Molecular genetics of Müllerian duct formation, regression and differentiation

During male development mesenchyme-epithelia interactions mediate MD regression to prevent its development into a uterus and oviduct ( Fig. 1 ). MD regression requires binding and signal transduction from the transforming growth factor-β (TGF-β) family member anti-Müllerian hormone (AMH) secreted from the Sertoli cells of the fetal testis and its type 1 and 2 receptors expressed in MD mesenchyme (reviewed in ( Josso et al., 1993 ; Kobayashi and Behringer, 2003 ). The transcription of the Amh gene is directly regulated by multiple factors in the testis-determining pathway including SRY-box containing gene 9 ( Sox9 ), steroidogenic factor 1 ( Sf1 ), Wilms tumour homologue ( Wt1 ), and DSS-AHC critical region on the X-chromosome gene 1 ( Dax1 ) ( Arango et al., 1999 ; De Santa Barbara et al., 1998 ; Nachtigal et al., 1998 ; Shen et al., 1994 ). Females do not express AMH during fetal development thus allowing differentiation of the MD. The first observable histological change during regression in males is the appearance of the “sworl” pattern of the mesenchymal cells surrounding the MD in the most rostral region ( Dyche, 1979 ; Orvis and Behringer, 2007 ). At the onset of regression, differences in the MD also appear at the cellular level between the sexes. Initially the forming MD is mesoepithelial in nature with cell markers consistent with a mesenchymal cell tube however the morphology is consistent with a true epithelial cell tube ( Dyche, 1979 ; Orvis and Behringer, 2007 ). Beginning at E13.5, the female MD begins to express the epithelial cell marker E-cadherin (CDH1) apically and show evidence of apicobasal polarity while the MD in males remains unchanged ( Orvis and Behringer, 2007 ). This has also been observed in several other species including rat ( Dohr et al., 1987 ; Paranko and Virtanen, 1986 ), human ( Magro and Grasso, 1995 ), chick ( Jacob et al., 1999 ), but not in golden hamster ( Viebahn et al., 1987 ). AMH-induced MD regression occurs in a specific window in time during development and after this time the MD is no longer sensitive to AMH-induced regression ( Josso et al., 1976 ). This window corresponds with the time frame in which the MD is not yet expressing epithelial-specific markers ( Orvis and Behringer, 2007 ). One hypothesis is that the mesoepithial nature of the MD may facilitate regression in males. Genetic experiments in mouse and naturally occurring mutations in the human have demonstrated that AMH is necessary and sufficient for Müllerian duct regression. Amh null male mice have normal development of the testis and male reproductive tract, however Müllerian-derived tissues develop, causing infertility by physically blocking sperm release ( Behringer et al., 1994 ). Further, female transgenic mice ectopically expressing human AMH lack MD-derived tissues ( Behringer et al., 1990 ). Additionally, mutations in the human AMH gene are causative of ~45% of Persistent Müllerian Duct Syndrome (PMDS) cases, a rare autosomal recessive disorder. Like Amh null mice, male patients with PMDS are normally virilized, but have female reproductive organs including a uterus and fallopian tubes. PMDS is most often diagnosed because of cryptorchidism, a failure of the testis to descend, and/or inguinal hernia ( Belville et al., 1999 ; Belville et al., 2009 ; di Clemente and Belville, 2006 ; Salehi et al., 2012 ). Amhr2 positive cells in the MD mesenchyme transduce the AMH hormone signal secreted from the fetal testis ( Mishina et al., 1999 ). AMH signaling occurs in a paracrine manner and begins with AMH binding to its type 2 receptor (AMHR2). AMHR2 then forms a heteromeric complex with and then phosphorylates and activates a type 1 receptor. This activation results in the phosphorylation of an R-Smad and supposedly formation of an R-SMAD/SMAD-4 complex that translocates into the nucleus to transcriptionally activate AMH signaling pathway target genes. AMH type 1 receptors ALK2 (AVCR2) and ALK3 (BMPR1A), and AMH R-Smad effectors (SMAD1, SMAD5 and SMAD8) function redundantly in MD regression and are shared with the bone morphogenetic protein (BMP) pathway. ALK3 is considered the primary type 1 receptor required for regression however ALK2 is capable of transducing the AMH signal in the absence of ALK3. Conditional knockout of Alk2 in the MD mesenchyme does not block MD regression. However approximately half of all Alk3 conditional mutant males and 100% of mutants with conditional knockout of both Alk2 and Alk3 failed to regress the MD ( Jamin et al., 2002 ; Orvis et al., 2008 ). ALK2 and ALK3 are believed to function in a distinct temporal and spatial manner during regression. Initially, Amhr2 is expressed in the mesonephric epithelium. In the mesonephric epithelium, ALK2 appears to mediate AMH signaling which directs Amhr2 positive cells to undergo an epithelial to mesenchymal transition and migrate to surround the MD epithelium forming the distinct “sworl” pattern observed at E15.5 ( Zhan et al., 2006 ). This is followed by the breakdown of the basement membrane and subsequent loss of the epithelium ( Dyche, 1979 ; Orvis and Behringer, 2007 ; Trelstad et al., 1982 ). In females, Amhr2 expressing cells are found in the mesonephric epithelium on the antimesometrial side of the MD, but without AMH signaling these cells do not undergo migration to surround the MD epithelium or cellular changes up to at least E15.5 ( Fig. 3 ) ( Arango et al., 2008 ; Orvis et al., 2008 ; Zhan et al., 2006 ). Mechanisms known to be involved in regression include epithelial cell migration, epithelial to mesenchymal transformations and apoptosis ( Allard et al., 2000 ; Austin, 1995 ; Hutson et al., 1984 ). The pattern of regression of the MD is hypothesized to occur in a rostral to caudal wave corresponding to Amhr2 expression. A statically imaged time course of ex vivo cultured rat urogenital ridges (males and AMH-treated females) during regression showed rostrally a reduction in MD diameter ( Picon, 1969 ; Tsuji et al., 1992 ). Further Allard et al. showed a positive correlation between the rostral-caudal wave of Amhr2 expression and the pattern of increased apoptosis of the MD epithelium during regression in rat ( Allard et al., 2000 ). In contrast, studies from the Amhr2-lacZ knock-in mouse model indicate that although Amhr2 expression initiates in a rostral to caudal wave, it is expressed along the entire length of the MD at E13.5 well before any overt changes in the male MD ( Fig. 3 ) ( Arango et al., 2008 ). Further studies will be needed to clarify the spatial-temporal patterning and cell behaviors, including migration and apoptosis, during MD regression. Genetic studies in mice and human indicate that AMHR2 is the sole type 2 receptor required for AMH signaling and is likely dedicated to the AMH-signaling pathway. Male Amhr2 and Amh mutant mice have identical phenotypes; normally virilized with persistent MD-derived organs ( Behringer et al., 1994 ; Mishina et al., 1996 ; Mishina et al., 1999 ). Additionally, the phenotype of hAMH expressing transgenic female mice is rescued by Amhr2 mutation ( Behringer et al., 1990 ; Mishina et al., 1999 ). Further, mutation of the AMHR2 gene accounts for about half of the cases of PMDS with known molecular etiology ( Belville et al., 2009 ; di Clemente and Belville, 2006 ; Salehi et al., 2012 ). The Wilms’ tumor 1 ( Wt1 ) transcription factor gene is a direct activator of Amhr2 transcription. Amhr2 transcript levels are reduced in Wt1 null mice and both transcripts are co-expressed in the developing MD. Additionally, Wt1 expression mirrors the sexually dimorphic pattern observed for Amhr2 expression during MD regression in the urogenital ridge. Further, in vitro assays show WT1 activates transcription of Amhr2 and binds to elements in the Amhr2 proximal promoter ( Klattig et al., 2007 ). Two alternate splice variants of Amhr2, Amhr2 -Δ exon 2 and Amhr2 -Δ exon 9 and 10, are found in adult rat and mouse. The Amhr2 splice variants act in a dominant negative manner when co-transfected with full length AMHR2 and AMH in in vitro luciferase reporter assays. The dominant negative effect of the splice variants was reduced at higher levels of AMH concentration corresponding to local levels of AMH in the gonads. The authors hypothesize that splice variants may be expressed at high levels only in a particular subset of gonadal cells and regulate AMH signaling in these cells. Alternatively, splice variants may have site specific effects independent of the presence of AMH ligand or may have a role in the transport of ligand into the cell and might be important in trafficking across the blood brain barrier ( Imhoff et al., 2013 ). Expression of these splice variants during MD regression has not been determined and it is currently unknown what if any role they have during reproductive tract development. Wnt signaling plays multiple roles in MD development and is needed for formation, regression and differentiation. Prior to the onset of MD regression in males, WNT7A signaling from the epithelium to mesenchyme of the MD activates Amhr2 expression in both sexes and is also required for appropriate differentiation of the MD. Wnt7a mutant males retain MD-derived organs because Amhr2 expression is lost in the MD mesenchyme, thus blocking the AMH-signaling pathway. Consistent with the differentiation defects observed in the Wnt7a mutant female reproductive tract, in mutant male, the ectopic female reproductive tract shows no evidence of oviduct coiling and is a simple epithelial tube ( Parr and McMahon, 1998 ). This is in contrast to Amh and Amhr2 null mouse models where the differentiation of the mutant female and male MD-derived organs is relatively normal ( Behringer, 1994 ; Behringer et al., 1994 ; Mishina et al., 1996 ). The Frizzled ( Fzd ) genes encode the seven trans-membrane protein receptors for the WNT ligand which are required for both canonical and non-canonical