Are homologies in vertebrate sex determination due to shared ancestry or to limited options?

The same candidate genes and the same autosomes are repeatedly used as sex chromosomes in vertebrates. Are these systems identical by descent, or are some genes or chromosomes intrinsically better at triggering the first steps of sex determination?

Review

In all but a few vertebrates there are two sexes; females make large sessile gametes (eggs) and males make small motile gametes (sperm). Yet, despite the commonality of the endpoint, there seem to be endless ways in which the sex of an animal is determined. This is not because of differences in the program of cell differentiation, which is almost unchanged throughout vertebrates, and is governed by much the same team of genes. The chaotic variation is generated almost entirely by the diversity of factors that trigger this sex-determining pathway.

In many vertebrate lineages, the trigger for the sex-determining pathway is a gene or genes, and this type of sex determination is termed genetic sex determination (GSD). In others, an environmental cue, usually temperature, is the trigger, and this mode is called environmental (temperature) determined sex (TSD). GSD may be controlled by a male-inducing factor that defines a Y chromosome in an XX female:XY male system of male heterogamety (as in humans and other mammals), or a gene that defines a female-specific W chromosome in a ZW female:ZZ male system of female heterogamety (as in birds). In some systems, the two sex chromosomes are almost identical, differing at only one or a few loci. In others, members of the pair are highly differentiated, with a large gene-rich X (or Z) and a Y (or W) with much repetitive sequence but few active genes.

Yet among this hegemony, homologous sex chromosomes, and homologous sex-determining genes, reappear again and again. One explanation for this could be that sex-determining genes that are ancestral to all vertebrates re-emerge in different lineages. An alternative hypothesis could be that there are only a few genes, on a few chromosomes, that are suited for the task of sex determination, and evolution keeps rediscovering them. To distinguish between these hypotheses, we will discuss the conserved sex-determining pathway, then the genetic triggers that activate it and the chromosomes they define, before marshalling the evidence in favor of each hypothesis. We will start by briefly describing the well-known mammalian XY system of sex determination, and compare it with the stable ZW systems in snakes and birds, and the plethora of systems in other reptiles, amphibians and fish.

Vertebrate sex-chromosome systems and their homologies

Sex determination in humans and other mammals is accomplished by a conserved XY male:XX female system. Figure 1 shows the sex-determination systems that can be found in the main vertebrate groups. All placental mammals share an almost identical large, gene-rich X chromosome, and have a degenerate Y chromosome that contains the male sex-determining gene SRY and paralogs of a few X-borne genes [1]. Marsupial mammals have a smaller X that is homologous to only part of the placental mammal X, and the SRY-containing marsupial Y is minute [1]. Monotreme mammals (platypus and echidna) have a bizarre complex of 5X and 5Y chromosomes [2], which have homology not to the XY of therian mammals (marsupial and placental mammals) but to the bird Z chromosome [3].

Figure 1

Phylogeny and sex-determination systems in vertebrates. (a) Tetrapods; (b) teleost fishes. The different sex-determination systems found in the different lineages are indicated on the right of the figure. For the fish lineages (b), homologies of identified sex chromosomes to the ancestral teleost karyotype (TEL) are shown [67]. See Table 2 for details.

Sex determination in birds and snakes is also accomplished by means of sex chromosomes, but they are completely the converse of the mammalian system. Males have two copies of a gene-rich Z chromosome, and females a single copy of the Z and a small heterochromatic W chromosome. The Z has been shown by chromosome painting and gene mapping to be identical in all bird species, but is not homologous with the mammalian X [4]. The W is a degenerate version of the Z, and ranges from highly conserved in flightless ratite birds (for example, the emu) to highly differentiated in carinate birds (for example, the chicken) [4]. Snakes also have a ZZ male:ZW female system of chromosomal sex determination. As in birds, the Z chromosome is highly conserved between all snakes, whereas the W shows various degrees of degeneration [4]. The snake ZW pair superficially resembles the bird ZW pair, but comparative gene mapping shows that they share no genetic homology [5,6].

