Introduction

Sex determination and differentiation are two important events in the development of gonads in vertebrates (Hughes 2001). Sex determination is a complex process that may be affected by direct environmental induction or genetic regulation (Devlin and Nagahama 2002; Raghuveer et al. 2011). For example, offspring of some fishes and reptiles can become male or female in response to an environmental factor(s) such as temperature (e.g., Janzen 1994), whereas the sex of many birds and mammals is determined by genotype at conception (Bull 1983; Conover and Heins 1987). In vertebrates, sex determination is characterised by a lack of conservation (Ijiri et al. 2008; Raghuveer et al. 2011). Sex differentiation of gonads after sex determination is a developmental process of most species and is considered as the differentiation from an undifferentiated gonadal primordium towards a testis or an ovary (Baron et al. 2005). At least some of the genes involved in the process of gonadal sex differentiation have now been identified, e.g., Sry, Sf-1, Wt1, Sox9, Dmrt1, Amh, Gata4, Dax1, Wnt4, Cyp19a1a and Foxl2 (Devlin and Nagahama 2002; Koopman 2001; Matsuda et al. 2002; Wilhelm et al. 2007). However, various genes that regulate sex differentiation are relatively conserved in most vertebrates (Angelopoulou et al. 2012; Baron et al. 2005; Devlin and Nagahama 2002; Raymond et al. 1998; Smith et al. 2013). For example, some mammalian sex differentiation-related genes have been identified in teleosts, including the Dmy/Dmrt1, Sox9 and Sry genes that promote testicular differentiation, while Cyp19a1a and Foxl2 are implicated in ovarian differentiation in most teleosts (Devlin and Nagahama 2002; Sandra and Norma 2010). In teleosts, the proportion of androgens and oestrogens is of great importance for gonadal differentiation, and this hormonal balance is maintained by aromatase which is a key enzyme catalysing the conversion from androgens to oestrogens (Guiguen et al. 2010). The expression of aromatase is regulated by several factors such as Foxl2 (forkhead box L2) and Sf-1 (steroidogenic factor-1) (Pannetier et al. 2006; Wang et al. 2007). While extensive studies have been carried out on aromatases during sexual differentiation (Guiguen et al. 1999, 2010; Zhang et al. 2014), there has been limited research conducted on the role of Foxl2 during sexual differentiation in teleosts (Wang et al. 2004; Sridevi and Senthilkumaran 2011). In this context, it is of interest to clone the Foxl2 gene to quantify the expression pattern of Foxl2 during gonadal differentiation in a teleost. This analysis might provide further valuable evidence for the significance of Foxl2 in the regulation of aromatases.

Forkhead transcription factors play critical roles in the regulation of cellular differentiation and proliferation (Cunningham et al. 2003; Vaquerizas et al. 2009). They are also involved in several other biological processes, including tissue development, establishment of the body axis and metabolic processes (Carlsson and Mahlapuu 2002). Forkhead box L2 (Foxl2) is a member of the forkhead family of transcription factors involved in ovarian development, granulosa cell differentiation and the maintenance of ovarian function (Cocquet et al. 2003; Nakamoto et al. 2006). In humans, Foxl2 gene mutations cause blepharophimosis ptosis epicanthus inversus syndrome (BPES), characterised by eyelid malformations and premature ovarian failure (POF) (Crisponi et al. 2001). The same phenomenon is observed in the disruption of Foxl2 function of mice (Uda et al. 2004). The genetic programme for somatic testis determination is activated in an XX gonads mouse lacking Foxl2 from meiotic prophase oocytes, implying the pivotal function of Foxl2 to repress the male gene pathway at several stages of female gonadal differentiation (Ottolenghi et al. 2005). Moreover, genes involved in testis differentiation such as Sox9 are sharply upregulated after birth in Foxl2 null mice (Ottolenghi et al. 2005). As far as is presently known, Foxl2 is the first gene in the developmental pathway showing differential expression between genders for gonadal differentiation in vertebrates, including teleosts (Cocquet et al. 2003; Ijiri et al. 2008; Leet et al. 2011; Loffler et al. 2003). In goats, rainbow trout and tilapia, the expression of Foxl2 is specifically initiated before morphological sex differentiation in female gonads and can persist until adulthood (Baron et al. 2004; Wang et al. 2004; Pailhoux et al. 2001). These results suggest that Foxl2 is evolutionarily conserved among vertebrates and may be involved in the early stages of female sex differentiation. On the other hand, Foxl2 is also involved in the transcriptional regulation of aromatase. Goats with polled intersex syndrome (PIS), whose Foxl2 function has been disrupted, display a reduction in the expression of aromatase (Pailhoux et al. 2001, 2002). Studies of chicken and rainbow trout demonstrate that the spatiotemporal expression profile of Foxl2 is strongly correlated with the expression of aromatase during sex differentiation and later follicular development (Baron et al. 2004; Govoroun et al. 2004). Foxl2 is also involved in the transcriptional regulation of aromatase in teleosts such as Japanese flounder (Yamaguchi et al. 2007) and medaka (Nakamoto et al. 2006). These results suggest that Foxl2 is involved in regulation of oestrogen synthesis via transcriptional regulation of aromatase during ovarian development.

