The Drosophila homeobox gene fushi tarazu (ftz) was initially identified as an important factor for segmentation, as down-regulation of ftz led to the development of fewer body segments (Kuroiwa et al., 1984; Wakimoto et al., 1984). The fushi tarazu factor-1 (FTZ-F1) was subsequently identified as the transcriptional regulator of ftz gene expression (Ueda et al., 1990; Lavorgna et al., 1991). FTZ-F1 and genes homologous to FTZ-F1 have since been isolated in numerous species in different phyla. Different names have been assigned to these genes, such as steroidogenic factor-1 (SF-1), adrenal-4-binding protein (Ad4BP), embryonal long terminal repeat-binding protein (ELP), α-fetoprotein transcription factor (FTF) and liver receptor hormone-1 (LRH-1). However, according to the novel nomenclature system for all nuclear receptors, the FTZ-F1 homologs constitute a separate phylogenetic clade named NR5A (Nuclear Receptors Committee, 1999). Among the members of the NR5A group there are certain traits, which are conserved and links the group members together. The vertebrate FTZ-F1 homologs (NR5A) have been divided in two groups (NR5A1 and NR5A2), mainly based on sequence homology, function, and expression patterns. NR5A1 contains genes closely connected to steroidogenesis and are identified as transcriptional regulators of steroidogenesis (Ikeda et al., 1993, 1994). SF-1 and genes homologous to SF-1 are placed in the NR5A1 group and, in mammals, are expressed in steroidogenic tissues and have been linked to sex determination.
The NR5A2 group contains genes coding for proteins linked to regulation of α-fetoprotein (Galarneau et al., 1996), an estrogen-binding protein expressed in mammalian embryos and fetuses. In adults, NR5A2 homologs are highly expressed in liver and have been shown to be involved in cholesterol metabolism and production of bile acids (Nitta et al., 1999; del Castillo-Olivares and Gil, 2000). Expression has also been observed in intestine, pancreas, and chondroid tissue during development (Rausa et al., 1999). The mammalian NR5A2 genes, to date, are not connected to steroid synthesis or sex determination. However, studies of teleost FTZ-F1 genes have shown that the mammalian classification system may not be appropriate to apply on fish.
Four zebrafish FTZ-F1 genes have been identified so far: ff1a (Liu et al., 1997), ff1b (Chai and Chan, 2000), and the previously uncharacterized ff1c (GenBank accession no. AF327373) and ff1d (GenBank accession no. AY212920). The functions of ff1a have been linked to that of NR5A1 by regulating the LHβ receptor in synergy with the estrogen receptor (ER) (Liu et al., 1997). Sequence homology links ff1a to NR5A2 and the embryonic expression pattern overlaps that of both NR5A1, 2, and 3 (von Hofsten et al., 2001), suggesting a broader function of ff1 genes in zebrafish compared with mammals. ff1b is officially designated NR5A4 by the nuclear receptors committee, but it has been suggested to be the zebrafish homolog to SF-1 due to its involvement in the differentiation of steroidogenic interrenal cells (Hsu, et al., 2003, Liu et al., 2003).
SF-1 regulates the expression of anti-Mullerian hormone (AMH) together with GATA4 and Wilms tumour-1 (WT1; Giuili et al., 1997). AMH is a dimeric glycoprotein belonging to the transforming growth factor-beta (TGFβ) superfamily of growth factors and initiates the regression of the Mullerian ducts. The Mullerian ducts and the Wolffian ducts are embryonic tissues present in both genetically male and female individuals before sex differentiation. The Mullerian ducts will in XX individuals and in the absence of AMH develop into parts of the female sex organ (oviducts and part of uterus), and the Wolffian ducts will regress. In genetical males, AMH will initiate the regression of the Mullerian ducts, which paves the way for the Wolffian ducts to continue differentiation into vas deferens and parts of the male sex organ. Although fish lack Mullerian ducts, other AMH functions may be important for gonad development. In mammals, AMH is involved, in addition to Mullerian degeneration, in regulation of steroidogenesis in the gonad. AMH inhibits the expression of aromatase in developing gonads (di Clemente et al., 1992) and negatively modulates the differentiation and function of Leydig cells (Racine et al., 1998) by down-regulating several enzymes involved in the steroidogenic pathway. Ovarian cell growth is also inhibited by AMH in vitro (Ha et al., 2000). No study has been published on teleost AMH to date, but an AMH related gene, designated eel spermatogenesis related substances 21 (eSRS21), has been studied in the Japanese eel (Miura et al., 2002). The eSRS21 gene was expressed in Sertoli cells and down-regulated 11-keto-testosterone (11-KT) induced spermatogenesis. This finding indicates that eSRS21 and genes related to AMH may have functions important both for reproduction as well as sex determination and differentiation in fish. In the present study, we have characterized the zebrafish ff1d gene expression pattern during development and in adult tissues and compared it with the expression of ff1a and ff1b. In addition, we cloned the zebrafish AMH gene to determine whether ff1d and AMH are coexpressed in testis and ovaries.
