Requirement of Lim1 for female reproductive tract development

Lim1 encodes a LIM-class homeodomain transcription factor that is essential for head and kidney development. In the developing urogenital system, Lim1 expression has been documented in the Wolffian (mesonephric) duct, the mesonephros, metanephros and fetal gonads. Using, a Lim1 lacZ knock-in allele in mice, we identified a previously unreported urogenital tissue for Lim1 expression, the epithelium of the developing Müllerian duct that gives rise to the oviduct, uterus and upper region of the vagina of the female reproductive tract. Lim1 expression in the Müllerian duct is dynamic, corresponding to its formation and differentiation in females and regression in males. Although female Lim1-null neonates had ovaries they lacked a uterus and oviducts. A novel female mouse chimera assay was developed and revealed that Lim1 is required cell autonomously for Müllerian duct epithelium formation. These studies demonstrate an essential role for Lim1 in female reproductive tract development.


Introduction
The mammalian female reproductive tract is an organ system that is composed of the oviducts, uterus, cervix and vagina. The primary role of this organ system is reproduction, i.e. the continuation of the species. The female reproductive tract organs are also a significant concern for women's health and disease, notably infertility and cancer. Some diseases of the female reproductive tract are caused by abnormalities that occur in the female fetus during embryogenesis. These include Müllerian agenesis and vaginal septum, which can prevent normal reproduction (Gidwani and Falcone, 1999). Agenesis of the female reproductive tract has been estimated to occur in one in 4000 to 20,000 women (Kim and Laufer, 1994). It is also suggested that deregulation of embryonic genes is observed in some types of cancers in several organs (Peifer and Polakis, 2000;Chi and Epstein, 2002;Ruiz i Altaba et al., 2002). Most studies of the female reproductive tract are focused upon its biology in adults. Thus, the molecular mechanisms that regulate female reproductive tract development are largely unknown (Kobayashi and Behringer, 2003).
During vertebrate embryogenesis, the urogenital system derives from the intermediate mesoderm of the gastrula. The female reproductive tract system develops primarily from the Müllerian (paramesonephric) duct and the male reproductive tract forms from the Wolffian (mesonephric) duct. In the mouse, the Wolffian duct is first formed from the intermediate mesoderm by embryonic day 9 (E9). Subsequently, the Müllerian duct starts to form by invagination of the surface epithelium of the anterior mesonephros around E11.5 in the developing urogenital ridge. This epithelial invagination extends caudally along the Wolffian duct laterally and then medially towards the cloaca to form the primordium of the female reproductive tract (Gruenwald, 1941;Kaufman and Bard, 1999). Thus, embryos have both male and female reproductive tract primordia regardless of their genetic sex before sexual differentiation occurs. The Müllerian duct can differentiate into the oviduct, uterus, cervix and upper part of the vagina of the female reproductive tract. The Wolffian duct can differentiate into the epididymis, vas deferens and seminal vesicle of the male reproductive tract.
Mammalian sex determination depends on the genetic sex in the gonad (Swain and Lovell-Badge, 1999;Capel, 2001). XY embryos usually become males and XX embryos usually become females. It is also known that, XX↔XY chimeric mice mostly develop into males (Tarkowski, 1998) and Sertoli cells of the testis from these chimeric animals are predominantly XY (Burgoyne et al., 1988;Palmer and Burgoyne, 1991;Patek et al., 1991). This indicates that the testis-determining gene acts cell-autonomously in this cell lineage and that high contribution of XY cells in this cell lineage in the gonads of XX↔XY chimeric mice can result in the male phenotype. The testis-determining gene on the Y chromosome, Sry, is both essential and sufficient for triggering testis differentiation to cause male differentiation (Gubbay et al., 1990;Koopman et al., 1991). Sry is expressed transiently and dynamically in the bipotential gonad of XY males (Bullejos and Koopman, 2001). It has been suggested that Sry is expressed in precursor cells of Sertoli cells (Albrecht and Eicher, 2001). In XX females, the absence of the Y chromosome permits the bipotential gonad to differentiate into an ovary leading to the female Lim1 encodes a LIM-class homeodomain transcription factor that is essential for head and kidney development. In the developing urogenital system, Lim1 expression has been documented in the Wolffian (mesonephric) duct, the mesonephros, metanephros and fetal gonads. Using, a Lim1 lacZ knock-in allele in mice, we identified a previously unreported urogenital tissue for Lim1 expression, the epithelium of the developing Müllerian duct that gives rise to the oviduct, uterus and upper region of the vagina of the female reproductive tract. Lim1 expression in the Müllerian duct is dynamic, corresponding to its formation and differentiation in females and regression in males. Although female Lim1-null neonates had ovaries they lacked a uterus and oviducts. A novel female mouse chimera assay was developed and revealed that Lim1 is required cell autonomously for Müllerian duct epithelium formation. These studies demonstrate an essential role for Lim1 in female reproductive tract development.
phenotype. Although the loss of Y chromosome is known to cause Turner's syndrome in humans, XO mice are phenotypically normal females except for a transient developmental delay until early mid-gestation stage and early loss of oocytes after birth (Morris, 1968;Lyon and Hawker, 1973;Burgoyne and Baker, 1981;Burgoyne et al., 1983).