WNT signaling pathways. A dedicated FZD receptor has not been identified for WNT7A in the MD. Previous studies showed interactions between WNT7A and FZD10 activated the WNT pathway ( Kawakami et al., 2000 ). Additionally, in mouse, Wnt7a and Fzd10 have overlapping expression patterns in the MD ( Nunnally and Parr, 2004 ). Although this identified FZD10 as a potential candidate receptor for WNT7A, Fzd10 knockout mice have no reproductive tract phenotype (Fzd 10: MGI Direct Data Submission MGI:3604450). Similar to Fzd10, Fzd1 expression is found in the MD mesenchyme and epithelium at E14.5, but Fzd1 knockout males have no MD regression defects ( Deutscher and Hung-Chang Yao, 2007 ; Lapointe et al., 2012 ). This suggests multiple FZD receptors are capable of interaction with WNT7A and function redundantly during MD regression. Several studies suggest WNT signaling is also important to the downstream molecular signaling cascade required for MD regression during male reproductive tract differentiation. However, the exact role of WNT signaling following activation of the AMH signaling pathway remains unclear. Either inactivation or constitutive activation of β-CATENIN (CTNNB1) in the MD mesenchyme causes retention of MD-derived tissues in mutant males independent of AMH expression suggesting tight control of CTNNB1 activation is necessary for MD regression ( Kobayashi et al., 2011 ; Tanwar et al., 2010 ). In the canonical WNT signaling pathway, nuclear localized CTNNB1 in a complex with T-cell factor/ lymphoid enhancer factor (TCF/LEF) transcription factors regulate expression of target genes. CTNNB1 is also known to have roles in cell tight junction formation and adhesion and may therefore be functioning independent of WNT signaling during regression ( Brembeck et al., 2006 ). CTNNB1 activates Lef1 transcription and upregulates Lef1 promoter activity in vitro ( Filali et al., 2002 ; Vadlamudi et al., 2005 ). Additionally, Ctnnb1 inactivation results in the loss of LEF1 up-regulation normally observed in the MD mesenchyme of males during regression ( Kobayashi et al., 2011 ). This suggested LEF1 may be required downstream of WNT/β-CATENIN signaling to induce MD regression during male reproductive tract differentiation. However, Lef1 null male mice have normal MD regression (Mullen and Behringer, unpublished observations) ( van Genderen et al., 1994 ). Several additional WNT pathway factors have also been identified that are expressed in a sex-specific pattern in the mesenchyme during AMH-induced MD regression including Wnt4, Wnt5a , and Frizzled-related Wnt pathway genes Sfrp1, Sfrp2, and Sfrp5 . Knockout of Wnt4 in the MD mesenchyme does not interfere with MD regression ( Kobayashi et al., 2011 ). No defects in MD regression have been reported in Wnt5a −/− mice ( Mericskay et al., 2004 ). Likewise, loss of function of Sfrp2 and Sfrp5 caused no defects in MD regression ( Cox et al., 2006 ). Double knockout Sfrp1 −/− /Sfrp2 −/− mice appear to have a slight delay, but MD regression is complete at later embryonic stages ( Warr et al., 2009 ). These results suggest that WNT pathway factors have redundant function during MD regression or alternatively are not required for regression. Further studies will be needed to clarify the roles of WNT signaling during later stages of MD regression. Sexually dimorphic expression patterns during regression have also been identified for Matrix metalloproteinase 2 ( Mmp2 ). Mmp2 is upregulated in the male MD mesenchyme during regression and this up regulation is lost in Amh null males. Morpholino knockdown of Mmp2 in organ culture blocks regression and decreases MD epithelium apoptosis ( Roberts et al., 2002 ). Null Mmp2 -mutant mice however have no defects in MD regression. This may suggest redundant function with other genes ( Itoh et al., 1997 ; Roberts et al., 2002 ). The PI3K/ AKT pathway may also have a role in AMH signal transduction during regression. Activated phospho-AKT (p-AKT) is present in equal amounts in the WD and MD in both sexes prior to regression. In rats, synchronous with the initiation of MD regression, p-AKT is decreased in males at E15.5 and is undetectable at E16.5. Females maintain p-AKT expression. This pattern is also observed in mice ( Fujino et al., 2009 ). Although PI3K signaling has been shown to prevent apoptosis and the epithelial to mesenchymal transitions that take place during MD regression, it is not clear if this reduction in p-AKT is a cause or effect of regression ( Allard et al., 2000 ; Dyche, 1979 ; Fujino et al., 2009 ; Trelstad et al., 1982 ; Zhan et al., 2006 ). AMH signaling has been shown to inactivate the PI3K pathway by blocking autophosphorylation of the EGF receptor in the MD epithelium by inhibiting tyrosine kinase ( Hurst et al., 2002 ; Hutson et al., 1984 ). Although multiple genes have been identified using candidate approaches, the role of many of these signaling pathways and molecules remains unclear due to the possibility of functional redundancy.

