Although sexual reproduction is the primary means of maintaining variation for evolutionary survival, the differentiation of dimorphic gonads is one of the most rapidly evolved developmental events in the sex determination cascades (Marin and Baker,1998). In general, the primordial germ cells in developing embryos are considered to be sexually bipotent (McLaren,1983,1991,1997; Jameson et al.,1997,1998). They differentiate into male germ cells in the developing testis and into female germ cells in the developing ovary. In nonmammalian vertebrates, the sex reversal of gonads has been demonstrated in both sexes (Reinboth,1983; Scheib,1983). In teleost fish, in particular, a complete sex reversal of gonads in both sexes can be induced by the administration of sex steroid hormones, and in some cases, by modulating environmental factors (Yamamoto,1953,1958,1969; Nakamura et al.,1998; Baroiller et al.,1999).
In contrast to many developmental processes, sex-determining mechanisms show no clear evolutionary conservation between phyla. In most mammals, the male-inducing master gene, SRY, is located on the Y chromosome and is, therefore, absent in XX females. SRY seems to be specific to mammals (Capel,2000). However, recent studies indicate that some downstream gene products of sex determination genes are functionally similar in diverse species. One of these genes, the doublesex/mab-3 conserved domain (DM) domain-containing gene (DMRT1; doublesex/mab-3 related transcription factor-1), has been found in mammals, a bird species, a reptile species, a frog species, teleosts, nematodes, and flies (Shen and Hodgkin,1988; Coschigano and Wensink,1993; Raymond et al.,1998,1999; Smith et al.,1999; De Grandi et al.,2000; Kettlewell et al.,2000; Guan et al.,2000; Marchand et al.,2000; Shibata et al.,2002). The DM domain, a zinc finger-like DNA-binding motif (Erdman and Burtis,1993; Raymond et al.,1998), was first identified in the sexual regulatory proteins Drosophila doublesex and C. elegans mab-3 and is widely conserved among metazoans. DMRT1 appears to be involved in a certain type of XY sex reversal in humans (Bennet et al.,1993; Flejter et al.,1998; Yi et al.,2000; Veitia et al.,2001). In chickens, aromatase inhibitor-treated ZW embryos showed sex reversal, accompanied by DMRT1 expression (Smith et al.,2003). In teleost fish also, DMRT1 has been linked to both natural sex change from female to male, and to sex hormone-dependent sex reversal (He et al.,2003; Guan et al.,2000; Marchand et al.,2000). These findings strongly suggest that DMRT1 homologs are involved in testicular differentiation in the process of normal sex development and sex reversal in nonmammalian vertebrates. Unfortunately, in teleost fish, the expression profiles of DMRT1 compared with those of morphological sexual dimorphisms in terms of germ cell number and histogenesis have not been reported in any species except for medaka (Kobayashi et al.,2004). In medaka (Oryzias latipes), the sex-determining gene DMY has been identified, and this gene is a DMRT1 homolog (Matsuda et al.,2002,2003). Also, DMY is found in two subspecies of medaka (Matsuda et al.,2003), but in no other medaka species. Thus, it is not easy to clarify the role of DMRT1 during sex differentiation by studying only the medaka. It has also been shown that the SRY-related gene Sox9 (SRY-like HMG-box 9) is necessary and sufficient to cause testicular differentiation in mammals (Vidal et al.,2001). However, no male-specific expression of Sox9 is seen until the end of the temperature-sensitive period in some reptiles (Western et al.,1999; Moreno-Medoza et al.,1999; Valleley et al.,2001). In addition, a few reports on the Sox9 expression profiles in teleosts have been published (Nakamoto et al.,2005; Rodriguez-Mari et al.,2005), suggesting that male-specific expression is seen after morphological sex differentiation. Although a little information on DMRT1 and Sox9 during gonadal sex differentiation has been reported in teleost fish, to date, no information on DMRT1 and Sox9 expression during gonadal sex differentiation has been reported from the viewpoint of a comparison with the signs of morphological sex difference.
