Gene dosage-dependent effects of the Hoxa-13 and Hoxd-13 mutations on morphogenesis of the terminal parts of the digestive and urogenital tracts

The 39 murine Hox genes belong to four linkage groups (the HoxA, HoxB, HoxC and HoxD clusters), related to the Drosophila homeotic gene complexes (McGinnis and Krumlauf, 1992; Scott, 1992; Duboule, 1994; and references therein). All genes of a given murine cluster are transcribed from the same DNA strand and are expressed in successive, but partially overlapping, domains along the anteroposterior (rostrocaudal) axis of the mouse embryo, according to their 3′ to 5′ order. Hox gene paralogues related to Drosophila Abdominal-B are also sequentially expressed in the developing limbs (Dollé and Duboule, 1993; Haack and Gruss, 1993; and refs. therein). Thus, expression of the most 5′-located paralogues Hoxa-13 and Hoxd-13 is restricted to the distal region of the developing limbs (autopod) and to the most posterior regions along the embryo axis. In particular, both genes are strongly expressed in the mesenchyme of the genital tubercle, as well as in the hindgut and cloacal region, from which the terminal part of the intestine (rectum) and urogenital tracts will develop (Dollé et al., 1991a,b; see below). The cognate zebrafish and chicken genes are similarly expressed in the terminal region of the developing hindgut (Yokouchi et al., 1995a; van der Hoeven et al., 1996). In the mouse, the expression of these genes persists until, at least, perinatal stages (Dollé et al., 1991a; Podlasek et al., 1997; see below).

The Hoxa-13 and Hoxd-13 genes have been knocked-out to analyze their developmental functions (Dollé et al., 1993; Fromental-Ramain et al., 1996). Compound mutant genotypes were produced by intercrosses, and we have previously shown that both genes act in a partly redundant manner for patterning the limb autopod (Fromental-Ramain et al., 1996). Several lines of evidence suggest that the same genes may control the development of posterior digestive and/or genito-urinary structures: (i) inactivation of other Abdominal-B-related genes have resulted in defects of kidney (in Hoxa-11/Hoxd-11 compound mutants; Davis et al., 1995) or urogenital tract morphogenesis (in Hoxa-10 and Hoxa-11 mutants; Hsieh-Li et al., 1995; Satokata et al., 1995; Benson et al., 1996; Gendron et al., 1997); (ii) Hoxd-13^−/− males are hypofertile and display subtle alterations of the penian bone, the accessory sex glands and the smooth muscle layer of the anal sphincter (Dollé et al., 1993; Kondo et al., 1996; Podlasek et al., 1997); (iii) the human handfoot-genital hereditary syndrome was shown to result from disruption of the HOXA13 gene (Mortlock and Innis, 1997). In mice, the homozygous disruption of Hoxa-13 results in fetal lethality between 11.5 and 15.5 days post-coitum (dpc) (Fromental-Ramain et al., 1996), in contrast to other Abdominal-B-related gene mutations, which are not lethal.

We report here urogenital and rectal abnormalities in Hoxa-13/Hoxd-13 compound mutant mice. We show that the Hoxa-13^+/−/Hoxd-13^−/− genotype results in specific abnormalities of the male and female genital tracts, lower urinary tract and rectum, leading to postnatal death of some of the mutants. Furthermore, compound homozygous mutant (Hoxa-13^−/−/Hoxd-13^−/−) fetuses show virtually no development of the genital tubercle and the caudal parts of the hindgut (cloaca) and mesonephric ducts. These data show that the Hoxa-13 and Hoxd-13 genes are essential for the morphogenesis of terminal structures of the embryonic trunk.