Aberrant development of the MD is a relatively frequent cause of human birth defects. Defects include MD aplasia, MD persistence, and MD fusion and patterning defects. Multiple medical syndromes are associated with female reproductive tract abnormalities. Molecular genetic studies of these patients have identified candidate factors involved in some cases. However the molecular genetic cause of these syndromes remains unknown in the majority of cases ( Kobayashi and Behringer, 2003 ). Several well-characterized syndromes are described; Mayer-Rokitansky-Küster-Hauser (MRKH) syndrome (OMIM 277000), maturity-onset diabetes of the young type 5 (MODY5) (OMIM 604284), persistent Müllerian duct (PMDS) syndrome (OMIM 261550), Urioste syndrome (OMIM 235255), Denys-Drash syndrome (DDS) (OMIM 194080) and Hand-foot-genital (HFG) syndrome (OMIM 140000). MRKH occurs in about 1 in 4500 female births and is characterized by the absence of the uterus and upper vagina in genetic females (46XX) ( Folch et al., 2000 ). In the majority of affected patients, ovarian development is normal and the lower third of the vagina is present. The molecular etiology of most cases of MRKH is unknown and occurs most often due to sporadic mutations but familial cases have been described with autosomal dominant inheritance with incomplete penetrance and variable expressivity. MRKH syndrome is further classified as type I (typical; isolated to the reproductive tract) or type II (atypical; associated with additional developmental defects) ( Ledig et al., 2011 ). Müllerian duct aplasia, unilateral renal agenesis, and cervicothoracic somite anomalies (MURCS) association is a severe form of MRKH type II. Patients with MURCS association have renal and skeletal defects in addition to uterine and vagina aplasia or agenesis ( Oppelt et al., 2006 ). Plausible causes of MRKH are mutations increasing either AMH or AMHR2 activity and/or expression that could result in MD regression in females. However, no defects in either the AMH or AMHR2 genes have been discovered to date ( Oppelt et al., 2005 ). Multiple genes have been associated with MRKH including WNT4, TCF2 (also known as HNF1β or v-HNF1 ), LHX1 and short stature homeobox ( SHOX ). Heterozygous loss of function mutations in the human WNT4 gene causes a complete absence of a uterus and upper vagina. In addition, female patients with WNT4 mutations also have excess androgens and symptoms of virilization ( Biason-Lauber et al., 2007 ; Biason-Lauber et al., 2004 ). This is consistent with the phenotype of Wnt4 null mice in which the MD does not development and testosterone biosynthesis is ectopically activated in the ovary ( Vainio et al., 1999 ). The most common mutations associated with MRKH are deletions of chromosomal region 17q12 which contains both TCF2 and LHX1 genes with approximately 6% of examined cases of MRKH carrying this deletion. TCF2 is a POU domain containing transcription factor widely expressed during development with function in epithelial differentiation ( Coffinier et al., 1999a ; Kolatsi-Joannou et al., 2001 ). Heterozygote mutations of TCF2 were first associated with MODY5 with additional malformations in renal development and function. A subset of female patients with heterozygote mutations in TCF2 have malformations of the reproductive tract including bicornuate uterus, uterus didelphys, and Müllerian aplasia and renal defects in the absence of diabetes ( Bingham et al., 2002 ; Lindner et al., 1999 ). This suggests an important role for TCF2 in urogenital tract formation and maintenance. In mouse, Tcf2 is expressed in the reproductive tract epithelium during development and persists in the adult ( Coffinier et al., 1999a ; Reber and Cereghini, 2001 ). However the function of Tcf2 in mouse urogenital development remains undetermined due to the early embryonic lethality of Tcf2 null mutant mice and the normal phenotype of the heterozygous mutants ( Barbacci et al., 1999 ; Coffinier et al., 1999b ). To date, MKRH syndrome has also been associated with five heterozygous mutations in the human LHX1 gene; four missense mutations and a frame shift mutation leading to a stop codon ( Ledig et al., 2012 ; Ledig et