Snakeheads (Teleostei: Channidae) are a group of freshwater fishes native to Asia and Africa (Ng and Lim 1990). The northern snakehead (Channa argus), which is mainly distributed in the Yangtze River and the Amur River basin, is one of the most commercially important fish species in China. Total production of the snakehead reached 480,594 tonnes in 2012, ranking it ninth in all freshwater fish species production in China (Fisheries Bureau of the Agriculture Ministry of China 2013). Aquaculture practices and studies show that males of C. argus grow faster than females and the average weight of males is 30 % larger than females at the same age (Jiang et al. 2013). It may therefore be possible to improve aquaculture production of C. argus through genetic manipulation of the sex ratio by obtaining all-male snakehead broods. However, there are few studies of sex differentiation of C. argus, except for the screening of sex-linked molecular markers (Jiang et al. 2013). In order to explore whether Foxl2 plays a role in the sexual development of C. argus, we investigated early gonadal development and sex differentiation by histological analysis and also identified the full length of Foxl2 cDNA. We analysed its expression patterns by qPCR during gonadal sex differentiation and the localisation of the Foxl2 protein by immunohistochemistry in mature gonads of C. argus. Our data provide valuable information for further studies of the sex differentiation mechanism of C. argus and will contribute to ongoing work focusing on the development of this species as a major aquaculture species in China.

Materials and methods

Animals and sampling

Ten C. argus specimens with an average weight 500 g were obtained from Liangzi Lake, Hubei Province. The fish were transported to Huazhong Agricultural University (HZAU) and reared for 1 week in a circulating water system at the College of Fisheries. Six individuals (three males and three females) were anesthetised with 1000 mg/L MS-222 before dissection. Various tissues (spleen, kidney, liver, gonad, gill, brain, pituitary, intestines, heart and muscle) were collected for gene cloning and tissue distribution analysis of gene expression.

To study the expression of the Foxl2 gene in gonads during early development, fertilised eggs of C. argus were collected from Wusi Lake in Ezhou City, Hubei Province. The eggs were incubated at 25 ± 1 °C in indoor tanks with an air pump for sufficient dissolved oxygen. Fry hatched on the third day after fertilisation. After hatching, larvae were transferred to rectangular plastic boxes for rearing and were fed with yolk and zooplankton after mouth-opening formation. The zooplankton was collected daily from South Lake next to the HZAU campus. During the period of artificial cultivation, the fish were kept at a water temperature of 25 ± 1 °C and with a 14:10-h light–dark cycle. Samples were collected at 1, 4, 7, 11, 17, 23, 30, 34, 41 and 48 days posthatching (dph). Six individuals were sampled each time at 1–23 dph and three individuals at 30–48 dph. Sampling was repeated five times on each occasion. For smaller fry at 1–30 dph stages, the whole body was collected for subsequent experiments. For bigger larvae at 34–48 dph stages, the fish head and posterior segment (from cloaca to tail end) were removed after anesthetising with 100 mg/L MS-222, and the abdomen segment was collected for subsequent experiments.

All adult tissues and larval samples were immediately immersed in liquid nitrogen and stored at −80 °C until RNA extraction. This experiment was conducted in accordance with the guidelines of the Institutional Animal Care and Use Committees (IACUC) of HZAU, Wuhan, P. R. China.

