Upon arriving in the undifferentiated genital ridge, by the end of the 5th week, germ cells continue to proliferate by mitosis and maintain bipotentiality for approximately one week. Then germ cells in the male gonad become enclosed in the seminiferous cords and differentiate into the spermatogonial lineage, which does not enter meiosis until the onset of puberty. It has recently been shown that gonocyte proliferation in the fetal testis is inhibited by androgens (58). In the female gonad, germ cells continue to proliferate by mitosis and enter meiosis in the 10th week (Table 1). Prevention of entry into meiosis was first thought to be a specific effect of male somatic cells since germ cells entering a prospective ovary or those which have failed to enter gonads of either sex enter meiosis at approximately the same time and develop into oocytes, irrespective of their chromosomal pattern (59). The existence of a testicular meiosis-preventing factor and an ovarian meiosis-inducing factor was initially postulated (60, 61), but these factors could not be isolated. Recent studies have shown that retinoic acid produced by the mesonephros, as well as by the fetal lung and adrenal, may act as a meiosis inducer (62, 63). Germ cells embedded in the seminiferous cords do not enter meiosis because they are protected from retinoic acid action: Sertoli cells express CYP26B1, an enzyme that catabolizes retinoic acid, and could then be the postulated meiosis-preventing factor (62, 63). As previously stated, chromosomal constitution does not influence their sex differentiation of germ cells: XX germ cells surrounded by Sertoli cells differentiate into spermatogonia while XY germ cells in an ovarian context differentiate into oogonia and then enter meiosis (64). However, germ cells whose karyotype is discordant with the somatic lineages fail to progress through gametogenesis and enter apoptosis later in life. This is also true for germ cells with an abnormal sex chromosome complement, e.g. 45,X (Turner syndrome), 47,XXY (Klinefelter syndrome).

The influence of germ cells on the developing gonad is sexually dimorphic: Germ cells are essential for the maintenance and differentiation of the fetal ovary, otherwise prospective follicular cells degenerate and streak gonads result. In contrast, the development of the testes is not hindered by the lack of germ cells (64, 65).

No sexual difference can be observed in the gonads until the 6th week of embryonic life in humans and 11.5 days post-coitum (dpc) in mice. Undifferentiated gonads of XX or XY individuals are apparently identical and can form either ovaries or testes. This period is therefore called indifferent stage of gonadal development.

Compelling evidence for the importance of the Y chromosome for the development of the testes, irrespective of the number of X chromosomes present, has existed for over 50 years (for review see refs. 12, 66). However, the identification of the testis-determining factor(TDF) on the Y chromosome did not prove easy and several candidates (e.g. HY antigen, ZFY) were successively proposed and rejected until the SRY gene was cloned in 1990 in man (67) and mouse (68). Experimental (69, 70) and (67, 71) evidence has clearly established that SRY is a testis determining factor.

SRY is a member of a family of DNA-binding proteins bearing a high mobility group (HMG) box and maps to the short arm of the Y chromosome Yp11.3, very close to the pseudoautosomal region 1 (PAR1). PAR1 on Yp and PAR2 on Yq are the only regions of the Y chromosome that undergo meiotic recombination with homologous sequences of the X chromosome during male spermatogenesis (Fig. 2). The proximity of SRY to PAR1 makes it susceptible to translocation to the X chromosome following aberrant recombination and provides an explanation for 80% of XX males (72) and for a low proportion of XY females. Surprisingly, mutations and deletions of the SRY locus only account for 15% of these patients (73, 73). The physiologically relevant mechanism of action of SRY in gonadal differentiation still remains to be completely elucidated. There are reports suggesting that it may act as a transcriptional activator, but also as a repressor, or as an architectural factor allowing DNA bending for other factors to regulate tagert gene transcription (for review, see ref. 12). Proteins that interact with Sry and could have a relevant function in gonadal differentiation include SIP-1/NHERF2 (74, 75) and KRAB-O (76).

