexual dimorphic differentiation occurs(Kobayashi and Behringer,2003). Thus, the formation of the genital ducts of the reproductive system is linked with the development of the excretory system.

The LIM-class homeodomain transcription factor, Lim1 (Lhx1 – Mouse Genome Informatics), is expressed in a dynamic pattern throughout urinary system development (Barnes et al.,1994; Fujii et al.,1994; Karavanov et al.,1998; Kobayashi et al.,2004; Kume et al.,2000; Tsang et al.,2000). In the mouse, Lim1 is expressed in the intermediate mesoderm at E7.5 and the nephric duct and mesonephric tubules by E10.5. Lim1 is also expressed in the metanephros. Lim1 is expressed in the tip of the ureteric bud in the cortical region. In metanephric mesenchyme-derived tissues, Lim1 expression is observed in the pretubular aggregate, the comma- and S-shaped bodies, and the podocyte of the immature glomerulus, and diminishes in the mature glomerulus(Karavanov et al., 1998; Sariola, 2002). Although almost all Lim1-null mutants die around E10.0 because of failure of chorioallantoic fusion for placenta formation, a few rare Lim1-null neonates are born and lack the metanephros(Shawlot and Behringer, 1995). This agenesis of the metanephros results from the failure of proper formation of the nephric duct (Tsang et al.,2000), which secondarily causes the lack of metanephros induction. Because there is both embryonic lethality and failure of proper formation of the nephric duct in Lim1-null mutants prior to the initiation of metanephros development, Lim1 function in development of the mesonephros and metanephros remains unclear.

Here, we have examined Lim1 function in mesonephros and metanephros development by avoiding the early defects of Lim1-null mutants, using BAC transgene rescue, tissue-specific knockout and chimera analysis. Our results demonstrate sequential, dosage-sensitive and distinct tissue-specific roles for Lim1 in tubular morphogenesis during kidney organogenesis.

Rarb2-Cre was generated using a 4.8 kb DNA fragment from the 5′ promoter region of the Rarb2 gene(Mendelsohn et al., 1991)fused with Cre-Mt1pA (Sauer and Henderson, 1990). BAC and Rarb2-Cre transgenic mice were generated by pronuclear injection of C57BL/6 (B6)×SJL F2 zygotes(Nagy et al., 2003). The 7.6 kb KpnI-NotI Rarb2-Cre-Mt1pA fragment was isolated and injected at 2 ng/μl. BAC DNA was linearized with PI-SceI, and injected at 0.2 ng/μl.

Lim1^lz/+ (Kania et al., 2000) and Lim1^flox/flox(Kwan and Behringer, 2002)mice were maintained on a B6×129/SvEv mixed genetic background, Lim^neo/+ (Shawlot and Behringer, 1995) on a B6 congenic background, Hoxb7-Cre^tg/+ (Yu et al., 2002) on a B6 inbred genetic background, and R26R(Soriano, 1999) and Wnt4^+/– (Stark et al., 1994) on a B6 x 129/Sv mixed background.

The following primers were used for genotyping mice by PCR. For BAC-rescued Lim1 mutant mice: mLim1-Fw8, GGCTACCTAAGCAACAACTACA; PGK-FX3,AGACTGCCTTGGGAAAAGCGC; BAC-Fw5, GTAAAACAAGCCACAGTTCTGAC; BAC-Rv6,TATTCAGCTACTGTTCCGTCAGC; Rap-A, AGGACTGGGTGGCTTCCAACTCCCAGACAC; Rap-B,AGCTTCTCATTGCTGCGCGCCAGGTTCAGG; lacZ-A, GCATCGAGCTGGGTAATAAGGGTTGGCAAT; and lacZ-B, GACACCAGACCAACTGGTAATGGTAGCGAC (Lim1neo, ∼230 bp; BAC, 281 bp;Rap, 590 bp; lacZ, 822 bp). For Hoxb7-Cre^tg/+ mice:mHoxb7-Fw1, TGGGCCGGGGTCACGTGGTCAGA; Cre-Rv2, CGACGATGAAGCATGTTTAGCTG(∼500 bp). For Rarb2-Cre^tg/+ mice: Cre-Fw6,GAAACAGGGGCAATGGTGCGCCTGCTG; mMt1-Rv1, AGGAAGACGCTGGGTTGGTCCGATACT (∼1.1 kb). Lim1^lz/+, Lim1^neo/+, Lim1^flox/flox and Wnt4^+/– mice were genotyped as described previously(Kobayashi et al., 2004; Kwan and Behringer, 2002; Stark et al., 1994).

X-gal staining was performed as described(Nagy et al., 2003). For histological analysis, paraffin embedded tissues were sectioned at 7 μm and counterstained with 0.33% eosin Y.

Whole-mount in situ hybridization was performed as described(Shawlot and Behringer,1995).

Lim1^+/+; R26^tg/+ ↔ wild-type and Lim1^–/–; R26^tg/+ ↔ wild-type chimeric mice were generated as previously described(Shawlot et al., 1999).