Cloning and sequencing of Foxl2 cDNA

Total RNA was isolated from ovary tissue of adult C. argus individuals using TRIzol Reagent (Invitrogen, USA) following the manufacturer’s instructions. The cDNA synthesis was carried out using a Molony murine leukaemia virus (M-MLV) Reverse Transcriptase kit (Promega, USA) according to the manufacturer’s protocol. To obtain partial sequence of C. argus Foxl2 cDNA, a pair of primers (Foxl2-F, Foxl2-R) was designed based on the conserved domains of the known Foxl2 sequences (Nile tilapia Oreochromis niloticus, NM_001279778; honeycomb grouper Epinephelus merra, EU555180; and three-spot wrasse Halichoeres trimaculatus, AB547448). To obtain the full-length cDNA sequence of Foxl2, rapid amplification of cDNA end (RACE) was performed by the standard method (Sambrook et al. 2001) using the gene-specific primers and adaptor primers (Table 1). The primers were designed using the software Primer Premier 5.0.

Table 1 Primers designed for cloning and expression analysis of Foxl2
Full size table

PCR amplifications were performed in a total volume of 10 μL, including 1.0 μL 10 × PCR buffer, 1.5 mM MgCl2, 0.4 U Taq DNA polymerase (Fermentas, Canada), 0.2 mM dNTP, 0.4 μM of each primer and 1.0 μL cDNA template. The PCR conditions were as follows: 3 min at 95 °C for predenaturation; 35 cycles of 30 s at 94 °C for denaturation, 30 s at 62 °C for annealing and 45 s at 72 °C for extension; 10 min at 72 °C for final extension.

All amplified products were checked by electrophoresis on a 1.5 % agarose gel. The target DNA fragments were purified using an AxyPrep™ gel extraction kit (Axygen, USA) and ligated into a pMD18-T vector (TakaRa, Japan). Following transfection into Escherichia coli DH5α competent cells, recombinants were identified by blue and white spot selection. Putative clones were further screened by PCR amplification and were sequenced by the Sangon Biotech Company (Shanghai, China).

Sequence analysis

Open reading frame (ORF) and protein predictions were performed using the ORF finder software (http://www.ncbi.nlm.nih.gov/gorf/gorf.html). The deduced protein sequence was analysed with the BLAST programme on the National Center for Biotechnology Information (NCBI) website (http://blast.ncbi.nlm.nih.gov/). Clustal X 1.83 was used to perform multiple alignment of amino acid sequence. A phylogenetic tree was constructed by the neighbour-joining (NJ) method using MEGA 5.03 software.

Real-time quantitative PCR

Gene expression was determined by real-time quantitative PCR (qPCR) using a Rotor-Gene 6500 Thermocycler (Corbett Research, Australia). The gene-specific primers were designed based on the full-length Foxl2 cDNA sequence (Table 1). The primer sequences of β-actin were obtained from Jia and Guo (2008). The reaction mixture of the qPCR consisted of 10 µL 2× GoTaq® qPCR Master Mix (Promega, USA), 0.2 µL 100× CXR reference dye, 0.4 µL of each gene-specific primer (10 µM), 2.0 µL cDNA template and 7 µL nuclease-free water in a total volume of 20 µL. The qPCR cycling parameters were 95 °C hold for 2 min, then 40 cycles at 95 °C for 15 s, 60 °C for 30 s, and 72 °C for 20 s. Melt curve analysis was carried out over a range from 55 to 99 °C at the end of each PCR run. All qPCRs were performed in triplicate biological replicates using standard reagents. The standard curve quantification method was adopted to analyse the data. Standard curves for each gene were obtained through a tenfold dilution series of plasmids. The β-actin gene was used as an internal control. The relative expression levels of the target genes were expressed as ratios of the copy numbers of target gene to copy numbers of β-actin gene.

Expression patterns of Foxl2 mRNA

Total RNA was extracted from various adult tissues and fry using TRIzol Reagent as described above. To avoid genomic DNA interference on qPCR, the total RNA was treated with RNase-free DNase I (TaKaRa, Japan). Approximately 1 µg of DNase I-treated total RNA was used for the synthesis of first-strand cDNA. Then, the cDNA was stored at −20 °C for qPCR and expression analysis as described above.