Figure 2. Mapping of genes involved in sex determination and differentiation (in blue) residing on sex chromosomes. AR: Androgen receptor; ATRX: Alpha-thalassemia/mental retardation syndrome X-linked; AZF: azoospermia factor; CSF2RA: Colony-stimulating factor 2 receptor alpha; DAX1: DSS-AHC critical region X chromosome gene 1; DAZ: Deleted in azoospermia; FRA-X: Fragile X syndrome; DMD: Duchenne muscular dystrophy; GK: Glycerol kinase; HY: Histocompatibility antigen Y; IL3RA: Interleukin 3 receptor alpha; IL9R: Interleukin 9 receptor; Kal1: Kallmann syndrome 1; PAR: Pseudo-autosomal regions; POLA: DNA polymerase alpha; RBMY: RNA-binding motif protein Y chromosome; SHOX: Short stature homeo box; SRY: Sex-determining region Y chromosome; USP9Y: Ubiquitin-specific protease 9 Y chromosome; XIST: X inactivation-specific transcript; ZFX: Zinc finger protein X-linked; ZFY: Zinc finger protein Y-linked.

A tight regulation of SRY expression is essential for fetal gonadogenesis: both timing and level of expression are determinant, as revealed by experiments in mouse showing that Sry levels has to reach a certain threshold at a certain stage of fetal development to induce testis differentiation (77).

The mechanisms underlying the initiation of SRY expression begin to be unraveled (for review, see refs. 78, 79). The +KTS splice variant of WT1 (27, 80, 81), SF1 (82), GATA4/FOG2 (83) and Sp1 (84, 85) are able to activate SRY transcription. Others postulate that the +KTS isoform of WT1 acts as a posttranscriptional stabilizer of SRY mRNA (79). Whether members of the insulin receptor family have a direct implication on SRY expression still needs further investigation (6). Sox9 seems to downregulate Sry expression in the mouse (86-88); in man, however, SRY expression persists after testis differentiation (89).

Once the importance of SRY as a mammalian testis-determining gene was established, scientists have been prompted to search for a testis-specific target gene. SOX9, an autosomal member of the HMG-box protein superfamily (90), has become a firm candidate. In the mouse, Sox9 is expressed in indifferent gonads of both sexes, but persists only in Sertoli cells after Sry expression has peaked (91, 92) (Fig. 3). Available evidence indicates that Sox9 is adirect target of Sry in mice (88). However, it is also possible that indirect mechanisms mediate Sox9 activation, in line with the hypothesis indicating that SRY might act as a repressor of a negative regulator of the male cascade (93). For instance, targeted disruption of Foxl2 leads to Sox9 upregulation in the XX gonad (94), and prostaglandin D2 has been shown to uregulate Sox9 in the absence of Sry (95, 96).

Overexpression of Sox9 during early embryogenesis induces testicular differentiation in two different models of transgenic XX mice (13, 97). Functional analysis of Sox9 during sex determination, by conditional gene targeting in mice, has shown that homozygous deletion of Sox9 in XY gonad interferes with sex cord development and with activation of testis specific markers (87). Further evidence for the role of SOX9 in testicular development comes from observations in humans, where a double dose of SOX9 expression is required. Heterozygous mutations result in haploinsufficiency resulting in campomelic dysplasia, a polymalformative syndrome that includes sex-reversal due to gonadal dysgenesis in XY individuals (90). Sox9 also affects differentiation of the reproductive tract by upregulating the expression of anti-Müllerian hormone (AMH) (98), a Sertoli cell-specific factor involved in male differentiation of the internal genitalia (see below).

Figure 3. Sry, Sox9 and Dax1 expression in fetal mouse gonads. Note the transient expression of Sry, while Sox9 continues to be expressed in the male. Data from refs. 92, 99.

Sox8 is another member of the Sox family. Like Sox9, Sox8 is expressed in the central nervous system, limbs, kidneys, gonads and craniofacial structures. During mouse embryo development, Sox8 expression can be observed in the developing testis around the time of sex determination (100). Sox8 can bind the canonical target DNA sequences and activate AMH transcription acting synergistically with SF1, but with less efficiency than Sox9 (101).

Observations made in XY intersex patients with normal SRY together with the discovery of proteins showing a sexually dimorphic pattern of expression in the gonads following SRY peak have helped to identify other loci, likely to be involved in testicular differentiation, which are discussed hereafter.