Nile tilapia, Oreochromis niloticus, is a diploid fish (XX-XY-type) well characterized as a model for gonadal sex differentiation (Nakamura et al.,1998,1999). In this species, sex reversal is easily inducible using hormones, and sex-reversed males (XX) and supermales (YY) have been produced (Nakamura et al.,1998; Carrasco et al.,1999; Kobayashi et al.,2003). Consequently, all-male or all-female populations can be easily produced using these males (Kobayashi et al.,2000). In this study, we first investigated the DMRT1 expression profiles during gonadal differentiation and development in normal fish and in those subjected to hormone-induced sex-reversal (XY and XX sex reversal). In addition, we examined the Sox9 (Sox9a) expression profiles during gonadal differentiation and development in normal and sex-reversed (XY and XX sex reversal) fish. In this study, these expression profiles were examined in comparison with morphological sex differences in terms of germ cell number and histogenesis.
Gonadal Sex Differentiation
The morphological sexual dimorphism during gonadal differentiation consisted of changes in germ cell number and histogenesis (Figs. 1, 2). Primordial germ cells (PGCs) migrated into the gonadal anlagen 3 days after hatching (dah), 7 days after fertilization (Kobayashi et al.,2000,2002). Thereafter, the germ cell numbers did not significantly change in either sex from 5 to 8 dah. After 8 dah, however, the XX female germ cells continued to proliferate, whereas the germ cell numbers did not change from 9 to 14 dah in XY male gonads (Fig. 1). As described in a previous report (Kobayashi et al.,2000), the number of germ cells in XY male gonads increased again after 14 dah, but spermatogenesis was not observed until 70 dah. The formation of the ovarian cavity or the intratesticular efferent duct occurred between 20 and 25 dah in the XX and XY gonads, respectively. Figure 2 shows the gonadal differentiation in tilapia. In XY gonads, the medullary cell mass derived from the germinal epithelium, neighbored the germ-cell–surrounding cells and developed into a cord-like structure as the intratesticular efferent duct. Although no medullary cell mass developed in the XX gonads, an ovarian cavity was formed by the extension of both tips of the gonads (Nakamura et al.,1998).
tDMRT1 Expression During Gonadal Differentiation and Development
Using all-XY male and all-XX female populations, tDMRT1 expression was examined during gonadal differentiation using reverse transcriptase-polymerase chain reaction (RT-PCR) and in situ hybridization (Figs. 3–5). Strong tDMRT1 expression was detected in XY gonads by RT-PCR from 6 dah through to the development of mature testes, whereas tDMRT1 expression was not detectable in the XX gonads (Fig. 3). In XY gonads at 6 dah, in situ hybridization revealed weak signals for tDMRT1 in the inner cells located around the germ cells (Fig. 4A), although no signals were observed in XX gonads (Fig. 4J). In XY gonads from 7 to 8 dah, the tDMRT1 signal intensity increased and was localized in the epithelial cells surrounding the germ cells (Fig. 4B). In contrast, signals for tDMRT1 were not detected during the gonadal differentiation and development of XX gonads (Fig. 4J–M). With the progression of gonadal differentiation, tDMRT1 expression was seen not only in the germ-cell–surrounding cells (pre-Sertoli cells) but also in the medullary cell mass neighboring the germ-cell–surrounding cells of the XY gonads (Fig. 4C).
Around 25 dah, when the formation of the intratesticular efferent duct anlagen was observed, tDMRT1 expression was found in the Sertoli cells, medullary cell mass and epithelial cells of the anlagen of the efferent duct (Figs. 2, 4D,E). From the localization of tDMRT1-positive cells, it appears that some portion of the tDMRT1-positive medullary cell mass neighboring the Sertoli cells differentiates into the precursor cell mass of the efferent duct, and onward into the anlagen of the intratesticular efferent duct. In mature fish, the specific expression of tDMRT1 was localized only in the Sertoli and epithelial cells of the intratesticular efferent duct in the mature testes and was undetectable in the ovaries (Fig. 4F–H,M). Using laminin antibody, further analysis of tDMRT1 mRNA localization in the testis definitely showed that tDMRT1-positive cells were localized only within the seminiferous tubules but not in the interstitium (Fig. 5), indicating that tDMRT1 is expressed in Sertoli-cell-lineage cells specifically. We did not detect any tDMRT1 expression in any of the germ cells during gonadal differentiation and development.