After gonadal sex is determined, the differentiating gonads secrete sexual hormones to promote sexual differentiation of the body. In males, the fetal testis secretes hormones including Müllerian inhibiting substance (MIS; AMH -Mouse Genome Informatics), testosterone and insulin-like 3 (Insl3) (Nef and Parada, 2000). MIS causes the elimination of the Müllerian duct and testosterone promotes the differentiation of the Wolffian duct. In the mouse, regression of the Müllerian duct system is observed cytologically from E13.5 (Dyche, 1979). All three hormones are involved in testicular descent. In humans and mice, males deficient for MIS or its type II receptor (MISRII; AMHR2 -Mouse Genome Informatics) are normally virilized and possess a male reproductive tract but fail to regress the Müllerian duct and retain ectopic female reproductive tract organs (Behringer et al., 1994;Mishina et al., 1996;Belville et al., 1999). In female fetuses, the differentiating ovaries do not produce MIS, testosterone or Isl3, which allows the Müllerian duct to differentiate into the female reproductive tract, the Wolffian duct to degenerate and the ovaries to remain in an intra-abdominal position (Kobayashi and Behringer, 2003).
Lim1 (also known as Lhx1) encodes a transcription factor with a DNA-binding homeodomain and two cysteine-rich LIM domains that are thought to be involved in protein-protein interactions Bach, 2000;Hobert and Westphal, 2000). During mouse urogenital system development, Lim1 is expressed in the intermediate mesoderm at E7.5 and this expression is subsequently restricted to the nephric duct that differentiates into the Wolffian duct (Barnes et al., 1994;Tsang et al., 2000). Lim1 is also expressed in the developing mesonephros in embryos and in the definitive kidney (metanephros) in both embryos and adults (Barnes et al., 1994;Fujii et al., 1994;Karavanov et al., 1996;Karavanov et al., 1998). Lim1 expression is also detected in the fetal gonad (Nagamine and Carlisle, 1996;Nagamine et al., 1999;Birk et al., 2000;Bouchard et al., 2002). These expression data indicate that Lim1 may play important roles in multiple processes of urogenital system development, including the reproductive organs. The Lim1 gene was previously mutated in the mouse and was found to be required for head and kidney formation (Shawlot and Behringer, 1995). However, except for very rare neonates, most Lim1-null mutants die around E10 probably owing to the failure of chorio-allantoic fusion to form the placenta. Therefore, the roles of Lim1 at later stages of development have remained unclear.
In this study, we show that Lim1 is also expressed in the developing Müllerian duct during female reproductive tract development. This expression for the first time allows for the visualization of Müllerian duct formation and regression in embryos. In the absence of Lim1 function, the female reproductive tract is absent in female neonates. We also show, using a novel female mouse chimera assay, that Lim1 activity is required cell-autonomously in the epithelium of the developing Müllerian duct. These data establish a new and essential role for Lim1 in female reproductive tract development.
X-gal staining of embryos X-gal staining for β-gal activity was performed as described (Nagy et al., 2003). After overnight post-fixation with 4% paraformaldehyde in PBS, photographs were taken. For histological analysis, paraffin waxembedded tissues were sectioned at 7 µm and counterstained with 0.33% eosin-Y.

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

Dynamic expression of Lim1 during Müllerian duct development
To further understand Lim1 function during urogenital organogenesis, we examined Lim1 expression in developing urogenital tissues using mice that were heterozygous for a Lim1-tau-lacZ (Lim1 tlz ) knock-in allele (Shawlot et al., 1999;Kania et al., 2000). β-galactosidase (β-gal) activity was observed in the Wolffian ducts and mesonephric tubules ( Fig.  1A-F), consistent with previous reports of Lim1 expression in these tissues (Barnes et al., 1994;Fujii et al., 1994;Karavanov et al., 1998). Surprisingly, we also detected β-gal activity in the Müllerian duct, which was very dynamic during embryogenesis (Fig. 1). The onset of Lim1 tlz expression in the Müllerian duct was detected at embryonic day 11.5 (E11.5) in the most anterior region of the urogenital ridge ( Fig. 1A). At this stage, the Müllerian duct has just formed as a small invagination of the surface epithelium of the mesonephros, located adjacent to the cranial end of the Wolffian duct. Subsequently, Lim1 tlz expression in the Müllerian duct extends caudally parallel to the Wolffian duct ( Fig. 1B). At E12.5, Lim1 tlz is expressed in the Müllerian duct along the cranial twothirds of the urogenital ridge (Fig. 1C). Lim1 tlz expression was also detected in the developing metanephros, the definitive kidney in mammals (Fig. 1C,E,F). At E13.5, Lim1 tlz expression in the Müllerian duct further extended caudally towards the cloaca and was observed in both Müllerian and Wolffian ducts along the entire urogenital ridge in both sexes (Fig. 1E,F). No obvious sexual dimorphic pattern of Lim1 tlz expression in the reproductive ducts was observed up to this stage.