ATRX, also known as XH2, is an X-encoded DNA-helicase whose mutation results in mental retardation, a-thalassemia and gonadal dysgenesis in XY individuals (18, 19). M33, whose homolog in human is CBX2 (16) is a transcriptional regulator whose disruption in mice causes several defects including retardation of embryonic gonadal formation and gonadal sex reversal in XY fetuses (17). Like SRY, ATRX and CBX2 might be involved in chromatin remodeling, which seems to play an important role in sex determination (102).

Although SRY was identified more than 10 years ago, the molecular mechanisms by which it triggers testicular differentiation from the gonadal ridge are as yet poorly understood. As already mentioned, two hypotheses have been proposed: one holds that SRY upregulates testis-specific genes, another that SRY antagonizes a gene that inhibits testicular development (93). SRY-negative XX maleness could then be explained by a mutation in the putative anti-testis gene. Several candidates have been proposed: DAX1, encoding for an orphan nuclear receptor and mapping to the DSS (Dosage Sensitive Sex-reversal) region on Xp21 was the first putative testis repressor and/or ovarian determining gene. A duplication of DSS results in sex-reversal in 46,XY patients with a normal SRY gene (103); XY transgenic mice that overexpress Dax1 may also be sex-reversed (104). However, the disruption of Dax1 in XX mice does not prevent ovarian differentiation (105) and no DAX1 mutation has yet been described in SRY-negative XX males. Furthermore, DAX1 seems to be essential for normal testicular cord formation in rodents (106). Interestingly, expression of both SRY and DAX1 is upregulated by WT1 in the indifferent gonad (27, 81, 107). These controversial findings may be explained by the hypothesis that a precise dosage of DAX1 is required for testis determination (108).

SOX3 is another candidate "anti-testis" gene (109), mapping to Xq27 (110). Yet, compelling evidence, e.g. a mutation of SOX3 in a SRY-negative XX male, is still lacking.

In goats, XX males develop in the event of a deletion in the autosomal PIS locus (111, 112). A winged helix/forkhead transcription factor gene, FOXL2, has been identified at this locus. FOXL2 is expressed in the mouse and human ovary (113), where it is involved in granulosa cell differentiation (114). While FOXL2 mutations result in a variety of phenotypes, from adult ovarian failure to development of streak gonads (113, 115-117), no XX males have been reported.

Finally, a member of the Wnt family of intracellular signaling molecules, Wnt4, has been shown to inhibit the development of androgen-producing Leydig cells in early mouse embryos. Wnt4 null XX mice show partial sex-reversal of the gonadal phenotype (118) and a duplication of 1p31-35, where human WNT4 maps, causes ambiguous genitalia of XY patients (119). Wnt4 is also involved in sex differentiation of the internal genital tract (see below).

SRY is expressed in pre-Sertoli cells having delaminated from the coelomic epithelium in the central part of the indifferent gonad and induces their differentiation (125, 126). SRY-expressing pre-Sertoli cells lying beneath the coelomic epithelium play a central role in the migration of cells from the mesonephric mesenchyme into the differentiating gonad to originate peritubular and endothelial cells (120). Experimental work using XX-XY chimeras has shown that not 100% of Sertoli cell precursors need to express Sry to differentiate along the male pathway: in fact up to 10% of Sertoli cells were XX. However, a threshold number of SRY expressing –i.e. XY– cells seems to be essential in order for Sertoli cell differentiation, and thus testicular development, to be guaranteed (127).

Along with SRY, FGF9 as well might have a role in inducing mesonephric cell migration into the developing fetal testis and Sertoli cell differentiation. FGF9 is expressed in Sertoli cells of the fetal testis and Fgf9 null mice have dysgenetic gonads (128) (see below).

Testicular cord formation, the first sign of testicular differentiation, can be detected in human fetuses 13-20 mm crown-rump length (43-50 days) beginning in the central part of the gonad (129). Cord formation is heralded by the development of a new type of cell, the primitive Sertoli cell, characterized by a polarized, large and clear cytoplasm with abundant rough endoplasmic reticulum and complex membrane interdigitations (130), a downregulation of desmin and an upregulation of cytokeratins (131), and the expression of AMH and DHH (132). Sertoli cells aggregate around large, spherical germ cells, with a large nucleus and pale cytoplasm, called gonocytes at this stage, which can be observed in the center of testicular cord cross-sections. The structural basis of cord formation seems to be dependent on basal lamina deposition between Sertoli and peritubular cells with myofibroblastic characteristics (122). In the interstitial compartment, connective tissue, blood vessels and Leydig cells can be observed. One particular feature of testicular vasculature is the formation of the coelomic vessel, a large vessel that appears below the coelomic epithelium very early in testicular differentiation (133). Surrounding the gonad, the basement membrane layer underlying the coelomic epithelium thickens to form the tunica albuginea.