A trait shared by all FTZ-F1 homologs is the FTZ-F1 Box region, a region in the DNA binding domain responsible for targeting the FTZ-F1 response element during transcriptional activation. The zebrafish FF1a, FF1b, FF1c, and FF1d protein sequences were aligned to identify common regions of the four gene products. An FTZ-F1 box was identified in the DNA binding domain (DBD) of all four FF1 proteins (Fig. 1). An AF-2 domain (LLIEML) and an I-box (LLRLPE) were also found in the ligand binding domains, although the I-Box core differed in a single position for FF1b (ILRLPE) and the AF-2 domain of FF1c was LLTEML.
By aligning the four zebrafish proteins to representatives of all other NR5A subgroups with a Clustal W algorithm, we intended to establish the position of FF1a, b, c, and d in an NR5A phylogenetic tree (Fig. 2). The previously established representatives of NR5A1 (mELP, rSF-1, and cSf-1) and NR5A2 (mLRH1, rFTF, cFTF, rrFTZ-F1a) were grouped, respectively, in the analysis. The teleost acFF1, rtFF1, and zFF1a also aligned within this clade. The zebrafish FF1b aligned with mFTZ-F1 in NR5A4, outside of the NR5A1 clade. Whereas FF1c did not align clearly within any of the previously defined NR5A subgroups, FF1d aligned in NR5A4 with FF1b as its closest match. Interestingly, FF1d also shows homology with the NR5A1 group (53%), at an equal extent to that of FF1b (54%). The shared identities between FF1b and FF1d were at 62%, whereas the similarity was at 57% between FF1a and FF1d. Due to the significant homology between ff1b and ff1d, the ff1d probe used in subsequent in situ analyses was tested against ff1b and ff1d templates by Southern blot. The ff1d probe was specific to ff1d and did not show affinity to the ff1b template (Fig. 3).
The developmental expression pattern of ff1d was studied using whole-mount in situ hybridization. The first expression domain detected was an area in the rostral diencephalon at the Prim5 stage, equivalent with the pituitary expression of ff1a during the same stage (Fig. 4A–C). At 30 hr, an expression domain in an area corresponding to that of the interrenal cells was observed (Fig. 4H). This domain was identical to that of ff1b (Fig. 4I), which during the same developmental stage is expressed in the developing interrenal tissue (Hsu et al., 2003). However, the ff1d expression was only observed in the interrenal tissue during this stage and disappeared until the Long Pec stage. The expression in the pituitary area was persisting during Prim 16 (30 hr) stage (Fig. 4D–F), equivalent with ff1b expression in the pituitary (Fig. 4G). At Long Pec (48 hr) and at the Protruding mouth (72 hr) stages the expression domain was migrating dorsocaudally to the center of the brain, corresponding to the location of the pituitary/hypothalamus area (Fig. 4J–O). Using two color in situ hybridization the pituitary expression was found to overlap that of ff1a at 48 hr (Fig. 5). The symmetrical bilateral pituitary expression domain was merged after 72 hr. At 30 days of development, ff1d expression was detected in hypothalamus and neurons descending to the anterior pituitary target cells (Fig. 6).
By searching the MedLine database a genomic sequence corresponding to AMH was identified (GenBank accession no. BX00598). Oligos were designed against the sequence and a 786-bp cDNA fragment, enough to use as a riboprobe, was isolated. The sequence obtained was translated and aligned to the corresponding amino acid sequences of Anguilla japonica eSRS21, chick, and human AMH and a TGFβ-like region was found in the C-terminus (Fig. 7).