Because the lacZ reporter gene that was introduced into the mouse Lim1 locus replaces the entire Lim1-coding region including all four introns (Shawlot et al., 1999;Kania et al., 2000), there was concern that the inserted lacZ transgene might disturb the transcriptional regulation of Lim1 gene expression and thus the lacZ reporter mice may not reflect endogenous Lim1 expression patterns. To exclude this possibility, we performed whole-mount in situ hybridization of urogenital tissues from wild-type embryos. The distribution of Lim1 transcripts in the Wolffian and Müllerian ducts of wild-type embryos was identical to the β-gal expression pattern in Lim1 tlz heterozygotes. Lim1 is expressed in the Wolffian duct along the entire urogenital ridge, and Lim1 expression in the Müllerian duct dynamically extends caudally between E11.5 and E13.5 ( Fig. 1G,H, data not shown). These data indicate that the β-gal expression pattern in Lim1 tlz mice faithfully reflects endogenous Lim1 expression in the developing genital ducts and validate our expression data using the Lim1 tlz reporter.
To determine which tissues specifically express Lim1 in the urogenital tract, we performed a histological analysis. We found that Lim1 tlz is specifically expressed in the epithelium of the Müllerian and Wolffian ducts and the mesonephric tubules ( Fig. 1D), but not in the surrounding mesenchyme of these Ventral view at E12.5 (C) and at E13.5 in females (E) and males (F). Cross-section of the anterior mesonephric region at E12.5 counterstained with Eosin (D). (G,H) Lim1 expression detected using whole-mount in situ hybridization at E12.5 (G) and E13.5 (H) in the female genital system. Asterisks point to the posterior end of the Müllerian duct and broken lines indicate the gonad. g, gonad; k, kidney (metanephros); m, mesonephros; md, Müllerian duct; wd, Wolffian duct. Scale bars: 250 µm in A-C,G,H; 100 µm in D; 500 µm in E,F. tissues. At this stage (E12.5), the mesonephric tubules begin to regress. Histological analysis of serial sections along the longitudinal axis of the urogenital ridge revealed that no epithelial structure of the Müllerian duct was detected posteriorly to the distal tip of Lim1 tlz expression in the Müllerian duct (data not shown). These data indicate that Lim1 expression is coincident with Müllerian duct formation and suggests a role in female reproductive organ formation.

Sexual dimorphic expression of Lim1 in the developing reproductive tract
We further examined Lim1 expression at later stages of urogenital development using the Lim1 tlz reporter. Sexual dimorphic patterns of Lim1 tlz expression were observed in the reproductive tract beginning at E14.5. In females, strong Lim1 tlz expression was observed in the Müllerian duct at E14.5 ( Fig. 2A) and this strong expression persists until E15.5 ( Fig.  2C,E). Lim1 tlz expression becomes weaker throughout the Müllerian duct at E16.5 (Fig. 2G). By E17.5, Lim1 tlz expression in the Müllerian duct becomes restricted anteriorly, to the prospective oviduct. Lim1 tlz expression in the posterior presumptive uterus region is downregulated and becomes undetectable (Fig. 2I). Lim1 tlz expression in the Müllerian duct was not observed at later stages (data not shown). In the Wolffian duct of females, Lim1 tlz expression is lost from the anterior gonadal region around E15.25 and becomes undetectable at later stages ( Fig. 2E,G,I, data not shown).
In males, Lim1 tlz expression in the Müllerian duct appears thinner at E14.5 (Fig. 2B) compared with the expression in females at the same stage ( Fig. 2A). At E15.25, Lim1 tlz expression in the Müllerian duct becomes discontinuous from its cranial region (open arrows in Fig. 2D). Subsequently, this discontinuity of Lim1 tlz expression spreads caudally and Lim1 tlz expression in the Müllerian duct shows a fragmented pattern throughout the entire reproductive tract region at E15.5 ( Fig. 2F). At E16.5, Lim1 tlz expression in the Müllerian duct becomes undetectable except for expression in a few vesicular structures (arrows in Fig. 2H). Histological analysis of serial sections of the male urogenital system at E15.5 showed that there is no epithelium tissue of the Müllerian duct where Lim1 tlz expression is absent (data not shown). This indicates that the Lim1 tlz expression fragmentation is likely caused by loss of the epithelial tissue during the process of Müllerian duct regression. Lim1 tlz expression in the Wolffian duct and its derivatives such as the epididymis and vas deferens persists in males during these stages (Fig. 2B,D,F,H,J) and becomes upregulated around E17.5 (Fig. 2J).

