Defects of urogenital development in mice lacking Emx2

The kidney has been widely exploited as a model system for the study of tissue interactions regulating organogenesis. Its development in mammals proceeds in three major stages: pronephros, mesonephros and metanephros. The nephric duct (Wolffian duct) develops in craniocaudal succession from the intermediate mesoderm and acts upon surrounding mesenchyme as an ‘inducer’ of epithelial transformation to nephric tubules (Saxén, 1987). The pronephric tubules, mesonephric tubules and the anterior portion of the Wolffian duct eventually degenerate, and it is the metanephros that becomes the permanent kidney. In metanephrogenesis, the metanephric blastema induces the sprouting of the ureteric bud from the caudal region of the Wolffian duct and the growth of this bud into the mesenchyme. The bud invades the mesenchyme and then induces epithelialization of the mesenchyme and its differentiation into the nephron. In turn this metanephric mesenchyme induces the branching of the ureteric bud. Thus, the permanent kidney develops by the reciprocal induction of the mesenchyme upon the epithelium and the epithelium upon the mesenchyme. Both are organized into functional units called uriniferous tubules, which are composed of duct-derived collecting system and mesenchyme-derived nephron (Grobstein, 1953, 1955; Saxén, 1987).

Information has accumulated especially with mutant mice about molecular events underlying the mutual interactions between the ureteric bud and mesenchyme (Bard et al., 1994; Sariola, 1996; Davis and Bard, 1996; Rothenpieler, 1996; Ekblom, 1996; Schofield and Boulter, 1996). In the null mutants of Pax-2, a transcriptional regulator of the paired-box family, the Wolffian duct develops only partially and metanephric development does not take place (Torres et al., 1995). The metanephrogenic mesenchyme deficient in Wilms tumor gene encoding a transcriptional factor with zinc finger motif, WT-1, lacks the ability to induce the formation of the ureteric bud (Kreidberg et al., 1993). The null mutants of c-Ret, a receptor tyrosine kinase, and of GDNF, a member of the transforming growth factor (TGF)-β superfamily, also fail to form ureteric bud (Schuchardt et. al., 1994, 1996; Moore et al., 1996; Sánchez et al., 1996; Pichel et al., 1996). In Danforth’s short tail mutants, the ureteric bud is initiated, but does not enter the metanephrogenic mesenchyme (Gluecksohn-Schoenheimer, 1943; Mesrobian and Sulik, 1992). Mice lacking Wnt-4, a secreted glycoprotein, do not form nephron (Stark et al., 1994). The disruption of BF-2 encoding a Winged Helix transcription factor reduces the rate of differentiation of the condensed mesenchyme into tubular epithelia as well as the rate of growth and branching of the ureter and collecting system (Hatini et al., 1996). Bmp-7 is another member of the TGF-β superfamily of secreted growth factors and, in its mutants, early metanephrogenesis is unaffected, but the mesenchyme cannot continue to proliferate, differentiate and survive (Dudley et al., 1995; Luo et al., 1995).

The homeobox genes Emx2 and Emx1 are mouse homologues of a Drosophila head gap gene ems (Simeone et al., 1992a,b). Homozygous Emx2 mutant mice display defects in development of the dorsal telencephalon that develops from the Emx2-positive and Emx1-negative domain in the palliochoroidial boundary (Yoshida et al., 1997). Emx2 mutant mice die due to failure of urogenital system development and we have focused our present study on this kidney defect.

Emx2 mutant mice were previously generated (Yoshida et al., 1997) and the genotype of each mouse or embryo was routinely determined on tail or yolk sac extracts by PCR (polymerase chain reaction) with the described primers (Yoshida et al., 1997). The mice were housed in an environmentally controlled room of the Animal Facility in Kumamoto University School of Medicine under the guidelines for recombinant and animal experiments of the School.

Fetus by crosses among heterozygous mice were collected in PBS and fixed in Bouin’s fixative for histological analysis or in 4% paraformaldehyde (PFA) solution for in situ hybridization analysis and germ cell analysis. For the histological and in situ hybridization analyses, embryos were embedded in paraffin wax (Paraplast) and sectioned at 8 and 10 μm thickness, respectively. Sections were stained with Haematoxylin-Eosin. In the analysis of germ cells, the fixed embryos were soaked in 30% sucrose in phosphate-buffered saline (PBS), embedded in OCT compound (TISSUE-TEK, USA) and frozen. Frozen samples were sectioned at 10 μm thickness with a microtome cryostat.