Pregnant females from natural overnight matings (morning of vaginal plug was considered as 0.5 days post-coitum [dpc]) were killed by cervical dislocation. The embryos were explanted in phosphate-buffered saline, placed in moulds containing embedding medium and frozen on the surface of dry ice. Cryosectioning, riboprobe synthesis and in situ hybridizations were performed as described previously (Décimo et al., 1995). The DNA template used to generate the Hoxd-13 riboprobe was described previously (Dollé et al., 1991b). A 800 bp SpeI-EcoRI fragment of the Hoxa-13 gene, located 3’ to the homeobox, was used to generate the corresponding riboprobe.

The establishment of Hoxd-13 and Hoxa-13 mutant lines has already been described (Dollé et al., 1993; Fromental-Ramain et al., 1996). Mutant lines were maintained in a mixed (129/Sv-C57Bl/6) or inbred (129/Sv) background. To produce double heterozygous mutants, Hoxa-13^+/− mutant mice were crossed to Hoxd-13^+/− or Hoxd-13^−/− mutants. The resulting double heterozygotes were then intercrossed to obtain other genotypic combinations. Genotyping was performed by Southern blot analysis of tail tip (for mice) or yolk sac membrane (for embryos and fetuses) DNA, as described in the above references.

For analysis of adult organs in situ, animals were anaesthetized and fixed by intracardiac perfusion of formalin (10% formadehyde in PBS). After dissection and photography of the urogenital tract, the individual organs were fixed in Bouin’s solution and processed for histological analysis (see below).

Embryos at 12.5 to 14.5 dpc were fixed in Bouin’s solution for at least 24 hours, dehydrated, embedded in paraffin, serially sectioned (7 pm thickness) and stained with hematoxylin and eosin. Newborn and 1- or 2-week-old animals were skinned, fixed in Bouin’s solution for at least one week, decalcified in Jenkin’s solution, dehydrated and embedded in paraffin, sectioned and stained with Groat’s hematoxylin and Mallory’s trichrome (Mark et al., 1993). Adult organs were fixed in Bouin’s solution for at least 2 days and treated as described for the embryos.

The expression of Hoxd-13 (Oefelstein et al., 1996; Podlasek et al., 1997), but not of Hoxa-13, has been studied in detail in developing mouse genito-urinary structures. In order to investigate whether both genes may be coexpressed at these levels, we performed comparative in situ hybridization on sections of 12.5 dpc and 14.5 dpc fetuses, as well as newborn animals. At 12.5 dpc, both genes were expressed in the mesenchyme of the genital tubercle, the mesenchyme surrounding the urogenital sinus and, to a lower extent, in the sinus epithelium (Fig. 1A). However, the distributions of both transcript species differed slightly, Hoxd-13 transcripts being more abundant towards the distal tip of the genital tubercle (Fig. 1A). At 14.5 dpc, both genes remained coexpressed in the genital tubercle and around the caudal portion of the Wolffian and Müllerian ducts (Fig. 1B, and data not shown). Coexpression was also seen in the wall of the developing urinary bladder. Interestingly, only Hoxa-13 transcripts extended within the wall of the umbilical arteries (Fig. 1B).

In newborn males, both genes were most strongly expressed in the developing prostate and seminal vesicles (data not shown; see also Oefelstein et al., 1996; Podlasek et al., 1997). In females, expression of both genes was strong in the developing uterine cervix and vagina, and weaker in the urethral epithelium and mesenchyme (Fig. 1C). Only Hoxa-13 transcripts could be clearly detected in the caudal portion of the ureters, including at the level of their entry sites into the bladder (Fig. 1B,C). Throughout development, Hoxa-13 transcripts were detected in the epithelium, but not the mesenchyme, of the hindgut and rectum, whereas Hoxa-13 transcripts were markedly expressed in both the epithelial and mesenchymal (developing muscular) layers of the rectum (Fig. 1B,C).