Histology of early gonad development

Five to ten fry were collected at 6, 15, 21, 27, 30, 34 and 48 dph. The samples were immersed in Bouin’s solution at room temperature for at least 48 h and stored in 70 % ethanol until histological processing. Tissues were dehydrated in a series of alcohol, clarified in benzene and embedded in paraffin. Cross sections were cut into 6-μm slices and were stained with haematoxylin–eosin. A thorough observation of gonads was performed by cutting the whole body or abdomen segment into serial sections (Gao et al. 2009) to observe the differentiation of the gonads. The sections were observed and photographed using a Nikon 80i microscope (Japan).

Localisation of Foxl2 protein in the mature gonads by immunohistochemistry

Live, wild adult individuals of C. argus were collected from markets in Wuhan. The fish were kept in a circulating water system at the College of Fisheries for 1 week. Two individuals (one male and one female) were anesthetised with 1000 mg/L MS-222 before dissection. A small segment of gonad was collected from both fishes and was fixed in Bouin’s solution for immunohistochemical analysis. An anti-Foxl2 polyclonal antibody (PA1-802, Thermo, USA) was used to determine the cellular localisation of the Foxl2 protein in the gonads. Histological sections were prepared as described above. After paraffin removal and dehydration, the sections were washed with citric acid buffer (0.1 M citric acid and sodium citrate, pH 6.0), incubated in 3 % (v/v) H2O2 and 10 % (v/v) normal goat serum to block nonspecific binding and then were incubated overnight at 4 °C with the primary antibody (diluted 1:50), which was a synthetic peptide corresponding to residues M(1)MASYPEPEDTAGT(14) of mouse Foxl2, derived from a rabbit host. After incubation with the secondary antibody which was labelled by peroxidase (anti-mouse/rabbit, DAKO K5007), the sections were exposed to 3,3′-diaminobenzidine (DAB) and stained with haematoxylin to visualise the nuclei in the gonadal tissues. As a negative control, the sections were treated in the same way but with Tris-buffered saline instead of the primary antibody. Finally, the sections were observed and photographed using a Motic BA310 microscope (China).

Statistical analysis

All data from qPCRs were expressed as the mean ± SE. Statistical differences were tested by one-way ANOVA using STATISTICA 6.0. P < 0.05 was considered to be statistically significant.

Results

Cloning and phylogenetic analysis of Foxl2 gene

From RACE PCR, a 1966-bp full-length Foxl2 cDNA (GenBank accession no. KF746072) was obtained, including a 213-bp 5′-untranslated region (5′ UTR), an 832-bp 3′ UTR and a 921-bp open reading frame (ORF), encoding a putative 306 amino acids. The putative Foxl2 amino acid sequence contained the characteristic forkhead (FH) domain ranging from residue 47 to 157 (Fig. 1). The C. argus Foxl2 amino acid sequence showed a high level of homology to those of other vertebrates. As expected, it was very similar to Foxl2 orthologs in teleost fishes such as Epinephelus merra (97 %), Oreochromis niloticus (95 %), Oryzias latipes (95 %), Oncorhynchus mykiss (92 %) and Danio rerio (81 %). It also showed 79 % similarity to Mus musculus and 61 % similarity to Homo sapiens. The forkhead domain of Foxl2 was highly conserved among these vertebrate species, whereas the 14-polyalanine tracts (A), glycine (G) and proline (P) repeats were only present in mammalian orthologs (Fig. 1). Phylogenetic (NJ) analysis demonstrated that C. argus Foxl2 clustered with other teleosts, and it was closest to the Foxl2 homolog of the honeycomb grouper (E. merra) and the Nile tilapia (O. niloticus). Teleostean Foxl2 was separated from elasmobranch, amphibian, avian and mammalian Foxl2 (Fig. 2).

