Hedgehog–BMP signalling establishes dorsoventral patterning in lateral plate mesoderm to trigger gonadogenesis in chicken embryos

The gonad appears in the early embryo after several events: cells at the lateral plate mesoderm (LPM) undergo ingression, begin gonadal differentiation and then retain primordial germ cells (PGCs). Here we show that in the chicken embryo, these events are triggered on the basis of dorsoventral patterning at the medial LPM. Gonadal progenitor cells (GPCs) at the ventromedial LPM initiate gonadogenesis by undergoing ingression, whereas mesonephric capsule progenitor cells (MCPCs) at the dorsomedial LPM do not. These contrasting behaviours are caused by Hedgehog signalling, which is activated in GPCs but not in MCPCs. Inhibiting Hedgehog signalling prevents GPCs from forming gonadal structures and collecting PGCs. When activated by Hedgehog signalling, MCPCs form an ectopic gonad. This Hedgehog signalling is mediated by BMP4. These findings provide insight into embryonic patterning and gonadal initiation in the chicken embryo.

I n animals, the gonads (testis and ovary) are essential for differentiation and maintenance of germ cells, which are required to create the progeny 1,2 . In the testis, somatic Sertoli cells support spermatogenesis through intimate interaction with germ cells 1 . Likewise, in the ovary, the oocyte develops through bidirectional signal exchanges with neighbouring somatic granulosa cells 2 . The testis and ovary secrete sex hormones to differentiate and maintain male and female characteristics, including the internal sex duct and other sexually dimorphic features [3][4][5] . Consequently, failure of gonadogenesis leads to infertility and disorders of sex differentiation 6,7 .
In vertebrates, gonadogenesis starts with the formation of sexually indifferent gonads, which emerge as genital ridges bilateral to the mesentery 8,9 . At early stages, the genital ridges do not exhibit any structural sexual dimorphisms, and maintain the capacity to differentiate into both testis and ovary. Subsequently, the bipotential gonad develops into either a testis or an ovary. The molecular mechanisms underlying sex determination (and subsequent sex differentiation of the gonad) have been investigated extensively in mice 3,10,11 . By contrast, the molecular mechanisms underlying the early stages of gonadal cell differentiation remain to be elucidated.
Bipotential gonads arise from a particular part of the coelomic epithelia 12 , which originate from the lateral plate mesoderm (LPM) in vertebrates 13 . The epithelia of the LPM undergo ingression to initiate genital ridge formation 14,15 . These cells start gonadal differentiation by expressing transcription factors, such as Gata4, Lhx9, Wt1 and Nr5a1 in mouse [16][17][18] . At nearly the same time, primordial germ cells (PGCs), which emerge in the extraembryonic region, migrate and localize to the genital ridge 19,20 in all vertebrates. Accordingly, multiple processes must occur in parallel to accurately form the early stage of the gonad, including formation of the sexually bipotential genital ridge and acquisition of the capability to attract and retain PGCs. However, it remains unclear how these processes are triggered and orchestrated at the correct part of the vertebrate LPM.
Organogenesis is triggered based on early embryonic patterning. Hedgehog and bone morphogenetic protein (BMP) signals determine embryonic pattern formation and regulate various cellular behaviours, including migration and differentiation, in developing embryos [21][22][23][24] . In particular, Sonic hedgehog (SHH) and BMP4 play pivotal roles in the development of the LPM. In chicken embryo, LPM is formed based on mediolateral patterning of mesoderm, which is regulated by BMP4 (ref. 25). By contrast, in mouse and chicken embryo, gut mesenchymal cells derived from splanchnic mesoderm (ventral LPM) undergo orderly differentiation along the radial axis, established by SHH 26,27 . However, it remains unknown what kind of embryonic patterning is necessary for gonadogenesis, and how this embryonic patterning regulates the initiation of gonadogenesis in vertebrates.
In this study, we used in ovo electroporation to precisely localize the gonadal progenitor cells (GPCs) in a particular region of the LPM of day 2 chicken embryos (E2.0). Moreover, we found that Hedgehog and BMP4 signalling play crucial roles in the localization of GPCs by establishing a dorsoventral axis in the medial LPM, followed by the onset of gonadogenesis, in chicken embryo. Our results elucidate the molecular mechanisms that trigger gonad formation.

