Early gonadogenesis in mammals: Significance of long and narrow gonadal structure


Department of Veterinary Anatomy, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo, 113-8657, Japan. E-mail: aykanai@mail.ecc.u-tokyo.ac.jp


In mammalian embryogenesis, the gonadal primordium arises from the thickening of the coelomic epithelium, which results in a pair of extremely long and narrow gonadal structures along the anteroposterior axis. These gonadal structures are conserved in various mammalian species, suggesting a great advantage in properly receiving migrating primordial germ cells (PGCs) that are widely scattered throughout the hindgut tube. Soon after the PGCs settle, the bipotential gonads undergo sex determination into testes or ovaries by the sex-determining gene, Sry, which is expressed in supporting cell precursors in a center-to-pole manner. Such a long, narrow gonadal structure bestows a considerable time lag on Sry expression between the center and pole regions, but testiculogenesis with cord formation and Leydig cell differentiation occurs synchronously throughout the whole organ. This synchronous testiculogenesis could be explained by a positive-feedback mechanism between SOX9 (another SRY-related transcription factor) and FGF9 downstream of Sry. FGF signals are likely secreted from the center region, rapidly diffuse into the poles, and then induce the establishment of SOX9 expression in Sertoli cells in the pole domains. This work focuses on recent knowledge of the molecular and cellular events of PGC migration, gonadogenesis, and testiculogenesis, and their biological significance in mammalian embryogenesis. Developmental Dynamics 242:330–338, 2013. © 2012 Wiley Periodicals, Inc.


Germ cells originate from outside the gonads (see reviews by Saga, 2008; Richardson and Lehmann, 2010). In the mouse, primordial germ cells (PGCs) arise from the proximal epiblast cells through extra-embryonic ectoderm-derived BMP (bone morphogenetic protein) signals in the early pre-gastrulation stages of mouse embryos (∼6.0 days post-coitum: dpc; Lawson et al., 1999). They subsequently move to the base of the allantois, the extraembryonic mesoderm, at the early gastrulation stage (∼7.25 dpc; see review by Sasaki and Matsui, 2008, Saitou et al., 2012). These PGCs also express several key genes, such as Nanos3, Dppa3/Stella/Pgc7, Prdm1/Blimp1, and Prdm14 (Yabuta et al., 2006; Kurimoto et al., 2008; Saitou et al., 2012). In contrast, formation of the genital ridge begins in the intermediate mesoderm at the early-organogenic stage, approximately 3 days after PGC specification (∼10.0 dpc). Until the onset of the gonadogenesis, PGCs remain at the base of the allantois as a cluster of approximately 40 alkaline phosphatase-positive cells (7.25 dpc; Fig. 1A), and then move toward the genital ridge within the posterior trunk (Molyneaux et al., 2001) using the morphogenic movement of the hindgut endoderm sheet in a conveyer-belt-like manner (Hara et al., 2009; Fig. 1B–E).

Figure 1.

Spatiotemporally close association between PGCs migration and hindgut morphogenesis at the late gastrulation/early somite stages. A–E: Schematic representation (top, sagittal section views; bottom, posteroventral views), showing spatiotemporal patterns of PGCs migration (red and pink round cells) and hindgut morphogenic movement (solid black arrows) in the hindgut entrance sites (arrowheads between “ge” and “exe”) of mouse wild-type embryos. PGCs appear to move from the allantois into the outer endoderm layer simultaneously with the morphogenic movement of hindgut progenitor cells (black solid arrows in A–C). They are then non-cell-autonomously transferred from the hindgut entrance into the intra-embryonic hindgut tube in a conveyor belt-type manner (C, D), resulting in wider PGC distribution along the AP axis near the pre-genital ridge regions (gr) of the posterior trunk (E). exe, extraembryonic visceral endoderm; ge, gut endoderm; gr, coelomic epithelial region corresponding to the pre-genital ridge. F: Schematic representation (left, ventral view; right, transverse section view) of the formation of an extremely long and narrow gonadal structure covering the hindgut tube including PGCs (red cells). Broken red arrows indicate the direction of PGCs' movement. md, mullerian duct; wd, wolffian duct.