Hoxd-13^−/− mutants are hypofertile; the size of their accessory sex glands is either normal or slightly diminished, but their ductal morphogenesis is deficient (Podlasek et al., 1997). The majority of Hoxa-13⁺/⁻/Hoxd-13^−/− compound mutant males died within a few weeks post-partum, and all surviving males were sterile. The seminal vesicles of these mutants were severely hypoplastic (SV, compare Fig. 2A,B and C,D). The dorsal prostate was present, but its cranial counterpart (coagulating gland) was not detectable (CG, compare Fig. 2B and D). Moreover, both the bulbourethral (not shown) and preputial glands (PG) were missing (compare Fig. 2A and C). Note that agenesis of the bulbourethral glands occurs in half of the Hoxd-13^−/− mutants, but that the preputial glands are present, although hypoplastic, in these mutants (Podlasek et al., 1997; our observations). In contrast to these deficiencies of accessory sex glands, the testis, epididymis and ductus deferens of Hoxa-13^+/−/Hoxd-13^−/− mutants were morphologically normal (compare Fig. 2A and C).

The development of genital structures was analyzed histologically in newborn, 1-week-old and 2-week-old Hoxa-13^+/−/Hoxd-13^−/− males (n=8).

As expected, the preputial glands were absent (not shown) and the prostatic epithelial buds were poorly developed (compare Fig. 3A and B). However, the seminal vesicle primordia appeared reasonably well developed in newborns (not shown), indicating that growth deficiency of these primordia occurs postnatally. At all these stages, the developing erectile tissues (corpora cavernosa) were clearly hypoplastic (not shown).

All Hoxa-13^+/−/Hoxd-13^−/− females were sterile. Two (out of ∼20 examined) Hoxa-13^+/−/Hoxd-13^−/− females had abnormal external genitalia. In one of these, the spacing between genitalia and anus was reduced (compare Fig. 4A and B). Histological analysis further showed that the caudal portions of the urethra and vagina ended together, instead of being separated by a cutaneous fold (compare Fig. 4C and D). The other female displayed a common ending of the vagina and anus, which resulted in uterine infection (not shown).

The female genital tract was analyzed histologically in newborn, 1-week-old and 2-week-old compound mutants (n=6). One newborn displayed agenesis of the caudal portion of the right uterine horn (compare Fig. 5A and B), the cervix and the cranial vagina (compare Fig. 5C and D). These defects probably result from agenesis of the terminal portion of the right Müllerian duct. Vaginal anomalies were also seen in the 1-week-old mutant females. The caudal portion of the vagina remained connected to the urethra (compare Fig. 6B and E), as it is normally at birth, and its cranial portion was septated (compare Fig. 6C and F). These malformations are likely to result from improper separation of the embryonic vaginal plate from the urogenital sinus and from insufficient fusion of the Müllerian ducts, respectively (see Discussion). As described in males, mutant females also showed agenesis of the preputial glands and very poorly developed erectile tissues (corpora cavernosa) (compare Fig. 6A and D).

Several Hoxa-13^+/−/Hoxd-13^−/− adult mutants showed swollen abdomens, which at autopsy were found to result from severe hydronephrosis (dilatation of renal cavities). More or less severe dilatation of renal cavities was seen in most (8 out of 12) of the young (newborn to 2-week-old) mutant animals of both sexes, sometimes unilaterally with a predominance on the left side (Fig. 3C and data not shown). The ipsilateral ureter(s) were also dilated, both outside and within the urinary bladder cavity (Fig. 3B). There was no sign of urethral obstruction, suggesting that the obstacle to urine excretion may take place at the level of the entry site of the ureter(s) into the bladder (see Discussion).

Ageing Hoxd-13^−/− males eventually display rectal prolapses, likely resulting from a deficiency in the outer smooth muscle layer already seen in young mutants of both sexes, whereas the rectal epithelium remains normal (Kondo et al., 1996). Hoxa-13^+/−/Hoxd-13^−/− newborn mutants displayed dilatation of the rectum with abnormalities of the epithelial layer, which lacked rectal glands, and of the smooth muscle layer, which appeared abnormally distant from the mucosa and interrupted in some areas (compare Fig. 7A and B). Moreover, one (out of 14 analyzed) newborn showed an almost complete absence of the terminal part of the rectum (compare Fig. 7C and D). Hoxa-13 transcripts were detected by in situ hybridization in the epithelial, but not the muscle layer of the differentiating rectum (see Fig. 1). The rectal defects seen in compound mutants suggest, however, that Hoxa-13 is expressed at low levels in the muscle layer and that its haploinsufficiency potentiates the muscular defect of Hoxd-1^−/−- mutants.