Morphologically and functionally distinct from testicular cords, the interstitial compartment contains developing Leydig cells, the most important androgen producing cells in the male. Leydig cells derive from the coelomic epithelium and also from mesenchymal cells of mesonephric origin (8, 134, 135). They differentiate in the interstitial tissue by the beginning of the 8th week in the human embryo, after testicular cords have completely formed, and soon begin to produce testosterone, which plays an essential role in the stabilization of Wolffian ducts and the masculinization of external genitalia. Leydig cells also produce insulin-like growth factor 3 (INSL3), a growth factor responsible for the transabdominal phase of testicular descent (136). Fetal Leydig cells are large, eosinophilic cells with an abundant smooth endoplasmic reticulum and numerous mitochondria, but no Reinke's crystalloids, which are restricted to adult Leydig cells. Although the initial differentiation of fetal Leydig cells may occur independently of gonadotropin action in rodents (137), further Leydig cell differentiation and proliferation depends on placental hCG in the first and second trimesters of fetal life and on fetal pituitary LH thereafter (138). Leydig cell number peaks at mid-gestation (Fig. 4) and then slightly decreases. After birth, Leydig cells disappear from the interstitial tissue of the testis until puberty (for review, see ref. 139).

Figure 4. Leydig cells accumulate in the testicular interstitium of a 90 mm male human fetus. Large eosinophilic Leydig cells with a prominent nucleus are interspersed with mesenchymal cells.

The morphology of the testes changes very little during fetal life and after birth, until pubertal onset. Testicular cords maintain the same appearance, with only slight changes. In the perinatal period, gonocytes are progressively replaced by spermatogonia, which are smaller and located in contact with the basement membrane. It is not until puberty, when germ cells enter meiosis and Sertoli cells acquire adult characteristics, that a lumen develops and testicular cords become seminiferous tubules.

In order for the fetal testis to adequately differentiate and secrete masculinizing hormones, not only do all these factors need to be present at sufficient levels in the right cell lineage, but their expression must also be initiated within a narrow time window. In XY mice, the delay of Sry expression has been postulated to result in higher Dax1 levels in the genital ridge at the time of gonadal differentiation. Consequently, ovarian development proceeds and primordial follicles develop in the center of the gonad, resulting in an ovotestis (77, 104), or in the whole gonad, with the formation of an ovary (165). These ovarian structures subsequently degenerate since germ cells, which are essential for follicular development, fail to enter meiosis and undergo apoptosis owing to the lack of two X chromosomes.

The mechanisms involved in ovarian differentiation are still poorly understood, and no ovary-inducing gene has been characterized. DAX1, an initial candidate, does not seem to be involved, since ovarian development occurs in knockout mice (105). WNT4 represses mesonephric endothelial and steroidogenic cell migration in the XX gonad, preventing the formation of a male-specific coelomic blood vessel and the production of steroids (118, 119, 140, 173, 174). Follistatin seems to be a gene downstream of WNT4 (175). Wnt4- and follistatin-null ovaries develop the male-specific coelomic vessel and Leydig cell-like androgen producing cells. There is also a massive loss of germ cells by apoptosis which depletes the oocyte pool during fetal life (173, 175). However, some of the abnormalities in gonadal development might not be sex-specific, but rather derive from the lack of expression of Wnt4 in the undifferentiated gonadal ridge, since testes from Wnt4 knockout XY mice also show disorganized testicular cords (173). FOXL2 is a transcription factor expressed in germ and somatic cells more strongly in the female than the male fetal gonad (114, 117). Mutations in the human FOXL2 gene have been found in females with premature ovarian failure associated with eyelid abnormalities characterized by blepharophimosis, ptosis and epicantus inversus (BPES) (113, 115). Although important to granulosa cell differentiation, FOXL2 does not seem essential to early ovarian development (94, 117).