Sox9a Expression During Gonadal Differentiation and Development
Although Sox9a was expressed in the germ-cell–surrounding cells, Sox9a expression showed no difference between the sexes until 25 dah, according to the RT-PCR and in situ hybridization results (Figs. 3, 6). The first signs of the sexual dimorphic expression of Sox9a were observed at 25 dah, when sex differences in histological architecture began to be seen, such as the incipient formation of the intratesticular efferent duct and the ovarian cavity in the males and the females, respectively (Fig. 6). The localization of Sox9a showed sex differences at this stage. In the XY gonad, signals for Sox9a are seen in the germ-cell–surrounding cells and the medullary cell mass (Fig. 6E). In the XX gonad, signals are seen in the germ-cell–surrounding cells and interstitial cells in the tip of the gonad (Fig. 6F). After 25 dah, Sox9a expression was seen specifically in the germ-cell–surrounding cells of the male gonads. In contrast to tDMRT1, however, Sox9a was not expressed in the epithelial cells of the efferent duct in the testis (Fig. 6G). Taken together, these results indicate that tDMRT1 is expressed in XY gonads specifically before the appearance of any morphological sex differences, and that Sox9a is expressed in XY gonads specifically after the appearance of sex differences in histological architecture, such as the formation of intrastesticular efferent duct or ovarian cavity. In mature gonads, Sox9a is expressed in males but not females (Figs. 3, 6G,H).
tDMRT1 Expression Is Induced in XX Sex Reversal and Suppressed in XY Sex Reversal
A previous study has indicated that tDMRT1 is expressed in the phenotypic testis irrespective of the genetic sex, according to the Northern analysis results (Guan et al.,2000). In our present study, in situ hybridization showed that, in mature sex-reversed fish, the specific expression of tDMRT1 was localized only in the Sertoli and epithelial cells of the intratesticular efferent duct in the mature testis, irrespective of the genotypic sex (Fig. 4G–I) and was undetectable in the phenotypic ovary (Fig. 4M,N). In sex-reversed XX testes induced by androgen, tDMRT1 localization was similar to that in XY normal testes (Fig. 4I). Also, in sex-reversed XY ovaries induced by estrogen, no tDMRT1 expression was detected, just as in XX normal ovaries (Fig. 4N).
We also examined the expression profiles for tDMRT1 during sex- reversed gonadal differentiation. In estrogen-induced XY sex reversal, RT-PCR analysis and in situ hybridization both indicated that tDMRT1 expression decreased after the induction of sex reversal and then disappeared (Figs. 3, 7D,E). On the other hand, tDMRT1 expression was induced specifically in the germ-cell–surrounding cells after the induction of XX sex reversal by 17α-methyltestosterone (MT; Figs. 3, 7I, J). At 15 to 20 dah, the XX gonads of MT-treated XX fry showed a testicular structure characterized by the formation of the intratesticular efferent duct, and strong tDMRT1 expression was observed in the germ-cell–surrounding cells, medullary-cell-mass cells neighboring the germ-cell–surrounding cells and efferent duct cells (Fig. 7J), as in the XY male gonad. In both sex reversal experiments, in situ hybridization and RT-PCR revealed that tDMRT1 expression in the germ-cell–surrounding cells and medullary-cell-mass cells was up-regulated during XX sex reversal and down-regulated during XY sex reversal (Figs. 3, 7). In these stages, XX gonads have steroid-producing cells located near the blood vessels, but not in XY gonads (Kobayashi et al.,2003). Although streoidogenic enzymes are down-regulated completely in the steroid-producing cells of sex-reversed XX gonads after the treatment of androgen as in XY gonads (Bhandari et al.,2006), no tDMRT1 expression is detectable in those cells (data not shown). These results indicate that tDMRT1 is an ideal molecular marker for the Sertoli cell lineage, irrespective of the genetic sex. In contrast to tDMRT1, Sox9a was not affected by the hormonal treatment for sex reversal applied in this study. Just as in normal fry, the sexual dimorphic expression of Sox9a was seen after 25 dah (data not shown).