Visualization of Müllerian duct regression by Lim1 tlz expression
MIS signaling is both essential and sufficient for Müllerian duct regression. MIS ligand is expressed by Sertoli cells in the testis from E11.5 (Hacker et al., 1995;Swain and Lovell-Badge, 1999) and the MIS type II receptor, Misr2, is expressed in the mesenchyme of the Müllerian duct from E13.0 in an anterior to posterior manner (A.K. and R.B., unpublished) in the mouse. We observed sexual dimorphic expression patterns of Lim1 tlz expression in the Müllerian duct from E14.5 (Fig.  2), 1 day after Misr2 is expressed in the mesenchyme along the entire length of the Müllerian duct. The timing of this differential expression pattern suggests that the fragmentation of Lim1 tlz expression reflects MIS-induced Müllerian duct Development 131 (3) Research article   (Fig. 3A-D). Lim1 tlz expression in the Wolffian duct and kidney was not affected in these mutant males. These data indicate that the fragmentation of Lim1 tlz expression in the Müllerian duct depends on MIS signaling. We also examined Lim1 tlz expression in female transgenic mice, in which human MIS is ectopically expressed (Behringer et al., 1990). This ectopic human MIS expression in females causes Müllerian duct regression. We detected no difference in Lim1 tlz expression patterns until E13.5 between wild-type and transgenic females (data not shown). However, Lim1 tlz expression in the Müllerian duct of Mis transgenic female mice (Fig. 3E) becomes thinner than the Lim1 tlz expression of wildtype littermate females at E14.5 (Fig. 3F). At E15.5, Lim1 tlz is still strongly expressed in the Müllerian duct of wild-type females (Fig. 3G). At the same stage, the Lim1 tlz expression in the Müllerian duct becomes discontinuous in Mis transgenic female mice (Fig. 3H). At E16.5, Lim1 tlz is weakly expressed along the entire Müllerian duct in wild-type females (Fig. 3I). At the same stage, Lim1 tlz expression in the Müllerian duct is restricted to the cranial end and within a few vesicles in Mis transgenic female mice (Fig. 3J). Ectopic expression of Mis causes thinner and flattened ovaries (Fig. 3E-J) (Lyet et al., 1995) but did not affect Lim1 tlz expression in the Wolffian duct and kidney (Fig. 3E-L). These data suggest that MIS signaling is sufficient to cause fragmentation of Lim1 tlz expression in the Müllerian duct. We also examined Lim1 tlz expression in Mis transgenic males. The fragmentation of Lim1 tlz expression is enhanced in Mis transgenic males compared with wild-type males at E15.5 (Fig. 3K,L).

Absence of the female reproductive tract in female Lim1-null mice
The dynamic expression pattern of Lim1 in the developing Müllerian duct indicates that Lim1 may have an important function during female reproductive tract organogenesis. Unfortunately, most Lim1 mutants die at around E10 probably because of a failure of chorion-allantois fusion before the formation of the Müllerian duct that is initiated around E11.5. Previously, we found that rare Lim1 mutants escaped this embryonic lethality and survived to the birth. These Lim1-null escapers lacked anterior head and kidney formation. However, only four escapers had been obtained out of more than 1000 (<0.4%) pups born (Shawlot and Behringer, 1995).
One explanation for Lim1-null mutant survival to birth is that the embryonic lethality is affected by genetic modifiers. Therefore, we modified the genetic background of the Lim1 mutation to increase phenotypic variety by outcrossing Lim1 heterozygous mice (maintained in C57BL/6J×129/SvEv mixed background) with Swiss Webster (Taconic) outbred mice. Subsequently, we intercrossed the resulting Lim1 heterozygous offspring from different pedigrees to screen for Lim1-null escapers. We found one breeding pair that produced two Lim1null escapers out of 76 (2.6%) pups. These escapers were genotyped and one escaper was an XX female and the other an XY male. We also intercrossed Lim1 heterozygous offspring from the mating pair that produced the two escapers and obtained two additional XX Lim1-null escaper neonates.
We analyzed the three XX Lim1-null neonates and one of the original four escapers that was also XX for female reproductive organ development. All of four Lim1-null female neonates had the identical phenotype. There was no anterior head formation (Fig. 4A,B). Examination of the internal reproductive organs of these female Lim1-null neonates showed that they had ovaries that were morphologically normal ( Fig. 4D,F). The male Lim1-null escaper had testes that were indistinguishable from the testes of wild-type animals (data not shown). Although the female Lim1-null neonates had ovaries, they completely lacked derivatives of the Müllerian duct, including oviducts, uterus, cervix and the upper region of the vagina (Fig. 4C-H). Although Lim1 is expressed only within the epithelium of the Müllerian duct (Fig. 1D), both the epithelium and the mesenchyme of the female reproductive tract were completely absent in the female Lim1-null neonates (Fig. 4G-H). These data suggest that Lim1 is essential for Müllerian duct development during embryogenesis and that lack of Lim1 activity results in Müllerian agenesis in females.

