Molecular players involved in temperature-dependent sex determination and sex differentiation in Teleost fish

The dmy gene

The dmy gene is a master sex-determining gene that was first described in 2002 in the medaka, Oryzias latipes, which has an XX/XY (female homogamety) sex-determining system [31, 32]. It is an excellent candidate as a primary male determining gene equivalent to the sry gene of mammals. The gene, designated dmy[31] or dmrt1Y[32], was located in the Y-specific chromosomal region that contains the male determining gene. It is important to note that this is the only structural gene, which specifies a functional protein in the Y-specific chromosome region [32, 33]. The product of this gene contains a DNA-binding domain called the DM domain, a structural motif present in a family of genes that is found in a wide range of invertebrates and vertebrates from nematodes and flies to humans [34–37]. The dmy/dmrt1Y gene is assumed to have arisen from a recent duplication event (5 to 10 million years ago) of the autosomal dmrt1 gene [33, 38]. A fragment of the medaka linkage group 9 that contains the dmrt1 gene was duplicated and inserted into the chromosome of linkage group 1, which subsequently became the Y chromosome [39].

The dmy gene is expressed before the sex-determining period, specifically, before the first appearance of morphological sex differences at the hatching stage when the male exhibits a decreased number of primordial germ cells [39]. The level of expression of the dmy gene during the sex-determining period appears to be critical for its function, since mutants that do not express this gene fail to become males and develop as sex-reversed XY females [31, 40]. In sex-reversed XY females induced by estrogen treatment, the dmy gene is expressed in the ovaries at a level similar to that in the testes, which means that the expression of this gene is not affected by the administration of exogenous estrogen [32]. Furthermore, the early expression of dmy in the somatic cells of the undifferentiated gonad is not influenced by 17β-estradiol (E₂, natural estrogen) treatment [41]. In addition, a high-temperature treatment (above 27°C) during the sex-determining period fails to induce expression of dmy although it leads to the masculinization of medaka XX females [42]. In summary, the expression of dmy is in perfect agreement with its function as a male upstream determining gene in medaka.

Two distinct natural mutations in the dmy gene present in wild medaka populations have been shown to induce XY genotypes to become fertile females [31]. A subsequent investigation on a natural mutant of medaka fry also showed an increased number of germ cells at day 0 post-hatching due to the low expression of dmy, with fry developing into females [40]. A gene knockdown experiment that used gripNA antisense oligonucleotides directed against dmy transcripts also showed that dmy knockdown XY medaka fry and control XX females had comparable germ cell numbers, which indicates that the disruption of the dmy gene resulted in the gonads entering the female pathway [43]. Thus, the dmy gene is necessary for the development of males in medaka. Moreover, over-expression of dmy by injecting dmy genomic DNA fragment into XX female eggs or over-expression of the dmy cDNA under the control of the CMV promoter in XX females resulted in XX individuals developing into males [44]. These results indicate that expression of dmy may be sufficient to induce male development in XX females.

Collectively, these results suggest that the dmy gene is a good candidate male determining gene at least in some of strains of medaka and its relative species Oryzias curvinotus[37, 40]. In terms of temperature effects on sex determination and differentiation, it is not yet known whether temperature can affect the expression of dmy[42] (see next section) although high-temperature treatments during the early stages of development result in females having a male phenotype. It is interesting to note that the thermosensitive period of sex differentiation in medaka lies between developmental stages 5 to 6 (8 to 16 cells) and 36 (heart development stage) [42], which is just before expression of dmy begins (stage 36). It is assumed that the function of dmy in the male embryo during the sex-determining period is to control the proliferation of primordial germ cells [39]. Likewise, Selim et al. [45] observed that a high-temperature treatment before hatching inhibited the proliferation of germ cells and the development of oocytes and consequently resulted in sex reversal. The fact that dmy and temperature have the same effects on the proliferation of primordial germ cells, and that genotypic (XY) males and genotypic (XX) females possess a dimorphic sensitive pattern to temperature during TSP [42] show that temperature may play a pivotal role in female fate as does dmy in male fate during evolution. These reports also suggest that genetic as well as environmental factors are not incompatible in terms of effects on sex determination or differentiation with the existence of genotype by environment interactions, as already mentioned in fish species and reptiles [1, 4, 46, 47]. To understand the interaction between the sex-determining gene (dmy) and the environment factor (temperature), the effects of temperature on the expression of dmy and its downstream targets should be investigated further.