Induction of Sex Reversal Is Accompanied by Phenotypic Sex-Specific Gonial Proliferation
As mentioned above, the gonial germ cells of XY fry did not proliferate from 9 to 14 dah, but the gonial germ cells of XX fry continued to proliferate. In XX fry treated with MT from 12 to 14 dah, tDMRT1 expression was detected in the germ-cell–supporting cells by 15 dah (Fig. 7). In XY fry treated with estrogen from 4 to 6 dah, tDMRT1 expression was undetectable in the gonads by 15 dah. We also found changes in the germ cell numbers in androgen-induced XX sex reversal (Fig. 8A). The mean number of germ cells in MT-treated XX individuals (98.0 ± 9.0) was significantly lower than that for the control XX fry (135.0 ± 13.0) and similar to that for the initial controls (91.3 ± 13.2). During estrogen-induced XY sex reversal, on the other hand, the number of germ cells was significantly higher than that for the control XY fry and similar to that for the control XX fry (Fig. 8B). These results indicate that hormone-dependent sex reversal is accompanied by sex-specific tDMRT1 expression and germ cell proliferation, as in normal gonadal differentiation.
This study indicates that, in tilapia, DMRT1 expression occurs specifically in the Sertoli cell lineage during testicular differentiation, irrespective of the genetic sex. The androgen-induced testicular differentiation of XX fry caused the induction of tDMRT1 expression in the germ-cell–surrounding cells and the precursor cell mass for the efferent duct. In contrast, the estrogen-induced ovarian differentiation of XY fry caused the suppression of tDMRT1 expression in the germ-cell–surrounding cells and the precursor cell mass for the efferent duct. Although Sox9a was expressed in the germ-cell–surrounding cells specifically, it was not observed in the epithelial cells of the efferent duct. In contrast to tDMRT1, the sexual dimorphic expression of Sox9a occurred after 25 dah, when sex differences in histological architecture first appeared, such as the efferent duct and the ovarian cavity in the males and the females, respectively. This finding indicates that tDMRT1 expression precedes any morphological signs of testicular differentiation and is a superior marker for Sertoli-cell-lineage cell differentiation in tilapia.
To date, DMRT1 has been cloned and its expression pattern was reported in several vertebrates (mouse: Raymond et al.,1999; chicken: Smith et al.,1999; turtle: Kettlewell et al.,2000; tilapia: Guan et al.,2000; trout, Marchand et al.,2000). However, information on the distribution of DMRT1 within the gonads is limited to the mouse (Raymond et al.,1999) and the chicken (Smith et al.,2003). Although the distribution of DMRT1 in the gonads in medaka (Kobayashi et al.,2004) and zebrafish (Guo et al.,2005) was reported, these reports were not as thoroughly described. These studies have all shown that DMRT1 is dominantly expressed in the testis. During gonadal sex differentiation in the chicken, in particular, DMRT1 was found by in situ hybridization to be expressed almost exclusively in the testis, and the signals were localized in the Sertoli cells (Smith et al.,1999,2003). Evidence of the male-specific expression of DMRT1 was also obtained in humans, a reptile species and a frog species (Raymond et al.,1999; Smith et al.,1999; Kettlewell et al.,2000; Moniot et al.,2000; Shibata et al.,2002). In teleost fish, several reports indicate that DMRT1 expression is related to the development of the testes (Guan et al.,2000; Marchand et al.,2000; He et al.,2003). The present study indicates clearly, for the first time in fish, that tDMRT1 is expressed in the germ-cell–surrounding cells and medullary-cell-mass cells, which differentiate into the efferent duct during testicular differentiation, irrespective of the genetic sex. A recent report on a study using chickens showed that ZW sex reversal from female to male was accompanied by DMRT1 expression in medullary cells, which developed into Sertoli cells (Smith et al.,2003). Furthermore, our recent study indicates that medaka DMRT1 is expressed in the germ-cell–surrounding cells and the intratesticular efferent duct, although medaka DMRT1 is expressed after testicular differentiation, unlike in tilapia (Kobayashi et al.,2004). In medaka, however, sex determining gene DMY, a DMRT1 orthologue, is expressed in germ-cell–surrounding cell lineage cells before sex differentiation (Kobayashi et al.,2004). Taken together, these results indicate that DMRT1 is expressed in Sertoli-cell-lineage cells specifically, suggesting that the findings from human, bird, and reptile models are consistent with the results for tilapia.