We looked for abnormalities of the urinary and genital systems in Hoxa-13^−/− fetuses at 13.5 and 14.5 dpc. Note that the fraction of the homozygous mutants that were not dead in utero (n=12) may correspond to the less expressive phenotypes. At 14.5 dpc, the urogenital tract is still at the undifferentiated stage: both the mesonephric (Wolffian) and paramesonephric (Müllerian) ducts are present in WT embryos of both sexes (Kaufman, 1992, and refs. therein; Fig. 8A,B). The caudal portion of the Müllerian ducts was consistently lacking in Hoxa-13^−/−- mutants (Fig. 8D,E). This abnormality appeared to be correlated with an abnormal location of the ureter extremities. During normal development, the sprouting of the ureteric buds from the Wolffian ducts occurs rostrally to the urogenital sinus, and the Wolffian duct/ureter junction eventually becomes incorporated, together with the caudal part of the Wolffian ducts, in the growing sinus wall (Larsen, 1993, and refs. therein). At 13.5 dpc, the Wolffian duct/ureter junction has reached the urogenital sinus in WT fetuses (not shown). In contrast, this junction remained abnormally distant from the sinus in Hoxa-13^−/− fetuses, both at 13.5 and 14.5 dpc (Fig. 8D,E and data not shown). We therefore suggest that the rostral-to-caudal progression of Müllerian duct formation is either arrested or markedly delayed in Hoxa-13^−/− mutants, correlated with the insufficient displacement of the ureter openings towards the urogenital sinus. In addition, the entire urogenital sinus appeared hypoplastic in Hoxa-13^−/− fetuses. In particular, its cranial extension, corresponding to the urinary bladder anlage, was missing (compare Fig. 8C and F).

Whereas some Hox gene mutations are lethal at birth (e.g. Chisaka and Capecchi, 1991; Rijli et al., 1993), the Hoxa-13 mutation is the only one to be lethal at fetal stages (i.e. between 11.5 and 15.5 dpc; Fromental-Ramain et al., 1996). There are two well-established causes of mid-gestational lethality: cardiac failure (e.g. Kastner et al., 1994; Kwee et al., 1995; Chen et al., 1994) and defects of the chorioallantoic placenta (Kwee et al., 1995; Gurtner et al., 1995). Hoxa-13 transcripts are not detectable in the developing heart and Hoxa-13^−/− fetuses do not display overt cardiac defects (data not shown). Although Hoxa-13 is weakly expressed in the chorioallantoic placenta, this organ was histologically normal in Hoxa-13^−/− mutants (data not shown). However, about one third of Hoxa-13^−/− fetuses showed marked stenosis of one umbilical artery at the level of its intraabdominal portion (i.e. along the wall of the abnormal urogenital sinus) (Fig. 8F, and data not shown). Whereas stenosis of one umbilical artery usually occurs during late fetal development in WT fetuses (Kaufman, 1992), the stenosed arteries of Hoxa-13^−/− mutants were seen at stages when both arteries are of comparable diameters in WT fetuses (i.e. at 13.5 dpc; compare Fig. 8C and F). As the viable Hoxa-13^−/− fetuses may exhibit milder phenotypic abnormalities, we suggest that premature or bilateral stenosis of the umbilical arteries, resulting in insufficient blood flow towards the placenta, could account for the death of Hoxa-13^−/− mutants at various stages of fetal development.