The amhy, gsdf, amhr2, and sdY genes

Just prior to and during the preparation of the present review, four additional strong candidate master sex-determining (SD) genes were reported in fish, which indicates that this area of study is moving fast. These include the genes amhy in Patagonian pejerrey Odontesthes hatchery[48], gsdf in Oryzias luzonensis[49], amhr2 in fugu (tiger pufferfish) Takifugu rubripes[50], and sdY in rainbow trout Oncorhynchus mykiss[51]. All master SD genes reported to date are in Figure 2, which also includes the mammalian sry gene, the dmrt1 gene in birds, the DM-W gene in Xenopus laevis, and the dmy gene in medaka O. latipes.

These novel master SD genes highlight the importance of non-transcriptional factors in sex determination since three of these genes i.e. amhy, amhr2 and gsdf are involved in the TGF-β signaling pathway, while the sry, dmrt1, DM-W and dmy genes code for transcription factors (Figure 2). In mammals, the TGF-β signaling pathway has been shown to play important roles in the development of ovarian and testicular functions [52, 53] but there is no evidence that it may be involved in sex reversal. Nevertheless, the identification of the three SD genes amhy, amhr2, and gsdf[48–50] indicates that the TGF-β signaling pathway plays a critical role in the commitment to the fate of either testicular or ovarian development. It has been hypothesized that this pathway may have a more dominant role in gonadal sex determination in non-mammalian vertebrates than in mammals [54]. For example, while amh clearly lies far downstream in the sex determination pathway and is controlled by sox9, it has been shown in chicken that the expression of amh precedes that of sox9 in males [55, 56]. In the American alligator, in which TSD is present, expression of amh precedes the onset of sox9 expression during testis differentiation [57]. Analysis of the expression patterns of sox9a, sox9b and sox8 compared to that of amh in different cell types ruled out the hypothesis that amh is regulated directly by sox9 or sox8 at least in the granulosa cells of adult zebrafish ovaries [58]. These findings combined with the aforementioned results in different fish species corroborate or confirm this hypothesis.

To date, one of the four novel SD genes, sdY, which is a truncated, divergent form of the immune-related gene irf9[51], has not been reported in the literature unlike the three others, which have been characterized and shown to play important conserved roles in the gonadal differentiation pathway across vertebrates. The sdY gene, which is expressed in the somatic cells that surround the germ cells, encodes a novel protein that displays sequence homology with the carboxy-terminal domain of irf9[51]. IRF9 is a transcription regulatory factor that mediates signaling by type I interferon in mammals [59]. The sdY sequence is highly conserved in all salmonids and is a male specific Y-chromosome gene in the majority of these species except in two whitefish species (subfamily Coregoninae) [60]. These results indicate that most salmonids share a conserved master sex-determining gene and that an alternative sex-determining system may have also evolved in this family. For more details, refer to Kikuchi and Hamaguchi [61].

Growing evidence demonstrates that the sex-determining pathways in fish species are conserved and diverse (Figure 2, see O. latipes and O. luzonensis). The dmy in O. latipes and gsdf genes in O. luzonensis (a close relative of O. latipes) are estimated to have appeared about 10 and 5 million years ago, respectively [49]. It is very interesting to note that the expression of the gsdf ^Y gene can lead to a male phenotype in XX medaka O. latipes in the absence of dmy expression. The gsdf ^Y gene is located downstream to dmy in O. latipes and originates from gsdf[49], which means that, at present, it functions independently of the existing sex-determining gene, and has usurped the control of the downstream cascade of sex determination within 5 million years.

Genetic by environment interactions are a hot topic. Nevertheless, no evidence has shown that temperature influences expression of sex-determining genes during the TSP. TSD is claimed in more and more fish species but how temperature can act as a signal for the undifferentiated gonad to generate a male or female pathway and what are the downstream target(s) of most sex-determining genes, remain to be elucidated (Figure 2).

The dmrt1 gene

The mechanisms that control sex determination and sex differentiation are highly variable among different phyla [62]. However, genes that are located downstream in the sex determination pathway are conserved [63]. For example, the dmrt1 gene plays a pivotal role in the fate of gonads in fish, reptiles, birds and mammals, and is expressed in the developing gonads, or in the adult testis and/or in the ovary. The dmrt1 gene encodes a putative transcription factor containing a zinc-finger-like DNA-binding motif (DM domain) and was initially identified in nematodes [64] and flies [34]. This gene is regarded as a crucial regulator of male sexual development from invertebrates to humans [65] and also as evidence that the sex determination and gonad differentiation mechanisms are conserved across different lineages [35, 66, 67].