Fig. 1
figure 1

Amino acid sequence comparison of Channa argus Foxl2 with other known orthologs. The glycine-rich repeats (G), proline repeats (P) and polyalanine tracts (A) are boxed. The underlined part indicates the forkhead (FH) domain. GenBank accession numbers of the Foxl2 amino acid sequences used are as follows: northern snakehead Channa argus (KF746072); honeycomb grouper Epinephelus merra (ACD62374); Nile tilapia Oreochromis niloticus (NP_001266707); medaka Oryzias latipes (NP_0010988358); rainbow trout Oncorhynchus mykiss (NP_001117957); zebrafish Danio rerio (NP_001038717); chicken Gallus gallus (AEE80502); pig Sus scrofa (NP_001231594); house mouse Mus musculus (NP_036150); and human Homo sapiens (AAY21823)

Full size image
Fig. 2
figure 2

Phylogenetic relationships of Channa argus Foxl2 with other species based on deduced amino acid sequences. The numbers in each branch represent the bootstrap values obtained using the neighbour-joining (NJ) method. GenBank accession numbers of the sequences used are as follows: northern snakehead Channa argus (KF746072); honeycomb grouper Epinephelus merra (ACD62374); Nile tilapia Oreochromis niloticus (NP_001266707); medaka Oryzias latipes (NP_0010988358); rainbow trout Oncorhynchus mykiss (NP_001117957); zebrafish Danio rerio (NP_001038717); half-smooth tongue sole Cynoglossus semilaevis (ACY05959); oriental weather fish Misgurnus anguillicaudatus (BAJ19137); chicken Gallus gallus (AEE80502); pig Sus scrofa (NP_001231594); house mouse Mus musculus (NP_036150); human Homo sapiens (AAY21823); small-spotted catshark Scyliorhinus canicula (ABP63571); and African clawed frog Xenopus laevis (BAH22852)

Full size image

Tissue distribution of Foxl2 mRNA expression levels in adults

qPCR standard curves showed a significant linear relationship between the values of threshold cycle (CT) and the logarithm of gene copy number in both Foxl2 gene (R 2 = 0.9990, P < 0.0001, Fig. 3a) and β-actin gene (R 2 = 0.9985, P < 0.0001, Fig. 3b), and the amplification efficiencies were both above 93 %. Therefore, the two standard curve equations can be used to reliably calculate gene copy number of the two genes.

Fig. 3
figure 3

Standard curves of Foxl2 gene (a) and β-actin gene (b) showing a linear relationship between the values of threshold cycle (CT) and the logarithm of gene copy number. conc concentration of gene copy number (copies/µL)

Full size image

Expression of Foxl2 mRNA in females had the highest tissue-specific level in ovary, followed by a high level in pituitary and gill, and a very low level in other tissues (Fig. 4a). The expression level of Foxl2 mRNA in males was the highest in gill, followed by a high level in pituitary, a moderate level in brain, whereas the expression levels were very low in testis and other tissues (Fig. 4b).

Fig. 4
figure 4

Tissue distributions of relative Foxl2 mRNA expression levels in female (a) and male (b) of Channa argus by qPCR. The relative expression level was measured as a ratio of copy numbers of Foxl2 gene to copy numbers of β-actin gene. Columns represent the means of three replicates for each treatment. Error bars represent standard error of the means. Different letters above the bars indicate significant difference at P < 0.05 level

Full size image

Expression of Foxl2 gene during the early development stages

The mRNA expression levels of Foxl2 gene were detected during the early developmental stages from 1- to 48-day posthatching (dph) larvae. The expression of Foxl2 mRNA increased significantly from a moderate level at 1 dph to the highest level at 4 dph and then decreased significantly to a moderate level (as judged by relative expression) at which it was maintained until 11 dph. Afterwards, the expression level decreased significantly to a low level at 17 dph where it was maintained until 34 dph (but with a minor increase at 23 dph), and further decreased significantly to a very low level at 41 until 48 dph (Fig. 5).

Fig. 5
figure 5

Expression of Foxl2 gene during early development stage of Channa argus. The relative expression level was expressed as a ratio of copy numbers of Foxl2 gene to copy numbers of β-actin gene. 1–48 are days posthatching (dph) at early development stages. Columns and bars represent the means and standard errors, respectively. Different letters above the bars indicate significant difference (P < 0.05)