Other reptile lineages show a variety of sex-determining systems, including XY, ZW and TSD (Figure 1). Different systems are often found in the same clade [4,7] or even in the same species [8]. For example, the dragon lizard Pogona vitticeps has a micro-ZW system with no homology to either the snake or the bird ZW sex chromosomes, or to the mammalian XY [9]. Remarkably, though, a gekko lizard (Gekko hokunensis) with female heterogamety was recently discovered to have a Z chromosome with the same genes (including DMRT1, the bird sex-determining gene) in the same order as on the bird Z [10].

In both amphibians and fish, GSD is common (Figure 1). Morphologically distinguishable sex chromosomes are relatively rare [11,12], but studying a wider variety of species with more sophisticated cytogenetic tools, such as comparative genome hybridization, may reveal cryptic morphological differences (heteromorphy) between sex chromosomes [13,14]. Even very closely related species of fish or frogs, and even different populations within the same species, can have different sex-determination mechanisms or non-homologous sex chromosomes, as evidenced by the presence of both XY and ZW sex-chromosome systems within cichlids, sticklebacks and medaka fishes (Oryzias) [15-20]. In the wrinkled frog Rana rugosa, XY and ZW populations inhabit different Japanese islands, and there is a hybrid zone with every mixture of sex chromosomes [21,22]. Remarkably, the ZW pair of the female heterogametic population is homologous to the XY pair in the male heterogametic population [23].

Thus, outside the well-known stable mammal XY and bird and snake ZW systems, the overall picture of vertebrate sex chromosomes is one of bewildering variety (Figure 1).

The conserved sex-determination pathway

At the histological level, however, vertebrate sex determination is highly conserved. A ridge of cells on the embryonic kidney (the genital ridge) differentiates into either testis (obvious from the presence of testis cords) or ovary (obvious by the presence of large follicles surrounding developing eggs). The differentiation pathways are shown in Figure 2. The same sets of genes appear to control these alternative processes in all vertebrates, although the timing of some steps and the tissues in which the genes are expressed may differ.

Figure 2

The antagonistic relationship between the pathways determining testis or ovary development in mammals. The male-determining pathway is shown on the left and the female pathway on the right. In males, SRY on the Y chromosome is activated by various upstream factors (small green arrows), including SF1, and upregulates the autosomal gene SOX9, which then maintains its own activity with the help of several other factors (small green arrows), including FGF9, whose expression is in turn upregulated by SOX9. FGF9 induces a cascade of downstream steps that is controlled by several genes, including DMRT1. These steps culminate in the differentiation of Sertoli cells, which are crucial to the development of the gonad as a testis. SOX9 and FGF9 repress RSPO1 and WNT4, which promote the development of the gonad as an ovary. In females, RSPO1 upregulates WNT4, which is accompanied by the stabilization of β-catenin, which is probably the mediator by which RSPO1 and WNT4 repress SOX9 and FGF9. Activation steps are represented by green arrows and inhibition by red bars.

To determine sex, the bipotential gonad must choose between the male or female pathway. This choice is governed by sets of genes that act to suppress each other, reinforcing one or the other cell fate [24]. In all vertebrates (TSD as well as GSD species), the autosomal gene SOX9 seems to have a pivotal role in early testis determination. SOX9 is present in both sexes but, in one of the first molecular events in the male pathway, it is upregulated in Sertoli cell precursors (which are essential for testis differentiation) by the primary sex-determining trigger interacting with an evolutionarily conserved sequence within SOX9 [25]. In mammals, a deficiency of SOX9 in XY animals produces female development, and SOX9 duplication causes male development in the absence of a Y chromosome [26].

In mammals, SOX9 expression in males is maintained by positive feedback autoregulation, and by the action of the growth factor FGF9 (FGF9 is itself upregulated by SOX9) and several other secreted signals (Figure 2; see Table 1 for names and descriptions of the genes mentioned in the text) [27]. In males, the actions of SOX9 and FGF9 repress the transcriptional regulator β-catenin and the secreted intercellular signal WNT4, which promote the development of ovaries in females [24,27]. On the other hand, in females, the secreted protein R-spondin 1 (RSPO1) acts together with WNT4 to stabilize β-catenin, which then represses SOX9 and FGF9 [27]. Thus, there is an antagonistic relationship between testis and ovary pathways.