The ray-finned fish (Actinopterygii) has two paralogous copies for many genes (e.g. dmrt1a and dmrt1b, cyp19a1a and cyp19a1b, sox9a and sox9b) due to the hypothesized fish-specific genome duplication that is dated between 335 and 404 million year ago [68]. With the increasing availability of whole-genome sequences, the comparative analysis of genes and genomes will reveal the evolution and phenotypic diversification of the third round (and fourth round in some fish species such as common carp Cyprinus carpio) of genome duplication [69–72]. Some duplicated genes have evolved new functions, while others have disappeared [71]. In this review, we focus mainly on the dmrt1 gene among the many duplicated genes that are mostly related to sex determination and differentiation and have been extensively investigated.

The most direct evidence for the important role of the dmrt1 gene has come from the discovery that the sex-determining gene of medaka, dmy originates from a duplicated copy of the autosomal dmrt1 gene [44]. In teleost fish, the expression of dmrt1 is associated with temperature effects and displays variable patterns in different species (See Additional file 1: Table S1). In tilapia (Oreochromis niloticus) and trout (O. mykiss), dmrt1 is expressed in males prior to sex differentiation but not in females, which indicates that, in these fish species, it is involved in testis formation and differentiation [73–75]. However, in other fish species, like medaka, pejerrey (Odontesthes bonariensis), and European sea bass (Dicentrarchus labrax), sexually dimorphic expression of dmrt1 in males and females was reported (See Additional file 1: Table S1), which indicates that, in these cases, dmrt1 participates in testis and ovarian development.

Regarding temperature effects on sex ratios, although expression profiles of dmrt1 have been described in a limited number of fish species, the results suggest that it plays an essential function in male development and testis formation in fish species with TSD or GSD + TE. In pejerrey, a fish species with pure TSD (based on the criteria of Ospina-Álvarez and Piferrer [4]), the abundance of dmrt1 transcripts differed clearly between larvae reared at female producing (or promoting) temperature (FPT, 100% female) and larvae reared at male producing temperature (MPT, 100% male). The expression of dmrt1 was significantly higher at MPT than at FPT during the two weeks before the first signs of morphological differentiation of the testis, and remained high during sexual differentiation at MPT, which highlights the importance of dmrt1 during the first stage of the gonadal sex differentiation cascade, rather than during the morphological differentiation of the gonads. However, it is interesting to note that, in larvae reared at FPT, dmrt1 expression remained low throughout the experiment (See Additional file 1: Table S1), which opens the question on the function of dmrt1 in ovarian development in pejerrey [76]. Similar results have been reported in reptiles with TSD [30, 77–79].

In red-eared slider turtle (Trachemys scripta) embryos, expression of dmrt1 in the gonads is up-regulated with a large difference between embryos reared at FPT and at MPT, which indicates that dmrt1 is necessary for male development [80]. In addition, if the eggshells of developing T. scripta embryos are treated with estrogen before the thermosensitive period, dmrt1 expression is inhibited during this period [81]. Suzuki et al. [41] also found that the level of dmrt1 expression is very low (even undetectable) in testes treated with 17β-estradiol (E₂, natural estrogen) compared to control untreated testes. In zebrafish, exposure of larvae to environmentally relevant concentrations of 17α-ethinylestradiol (EE₂, synthetic estrogen) suppresses dmrt1 expression during gonad differentiation [82]. To our knowledge, there is no evidence on the possible implication of exogenous estrogen treatment on dmrt1 expression in the induction of sex reversal in fish species with TSD. Taken together, these data strongly indicate that dmrt1 is involved in testis formation and differentiation, and that its expression is sensitive to both temperature and estrogen in turtles with TSD and fish species with GSD + TE.

Fernandino et al. [83] reported that the expression level of dmrt1 was not proportional to exposure temperature [83], which suggests that other genes/factors acting as sex inducers and located upstream of dmrt1 are involved in the transduction of temperature and the gonad differentiation cascade. As in turtles, it is assumed that dmrt1 still holds the capacity of being a master sex-determining gene in several species with GSD and that it can probably be directly modulated by temperature in species with TSD [30].