Lim1 is expressed in multiple tubule-forming tissues during kidney organogenesis and the genital ducts of the reproductive system. We found that Lim1-lacZ knock-in (Lim1^lz/+) mice(Kania et al., 2000)recapitulated the complex endogenous Lim1 expression patterns during kidney development (see Fig. S1 in the supplementary material). Briefly, at E7.5 and E8.5, Lim1-lacZ expression was detected in the lateral mesoderm that comprises the intermediate mesoderm and lateral plate, and subsequently this expression was restricted to the nephric duct epithelium that forms within the intermediate mesoderm by E9.5. At E10.5, Lim1-lacZ expression was observed in the epithelial mesonephric vesicles. At this stage, Lim1-lacZ is weakly expressed in the epithelium of the nephric duct, but upregulated in the ureteric bud. In the metanephros, Lim1-lacZ is expressed weakly in the growing ureteric bud in the cortical region of the metanephros but strongly in the pretubular aggregate. Lim1-lacZ expression persists in the pretubular aggregate-derivatives, including the comma- and S-shaped bodies, and Bowman's capsule of the immature glomerulus, but is downregulated in the mature glomerulus. Lim1-lacZ expression is also detected in the epithelium of the Wolffian (mesonephric) and Müllerian (paramesonephric) ducts in both sexes.

Although most Lim1-null mutants die during embryogenesis around E10, there are rare homozygous mutants that can survive to birth, which appears to be influenced by genetic background(Kobayashi et al., 2004). Two male and four female Lim1-null neonates were generated and all lacked the anterior head, metanephros and ureter(Fig. 1D,H). Lim1-null males had normal appearing testes but lacked the male reproductive tract, and female Lim1-null mutants had normal ovaries but lacked the female reproductive tract (Fig. 1H and data not shown) (Kobayashi et al.,2004).

To understand Lim1 function in urogenital system development, we rescued the embryonic lethality of Lim1-null mutants using a mouse Lim1-containing bacterial artificial chromosome (BAC) transgene. We identified a ∼208 kb C57BL/6J strain BAC clone (RP23-123A17) with the Lim1 locus centrally located (Fig. 1A). Five Lim1 BAC transgenic founders were generated and bred with Lim1 heterozygous mice of two different null alleles, Lim1^neo/+ and Lim1^lz/+, to place the BAC transgenes on a Lim1-null background(Fig. 1B). BAC-rescued Lim1 mutants (Lim1^lz/neo; BAC^tg/+)were readily recovered at birth using two independent BAC transgenic lines(lines 5 and 7) (Fig. 1C).

We found that all BAC-rescued Lim1 mutants died within 1 day of birth for both transgenic lines, and displayed hypomorphic phenotypes in tissues requiring Lim1 for their development, including the head,kidney and Müllerian duct derivatives(Fig. 1E,F,I,J). For line 7,some BAC-rescued Lim1 mutants (21.7%, n=23) had craniofacial abnormalities, including a shortened snout, eye defects and loss of the lower jaw (Fig. 1F and data not shown). The female BAC-rescued mutants also frequently (46.2%, n=13)lacked the posterior part of the uterus(Fig. 1J). All of the BAC-rescued mutants for line 7 (n=23) had metanephroi that were smaller in comparison with wild-type controls(Fig. 1J,K). For line 5, the phenotypes of BAC-rescued Lim1 mutants were more severe than line 7. All of the BAC-rescued Lim1 mutants for line 5 (n=4) lacked most of the anterior head, which was smaller than those of rescued mutants for line 7 (Fig. 1E-G). These BAC-rescued Lim1 mutants had two separate ears(Fig. 1E), whereas Lim1-null neonates usually have a single fused ear at the midline(Fig. 1D). All of the BAC-rescued Lim1 mutants for line 5 lacked most of the uterus in females (n=3) (Fig. 1I) and had smaller metanephroi in both sexes (n=4) than those of BAC-rescued Lim1 mutants for line 7 or wild type(Fig. 1I-K).

All of the BAC-rescued Lim1 mutants for both transgenic lines 5 and 7 (n=27) had a shrunken bladder without urine (data not shown),suggesting that the smaller metanephroi were not functional. Histological analysis of the metanephroi of BAC-rescued Lim1 mutants suggested retarded development of the ureteric bud and abnormalities of the medulla(Fig. 1L-N), although collecting ducts (the ducts of Bellini) were present(Fig. 1O-Q). The metanephros of BAC-rescued Lim1 mutants also completely lacked nephrons, including glomeruli and associated tubules of the nephron as well as comma- and S-shaped bodies (Fig. 1O-Q).

Using line 7, we examined Lim1 expression in BAC-rescued Lim1 mutants. The level of BAC-derived Lim1 expression was very low in all organs examined, including the metanephros, Müllerian duct and neural tube (Fig. 1R-Y, data not shown).

Lim1 is expressed in both ureteric bud- or metanephric mesenchyme-derived tissues and both ureteric bud- or metanephric mesenchyme-derived tissues have defects in BAC-rescued Lim1 mutants. Because there are reciprocal interactions between these tissues, it is not clear if Lim1 activity is required in the ureteric bud- or metanephric mesenchyme-derived tissues, or both. Therefore, we inactivated Lim1 function individually in two distinct tissues of the metanephros using the Cre-loxP recombination system in vivo to bypass the nephric duct defect and early lethality of Lim1-null mutants.

For Lim1 inactivation in nephric duct- and ureteric bud-derived tissues, we used Hoxb7-Cre transgenic mice(Yu et al., 2002). To inactivate Lim1 in the nephric mesenchyme, including the metanephric mesenchyme, we generated a new Cre transgenic mouse line using a retinoic acid receptor β isoform 2 (Rarb2) gene promoter(Mendelsohn et al., 1991). To examine Cre recombinase activity in these transgenic mice, we crossed them with R26R Cre reporter mice (Soriano,1999). Cre reporter expression was detected in the urogenital system from E9.5 in both Hoxb7- and Rarb2-Cre mice(Fig. 2A,B). For Hoxb7-Cre, reporter expression was detected in the entire nephric duct epithelium, but not in the nephric mesenchyme, at E9.5(Fig. 2A,C). At the same stage,for Rarb2-Cre, reporter expression was restricted to the nephric mesenchyme in the anterior region of the mesonephros, but not in the nephric duct epithelium (Fig. 2B,D). Reporter expression for Rarb2-Cre was also detected in the anterior central nervous system (Fig. 2B; data not shown).

