The SRY-related HMG box 9 (SOX9) gene, a member of the gene family of SOX transcription factors, is implicated in different processes of organogenesis in vertebrates (for a review, see Kiefer,2007; Lefebvre et al.,2007). In mammals, SOX9 has been shown to regulate both chondrogenesis and sex determination. Heterozygous mutations in the SOX9 gene result in Campomelic Dysplasia (CD), a lethal human disorder characterized by skeletal malformations associated with male to female sex reversal in 75% of XY patients (Foster et al.,1994; Wagner et al.,1994). SOX9 is expressed in pre-Sertoli cells just after the SRY gene (Kent et al.,1996; Sekido et al.,2004) and recently it has been shown that SRY directly enhances SOX9 transcription (Sekido and Lovell-Badge,2008). SOX9 deletion causes male to female sex reversal in mice (Chaboissier et al.,2004) and humans (Wagner et al.,1994; Smyk et al.,2007). Duplication or overexpression of this gene in XX individuals leads to sex reversal in the opposite direction (female to male; Huang et al.,1999; Bishop et al.,2000; Vidal et al.,2001). The SOX9 protein can functionally substitute for the SRY protein to induce testicular differentiation (Bergstrom et al.,2000). SOX9 is a highly conserved gene of the testis determination pathway in vertebrates (Foster et al.,1994; Wagner et al.,1994; Kent et al.,1996; Morais da Silva et al.,1996) as well as in Drosophila (DeFalco et al.,2003). It has been reported that SOX9 is up-regulated in the male gonad and down-regulated in the female gonad in both mammalian and nonmammalian species such as birds and reptiles (for a review, see Morrish and Sinclair,2002). However, in the anamniote species, the putative sex-determining role of SOX9 remains unclear because SOX9 transcripts were found in the developing and adult gonad in both sexes of fishes (Yokoi et al.,2002; Nakamoto et al.,2005) and amphibians (Takase et al.,2000; Osawa et al.,2005). Although these studies suggested that SOX9 might not be important for male sex determination, they did not exclude a conserved role of SOX9 in gonad differentiation. It has been well established that the key factor for the biological activity of the SOX9 protein in gonad morphogenesis is its nuclear subcellular localization within the differentiating Sertoli cells (for a review, see Smith and Koopman,2004). We reinvestigated the role of the SOX9 gene in gonadogenesis in amphibians, a class of vertebrate, which occupies a key phylogenetic position in vertebrate evolution. X. tropicalis has emerged recently as a model amphibian because of its numerous advantages for genomic and developmental studies, such as its sequenced genome and short life cycle (Hirsch et al.,2002; Song et al.,2003; Fort et al.,2004; Grammer et al.,2005; El Jamil et al.,2008). We have recently reported a detailed description of the process of gonadal development in this species (El Jamil et al.,2008). Based on these results, we have attempted to tackle the following questions: (1) in which type of cells does SOX9 expression take place in the amphibian gonad, (2) is the SOX9 mRNA translated into a protein, and (3) what is the sub-cellular localization of this protein in both male and female gonads? With this aim, we have analyzed the spatiotemporal expression of the SOX9 gene in the developing gonad of X. tropicalis by in situ hybridization and immunofluorescence labeling. Our results show that SOX9 mRNA and protein are expressed in both male and female gonads after metamorphosis when the gonads are well differentiated. In the male gonad, SOX9 expression takes place within the supporting cells whereas its expression is restricted to primary oocytes in the female gonad suggesting that the SOX9 gene has a different role in male and female amphibian gonadogenesis.
A search of Expressed Sequences Tags (EST) databases of X. tropicalis, showed that sequences with a high homology to vertebrate SOX9 correspond to a single gene (NCBI: gene ID 549607). Phylogenetic analysis using the neighbor-joining method (PHYLIP program) revealed that the X. tropicalis SOX9 gene is more closely related to the tetrapode SOX9 than to those of fishes (Fig. 1).
To examine the pattern of SOX9 gene expression in the male and female gonads of X. tropicalis and identify the cell type(s) expressing it, a X. tropicalis cRNA probe served to perform in situ hybridization on tissue sections spanning the entire process of gonad differentiation from the NF stage 55 (Nieuwkoop and Faber,1967) to the adult stage. In parallel, an immunohistological study was conducted using polyclonal anti-SOX9 antibody raised against the N-terminal (1-62 amino acids [aa]) region of human protein SOX9 which shows 90% identity with the corresponding region of X. tropicalis (Fig. 2; see the Experimental Procedures section).