Table 1

Genetic loci involved in sex determination in vertebrates

Locus	Gene name	Type of protein product
SRY	Sex-determining region on the Y chromosome	HMG-box transcription factor
SOX3	SRY-like, HMG-box-containing gene family, member 3	HMG-box transcription factor
SOX9	SRY-like, HMG-box-containing gene family, member 9	HMG-box transcription factor
SF1	Steroidogenic factor 1	Transcription factor of the steroid receptor family
FGF9	Fibroblast growth factor 9.	Secreted intercellular signal
WNT4	Wingless-type MMTV integration site family, member 4	Secreted intercellular signal
RSPO1	R-spondin-1	Secreted intercellular signal
DMRT1	Doublesex and mab-3 related transcription factor 1	Transcription factor
DMY	DMRT1 homolog on the Y in Oryzias latipes	Transcription factor
DM-W	DMRT1 homolog on the W in Xenopus laevis	Transcription factor
DGCR8	DiGeorge syndrome critical region, gene 8	microRNA biogenesis protein
RBMY	RNA-binding motif protein, Y-linked	RNA binding protein
TSPY	Testis-specific protein, Y-linked	Growth promoting factor and candidate gonadoblastoma gene.

A single gene - SRY - sets off the mammalian testis-determining cascade. SRY was discovered on the human Y chromosome [1] and encodes a transcription factor that briefly upregulates SOX9, which then maintains its own expression as described above. SRY is mammal-specific but is the defining member of the large SOX gene family, members of which share the HMG-box motif that binds to DNA and bends it at a specific angle, changing chromatin structure and permitting transcription. There seems to be no comparable 'ovary-determining factor'; instead, development of the ovary is induced if the level of SOX9 is insufficient to suppress the female-promoting genes RSPO1 and WNT4 [24,27].

Another gene involved in vertebrate sex determination is that encoding the transcription factor DMRT1 (doublesex and mab-related transcription factor 1). DMRT1 is the vertebrate-specific homolog of the genes doublesex in Drosophila melanogaster and mab-3 in Caenorhabditis elegans (from whence it derives its name), which are involved in the downstream events of sex differentiation, rather than the initial sex-determination switch, in invertebrates [28,29]. The transcription factors encoded by these genes all share a zinc finger-like DM domain that binds DNA and regulates transcription [30,31]. Across mammals, birds, reptiles, amphibians and fish, DMRT1 is expressed specifically in male gonads just after sex determination [28,29]. Knockdown experiments in chickens recently confirmed that DMRT1 is the bird sex-determining gene [32]. It lies on the Z chromosome and has no allele on the heterochromatic W: the double dose of DMRT1 is needed to form a testis in ZZ male birds, whereas the single dose in ZW females is insufficient. How DMRT1 dosage could have a cell-autonomous effect in ZZ/ZW chimeras, as determined by Zhao et al. [33] is not clear. In humans, DMRT1 is not the sex-determining trigger, but lies downstream in the sex-determining pathway. It maps to a region of chromosome 9p that is deleted in cases of XY testicular dysgenesis, and Dmrt1 loss-of-function mutations in mice result in defects in testis development [28,29]. Thus, DMRT1 appears to be a critical gene near the top of the sex-determination cascade in both vertebrates and invertebrates.

In addition to SRY and DMRT1, only two other vertebrate sex-determining genes are known, and remarkably they are both homologs of DMRT1. In the Japanese medaka fish (Oryzias latipes), which has an XY system, a duplicated copy of DMRT1 (DMY) defines a novel Y chromosome. This novel Y chromosome is genetically the same as the X chromosome, with the addition of around 258 kb of sequence that includes the DMY gene. DMY encodes a fully functional DMRT1-type protein and has been shown to be necessary and sufficient to turn on male development [34-37]. The only known amphibian sex-determination gene, DM-W, is a truncated copy of DMRT1 on the W chromosome in the frog Xenopus laevis. ZZ tadpoles transgenic for DM-W were feminized, implying that DM-W acts as a dominant-negative, antagonizing DMRT1 function and repressing testis development in ZW frogs [38].