Results
Ventromedial LPM cells initiate gonadogenesis at E2.0. It is generally accepted that coelomic epithelial cells of the LPM undergo ingression to form the gonadal primordium 14,15 . In chicken embryos, such events should occur from E2.0 to E2.7. However, it remains unclear what subset of coelomic epithelial cells gives rise to the gonad primordium, as well as when these cells start migrating. To address these questions, we traced the lineage of LPM epithelial cells in E2.0 chicken embryos by labelling the cells with DiI (Fig. 1a,f, n ¼ 4; Fig. 1k). The ventromedial aspect of the labelled cells began to migrate soon after DiI injection (Fig. 1b,c,g,h, n ¼ 5; Fig. 1l,m), and obvious thickening was observed at the ventromedial aspect at E3.0 ( Fig. 1d,i, n ¼ 5; Fig. 1n). These cells differentiated into gonadal cells, as revealed by the expression of its marker GATA4 at E4.5 (Fig. 1e,j, n ¼ 4; Fig. 1o). These observations indicate that the cells at the ventromedial aspect of LPM are determined as the GPCs as early as E2.0 ( Supplementary Fig. 1). This result is consistent with the idea that thickening of a particular region of the LPM in the E3.0 embryo is the first sign of gonadogenesis in the chicken 28 .
At E2.0, the ventromedial LPM cells formed an epithelial structure with a laminin-1-positive basement membrane and an atypical protein kinase C (PKC)-positive apical surface ( Supplementary Fig. 2). These cells partially lost epithelial integrity at E3.0; the apical surface was maintained, but laminin-1 accumulated discontinuously ( Supplementary Fig. 2). This degradation of epithelial integrity may be related to ingression of GPCs. The cells in the dorsomedial LPM covering the mesonephros (MCPCs; mesonephric capsule progenitor cells), which we previously termed Neph-CE 29 , were behaviourally distinct from GPCs at the ventral side. Unlike GPCs, the MCPCs did not undergo ingression, but remained as epithelia to form mesonephric capsule ( Fig. 1, n ¼ 4; Supplementary Fig. 1) as we previously showed 29 . Subsequently, PGCs were attracted by the GPC-derived gonadal cells, but not by the MCPCs (Supplementary Fig. 3 Hedgehog signalling is activated in GPCs but not in MCPCs. On the basis of the observations described above, we assumed that the mechanisms that caused the differences between GPCs and MCPCs were crucial for triggering gonadogenesis in GPCs. SHH induces cell differentiation and migration in various contexts, including embryonic development and cancer metastasis 22,30 . SHH is expressed in endoderm adjacent to the GPCs, and the site of SHH expression is more distant from the MCPCs in both mouse and chicken embryos 31,32 ; therefore, we assumed that the higher SHH concentration in GPCs relative to MCPCs is responsible for the differences in Hedgehog signalling activities and responses in GPCs and MCPCs. To test this idea, we investigated the expression of a SHH downstream gene, PATCHED, in GPCs and MCPCs in E2.0 chicken embryos 33,34 . As reported previously, we detected SHH mRNA in the endoderm at E2.0, when GPCs begin to form gonadal primordium by undergoing ingression (Fig. 2a-c,g-i) 31 . Meanwhile, mRNA expression of PATCHED grew stronger in the GPCs. By contrast, expression of PATCHED was absent from the MCPCs (Fig. 2d-i).
SHH was expressed not only in the endoderm but also in axial structures (notochord and floor plate; Fig. 2a-c,g-i). However, even when the axial structures were removed, PATCHED expression was maintained in the GPCs ( Supplementary Fig. 4, n ¼ 4). Furthermore, IHH (Indian Hedgehog), the other Hedgehog ligand in chicken embryos 35 , was not expressed in cells surrounding the GPCs (Supplementary Fig. 5). These results indicate that Hedgehog signalling is activated in the GPCs but not in the MCPCs of E2.0 chicken embryos, and that this differential activation pattern is most likely established by SHH secreted from the endoderm.