Using reverse transcriptase-polymerase chain reaction (RT-PCR) the temporal expression patterns of AMH, ff1a, ff1b, ff1d, and 18S RNA were examined in zebrafish embryos. Maternal transcripts deriving from all examined genes were detected in unfertilized oocytes. Zygotic ff1d expression was detected from the 1K stage until 30 days postfertilization (dpf; Fig. 8). The expression of ff1d exhibited three peaks, with the first occurring at 1K and the second at hatch. A third increase in ff1d was apparent after hatch and increased to 30 dpf. The other examined genes (ff1a, ff1b, and AMH) were all expressed from the one-somite stage until 30 dpf. ff1a showed expression from one-somite with a moderate increase throughout the studied period. The ff1b expression pattern indicated increasing levels during development, with a peak at hatch and reduced levels thereafter. AMH was first detected at 50% epiboly. AMH expression was observed from one somite, through development, and, after hatch, AMH peaked at 20 dpf.
The tissue distribution in adults was determined for brain, heart, intestine, gonads, liver, muscle, and eye from males and females. Of the examined tissues, ff1d was expressed in brain and in gonads (Fig. 9). The expression pattern of ff1d was similar to that of ff1b, which was found expressed in the same tissues. However, while ff1b showed equal expression in testis and ovary, the ff1d levels were highest in the testis. ff1a was expressed in all tissues examined. In female eye and brain, the ff1a levels were significantly lower than in males. Whereas AMH was only expressed in gonads, it showed the same expression pattern as ff1d, with the highest expression in the testis.
The RT-PCR indicated a strong ff1d expression in testis, and testis sections were subsequently examined for ff1d expression. Adult male testes expressed ff1d, mainly in interstitial Leydig cells, but also in Sertoli cells (Fig. 10A,B). AMH is a marker for Sertoli cells in other organisms and was detected in these cells in the testis sections (Fig. 10C). Double staining of ff1d and AMH showed colocalization of the signals to the Sertoli cells (Fig. 10D). In the ovary, both ff1d and AMH were detected inside the oocyte in previtellogenic follicles (Fig. 11). AMH and ff1d was coexpressed in the area inside of the vitelline envelope in stage I to stage III follicles. In stage II and stage III follicles, both ff1d and AMH were expressed within the follicle layer containing the granulosa and Theca cells. ff1d was also expressed in the inner region of the vitelline envelope. Thus, the expression pattern of ff1d and AMH was comparable in both testis and ovaries, with colocalization to steroidogenic cells in both tissues.
ff1d is a novel zebrafish FTZ-F1 gene with protein sequence characteristics indicating similar basic regulatory mechanisms as the previously characterized ff1. FF1d aligns within the NR5A4 clade, and the FF1d protein is similar to FF1b in two respects. First, the FF1d protein sequence closest match was FF1b, although the similarity was modest and the similarity to FF1a was nearly as high. However, the expression pattern of ff1d and ff1b links them together as well, both in their expression in the brain and in the interrenal tissue. ff1b in the pituitary has been detected previously by in situ hybridization and coexpression with the pituitary marker pit-1 (Chai and Chan, 2000). ff1a is expressed in the same region, but in addition to the pituitary expression, ff1a is expressed in a broad variety of tissues during development (von Hofsten et al., 2001). Brain and eye ff1a expression was higher in males than in females (Fig. 9). Sex differences has been observed previously for Arctic char ff1 (NR5A2) expression, including brain (von Hofsten et al., 2003), suggesting a conserved teleost sex-related difference in NR5A2 expression. ff1d was expressed in the same pituitary area as ff1a and ff1b during embryogenesis. However, the coexpression of ff1a and ff1d is restricted to the pituitary area (Fig. 5). ff1d was expressed in an area comparable to ff1b interrenal tissue expression during 30 hr after fertilization. Previous studies have shown that ff1b is an essential regulator of the differentiation of the steroidogenic interrenal cells, and based on that, ff1b has been designated a functional homolog of mammalian SF-1 (Hsu et al., 2003). The fish interrenal tissue is a major site of steroid synthesis in teleosts and the equivalent to the mammalian adrenal cortex. SF-1 is a transcriptional regulator of many genes expressed in the adrenal (reviewed in Hammer and Ingraham, 1999) and SF-1 knockout mice die due to lack of adrenals (Sadovsky et al., 1995). There is no direct evidence that ff1b trancriptionally regulates steroidogenic downstream genes, but ff1b morphant embryos lack 3βHSD and CYP11A expression (Hsu et al., 2003), either as a consequence of the lack of a positive regulator or due to the lack of differentiated interrenal cells.