Sexual genotypes of mutant embryos were determined by PCR analysis with primers specific for the Y chromosome-encoded Sry gene as described (Hogan et al., 1994). PCR primers used were 5′-GAGAGCATGGAGGGCCAT-3′ and 5′-CCACTCCTCTGTGACACT-3′ to amplify a 200 base pair (bp) fragment. A fragment from the Zfy gene, which is present in both male and female genomes, was amplified as a control; PCR primers were 5′-GACTAGACATGTCT-TAACATCTGTCC-3′ and 5′-CCTATTGCATGGACTGCAGCT-TATG-3′ to detect a 120 bp fragment.

In situ hybridization analysis was performed using digoxigenin-labeled antisense riboprobe as described (Wilkinson, 1993). The probes used were as described for Emx2 (Yoshida et al., 1997), WT-1 (Kudoh et al., 1995), Pax-2 (Dressler et al., 1990), Lim1 (Barnes et al., 1994) and Wnt-4 (Stark et al., 1994). pKSmcRetXR is a plasmid for c-ret riboprobe synthesis and was kindly provided by Dr M. Yanagisawa. GDNF cDNA was isolated by PCR from a cDNA pool prepared with 9.5 dpc (days post coitus) ICR mouse embryos and subcloned into pBluescript SK. Sequence data of GDNF was obtained from GenBank (Accession Number; U37549).

Endogenous alkaline phosphatase activity of primordial germ cells (PGCs) was determined using the nitroblue tetrazolium reaction with frozen sections of 11.5 dpc embryos as described (Harlow and Lane, 1988).

Cultures were set up essentially as described by Saxén and Lehtonen (1987). The isolated metanephros was soaked in PBS containing 1 mM ethylendiamine tetraacetic acid (EDTA, pH 8.0) for 15 minutes, and the ureteric bud and metanephric mesenchyme were mechanically separated with a sharpened tungsten needle and co-cultured in minimal essential medium (MEM) supplemented with 10% fetal bovine serum.

The Emx2 mutants were previously generated (Yoshida et al., 1997). In brief, the gene was disrupted in front of its homeodomain by inserting neomycine phosphotransferase gene. The homozygous mutants obtained by intercrosses of heterozygotes died soon after birth. PCR analysis with primers specific for the Y chromosome-encoded Sry gene proved that both male and female mutants exhibited the same phenotype. They lacked kidneys, ureters, gonads and genital tracts completely; vas deferens, epididymis, ductulus efferentes and seminal vescicles were lost in male mutants, and oviducts, uterus and the upper part of the vagina in female (Fig. 1). Adrenal glands and bladder developed normally. No defects were found in the heterozygous urogenital system.

The development of metanephros was grossly normal till 11.5 dpc in the mutants. The metanephric blastema was apparently normal at 10.5 dpc (Fig. 2A,B), and the invasion of the ureteric bud to the metanephric blastema was found at 11.5 dpc mutants (Fig. 2C,D). However, the dilation of the tip of the ureteric bud was never found (Fig. 2D). In 12.5 and 13.5 dpc wild-type embryos, the invading ureteric bud branches several times, the metanephric mesenchyme cells are induced to transform into epithelial cells around each bud and the nephron is actively formed (Fig. 2E,G). Neither the branching of the ureteric bud nor the epithelial transformation of mesenchyme was observed in the mutant metanephros (Fig. 2F,H); pretubular cell aggregates, comma- or S-shaped bodies, were never observed in the mutant blastema. Instead, the invading ureteric bud had begun to degenerate at 12.5 dpc (Fig. 2F) and had disappeared in the metanephric blastema by 13.0 dpc (Fig. 2H). The size of the mutant blastema did not increase at 12.5 dpc (Fig. 2F), and a number of apoptotic cells with dark nuclear fragments were found at 13.0 dpc (Fig. 2H,I).