Full size image

Histology of early gonad development

Histological sections were employed to trace early gonad development and sex differentiation of C. argus from 6 to 48 dph. After hatching, gonadal primordial germ cells (GCs) were present under the mesonephric duct at 6 dph (Fig. 6a). The germ cells were morphologically distinguished from somatic cells by their relatively large diameter and their histological features. Primitive gonads appeared under the dorsal coelomic epithelium. At 15 dph, the gonads projected into the abdominal cavity and the germ cells had increased in number by active mitosis (Fig. 6b). A pair of pear-shaped gonads was present under the abdominal cavity and was linked to a cord-like tissue from the dorsal coelomic epithelium at 21 dph. The gonads did not exhibit any morphological characteristics indicative of a differentiating ovary or testis until 21 dph (Fig. 6c). Gonad size and number of germ cells increased dramatically between 21 and 27 dph. An apparent change in larval gonad histology occurred at 27 dph. The presumptive initial ovarian cavity formation was indicated by the presence of two elongated aggregations of germ cells in the gonads at 27 dph (Fig. 6d). In contrast to ovarian development, the presumptive testis had germinated to be two aggregations of stromal cells at 27 dph. The initial testes retained the original pear-like shape of undifferentiated gonads and were much smaller than the ovaries at the same developmental stage of 27 dph (Fig. 6e). At 30 dph, the ovarian cavity was completely formed and oogonia were undergoing active mitosis to become oocytes (Fig. 6f). At 34 dph, ovarian gonads were observed to contain numerous oocytes at chromatin-nucleolus phase, and the ovarian cavity was clear in the central part of the ovary (Fig. 6g). Some slit-like spaces in the central stromal tissue of testes formed the efferent duct anlages at 34 dph, and blood vessels were observed in the dorsal region of testes (Fig. 6h). At 48 dph, numerous perinucleolus oocytes were found in the ovarian gonad (Fig. 6i). Some spermatogonia were undergoing mitotic proliferation at 48 dph, suggesting the onset of meiosis in the development of testes. The two gonadal tissues of testes that were attached at both sides of the mesentery were observed to fuse together (Fig. 6j).

Fig. 6
figure 6

Histological sections of larval and juvenile Channa argus gonad showing the early gonad development and sex differentiation. a Primordial gonads at 6 dph. A pair of primordial gonads is indicated by arrows. b Primordial gonads at 15 dph. Germ cells are undergoing early mitosis. c Undifferentiated gonads at 21 dph, showing the multiplication of germ cells in number. d Presumptive initial ovary at 27 dph, showing somatic elongations. Two somatic elongations forming the initial ovarian cavity are indicated by the thick arrows. e Presumptive testis at 27 dph. The aggregations of stromal cells are indicated by the thick arrows. f Ovary at 30 dph, showing the ovary cavity (OC) and oogonium. g Ovary at 34 dph, showing oocytes at chromatin-nucleolus stage (CNO). h Testis at 34 dph, showing the efferent duct anlage and blood vessel. i Ovary at 48 dph, showing some oocytes at perinucleolus stage (PO). j Testis at 48 dph, showing evident efferent duct anlage and spermatogonia undergoing mitosis. BV blood vessel, CNO chromatin-nucleolus oocyte, EDA efferent duct anlage, G gut, GC germ cells, MD mesonephric duct, OC ovarian cavity, og oogonium, PO perinucleolus oocyte, SG spermatogonium

Full size image

Localisation of the Foxl2 protein in the mature gonads

In the ovary, Foxl2 immunoreactivity was detected in the granulosa cells surrounding the oocytes, but not in the oocytes (Fig. 7a, b). In contrast, no specific signals were detectable in the testis (Fig. 7c, d). No positive signals were observed in the negative control of the ovary (Fig. 7e), nor in the testis (Fig. 7f).

Fig. 7
figure 7

Immunohistochemical analysis of Foxl2 in mature gonads of Channa argus. a–d Immunohistochemical analysis of Foxl2 in the ovaries (a, b) and testes (c, d). b, d Enlarged areas of a and c, respectively. The positive antigen was dyed brown with 3,3′-diaminobenzidine (DAB) (arrows). e, f Negative controls

