Correspondence to: Jennifer A. Marshall Graves, La Trobe Institute of Molecular Science, La Trobe University, Bundoora Vic 3186, Australian National University, Melbourne, VIC 3186, Australia. E-mail: email@example.com
Sex determination is a trait that reveals general principles of gene evolution, but offers the advantage that evolution is rapid and quixotic, and that enormous variability can be found even within closely related taxa.
The discovery of the gene that initiates testis determine in humans and other placental mammals (Sinclair et al., 1990) ignited research on many fronts. First, attempts were made to understand the pathway that SRY triggers, then a search for generalities was undertaken, as well as intensive work on the evolution of sex chromosomes and sex determining genes.
The first task proved much more difficult than expected but is now well in hand (Kashimada and Koopman, 2010). The second drew a complete blank, for there is no SRY in other vertebrates, or even in prototherian mammals like the platypus (Wallis et al., 2007). Instead, a completely unrelated gene DMRT1 appears to have this role in birds, a quite different (and as yet unknown) gene has it in snakes, and other yet unknown genes in lizards and turtles, frogs, and fish. In salmon, an unrelated transcription factor has this function, and in some fish and amphibians, copies of SOX genes or DMRT1. In many reptiles and a few fish, there is no sex determining gene; instead, sex is determined by an environmental stimulus, usually temperature.
How can these completely unrelated stimuli initiate essentially the same pathway? And how did it happen that different sex determining genes have been chosen for a sex determining function in different lineages? It becomes necessary to understand how sex determining genes, and sex chromosomes, arise in different lineages.
SRY AND THE SEX DETERMINING PATHWAY
SRY (named for the Sex Region on the Y chromosome) is an odd candidate for one of the most critical functions that ensures an organism's survival, being poorly conserved, even between quite closely related mammals such as human and mouse (Kashimada and Koopman, 2010). Indeed, the mouse version of the gene codes for a 3′ polyglutamate tail that is completely lacking in nonrodents.
It has not been easy to define exactly what SRY does. The single exon of SRY codes for a region (the HMG box) of 81 amino acids that binds DNA at short consensus sites and bends it through a specific angle. This must be the business end of the gene, because mutations in this region that affect binding or bending cause sex reversal (Hawkins, 1993). SRY defined a whole family of SOX genes that share the HMG box and appear to have important functions in development. The SOX subgroup of genes to which it is most similar, are also single exon, but have long transactivating 3′ and 5′ regions, and are highly conserved between all vertebrates.
In humans and other animals, several genes up and downstream of SRY have been discovered by probing sex reversed human patients, mouse models, and nonmodel animals like goat and dog. The first critical gene in the pathway was discovered by positionally cloning an autosomal gene, haploinsufficiency for which causes the bone disease campomelic dysplasia which is often accompanied by sex reversal. This resulted in the discovery of an SRY relative, SOX9 (Foster et al., 1994; Wagner et al., 1994). This gene is expressed soon after SRY in the genital ridge in all mammals, and is the first gene to show sex specific expression in other vertebrates such as reptiles, amphibians and fish (Morais da Silva et al., 1996).
SRY interacts with other factors to ensure that SOX9 reaches a threshold level sufficient to induce testis (Kashimada and Koopman, 2010). Several other genes that either promote or inhibit testis formation downstream of SRY have since been discovered, and we now understand gonad differentiation as the product of a balance between these promoting and inhibiting forces (di Napoli and Capel, 2008). Sex reversing genes in human and mice include the autosomal DMRT1 that is required to maintain testis differentiation in the embryo and suppress female promoting genes in the adult, the X-linked ATRX, and female-promoting genes such as FOXL2, WNT4, β-catenin, and RSPO1.
EVOLUTION OF SRY AND THE HUMAN Y CHROMOSOME
The SOX gene that most closely relates to SRY, SOX3, was found to lie on the X chromosome, and it was proposed that SRY evolved by truncation of this gene (Foster and Graves, 1994). Given that SOX3 is expressed largely in the central nervous system and in germcells (but not somatic cells of the testis), it was not immediately clear how it acquired its sex determining function. However, it was recently shown that mis-expression in the somatic cells of the testis can induce testis in XX male patients and transgenic mice (Sutton et al., 2011), suggesting that ectopic expression, perhaps by fusion of the gene with control sequences from another gene on the X (Sato et al., 2010), gave SRY its start (Fig. 1).
Once SRY became male-determining, it defined a male-specific version of the X that differentiated rapidly from its partner as more and more genes near SRY took on a male-advantage function and recombination with the X was suppressed to keep this male package together (reviewed Graves, 2006).