In mesonephros, the Wolffian duct developed normally along the urogenital ridge in 10.5 dpc homozygous mutant embryos (Fig. 3B). Mesonephric tubules adjacent to the Wolffian duct were also apparently normal at 10.5 dpc. The11.5 dpc mutant mesonephros was grossly normal (Fig. 3D), but degeneration of the Wolffian duct was found in several sites (Fig. 3E). Degeneration of the duct never occurs at this stage in the wild-type embryos (Fig. 3C). In wild-type 11.5 dpc embryos, the thickening of the epithelium on the coelomic surface of the urogenital ridge marks the first stage of the gonadal development (Fig. 3C). Thickening of the coelomic epithelium was not prominent in the 11.5 dpc mutants (Fig. 3D,E); gonadal cells were sparse and the coelomic surface was rough. At 13.0 dpc, the invaginations of the mesonephros and gonad into the coelom are prominent in the wild-type embryos (Fig. 3F), but these invaginations were very poor in the mutants (Fig. 3G). The PGCs migrate into the gonad from the endodermal region of the yolk sac at 11.5 dpc. Alkaline phosphatase staining showed that the migration of the PGCs to the genital ridge normally occurred in the mutants (Fig. 3H,I). In both male and female, the Müllerian duct normally develops in parallel with the Wolffian duct around 13.0 dpc (Fig. 3F), from which the oviducts, uterus and upper part of the vagina are formed in female. Müllerian duct was never formed in the mutants (Fig. 3G).

To assess the site affected by the Emx2 mutation, the Emx2 expression was examined in the developing kidney. At 8.5 dpc, this expression was detected in the nephrogenic cord (Fig. 4A) and, at 9.5 dpc, it was observed in the Wolffian duct (Fig. 4B). In the developing mesonephros of wild-type mouse embryos at 10.5 and 11.5 dpc, Emx2 transcripts were found in the Wolffian duct, mesonephric tubules and coelomic epithelium (Fig. 4C,D). The expression was not detectable in the mesenchyme cells of the mesonephros. In metanephros at 11.5 dpc, the invading ureteric bud hybridized with the Emx2 probe, whereas the surrounding mesenchyme including pretubular cell aggregates did not express Emx2 at all (Fig. 4E). At 12.5 dpc, Emx2 expression was also detected in the newly formed Müllerian duct (Fig. 4F). The expression was continuous in 13.5 dpc Wolffian duct, mesonephric tubules, the Müllerian duct and gonad (Fig. 4H). In the 12.5 and 13.5 dpc metanephros, Emx2 was expressed not only in the branched ureter, but also in an unique population of metanephric mesenchyme (Fig. 4G,H). Later, at 16.5 dpc, the Emx2 expression became confined to the cortical zone of the kidney, where new nephric tubules are being formed (Fig. 4I). At this stage, the expression in the collecting ducts of the medial zone was decreased.

Of note is that, in the process of epithelialization of the metanephric mesenchyme, Emx2 expression in the mesenchyme occurs at a later stage of epithelialization. WT-1 and GDNF are expressed in the metanephric blastema in advance of the invasion of ureteric bud (Fig. 6C; Armstrong et al., 1992), and their expressions at 11.5 dpc are found througout the metanephric mesenchyme with the invading ureteric bud (Fig. 6A,E; Armstrong et al., 1992; Sánchez et al., 1996). At subsequent stages, their expressions are found in the condensed mesenchyme surrounding the bud of each ureteric branch (Fig. 4J; Pelletier et al., 1991; Armstrong et al., 1992; Moore et al., 1996; Sánchez et al., 1996; Hellmich et al., 1996). Pax-2 expression is first observed in the metanephric mesenchyme when the ureteric bud invades (Fig. 5F; Dressler et al., 1990). Thereafter, it is continuously expressed in the condensed and epithelializing mesenchyme as well as ureter (Fig. 4K; Dressler et al., 1990; Dressler and Douglass, 1992). Wnt-4 expression is restricted to pretubular cell aggregates on both sides of the 11.5 dpc invading ureteric bud (Fig. 6G; Stark et al., 1994). At 12.0 and 13.5 dpc, it was found in an increasing number of pretubular cell aggregates, comma- and S-shaped bodies (Figs 4L, 6I; Stark et al., 1994). In contrast, Emx2 was not expressed in pretubular cell aggregates at 11.5 dpc (Fig. 4E), and was expressed at 12.5 and 13.5 dpc in more limited populations of mesenchyme that would be in a later stage of epithelialization (Fig. 4G,H).