We have previously shown that disruption of all Hoxa-13 and Hoxd-13 alleles results in much more severe defects in developing limbs than those occuring when at least one allele remains functional (Fromental-Ramain et al., 1996). We therefore investigated whether such synergistic mutational effects could also occur in the developing digestive or urogenital apparatus. Both Hoxa-13 and Hoxd-13 genes are expressed in the developing genital tubercle, which gives rise to the penis or clitoris (see Fig. 1). While all genotype combinations with at least one (out of four) functional allele resulted in morphologically normal genital tubercles, this structure was absent in double homozygous (Hoxa-13^−/−/Hoxd-13^−/−) mutants (Fig. 9A–C, and data not shown). Additional abnormalities were investigated histologically in 12.5 and 13.5 dpc compound mutant fetuses.

In 12.5 dpc WT specimens, the terminal hindgut (presumptive rectum) and urogenital sinus are separated by the urorectal septum and two regions can be distinguished in the urogenital sinus: a rostral portion, the presumptive bladder, alongside of which pass both umbilical arteries (Fig. 9G), and a caudal ‘definitive’ sinus or presumptive urethra (Fig. 9D). The entry points of the mesonephric (Wolffian) ducts mark the limit between the two regions. Strikingly, the Hoxa-13^−/−/Hoxd-13^−/− mutants (n=2) showed no distinct hindgut and urogenital sinus. Instead, the Wolffian ducts opened in a small dilated cavity (cloaca, * in Fig. 9F) linked rostrally to the herniated midgut (Fig. 9I). As there was no bladder anlage, the umbilical arteries were buried in the abdominal wall (Fig. 9I). In addition, the Wolffian duct/ureter junction remained far too rostral, i.e. close to the metanephroi (compare Fig. 9D and F, which correspond to sections at the level of the junction in the WT embryo, and Fig. 9G and I, which are sections at more rostral locations). More rostrally, the metanephroi and genital ridges were normally developed (data not shown). The same abnormal features were seen in a 13.5 dpc double null mutant (data not shown), indicating that there has been no significant progress in the morphogenesis of terminal urogenital and digestive structures.

The other compound mutant genotypes produced less severe alterations. Hoxa-13^−/−/Hoxd-13^+/− mutants (n=3) had histologically normal genital tubercles (Fig. 9B) and a proper separation between hindgut and urogenital sinus (Fig. 9E). However, the entire urogenital sinus was rudimentary and lacked a presumptive urinary bladder primordium (Fig. 9E,H). Consequently, the umbilical arteries were buried in the abdominal wall, as in double null mutants (Fig. 9H). In addition, the ureter and mesonephric duct junction occurred too rostrally (compare Fig. 9E,H and D,G). Hoxa-13^+/−/Hoxd-13^−/−mutants (n=2) showed somewhat milder defects of the urogenital sinus (data not shown).

Hoxa-13 and Hoxd-13 genes have two additional paralogous genes, Hoxb-13 (Zeltser et al., 1996) and Hoxc-13 (Peterson et al., 1994). Hoxb-13 is not expressed in the secondary axes of the body, i.e. those of the limbs and the genital tubercle (Zeltser et al., 1996), and Hoxc-13 expression in the genital region has not yet been investigated. The genital eminence is lacking in Hoxa-13^−/−/Hoxd-13^−/− double null fetuses, but is morphologically normal in other compound mutants with at least one functional Hoxa-13 or Hoxd-13 allele, indicating that these genes are indispensable (but act in a redundant manner) for the induction and/or the growth of the genital tubercle. Hoxc-13, if expressed in this structure, cannot functionally compensate for their inactivation. The most 5′-located genes in the HoxD cluster (Hoxd-11 to Hoxd-13) are coexpressed in the developing genital tubercle, Hoxd-13 showing the highest expression levels (Dollé et al., 1991a). It is unclear whether expression of Hoxd-11 and Hoxd-12 in the developing genital tubercle has any functional significance since (i) in contrast to Hoxd-13^−/− mice, Hoxd-11^{−/− -}and Hoxd-12^−/− mutants exhibit no detectable abnormality of genital tubercle derivatives and (ii) these genes are apparently not functionally redundant with Hoxa-13 and Hoxd-13. Compound mutations involving Hoxd-11 and Hoxd-12 will reveal whether these genes could be involved in some morphogenetic events in genitalia.