Expression of the SOX9 mRNA and Protein in the Developing Male Gonad of X. tropicalis
On the basis of our previous morphological description of the testicular differentiation process throughout the larval period, the process of testicular differentiation was divided into four stages (named Test 1, Test 2, Test 3, Test 4) according to the shape of the gonad as seen in cross sections, the relationships between germ cells and somatic cells and the distribution of the laminin over the gonad (El Jamil et al.,2008). At early stages of testicular differentiation (Test 1, Test 2, Test 3), when the formation of testicular cords had not already taken place there was no detectable SOX9 mRNA and protein expression (Fig. 3A–E). However, a strong positive in situ hybridization signal was detected from Test 4 stage (Fig. 4B) when the testicular cord were clearly formed and surrounded by extracellular matrix (Fig. 4D,F,G,I). No in situ hybridization signal was detected using SOX9 sense RNA probe (Fig. 4A). Double labeling of the same sections with a polyclonal antibody directed against fibronectin clearly showed that SOX9 expression was restricted to differentiating testicular cords, which were fibronectin-negative (Fig. 4C). Consistent with these in situ hybridization results, the SOX9 protein was first detected at the same Test 4 stage and the immunostaining signal was restricted to the nuclei of somatic cells, which support the germ cells inside the testicular cords (Fig. 4E,H).
Expression of the SOX9 mRNA and Protein in the Adult Male Gonad of X. tropicalis
As observed by light microscopy, the adult male gonad of X. tropicalis is comprised of seminiferous tubules (Fig. 5A), separated by an interstitial tissue and composed of cysts of germ cells surrounded by somatic cells. Germ cells occurred in mitotically synchronous groups whose sizes depend on the stage of spermatogenesis (Fig. 5B–D). In situ hybridization showed that the SOX9 gene is expressed in the adult male gonad (Fig. 5E). Observations with Nomarski optics showed that in situ hybridization positive signals were only detected within the somatic epithelial cells surrounding the germ cysts (Fig. 5F). Consistently, a strong signal was detected with the SOX9 antibody in all developmental stages at the level of the nuclei of somatic cells, which support germ cells within the seminiferous tubules (Fig. 5H–M). To determine whether these SOX9 expressing cells have a steroidogenic activity, we carried out histochemical detection for 3β-hydroxysteroid dehydrogenase (3βHSDH), a key enzyme involved in steroid hormones biosynthesis. Our observations clearly showed that the enzymatic activity of 3β-HSDH was localized within the interstitial cells located outside the seminiferous tubules. In contrast, there was no detectable enzymatic reaction within somatic SOX9-expressing cells located inside seminiferous tubules (Fig. 5G). At the ultrastructural level we observed that the somatic cells surrounding germ cell cysts appeared in variable shapes depending on the stage of germ cell development. At the spermatogonia or spermatocyte stages, these cells, which appeared flattened and connected by desmosomes, had elongated nuclei and an abundant rough Endoplasmic Reticulum (r-ER). The cytoplasmic extensions of these somatic cells did not enter the germ cell cysts (Fig. 6A–C). At later stages of spermatogenesis (Fig. 6D), somatic cells were connected by ectoplasmic specializations, which were formed by tight junctions and parallel-arranged laminae of r-ER. (Fig. 6E). At these stages, somatic cells also exhibited oval-shaped nuclei and embedded the elongated spermatids or spermatozoa (Fig. 6F). Outside seminiferous tubules, the interstitial tissue was filled with other types of somatic cells, which exhibited characteristic features of steroidogenic cells in so far as their cytoplasm contained many lysosome-like electron dense bodies and a large number of lipid droplets (Fig. 6G).