At E10.5, Hoxb7-Cre reporter expression remains restricted to the nephric duct epithelium, including the ureteric bud(Fig. 2E,G,I,K). At the same stage, Rarb2-Cre reporter expression extended posteriorly along the entire nephric duct mesenchyme, but not in the ureteric bud except for a few rare cells (Fig. 2F,H,J,L; data not shown). It has been proposed that cranial mesonephric tubules are derived from the nephric duct and caudal mesonephric tubules are derived from the nephric mesenchyme (Sainio et al.,1997). Consistent with this, Hoxb7-Cre and Rarb2-Cre reporter expression were also detected in the cranial and caudal mesonephros, respectively (Fig. 2G,H). Rarb2-Cre reporter expression was also observed in the limb mesenchyme and neuroretina of the eye(Fig. 2F; data not shown).

At E12.5 and E14.5, Hoxb7-Cre reporter expression remained in the nephric duct epithelium derivatives such as the epithelium of the Wolffian duct, ureter and ureteric bud (Fig. 2M,O,Q,S,U,W). At the same stages, Rarb2-Cre reporter expression remains restricted to the metanephric mesenchyme derivatives,including the condensed mesenchyme, cortical stroma, pretubular aggregate,comma- and S-shaped bodies, Bowman's capsule, proximal and distal tubules of the nephron and mesenchyme of the ureter and reproductive tracts(Fig. 2P,R,V,X). Rarb2-Cre reporter expression was also detected in the gubernaculum(Fig. 2N,T). No Cre reporter expression was detected in medullary (inner) stromal cells and glomerular capillary cells that are surrounded by Bowman's capsule, which include endothelial and mesangial cells, in either Hoxb7-Cre or Rarb2-Cre mice (Fig. 2U-X).

Using the Hoxb7- and Rarb2-Cre mice with a Lim1conditional null (flox) allele, we inactivated Lim1 function in nephric duct epithelium- or nephric mesenchyme-derived tissues during kidney development, respectively. In Hoxb7-Cre^tg/+;Lim1^lz/flox mutants, we found that Lim1 expression in the ureteric bud was absent but Lim1 expression in the pretubular aggregate and its derivative tissues was detected(Fig. 3B,E). In Rarb2-Cre^tg/+; Lim1^lz/flox mutants, Lim1 expression in the pretubular aggregate-derived tissues was not detected but Lim1 expression in the ureteric bud was intact(Fig. 3C,F).

When Lim1 was inactivated specifically in nephric duct-derived tissues using Hoxb7-Cre, the Hoxb7-Cre^tg/+;Lim1^lz/flox mutants were viable and could survive to adulthood. However, when Lim1 in metanephric mesenchyme-derived tissues was specifically inactivated by Rarb2-Cre, all of the Rarb2-Cre^tg/+; Lim1^lz/flox mutants died within the first day of birth. All other mice with genotypes including Lim1^lz/flox, Hoxb7-Cre^tg/+, Rarb2-Cre^tg/+, Hoxb7-Cre^tg/+;Lim1^flox/+ and Rarb2-Cre^tg/+;Lim1^flox/+ were phenotypically identical to wild type.

At birth, the Hoxb7-Cre^tg/+; Lim1^lz/floxmutant neonates had small metanephroi (Fig. 3H,N). Urine was observed in the bladder (data not shown),suggesting that the smaller metanephroi were functional. In some Hoxb7-Cre^tg/+; Lim1^lz/flox mutants (40.0%, n=15), hydronephrosis and hydroureter were observed(Fig. 3N), suggesting urine production but abnormal development of the distal ureter, which results in obstruction of urinary excretion into the bladder(Batourina et al., 2002). Serial sections of these mice with hydronephrosis and hydroureter revealed that the distal ureter was closed in both sexes or ended abnormally into the uterus in some females (data not shown). These nephric duct-specific Lim1 mutants also had abnormal reproductive tracts. In all Hoxb7-Cre^tg/+; Lim1^lz/flox mutant males(n=8), most parts of the male reproductive tract such as the epididymis and vas deferens were absent, except for some residual tissue(Fig. 3K). In some Hoxb7-Cre^tg/+; Lim1^lz/flox mutant females(57.1%, n=7), uteri were completely or partially absent with residual uterine tissue discontinuously present(Fig. 3Q and data not shown). The posterior uterus was more frequently absent compared with its anterior region (see Fig. S2 in the supplementary material; data not shown).

Rarb2-Cre^tg/+; Lim1^lz/flox mutant neonates also had small metanephroi with normal ureters (n=10)(Fig. 3I,O). However, these small metanephroi were slightly bigger than those of Hoxb7-Cre^tg/+; Lim1^lz/flox mutants(Fig. 3H,N). In addition, Rarb2-Cre^tg/+; Lim1^lz/flox mutants had a shrunken bladder with no urine (data not shown). Rarb2-Cre^tg/+;Lim1^lz/flox mice had normal reproductive tracts both in males and females (Fig. 3L,R).