Thus, the vertebrate sex-determining pathway is extremely conserved at the molecular as well as the physiological level, but many different factors may trigger it. It has been proposed that evolutionary stability at the bottom of the sex-determination hierarchy is coupled with lability at the top [39], and the molecular and physiological conservation described above fits this notion very well. Few vertebrate sex-determining genes have been identified: SRY in mammals, and DMRT1 in birds and its homologs in fish and frogs. We can deduce the deep evolutionary history of these two genes by comparing them in different vertebrate lineages in order to gain insights into the evolution of sex determination.

Evolution of sex-determining genes

When SRY was identified it was initially assumed to be unique to the Y chromosome. A search for SRY in kangaroos, however, identified a homolog on the X chromosome, termed SOX3. The sequence of the HMG-box in SOX3 most closely resembled that of SRY, so it was suggested that SOX3 was the ancestor of SRY [40]. Most other genes on the Y (for example, RBMY, TSPY), including several with male-specific roles in spermatogenesis, were subsequently found to have homologs on the X from which they had obviously evolved [1].

How did SOX3 evolve into the testis-determining SRY? SOX3 is expressed strongly in the gonads as well as in the central nervous system, at least in mouse and humans, but its deletion or duplication in human males affects fertility and intelligence rather than sex determination [41]. Its sequence similarity to SOX9 initially prompted the suggestion that SOX3 in a therian ancestor was originally a dosage-regulated inhibitor of SOX9 in females, and so a null mutant of SOX3 could have permitted activation of SOX9 and male development. Truncation of the null SOX3 allele was suggested to have subsequently turned it into an inhibitor of normal SOX3, and thus an activator of SOX9 [42]. However, it now seems more likely that SRY interacts with steroidogenic factor 1 (SF1) to activate SOX9 directly [43], suggesting that acquisition of a testis-determining function was due to a changed SOX3 expression pattern or to its association with different binding partners. A recent bioinformatics search of protein databases using untranslated regions of SRY hints that SRY evolved from fusion of the HMG-box of SOX3 with another X-linked gene, DGCR8 [44]. SOX3 is also found on the sex chromosomes of two distantly related non-mammalian vertebrates, the frog Rana rugosa [23] and the fish Oryzias dancena (Table 2) [45]. However, it has not yet been shown to be sex-determining in either species, and is not known to be sex-linked in any other vertebrate species. Mouse knockouts indicate that Sox3 is involved in spermatogenesis and it is a developmental regulator like Sox1 and Sox2, specifying neuronal fate in fish and mammals.

Table 2

Known homologies of sex chromosomes in teleost fish

Species	Linkage group (LG)	Genes	Gac	Tni	Ola	Dre	TEL
Threespine sticklebackGasterosteus aculeatus	XX/XY - LG 19	CYP19BWT1A	LG 19LG 19	LG 13Sc14539	LG 6LG 6	LG 25LG 25	7
Ninespine sticklebackPungitius pungitius	XX/XY - LG 12	PAX7MITFB	LG 12LG 12	LG 11LG 9	LG 5LG 7	LG11LG 23	3
Nile tilapiaOreochromis niloticus	XX/XY - LG 1	CYP19AWT1B	LG 2LG 2	LG 5LG 5	LG 3LG 3	LG 18LG 18	7
Spotted tilapiaTilapia mariae	ZZ/ZW - LG 3	TRP1/TYRP1ADMO/DMRT4	LG 7LG 7	Sc14681Sc7577	LG 18LG 18	LG 7NF	?
Lake Malawi cichlidsnon-OB phenotype	XX/XY - LG 7	IGF2WT1A	LG 19LG 19	LG 13Sc14539	Sc1060LG 6	LG 25LG 25	7
Lake Malawi cichlidsOB phenotype	ZZ/ZW - LG 5	PAX7Opsins	LG 12LG 17	LG 11LG 11	LG 5LG 5	LG 11LG 6,11	3
Tiger pufferfishTakifugu rubripes	XX/XY - LG 19	overallAMHR2	LG 17LG 17	LG 11LG 11	LG 5LG 7	NANF	3
Japanese medakaOryzias latipes andOryzias curvinotus	XX/XY - LG 1	TYRP1BLEF1	LG 9LG 9	Sc13631LG 18	LG 1LG 1	LG 1LG 1	4
Oryzias luzonensis	XX/XY - LG 12	SLC45A2(b locus)	LG 14	LG 4	LG 12	LG 21	6
Oryzias mekongensis	XX/XY - LG 2	XDHPOMC	LG 1	LG 3	LG 2	NF	1
Oryzias minutillus	XX/XY - LG 8	HOXBASOX9B	LG 11LG 11	Sc14653LG 3	LG 8LG 8	LG 3LG 3	1
Oryzias dancena	XX/XY - LG10	SOX3FGF9	LG 4LG 4	LG 1LG 1	LG 10LG 10	LG 14LG 14	5
Oryzias hubbsi	ZZ/ZW - LG 5	OpsinsWNT4A	LG 17LG 17	LG 11LG 11	LG 5LG 5	LG 6,11LG 11	3
Oryzias javanicus	ZZ/ZW - LG 16	HOXABWNT4BRSPO1	LG 10NFLG 20	LG 8UNNF	LG 16LG 16LG 16	LG 16LG 16LG 16	8
GuppyPoecilia reticulata	XX/XY - LG 12	SLC45A2	LG 14	LG 4	LG 12	LG 21	6
PlatyfishXiphophorus maculatus	XX/XY - LG 24	MC4R	LG 21	Sc14565	LG 20	LG 2	12
		DMRT1	LG 13	LG 12	LG 9	LG 5	6