The birth of SRY can be easily dated by following its presence in different mammal groups. It is present in all placental mammals except a few rodent groups that have secondarily lost the Y. It is also present in marsupial mammals, which last shared a common ancestor with placental mammals 148 MYA. However, there is no direct evidence that SRY is male-determining in this group, and in fact there is a rival in ATRY, a Y-borne paralogue of the sex-reversing gene ATRX on the X (Pask et al., 2000). The real surprise was the discovery that the complex XY sex chromosome system of the most ancient mammalian offshoot, the monotremes (including platypus and echidna) has homology, not to the therian mammal Y, but to the ZW system of birds (Veyrunes et al., 2008). SRY was, therefore, born only 148-166 MYA (Graves, 2006) (Fig. 2).
SEX DETERMINING GENES AND SEX CHROMOSOMES IN OTHER VERTEBRATES; CHOOSING THE RAW MATERIAL
Other vertebrates do not necessarily share the mammal male-dominant mode of sex determination, in which a gene on a male-specific Y chromosome initiates testis development on a background shared by the female. Either the male, or the female may be the heterogametic sex (that is the sex that produces two types of gametes). Equally possible is a female-dominant gene on a female-specific W chromosome that initiates ovary development (Fig. 3). An even more common way to determine sex is by control of the dosage of a gene; in female heterogametic species with a ZZ male: ZW female sex chromosome constitution, the product of a gene present on the Z but not the W may initiate testis at a threshold concentration. The opposite can occur in a male heterogametic species if a gene on the X but not the Y promotes ovary development.
Outside of mammals, positional cloning of sex genes has proved difficult, even in lineages with extremely conserved sex chromosomes. Birds have a ZZ male: ZW female system in which the Z is very conserved (Shetty et al., 1999) and the W, like the mammal Y, is a degraded version of the Z, being a small relic in carinate birds, but largely undifferentiated in the flightless ratites. The ZW pair shares no homology with the human XY system, but shows homology with human chromosomes 9 and 5 (Nanda et al., 1999). Sex determination has been long suspected to depend on dosage of a Z-borne gene, and DMRT1 was cloned from the chicken Z many years ago was suspected to have this function (Smith et al., 1999: Raymond et al., 1999), a proposal consistent with the presence of this gene in the small differentiated region of the Z in emus (Shetty et al., 2002) and the sex reversing effects of a deletion of human chromosome 9 carrying the DMRT1 gene (Ferguson-Smith, 2007). Knocking down DMRT1 in chicken embryos was recently shown to produce XX females (Smith et al., 2009); however, there must be another cell autonomous factor up or downstream of this gene (Zhao et al., 2010).
Snakes, too, possess a ZZ male ZW female system, in which the Z is very conserved but the W shows different degrees of differentiation in different snake families (this is the data that Ohno used to formulate the theory that the sex-specific member degrades over evolutionary time; Ohno, 1967). Although the Z and W chromosomes are superficially similar to bird sex chromosomes, gene mapping shows that the snake Z has homology to the bird chromosome 2 and vice versa, suggesting a rearrangement in a common ancestor (Matsubara et al., 2006). We do not yet know which gene is sex determining in snakes, but it is not DMRT1.
Nor is it DMRT1 in a ZW gekko lizard, in which DMRT1 maps to an autosome (Trifonov et al., 2011). A phylogeny comprising 27 related agamid lizards contains species having sex chromosomes with different markers, as well as species with temperature dependent sex (Ezaz et al., 2009). Frogs, too, have diverse sex chrimosomes that do not include DMRT1 (Uno et al., 2008).
What was the ancestral sex determining system in the common ancestor of mammals and reptiles 310 MYA? The finding that in one gekko lizard, the Z is identical to the bird Z (Kawai et al., 2009), and the monotreme XY complex has homology to the bird ZW (Veyrunes et al., 2008) initially suggested that this system was ancestral to all amniotes (Graves, 2006). However, it seems unlikely that DMRT1 is the sex determining gene in gekkos, because both Z and W possess this gene, and it is autosomal in related species (Pokorná et al., 2011); and in monotremes, the dosage is all wrong because females have two copies and males only one. It could be, therefore, that gekkos, birds, and monotremes chose the same genomic region to act as sex chromosome because it contains gene(s) that are good at this function. This proposal receives support from an analysis of fish sex chromosomes (Graves and Peichel, 2010), which show that particular ancestral regions, inferred by homologies between synteny groups, are sex linked in a variety of species. For instance, chromosome 3 of the teleost protokaryotype appears as an XY chromosome in the ninspine stickleback and tiger pufferfish species, and as a ZW chromosome in Lake Malawi cichlids and a medaka species.
GENE COPIES AS SEX DETERMINERS
Of special interest is evidence that two of the known sex determining genes have spawned copies that have independently taken over sex determination in fish and frogs. A stunning finding (reviewed Kondo et al., 2009) was that one species of medaka fish (Oryzias latipes) and its closest relatives have a newly minted Y chromosome defined by a copy of DMRT1. This male-dominant DMY gene (but not its autosomal ancestor DMRT1) is expressed in testis under control of a transposed promotor (Herpin et al., 2010) to promote testis development in XY males. This neo-Y is not shared even by quite closely related species, so that the event that transposed it to its new position occurred only approximately 10 million years ago.