We found that the nephric duct was initially formed normally in Hoxb7-Cre^tg/+; Lim1^lz/flox mice(Fig. 4C,F), which allowed induction of the metanephros by the ureteric bud and subsequent formation of the ureter (Fig. 3H,N). However, after formation of the nephric duct, Lim1 expression in the nephric duct epithelium is inactivated by Hoxb7-Cre from E9.5(Fig. 2A). The Lim1^lz reporter was used to follow Lim1-null cells in Hoxb7-Cre^tg/+; Lim1^lz/flox mice. At E11.5, Lim1-lacZ expression in the nephric duct epithelium was discontinuous (Fig. 4H,J). Histological analysis of Hoxb7-Cre^tg/+;Lim1^lz/flox mice revealed that the epithelial tissues of the nephric duct were not present where lacZ expression was absent (data not shown). This degeneration of the nephric duct epithelium results in the absence of most of the male reproductive tract such as the epididymis and vas deferens (Fig. 3K, Fig. 4R; see Fig. S2 in the supplementary material). The Lim1-lacZ reporter also revealed a loss of caudal mesonephric tubules, which are derived from the nephric mesenchyme(Fig. 2H), although the cranial mesonephros, which is derived from the nephric duct(Fig. 2G), was present(Fig. 4J).

Lim1 expression is inactivated only in the nephric (Wolffian) duct epithelium, but not in the Müllerian duct of Hoxb7-Cre^tg/+; Lim1^lz/flox mice(Fig. 2M,Q,S,), which we confirmed by in situ hybridization (data not shown). However, we observed Müllerian aplasia in Hoxb7-Cre^tg/+;Lim1^lz/flox mice (Fig. 3Q; see Fig. S2 in the supplementary material). In Hoxb7-Cre^tg/+; Lim1^lz/flox mice, initiation of Müllerian duct formation was normal at E11.5(Fig. 4H,J). However, posterior elongation of the Müllerian duct was impaired(Fig. 4L). Müllerian duct elongation was blocked where there was a discontinuity in the Wolffian duct(Fig. 4N). Interestingly the most anterior gaps of the Wolffian duct did not affect initial formation of the Müllerian duct (Fig. 4N, white arrowheads). After the Müllerian duct reaches the cloaca, the Wolffian duct further degenerates. We found that the Müllerian duct also starts to degenerate adjacent to where the Wolffian duct is lost (Fig. 4P).

The metanephros of Hoxb7-Cre^tg/+;Lim1^lz/flox mice had morphologically normal medulla and glomeruli (Fig. 5B,F). However,the metanephroi of Hoxb7-Cre^tg/+; Lim1^lz/floxmice was hypoplastic with reduced number of glomeruli(Fig. 3H,N; Fig. 5B,F; data not shown).

To understand Lim1 function in the nephric duct epithelium derivatives, we examined Lim1-lacZ expression in the metanephros of Hoxb7-Cre^tg/+; Lim1^lz/flox neonates. In neonates, Lim1-lacZ staining continues to show strong expression in the developing nephrons and weak expression in the ureteric bud(Fig. 5I,M). We found that the number of developing nephrons was greatly reduced at birth(Fig. 5J). Furthermore, the distribution of nephron progenitors is disorganized and there are scattered regions lacking nephron progenitors (Fig. 5J). In the metanephros of Hoxb7-Cre^tg/+;Lim1^lz/flox embryos, nephrons are found at every ureteric bud branching tip (Fig. 5R).

By tail somite (ts) stage 6, Lim-lacZ expression revealed that the nephric duct reaches the posterior region of the urogenital system adjacent to the metanephric mesenchyme in both control and Hoxb7-Cre^tg/+;Lim1^lz/flox embryos (Fig. 6A,B), which is consistent with the previous observation of normal formation of the nephric duct in Hoxb7-Cre^tg/+;Lim1^lz/flox embryos initially(Fig. 4F). However, induction of the ureteric bud is delayed in Hoxb7-Cre^tg/+;Lim1^lz/flox embryos (Fig. 6B), although Lim1-lacZ expression is upregulated for ureteric bud induction in the posterior part of the nephric duct(Fig. 6B).

Several molecules have been shown to be required for ureteric bud development (Carroll and McMahon,2003; Lechner and Dressler,1997; Pohl et al.,2000; Shah et al.,2004). Ret is required for formation, growth and branching of the ureteric bud (Schuchardt et al., 1994). It is weakly expressed in the nephric duct and its expression is upregulated in the ureteric bud tip(Fig. 6C). However, in Hoxb7-Cre^tg/+; Lim1^lz/flox embryos, the upregulation of Ret at the posterior region of the nephric duct was not as robust as their control littermates(Fig. 6D). Glial cell line-derived neurotrophic factor (GDNF) is a ligand for Ret expressed in the metanephric mesenchyme and is also required for ureteric bud formation, growth and branching (Moore et al.,1996; Pichel et al.,1996; Sanchez et al.,1996). We found that Gdnf is normally expressed in the metanephric mesenchyme of Hoxb7-Cre^tg/+;Lim1^lz/flox embryos at ts9(Fig. 6G,H). Other molecular markers for the metanephric mesenchyme, such as Eya1 and Pax2, were also normally detected in Hoxb7-Cre^tg/+;Lim1^lz/flox embryos at the same stage(Fig. 6I,J; data not shown). Wnt11 is expressed in the tip of the ureteric bud(Kispert et al., 1996) and its function is required for proper ureteric branching interacting with Ret(Majumdar et al., 2003). At ts9, Wnt11 expression is detected in the swollen-shaped tip of the ureteric bud (Fig. 6E). Although the pattern of Wnt11 expression appears thinner, its expression is correctly upregulated in the posterior nephric duct of Hoxb7-Cre^tg/+; Lim1^lz/flox embryos(Fig. 6F).