As the next step in determining the defect caused by the Emx2 mutation, the expressions of several molecular marker genes during nephrogenesis were examined. Pax-2 is normally expressed in the Wolffian duct and mesonephric tubules (Fig. 5A,C; Dressler et al., 1990). Its expression at 10.5 dpc was apparently normal in the mutant mesonephros (Fig. 5B). At 11.5 dpc, however, the Pax-2 expression was lost in several portions even where the Wolffian duct and mesonephric tubules were morphologically normal (Fig. 5E); its expression in the neural tube was unaffected (Fig. 5D,E). In metanephros, Pax-2 is expressed not only in the ureteric bud but also in the mesenchyme (Fig. 5F; Dressler et al., 1990). In the Emx2 mutants, Pax-2 expression in the mesenchyme was normally found, but that in the ureteric bud was greatly reduced; it was faint at the tip of the bud and lost in the proximal part (Fig. 5G). Lim1 transcripts are also present in the wild-type ureteric bud (Fig. 5H; Fujii et al., 1994). Its expression was completely lost in the tip of the mutant ureteic bud and was weak in the proximal part of the bud (Fig. 5I). The c-ret is another epithelial marker (Fig. 5J,L,N; Pachinis et al., 1993; Tsuzuki et al., 1995). The c-ret expression in the mutant Wolffian duct was apparently normal at 10.5 dpc (Fig. 5K), but its expression in the 11.5 dpc mutant ureteric bud was greatly diminished (Fig. 5M). No c-ret expression was found in the tip of the 12.0 dpc mutant bud that had begun to degenerate (Fig. 5O). The WT-1 expression in the metanephric mesenchyme was normally found (Fig. 6A,B). The GDNF expression in the 10.5 dpc metanephric blastema before the invasion of the ureteric bud was normal (Fig. 6C,D), but the 11.5 dpc expression around the invading ureteric bud was greatly reduced in the mutant (Fig. 6E,F). The Wnt-4 expression in pretubular cell aggregates was never found in the mutant metanephric mesenchyme (Fig. 6H,J).

To determine whether it is the ureteric bud or the mesenchyme that is lesioned by the Emx2 mutation, ureteric bud and metanephric blastema were cultured in vitro (Fig. 7A,B; Grobstein, 1955). Neither epithelialization of the mesenchyme nor branching of the ureteric bud occurred in co-culture of the 11.5 dpc mutant ureteric bud and mesenchyme (Fig. 7C,D). In addition, when the mutant ureteric bud was co-cultured with the wild-type metanephric blastema, neither the epithelial transformation of the wild-type mesenchyme nor the branching of the mutant bud was observed (Fig. 7E,F). In contrast, when the isolated wild-type ureteric bud was co-cultured with the mutant metanephric blastema, epithelial transformation occurred in the mutant mesenchyme and the branching was induced in the wild-type ureteric bud (Fig. 7G,H), though the number of branchings was somewhat less than that in co-culture of wildtype counterparts (Fig. 7A,B). Spinal cord is also known to stimulate epithelial transformation of the metanephric blastema (Fig. 7I; Grobstein, 1955). Indeed, the isolated mutant metanephric mesenchyme was successfully epithelialized by the wild-type dorsal spinal cord (Fig. 7J).

In metanephrogenesis, the ureteric bud sprouts from the caudal region of the Wolffian duct and invades the metanephric mesenchyme, possibly guided by a chemotactic signal from the mesenchyme (Kreidberg et al., 1993; Moore et al., 1996; Sánchez et al., 1996; Pichel et al., 1996; Schuchardt et al., 1996). The invading ureteric bud activates the mesenchyme to transform into the epithelia (Grobstein, 1955). The tip of the bud dilates and, at both sides of this dilated tip, the bud induces the mesenchymal cells to form cell aggregates. These pretubular cell aggregates elongate into a ‘comma’ shape and then form a characteristic S-shaped body. The body fuses with the ureter yielding the nephron. At the same time, the growth and dichotomous branching of the ureteric bud is supported by the nephrogenic mesenchyme to give rise to the collecting duct (Saxén, 1987). The present analysis strongly suggests that the defect in metanephrogenesis by the Emx2 mutation resides in the ureteric bud, and the Emx2 function in the epithelium is crucial in the reciprocal induction with the mesenchyme for nephrogenesis.