Recently, it was discovered that the clawed toad Xenopus laevis has a ZW system in which the W is female-dominant; remarkably the gene that defines this neo-W is also a copy of DMRT1 (Yoshimoto et al., 2008). This DMW has evidently been degraded into a competitive inhibitor of DMRT1 (Yoshimoto et al., 2010, Mawaribuchi et al., 2011). These copies of DMRT1 both appear to influence DMRT1 dosage; DMY on the medaka Y augmenting it and DMW on the toad W diminishing it. Thus, new sex determining genes may arise from a copy of a gene that acts by differential dosage (Fig. 4).
It seems that SOX3, too, has spawned three independent sex chromosome systems. This gene is known to be expressed downstream in the ovary in frogs (Koyano et al., 1997), but is not sex-determining. However, it may be the sex switch in the Japanese wrinkled frog, Rana rugosa, which has a ZW system on one side of an island in Japan, and an XY system on another. Remarkably, the Z and X chromosomes are identical, and have homology to the human X, including the SOX3 gene. Mapping and characterizing SOX3 alleles revealed a Y-specific allele (Miura et al., 2009). Very recently, too, another medaka species O. dancera, has been shown to have XY chromosomes that share homology with the human X and include SOX3 (Y. Takehana personal communication). It will be interesting to discover whether these Y-specific SOX3 alleles have, like mammalian SRY, acquired expression in the somatic cells of the testis, providing independent examples of how new sex determining genes may arise as autosomal genes change the tissue of expression (Fig. 5).
It is difficult to construct a phylogenetic tree in which DMRT1 and SOX3 are both sex determining in a common ancestor; rather, it seems that these two genes are just particularly good raw material for making new sex determining genes.
OTHER SEX DETERMINING GENES: CAN ANY GENE BE SEX DETERMINING?
Recently new fish sex determining genes have been characterized that are unrelated to DMRT1 or SOX3. Two are involved in sex in other species, but the third seems to be a complete ring-in.
A novel gene was recently discovered that determines sex in another medaka species, O. luzonensis. GsdfY (Gonadal Soma Derived growth Factor on the Y chromosome) is highly expressed in males during sex determination, and transgenesis produces XX males (Myosho et al., 2012). The sex determining region is not the gene itself but an upstream promotor that overexpresses the gene specifically in the genital ridge in males, suggesting another way that novel sex determining genes may be created (Fig. 5). This gene is involved with sex determination in O. latipes, acting downstream of DMY in the sex determining cascade. It seems to have replaced it only recently, because it is nested within medaka species with DMY. Very recently, sex specific polymorphisms have been found in fugu species in the anti-Mullerian hormone receptor type II (Amhr2) gene (Kamiya et al., 2012), and a duplication of this gene in the Patagonian pejerry (Hattori et al., 2012). AMH is downstream of SOX9 in mammals, but expressed earlier in alligators (Western et al., 1999).
In the brown trout, a novel gene sdY colocalizes with sex determination: targeted disruption produces XY female sex reversal, and over-expression causes XX testis development (Yano et al., 2012). Unlike the other genes so far characterized, Sdy has no homology with any known gene in the sex differentiating pathway. Rather, it has homology with interferon regulatory factor 9, an immune regulatory gene not known to be involved in sex determination in any lineage. Does this just reflect our ignorance of pathways involved in sex determination? Or does it mean that unrelated genes can acquire a sex determining function by, for instance changing their binding domain or target sequence?
Many different genes may be dragooned into a sex determining role, either by increasing or decreasing the gene dosage (e.g., DMRT1 copies DMY and DMW), increasing the level of expression (e.g., GsdfY) or changing tissue specificity (e.g., SRY). Some of these genes have been recruited from ancestral sequences that already have a function in sexual differentiation (e.g., fish and frog DMRT1, fish Amhr2), others are related to genes in the pathway (e.g., mammalian SOX3); still others do not, as far as we know, have a function in sexual differentiation (e.g., trout sdY). This will not be news to those studying sex determination in insects, in which there is a bewildering variety of genetic triggers (Gempe and Beye, 2011).
Is the repeated discovery of SOX3 and DMRT1-related genes involved in sex determination just an artifact of our search by homology? Or do these genes make particularly good raw material (O'Meally et al., 2012)? It will be important to carry out unbiased searches in species with nonhomologous sex chromosomes, such as snakes and other lizards.
A big leap forward is likely to be made by whole genome sequencing, and comparison of sequence between of males and females of the same species (Genome 10K). This will be especially valuable for species with cryptic, or barely differentiated sex chromosomes. For instance, in boid snakes and ratite birds, Z and W chromosomes are almost identical, and a sex-specific region will be obvious from dosage differences. The sex gene will find it hard to hide in the tiny region that is absent from the Z and differentiated on the W.
I predict that many vertebrate sex determining genes will be discovered by such unbiased searches in the next few years. This will make it easier to detect patterns in the evolution of new sex determining genes, and answer some of the questions raised in this review.