Two major events can be distinguished in ovarian development: germ cell migration, proliferation and meiosis, and folliculogenesis. For a long time, it has been known that two intact X chromosomes are required in the human -in contrast to the mouse- for ovarian differentiation and development (176). The lack of two X chromosomes results in germ cell loss and, subsequently, absent folliculogenesis. Therefore, all the factors involved in the proliferation and migration of PGCs in early embryogenesis (see The germ cells section) are essential for ovarian formation. Factors involved in germ cell meiosis are also important. Stabilization of oocytes requires the expression of DAZLA, a cytoplasmic protein present in female and male germ cells (177), and Msh5, a protein involved in DNA mismatch repair (178). In Msh5 null mice, oocytes are lost before the diplotene stage resulting in ovarian dysgenesis. Stra8 gene, an early marker of female germ cells, is upregulated in embryonic germ cells of XX gonads prior to meiotic entry and is not expressed in male embryonic germ cells. Stra8 expression in germ cells of XX gonads takes place in an anterior-to-posterior wave and is followed by the upregulation of the meiotic gene Dmc1 (179). A number of other factors involved in oocyte development has been described (for review, see ref. 180).

Although not essential to ovarian differentiation, several factors are involved in the development of ovarian follicles. FIGα is crucial for the formation of primordial follicles (181). AMH regulates the recruitment of primordial follicles into subsequent steps of folliculogenesis (182), and GDF9 is important for follicle growth beyond the primary stage (183, 184).

The Wolffian duct is the excretory canal of the primitive kidney, the mesonephros. In the human embryo, Wolffian ducts originate dorsolaterally to the segmented pronephric blastema and laterally to somites 8-13 in embryos 25 to 32 days old (Table 1). In normal embryogenesis, a single ureteric bud evaginates from the Wolffian duct and grows dorsally, in response to inductive signals from the metanephric mesenchyme (185). Growing caudally, Wolffian ducts progressively acquire a lumen and reach the hindgut, which caudally becomes the cloaca. The Wolffian ducts become incorporated into the male genital system when renal function is taken over by the definitive kidney, the metanephros. As pointed out by Dorothy Price (186), the duration of the ambisexual stage of Wolffian development is determined by the duration of mesonephric activity.

Müllerian ducts arise in 10-mm human embryos as a cleft lined by the coelomic epithelium, between the gonadal and mesonephric parts of the urogenital ridge. This coelomic opening will later constitute the abdominal ostium of the Fallopian tube. The cleft is closed caudally by a solid bud of epithelial cells, which burrows in the mesenchyme lateral to the Wolffian ducts and then travels caudally inside their basal lamina. In birds, integrity of the Wolffian duct is required for Müllerian growth : experimental interruption of the Wolffian duct arrests the growth of the Müllerian duct at this point (187). At 8 weeks of development, the growing solid tip of the Müllerian duct, now in the pelvis, lies medial to the Wolffian duct, having crossed it ventrally in its downward course and entered its basal membrane. For a while, the two Müllerian ducts are in intimate contact, then they fuse, giving rise to the uterovaginal canal (Fig. 8), which makes contact with the posterior wall of the urogenital sinus, causing an elevation, the Müllerian tubercle, flanked on both sides by the opening of the Wolffian ducts (Fig. 7). Lineage-tracing experiments in chicks and mice have shown that coelomic epithelial cells contribute to all components of the Müllerian duct at all times and suggest that Wolffian ducts do not provide cellular building material (188).

Figure 8. Fused Müllerian ducts flanked by Wolffian ducts in the lower reproductive tract of a 50 mm female human fetus

Müllerian regression, the first sign of male differentiation of the genital tract, occurs in 55 to 60 day-old human embryos in the area where the ducts are closest to the caudal pole of the testis (189) (Fig. 9). Once initiated, the regression of the Müllerian duct extends caudally as well as cranially, sparing the cranial tip which becomes the Morgagni hydatid, and the caudal end, which participates in the organogenesis of the prostatic utricle. Müllerian regression of the cranial part of the Müllerian duct begins while the duct is still growing caudally towards the urogenital sinus (190) and is characterized by a wave of apoptosis spreading along the Müllerian duct (191, 192). Shortly afterwards, the peri-Müllerian mesenchyme condenses to form a fibrous whorl, which progressively strangles the Müllerian duct and finally remains the only witness of its former existence. Mesenchymal changes are preceded by the dissolution of the basement membrane, which precipitates apoptosis and allows extrusion of epithelial cells and their transformation into mesenchymal cells (192, 193). Epithelial-mesenchymal transformation is an important factor of epithelial cell loss during Müllerian regression. From a molecular standpoint, Müllerian regression is marked by the deposition of peri-epithelial extracellular matrix (194) by an increased expression of Mmp2, a member of the matrix metalloproteinase family, involved in apoptosis (195) and by the migration of anti-Müllerian hormone type II receptors from the coelomic epithelium to the peri-Müllerian mesenchyme (196).