In a study using RT-PCR, DMRT1 expression was induced after the androgen-induced sex reversal of a frog (Shibata et al.,2002). In a study using semiquantitative RT-PCR in trout, estrogen-induced XY sex reversal was associated with a reduction in DMRT1 expression in the gonads (Marchand et al.,2000). In tilapia, the Northern analysis has shown that DMRT1 is expressed in XX testes, just as in XY testes (Guan et al.,2000). From these findings, we hypothesized that the sex steroids inducing sex reversal regulate the expression of DMRT1 directly or indirectly. The present study indicates clearly that the estrogen-induced ovarian differentiation of XY fry and the androgen-induced testicular differentiation of XX fry are accompanied by the suppression and induction of DMRT1 in germ-cell–surrounding cells, respectively. Together, these results suggest that DMRT1 might be regulated by sex steroids in nonmammalian vertebrates that can be induced to undergo sex reversal, although, in rats, it has been reported that follicle-stimulating hormone regulates DMRT1 expression after birth (Chen and Heckert,2001). Currently, little information on the 5′-flanking region of DMRT1 is available for nonmammalian vertebrates. In tilapia, no estrogen- or androgen-responsive elements have been found in the ca. 600-bp-long sequence of the 5′-flanking region of DMRT1, although they were found at other sites (e.g., SRY and Gata-1, etc.) (Guan et al.,2000). Further investigation will be necessary to clarify the mechanisms of DMRT1 transcriptional regulation.
As for teleosts, it was reported that two types of Sox9 were reported as Sox9a and Sox9b/a2 (Zebrafish, Chiang et al.,2001; medaka, Nakamoto et al.,2005). However, testis-dominant type of Sox9 was different in each species (Zebrafish, tilapia, Sox9a; medaka, Sox9b/a2). Sox9 expression is observed in the germ-cell–surrounding cells of both sexes before gonadal sex differentiation in tilapia and medaka (this study; Nakamoto et al.,2005). After the appearance of morphological sex differences, Sox9 is expressed in the germ-cell–surrounding cells of XY gonads specifically. These results coincide with those for reptiles (Western et al.,1999). Furthermore, Sox9 is not expressed in the epithelial cells of the intratesticular efferent duct in tilapia and medaka, in contrast to DMRT1, whereas both genes are expressed in the germ-cell–surrounding cells of the testis, suggesting that Sox9 is not involved in the differentiation of the intratesticular efferent duct. Further study is necessary to clarify the role of Sox9 in the testicular differentiation of lower vertebrates.
In conclusion, this study indicates that, in tilapia, DMRT1 expression precedes testicular differentiation, and expression occurs specifically in the Sertoli cell lineage, including the differentiation of the intratesticular efferent duct. These pieces of evidence suggest that DMRT1 is involved in Sertoli-cell-lineage cell differentiation. On the other hand, the male-specific expression of Sox9a occurs after morphological testicular differentiation, as in reptiles. Furthermore, Sox9a is not expressed in the epithelial cells of the intratesticular efferent duct. Taken together, the results of this study suggest that DMRT1 is a superior molecular marker for somatic testicular differentiation, and that DMRT1 and Sox9a play different roles in the testicular differentiation of tilapia.
Nile tilapia were maintained in re-circulating 500-L freshwater tanks at 26 ± 1°C until use. All genetic females (XX) and males (XY) were obtained by the artificial fertilization of eggs from a normal female (XX) and sperm from a sex-reversed male (XX) or a supermale (YY), respectively, as described in our previous report (Kobayashi et al.,2000). The fertilized eggs were cultured in re-circulating water at 26 ± 1°C. One hundred fish each from the all-female and all-male populations were sexed histologically. The all-female population was found to contain 100% females (100/100), and the all-male population was found to contain 100% males (100/100). No hermaphroditic gonads were observed (data not shown).