Hedgehog signalling is required for GPCs to form gonad. Next, we examined how differential Hedgehog signalling activity induces differential behaviours of GPCs and MCPCs. First, we inhibited Hedgehog signalling to investigate whether GPCs required the signal to form gonadal primordium. Because impairment of Hedgehog signalling throughout the body causes embryonic lethality before the formation of gonadal primordium 32 , we blocked Hedgehog signalling in a spatiotemporally controlled manner. We performed regionspecific gene manipulation by a modification of our previously established method for electroporation of genes into MCPCs 29 . Specifically, we adjusted the position of the electrode slightly to direct the electric pulse ventrally into E2.0 embryos so that the genes were transferred mainly into GPCs (Fig. 3a).
Hedgehog-interacting protein (Hip) lacking the last 22 C-terminal amino-acid residues (HipDC22; Fig. 3a) competitively inhibits binding of Hedgehog ligand to its receptor in the extracellular space 34,36 . Forced expression of HipDC22 in GPCs led to the downregulation of Hedgehog signalling specifically in these cells 12 h after electroporation ( Supplementary Fig. 6, n ¼ 6). In enhanced green fluorescent protein (EGFP)electroporated (control) embryos, GATA4-expressing gonadal primordium started to form ridge structures bilateral to the mesentery at E4.5 (Fig. 3b,e, n ¼ 10; Fig. 3d,g). By contrast, when Hedgehog signalling was interrupted by HipDC22 at E2.0, no genital ridge was formed at the presumptive gonadal area on the electroporated side. Consistent with this, cells in this region failed to express GATA4 (Fig. 3h,k, n ¼ 5; Fig. 3j,m). Moreover, when HipDC22 was overexpressed in GPCs at E3.0, they formed a GATA4 þ genital ridge ( Supplementary Fig. 7, n ¼ 6). These results indicate that Hedgehog signalling activated in GPCs from E2.0 to E3.0 causes them to differentiate into gonadal cells and form a genital ridge.
To investigate whether Hedgehog signalling endows the gonadal primordium with the ability to attract and retain PGCs, we examined the effect of HipDC22 overexpression on the localization of PGCs marked by the expression of SSEA1 or CVH. In chicken embryos, PGCs floating in the blood flow migrate almost equally to the left and right gonad through prospective mesentery ( Supplementary Fig. 3, n ¼ 4; Fig. 3c,f, n ¼ 10; Fig. 3d,g; Supplementary Fig. 8) 19,37 . However, when Hedgehog signalling was inhibited as described above, the PGCs localized ectopically; moreover, they were less abundant in the gonad at electroporated side and more abundant on the untreated side ( Fig. 3i,l, n ¼ 5; Fig. 3j,m; Supplementary Fig. 8), possibly because the gonads did not attract or retain PGCs on the electroporated side and more PGCs were attracted to the gonad on the untreated side. Thus, Hedgehog signalling is indispensable for the settlement of PGCs during the early stage of gonadal development.
SHH is sufficient to trigger gonadogenesis in MCPCs. We next investigated whether overexpressed SHH is capable of inducing gonadal development in MCPCs (Fig. 4a). As shown previously 29 , EGFP-electroporated (control) MCPCs remained as epithelia on a laminin-1-positive basement membrane at E4.5 (Fig. 4b These results indicate that SHH signalling can trigger ectopic gonadogenesis in MCPCs by inducing ingression and well-organized differentiation into gonadal cells that can attract and retain PGCs (Fig. 4m,q). SHH appeared to activate its downstream signalling pathway specifically in early MCPC-derived cells. PATCHED was expressed in the MCPCderived GFP þ cells stimulated by SHH at E3.5, but not E4.5. Furthermore, even in E3.5 embryos, Hedgehog signalling was not upregulated in the underlying mesonephric cells ( Supplementary  Fig. 10, n ¼ 5). Taken together, these results show that differences in Hedgehog signalling activity are responsible for the distinct behaviours of GPCs and MCPCs of E2.0 embryo, and that Hedgehog signalling orchestrates the onset of gonadogenesis in the GPCs. The gonadal competence of the LPM in response to SHH appeared to depend on the proximodistal axis in E2.0 embryos. The lateral region of the LPM, neighbouring the MCPC, is called the somatopleural mesoderm, and it forms the body wall by undergoing ingression (Supplementary Fig. 11a) 13 . When SHH was overexpressed in the somatopleural mesoderm proximal to the MCPCs, these cells ectopically formed a well-organized genital ridge structure (in this structure, GATA4 was expressed broadly and LHX9 is expressed in the cortex; all of these properties were similar to those of the untreated gonad; Supplementary Fig. 11b-k, n ¼ 5). By contrast, the overexpression of SHH in somatopleural cells distal to MCPC did not form this structure ( Supplementary Fig. 11q-u, n ¼ 5).