At ts18, strong expression of Ret persists in wild-type ureteric bud tips (Fig. 6K). This Ret expression was not observed in the retarded ureteric bud of Hoxb7-Cre^tg/+; Lim1^lz/flox mutants(Fig. 6L). At the same stage, Wnt11 expression indicated that the ureteric bud tips persist in Hoxb7-Cre^tg/+; Lim1^lz/flox embryos(Fig. 6O,P). However, Lim1-lacZ and Wnt11 expression patterns revealed a shorter distance between two smaller ureteric bud tips(Fig. 6M-P), suggesting that ureteric bud growth is retarded in the metanephros of Hoxb7-Cre^tg/+; Lim1^lz/flox embryos. The metanephric mesenchyme marker Pax2 and the stroma marker Bf2(also known as Foxd1) (Hatini et al., 1996) are normally expressed in both wild-type and Hoxb7-Cre^tg/+; Lim1^lz/flox mutants at this stage (data not shown).

Histological analysis of the metanephroi of Rarb2-Cre^tg/+;Lim1^lz/flox neonates revealed that the medulla was not correctly formed and there were no nephron tissues such as glomeruli and their associated tubules (Fig. 5C,G). However, Lim1-lacZ expression in the metanephros of Rarb2-Cre^tg/+; Lim1^lz/flox embryos at E12.5 showed the presence of nephron progenitors(Fig. 5S).

Although the number of developing pretubular aggregates was slightly reduced in the smaller metanephroi of Rarb2-Cre^tg/+;Lim1^lz/flox mutants, we found that the distribution pattern of the nephron progenitors was normal (Fig. 5K), in contrast to the abnormal pattern found in the metanephroi of Hoxb7-Cre^tg/+; Lim1^lz/flox mice(Fig. 5J). Histological analysis revealed that nephron development is arrested at the stage of the renal vesicle, which are Lim1-lacZ positive(Fig. 5O). Apparently, these Lim1-lacZ positive tissues subsequently degenerate(Fig. 5O, white arrowheads).

It has previously been shown that Wnt4-null mutants also lack nephrons (Stark et al., 1994). Because Lim1 and Wnt4 are co-expressed in the pretubular aggregate, it has been proposed that these genes may be involved in the same pathway for nephron development (Sariola,2002). Therefore, we examined this possibility by comparing Rarb2-Cre^tg/+; Lim1^lz/flox mutant mice with Wnt4-null mice. We found that the metanephroi of Wnt4-null mice were severely hypoplastic compared with those of Rarb2-Cre^tg/+; Lim1^lz/flox mutant mice(Fig. 5C,D). Surprisingly, we also found that the small metanephroi of Wnt4-null mice can form a few glomeruli, nephron tubules and medulla-like structures(Fig. 5D,H), which are completely absent in Rarb2-Cre^tg/+; Lim1^lz/floxmutant mice.

We found that Wnt4 expression in the pretubular aggregate(Fig. 7C; data not shown) is broader than Lim1 expression, which is restricted only in the presumptive renal vesicle (see Fig. S1 in the supplementary material). Although Lim1-lacZ expression in nephron progenitors on a Wnt4-null background was absent at E12.5(Fig. 5T), we observed a few Lim1-lacZ-positive developing nephrons in Wnt4-null mutants later at E14.5 (Fig. 5L,P).

During nephron development, Wnt4, Pax8 and Fgf8expression are initiated in the pretubular aggregate and persist in the renal vesicles (Fig. 7A,C; data not shown) (Stark et al., 1994). In the Rarb2-Cre^tg/+; Lim1^lz/flox mutants, the expression patterns of Wnt4, Pax8 and Fgf8 are correctly induced and persist in the renal vesicle(Fig. 7B,D; data not shown). Dll1 expression is initially expressed very weakly in the renal vesicle and Dll1 expression is upregulated in a subset of cells in the renal vesicle during its maturation and persists in the presumptive proximal tubule of the S-shaped body (Fig. 7E,G; data not shown)(Leimeister et al., 2003). In the Rarb2-Cre^tg/+; Lim1^lz/flox mutants, the upregulation of Dll1 expression was not observed(Fig. 7F,H). Expression of Brn1 (also known as Pou3f3) is initiated in a subset of cells in the renal vesicle proximal to the ureteric bud and its function is required for the loop of Henle and distal tubule(Fig. 7I; data not shown)(Nakai et al., 2003). In the renal vesicles of the Rarb2-Cre^tg/+;Lim1^lz/flox mutants, the polarized Brn1 expression was not observed (Fig. 7J). Expression patterns of a ureteric bud marker Wnt9b(Fig. 7K,L) and a stroma marker Bf2 (Fig. 7M,N) did not change in the Rarb2-Cre^tg/+; Lim1^lz/floxmutants.

To obtain more information about Lim1 function during kidney development, we performed a chimera analysis(Tam and Rossant, 2003). To generate chimeric mice, we injected Lim1^–/–;Rosa26^tg/+ embryonic stem (ES) cells into wild-type blastocysts (Shawlot et al.,1999). Subsequently, we examined the distribution of Lim1-null cells by lacZ expression in the chimeric embryos. In control experiments, Lim1^+/+; Rosa26^tg/+ ES cells were injected into wild-type blastocysts. Chimeric kidneys from 57 embryos were analyzed. In control chimeras, Lim1^+/+ cells could extensively contribute to all cell types in the urogenital systems(Fig. 8A,C,E,G,I,L,O,Q).