Figure 9. Regressing Müllerian duct in a 35 mm male human fetus. Note the fibroblastic ring surrounding the epithelium.

The second aspect of male differentiation of the internal genital tract is the stabilization and differentiation of the Wolffian ducts (reviewed in ref. 197). After the loss of mesonephric functional activity, the mesonephric nephrons and caudal tubules degenerate but the cranial tubules persist to form the male efferent ducts. The connections between the mesonephric tubules and the gonadal primordium are permanently established in the sixth week; in the male, they give rise to the rete testis, while in the female, they form the rete ovarii. Between weeks 9 and 13 in the human embryo the upper part of the Wolffian duct differentiates into the epididymis. Below, it is surrounded by a layer of smooth muscle and becomes the vas deferens, which opens into the urogenital sinus at the level of Müllerian tubercle. In sexually ambiguous individuals, in whom Wolffian and Müllerian ducts coexist, the vas deferens is embedded in the uterine and vaginal walls (198, 199). The seminal vesicle originates from a dilatation of the terminal portion of the vas deferens in 12 wk-old fetuses.

During human fetal development, the testis migrates from its initial pararenal position to its terminal location in the scrotum. Testicular descent is a complex process that has been subdivided into several phases. Transbdominal movement brings the testis down to the internal inguinal ring at 12 weeks (Fig. 10). As the testis approaches the inguinal ring, the epididymis is pulled into the canal and serves as a wedge for opening it (200). The gubernaculum testis, a jelly-like structure which connects the lower pole of the testis and the tail of the epididymis to the scrotum, swells and pulls the testis down to the inguinal ring. while the cranial suspensory ligament dissolves. In the female, the cranial ligament holds the ovary in a high position and the gubernaculum remains long and thin as the round ligament (201). The second –inguinoscrotal- phase of testicular descent occurs between 27 and 30 weeks after conception (202). « Physiological » cryptorchidism is frequent in premature infants. After 25 weeks, the gubernaculum bulges beyond the external inguinal ring and is hollowed out by a peritoneal diverticulum called the processus vaginalis.

Figure 10. Testicular descent. Initially (left) the primitive gonad is located near the kidney, held by the cranial suspensory ligament (CSL) and the gubernaculum testis. Androgen-mediated dissolution of the CSL and insulin like factor 3 (INSL3) mediated swelling of the gubernaculum brings the testis to the internal orifice of the inguinal canal (center). Inguino-scrotal migration consists of the passage of the testis through the inguinal canal into the scrotum and is thought to be androgen-dependent (right).

Male orientation of the urogenital sinus is characterized by prostatic development and the repression of vaginal development. Prostatic buds appear at approximately 10 weeks at the site of the Müllerian tubercle and grow into solid branching cords. Maturation of the prostatic gland is accompanied by development of the prostatic utricle. Two buds of epithelial cells, called the sino-utricular bulbs in the male, develop from the urogenital sinus close to the opening of the Wolffian ducts and grow inwards, fusing with the medial Müllerian tubercle, to form the sino-utricular cord, enclosed within the prostate gland, which canalizes at 18 weeks to form the prostatic utricle, the male equivalent of the vagina (203).