Medaka, which is a widely used research model, did not pass the criteria to be diagnosed as a true case of species with TSD because the temperature that causes sex reversal is not within the range of temperatures to which medaka are exposed during the development in the wild [4]. Incubation of medaka at a high temperature (> 30°C) induces sex reversal from genotypic (XX) females to phenotypic males and 100% males are obtained at 34°C [42, 84], which means that the sex ratio in medaka is genetically determined with strong temperature effects (GSD + TE). During the temperature-dependent sex reversal of medaka, substantial amounts of dmrt1 mRNA were detected in six of 12 XX embryos incubated at 32°C but not at the “sexually neutral” temperature of 25°C. Thus, the expression of dmrt1 could be considered as a marker of sex reversal. This is in a good agreement with the results reported by Hattori et al. [42] who observed 40% of sex-reversed males at 32°C, which confirms that dmrt1 may have an essential role in temperature-dependent sex reversal from genetic females to phenotypic males. However, other studies on medaka reported that dmrt1 was expressed at very low levels up to 15-20 days post-hatching (DPH), irrespective of the genetic sex [32, 85]. These results lead us to raise two relevant questions: (1) does a high temperature accelerate the cascade of molecular events that lead to testis differentiation (but not ovarian differentiation) and result in earlier expression of dmrt1? and (2) is the sex-specific (32°C, embryogenesis) vs. non-sex-specific (possibly 25°C, after hatching) expression of dmrt1 due to the influence of incubation at high-temperature or is it necessary for natural ovary development? More studies on dmrt1 expression at various developmental temperatures are necessary to clarify these questions and to understand how this gene contributes to the fate, development, and/or maintenance of the testis as well as the ovary in fish species.

It has been reported that in European sea bass that has GSD + TE, the expression level of dmrt1 mRNA increased continuously in differentiating and differentiated testes between 150 and 300 DPH, while in adult ovaries it increased until 200 DPH and then decreased to an undetectable level [86], which means that dmrt1 is not necessary for ovarian maintenance in this fish species. The delayed expression of dmrt1 suggested that it is neither involved in the formation of the undifferentiated gonad (formed around 90 DPH), nor in the high-temperature induced masculinization mechanism. In contrast, in the Nile tilapia in which the sex ratio is genetically determined with strong temperature effects, dmrt1 expression presented a rapid up-regulation pattern during the critical period of sex differentiation at high temperature in both the XY and XX male population [5], which implies that in this species dmrt1 is involved in testicular differentiation and high-temperature induced sex reversal. In the pufferfish (T. rubripes), dmrt1 appears to be involved in the degeneration of germ cells in the ovary, which in turn, causes ovary to testis sex reversal induced by high-temperature exposure during the early development of gonads [87].

Based on the results of the literature on gene expression patterns and on the pathways involved in temperature-induced sex reversal from XX females to phenotypic males in Nile tilapia, we propose a scheme that depicts the putative male pathway induced by high-temperature/MPT (Figure 3). However, how temperature triggers the sex determination of undifferentiated gonads remains to be investigated.

In spite of the diverse and seemingly paradoxical expression patterns of dmrt1 found among different types of sex determination systems involved in temperature-dependent sex reversal, it is clear that dmrt1 plays an essential role in testis differentiation and in the temperature signal transduction pathway to the gonad at least in fish species with TSD, and that the up-regulation of dmrt1 expression is correlated with the temperature-induced male phenotype. Nevertheless, further studies are necessary to elucidate all the mechanisms that involve the dmrt1 gene.

The gene amh/mis

Anti-müllerian hormone (AMH), also known as müllerian-inhibitory substance (MIS), belongs to the transforming growth factor β (TGF-β) family and is secreted by the Sertoli cells. It is responsible for the regression of Müllerian ducts during male fetal development in mammals, birds, and reptiles [88–90] and is involved in both early sex determination and later gonadal development in higher vertebrate species [91]. Although teleost species do not have Müllerian ducts, dimorphic expression of amh was detected in developing and/or adult gonads and amh seems to play a role in gonadal differentiation and maintenance of both sexes. For example, in Japanese flounder [92], zebrafish [58, 93], Atlantic salmon [94] and Squalius alburnoides complex [95], amh is initially expressed at low levels in the undifferentiated gonads of both sexes and then at higher levels in the testis compared to the ovary during sex differentiation. In medaka, however, this is not the case since from hatching to adult, no sex-specific difference of amh expression was found [96].