Generation of female mouse chimeras
To obtain further clues about action of Lim1 during female reproductive tract development, we generated chimeric mice composed of Lim1 -/and wild-type cells by injection of mutant ES cells into wild-type blastocysts (Tam and Rossant, 2003). In mammals, testis determination depends on the expression of the Y chromosome-linked Sry gene in the gonad of embryos. Most ES cell lines commonly used are genetically XY. The incorporation of XY cells into the gonad of chimeric animals can cause female-to-male sex reversal, resulting in a male phenotype and therefore regression of the Müllerian duct system (Tarkowski, 1998). Therefore, to avoid this aspect of chimera biology and study the role of Lim1 in female reproductive tract development, we developed a novel female mouse chimera strategy (Fig. 5A).
XY ES cells spontaneously lose the Y chromosome and XO cells are found at a ~1-2% frequency (A. Bradley, personal communication). Therefore, we exploited this phenomenon and generated Lim1 -/and Lim1 +/+ Rosa26-marked XO female ES cell lines. To establish XO ES cell lines, we plated Lim1 -/and Lim1 +/+ Rosa26-marked XY ES cells (Shawlot et al., 1999) at clonal density. 576 ES cell colonies were recovered and examined. A Y chromosome-specific repeat probe, Y353/B, was used to screen for Y-chromosome deficiency (Bishop and Hatat, 1987). Loss of the Y chromosome was observed in nine independent clones, including five Lim1 -/lines and three Lim1 +/+ lines (Fig. 5B). All XO ES cell lines were used to generate chimeric mice.
To generate female chimeras, Lim1 -/and Lim1 +/+ XO ES cells were injected into wild-type XX blastocysts (Fig. 5A). XX blastocysts were distinguished from XY blastocysts using X-linked GFP transgenic male mice (Hadjantonakis et al., 1998) for matings with wild-type females. In this strategy, only chimeric embryos derived from XX blastocysts are GFP positive. Chimeric embryos were harvested and stained with X-gal to distinguish ES cell-derived cells from blastocystderived cells. The sex genotype of the recipient blastocysts was further confirmed by PCR genotyping for Sry using the yolk sac of the chimeras.

Cell-autonomous requirement of Lim1 for Müllerian duct epithelium formation
We generated 27 female chimeras and analyzed the distribution of ES-derived cells (Fig. 6A,B). Identical results were obtained for all five Lim1 -/-XO ES cell lines and for all three Lim1 +/+ XO ES cell lines examined. High contribution of XO Lim1 -/cells in chimeric animals caused craniofacial abnormalities. These included loss of the lower jaw (data not shown) and, in more severe cases, head truncation (Fig. 6B). However, the Lim1-null urogenital defects were completely rescued in all chimeric animals composed of Lim1-null and wild-type cells that were recovered. Histological analysis was performed to understand the tissue distribution of Lim1 -/cells in chimeric female mice. In control experiments, XO wild-type cells could contribute to both the epithelium and mesenchyme of the uterus and oviduct in chimeric females at E18.5 (Fig. 6C,E). We also examined other organs and did not observe any bias of distribution in these chimeras. This indicates that XO cells can extensively contribute to the somatic tissues of chimeric mice when XO cells are mixed with XX cells. By contrast, when XO Lim1 -/cells were used to generate female chimeric mice, we found that these Lim1 -/cells Development 131 (3) Research article did not contribute to the epithelium of the uterus at E18.5, although these Lim1 -/cells could contribute extensively to the uterus mesenchyme (Fig. 6D). We also found that Lim1 -/cells were not present in the epithelium of the oviduct at the same stage (Fig. 6F). These data suggest that Lim1 function is required cell-autonomously for epithelium development of the uterus and oviduct.
To test if Lim1 is required for formation of the Müllerian duct, we also analyzed chimeric mice at E12.5 when the Müllerian duct is being formed prior to its most caudal extension along the urogenital ridge (Fig. 1C,G). In control experiments, XO Lim1 +/+ cells contribute to the both epithelium and mesenchyme of the Müllerian and Wolffian ducts of chimeric mice along with wild-type cells at this stage (Fig. 6G). However, in chimeric mice composed of XO Lim1 -/cells, no Lim1 -/cells were observed in the Müllerian duct epithelium (Fig. 6H). This indicates that Lim1 function is required cell-autonomously during Müllerian duct epithelium formation. We also found that Lim1 -/cells were not found in the Wolffian duct epithelium at this stage. This suggests that Lim1 function is also required cell-autonomously for epithelium development of the male reproductive tract.