We found that Lim1^–/– cells could not contribute to the epithelium of the nephric duct in chimeric mice at E9.5(Fig. 8B). By contrast, Lim1^–/– cells could contribute to the nephric mesenchyme at the same stage (Fig. 8B). In the caudal mesonephros of chimeric mice at the E10.5, Lim1^–/– cells were excluded from the epithelial renal vesicles of the mesonephros(Fig. 8D). The renal vesicles of the caudal mesonephros are derived from the nephric mesenchyme(Fig. 2H), where Lim1^–/– cells could contribute one day before at E9.5 (Fig. 8B). At the posterior end of the nephric duct at E10.5, Lim1^–/– cells could contribute to the metanephric mesenchyme, although Lim1^–/– cells remained excluded from the ureteric bud, which is derived from the nephric duct epithelium (Fig. 8F). These findings suggest that Lim1 acts cell autonomously for epithelium development of the nephric duct by E9.5 and for renal vesicle formation in the mesonephros between E9.5 and E10.5.

At later stages of development, Lim1^–/–cells could contribute to the condensed mesenchyme and pretubular aggregate in the metanephros (Fig. 8H). Although Lim1^–/– cells could initially contribute to the renal vesicle at early stages(Fig. 8J), Lim1^–/– cells were not found in the region of the renal vesicle proximal to the ureteric bud(Fig. 8K), although Lim1 is expressed in the entire renal vesicle at this stage (see Fig. S1 in the supplementary material; data not shown)(Karavanov et al., 1998). The epithelial tissue of the renal vesicle that lacks Lim1^–/– cells invaginates to form the cleft of the comma-shaped body (Fig. 8M,N). Subsequently, Lim1^–/– cells were not present in the region opposite to the first cleft of differentiating comma-shaped bodies (Fig. 8N,white arrow), which subsequently invaginates to form the second cleft of the S-shaped body (Fig. 8P). Lim1^–/– cells could contribute to the proximal tip of the S-shaped body (Fig. 8P, dashed box). Subsequently, Lim1^–/– cells were not found in a subset of cells that faces towards the cleft within the proximal tip of the S-shaped body (Fig. 8R). This small epithelial tissue where Lim1^–/– cells could not contribute expanded extensively and formed podocytes of the glomerulus(Fig. 8S,T). Lim1^–/– cells could contribute to the glomerular capillaries and outer epithelium layer of Bowman's capsule(Fig. 8T).

The LIM-class homeobox gene Lim1 is expressed in tubule-forming tissues of the pronephros, mesonephros and metanephros, and genital ducts(Barnes et al., 1994; Fujii et al., 1994; Shawlot and Behringer, 1995). Lim1-null neonates lack the urogenital system, including the metanephros, ureter, and male and female reproductive tracts, demonstrating a fundamental role in kidney and genital tract formation(Kobayashi et al., 2004; Shawlot and Behringer, 1995). However, the requirement of Lim1 for nephric duct formation has hindered an understanding of its later roles in kidney development(Tsang et al., 2000). Here, we demonstrate that Lim1 is required at multiple steps of kidney development.

It is likely that the 208 kb BAC clone contains most, if not all,tissue-specific transcriptional enhancer elements for Lim1expression. It is currently unclear why none of our BAC transgenes could completely rescue the defects of Lim1-null mutants. Perhaps overexpression of Lim1 and/or other genes from the BAC clone during embryogenesis or after birth is lethal, leading to a pre-selection for lower expressing transgenic mice. These BAC transgene rescue findings suggest that Lim1 function is dose sensitive, i.e. higher expression levels of Lim1 are required for nephron formation.

Although the Hoxb7-Cre mouse line used in our study was previously generated for Cre-mediated gene manipulations in the nephric duct and ureteric bud (Basson et al., 2005; Rubera et al., 2003; Yu et al., 2002), there was no Cre mouse line specifically for the metanephric mesenchyme, but not in the ureteric bud, from its early stage of development(Bouchard et al., 2004; Ohyama and Groves, 2004; Oxburgh et al., 2004). In this study, we generated a novel resource for metanephric mesenchyme-specific gene modifications, the Rarb2-Cre mouse. These two transgenic mouse lines will serve as basic tools to dissect tissue-specific gene functions in the ureteric bud or metanephric mesenchyme.

In the metanephros, Rarb2 gene expression is restricted to the kidney stroma (Batourina et al.,2001; Mendelsohn et al.,1999). The 4.8 kb Rarb2 sequences that we used to generate the Rarb2-Cre transgenic mouse line were previously tested with a lacZ reporter; however, Rarb2-lacZ transgenes did not direct metanephric mesenchyme-specific expression(Dolle et al., 1990; Mendelsohn et al., 1991; Reynolds et al., 1991; Shen et al., 1992). Thus, we were surprised to find that our Rarb2-Cre transgenic mouse line directed Cre reporter expression in metanephric mesenchyme and its derivatives. One possibility is that the Cre reporter assay is more sensitive than a lacZ transgene. Alternatively, sequences around the site of Rarb2-Cre transgene integration were either permissive or fortuitously provided enhancer elements for expression in the metanephric mesenchyme.