ff1a is expressed the differentiating pronephros, suggesting that ff1a may have a role in the early urogenital differentiation process (von Hofsten et al., 2001). Neither ff1b nor ff1d can be detected in the pronephros during embryogenesis, but ff1d is highly expressed in the adult testis (Figs. 9, 10). This finding suggests that the zebrafish genes have both overlapping and separate functions during development. ff1d is expressed in testicular tissue and, in particular, the interstitial Leydig and Sertoli cells (Fig. 10), suggesting an involvement in gonadal steroid synthesis.
SF-1, in addition to its role in steroidogenesis, is an important factor in mammalian sex determination by regulating the AMH in the developing testis (Giuili et al., 1997). AMH initiates the regression of Mullerian ducts in mammals and, therefore, is important for the formation of male sex characteristics. AMH is also expressed early in differentiating Sertoli cells and can be used as a Sertoli-specific marker (Vigier et al., 1984). Later in development, AMH is also expressed in granulosa cells and where it is involved in the regulation of follicle growth (Vigier et al., 1984; Salmon et al., 2004). Both ff1d and AMH were expressed in growing follicles in the oocyte and in the follicular region of growing follicles (Fig. 11). The expression declined in mature follicles, suggesting an involvement in follicle growth. Mammalian AMH is expressed in growing follicles and disappears in matured follicles in a similar way as seen in this study (Ueno et al., 1989, Durlinger et al., 2002). In zebrafish, the AMH gene is still not characterized, but its expression is ovary- and testis-specific (Figs. 9–11). The testis expression was restricted to Sertoli cells (Fig. 10), similar to the mammalian pattern, suggesting a conserved role between zebrafish and mammalian AMH. AMH expression was first detected by RT-PCR between stages 50% epiboly and one-somite during embryogenesis but was only detectable in yolk syncytial layer by whole-mount in situ hybridization. The highest AMH expression level was detected by RT-PCR at 20 dpf, which is the approximate time when sex differentiation is initiated in zebrafish.
Teleosts lack a gene similar to SRY, which constitute the universal testis-determining factor in mammalian sex determination. Fish sex is decided by a delicate process, which involves hormonal secretion during embryogenesis. Because zebrafish lack sex chromosomes (Wallace and Wallace, 2003) and no testis-determining factor has been found, a regulatory pathway involving AMH would be an interesting candidate for zebrafish sex determination and differentiation. Zebrafish are considered to be juvenile hermaphrodites where the gonads initially develop into ovaries in all individuals. Between 20 and 30 dpf, the oocytes undergo apoptosis in presumptive males, leading to the development of testes rather that ovaries (Uchida et al., 2002). It is intriguing, therefore, that AMH expression peaked at 20 dpf in this study. Future studies are needed to establish the precise roles for AMH and ff1d during gonad differentiation, but by comparing sequences and expression patterns, we conclude that ff1d and AMH are new candidates for sex determination and differentiation in fish.
Isolation of cDNA
Oligos complementary to the GenBank sequences for ff1a (NM_131463), ff1b (NM_131794), ff1c (AF327373), and ff1d (AY212920) were constructed and used to PCR amplify cDNA from whole embryo (48 hr) after reverse transcription of total RNA preparations. Primers were as described previously for ff1a and ff1b (von Hofsten et al., 2001), and the following primer pairs were designed for ff1d (fwd: 5′-gcagcaggagcagaactcacggag-3′ and rev: 5′-agctcaggaaggcgcagcagaatc-3′). AMH was identified by searching the zebrafish genome in GenBank using the C-terminal sequence of Anguilla japonicaAMH (GenBank accession no. BAB93107). Zebrafish AMH cDNA was subsequently isolated from testicular cDNA using PCR with forward oligo: 5′-taagggacatcctgcctcagagc-3′, and reverse: 5′-tcagcggcattcgcacttggtcgc-3′ at 95°C for 30 sec, 60°C for 40 sec, and 72°C for 45 sec for 40 cycles. All PCR products were subcloned into pGEMt (Promega), the inserts were subsequently identified by sequencing using Dyenamic ET (Amersham), and reactions were resolved on an ABI Prism 377 DNA Sequencer (Perkin-Elmer). The data obtained were analyzed using EditView (version 1.0.1; Perkin-Elmer). The zebrafish AMH sequence was published in GenBank (accession no. AY677080).