Full size image

Discussion

Two Foxl2 paralogs named Foxl2a (Foxl2) and Foxl2b (Foxl3) have been reported in some teleost species (Baron et al. 2004; Crespo et al. 2013). They are identified by blast searches against the available genomic database in fish, which is in agreement with the fish genome duplication event (Jiang et al. 2011). In rainbow trout, Foxl2a and Foxl2b genes were specifically expressed in the ovary, but they displayed different temporal patterns of expression (Baron et al. 2004). In European sea bass, however, the expressions of Foxl2 in ovary and Foxl3 in testis showed a strong sexual dimorphism, and they varied significantly during the reproductive cycle (Crespo et al. 2013). In most mammalian species and teleosts, only one form of Foxl2 gene with a conserved DNA-binding domain has been identified (Cocquet et al. 2003). In the present study, only one type of Foxl2 gene cDNA was cloned from the C. argus ovary using the RACE strategy. Alignment of the putative Foxl2 amino acid sequences indicated that the C-terminal region and the forkhead domain of the Foxl2 were highly conserved, but the N-terminal region was divergent among teleost fishes, birds, amphibians and mammals. The forkhead and C-terminal region might be conserved in their functions through evolution, whilst the N-terminal region might have evolved under weaker constraints (Cocquet et al. 2002). The forkhead domain of Foxl2 contributed to the nuclear localisation of this protein by nuclear localisation signal (NLS) at the C-terminal (Berry et al. 2002; Romanelli et al. 2003). Homopolymers of amino acids, such as polyalanine tracts (A), glycine-rich repeats (G) and proline repeats (P), were present in the mammalian orthologs but not in those of nonmammalian vertebrates including C. argus. The elongation of Foxl2 in mammals during evolution may increase protein sizes and potentially promote acquisition of new functions (Mortlock et al. 2000). The NJ phylogenetic tree reveals that the C. argus Foxl2 has greatest homology with its teleostean counterparts, the honeycomb grouper and the Nile tilapia. Molecular phylogenetic analysis agrees with the traditional taxonomy because these three fish are all members of the order Perciformes.

Tissue distribution analysis revealed that C. argus Foxl2 in females was predominantly expressed in ovary, pituitary and gill, with a relatively lower level of expression in brain. In contrast, the expression levels of Foxl2 were high in gill, pituitary and brain of male C. argus, but with low expression level in testis, which revealed an obvious sexually dimorphic pattern of expression in the gonads. The gonadotropin-releasing hormone receptor (GnRHR) is a composite regulatory element that can be activated by Smads, AP-1 and Foxl2 in mammals (Ellsworth et al. 2003). A detectable level of Foxl2 transcript was found in C. argus brain of both sexes, implying that Foxl2 might be involved in the transcriptional regulation of GnRHR in this species. In female C. argus, the transcript of Foxl2 was much higher in gonad and pituitary than that in the brain, implying that Foxl2 probably executed its functions via the transcriptional regulation of the GnRH–GtH–sex steroids pathway. Similarly, it has been demonstrated that Foxl2 may be involved in the regulation of the hypothalamus–pituitary–gonadal axis in Nile tilapia (Wang et al. 2004), honeycomb grouper (Alam et al. 2008) and protogynous wrasse (Kobayashi et al. 2010). In mammals, birds and teleosts, Foxl2 is reported to be highly expressed in ovary but barely detectable in testis (Cocquet et al. 2002; Govoroun et al. 2004; Wang et al. 2004; Loffler et al. 2003; Nakamoto et al. 2006). Similarly, Foxl2 was highly expressed in the ovary and poorly expressed in the testis of C. argus. This expression pattern is consistent with the conserved functions of Foxl2 across these different species. However, the expression of Foxl2 in protogynous wrasse was abundant in the ovary and testis (Kobayashi et al. 2010). This situation is most likely attributed to the process of natural sex change in protogynous wrasse. It is interesting that Foxl2 was also expressed in the gill of C. argus, an organ unique to aquatic fish, which may indicate a possible new function of Foxl2 in vertebrate evolution.