Recombinase reporter systems can be used for cell-lineage analysis and fate mapping (Lewandoski, 2001). Recently, it has been suggested that some glomerular capillary cells may derive from the metanephric mesenchyme(Woolf and Loughna, 1998). However, there was no definitive cell-lineage analysis for these cell types. In this study, we found that glomerular capillary cells were Cre reporter negative in Hoxb7-Cre; R26R and Rarb2-Cre; R26R mice, suggesting that glomerular capillaries are not derived from either the ureteric bud or metanephric mesenchyme. Similar observations were obtained for the medullary stromal cells. We suggest that glomerular capillary and medullary stromal cells are derived from precursors that arise outside of the ureteric bud and metanephric mesenchyme and that these precursor cells migrate into the developing metanephros after E10.5 in the mouse. Consistent with this idea, it is known that endothelial cells do not form when ureteric bud and metanephric mesenchyme are isolated from embryos at E10.5, recombined, and cultured in vitro to form nephrons(Vainio and Lin, 2002).

It is widely believed that the upper region of the vagina derives from the Müllerian duct and the lower part from the urogenital sinus(Forsberg, 1973). However, a recent study suggests that the entire vagina derives from the Müllerian duct (Drews et al., 2002). In this study, we found that loss of the posterior part of the Müllerian duct causes the complete loss of the vagina. One explanation is that the entire vagina derives from the Müllerian duct. Alternatively, the urogenital sinus may be induced by the Müllerian duct to form the lower region of the vagina.

During embryogenesis, the Wolffian duct forms first, and subsequently the Müllerian duct forms adjacent to the Wolffian duct(Kobayashi and Behringer,2003). Experimental interruption of the Wolffian duct in chick embryos resulted in Müllerian duct growth defects(Gruenwald, 1941); however, it is unclear how the Wolffian duct is required for Müllerian duct development. Our data suggest that the Wolffian duct is essential for both elongation and maintenance of the Müllerian duct, but not for initiation.

It has previously been shown that Lim1 is essential for proper nephric duct formation (Tsang et al.,2000). Here, we found that Lim1 function is also required for maintenance of the nephric duct after its formation. Furthermore, tubule formation of the caudal mesonephros from the nephric mesenchyme is also disturbed when Lim1 is inactivated in the nephric duct epithelium. These data suggest that there are nephric duct-derived inductive molecules for mesonephric tubule formation from mesonephric mesenchyme.

Using the Hoxb7-Cre and Rarb2-Cre transgenic mouse lines,we could dissect Lim1 functions tissue specifically in the ureteric bud- and metanephric mesenchyme-derived tissues, respectively. Although both tissue-specific mutants have hypoplastic kidneys, Hoxb7-Cre; Lim1mutants are alive and have metanephroi with all structures but with abnormal ureter development, whereas Rarb2-Cre; Lim1 mutants are lethal at birth and have metanephroi without nephrons but with a proper ureteric branching pattern. These observations also indicate that Lim1functions within these two tissues may be largely independent during metanephros development. Consistent with this idea, the metanephros of Hoxb7-Cre^tg/+; Rarb2-Cre^tg/+;Lim1^lz/flox mutants was as small as Hoxb7-Cre; Lim1mutants and lacked nephrons in the same way as did Rarb2-Cre; Lim1mutants (A.K., unpublished), indicating that defects of Lim1 mutants in ureteric bud- and metanephric mesenchyme-derived tissues are additive rather than synergistic.

Ureteric bud growth and branching is severely perturbed in Retmutants (Schuchardt et al.,1994), whereas Wnt11 mutants have relatively moderate defects in ureteric bud development(Majumdar et al., 2003). It has also been shown that Wnt11 and Gdnf/Ret pathways synergistically interact with each other in ureteric bud development(Majumdar et al., 2003). In Hoxb7-Cre; Lim1 mutants, upregulation of Ret expression was not observed in the ureteric bud tip, although Wnt11 expression was normally initiated and maintained. Consistent with these expression patterns,the metanephros was more hypoplastic and ureteric bud growth is severely disturbed in Hoxb7-Cre; Lim1 mutants compared with those in Wnt11 mutants. Thus, Lim1 may act as an upstream regulator for upregulation of Ret expression in the ureteric bud tips to regulate ureteric bud growth. Indeed, Lim1 and Retexpression patterns are similar in the developing ureteric bud with highest levels at the ureteric bud tips.

Retinoic acid receptor (Rar) genes in the stromal component are required for Ret expression in the ureteric bud(Batourina et al., 2001). It is unlikely that Lim1 functions through the stroma because Lim1is required to upregulate Ret at E10.5 before Rar gene expression initiates in the stromal components around E11. Consistent with this idea, Bf2-positive stromal components are normally formed in Hoxb7-Cre;Lim1 mutants at E11.5 (data not shown).

The basic unit of the kidney, the nephron, is a long tubule with highly specialized domains, including the glomerulus, distal and proximal tubules,and loop of Henle. It is not clear how the complicated tubular structure of the nephron develops from a simple epithelial sphere, the renal vesicle. Here,we show that Lim1 is required for the initial step of nephron patterning, renal vesicle patterning. Our chimera analysis indicates that successive cell-autonomous Lim1 function is also required for comma-to S-shape body morphogenesis and podocyte development.

Although Lim1 is homogeneously expressed in the renal vesicle,comma-shaped bodies and Bowman's capsule of immature glomeruli, our chimera analysis revealed that Lim1 function is cell-autonomously required only in specific subregions in these tissues. This may indicate that Lim1 expression establishes a permissive tissue identity that can respond to unidentified regional signals that induce epithelial tissues to undergo morphological changes. Indeed, involution of the renal vesicle to form the comma-shaped body is initiated at the same region relative to the position of the ureteric bud, indicating instructive signals that determine renal vesicle patterning emanate from the ureteric tip or stalk, or the surrounding stroma.