Sequence Similarity Analysis
A comparative sequence analysis of NR5A proteins was performed. The tree was constructed using Tree View (version 1.6.2) after alignment of the protein sequences by the Clustal W algorithm (version 1.7). A confirmation of each clade division was established by bootstrap analysis using 1,000 repeats. GenBank accession nos.: Salvelinus alpinus FF1a (acFF1a; AAL85312); Mus musculus LRH-1 (mLRH-1; P45448); Rattus norvegicus SF-1 (rSF-1; P50569); Mus musculus ELP (mELP; P33242); Rana rugosa FTZ-F1 (rrFTZ-F1; BAA94077); Danio rerio FF1b (zFF1b; AAF43283); Danio rerio FF1a (zFF1a; AAC60274); Danio rerio FF1d (zFF1d; NM_212834); Danio rerio FF1c (zFF1c; NP_999944); Rattus norvegicus FTF (rFTF; AAC52645); Oryzias latipes FTZ-F1 (mFTZ-F1; BAA32394); Oncorhynchus mykiss FTZ-F1 (rtFTZ-F1; BAB11689); Gallus gallus SF-1 (cSF-1; BAA22839); Gallus gallus FTF (cFTF; BAA22838); and Drosophila melanogaster FTZ-F1 (dmFTZ-F1; NP_524143). The zebrafish FF1 protein alignment was constructed in Omiga using the protein sequences described above. The AMH alignment was made using the translated sequence of the partial cDNA (AY677080) and was aligned with the same protein region of Anguilla japonica AMH (ajAMH; BAB93107); Gallus gallus AMH (cAHM; CAA61536), and Homo sapiens AMH (hAMH; AAH49194).
cDNA was made using Super Script II reverse transcriptase (Invitrogen) on whole embryo or tissue total RNA preparations obtained by Tri Reagent (Sigma-Aldrich). The cDNA was subsequently used in PCR to analyze the expression of ff1a, ff1b, ff1d, and AMH using the conditions described above. The embryonic stages were determined according to Kimmel et al. (1995). In the adult tissue distribution analysis, male and female tissue was prepared separately. All RT reactions were made in triplicate, and the most representative PCR reaction was used to illustrate the results for each category. Amplified products from each primer combination were subcloned and sequenced to confirm the identities. Sequenced plasmids were used as positive control templates. Negative controls contained no template.
Southern and In Situ Hybridizations
cDNA encoding ff1a, ff1b, ff1d, and AMH were isolated and sequenced as described above. Digoxigenin- (DIG) and biotin-labeled anti-sense RNA probes were generated using the DIG RNA Labeling Kit (Roche Diagnostics Scandinavia AB, Bromma, Sweden). Pigmentation was blocked according to Karlsson et al. (2001). Whole-mount in situ hybridization and visualization by nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) or Fast Red were performed according to Jowett (1999). The 10- to 12-μm cryosections were made according to Westerfield (2000) and subsequently used to detect ff1d and AMH expression in situ in testis, ovary, and in 30 dpf zebrafish, using standard procedures and the riboprobes described above. Staging of zebrafish oocytes was made according to Selman et al. (1993). The ff1d probe specificity was examined by using the complete coding sequences from ff1b and ff1d as templates for Southern hybridization. Before loading samples on an agarose gel for electrophoretic separation, 0.5 μg of sequenced ff1b and ff1d pGEM-T (Promega) clones was digested with 10 U of NotI and NcoI to cut the inserts out. The DNA was transferred to Hybond-Nfp membranes (Amersham Pharmacia Biotech) by capillary transfer in 20 × standard saline citrate and was crosslinked before Southern hybridization according to Sambrook et al. (1989).
The Swedish Research Council, the Magnus Bergwall Foundation, and the Kempe Memorial Foundation supported this work.