Expression of the SOX9 mRNA and Protein in the Developing Female Gonad of X. tropicalis
We previously carried a detailed histological description of the differentiation of the female gonad in X. tropicalis throughout the larval period, and divided this process into four stages (named Ov 1, Ov 2, Ov 3, Ov 4) according to the shape and the histological architecture of the female gonad (El Jamil et al.,2008). Similarly to the male gonad, in situ hybridization and immunohistological studies showed no positive SOX9 signal at early stages of ovarian development (Fig. 7A–D). Expression of the SOX9 mRNA and protein started to be detectable after metamorphosis when the ovaries reached the Ov 4 stage (Fig. 7E–H).At this stage, the germ cells were oogonia and primary oocytes at the previtellogenic phase. Oogonia and oocytes were found in cell nests ranging from the premeiotic stage to the late pachytene stage (Coggins,1973). Germ cells within these nests developed synchronously. In contrast, at the late pachytene and early diplotene stages, oocytes were surrounded by follicular cells, separated from one another, and then developed asynchronously and entered in their first growth phase (Fig. 7E). SOX9 transcripts were only detected within the growing oocytes by in situ hybridization (Fig. 7F,H), while immunohistology showed that the SOX9 protein was present within the cytoplasm at this stage (Fig. 7G). No signal was ever observed within the germ cell nests (Fig. 7G,H).
Expression of the SOX9 Protein in the Adult Female Gonad of X. tropicalis
To study the subcellular distribution of SOX9 protein throughout oogenesis, an immunohistological study of X. tropicalis adult ovary was conducted. At this stage, germ cells were mainly represented by primary oocytes at different stages ranging from I to VI according to Dumont (1972). These oocytes were in their previtellogenic phase (stages I, II) or vitellogenic phase (stages III, IV, V, VI) growth phase. Vitellogenic oocytes are easily distinguishable from previtellogenic oocytes by the presence of accumulated yolk platelets within the cytoplasm and the pigmentation lying beneath the cortical layer of oocytes. At previtellogenic stages, a strong signal was essentially found in the oocyte cytoplasm. While at vitellogenic stages, SOX9 protein was mainly detected in the nucleus (Fig. 7I,J).
In the present report, we investigated the role of the SOX9 gene during gonadogenesis in the amphibian X. tropicalis by determining when and where the SOX9 protein is expressed. Our observations provide evidence that the SOX9 mRNA and protein are expressed in both male and female gonads after metamorphosis when the gonads are well differentiated and that SOX9 expression persists in both sexes until the adult stage. These data suggest that the SOX9 gene is not necessary to trigger the early events of gonad differentiation in X. tropicalis and thus does not play a sex-determining role in amphibians. Consistent with this idea, previous reverse transcriptase-polymerase chain reaction (RT-PCR) analyses showed that the SOX9 gene is transcribed in developing male and female gonads in R. rugosa (Takase et al.,2000), and X. laevis (Osawa et al.,2005). In addition, we show that expression of SOX9 mRNA and protein is restricted to somatic cells in the male gonad whereas in the female gonad it is restricted to germ cells. Thus, the pattern of SOX9 expression in amphibians is different from that reported in mammals and other amniote species, in which SOX9 expression is restricted to somatic cells that support germ cells and never takes place in germ cells. This expression is down-regulated in ovaries and up-regulated in testes (Morais da Silva et al.,1996; for a review, see Morrish and Sinclair,2002). In contrast to amniotes species, SOX9 expression in X. tropicalis is similar to that reported in different teleost fish species such as zebrafish (Chiang et al.,2001) and medaka (Yokoi et al.,2002; Klüver et al.,2005; Nakamoto et al.,2005; Nakamura et al.,2008) in which the expression of SOX9 is maintained in the testis as well as in the ovary. However, contrary to the situation in X. tropicalis were the same SOX9 gene is expressed in both male and female gonads, two distinct SOX9 genes are expressed in fish gonads, a male-specific and a female-specific. Indeed, the medaka SOX9a2 (alternatively named SOX9b) is expressed in the somatic cells of the adult testis (Nakamoto et al.,2005), whereas the medaka SOX9a (formerly named SOX9) is predominately expressed in oocytes of the adult ovary (Yokoi et al.,2002). Similarly, two SOX9 genes were reported in zebrafish where SOX9a is expressed in the testis and SOX9b is expressed in the ovary (Chiang et al.,2001; Rodriguez-Mari et al.,2005). The SOX9 gene sequence is highly conserved among vertebrate species supporting the possibility that it evolved from a common ancestor. The existence in teleost fishes of two SOX9 genes with different functions has been suggested to be correlated with the genome-wide duplication which occurred before the divergence of teleosts and may have led to a lignage-specific subfunctionalization (Klüver et al.,2005). Interestingly, phylogenetic analyses indicate that teleost SOX9 genes form a group from which tetrapode SOX9 genes are more distantly related. These findings support the hypothesis that the X. tropicalis SOX9 expression patterns described here may reflect a diversified