In reptiles with TSD, amh has been studied in the red-eared slider turtle (T. scripta) and the American alligator (Alligator mississippiensis). In the red-eared slider turtle, expression analyses in both AKG (adrenal-kidney-gonad) complexes and isolated gonads showed that the levels of amh expression were higher at MPT than at FTP early in the bipotential gonad and throughout gonadogenesis [80, 97]. Moreover, amh expression decreased rapidly in gonads shifted from MPT to FPT but not in gonads shifted from FPT to MPT, which suggests that a testis-specific function is repressed and that expression of amh is modulated by temperature. In the American alligator, expression of amh was only detected in the developing testis of embryos incubated at MPT but not at FPT [57]. The dimorphic expression of amh in reptiles suggests that it is involved in temperature-dependent sex determination/differentiation although the mechanism remains unclear.

To our knowledge, the expression level of amh has been investigated in only one fish species with TSD, i.e. pejerrey (O. bonariensis) [76], in which it is much higher at MPT than at FTP and increases dramatically during gonad differentiation. In addition, a significant increase in amh expression is observed in putative males at MPT and at MixPT approximately one week before the first morphological signs of testis differentiation, which is similar to the expression profile in reptiles with TSD. However, the expression level of amh was comparable in putative males at MPT and at MixPT and in putative females at FPT and at MixPT, which indicates that, in this fish species, temperature does not modulate directly the expression of amh and thus that the expression of amh is the consequence rather than the cause of gonad sex differentiation. Thus, other gene(s) or sex inducer(s) are involved in the regulation of gonad sex differentiation in this fish species. Furthermore, it is assumed that amh expression is regulated by the level of 17β-estradiol and thus that cyp19a1 is involved in this regulation [76]. In zebrafish, Schulz et al. [82] reported that exposure to environmentally relevant concentrations of 17β-estradiol during early life suppressed both amh and dmrt1 expression, which was associated with cessation or retardation of testis development.

It is interesting to note that a Y-linked amh duplicated copy, termed amhy is supposed to be a master male determining gene in the Patagonian pejerrey (O. hatchery) [48], which is generally regarded as a species with TSD. The gene amhy was located on a single metacentric/submetacentric chromosome in XY individuals and was found to be expressed much earlier than the autosomal form of amh (6 days post-fertilization vs. 12 weeks post-fertilization). Furthermore, amhy knockdown in XY Patagonian pejerrey embryos results in the up-regulation of foxl2 and cyp19a1a mRNA and the development of ovaries. In the protandrous fish species, black porgy (Acanthopagrus schlegeli), expression patterns of amh during development and in the adult gonad indicate that amh plays important roles in early gonadal development in both sexes and later in ovary growth and natural sex change [98].

The amh gene appears to have lost part of its functions and to have acquired new ones, which are involved in sex determination and sex differentiation, during the vertebrate evolution from teleost fish to mammals. It will be interesting to compare the expression profile of amh in fish species with TSD and GSD, or fish species with different levels of TSD (varying levels of sex ratio response to temperature, e.g. Atlantic silverside) before and during sex differentiation and in different temperature conditions to address the adaptation of species with TSD. We expect that, in the near future, duplicated copies of other downstream genes involved in sex determination (e.g. amhy) will be identified and shown to have essential roles in this mechanism.

The sox9 gene

The gene sox9 is the direct target of the mammalian sex-determining factor SRY [99]. SOX9 and SRY belong to the same family of HMG-box containing transcription factors. SOX9 is a male determining factor both necessary and sufficient for testis formation in mammals. In mice [100, 101] and humans [102, 103], loss of sox9 results in male to female sex reversal, while transgenic XX mice that carry a copy of the sox9 gene develop into phenotypic males [104, 105]. Extensively investigated, the function of sox9 is known to be critical for many aspects of cell differentiation such as heart valve development, neural crest differentiation, chondrocyte specification and male sex determination in mammals [106–109]. Many of these functions have been demonstrated in various vertebrates, which suggests that the sox9 gene is conserved both structurally and functionally. In several reptiles with TSD, sox9 was shown to be expressed at higher levels in gonads at MPT than at FPT [110]. The comparison of expression profiles of sox9 at FPT, MPT, and temperature transfer (FPT to MPT) in reptiles indicates that sox9 may not be involved in the initial steps of sex determination as in mammals. However, a study on turtles, suggested that its function may be critical for final commitment to a testicular fate [30].