Genetic analyses using Sox17-null (hindgut endoderm-deficient) (Kanai-Azuma et al., 2002) and chimeric embryos mixed with wild-type hindgut endoderm cells has shown the importance of the synchronized movement of hindgut endodermal cells and PGCs at the posterior end of the primitive streak for proper migration of PGCs to genital ridges (Fig. 1A–C). This finding was based on the observation of ectopic PGC migration into the yolk sac visceral endoderm in the hindgut-deficient embryos (Hara et al., 2009). The use of endodermal morphogenic movement for PGC migration appears to be a conserved cellular event in invertebrates, such as Caenorhabditis elegans and Drosophila (“hitchhiking” endoderm movement and “remodeling” of endodermal epithelium; DeGennaro et al., 2011; Chihara and Nance, 2012; Seifert and Lehmann, 2012). Interestingly, Nanos1, a PGC specification factor, is required to prevent the expression of various key endoderm genes, including Sox17 in Xenopus PGCs (Lai et al., 2011, 2012). A similar association between these definitive endoderm and PGC lineages was also suggested by findings showing the transient expression of Sox17, the gut endoderm-determining gene, in mouse PGCs at the base of the allantois at around 7.25 dpc (Yabuta et al., 2006). In light of this co-localization of hindgut endodermal progenitors around a PGC cluster, these findings clearly suggest the regionally close association and segregation between these two lineages at the posterior end of the primitive streak. This idea may be consistent with our previous findings showing a highly efficient contribution of Sox17-null ES cells into the PGC lineage to compete with host-derived wild-type PGCs in chimeric embryos (Hara et al., 2009).

The elongation of the hindgut, including the PGCs, leads to the entry of PGCs from the extra-embryonic endoderm into embryonic regions near the genital ridges (Fig. 1D, E) and their subsequent migration away from the hindgut toward the genital ridges through the mesentery at around 9.5–10.0 dpc (Fig. 1F; Molyneaux and Wylie, 2004). This active migration of PGCs from the hindgut tube toward the genital ridge has been shown to be regulated through several survival and chemotaxis signals involving SDF1-CXCR4 (Ara et al., 2003; Molyneaux et al., 2003), SCF (secreted KitL)-ckit (Runyan et al., 2006; Gu et al., 2009), and WNT5a-ROR2 (Laird et al. 2011; Chawengsaksophak et al., 2012).


The use of hindgut morphogenesis in PGC movement could anatomically explain the wider distribution of PGCs throughout the hindgut region along the anteroposterior (AP) axis. Interestingly, the gonadal primordium arises from the thickening of the coelomic epithelium at the bilateral regions covering the whole hindgut area, resulting in the formation of an extremely long and narrow gonadal structure along the AP axis (Fig. 1E, F). This unique long, narrow gonadal structure is conserved in various mammalian species, suggesting a great advantage in properly receiving migrating PGCs that are widely scattered throughout the hindgut tube. Since several Hox genes, such as Hoxc5, Hoxc6, and Hoxd9, were previously shown to be expressed in the gonadal region (Dollé and Duboule, 1989; Gaunt et al., 1990; Spirov et al., 2002), these Hox gene actions may be involved in the determination of the position and length of the gonadal primordium along the AP axis. However, its precise mechanism remains unclear.

It became evident that various transcription factors, including EMX2, WT1, SF1/Ad4Bp/NR5A1, LHX9, GATA4/FOG2, CBX2/M33, and others, are crucial for the proper thickening of the gonadal primordium covered with the coelomic epithelium (see reviews by Wilhelm et al., 2007; Munger and Capel, 2012). Recently, Kusaka et al. (2010) used a cell-labeling method to demonstrate that some coelomic epithelial cells proliferate, undergo epithelial mesenchymal transition (EMT), and migrate into the dorsal inner layers of the coelomic epithelium through the basal lamina layer, resulting in the formation of the gonadal primordium at 10.0–11.0 dpc (Fig. 2). In Emx2-null gonads, coelomic epithelial cells have severe defects in their EMT and subsequent ingression toward the dorsal side due to the aberrant regulation of EGF receptor signaling, which results in the hypoplastic gonadal primordium covered by the irregularly protruding of the coelomic epithelium (Kusaka et al., 2010). In contrast, in Cbx2-null gonads, the proliferation and ingression of the coelomic epithelial cells appear to be normal, but the gonadal cells show defective proliferation, especially in the posterior half of the gonadal primordium (Katoh-Fukui et al., 2012). These Cbx2-null phenotype data also provide the first direct evidence of regionally distinct mechanisms between anterior and posterior halves in the thickening gonadal primordium of mammals.