role of an ancestral Sox9 gene, for which the role in the female gonad has been lost during evolution of the tetrapode species. What could be the biological significance of such an expression in X. tropicalis? In the male gonad, the expression of SOX9 is first observed after metamorphosis and restricted to the somatic cells that support germ cells. In amphibians different nomenclatures have been used to designate these supporting cells: follicle cells (Lofts,1974; Sàez et al.,1999,2004), follicular cells (Burgos et al., 1956; Chavadej et al.,2000), sustentacular cells (Kalt et al.,1975), and Sertoli cells (Penrad-Mobayed,1983; Risley,1990; Risley and Morse-Gaudio,1992). Our ultrastructural study shows that these supporting cells exhibit several of the morphological characteristics of mammalian Sertoli cells, such as a close association with the developing germ cells and the presence of ectoplasmic specializations that are formed by tight junctions and parallel-arranged laminae of r-ER (Meyer et al.,1977). In addition, our in situ hybridization and immunohistological studies indicate that these cells express the SOX9 gene, which is usually considered as a specific marker for Sertoli cells. Taken together, these data lead us to propose that these cells are equivalent to the Sertoli cells of mammals and should thus be referred to as Sertoli-like cells. The data presented here add strong support to the view that SOX9 expression in the Sertoli cell lineage has been conserved throughout vertebrate evolution not only in mammals (for a review, see Morrish and Sinclair,2002), birds (Vaillant et al.,2001; Oréal et al.,2002), reptiles (Western et al.,1999; Moreno-Mendoza et al.,2001), fishes (Chiang et al.,2001; Nakamato et al.,2005) but also in amphibians. In addition, our immunohistological study shows that the localization of SOX9 is restricted to the nuclei of the Sertoli-like cells. Nuclear localization confers to SOX9 protein a biological activity in the differentiation of the Sertoli cells and testicular histogenesis. It has been reported that SOX9 protein is initially detected within the cytoplasm of supporting cells, in both male and female gonad of mammals. At the onset of testicular differentiation SOX9 moves into the nucleus of Sertoli cells, where it is detected up to the adult stage. In contrast, the SOX9 protein remains cytoplasmic in the somatic cells of the female gonad, and its expression is down-regulated during ovarian differentiation (Morais da Silva et al.,1996; de Santa Barbara et al.,2000). It has been clearly demonstrated that inhibition of the nuclear export of SOX9 in mouse XX gonads induces a female to male sex reversal while the nuclear import of the SOX9 protein is a key event in the onset of the differentiation process of Sertoli cells and testicular cords (Gasca et al.,2002; Malki et al.,2005). Although the precise role of the SOX9 protein in the male gonad of X. tropicalis remains to be elucidated, it is noteworthy that the spatiotemporal expression of the SOX9 gene in the Sertoli-like cells coincides with the detection of laminin which surround testicular cord. It is thus possible that SOX9 contributes to the synthesis of the basement membrane, which lines up these cords. This putative role of SOX9 is consistent with previous data that clearly demonstrated a role of SOX9 as transcription factor for type II collagen (Bell et al.,1997). Such a role has been also recently suggested by Notarnicola et al. (2006) who showed that in the mouse ovary SOX9 is transiently expressed in the theca cells which line the basal follicular lamina, and therefore can play a role in follicular morphogenesis. Taken together, our findings demonstrate that expression of SOX9 occurs at later stages of testicular differentiation. This suggests that SOX9 is not a sex-determining gene. However, the nuclear localization of SOX9 in Sertoli-like cells, similarly to mammals, may confer to this gene a conserved role in the differentiation of Sertoli cells and the maturation of testicular cords throughout vertebrate evolution. In the female gonad, expression of SOX9 was also first observed after metamorphosis. Contrary to the situation in X. tropicalis male gonads, SOX9 mRNA and protein expression is restricted to the germ cells that have entered their first growth phase of early diplotene (primary oocytes). It is noteworthy that expression of SOX9 in the oocytes at the diplotene stage coincides with the development of lampbrush chromosomes that are highly active in transcription. The immunolocalisation results presented here show for the first time in vertebrates that SOX9 protein is found in the cytoplasm of previtellogenic oocytes and is subsequently imported to the nucleus in vitellogenic oocytes. Although Sox9 expression was reported to take place in differentiated oocytes of fishes (Yokoi et al.,2002; Chiang et al.,2001; Nakamoto et al.,2005), these studies were limited to the expression of SOX9 mRNA and not to the protein. Thus, we cannot conclude if this nuclear import is specific to the amphibian X. tropicalis or if it is present in all anamniote species. Nevertheless, because nuclear localization is an essential step in testis development, we can assume that the nuclear import of SOX9 in vitellogenic oocytes is also in favor for an unexpected biological role of this protein during the late phases of oogenesis, which remain to be investigated.