al., 2011 ; Sandbacka et al., 2013 ). Additionally, Lhx1 has been shown in mice to be essential for MD formation ( Kobayashi et al., 2004 ). Partial duplication of the SHOX gene was found in two daughters with MRKH type I and their unaffected father ( Gervasini et al., 2010 ). The development of a uterus and oviduct in human males has been noted in three syndromes; PMDS, Urioste syndrome and DDS. PMDS patients have normal testis development and the presence of MD-derived female reproductive organs. The syndrome is usually diagnosed while correcting undescended testes in pediatric patients. Reduced fertility is common in PMDS patients and potential causes include structural abnormalities caused by MD remnants, cryptorchidism past the age of 2 years, and damage to the vas deferens during orchidopexy. There is also an increased risk of malignancy in the ectopic MD-derived organs if not surgically removed and in the testes due to cryptorchidism. Eleven cases of malignancy in the retained MD organs have been reported and laparoscopic removal of the MD structures in PMDS patients is recommended ( Farikullah et al., 2012 ). PMDS is further classified with type I males having undetectable levels of AMH and type II males with normal AMH levels. The majority of PMDS cases are caused by mutations in the AMH (type I) or AMHR2 (type II) genes with each representing about half of the cases with known molecular etiology ( Belville et al., 2009 ; di Clemente and Belville, 2006 ; Salehi et al., 2012 ). Urioste syndrome is an autosomal recessive disorder associated with the retention of MD-derived tissues in males. In addition to a persistent MD phenotype patients also have lymphangiectasia and postaxial polydactyly. The molecular basis of this syndrome is currently unknown ( Urioste et al., 1993 ). Denys-Drash syndrome (DDS) is characterized by partial gonad dysgeneis, congenital or infantile nephropathy, and Wilms’ tumor. The molecular cause of DDS in almost all cases is dominant loss of function mutations in the zinc finger DNA binding domain of WT1. In multiple cases of DDS, patients have MD-derived uterus and vagina remnants in addition to developed vas deferens and epididymis ( Barakat et al., 1974 ; Denys et al., 1967 ; Manivel et al., 1987 ). Amh and Amhr2 are regulated by WT1 therefore reductions in AMH ligand and its receptor are postulated to cause the defects in MD regression seen in DDS patients ( Hossain and Saunders, 2003 ; Klattig et al., 2007 ; Nachtigal et al., 1998 ). The most common WT1 gene mutation in DDS is a missense mutation in exon 9, 1180C>T (R394W). The presence of retained MD structures is found in some but not all patients including those from the same family ( Coppes et al., 1992 ; Zhu et al., 2013 ). Mouse models heterozygous either for the Wt1 null allele or the R394W mutation have no evidence of MD regression defects ( Gao et al., 2004 ). Together this suggests that genetic background and/or environmental factors may play an important role in determining the penetrance of MD regression defects in DDS patients. HFG is an autosomal dominant syndrome that results in shortened thumbs and great toes and genital defects including hypospadias in males and a range of female reproductive tract defects from a longitudinal vagina or double vagina to double uterus and cervix. Incomplete MD fusion during embryogenesis gives rise to these defects in females with HFG ( Goodman and Scambler, 2001 ). The similarity of the Hypodactly ( Hd ) mutant mouse phenotype with a spontaneous dominant negative mutation in the first exon of the Hoxa13 gene to the limb and genital defects in humans first identified Hoxa13 as a potential candidate gene ( Post et al., 2000 ). Although, MD fusion defects are not present in Hd and Hoxa13 −/− mice, mild hypospadias of the vagina is observed in a portion of the mutant females ( Post et al., 2000 ; Warot et al., 1997 ). Additionally, one in six female compound mutants of Hoxa13 and its paralogue, Hoxd13 , had MD fusion defects ( Warot et al., 1997 ). Further, six heterozygous mutations in the human HOXA13 gene have been reported in families with HFG to date ( Goodman et al., 2000 ; Mortlock and Innis, 1997 ).