Figure 2.

Spatiotemporal expression pattern of Sry and Sox9 during early gonadogenesis and testiculogenesis. Schematic representation (left, lateral view; right, transverse section view in the center) showing a center-to-pole wave-like (left images) and ventral-to-dorsal pattern (right images) of Sry (blue broken lines) and Sox9 (green circles) expression in the developing XY gonads (Bullejos and Koopman, 2001; Sekido et al., 2004; Kidokoro et al., 2005). Testiculogenesis, including testis cord formation and subsequent Leydig cell differentiation, occurs synchronously throughout the entire organ from 12.5 dpc. Using tail somite (ts) stages, 10.5 dpc corresponds to approximately 8 ts, 11.0 dpc to 12 ts, 11.5 dpc to 18 ts, 12.0 dpc to 24 ts, and 12.5 dpc to 30 ts (Hacker et al., 1995).

During early gonadogenesis, several gonadal cell markers can allow visualization of the multiple layers of the thickening gonadal primordium along the dorsoventral (DV) axis. For example, Mfge8, which marks the gonadal somatic cells restricted to the border region between the gonads and mesonephros (Fig. 2: lightest gray area in images on right), begins to be expressed in the coelomic epithelium at 10.0 dpc, after which the Mfge8-positive cell layer expands and is mainly restricted to the border region of the genital ridge (Kanai et al., 2000). In the supporting cell lineages, male-type supporting cells (i.e., pre-Sertoli cells positive for either SRY or SOX9) appear to move from the coelomic epithelium during the time window of 11.0–11.5 dpc in XY gonads (Fig. 2: blue oblique lines and green circles; Karl and Capel, 1998; Capel, 2000; Sekido et al., 2004; Kidokoro et al., 2005). Subsequently, during testiculogenesis, the coelomic epithelium are completely separated from the gonadal parenchyma by the formation of the tunica albuginea (Fig. 2: white broken line in bottom right image), resulting in no ingression of the coelomic epithelium toward the inner testicular parenchyma at later stages (i.e., ∼12.5 dpc). In XY gonads, the male-specific coelomic epithelial proliferation appears to lead the recruitment of additional supporting (i.e., pre-Sertoli) cells during the later half of this time window (11.25–11.5dpc; Schmahl et al., 2000; Schmalh and Capel 2003; Fig. 2: dark gray region). In contrast, XX gonads appear to show the lack or reduced contribution of the coelomic epithelium to the supporting cell lineage due to the female-specific low proliferation activity in the coelomic epithelium of XX gonads at the same stages. Moreover, in XX gonads, the coelomic epithelia are anatomically connected with ovarian parenchyma by the postnatal stages. Recently, granulosa cells (i.e., primordial follicles) were shown to be newly recruited from the coelomic epithelium at the perinatal and early postnatal periods (Mork et al., 2012), suggesting a partial contribution of the coelomic epithelial cells to ovarian cortex formation in postnatal ovaries. Taken together, these findings suggest that a stage- and region-specific proliferation, EMT, and ingression of the coelomic epithelium roughly contribute to the gonadal somatic cells located in spatial order along the dorsoventral axis of developing gonads (Fig. 2; gray gradient in right images).