The WT-1 is expressed in the metanephric blastema (Fig. 6A; Armstrong et al., 1992) and is believed to be involved in outgrowth of the ureteric bud (Kreidberg et al., 1993). Recently, GDNF-deficiency has also been reported to fail to induce the ureteric bud (Moore et al., 1996; Sánchez et al., 1996, Pichel et al., 1996); GDNF is expressed in the mesenchyme of metanephros (Fig. 6C, E; Moore et al., 1996; Sánchez et al., 1996; Hellmich et al., 1996). c-Ret activity is stimulated by GDNF (Durbec et al., 1996; Jing et al., 1996; Trupp et al., 1996; Treanor et al., 1996). The receptor type tyrosine kinase is initially expressed in the Wolffian duct and, in the developing metanephros, its expression is gradually restricted to the tip of the ureteric bud (Fig. 5N; Pachinis et al., 1993; Tsuzuki et al., 1995). c-ret knockout mice fail to form the ureteric bud (Schuchardt et al., 1994). Thus the WT-1, GDNF and c-ret mutant phenotypes are closely related to each other; it is most plausible that the GDNF secretion from the mesenchyme is regulated by WT-1 and its signaling in the ureteric bud epithelium is mediated by c-Ret. In Emx2 mutants, the WT-1 expression was unaffected. The GDNF expression appeared also normal in the metanephric blastema before the invasion of ureteric bud (Fig. 6D). The c-ret was also expressed normally in the Wolffian duct before the budding of the bud (Fig. 5K). In all the Emx2 mutants, the ureteric bud was induced and invaded the metanephrogenic blastema. Thus, defects are not present in the inductive capability of the mesenchyme or in the reactivity of the epithelium for the initial growth of ureteric bud.

Pax-2 is one of the earliest responding genes in the metanephric mesenchyme upon induction by the ureteric bud (Fig. 5F; Dressler et al., 1990) and it is continuously expressed throughout the process of epithelialization. The antisense oligonucleotides to Pax-2 inhibits the condensation of mesenchyme cells in culture (Rothenpieler and Dressler, 1993). The Emx2 mutant bud could induce Pax-2 expression in the mesenchyme. In contrast, Wnt-4 expression that normally occurs in pretubular cell aggregates was never found in the Emx2 mutant metanephros. None of the mutants examined showed any histological evidence of mesenchymal aggregation, nor did we ever observe the comma-shaped or S-shaped body. Normal Emx2 expression is also found in mesenchyme cells, although here it occurs at a later stage of epithelialization, possibly in the S-shaped body, and the Emx2 mutant mesenchyme had deteriorated much earlier than this stage. Thus, the Emx2 mutant ureteric bud has a signal to induce Pax-2 expression in the mesenchyme, but lacks a signal to induce further epithelialization there.

The Wnt-4 mutant metanephros fails to form pretubular cell aggregates but undergoes considerable branching of the ureter (Stark et al., 1994), whereas the branching was never found in the Emx2 mutant metanephros. GDNF/c-Ret signaling is thought essential not only in the initial growth of the ureteric bud but also in branching of the bud (Schuchardt et al., 1996). In Emx2 mutant, GDNF and c-ret expressions were reduced in the mesenchyme and ureteric bud, respectively, when the bud invaded the blastema. A signal from the bud invading the metanephric blastema may be necessary to maintain GDNF expression in the blastema; this signal may be lost in the Emx2 mutant ureteric bud. At the same time, Emx2 may act to retain c-ret expression in the invading bud.

It is known that, in the absence of a signal(s) from the ureteric bud, metanephric mesenchyme cells not only fail to differentiate but they undergo apoptosis (Koseki et al., 1992; Kreidberg et al., 1993; Coles et al., 1993; Herzlinger, 1995; Camp and Martin, 1996). In the WT-1 and c-ret mutants, all of which fail to form ureteric bud, the metanephric mesenchyme undergoes prominent apoptosis (Kreidberg et al., 1993; Schuchardt et al., 1996). In contrast, the apoptoic figures are not prominent in Wnt-4 mutants that fail to epithelialize mesenchyme but have normal ureteric epithelium (Stark et al., 1994). These findings implicate a survival factor(s) from the ureteric bud that inhibits apoptosis of the mesenchyme (Camp and Martin, 1996). In Emx-2 mutants, the ureteric bud invaded the mesenchyme, but apoptotic figures were prominent. The Emx2 mutant ureteric bud may also be unable to generate the survival factor.

The results of co-culture experiments are also consistent with the deterioration of the Emx2 mutant ureteric bud, its mesenchyme being normal. The Emx2 mutant ureteric bud did not induce epithelialization of normal mesenchyme, whereas epithelialization of the mutant mesenchyme was induced by the isolated wild-type ureteric bud and dorsal spinal cord. Branching of the wild-type ureteric bud was also induced by the mutant mesenchyme, while it was not induced in the mutant ureteric bud by wild-type mesenchyme. In the wild-type ureteric bud growing into the mesenchyme not only c-ret but also Pax-2 and Lim1 are expressed. In Emx2 mutants, the expressions of these genes in the invading ureteric bud were greatly reduced or completely lost. The ureteric bud was lost in the Emx2 mutant embryos soon after its invasion into the metanephric blastema. The most conspicuous morphological feature was the non-dilatation of the tip of the invading bud, which is an early branching feature and may be no different from subsequent branchings, regulated by GDNF/c-Ret signaling. Alternatively, it might require a specific process in which Emx2 has a crucial function so that the gene is essential for all the subsequent steps regulating the epithelial functions of expressions of c-Ret, Pax-2 and Lim1 as well as signalings to the mesenchyme.