The positions of DMRT1 and various sex-linked genes in teleost fish species were identified by BLAT searches of the Gasterosteus aculeatus (Gac), Tetraodon nigrividis (Tni), Oryzias latipes (Ola), and Danio rerio (Dre) genomes in the Ensembl genome browser release 56. The positions of genes are indicated by linkage groups (LG), unless the gene was not found in a genome assembly (NF), in an unassembled region of the genome (UN), or on a scaffold (Sc) that has not yet been assigned to a linkage group. The homologies between different fish sex chromosomes can be seen in the column indicating its ancestral teleost protokaryotype (TEL), which was inferred on the basis of the assignment of sex-linked genes to the Tni, Ola, and Dre linkage groups [67]. AMHR2, anti-Müllerian hormone receptor, type II; CYP19A, cytochrome P450, family 19, paralogous gene a; CYP19B, cytochrome P450, family 19, paralogous gene b; DMRT4, doublesex and mab-3 related transcription factor 4; HOXAB, Hox cluster A, paralogous subgroup B; HOXBA, Hox cluster B, paralogous subgroup A; IGF2, insulin-related growth factor 2; LEF1, lymphoid enhancer-binding factor 1; MITFB, microphthalmia-associated transcription factor b; PAX7, paired box 7; POMC, proopiomelanocortin; MC4R, melanocortin 4 receptor; SLC45A2, solute carrier family 45, member 2; TYRP1A, tyrosinase-related protein 1; WT1A, Wilms tumor 1a; WT1B, Wilms tumor 1b; XDH, xanthine dehydrogenase.

DMRT1 is even more ubiquitous, being the sex-determination gene in birds, and the source of new sex-determination genes, DMY and DM-W, in fish and amphibians, respectively. DMY in O. latipes was evidently acquired about 10 million years ago, because other closely related Oryzias species do not share the neo-Y defined by DMY [46,47] and DMRT1 is not sex-linked in any Oryzias species, or in any other fish with known sex chromosomes (Table 2) [48]. DM-W in X. laevis must also have been recently acquired, as close relatives (such as X. tropicalis) have no female-specific copy of this gene [38,49].

DMRT1 is also a candidate (albeit an unlikely one) for the sex-determining gene in monotreme mammals (platypus and echidna). It lies on one X of the platypus sex-chromosome complex, so is present in a single copy in males and two copies in females, the wrong way around for it to act to control sex by dosage. No male-specific DMY-like gene is detectable, so it is not clear how DMRT1 could be implicated in monotreme sex determination.

Sex-chromosome evolution

The seemingly chaotic distribution of GSD and TSD, and XY and ZW systems described earlier, with a range of sex-chromosome differentiation between none and extreme, makes no functional sense. Rather, it is the result of evolutionary forces acting in parallel in many systems. Sex-chromosome evolution is rapid and quixotic, and plays by unique rules. We can begin to understand these rules by comparing the sex chromosomes in the different vertebrate lineages (Figures 1 and 3).