Lim1 expression in Müllerian duct precursor cells is independent of Wnt4 function
To understand interactions with other genes that are required for Müllerian duct development, we examined Lim1 tlz expression in Pax2, Wnt4 and Wnt7a mutants (Fig. 7). Wnt7a is expressed in the Müllerian duct epithelium from E11.5  (Shawlot et al., 1999) by spontaneous loss of the Y chromosome by screening with an Ychromosome-specific repeat probe, Y353/B. These XO ES cells were injected into XX blastocysts to generate chimeric female mice. XX blastocysts can be distinguished from XY blastocysts by using X-linked ubiquitous GFP male mice for breeding. Chimeric mice were recovered at different stages of embryogenesis and processed for β-gal staining to visualize Rosa26-marked ES cell-derived XO cells. The yolk sac was collected for reconfirming sexes of injected blastocysts by PCR genotyping for the Sry gene. (B) Southern blot analysis of XO ES cells. Genomic DNA of XO ES cells was blotted after EcoRI digestion and hybridized with a Lim1 probe. The same blotted membrane was hybridized again with Y-chromosome-specific Y353/B probe. Genomic DNA from male and female tails was used for comparison. Note that these XO cells show the same pattern as female tail.  (Vainio et al., 1999). In Wnt7a mutant females, the Müllerian duct is formed but tissues of the oviduct and uterus fail to form proper cytoarchitectures (Miller and Sassoon, 1998), indicating a requirement of Wnt7a for Müllerian duct differentiation. In Wnt7a mutants, Lim1 tlz expression in the urogenital system, including the Müllerian duct, was identical to wild-type at E12.5 (Fig. 7A,D). In Pax2 mutants, the Wolffian duct is formed only anteriorly at E11.5 but starts to degenerate at E12.5 (Torres et al., 1995). Pax2 mutants also lack mesonephric tubule formation and metanephros induction. Pax2 is also expressed in the epithelium of the developing Müllerian duct by E13.5 and Pax2 mutants form only the anterior region of the Müllerian duct by E13.0 but it degenerates subsequently by E16.5 (Torres et al., 1995), indicating a Pax2 requirement for maintenance of the Müllerian duct. Lim1 tlz expression in the Wolffian duct, mesonephros and metanephros was not observed in Pax2 mutants at E12.5 (Fig. 7C). However, the shortened Müllerian duct of Pax2 mutants still expressed Lim1 tlz at the same stage (Fig. 7C). Taken together, these data suggest that Lim1 expression in the Müllerian duct does not require Wnt7a or Pax2 function.
It was reported that the Müllerian duct is absent in Wnt4 mutants at E11.5 and E12.5 (Vainio et al., 1999), suggesting a Wnt4 requirement for Müllerian duct formation. Surprisingly, we found ectopic weak staining at the anterior end of the mesonephros of Wnt4 mutants at E12.5 (Fig. 7B). To understand the origin of these Lim1 tlz positive cells, we also examined Lim1 tlz expression in Wnt4 mutants at E11.5, when the Müllerian duct just begins to form. We found Lim1 tlz expression at the anterior end of the mesonephros of Wnt4 mutants (Fig. 7F), where the invagination of the Müllerian duct normally initiates (Fig. 7E). At this stage, the Müllerian duct shows a funnel shape of the invaginating surface epithelium of the mesonephros in wild-type or Wnt4 +/control embryos (Fig.  1A, Fig. 7E,G). However, the Lim1 tlz positive cells in the anterior mesonephros of Wnt4 mutants do not invaginate for tubulogenesis of the Müllerian duct, but remain as a small cluster (Fig. 7F,H). These observations suggest that induction of Lim1 expression in Müllerian duct precursor cells does not require Wnt4 function. These data also suggest that Wnt4 is required for invagination of the Müllerian duct, but not specification of its precursor cells. Lim1 tlz expression in the mesonephos of Wnt4 mutants is greatly reduced at E11.5 and almost absent at E12.5 (Fig. 7B,F,H), suggesting a Wnt4 requirement for mesonephros development.