Therefore, LPM cells appear to have different gonadal competence along the proximodistal axis at E2.0. Notably, when Hedgehog signalling was activated in MCPCs at E3.0 by SHH overexpression, they underwent neither ingression nor differentiation into gonadal cells ( Supplementary Fig. 12, n ¼ 10). This observation suggests that MCPCs lose competence to differentiate into gonad at E3.0. Collectively, these findings indicate that the ability of LPM cells to respond to SHH signalling is spatiotemporally regulated.
Regulation of BMP4 expression in the medial LPM. We next investigated which events were induced in GPCs and MCPCs by Hedgehog signalling. BMP4, which acts downstream of Hedgehog signalling, regulates cell differentiation and migration in many processes, such as embryogenesis and cancer progression 24,38 . Furthermore, BMP4 is upregulated by SHH in the developing gut mesenchyme 26,39 . Therefore, we examined the expression of BMP4 in early chicken embryos. At E1.7, BMP4 is expressed throughout the LPM, including both GPCs and MCPCs, and thus establishes the mediolateral axis in the mesoderm (Fig. 5a,d) 25 . At E2.0, after gonadogenesis has been triggered, BMP4 mRNA expression was downregulated in MCPCs, but maintained in GPCs (Fig. 5b,c,e,f). BMP4 expression at E2.0 and thereafter was similar to that of PATCHED (Fig. 2h,i). These observations raised the possibility that differential expression of BMP4 is due to the differences in Hedgehog signalling activity in GPCs and MCPCs after E2.0.
To address this question, we co-electroporated two plasmids, pCMV-SHH and pCAGGS-EGFP, into the MCPCs at E2.0 and investigated BMP4 mRNA expression at E3.0 (Fig. 6a). BMP4 was expressed by GPC-derived cells, but not MCPCs, on the untreated side (Fig. 6b, n ¼ 6), whereas on the side with SHH overexpression, BMP4 was ectopically expressed in EGFPpositive MCPC-derived cells and their neighbours (Fig. 6c, n ¼ 6). When Hedgehog signalling was inhibited in GPCs by electroporation of HipDC22 (Fig. 6d), BMP4 expression was downregulated in GPCs on the electroporated side but not on the unelectroporated side (Fig. 6e, n ¼ 6). These results indicate that Hedgehog signalling is responsible for differential expression of BMP4 between GPCs and MCPCs after E2.0.
These results indicate that BMP signalling induces ingression of MCPC and differentiation into gonadal cells that can retain PGCs. Moreover, when BMP4 was overexpressed in the somatopleure proximal to MCPC, which formed an ectopic gonad on SHH overexpression, gonadogenesis was not complete: although these BMP4-overexpressing cells ectopically expressed GATA4, they did not express LHX9 (Supplementary Fig. 11l- We next investigated whether GPCs require BMP signalling to initiate gonadal differentiation. To this end, we electroporated the secretory BMP antagonist Noggin 25,40 into GPCs and investigated whether the gonad was formed at E4.5 (Fig. 8a). We observed neither the expression of GATA4 nor the formation of a genital ridge at the Noggin-overexpressing side (Fig. 8h,k, n ¼ 5; Fig. 8m). By contrast, EGFP-overexpressing (control) GPCs or untreated GPCs formed a GATA4-positive genital ridge (Fig. 8b,e, n ¼ 10; Fig. 8d,g; Fig. 8h, n ¼ 5; Fig. 8j).
SSEA1-positive PGCs localized to the genital ridge derived from EGFP-overexpressing (control) GPCs (Fig. 8c,f, n ¼ 10; Fig. 8g). By contrast, in Noggin-overexpressing embryos, PGCs did not localize to the presumptive gonadal region at the electroporated side. Instead, more PGCs were retained at the genital ridge on the untreated control side (Fig. 8i,l, n ¼ 10; Fig. 8m; Supplementary Fig. 8). Although Noggin inhibits BMP2 and BMP7 as well as BMP4 (ref. 41), BMP2 and BMP7 mRNAs were not detected in the GPCs or surrounding cells at E2.0 ( Supplementary Fig. 13). Taken together, these findings indicate that altered BMP4 expression at E2.0 established a dorsal (MCPCs) and ventral (GPCs) patterning in the medial LPM and orchestrates GPC behaviours to initiate gonadogenesis downstream of Hedgehog signalling.