Figure 3

Orthology of sex chromosomes and autosomes in amniotes. Four conserved regions, containing orthologous genes, are represented as green, dark blue, red and yellow regions in a variety of amniotes, lying on sex chromosomes (background shaded in yellow, pink or blue) or autosomes (numbered where known). The human (eutherian) XY pair consists of the blue and green ancestral regions that are separate in reptiles and birds and in monotremes and marsupials; the blue represents the ancestral therian XY (XCR) and the green blocks an addition in eutherians (XAR or YAR). SOX3 in the dark-blue block evolved into the mammalian sex-determining gene SRY. The dark-blue and the green blocks are autosomal in birds, reptiles and monotreme mammals. In birds, a different ancestral block (red), containing the sex-determining gene DMRT1, forms the ZW sex-chromosome pair; this gene block is also present in the ZW sex-chromosome pair in a gekko lizard (not shown), and on the sex-chromosome complex of platypus. In eutherians and marsupials, this block is autosomal. The red block is also autosomal in snakes, in which a different chromosome (yellow) has taken on a sex-determining role; the snake ZW is autosomal in all other lineages. All these genome regions are autosomal in a turtle that has temperature sex determination (TSD).

Curiously, it was comparisons of snake sex-chromosome morphology that first prompted the idea that vertebrate sex chromosomes evolved from an autosomal pair. The conservation of a large Z chromosome in all families, but morphological differences among the W chromosomes (homologous to the Z in boids such as pythons and boas; rearranged and partly heterochromatic in colubrids, the majority of snakes, and a heterochromatic rump in the evolutionarily advanced vipers) led Ohno [50] to propose that snake sex chromosomes differentiated from an original autosome pair as a sex-specific W became genetically isolated and degenerated. The same argument can be made for the bird ZW pair, in which the minimally degenerated W of ratites represents the ancient autosome. Observations of genetic homology between the bird Z and W chromosomes confirms their origin from an autosome pair [4]. Although snake and bird sex-chromosome systems are non-homologous, the bird Z is equivalent to the snake chromosome 2p (which contains DMRT1), and the snake Z to bird chromosome 2p, suggesting that they are separated by a single reciprocal translocation [5,6].

Similarly, the mammalian XY chromosome pair was suggested to have evolved from an ancestral autosomal pair by conservation of the X and degeneration of the Y to a small element retaining only a few active genes. The human X has more than 1,000 genes, but the Y retains only 45 that make unique proteins that are the relic of the Y degradation process [1]. Evidence for an autosomal origin comes from sequence homology between the human X and Y chromosomes, both within the terminal pseudoautosomal regions and between most active Y genes and their copies on the X chromosome [1]. Y degradation is rapid, fuelled by the high variation induced in the testis and the inefficiency of selection of a non-recombining entity [51]. Further degradation at this rate would lead to the extinction of the mammalian Y in a few million years [1] and indeed, several rodent lineages have already dispensed with the Y and evolved new, yet to be characterized, sex-determining genes [52-54].

The autosomal origin of the human X and Y chromosomes was confirmed by comparative mapping, which showed that orthologs of human X-borne genes are located on autosomes in other vertebrates, and even in other groups of mammals (Figure 3). Mapping in marsupials identified a region shared by the X chromosomes of all therian mammals - the X conserved region (XCR) - and a region that is autosomal in marsupials and is proposed to have been recently added to the X chromosome in placental (eutherian) mammals (Figure 1) - the X added region (XAR) [1]. The Y chromosome, too, is made up of a comparable conserved region YCR and added region YAR; all but four of the genes on the human Y derive from the added YAR [55]. The XCR and the XAR represent two evolutionarily ancient blocks of genes that became fused before the radiation of placental mammals, approximately 105 million years ago [3]. They are present and separated in all other vertebrates; for instance, XCR genes map to chicken chromosome 4p, and XAR genes lie, in almost the same order as in humans, on the long arm of chicken chromosome 1 (Figure 3). Claims that part of the XCR represented a third ancient block that mapped elsewhere on the chicken genome [56] were based on misidentification of orthologs from the incompletely assembled chicken genome sequence [57].