Fertilized eggs were obtained from pairs of males (UYERE, Nigeria) and females (ADIOPODOUME, Ivory Cost) of X. tropicalis after injection with chorionic gonadotropin (100 UI). Tadpoles were maintained in dechlorinated water at 25°C and enlightened for a photoperiod of 12 hr. Experiments were conducted in accordance with the Gyeongsang National University Guide for Care and Use of Laboratory Animals. Before all experiments, animals were anaesthetized in 0.1% MS222 (Aminobenzoic Acid Ethyl, Fluka). Developmental stages were estimated according to Nieuwkoop and Faber (NF) for X. laevis (1967).
For histological observations, dissected gonads associated or not with the mesonephros, depending on the age of the animal, were fixed in Bouin's solution, embedded in paraffin and sectioned at 7μm. The sections were stained with hematoxylin and eosin. Histological preparations were examined under a Leica microscope with Nomarski optics and photographed with a NIKON DXM-1200 camera.
Tissue Preparation for Immunofluorescence and In Situ Hybridization
Gonads were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) 1× for 1 hr at 4°C and cryoprotected in a series of 12%, 15%, and 18% sucrose in PBS. Tissues were transferred to embedding Tissue-Tek OCT compound (Bayer Diagnostics). Cryostat sections (10 μM) were mounted onto slides treated with 2% 3-aminopropyltriethoxysilane (Sigma) and stored at −20°C until use for either immunofluorescence or in situ hybridization studies.
Preparation of RNA Probes and In Situ Hybridization Experiments
The cDNA used as a template for the synthesis of SOX9 probes was isolated from the X. tropicalis xtb > library of cDNAs prepared from the central nervous system of metamorphosing tadpoles. The xtb01E12 plasmid containing a full-length cDNA (2,538 bp) inserted into the pCMVSPORT6-xtbs plasmid, a modified version of the PCMVsport6 vector (Fierro et al.,2007) was linearized with EcoRI or BamHI. The sense and antisense SOX9 riboprobes were generated by in vitro transcription with digoxigenin-labeled deoxy-UTP and the appropriate SP6 or T7 RNA polymerase according to the manufacturer's instructions (Roche). For in situ hybridization experiments, frozen sections were thawed, delipidized in chloroform for 1 min and dehydrated in PBS pH 8. After 2 hr of prehybridization in 50% formamide, 2× sodium saline citrate, 5× Denhardt's solution, 50 μg/ml yeast tRNA, 250 μg/ml salmon sperm DNA, 4 mM ethylenediaminetetraacetic acid (EDTA), and 2.5% dextran sulfate at 55°C, hybridization was carried out overnight at 55°C in a moist chamber with the riboprobe diluted in the same buffer without salmon sperm DNA and EDTA. The slides were then washed, incubated with blocking buffer and treated with the alkaline phosphatase-conjugated anti-digoxigenin antibody. After extensive washing, the color reaction was performed using the NBT (nitro-blue tetrazolium chloride) and the BCIP (5-bromo-4-chloro-3′-indolyphosphate p-toluidine salt) substrates (Roche). Preparations for in situ hybridization were examined under a Leica microscope with or without Nomarski optics and photographed with a Nikon DXM-1200 60 camera.