The cyp19a1a gene

Due to the important role of estrogen in development, growth and reproduction in teleost fish, aromatase, which is the key enzyme that catalyzes the formation of estrogen from androgen, has been extensively analyzed in the undifferentiated, differentiating, and differentiated gonads as well as in adult fish [124, 125]. The gene that encodes aromatase is a duplicated gene in all investigated teleost fish [126–129], except in the eel which belongs to the ancient group of Elopomorpha [130]. The gene duplication gave rise to two different genes (isoforms), namely cyp19a1a and cyp19a1b, in most teleost fish. The cyp19a1a gene is also known as “gonadal aromatase” or “ovarian aromatase” (also referred to as p450aromA, cyp19a or cyp19a1) since it is mainly expressed in the differentiating and adult gonad of teleost fish. The cyp19a1b gene is called the “neural aromatase” or “brain aromatase” (also referred as P450aromB, cyp19b or cyp19a2) since it is highly expressed in the brain of both male and female teleost species [131] but no sexually dimorphic brain expression during gonad sex differentiation has been demonstrated [132]. Our review of the literature is restricted to studies on the ovarian gene (cyp19a1a) with regard to sex differentiation.

Expression of cyp19a1a has been detected prior to sex differentiation in all fish species investigated and is associated with temperature-induced sex reversal (See Additional file 1: Table S1). Kitano et al. [133] have studied the Japanese flounder (Paralichthys olivaceus), in which sex is genetically determined with strong temperature effects since high temperatures induced an all-male population from an all-female population. They reported that, in this species, the expression of cyp19a1a mRNA was the same between males and females during the sex-undifferentiated period up to 50 DPH but that 10 days later when gonads start to differentiate, a specific expression was detected in the females. The level of cyp19a1a expression increased rapidly in the female group but decreased slowly in the male group. A subsequent study conducted by the same team on Japanese flounder showed that follicle-stimulating hormone (FSH) signaling and FOXL2 are involved in the transcriptional regulation of the cyp19a1 gene during gonad sex differentiation [134]. These results indicate that cyp19a1a is involved in temperature-induced sex reversal and plays an essential role in ovarian differentiation. However, it seems that cyp19a1a is a downstream gene in the sex differentiation pathway and thus, other upstream genes probably trigger the development of the undifferentiated gonads to ovaries or testes in natural sex differentiation or temperature-induced sex differentiation.

In Nile tilapia, in which high temperature causes 100% masculinization, the difference in cyp19a1a expression between sexes (higher in females than in males) was shown to exist before the histological differentiation of the ovary and to occur when germ cells enter meiosis [135]. Furthermore, the expression of cyp19a1a followed an interesting pattern: higher in all-female (27°C) than in all-female (35°C) than in all-male (27°C) than in all-male (35°C), which indicates that cyp19a1a is involved in the high-temperature induced masculinization (See Additional file 1: Table S1). Similar results were also observed in Atlantic Halibut (Hippoglossus hippoglossus) in which expression levels of cyp19a1a were lower before sex differentiation (13°C < 10°C < 7°C) in the high-temperature treatment group [136]. In addition, the expression of cyp19a1a increased in the low-temperature group (7°C) compared to the higher-temperature groups. However, in European sea bass (Dicentrarchus labrax), no significant difference in the expression of cyp19a1a was detected between MPT and FPT prior to and during sex differentiation (See Additional file 1: Table S1), although the proportion of males was 73% in the high-temperature group compared to 23% in the low-temperature groups [137, 138]. In Nile tilapia, foxl2 was found to strongly activate the transcription of cyp19a1a by binding to the sequence ACAAATA in the promoter region of the cyp19a1a gene directly through its forkhead domain in vivo and in vitro studies [139]. In a subsequent study by Navarro-Martín et al. [138] in the European sea bass, methylation of the cyp19a1a promoter activated by high-temperature resulted in a lower expression of cyp19a1a in temperature-masculinized fish, by preventing binding of foxl2 and sf-1 to their sites, and in turn, blocking transcriptional activation of cyp19a1a, which agreed with previous studies on foxl2 and sf-1[139–143]. Interestingly, in Nile tilapia, the suppressive effect of dmrt1 on cyp19a1a expression was confirmed in vivo and in vitro via the repression of the activity of Ad4BP/sf-1, which suggests that dmrt1 suppressed the female pathway by repressing the transcription of the aromatase gene and production of estrogen in the gonads [144].