Discussion
In most multicellular organisms, the gonad is essential for the production of offspring. However, the molecular mechanism underlying the onset of gonadogenesis in vertebrates has remained largely unexplored, mainly because the location of GPCs within the LPM and the precise timing of the onset of gonadogenesis remained unknown. In this study, we demonstrated that GPCs are positioned at the ventromedial LPM, and that these cells initiate gonadogenesis by undergoing ingression in chicken 2-day (E2.0) embryos. Meanwhile, the dorsomedial LPM cells covering the mesonephros (MCPCs) maintain their epithelial integrity. These distinct behaviours of GPC and MCPC appear to be controlled by Hedgehog signalling. Subsequent to these events, Hedgehog signalling triggers gonadal differentiation by activating BMP4 expression in GPCs (Fig. 8n).
The earliest morphogenetic event of gonadogenesis is considered to be the emergence of the epithelial thickening called the genital ridge, which is identifiable starting at E3.0 in chicken and E10.0 in mouse embryos 28 . Here we showed that GPCs start undergoing ingression as early as E2.0 in chicken embryos (Fig. 1). As far as we know, this is the earliest tissue/cellular sign of gonadogenesis reported to date.
This earliest event of gonadal development is most likely evoked by SHH emanated from endoderm. Moreover, it is noteworthy that SHH overexpression also caused multiple features normally observed during gonadal development, including formation of the genital ridge-like structure and collection of PGCs. Collectively, these data show that SHH is the most upstream molecule involved in triggering and orchestration of gonadogenesis in chicken.
In this study, we demonstrated that GATA4 and LHX9, which are gonadogenesis-related transcription factors in mouse 16,17 , are downstream targets of Hedgehog signalling in chicken. Moreover, we found that BMP4 regulates gonadal initiation downstream of Hedgehog signalling. In chicken, forced expression of BMP4 enabled MCPCs to show several aspects of gonadal initiation, but could not induce the formation of a genital ridge-like structure in and around MCPCs. These results imply that Hedgehog signalling controls downstream molecules in addition to BMP4 to form the ridge structure.
In addition, we demonstrated that expression patterns of BMP4 dynamically change before and after gonadal initiation. Before E2.0, BMP4 is broadly expressed in LPM cells including MCPCs and GPCs, and form the LPM structure 25 . Subsequently, BMP4 expression is restricted only to GPCs after E2.0. An appropriate transition of BMP4 expression might be important for correct formation of the LPM and subsequent gonadal initiation. Future studies should seek to elucidate the mechanisms underlying this transition in BMP4 expression, in which Hedgehog signalling plays an essential role.
We also found that ectopic gonadal induction by forced expression of SHH or BMP4 depended on the position within the LPM, as well as on developmental stage. In E2.0, MCPCs could respond to both SHH and BMP4, nearby proximal somatopleural cells could respond to SHH but not to BMP4, and distal somatopleural cells far from MCPCs could respond to neither. At E3.0, MCPCs no longer responded to SHH. Thus, gonadal competence seems to be regulated within LPM cells in a spatiotemporal manner. This regulation might contribute to placing gonads in the correct position or to the robustness of gonadal formation.
Although SHH and BMP4 play central roles in the initiation of gonadal differentiation in chicken embryos, these molecules have not been shown to be necessary in mice so far 42,43 . However, Shh and Bmp4 are expressed in mice embryo in a manner similar to the chicken 32,44 . Furthermore, primary cilia responsible for the distribution of secreted Hedgehog affect the gonadal length, and BMP signalling is required for correct localization of PGCs at the early genital ridge in mice 44,45 . Hedgehog and Bmp4 signalling might be involved in the initiation of gonadal differentiation in mouse embryos. It would be interesting to determine whether, and to what extent, gonadal initiation processes vary among amniotes.