SOX9 protein immunolabeling was performed directly on frozen sections using polyclonal anti-SOX9 antibody raised against the N-terminal (aa 1-62) region of the human SOX9 protein (Notarnicola et al.,2006).The full-length sequence of X. tropicalis SOX9 protein (482 aa) showed 84% identity with the human SOX9 protein and the 1-62 N-terminal region 90% identity with this protein (Fig. 2). After washing with PBS, sections were incubated overnight at 4°C with rabbit anti-SOX9 antibody (Notarnicola et al.,2006) at a 1/200 dilution. Fibronectin immunostaining was performed on frozen sections previously treated for in situ hybridization experiments and incubated with rabbit anti-fibronectin (Sigma) at a 1/200 dilution. Laminin immunostaining was performed on frozen sections using rabbit anti-laminin anti-body (Sigma) at the dilution 1/300. After washing with PBS, the slides were incubated with Alexa fluor 488-goat anti-rabbit or Alexa Fluor 594 goat anti rabbit (Invitrogen) diluted at 1:1,000, DNA was stained with Hoechst 33342 (Invitrogen) to visualize the nuclei. Fluorescence labeling was observed under a Leica microscope and photographed with a Hamamatsu ORCA-ER camera.
The 3β-Hydroxysteroid Dehydrogenase (3β-HSDH) enzymatic activity was revealed by deposits of formazan in delipidized, and rehydrated sections of adult testes of X. tropicalis, after incubation with 5β-androstan-β-ol-17-one (3 β-etiocholanolone; Sigma) as substrate in the presence of nitro blue tetrazolium and NAD, according to the method described by Collenot and Collenot (1977).
Adult testes of X. tropicalis were dissected and fixed for 24 hr at 4°C in 2% glutaraldehyde in 0.1M cacodylate buffer (pH 7.2). After washing in the same buffer, the tissues were post-fixed for 1 hr at room temperature in 1% osmium tetroxide (OsO4) in 0.1 M cacodylate buffer (pH 7.2). Subsequently, tissues were dehydrated through a graded series of ethanol and embedded in Epoxy embedding resin (Penrad-Mobayed,1983). Semithin sections were stained with methylene blue. Selected blocks were ultrathin sectioned, and grids were stained with uranylacetate and lead citrate. Sections were observed and photographed with a Tecnai 12 electron microscope (FEI, Eindoven, The Netherlands). Digital acquisitions were made with a numeric camera Keen View and analyzed using the Item software (Soft Imaging System, Munster, Germany).
Multiple sequence alignments were made using CLUSTALX (Larkin et al.,2007). These alignments were treated with the PHYLIP program package version 3.6 (Felsenstein,1989) to generate a phylogenetic tree using the neighbor-joining algorithm. The robustness of each nod was estimated by bootstrapping analyses (1,000 replicates). The SOX9 protein sequences used to construct the tree are human (NCBI ID:NP_000337), mouse (NCBI ID:NP_035578), marsupial (NCBI ID:XP_001370056), chicken (NCBI ID:NP_989612), Rana rugosa_SOX9a (NCBI ID:BAA95427), alligator (NCBI ID:AAD117974), Rana rugosa_SOX9b (NCBI ID:BAA95428), X. leavis (NCBI ID:NP_001084276), X. tropicalis (NCBI ID:NP_NP_001016853), Danio rerio_SOX9a (NCBI ID:NP_571718), Danio rerio_ SOX9b (NCBI ID:NP_571719), Rainbow trout_ SOX9a (NCBI ID:NP_001117651), Rainbow trout_ SOX9a2 (NCBI ID:NP_001117742), Medaka_ SOX9a (NCBI ID:AAX62152), and Medaka_ SOX9a2 (NCBI ID:BAE02836).
We thank Dr. André Mazabraud for providing animals and Dr. Nicolas Pollet (Transgenèse et Génétique des amphibians, Université Paris Sud, Orsay) for providing us with the plasmid pCMVSPORT6-xtbs. We thank Mabel Jouve San Roman (Electron microscopy Core facility, Institut Jacques Monod, Paris) for electron microscopy, Christophe De Meiderios (IBAIC, Orsay) for animal care, and Drs. Anne-Lise Haenni and Jean-Antoine Lepesant (Institut Jacques Monod, Paris) for critical reading of the manuscript.