Recent studies have suggested that Hox genes may act by controlling local rates of cell proliferation (Davis and Capecchi, 1996; Zàkàny and Duboule, 1996; Goff and Tabin, 1997) and interactions between a homeobox gene and cell cycle regulators have been shown (Kawabe et al., 1997). Our results are consistent with such a link between Hox genes and cell proliferation, since the lack of genital tubercle could reflect a decrease or arrest of cell proliferation. Interestingly, Hoxa-13^−/−/Hoxd-13^−/− mutant fetuses exhibit drastic alterations of the forelimb and hindlimb extremities (autopods), which are severely truncated and show almost no patterning of preskel-etal elements (Fromental-Ramain et al., 1996). This phenotype may also result from a reduction of cell proliferation in the early autopod, which suggests that Hoxa-13 and Hoxd-13 could exert the same basic function in all secondary axes of the body. It has been previously proposed that the expression of the most 5′-located HoxA and HoxD genes in the autopodal region of the limb buds allowed the evolution of the distal limb skeleton in ancestral tetrapods (Sordino et al., 1995; Sordino and Duboule, 1996). Our data suggest that the expression of the same genes around the extremity of the urogenital duct system has been instrumental for the emergence of the external genital structures indispensable for the efficient internal fertilization that became a requisite for terrestrial life.

The developing digestive and urogenital duct systems of mammalian embryos initially share a common posterior outlet, the cloaca, which is partitioned into a ventral primitive urogenital sinus and a dorsal rectum by the growth of the urorectal septum. Expression studies in chick embryos have suggested that Abdominal-B-related HoxA genes could be involved in the regional specification of the proctodeal region (Roberts et al.,1995; Yokouchi et al., 1995a). Analyses in zebrafish (Danio rerio) embryos further suggested a role of Hoxa-13 and Hoxd-13 in the morphogenesis of the proctodeal area: Hoxd-13 was shown to be strongly expressed in the developing genitalia (van der Hoeven et al., 1996) and Hoxa-13 exhibits a peculiar expression pattern in the mesodermal cells separating the posterior extremities of the reproductive, excretory and digestive ducts, thus suggesting that it could play a role in the disappearance of the cloaca and the building of more complex terminal structures (Sordino et al., 1996). Our analysis of Hoxa-13 and Hoxd-13 compound mutant mice points to a crucial role of both genes in the morphogenesis of the extremities of the digestive and urogenital tracts. In double null mutant fetuses, there is no partition of the cloaca and, therefore, no urogenital sinus is formed. In Hoxa-13^−/−/Hoxd-13^+/− fetuses, the splitting of the cloaca occurs, but the development of the urogenital sinus is subsequently arrested.

During normal development, the Müllerian ducts grow caudally along the Wolffian ducts to reach the developing urogenital sinus. Concomitantly with the downgrowth of the Müllerian ducts, the distal parts of the Wolffian ducts and the ureters become incorporated in the dorsal wall of the sinus. In Hoxa-13^−/− embryos, the Müllerian ducts fail to grow caudally. The same abnormality is observed in Hoxa-13^−/−/Hoxd-13^+/− and Hoxa-13^−/−/Hoxd-13^−/− mutants. Moreover, in these mutants, the bifurcation point of the ureters from the Wolffian ducts remains too rostral. Thus, the morphogenetic movements in the caudal portion of the excretory and genital duct system are impaired in these mutants. By analogy with the effect of the compound mutation on genital tubercle development, these abnormalities could result from altered rates of cell proliferation in the region of the hindgut and urogenital sinus. Alternatively, and as previously suggested (Yokouchi et al., 1995b), altered cell adhesion properties could explain improper morphogenesis of the urogenital duct system and of the urorectal septum.