The Müllerian duct (MD) is the embryonic structure that develops into the female reproductive tract (FRT), including the oviduct, uterus, cervix and upper vagina. The FRT has essential functions in mammals, providing the site of fertilization, embryo implantation and fetal development. Defects in human FRT formation, thought to arise from abnormal embryonic development, are estimated to occur in up to 3% of births and often result in fertility problems. Diseases of the FRT are also prevalent in adult women and include uterine and cervical cancers, and endometriosis. Further, the reproductive tract of males and females initially contain identical pairs of fully formed Wolffian ducts (WDs) and MDs. During male sex differentiation, signaling between MD mesenchyme and epithelium mediate MD regression and prevent its development into a FRT. In males, defects in MD regression result in the retention of MD-derived organs and have been described in human persistent Müllerian duct, Urioste and Denys-Drash syndromes. Although to date not reported in humans, activation of the signaling pathways responsible for MD regression in females results in aplasia of the FRT ( Kobayashi and Behringer, 2003 ). While an understanding of MD development is clearly important to human health, the cellular and molecular mechanisms of these processes remain largely unknown. Recent molecular genetics studies of human disease and mouse models have identified multiple genes important for MD development ( Table 1 ). This review will provide an overview of MD formation, regression and differentiation and important genes and signaling mechanisms involved.

In vertebrates, the urogenital system originates from the intermediate mesoderm and consists of the kidneys, gonads, and urinary and reproductive tracts. Differentiation of the intermediate mesoderm into the urogenital tract begins shortly after gastrulation. First, signaling from the somite and surface ectoderm transduces mesenchymal-to-epithelial conversions in the intermediate mesoderm and the anterior to posterior formation of the nephric ducts, a pair of epithelial tubes joined at the cloaca ( Mauch et al., 2000 ; Obara-Ishihara et al., 1999 ). Next, the primary kidney or pronephros transiently forms in the posterior region of the nephric ducts and subsequently degenerates ( Bouchard et al., 2002 ; Saxen and Sariola, 1987 ). Then, posterior to the degenerating pronephros, the mesonephric duct (WD) develops and extends in an anteroposterior direction. The metanephros arises from inductive interactions between the ureteric bud that branches from the caudal WD and mesenchyme (reviewed in ( Little et al., 2010 )). Soon after formation of the WD, the paramesonephric duct (MD) appears and grows rostral to caudal adjacent to the WD until the duct joins at the urogenital sinus. Initially, the reproductive tracts of males and females are identical, containing two pairs of fully formed WDs and MDs. After sex determination, hormones produced in the fetal testis, anti-Müllerian hormone (AMH), testosterone, and insulin-like 3 (Insl3) trigger regression of the MD, differentiation of the WD into the male genital tract, consisting of the vas deferentia, epididymides, and seminal vesicles, and testicular descent, respectively. In females, lack of AMH, testosterone, and Insl3 in this developmental window permits differentiation of the MD into the female reproductive tract, consisting of the oviducts, uterus and upper vagina, passive degeneration of the WD, and maintenance of the ovaries in an abdominal position, respectively ( Fig. 1 ) (reviewed in ( Kobayashi and Behringer, 2003 ). The elongating MDs reach and fuse with the urogenital sinus, to form the utero-vaginal duct that will give rise to the caudal uterus, cervix and upper vagina ( Orvis and Behringer, 2007 ). The rostral region of the MD develops into the oviducts and rostral uterus. Uterine morphology between different mammalian species is highly diverse and varies in part because of differences in the extent of rostral MD fusion. For example, fusion in rodents is minimal, resulting in a duplex uterus (consisting of two individual uterine horns connected at the cervix) while in primates fusion extends more rostrally, resulting in a simplex uterus (consisting of a single uterine cavity) ( Kobayashi and Behringer, 2003 ).