Brn1 is essential for development of the distal tubule and loop of Henle (Nakai et al., 2003) and Notch signaling is required for formation of the proximal tubules, podocytes and glomerular capillaries (Cheng et al.,2003; McCright et al.,2001; Wang et al.,2003). Expression of these genes is restricted to a subset of cells in the renal vesicle, but we found that Brn1 and a component in Notch signaling pathway, Dll1, are not expressed in the renal vesicle of Rarb2-Cre; Lim1 mutants. These observations indicate that regional identities already established at the renal vesicle stage are not present in the absence of Lim1 function and the loss of these cell types may block further nephrogenesis.

Lim1 and Wnt4 are co-expressed in developing nephrons,leading to the idea that these genes may act in the same pathway for pretubular aggregate development (Sariola,2002). Surprisingly, we found nephrons and developing nephrons in the hypoplastic kidneys of neonatal and E14.5 Wnt4-null mutants,respectively. However, at E12.5 no developing nephrons were found in Wnt4-null mutants. Our observations are somewhat contrary to a previous report, indicating that Wnt4 is essential for the mesenchyme-to-epithelial transition during nephron development(Stark et al., 1994). This may be because of genetic background differences between the mice used in the two studies, indicating genetic modifiers that regulate nephron development in the absence of Wnt4.

The Wnt4 expression pattern in the pretubular aggregate is broader than the Lim1 expression pattern, which is restricted to the presumptive renal vesicle, and Lim1 expression is not detected in the pretubular aggregate of Wnt4-null mutants at early stages. This indicates that Wnt4 may function at an early stage of pretubular aggregate differentiation before Lim1 expression initiates in this tissue. It is possible that Lim1 and Wnt4 may regulate independent pathways in nephron progenitors. Alternatively, Wnt4 is essential for the normal onset of nephrogenesis acting as an upstream factor for Lim1 expression, but a Wnt4-independent process that upregulates Lim1 can give rise to a limited number of nephrons in Wnt4-null kidneys.

In addition to reduced and delayed nephron development, the growth of the metanephros and development of the ureteric bud is severely retarded in Wnt4 mutants compared with Rarb2-Cre^tg/+;Lim1^lz/flox mutants, in which metanephros growth and ureteric bud branching are relatively normal even when nephrons are completely absent. These data suggest that renal vesicles, which Wnt4-null mutants essentially lack, may transmit promoting signals to regulate kidney growth and ureteric bud development. Alternatively, Wnt4 in the pretubular aggregate itself may also be the promoting signal for ureteric bud development.

The embryonic development of the three successive kidneys of amniotes recapitulates the evolution of this organ system(Saxen, 1987). Whereas there are many genes with restricted expression patterns in diverse tissues and stages that regulate one particular aspect of kidney development, there are very few genes that are expressed in each of the three kidney types(Vainio and Lin, 2002). A small set of transcription factors (Pax2, Pax8 and Lim1) are expressed in tubule-forming tissues in each of the three kidney types. Mutants for Lim1 and Pax2 have nephric duct defects that lead to an absence of mesonephros and metanephros development(Bouchard et al., 2004; Shawlot et al., 1999; Torres et al., 1995; Tsang et al., 2000). Pax8 mutants do not have kidney defects but genetic studies indicate that Pax2 and Pax8 act redundantly for nephric duct formation, perhaps upstream of Lim1(Bouchard et al., 2002; Mansouri et al., 1998). In addition, overexpression of both XPax-8 and XLim-1 together in frog embryos leads to embryonic kidney overgrowth and ectopic pronephric tubules (Carroll and Vize,1999). The results presented here and by others suggest that Lim1 is crucially required in the tubule-forming tissues at multiple steps of kidney development. Although the precise relationships between Pax2, Pax8 and Lim1 are complex, it is clear that they are each used at multiple steps of kidney organogenesis.

The development of the amniote excretory system (pronephros, mesonephros,and metanephros) suggests that it has evolved by elaborating on a fundamental developmental process, tubulogenesis. A simple way to evolve such an organ system would be to use that process repeatedly and exploit a fundamental genetic program. It would appear that the set of transcription factors mentioned above serves as a fundamental genetic cassette for tubulogenesis during kidney development. Interestingly, the mechanism regulated by Lim1 for nephric duct, Wolffian and Müllerian duct, mesonephric tubule and ureteric bud formation appear to be similar, i.e. a cell-autonomous action for maintenance and/or growth of the ductal epithelium. However, this mechanism appears to be fundamentally different for nephron formation, i.e. formation of a permissive tissue identity, resulting in regional patterning of the renal vesicle for subsequent morphogenesis. Thus, it would appear that a new elaboration of our proposed primary genetic cassette evolved for nephron formation in the metanephros. This may have been facilitated by the ability of LIM-class homeodomain proteins to complex with other factors to regulate transcription (Bach et al.,1997).

We thank William Shawlot for Lim1^+/+;R26^tg/+ and Lim1^–/–;R26^tg/+ ES cells, Artur Kania and Thomas Jessell for Lim1^lz/+ mice, Allan Bradley for SNL 76/7 STO cells, and Robert McEvilly, Frank Costantini, Achim Gossler, Brigid Hogan and Peter Gruss for probes. These studies were supported by a grant from the National Institutes of Health (NIH) HD30284 to R.R.B., by DK61459 and DK55388 to C.M.,and by DK054363 to A.P.M. DNA sequencing and veterinary resources were supported by the NIH Cancer Center Support Grant CA16672.