Hox gene targeted mutations can be helpful to understand how novel structures have evolved in higher vertebrates. Hoxa-13/Hoxd-13 double homozygous mutants show almost no patterning of the limb autopod, except for a single condensed ray that could correspond to a truncated ‘metapterygial axis’ (Fromental-Ramain et al., 1996) and reflect an atavistic situation, present before the divergence between fishes and tetrapods. We report here that the same mutants have no genital tubercle and a persistent terminal cloacal cavity in which the two Wolffian ducts open. In chondrichthyans fishes, the cloaca is not a transient embryonic structure but persists in adults as a common outlet of the digestive, urinary and genital ducts (Romer and Parsons, 1986; Walker, 1987). A cloaca is also present in certain adult tetrapods (amphibians, reptiles and monotreme mammals; Romer and Parsons, 1986). In contrast, full partition of the adult urogenital and alimentary duct extremities is achieved in teleost fishes and in placental mammals. It would be of obvious interest to investigate whether Hoxa-13 and Hoxd-13 are expressed and functional in the proctodeal region of species that display an adult cloaca.

Most of the Hoxa-13^+/−/Hoxd-13^−/− mutants are viable, and their analysis at postnatal stages revealed specific alterations of the rectum, urinary and genital tracts. Some of these correspond to more severe versions of the Hoxd-13^−/− postnatal abnormalities (Kondo et al., 1996; Podlasek et al., 1997), while others are not seen in Hoxd-13^−/− mutants.

Both male and female Hoxa-13^+/−/Hoxd-13^−/− mutants exhibited genital abnormalities, which affected derivatives of the urogenital sinus and of either the Wolffian ducts in males or Müllerian ducts in females. The hypoplasia of some male accessory sex glands (seminal vesicles, coagulating gland) can be considered as more severe Hoxd-13^−/− defects (Podlasek et al., 1997), whereas other defects of compound mutants (hypoplasia of erectile tissue, absence of preputial glands) have no counterpart in Hoxd-13^−/− mutants. Hoxa-13^+/−/Hoxd-13^−/− mutant females display anomalies equivalent to those seen in males (e.g. the lack of erectile tissues), as well as vaginal abnormalities. Whether the vagina derives from the caudal Müllerian ducts, the urogenital sinus epithelium, or both of these structures, has been a matter of controversy (see Koff, 1933, Witschi, 1970). The most recent observations suggest that the vagina could have a dual origin, its cranial part deriving from the Müllerian ducts and its caudal part from the urogenital sinus (Juillard, 1972), or that it could be exclusively formed by a caudal extension of the Müllerian ducts (Thiedeman, 1987). Hoxa-13^+/−/Hoxd-13^−/− females showed a separation of the cranial region of the vagina in two horns, while its caudal part remained abnormally connected to the urethral groove, thus supporting a dual embryological origin of the vagina: the abnormal cranial region could result from a lack of mid-fusion of the Müllerian ducts, while the altered caudal region could reflect an abnormal partitioning of the urogenital sinus.

In addition, most Hoxa-13^+/−/Hoxd-13^−/− animals developed severe uretero-hydronephrosis (dilatation of renal and ureteric cavities) which was probably the cause of postnatal lethality. Since the ureters were dilated within the wall of the urinary bladder, and by analogy with a human malformative syndrome involving HOXA13 (see below), the uretero-hydronephrosis could result from a misposition of the caudal end of the Wolffian duct(s) within the posterior wall of the urogenital sinus during late fetal development.

Hox gene mutations have been found in two human hereditary malformations. A mutation in the amino-terminal region of the HOXD13 protein causes the semi-dominant synpolydactyly type II (SPD II) syndrome (Muragaki et al., 1996), whose abnormalities are clearly more severe than those found in the limbs of Hoxd-13^−/− mice, thus suggesting that the HOXD13 protein of SPD II patients could have a dominant negative effect (Muragaki et al., 1996; Zàkàny and Duboule, 1996). However, no genital abnormalities have been described in SPD II affected individuals.