Visualization of Müllerian duct development by Lim1 tlz
Analysis of the expression of the Lim1 tlz allele in mice has revealed a previously unrecognized tissue for Lim1 expression, the epithelium of the developing Müllerian duct and its derivatives in the uterus and oviducts. This Lim1 tlz allele provides a visual marker that reveals for the first time the dynamic morphological changes associated with the formation and differentiation of the Müllerian duct into the organs of the female reproductive tract, and the regression of the prospective female reproductive tract primordium in male fetuses. The Müllerian duct forms adjacent to the Wolffian (mesonephric) duct in an anterior to posterior manner by the invagination of the anteriolateral coelomic epithelium of the mesonephros (Gruenwald, 1941;Kaufman and Bard, 1999). Lim1 tlz expression was coincident with the initial formation of the Müllerian duct epithelium, suggesting that Lim1 may be essential in this tissue for the formation of the Müllerian duct. Expression was maintained in the developing Müllerian duct until E16.5 but was then downregulated. However, Lim1 expression persisted in the differentiating oviduct, indicating that Lim1 may have later roles in the regional differentiation of the female reproductive tract perhaps regulating oviductal morphogenesis by altering cell shape or regional proliferation. Interestingly, Hoxa10 is expressed in the developing uterus, but not in the oviduct at this stage (Benson et al., 1996;Ma et al., 1998). In Hoxa10 mutant females, the anterior uterus is transformed into the oviduct (Benson et al., 1996), suggesting that Hoxa10 may downregulate Lim1 expression in the presumptive uterus to inhibit oviduct differentiation.
During male sexual development, the Müllerian ducts are eliminated by the actions of the MIS signaling pathway (Josso et al., 1993). Regression of the Müllerian duct system visualized by the Lim1 tlz reporter was first observed at E14.5 Over the next 2 days of embryonic development, the Müllerian duct system in the male fetus is eliminated. In males that lack MIS, its type II receptor or Wnt7a, the Lim1 tlz reporter documented the persistence of the Müllerian duct in the absence of MIS signaling. Thus, the Lim1 tlz reporter provides the first opportunity to visualize Müllerian duct regression or persistence from the perspective of the ductal epithelium. In the mouse, MIS expression is first detected around E11.5 at the time of testis determination (Hacker et al., 1995;Swain and Lovell-Badge, 1999). MIS signaling requires the MIS type II receptor expressed in the mesenchyme cells adjacent to the Müllerian duct epithelium (Baarends et al., 1994;Mishina et al., 1996). This MIS type II receptor expression requires the function of Wnt7a, which is expressed in the Müllerian duct epithelium (Parr and McMahon, 1998). Using a Misr2-lacZ knock-in allele in mice, we have detected β-gal activity throughout the Müllerian duct mesenchyme by E13.5 (N. A. Arango,A.K. and R.R.B.,unpublished). Overexpression of human MIS in female transgenic mice causes the elimination of Müllerian ducts and therefore they do not have a uterus or oviducts (Behringer et al., 1990). Interestingly, the female Mis transgenic fetuses had an altered spatial pattern of Müllerian duct regression in comparison with normal males. Whereas fragmentation of the Lim1 tlz β-gal pattern (i.e. loss of the Müllerian duct epithelium) in males followed an anterior to posterior pattern, in females, fragmentation was throughout the anterior-posterior limits of the Müllerian ducts. These observations indicate that at E15.5 the posterior Müllerian duct is competent for regression, consistent with the establishment of Misr2 expression in the adjacent mesenchyme along the entire Müllerian duct at this stage. We also found that the regression of the Müllerian duct epithelium of male Mis transgenic fetuses was enhanced, suggesting that MIS is not at saturating levels for its signal transduction in vivo as shown previously using reconstruction from serial sections (Allard et al., 2000).

Role of Lim1 in female reproductive tract development
Most Lim1 mutants die at mid-gestation because of defects in allantois differentiation and the subsequent failure of chorioallantoic fusion to establish the maternal-fetal connection (Shawlot and Behringer, 1995). Lim1 is expressed in the primitive streak during gastrulation. Thus, these defects in allantoic development are likely to be the consequence of alterations in the posterior primitive streak directly or indirectly caused by the lack of Lim1. The frequency of Lim1 mutants surviving to birth appeared to be increased when the mutation was moved onto a more diverse, outbred genetic background. The phenotypes of the Lim1-null neonates were consistent between animals, suggesting that there are genetic modifiers that specifically influence the expressivity of the mutation during gastrulation to affect placentation. Analysis of four Lim1-null female neonates showed a complete absence of the derivatives of the Müllerian ducts, the oviducts and uterus, establishing an essential role for Lim1 in the formation of the female reproductive tract. The uterus and oviducts derive from the Müllerian duct epithelium but also from the surrounding mesenchyme. Thus, the complete absence of the uterus in the Lim1-null female neonates demonstrates that the Müllerian duct epithelium has an essential role in instructing the mesenchyme to participate in uterine organogenesis. In addition, the one Lim1-null male neonate that was obtained did not have Wolffian duct derivatives, identifying a role for Lim1 in male reproductive tract development. Whereas the Müllerian and Wolffian duct derivatives were absent in the Lim1-null neonates, there were gonads that were morphologically and histologically normal. It was previously reported that Lim1-null neonates lacked anterior head structures, kidneys and gonads (Shawlot and Behringer, 1995). However, it is now clear that Lim1-null mice can form gonads and that Lim1 is dispensible for gonad formation. Several groups have reported Lim1 expression in the developing gonad (Nagamine and Carlisle, 1996;Nagamine et al., 1999;Birk et al., 2000;Bouchard et al., 2002). Thus, it is still possible that Lim1 has a role in gonad development but such a role would be compensated by other factors in its absence.