Induction of Sex Reversal
Sex reversal in the XX fry was induced as follows. The XX fry were fed commercial food (OTOHIME, Marubeni-Nisshin Feed Co. Ltd., Japan) containing 10 μg of 17α-methyltestosterone (MT; Sigma Co. Ltd., St. Louis, MO) per gram of diet 7 to 15 dah or 12 to 14 dah. This food was given four times a day. The food was immersed in ethanol or ethanol containing MT for the control and the MT treatment, respectively. Then, the food was dried at 37°C overnight. MT containing food caused complete sex reversal to functional males (7 to 15 dah: 30/30 males and 0/30 females, 12 to 14 dah: 30/30 males and 0/30 females). XY sex reversal was induced by the immersion of the fry in 0.1 μg of 17α-ethynylestradiol (EE2, Sigma Co. Ltd.)/ml solution for 2 days, 4 to 5 dah, in accordance with a previous report (Kobayashi et al.,2003). As control, XY fry was immersed in steroid-free solution.
First-strand cDNA was synthesized from 400 ng of total RNA in 20 μl using Omniscript (QIAGEN) with oligo-dT primers. PCR was carried out in a 25-μl reaction mixture containing 1 μl of the first-strand cDNA DMRT1 of the gonads. The specific primers for DMRT1 and Sox9a were (5′-CGGATACGTGTCTCCACTGAAGGGAC-3′, 5′-CGATGGCGTCGACGGTGAAGC- CGGC-3′) and (5′-CGGGAGAGCATTCAGGTCAGTC-3′, 5′-CAGCTTTG- CTGGAGGGAAGG-3′), respectively. The specific primer set for β-actin was described in a previous report (Kobayashi et al.,2000,2002). Using these specific primer sets, PCR was performed in accordance with a previous report (Kobayashi et al.,2000).
In Situ Hybridization
The trunks from 0–10 dah fry were dissected out and fixed in 4% paraformaldehyde (4% PFA) in 0.1 M phosphate buffer (PB; pH 7.4) at 4°C overnight. The gonads of 10–100-dah fry were removed and fixed in 4% PFA. After fixation, the tissues were embedded in paraffin, and cross-sections were cut at a thickness of 5 μm. Sense and antisense digoxigenin (DIG)-labeled RNA probes were transcribed in vitro using an RNA labeling kit (Roche). In situ hybridization was performed as described previously (Kobayashi et al.,2000). In some cases, anti-DIG–conjugated horseradish peroxidase was used and diaminobenzidine staining was performed. To detect DMRT1 mRNA and Laminin, first the sections were stained with anti-Laminin antibody (Sigma) containing RNase inhibitor using Alexa-568 antibody (Invitrogen). Next, the sections were hybridized with DIG-labeled DMRT1 and then stained with TSA-biotin system (Perkin-Elmer). Finally, the sections were stained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; Invitrogen).
To describe the gonadal sex differentiation, gonads were fixed in Bouin's solution or 4% PFA. After fixation, the gonads were embedded in paraffin and cross-sectioned serially at a thickness of 5 μm. Sections were stained with Carazzie's hematoxylin.
Determination of the Germ Cell Numbers
To determine the changes in germ cell number during gonadogenesis, all-XY male and all-XX female fry were fixed daily between 5 and 15 dah in Bouin's solution or in 3.7% formaldehyde in TBS. The gonads from all-XY and all-XX fry at 20 dah were also fixed. After fixation, the gonads were embedded in paraffin and cross-sectioned serially at a thickness of 5 μm. As shown in Figure 1, the germ cell numbers were determined in accordance with a previous report (Satoh and Egami,1972). Briefly, all germ cells were counted in each individual. Five to six individuals derived from the same parents were examined at each developmental stage for each sex. After the cell counting, the means ± SE were calculated for each sex at each stage, and then the differences between the sexes were evaluated statistically by the paired t-test for each stage. In Figure 9, the gonads were also stained with anti-vasa antibody, which also specifically stains germ cells (Kobayashi et al.,2002), under whole-mount conditions. After staining, the germ cells were counted in eight individuals per sampling point. These determinations were performed using three independent series of tissues. Both methods yielded comparable germ cell numbers. After the cell counting, the means ± SE were calculated, and then the differences between the sexes were evaluated statistically using Duncan's new multiple range test, in accordance with a previous report (Kobayashi et al.,1993).
This work was supported in part by a Grant-in-Aid for Research for Priority Area from the Ministry of Education, Science, Sports and Culture of Japan, CREST, JST and Bio Design program from the Ministry of Agriculture, Forestry and Fisheries.