Methods
Chicken embryos. Fertilized chicken eggs were commercially obtained from the Shiroyama Farm (Sagamihara, Japan). All animal experiments were conducted with the ethical approval of Kyoto University (No.H2620).
DiI labelling. DiI (Invitrogen) was dissolved in ethanol (0.1%), heated at 45°C for 3 min and diluted 1:10 with 0.3 M sucrose. This solution was injected into the coelom of the right side of E2.0 embryos using a glass capillary. Thereafter, the embryos were incubated for various time periods, killed, fixed overnight at 4°C in PBS containing 4% paraformaldehyde (PFA) and sectioned by cryostat at 10-mm thick on Platinum Pro-coated glass slides (Matsunami). Images were obtained on an AxioPlanII microscope with the Apotome system (Carl Zeiss).
In ovo electroporation. Expression plasmids were suspended in EB buffer (Qiagen) containing 2% Fast Green FCF (Nacalai Tesque) and 8% sucrose. The Preparation of an anti-CVH antibody. A full-length Cvh cDNA fragment was subcloned in-frame into pGEX-5X3 (Pharmacia). The GST-CVH fusion protein was purified using a GST fusion system (Pharmacia) according to the manufacturer's instructions. The purified protein (300 mg) was injected into a rat. The resulting antiserum was purified by affinity chromatography using GST-CVH-conjugated agarose beads, and was used as an anti-CVH antibody. They were then incubated with hybridization buffer (Ultra hyb; Ambion) for 5 min at 65°C. Hybridization was carried out with a DNP-labelled RNA probe (1 ng ml À 1 ). The sections were washed with wash solution 1 (50% formamide, 5 Â SSC, pH 4.5, and 1% SDS), wash solution 2 (50% formamide and Â SSC, pH 4.5) at 65°C, and a 1:1 mixture of wash solution 2 and TNT. After washing three times in TNT for 5 min each at RT, the specimens were preblocked with blocking solution (1% blocking reagent (Roche)/TNT) for 1 h at RT, followed by overnight incubation at 4°C in the blocking solution containing 1:1,000 dilutions of an anti-DNP-HRP-conjugated antibody (PerkinElmer) and anti-GFP antibody (rabbit, Invitrogen). After three washes in TNT, the specimens were reacted with TSA plus Cy3 system for 10 min at RT. The reaction was terminated by washing three times in TNT. To visualize EGFP, the sections were incubated with an anti-rabbit IgG-Alexa 488-conjugated goat antibody (Invitrogen) diluted 1:500 in blocking solution, for 1 h at RT.
To detect mRNA with NBT/BCIP, sections were incubated with hybridization buffer for 5 min at 65°C after two washes in TNT. The solution was replaced with prewarmed hybridization buffer containing Dig-labelled RNA probes and incubated overnight at 65°C. The sections were washed with wash solution 1, wash solution 2, and a 1:1 mixture of wash solution 2 and TNT. After three washes in TNT, the samples were preblocked and incubated overnight at 4°C in the blocking solution containing an anti-DIG-alkaline phosphatase-conjugated antibody and anti-GFP antibody. The samples were washed three times in TNT, followed by washing in NTMT (100 mM Tris-HCl, pH 9.5, 100 mM NaCl, 50 mM MgCl 2 and 0.1% Tween 20)/2 mM Levamisole. The alkaline phosphatase activity was visualized by incubating the samples in NTMT containing nitroblue tetrazolium chloride, 5-bromo-4-chloro-3-indolyl phosphate and levamisole. The colour reaction was stopped by TNT washing, and then the sections were mounted. Images were obtained using a TCS SP6 microscope (Leica).
Depletion of axial structures and explant culture. The surgical manipulations were performed on E2 embryos. A slit was made between the neural tube and right somite along the anteroposterior axis, with a sharpened tungsten needle. The embryo was then separated into a right axial structure-depleted side and a left control side. Both parts were fixed with PFA soon after the manipulation or incubated for 4 h at 38°C on a 0.8-mm filter (Millipore) placed in a 6-cm Center-Well Culture Dish (Falcon) containing 10% fetal bovine serum/DMEM (Nissui).
Visualization of CVH and EGFP proteins in whole E4.5 embryos.
Data availability. Data supporting the findings of this study are available within the article and its Supplementary Information files and from the corresponding author on reasonable request.