In summary, in temperature-dependent sex determination of phenotypic males, temperature-induced methylation of the cyp19a1a promoter or temperature-induced high expression of dmrt1 are the cause of the down-regulation of cyp19a1a expression and the subsequent low level of estrogen. However, the status and relationship between dmrt1 and methylation of the cyp19a1a promoter are unclear and cyp19a1a is probably not the direct target of temperature. To our knowledge, in fish species, expression of cyp19a1a is generally thought to respond to temperature, namely, it is repressed with increasing temperatures (See Additional file 1: Table S1), which agrees well with the single sex ratio pattern dependent on temperature proposed by Ospina-Álvarez and Piferrer [4]. It could be argued that cyp19a1a suppression is the consequence rather than the cause of the suppression of female development. Nevertheless, cyp19a1a could be a good “indicator” to differentiate females from males prior to and/or during sex differentiation in some fish species. Undoubtedly, cyp19a1a plays an important role in sex differentiation in fish species.

The foxl2 gene

FOXL2 is a forkhead domain transcription factor, which is required for granulosa cell differentiation and ovarian maintenance [145]. Mutations of the foxl2 gene are responsible for blepharophimosis-ptosis-epicanthus inversus syndrome characterized by a distinctive eyelid abnormality and premature ovarian failure [146]. During the past decade, many studies have investigated the expression profiles of foxl2, its targets and its signaling pathway from mammals to teleosts and confirmed its pivotal role in the development and maintenance of female sexual characteristics.

Human FOXL2 and SF-1 (SF-1 is a steroidogenic factor-1, also known as Ad4BP and officially designated NR5A1) proteins interact in ovarian granulosa cells i.e. FOXL2 down-regulates the transcriptional activation of the steroidogenic enzyme, CYP17, through SF-1 [147]. Furthermore, patients with blepharophimosis-ptosis-epicanthus inversus syndrome type I present mutations in the foxl2 gene that result in the loss of the ability to suppress the induction of cyp17 mediated by SF-1. This demonstrates that mutations in the foxl2 gene are responsible for the disruption of normal ovarian follicle development. In goats, FOXL2 has been shown to be a direct transcriptional activator of the cyp19 gene via its ovarian-specific promoter 2 [148]. Goats with polled intersex syndrome in which the function of foxl2 is disrupted have a reduced expression of aromatase compared to control animals [149, 150]. The phenotype of foxl2 knockout mice comprises total absence of secondary follicles and oocyte atresia [145, 151] and mouse XX gonads without foxl2 develop into males [152], which suggests that foxl2 represses the male pathway during female gonadal development. For more details on the foxl2 function in ovarian development in mammals, refer to the review by Uhlenhaut and Treier [153].

In Japanese flounder, in which there is complete sex-reversal from males to females or from females to males by exposure to high (27°C) or low (18°C) temperatures during TSP respectively [133], expression of foxl2 in the gonads is suppressed at high temperature [134]. This indicates that foxl2 may also act as a female determinant in fish species with TSD.

The cortisol-mediated pathway

Hormones are considered as the primary communicators between external conditions and physiological activities because environmental information must first be transduced into a physiological signal to influence sex ratio [159]. As early as 1985, van den Hurk and van Oordt found [160] that, in rainbow trout larvae, exposure to cortisol and cortisone inhibited ovarian growth and increased the proportion of males. Cortisol is the major glucocorticoid produced by the interrenal cells and is used as an important indicator of stress since its production is increased by stressors such as handling, acid water and rapid temperature changes in fish [161]. Cortisol regulates a diverse array of systems including metabolism, ion regulation, growth and reproduction [162].