Once the MD is formed it differentiates into a functional oviduct, uterus, cervix and upper vagina. Correct patterning and differentiation of the MD is dependent on a complex network of Hox and Wnt genes. Further it is known that steroid hormones also regulate many of the genes necessary for proper MD differentiation during organogenesis and adulthood ( Masse et al., 2009 ). Abdominal B ( AbdB ) homeobox genes ( Hoxa9, Hoxa10, Hoxa11 and Hoxa13 ) of the mammalian Hoxa cluster are required for differentiation and segmental patterning of the MD. AbdB genes are expressed along the anterior-posterior axis of the MD according to their 3’ to 5’ order in the Hoxa cluster. Hoxa9 is expressed in the oviduct. Hoxa10 is expressed in the mesenchyme of the uterus. Hoxa11 is expressed in the posterior uterus and cervix. Hoxa13 is expressed in the cervix and upper vagina ( Taylor et al., 1997 ; Warot et al., 1997 ). Hox10 expression is necessary for correct specification of tissue boundaries in the male and female reproductive tract. In Hoxa10 null mice, male and female reproductive tracts display posterior to anterior homeotic transformation. At E17.5, Hoxa10 is present only in the portion of the MD that will differentiate into the uterus. Mutations in Hoxa10 result in homeotic transformation of 25% of the proximal uterus into oviduct ( Benson et al., 1996 ). Three heterozygous mutations in the HOXA10 gene with predicted loss of function have been associated with uterine malformations. Patients with HOXA10 mutations had uterine defects ranging from septate uterus and vagina with a duplex cervix to a didelphic uterus indicative of MD fusion defects ( Cheng et al., 2011 ; Ekici et al., 2013 ). In adult uterus, Hoxa10 represses Emx2 and is found in an inverse expression pattern suggesting a further role for Emx2 in MD patterning and differentiation ( Troy et al., 2003 ). Overlapping expression of patterns of Hoxa10 and Hoxa11 suggests they have partially redundant function during MD differentiation. Further, exposure to the non-steroidal estrogen diethylstilbestrol (DES) in mice caused a posterior shift in Hoxa9 expression likely a result of down regulation of Hoxa10 and Hoxa11 ( Block et al., 2000 ). Sex steroids mediate Hoxa10 and Hoxa11 expression levels. Women exposed to DES during development have malformations of the reproductive tract consistent with anterior transformation ( Cermik et al., 2003 ; Taylor et al., 1999 ). Hoxa11 null mice have a thinner and shorter uterus lacking glands consistent with a partial homeotic transformation ( Gendron et al., 1997 ). Mutant mice in which the Hoxa11 homeodomain was replaced with the Hoxa13 homeodomain displayed posterior homeotic transformation of the female reproductive tract with the posterior uterus becoming cervix/vagina. This demonstrates that Hoxa11 and Hoxa13 have unique functions in MD differentiation and that Hoxa13 is upstream of factors required for differentiation of the MD into cervix and vagina ( Zhao and Potter, 2001 ). Although Hoxa13 null mutants die between E13.5 to E14.5, mutant female embryos are missing the caudal portion of the MD suggesting Hoxa13 has function during MD formation in addition to MD differentiation. Defects in caudal MD formation were also observed in a Hoxa13 paralogue, Hoxd13 , mutant females at birth ( Warot et al., 1997 ). The WNT pathway is required for MD patterning and differentiation. Wnt7a has important function in both MD regression in males and MD differentiation in females. Wnt7a is expressed throughout the MD epithelium prior to birth. After birth expression is maintained in the oviductal and uterine epithelium but is down-regulated in the vaginal epithelium ( Miller et al., 1998 ). In adult and neonate Wnt7a mutant females, the uterus is smaller in length and diameter and the uterine wall is thinner with less smooth muscle. Oviduct differentiation occurs but coiling and elongation are absent and uterine glands are not present. Posteriorly, a homeotic transformation occurs of oviduct to uterus and uterus to vagina in the Wnt7a null females ( Miller and Sassoon, 1998 ; Parr and McMahon, 1998 ). Wnt7a also appears to be necessary for the maintenance of Hoxa10 and Hoxa11 expression with the Wnt7a null females showing reduced expression of the two genes. Additionally, the similarity between the Hoxa11 null and Wnt7a null mutant phenotype (thin, small uterus lacking glands) is consistent with upstream regulation of Hoxa11 by Wnt7a ( Miller and Sassoon, 1998 ; Parr and McMahon, 1998 ). Recent studies using conditional knockout and tissues explanted under the kidney capsule have also clarified the role for WNT signaling at later stages of MD differentiation. These WNT signaling molecules include Wnt4, Wnt5a and CTNNB1 , which were not previously described because of early embryonic lethality ( Wnt5a and CTNNB1 ) or early roles in MD formation ( Wnt4 ). The role of Wnt4 in MD formation is well established. A recent conditional knockout study