Phenotypic analysis of the female Lim1-null neonates established a requirement for Lim1 in female reproductive tract development but did not provide information about when, how and in which tissues Lim1 acted. To address these questions a chimera study was performed (Tam and Rossant, 2003). However, because there is an inherent bias towards male development when male and female cells are mixed, we devised a novel female chimera assay, exploiting the fact that genetically XO mice develop as females (Morris, 1968). Using this assay, we showed that Lim1 activity is required cell autonomously in the epithelium of the Müllerian duct for female reproductive tract development. This was true at E18.5 and at E12.5 when the Müllerian duct is just forming. Our data indicate that Lim1 function is required for the formation or very early steps of the differentiation of the Müllerian duct epithelium. However, because Müllerian duct formation initiates at E11.5, we cannot formally conclude that Lim1 is essential for the initial formation of the Müllerian duct epithelium. Because Lim1-null cells to do not contribute to the Müllerian duct epithelium of female chimeras, we could not assess the role of Lim1 at later stages of Müllerian duct differentiation where it might act to maintain this tissue or regulate oviductal morphogenesis. Conditional genetic strategies may be required to address this issue (Kwan and Behringer, 2002).
The role of Lim1 has previously been investigated using mouse chimeras to understand its function in motoneuron axon trajectories in the developing limb . These studies identified a cell-autonomous requirement for Lim1 in selecting a dorsal trajectory in the limb. These results and those presented in the current study that support a cell-autonomous action of Lim1 stand in contrast to another chimera analysis of Lim1 during gastrulation and head formation (Shawlot et al., 1999). In those studies, we showed that Lim1 expression in primitive streak-derived tissues and the visceral endoderm acted in a cell non-autonomous manner to regulate anterior head formation. However, recent studies in frogs and mice suggest that the abnormalities in head formation may be secondary to a cell autonomous defect in cell adhesion that alters mesoderm migration (Hukriede et al., 2003). Thus, it is possible that Lim1 may regulate a common set of downstream targets in different tissues.

Relationship of Lim1 with other genes that influence Müllerian duct development
We examined Lim1 expression in mutants with abnormalities of Müllerian duct development to understand the relationship of Lim1 with other genes in the genetic cascade of Müllerian duct development (Kobayashi and Behringer, 2003). Our chimera analysis indicated that Lim1 is required for Müllerian duct formation. This is consistent with our observations that Lim1 expression in the forming Müllerian duct does not require Pax2 or Wnt7a, which are essential for later events of maintenance or differentiation of the Müllerian duct system, respectively (Torres et al., 1995;Miller and Sassoon, 1998). Pax8 is also expressed in the Müllerian duct epithelium (Vainio et al., 1999) but mice lacking Pax8 do not have defects in female reproductive tract development (Mansouri et al., 1998). It is known that Pax2 and Pax8 are functionally redundant (Carroll and Vize, 1999;Bouchard et al., 2002). In Xenopus, Lim1 and Pax8 synergistically induce pronephric tissues in kidney development (Carroll and Vize, 1999). Although the anterior region of the Müllerian duct is initially formed in Pax2 mutants, it is possible that Lim1 and Pax2/8 also cooperate to regulate Müllerian duct formation in mice. Currently, it is not clear if Lim1 expression in the Wolffian duct requires Pax2 function (Fig. 7C) because the Wolffian duct starts to regress at this stage (Torres et al., 1995).
Wnt4 is one of relatively few molecules that have been shown to be required for the initial steps of Müllerian duct formation. It was reported that there is no Müllerian duct in Wnt4 mutants by molecular expression analysis of the Müllerian duct epithelium markers, Wnt7a and Pax8 (Vainio et al., 1999). Interestingly, we showed that Lim1 is expressed in Müllerian duct precursor cells of Wnt4 mutants and that these precursor cells do not invaginate to form the Müllerian duct. This suggests that Lim1 expression in Müllerian duct precursor cells does not require Wnt4 function. It is possible that Lim1 may be required for specifying these Müllerian duct precursor cells acting genetically upstream of Wnt4.
It is noteworthy that Lim1, Pax2, Emx2 and Wnt4 are involved in Müllerian duct formation and are also required for the initial steps of kidney (metanephros) development (Stark et al., 1994;Shawlot et al., 1995;Torres et al., 1995;Miyamoto et al., 1997;Tsang et al., 2000). This infers that similar mechanisms may be functioning in Müllerian duct formation and kidney development.