The Emx2 defects were apparent at 11.5 dpc not only in metanephros but also in mesonephros; no defects were detected at 10.5 dpc either morphologically or molecularly. Degeneration of the Wolffian duct and mesonephric tubules was found in several sites at 11.5 dpc. Studies of Pax-2 expression have shown that the degeneration progressed more widely in the epithelia that appeared morphologically normal. In contrast to Pax-2, Emx2 (cf. Fig. 4I) as well as Lim1 (Fujii et al., 1994) are expressed only in earlier phases of the nephrogenic epithelium, suggesting that they are not essential in the epithelium once established. However, in WT-1, c-ret and GDNF mutants that cannot form ureteric bud, Wolffian duct degeneration is not accelerated (Kreidberg et al., 1993; Schuchardt et al., 1994; Moore et al., 1996; Sánchez et al., 1996; Pichel et al., 1996), suggesting that the loss of the ureteric bud in the Emx2 mutant itself is not the cause of the mesonephric degeneration. It is possible that Emx2 acts to sustain the overall epithelium in the earlier stages, not specifically the ureteric bud. Then the question would be raised as to why the deterioration happened at the stage when the ureteric bud invades the metanephric blastema.

In several kidney mutants, the homozygous phenotype exhibits considerable variability and the defects were also found by haplo-insufficiency. Most GDNF homozygous mutants do not develop ureteric bud as noted above, but some do, while some heterozygotes do not (Moore et al., 1996; Sánchez et al., 1996; Pichel et al., 1996). Only one third of the c-ret newborn mutants display a complete absence of ureter and kidney bilaterally, and one-tenth have kidney rudiments bilaterally (Schuchardt et al., 1994, 1996). Several Wnt-4-deficient metanephroi develop limited small cell aggregates (Stark et al., 1994). The variability in phenotype may implicate the presence of closely related molecules with functions that overlap. In contrast, the Emx2 mutant phenotype displayed little variability. The Emx1, a cognate of Emx2, is also expressed in the kidney (Briata et al., 1996; our unpublished data). No defects were found in the kidney development by the Emx1 mutation that we previously generated, however (unpublished data).

Relatively little is known about the genetic basis of gonadal development at early stages when sexual differentiation by SRY has not yet occurred. WT-1 and an orphan nuclear receptor SF-1/Ftz-F1 are the only two genes known to be involved in this process (Kreidberg et al., 1993; Luo et al., 1994; Jiménez et al. 1996). The defect in gonadal development by WT-1 deficiency is apparent at 11.5 dpc as a markedly reduced thickening of the epithelium (Kreidberg et al., 1993). Ftz-F1-deficient phenotype is characterized by apoptosis in both the epithelium and mesenchyme of the gonadal primordium at 12.0 dpc when sexual differentiation normally becomes manifest. Emx2 transcripts are observed in the developing gonad. The thickening of the coelomic epithelium was poor at 11.5 dpc, although PGCs migrated from the endodermal region of the allantois to the genital ridge, and subsequent gonadal development was impaired in the Emx2 mutant embryos. The interaction between the mesonephros and gonad has to be kept in mind. Mesonephric mesenchyme cells contribute to the gonadal development (Buehr et al., 1993), and the mesenchyme was poor in Emx2 mutants. The Emx2 may function in proliferation, differentiation and/or survival of gonadal cells, subjects for future studies.

We thank Mr Naoki Takeda for production of mutant mice, Dr P. Gruss, Dr T. Akiyama, Dr B. Hogan, Dr S. Vainio, Dr A. P. McMahon and Dr M. Yanagisawa for providing us Pax-2, WT-1, Lim1, Wnt-4 and c-ret probes respectively. We are grateful to the Laboratory Animal Research Center of Kumamoto University School of Medicine for housing of the mice. This work was supported in part by grants-in-aid from the Ministry of Education, Science and Culture of Japan (Specially Promoted Research), the Science and Technology Agency, Japan and the Ministry of Public Welfare, Japan.