Masculinization of the external genitalia begins in human male fetuses 35-40 mm long (~9 weeks) in human male fetuses by lengthening of the ano-genital distance (189) (Fig. 12). Fusion of the labioscrotal folds, in a proximal to distal fashion, forms the epithelial seam (204), which closes the primary urethral groove. The urethral plate, an extension of the urogenital sinus present within the genital tubercle from the earliest stages of development (205), lies in the roof of the primary urethral groove and extends to the tip of the phallus. The literature concerning penile development is controversial. Most textbooks describe it as a two-step process, with the proximal urethra forming by fusion of the urethral folds around the urethral plate and the distal urethra arising from a invagination of the apical ectoderm. However, some authors (204, 206) believe that the entire human male urethra is of endodermal origin, formed by the urethral plate dorsally and the fused urethral folds ventrally. The seam is remodeled into the tubularized urethra without connection to the epidermis. The ventrally discarded excess epithelial cells migrate into the ventral skin of the penis. Abnormalities of seam formation or remodeling could explain the vast majority of cases of hypospadias in which defects of androgen synthesis or metabolism cannot be demonstrated (207).

Urethral organogenesis is complete at 14 weeks, apart from a physiological ventral curvature, which can persist up to 6 months of gestation (208). However, surprisingly, no size difference exists between penile or clitoral size before 14 weeks (209, 210), in spite of the fact that serum testosterone levels peak at that time (211). Maximal phallic growth occurs during the third trimester of fetal life, at a time when male testosterone levels are declining. At that time, the fetal pituitary has taken over control of Leydig cell production of testosterone in the fetus; indeed, microphallus is a classical sign of congenital gonadotropin deficiency. The insensitivity of the male genital tubercle to high levels of androgens during the second trimester is not due to a low expression of the androgen receptor or of 5α-reductase type 2 in the corpora cavernosa (212).

Molecular genetic studies in mouse have contributed to the identification of growth factors essential for the formation of the sexual ducts (Table 3).Homeotic genes Lim1, Emx2 and Hoxa-13 are also required for the normal development of a wide array of structures and organs. The function of the transcriptional regulators Pax2 is essential for the developmentof epithelial components derived from the intermediate mesoderm, including kidneys and ureters; in contrast the role of Wnt4a and Wnt7a, a subset of the Wnt family homologous to the Drosophila wingless gene, is restricted to reproductive organs.

Congenital bilateral absence of the vas deferens is responsible for 1-2 % of cases of male infertility and is present in 95% of patients affected with cystic fibrosis (217), a bronchial and pancreatic disease due to mutations in the cystic fibrosis transmembrane conductance regulator (CFTR). Whether efferent duct maldevelopment is a primary defect of cystic fibrosis or a secondary degenerative change resulting from obstruction by mucus is not known at the present time. Isolated absence of vas deferens in otherwise healthy men is often associated with the presence of a single CFTR allele mutation (218).

Development of the genital tubercle and limb bud has many similarities, including at the molecular level (Table 3). External genitalia are appendages emerging from the caudal body trunk, hence many genes which pattern distal limb development also play a predominant role during genital tubercle formation, for example BMPs (226) and members of the fibroblast growth factor, hedgehog and Hox gene families (for reviews, see refs. 227-229). Many of these genes interact to form a signaling cascade (227). Sonic hedgehog (Shh) signaling from the urethral epithelium plays a crucial role per se and by regulating the transcription of many of the other factors involved for outgrowth and patterning of external genitalia (229). Shh does not control the earliest initiation of genital swellings, a step regulated by Fgf8. The homeotic genes Hoxa13 and Hoxd13 act in a partially redundant manner since double null mutants, though more likely to survive than Hoxa13 null ones, show more severe urogenital abnormalities than those with at least one functional allele (222).

Correct vaginal development requires Wnt, Pax and Ltap genes (Table 4). Vaginal abnormalities similar to those elicited by diethylstilbestrol (DES) administration, i.e. vaginal clear-cell adenocarcinoma, vaginal adenosis, transverse vaginal ridges and structural malformations of the cervix and uterus, occur in transgenic mice deficient in Wnt7a, a signaling molecule expressed by the Müllerian epithelium (230), suggesting that DES exposure acts by deregulating Wnt7a during uterine morphogenesis (231). Wnt7a deficiency could act by interfering with normal mesenchymal-epithelial signaling, which is required for correct morphogenesis of the reproductive tract. Vaginal opening is regulated by Pax8 (221) and Ltap (loop-tail), a component of the frizzled pathway (232). Females heterozygous for the semi-dominant mutation Lp have tail loops and an imperforate vagina (233).