In recent years, several studies reported that exposure to high temperature elevated cortisol levels and led to the masculinization of fish species with TSD and GSD + TE. In 2010, Hayashi et al. [163] reported that, in medaka, exposure to a high-temperature (33°C) induced masculinization of XX females by elevating the cortisol level, which, in turn, suppressed germ cell proliferation and expression of fshr mRNA. Thus, cortisol can cause female-to-male sex reversal in this species. In Pejerrey, a fish species with TSD, individuals treated with cortisol presented elevated levels of 11-ketotestosterone (11-KT) and testosterone and typical molecular signatures of masculinization including up-regulation of amh expression and down-regulation of cyp19a1a expression [164]. Moreover, in the same species, it was observed that during high-temperature-induced masculinization, cortisol promoted the production of 11-KT by modulating the expression of hsd11b2. Cortisol also produces a dose-dependent sex reversal from females to males in the southern flounder (Paralichthys lethostigma) i.e. exposure to high (28°C) and low (18°C) temperatures produced a preponderance of males while an intermediate temperature (23°C) favored a 1:1 sex ratio [165]. In addition, in the Japanese flounder, exposure to cortisol caused masculinization by directly suppressing the expression of cyp19a1a mRNA by disrupting cAMP-mediated activation [166].

These results provide evidence on the relationships between temperature conditions and the responses of the organism and allow us to draw a picture of the endocrine-stress axis in terms of gonadal fate under temperature effects. They suggest that cortisol may be the “lost” link between temperature and the sex-determining mechanism in species with TSD and may, as a stressor indicator, be involved in the adaptive modification of sex ratio in a spatially and temporally variable environment during the evolution of such species. The relation between glucocorticoid production and androgen production during the masculinization process should be further investigated.

The epigenetic regulatory pathway

Epigenetics, a “hot spot” area of biology, is the focus of many studies. Epigenetics is defined as “the study of mitotically and/or meiotically heritable changes in gene function that cannot be explained by changes in DNA sequence” [167]. Epigenetic mechanisms that are involved in the regulation of gene expression typically include DNA methylation, which is relatively well-documented, modification of histones and histone variants, and presence of non-coding RNAs [168].

Gorelick [169] predicted that different methylation patterns of virtually identical sex chromosomes in species with TSD could be the result of small environmental changes (e.g. temperature variation), thereby determining the sex of each individual. He also hypothesized that homomorphic sex chromosomes are required to explain the origin of TSD; sex differences are initially determined by the different methylation patterns of nuclear DNA in females and males, which result in different sexual phenotypes of TSD. Based on an analysis on European sea bass, a fish with a polygenic system of sex determination in which genetics and temperature contribute almost equally to the sexual phenotype [170, 171], Navarro-Martín et al. [138] reported that the methylation levels of the promoter of the gonadal cyp19a1 gene are sex-specific and influenced by the temperature at which one-year-old juveniles are reared. The inverse relationship between methylation levels and cyp19a1 expression indicates that temperature-induced masculinization (high-temperature) involves DNA methylation-mediated control of cyp19a1. These findings are the first example of an epigenetic mechanism that regulates temperature effects on sex ratios in a vertebrate. Recently, similar but slightly different results were also found in red-eared slider turtle, a species with clear TSD [172]. DNA methylation levels of the promoter region of the cyp19a1 gene were significantly higher in embryonic gonads at the male-producing temperature (26°C) than at the female-producing temperature (31°C). Nevertheless, a switch from male-to-female rather than from female-to-male producing temperature during TSP significantly decreased the level of DNA methylation of the cyp19a promoter region in the gonads. These results indicate that temperature, specifically female-producing temperature, causes demethylation of the cyp19a promoter region which, in turn, leads to the temperature-specific expression of cyp19a and favors the female pathway. Moreover, in the protandrous fish species, black porgy (Acanthopagrus schlegeli), it has been reported that the methylation level of the cyp19a1 promoter was higher in inactive ovaries than in active ones, which suggests that the expression of cyp19a1 is controlled by an epigenetic mechanism in addition to the classical transcriptional activators of cyp19a1 such as SF-1 and FOXL2 [173].

Based on these studies, it is tempting to speculate that methylation of the cyp19a promoter may participate in the mechanism that links environmental temperature and sex ratios in vertebrate species with TSD but future research is necessary to support this hypothesis. In 2013, Piferrer published an extensive review on the epigenetic regulation of sex determination and gonadogenesis [174]. It would be very interesting to investigate whether the methylation patterns of related genes are conserved in vertebrates with different sex-determining mechanisms including GSD, TSD, as well as hermaphrodite species. We believe that the epigenetic regulation of sex determination and differentiation should be extensively studied in many vertebrates and that such studies would provide new insights on our present understanding of the origin, evolution, and maintenance of sex-determining mechanisms.