also demonstrates Wnt4 is important for MD differentiation. Conditional inactivation of Wnt4 in the uterine luminal and glandular epithelium, stroma and myometrium using Progesterone receptor ( PR )- Cre resulted in reduced uterine gland number and a stratified luminal epithelial layer instead of a simple columnar epithelial cell layer ( Franco et al., 2011 ). The receptor for Wnt4 during MD development has not been identified. However, Frizzled1 ( Fzd1 ) expression has been found in the developing mesenophros in both the MD mesenchyme and epithelium ( Deutscher and Hung-Chang Yao, 2007 ). Further, 3 of 17 Fzd1 −/− null females had a uterine phenotype similar to the Wnt4 conditional mutant females suggesting FZD1 may be acting as the WNT4 receptor ( Lapointe et al., 2012 ). However because of the low penetrance of uterine defects in the Fzd1 −/− mutant females it is likely other FZD receptors are also able to transduce the WNT4 signal. The Wnt5a gene is required for development of the posterior (caudal) region of the MD and glandular genesis. Wnt5a null mice have short, coiled uterine horns, but lack a cervix and vagina. In kidney capsule explant studies of mutant uterine horns, both Wnt7a in the luminal epithelium and Wnt5a in the uterine stroma were required for gland formation independent of canonical pathway member Lef1 ( Mericskay et al., 2004 ). Conditional deletion of Ctnnb1 in the MD mesenchyme using Amhr2-Cre causes a hypoplastic uterus with uterine hypotrophy, reduced uterine glands and uncoiled oviducts. In Ctnnb1 conditional mutants, reduced proliferation but not apoptosis contributes to the hypoplasia of the uterus. No differences were found in the expression patterns of Wnt4 and Wnt5a in the mesenchyme or Wnt7a in the epithelium ( Deutscher and Hung-Chang Yao, 2007 ). The mutant phenotype including lack of coiling mimics the Wnt7a null phenotype ( Parr and McMahon 1998 ). Distinct from the Wnt7a null mice, deletion of Ctnnb1 in the MD mesenchyme resulted in the differentiation of smooth muscle cells into adipose tissue postnatally causing the uterus to become fat filled. This suggests other WNTs may be needed for uterine differentiation or alternatively the role of Ctnnb1 in the mesenchyme is independent of WNTs and is functioning instead to control cell adhesion or the formation of cell tight junctions. These studies suggest that the initial differentiation of the myometrium does not require Ctnnb1 but in its absence there is a progressive shift from smooth muscle tissue to adipose tissue ( Arango et al., 2005 ; Deutscher and Hung-Chang Yao, 2007 ). Ctnnb1 conditional ablation using PR-Cre resulted in a thinner uterus of normal length with reduced gland numbers at sexual maturity. Constitutive activation of CTNNB1 using PR-Cre reduced uterus length and caused hyperplasia of uterine glands. Reductions in the size of the uterus as a result of ablation or activation of Ctnnb1 suggest tight control of WNT signaling is required for proper MD differentiation and development ( Jeong et al., 2009 ). The Wnt 4 and β- catenin ( Ctnnb1 ) conditional mutant females and Wnt7a −/− and Wnt5a −/− null females have similar defects in MD differentiation. These are also similar to the uterine phenotype caused by DES exposure suggesting a link between WNT signaling and estrogen signaling. WNT signaling ligands in the luminal epithelium may be required to prevent the formation of a stratified epithelial layer in the uterus in response to estrogen signaling ( Franco et al., 2011 ).

Much progress has been made in understanding the molecular genetics of MD development and several important signaling pathways have been identified ( Fig. 4 .). Study of knockout and conditional knockout mouse model phenotypes and molecular genetic studies of human diseases of FRT development have provided key insights into the complex signaling cascade involved. However potential functional redundancy of many of these factors including WNT signaling pathway members, MMPS and HOX genes has made it difficult to assess their in vivo function. For example, MD regression requires tight regulation of CTNNB1 activation but thus far a single WNT required for MD regression has not been identified. Further it is likely other factors yet to be identified are involved in MD development. Expression profiling using next generation sequencing technologies should identify genes that are differentially expressed during MD formation and differentiation. By using Cre recombinase lines expressing in the MD epithelium ( Wnt7a–Cre ) and mesenchyme ( Amhr2-Cre ) it will be possible to globally uncover the transcriptome of these juxtaposed tissue types. Mesenchyme-epithelia interactions are important regulators of development and many defects in FRT development are the result of aberrant cell-cell communication and signaling. Understanding the molecular and cellular mechanisms of MD formation and differentiation will give key insights into FRT development and disease.