AMH is secreted in high levels by Sertoli cells from the time of testicular differentiation until puberty (Figs. 15 and 16) and at lower levels thereafter (for reviews, see (243-245). In the female, AMH is produced by granulosa cells of growing follicles from birth to menopause (170, 246-248).

Figure 15. Expression of the AMH protein in 11-week-old male human fetus, using an AMH-specific polyclonal antibody and a peroxidase reaction. Note the strong staining of seminiferous tubules.

Figure 16. Ontogeny of testicular AMH production. AMH expression is switched on following SRY and SOX9 peaks. Expression levels are high during fetal life, decline transiently in the perinatal period but remain high in childhood, and are finally downregulated by androgens after pubertal onset.

The AMH protein is a 140-kD glycoprotein homodimer belonging to the TGF-β family (249) and as such is cleaved to generate N-terminal and C-terminal homodimers (250). The homology of AMH to other members of the transforming growth factor-β (TGF-β) family is restricted to the bioactive C-terminus (251, 252), for which a molecular model has been built, through analogy to crystallized members of the family (253) (Figure 17). The human 2.8-kb gene has been cloned (249) and mapped to chromosome 19p13.2-p13.3 (254). It consists of five exons, the last one coding for the C-terminal fragment.

Figure 17. A three-dimensional model of AMH C-terminal dimer, constructed by analogy with BMP2 and BMP7. Adapted from ref. 253.

The AMH gene has been cloned in other eutherian mammals such as the mouse (255) and rat (256) and in the marsupial tammar wallaby (257). AMH is also present in the chick (258, 259) and American alligator (260), both of which carry Müllerian ducts which regress in the male. More surprisingly, AMH orthologs (reviewed in ref. 261) and the AMH type II receptor (262) have been cloned from the gonads of modern teleost fish, which do not possess Müllerian ducts. In this case, AMH appears to be involved essentially in spermatogenesis, which suggests that AMH acquired its anti-Müllerian activity during the course of evolution. Indeed, in higher vertebrates, AMH continues to play a role upon gonadal differentiation and steroidogenesis both in males (263) and in females (248).

In the human fetal male gonad, AMH mRNA and protein can be detected from the 8th week, when Sertoli cells become identifiable and begin to arrange in cord-like structures, the future seminiferous tubules (264) (Fig. 15). High AMH production by Sertoli cells continues until puberty (Fig. 16), irrespective of the disappearance of Müllerian ducts.

AMH is measurable in human serum by ELISA; in 2007 two kits were commercially available, one from Beckman-Immunotech (265) and one from Diagnostic System Laboratoires (DSL) (266). Each uses different sets of antibodies and different AMH standards, which could be the reason why, although there is a high correlation between the two assays, the absolute values with the DSL assay are approximately half of those obtained with the Immunotech kit (267, 268). Measurement of AMH in serum has diagnostic applications as a marker of prepubertal testicular function in boys (269), of follicular reserve in women (270-273), in follow-up of granulosa cell tumors (246, 247, 274) and in prediction of successful in vitro fertilization (271), particularly when levels are measured in follicular fluid (275). By contrast, the clinical usefulness of AMH in seminal fluid in men with non-obstructive azoospermia still needs to be validated (276-279).

The chronological expression of AMH is of greatest importance in sex differentiation (Fig. 16). In the male, AMH must be expressed before Müllerian ducts lose their responsiveness, i.e. before the end of the 8th week in the human fetus (280, 281). Thus, the AMH gene is likely to be under tight transcriptional control. SOX9 –a member of the SRY gene family– plays an essential role in the initiation of AMH expression in the mammalian male fetus. Mutation of the SOX9 binding site of the AMH promoter completely abolishes initiation of AMH production in transgenic mice, while that of SF-1 decreases AMH transcription, but nevertheless allows Müllerian regression to take place (98). SF-1 and another transcription factor, GATA4, bind to response elements present on the AMH promoter to enhance SOX9-activated AMH expression (282, 283). The Wilms tumor associated gene WT-1 and the transcription factor GATA-4 synergize with SF-1 to increase AMH transcription, a synergy antagonized by the X-linked gene DAX-1 (39, 160, 284) (Fig. 18). In males, Sertoli cells continue to produce AMH until puberty when expression is curtailed by the synergistic negative action of androgens and meiotic entry (285).