Careful histological observations of the gonadal morphogenetic process are of primary importance for a precise understanding of the mechanism of gonadal sex differentiation (Nakamura et al. 1998). Histological observations of gonadal differentiation in fish are often described as either anatomical or cytological. For differentiated gonochorists in teleost fish, the formation of the ovarian cavity in females and the efferent ducts in males is generally accepted as the criterion of anatomical sexual differentiation (Nakamura et al. 1998; Strüssmann and Nakamura 2002). Cytological differentiation of the testis involves primordial germ cells (PGCs) undergoing mitotic division to become spermatogonia and then primary spermatocytes. In the ovaries, PGCs differentiate into oogonia which subsequently become oocytes (Sacobie and Benfey 2005). Generally, the first sign of fish gonadal development is the appearance of a primordial germ cell (PGC) (Gao et al. 2009; Meijide et al. 2005; Sacobie and Benfey 2005). In this study, the gonadal development of C. argus had already begun before the first sampling date of 6 dph. By this time, the size of the gonads and the number of PGCs had increased dramatically, but the gonads remained undifferentiated between 6 and 21 dph. Anatomical gonadal differentiation of C. argus occurred at approximately 27 dph when the initial ovarian cavity began to form, but testicular differentiation was observed until 34 dph with the formation of the efferent duct anlages. Therefore, ovarian differentiation precedes testicular differentiation in C. argus, which is consistent with many other teleosts (Devlin and Nagahama 2002; Meijide et al. 2005). Cytological gonadal differentiation in C. argus was observed between 34 and 48 dph. Similar to cichlid fish (Meijide et al. 2005), C. argus ovaries were easily identified by their well-developed perinucleolar oocytes, whereas testes did not reach a clear state of distinction at the cytological level. At 48 dph, spermatogonia undergoing mitotic proliferation could be observed, but spermatocytes were not formed, meaning that cytological testicular differentiation had not occurred by the end of this study. These results indicate that anatomical differentiation of fish gonads preceded cytological differentiation and confirms the pattern described for other teleosts (Sacobie and Benfey 2005; Sandra and Norma 2010). A longer sampling duration is required to reveal the cytological differentiation of testis in C. argus.

During the early developmental stages, Foxl2 transcription was at a moderate level at hatching day in C. argus, suggesting that the expression of Foxl2 was initiated around this time. The expression of Foxl2 increased significantly to the highest level at 4 dph and was maintained at a moderate level from 7 to 11 dph before further decreasing to a low level at 17 dph. These results indicate that the initiation of Foxl2 expression preceded the first anatomical and morphological differentiation of female and male gonads. Similarly, Foxl2 expression was strongest at 5 dph, the period preceding premeiotic proliferation in willow minnow (Ashida et al. 2013). In medaka, however, although Foxl2 was expressed in somatic cells surrounding germ cells in XX specimens from hatching day, the initiation of expression followed the first morphological difference between male and female gonads and before folliculogenesis (Nakamoto et al. 2006). In mouse, chicken and turtle, Foxl2 expression starts in female gonads and is upregulated shortly after the sex determination switch point (Loffler et al. 2003). Therefore, the start of expression of Foxl2 before folliculogenesis is well conserved among vertebrates (Nakamoto et al. 2006). Although the role of Foxl2 at this stage is unknown, the strong short-term upregulation of Foxl2 expression after hatching in C. argus and other fishes implies that Foxl2 might have a role in early gonadal differentiation and development. Changes in expression may promote or repress some cellular bioprocesses and then result in morphological change in the gonads (Cao et al. 2012).

Foxl2 is involved in the differentiation of granulosa cells and the maintenance of ovarian function in various vertebrates (Cocquet et al. 2003; Nakamoto et al. 2006). Previous research has demonstrated that the expression of Foxl2 in the ovary is restricted to the granulosa (follicular) cells of the oocytes in Nile tilapia (Wang et al. 2004), medaka (Nakamoto et al. 2006), catfish (Sridevi and Senthilkumaran. 2011) and rice-field eel (Hu et al. 2014). However, no signals have been detected in the oocytes of these fishes, as is also the situation in mammals (Cocquet et al. 2002). In this study, an immunohistochemical analysis showed that the Foxl2 protein was observed in the granulosa cells around the oocytes but not in the mature oocytes of C. argus. These results indicate that expression of Foxl2 is well conserved among vertebrates and that Foxl2 may have a basic function in the differentiation of granulosa cells and the maintenance of oocytes of C. argus.

In conclusion, Foxl2 was isolated from C. argus; it was expressed predominantly in the brain, pituitary, gonads and gill, indicating that the brain–pituitary–gonad axis is the main target tissue of C. argus. Foxl2 expressions were strongly upregulated from 1 to 11 dph, and the initiation preceded the first anatomical ovarian differentiation (27 dph), suggesting that Foxl2 may play a role in early gonadal differentiation and development in C. argus. The Foxl2 protein was detected in the granulosa cells surrounding the oocytes but not in the mature oocytes, implying that Foxl2 may have a basic function in granulosa cell differentiation and the maintenance of oocytes of C. argus.