Soon after the PGCs settle in the genital ridges (∼12 hr in mice), the gonadal primordium undergoes organ fate determination into the testis or ovary by the action of the sex-determining gene, Sry, which encodes an HMG-domain transcription factor (see reviews by Kanai et al., 2005; Kashimada and Koopman, 2010). The long gonadal structure has a great advantage in properly receiving migrating PGCs at around 9.5–10.5 dpc (Fig. 1F), but appears to bestow a considerable time lag (more than 6 hr in mice) on the onset of Sry expression between the center and pole regions at 11.0–12.0 dpc (Fig. 2; blue shaded lines in left images; Albrecht and Eicher, 2001; Bullejos and Koopman, 2001). In brief, Sry expression is first detected in the central region of the XY gonad at 11.0 dpc (approximately the 12 tail-somite [ts] stage) and extends to both the anterior and posterior ends by 11.5 dpc (18 ts). Thereafter, its expression is rapidly down-regulated in the anterior and middle regions, becoming restricted to the posterior pole before it completely disappears at around 12.5 dpc (30 ts) (Fig. 2). Since several gonadal markers of undifferentiated supporting cells (e.g., Sprr2d) show a similar center-to-pole expression pattern in early phases of gonadogenesis (Lee et al., 2009), these findings suggest that this center-to-pole pattern of SRY-positive cells reflects the spatiotemporal differentiation pattern of the supporting cell progenitors in the thickening gonadal primordium.

In contrast to such a center-to-pole wave-like pattern in the developing XY gonads, an anterior-to-posterior pattern was previously shown especially in the developing XX gonads: (1) meiotic induction of female germ cells (Menke et al., 2003; Koubova et al., 2006; Bowles et al., 2006; Chen et al., 2012), and (2) reduced expression of several undifferentiated pre-granulosa cell markers (Lee et al., 2009). Although this anterior-to-posterior wave-like pattern may be partially induced by several diffusible signaling factors enriched in the adjacent mesonephric tissue (e.g., WNT4 [Mizusaki et al., 2003] and retinoic acid [Bowles et al., 2006]), its precise mechanisms are still unclear.


SRY directly induces the up-regulation of Sox9 (Sry-related HMG box-9), a fundamental testis-differentiation gene common to various vertebrate species, in Sertoli cell precursors (Sekido and Lovell-Badge, 2008), which in turn upregulates other genes involved in the differentiation of Sertoli cells. In XY gonads, SOX9 expression subsequently induces male-specific FGF9 expression, after which FGF9 signals promote the maintenance of high-level SOX9 expression in differentiated Sertoli cells via a positive-feedback (Kim et al., 2006). In addition, female-specific WNT4/RSPO1-β-catenin signals have been shown to repress the male-specific SOX9-FGF9 positive-feedback loop in the developing XX ovaries (Kim et al., 2006; Chassot et al., 2008; Maatouk et al., 2008; Tomizuka et al., 2008; Bernard et al., 2012).

Sekido et al. (2004) reported that SRY expression is transiently limited to ∼8 hr in each supporting cell. To resolve the critical time window of SRY action, we established a heat-shock-inducible SRY transgenic (Tg) mouse system that allowed induction of testis development in cultured XX genital ridges at various time points during development (Fig. 3A; Hiramatsu et al., 2009). Using this transient SRY-inducible system, we found that the ability of Sry to determine testis development was limited to approximately 11.0–11.25 dpc (12–15 ts), a window of only 6 hr (Fig. 3A: blue arrows). After this critical time period (11.25–11.5 dpc), ectopic Sry induction initially induced SOX9 expression. However, high levels of Sox9 expression were not maintained, resulting in ovarian differentiation (Fig. 3A: purple arrows). We also showed that this unexpectedly narrow time window of testis determination is non-cell-autonomously defined by the availability of male-specific high-FGF9/low-WNT4 signaling states, which leads to the maintenance of high levels of SOX9 expression and subsequent testis cord formation (Colvin et al., 2001; Schmahl et al., 2004; Kim et al., 2006). Since early gonadal somatic cells are primed with a female bias despite their bipotentiality (Munger et al., 2009; Jameson et al., 2012b), these data strongly indicate the overarching importance of Sry action at the initial 6-hr phase (11.0–11.25 dpc), which is crucial to the switch from female- to male-specific patterns of FGF9/WNT4 signaling states in the developing gonads.