More recently, a nonsense mutation in HOXA13 was proved to cause the human hand-foot-genital (HFG) familial syndrome (Mortlock and Innis, 1997). This mutation is located in the third helix of the HOXA13 homeodomain. In one of our two Hoxa-13 mutant lines, the neomycin gene was inserted at nearly the same position (second helix) of the homeodomain, which should in principle result in a similarly truncated Hoxa-13 protein (Fromental-Ramain et al., 1996). The HFG syndrome, which is considered as genetically dominant (although no homozygous patients have yet been identified), includes subtle digit abnormalities, some of which are reminiscent of those found in Hoxa-13^+/− mouse mutants, as well as genital tract malformations of variable expressivity. The most severe defects are found in females and consist of ‘duplications’ of the uterus, resulting from insufficient mid-fusion of the terminal parts of the Müllerian ducts during development (Stern et al., 1970; Halal, 1988); in some cases, the female urinary tract malformations also include a displaced urethral opening and malpositioning of the ureteral orifices in the bladder wall. Male individuals exhibit various degrees of hypospadias (Giedion and Prader, 1975; Verp, 1989; Donnefeld et al., 1992; Fryns et al., 1993).

Hoxa-13^+/− heterozygous mutant mice are fertile and display no detectable urogenital tract anomalies, although subtle alterations may have been overlooked. The absence of major genital defects in Hoxa-13^+/− females may reflect the fact that humans and rodents have different uterus patterns. The human uterus has a single cavity (uterus simplex) generated by an extended mid-linear fusion between Müllerian ducts during development, up to quite rostral levels. In contrast, most rodent uteri are bipartite (uterus duplex) until the level of the cervix, and the mid-fusion between Müllerian ducts is restricted to more caudal levels. It is therefore likely that the lack of fusion leading to uterine duplication in HFG-affected women arises at levels where fusion naturally does not occur in mice. We note, however, that Hoxa-13^−/− fetuses show abnormal development of the distal part of the Müllerian ducts, which would result in uterine malformations if the homozygous mutation was not lethal. Furthermore, several defects exhibited by Hoxa-13^+/−/Hoxd-13^−/− compound mutant mice (misposition of the openings of the vagina and urethra, failure of distal fusion of the Müllerian ducts to form the upper vagina, hydronephrosis) presumably result from developmental abnormalities similar to those occuring in HFG patients.

Thus, nearly equivalent mutations of the Hoxa-13 gene have different outcomes in human and mouse urogenital tracts. It will be of interest to further investigate whether: (i) different expression boundaries of Hoxa-13/HOXA13 in the developing Müllerian duct system might correlate with the distinct human and rodent uterine morphologies; (ii) a differential pattern of HOXD13 and HOXA13 expression in the developing urogenital tract would make humans more susceptible than mice to HOXA13 gene dosage. Finally, we note that a number of familial or sporadic syndromes involve simultaneous malformations of the limbs and the urogenital tract (Pinsky, 1973; Halal, 1986). The Abdominal-B-related HOX genes clearly represent good candidates for some of these human syndromes.

We wish to thank Drs M. Mark, K. Niederreither and V. Sapin for discussions and critical reading of the manuscript, as well as C. Birling and B. Schuhbaur for excellent technical assistance. The help of M. Le Meur and her staff in animal maintenance is gratefully acknowledged. This work was supported by funds from the Institut National de la Santé et de la Recherche Médicale, the Centre National de la Recherche Scientifique, the Collège de France, the Centre Hospitalier Universitaire Régional, the Association pour la Recherche sur le Cancer, the Fondation pour la Recherche Médicale and the Ligue Nationale contre le Cancer. X. W. was supported by a fellowship from the Ministère de l’Éducation Supérieure et de la Recherche.