Figure 3.

Spatiotemporal pattern of the SOX9/FGF9 positive feedback system immediately downstream of SRY in mouse testiculogenesis. A: Critical time-window of SRY action in the heat-inducible HSP-Sry Tg gonads (Hiramatsu et al., 2009). In XX Tg gonads at 12–14 ts (11.0–11.25 dpc), artificial Sry induction cell-autonomously induces Sox9 activation in pre-Sertoli cells. Such SRY/SOX9 expression non-cell-autonomously leads to high FGF9 expression, which maintains high-level Sox9 expression (blue arrows). The delayed Sry induction at 16–21 ts can promote initial Sox9 activation, but it fails to maintain SOX9/FGF9 expression, resulting in the ovarian development at later stages (purple arrows). B: Mechanism of spatiotemporal testiculogenesis along the AP axis in the developing XY gonads (Hiramatsu et al., 2010). In the central domain, Sry is initially activated at 12 ts, which leads to Sox9 expression at 13–14 ts and Fgf9 expression at 15–16 ts. FGF9 signals maintain high-level SOX9 expression in a positive feedback manner in the center, and FGF9 rapidly diffuse toward the poles. A center-to-pole wave-like pattern of Sry expression causes its delayed expression in the pole domain, but the diffused FGF9 supports a rapid establishment of high-level SOX9 expression and ensures synchronized testis differentiation throughout the whole gonad.

To understand the vital importance of the action of SRY, we conducted segment, reconstruction, and partition culture assays (i.e., three equal [anterior, middle, and posterior] segments of the XY genital ridge and their recombinant gonad were cultured in the presence or absence of the partition between these segments) at this critical stage (11.0–11.25 dpc), in combination with the addition of various diffusible signaling factors and their inhibitors (Hiramatsu et al., 2010). This investigation was also based on our previous findings that showed the importance of the center domain in the testiculogenesis of the pole domains at this initial 6-hr stage by segment assay using XY genital ridge cultures of anterior, middle, and posterior segments (Hiramatsu et al., 2003). The results clearly demonstrated (1) that certain soluble/diffusible factors secreted from the gonadal center domain at this initial critical period contribute to proper testiculogenesis in the pole domains at the later stage; (2) FGF9 is one of the center-derived signaling factors that can rescue defective testiculogenesis in the pole domains; and (3) inhibition of FGF signaling represses the expansion of the Sox9 expression domain toward the poles, leading to typical ovotestis formation. To monitor two sites of Fgf9 expression and its signal transmission, we compared the spatiotemporal expression patterns of Fgf9 and Dusp6, a downstream mediator of the FGF signal, in the developing gonads in vivo using whole-mount in situ hybridization. Dusp6 expression was sexually dimorphic, being higher in XY gonads than in XX gonads at 11.5 dpc. Interestingly, Dusp6 expression in the coelomic epithelial and subepithelial cells was rapidly up-regulated over the whole gonadal area immediately after the onset of Fgf9 activation in the central domain at 11.3 dpc, despite incomplete expansion of Fgf9 expression toward the poles even at 11.5 dpc. Taken together, these data indicate that the overarching importance of SRY action at the initial 6-hr phase (11.0–11.25 dpc; Fig. 3) is due to a very rapid diffusion toward the poles of center-derived FGF9 molecules downstream of SRY, leading to the rapid establishment of high levels of SOX9 expression in pole domains (Fig. 3). This idea is consistent with the fact that XYPOS, XYAKR (“delayed” Sry expression) and Fgfr2-null XY gonads (loss of the FGF receptor with the highest affinity for FGF9) have the same phenotype with the feminization of pole regions (Schmahl et al., 2004; Kim et al., 2007; Wilhelm et al., 2009; Correa et al., 2012).


The center-to-pole expansion system with positive feedback loops likely ensures a rapid female-to-male switching throughout the whole organ. This mechanism may be especially important in the case of long, narrow structures, such as the mammalian gonadal primordium. FGF9 secretion from the gonadal center domain appears to rapidly diffuse toward the poles to stabilize high-level SOX9 expression in pre-Sertoli cells (Colvin et al., 2001; Schmahl et al., 2004) and to recruit pre-Sertoli cell precursors (Karl and Capel, 1998; Schmahl et al., 2004), consequently leading to the rapid progress of testiculogenesis in the pole domains. Such a mechanism would explain the synchronous tubulogenesis in the entire gonadal area (Coveney et al., 2008), despite the considerable time lag in the onset of Sry expression between the center and pole regions (Bullejos and Koopman, 2001; Albrecht and Eicher, 2001).

Moreover, other soluble/diffusible factors may be involved in the expansion system of testiculogenesis toward the poles in vivo. Soluble/diffusible factors that were up-regulated immediately downstream of SRY/SOX9 could support the establishment of high levels of SOX9 expression in a positive-feedback manner. Among FGF family members, except FGF9, Fgf10 has also been shown to be more highly expressed in XY gonads than in XX gonads at 11.5 dpc (Hiramatsu et al., 2010). Its expression was also higher in the central domain than in the pole domains, suggesting the functional redundancy of FGF9 and FGF10 in testiculogenesis. The PGD2 signaling pathway has been shown to amplify SOX9 activity immediately downstream of SRY/SOX9 in testis differentiation (Malki et al., 2005; Wilhelm et al., 2005; Moniot et al., 2009). Some ECM components were also shown to be involved in positive feedback regulating the maintenance of high levels of SOX9 expression, leading to proper tubulogenesis (Matoba et al., 2008). Taken together, this system of SOX9 and diffusible positive-feedback signaling factors is clearly advantageous to the rapid and synchronous switch of sex in the gonads, especially in the long and narrow structure of mammal gonads.


In this review, we have focused mainly on recent findings regarding the molecular and cellular events of PGC migration, gonadogenesis, and early testiculogenesis and have discussed their biological significance in terms of anatomy and morphogenesis. However, as these findings were derived mainly from mouse embryos, it remains unclear whether these events are mouse-specific, mouse/human-specific, or events common to various mammalian species. To resolve these questions, other novel animal models that allow evaluation of conserved or non-conserved molecular and cellular events in early gonadogenesis and testiculogenesis must be developed. Moreover, this review does not refer to an important finding regarding postnatal sex reversal in Dmrt1-null testis (Matson et al., 2011) and Foxl2-null ovary (Uhlenhaut et al., 2009). These two mutant gonads undergo proper sex differentiation and development until the late-fetal stage (Raymond et al., 2000; Ottolenghi et al., 2005, 2007; Kim et al., 2007). These data clearly imply that the maintenance of the phenotypic sex of supporting cells at later stages may be regulated in a manner distinct from the initial cellular and molecular events immediately downstream of SRY in mammals. These two distinct mechanisms could also explain the controversial actions of WNT4 and DAX1/NR0B1; i.e., their SOX9-inducible action in initial testiculogenesis during the SRY expression period (Meeks et al., 2003; Jeays-Ward et al., 2004; Bouma et al., 2005) and their anti-testis actions at later stages (Swain et al., 1998; Vainio et al., 1999; Jameson et al., 2012a; Ludbrook et al., 2012). Further detailed analyses of sex reversal processes at the fetal and postnatal stages are needed to resolve the molecular and cellular events common to these sex differentiation/sex reversal processes in mammals.


The authors thank Drs. Ryuji Hiramatsu and Kenshiro Hara for kindly providing the original illustration files and critical readings of this manuscript. Our thanks also to Prof. Drs. Tsunekawa Naoki, Masamichi Kurohmaru (University of Tokyo), and Prof. Dr. Masami Kanai-Azuma (Tokyo Medical and Dental University) for their kind and helpful support of this work. Thanks also to Drs. Yoshiko Kuroda and Miyuri Kawasumi for their kind technical support and to Ms. Itsuko Yagihashi for her secretarial assistance. Kyoko Harikae is a DC1 JSPS Research Fellow.