In monotreme mammals, the XCR and XAR gene blocks are present and independent. However, platypus orthologs of genes from the XCR as well as the XAR are autosomal, lying on chromosome 6 (Figure 3). This implies that about 166 million years ago, the XY chromosomes of therian mammals were still an autosomal pair [3]. Thus, our sex chromosomes are relatively young, and the decline of our Y chromosome is even more precipitous than we thought [1,58].

The non-homology of the bird ZW and the therian XY systems initially suggested that they evolved independently from different autosomes. This conclusion was challenged by the finding that mammalian X and bird Z markers are syntenic in a salamander (Ambystoma) [59], suggesting that these regions originally formed a 'super sex chromosome' that broke up differently in the two lineages. However, this association is lacking in other vertebrates, so their synteny may be a coincidence, made more probable by the large chromosomes of Ambystoma.

Although these mammal, bird, and snake systems appear to be evolutionarily stable, there are many vertebrate lineages in which sex chromosomes have changed very rapidly [4]. Such switches between sex-chromosome systems are particularly enlightening. Even in mammals, there have been recent switches from the standard XY system to new systems in three rodent lineages - mole voles [52], Japanese spiny rats [53] and the Mandarin vole [54] - but we do not yet know the identity or location of the new sex-determining loci, as discussed earlier. Perhaps the best documented switch in sex chromosomes is the evolution of a neo-Y in O. latipes (defined by the acquisition of DMY), and the evolution of a neo-W in X. laevis (defined by DM-W), as discussed above. In these systems, it is proposed that acquisition of a novel male- or female-dominant sex-determining gene overrode an old system.

How such transitions can occur is still quite mysterious because we might expect that morbidity and sub-fertility of the various hybrids would select against them. However, in Rana rugosa (which is polymorphic for ZW and XY systems), Ogata et al. [21] found that hybridization creates many different sex-chromosome combinations, all of which seem to be viable. Sex-ratio biases were seen in these crosses, and the authors suggested that the transition from an XY to a ZW system resulted from selection to maintain equal sex ratios after populations hybridized. Selection for optimal sex ratios is also invoked to explain the transition between XY and ZW sex chromosomes seen in other systems, such as cichlids [60,61]. Many shifts between GSD and TSD have been documented in reptiles, and sex-chromosome switches may be facilitated in species in which an underlying genetic system interacts with a continuous variable such as temperature [62]. Examples of interactions between GSD and TSD have been documented in the dragon lizard (P. vitticeps) [8], the three-lined skink Bassiana duperreyi [63,64], and the Atlantic silverside fish Menidia menidia [65]. For instance, in P. vitticeps, ZZ males and ZW females can be recognized cytologically and by a W-specific molecular marker. Eggs incubated over a range of temperatures hatch into equal numbers of ZZ males and ZW females, but at higher temperatures all hatchlings are female, and half of these are sex reversed ZZ [8].

Despite all the evidence of rapid changes of sex chromosomes and genes in many vertebrate lineages, closer examination reveals homologous sex chromosomes, and sex-determining genes, in distantly related animals. So the questions are: is this homology the result of shared ancestry? Or is it that some genes, or some chromosomes, are particularly good at doing the job?

Is homology a relic of shared ancestry?

We will first discuss the hypothesis that homology is the result of shared ancestry. Among the diversity of vertebrate sex-determining genes and chromosomes, DMRT1 and SOX3 stand out because they are implicated in sex in distantly related lineages. There is evidence, too, that some sex chromosomes reappear in diverse lineages. Perhaps the most compelling case for shared ancestry is the striking homology between the bird and gekko Z chromosome and the XY chromosomes of monotreme mammals, all of which contain DMRT1 [3,10,66]. This suggests descent from an ancestral amniote sex chromosome about 310 million years ago [58] (Figure 1a). SOX genes, too, turn up in distantly related lineages: as the SRY gene in mammals, as well as on the sex chromosomes of Rana rugosa [23], suggesting descent from a sex-chromosome system in an ancestral tetrapod about 400 million years ago [58].

How likely is such extraordinary sex chromosome stability? Although the mammalian, bird and snake sex-chromosome systems have been relatively stable for 166 and 80 million years, respectively, the rapid turnover of sex chromosomes within many reptiles, amphibians and fish [4] suggests that they arose independently in many different lineages. Indeed, comparing known sex chromosomes across mammals, birds, reptiles, amphibians, and fish reveals no clear ancestral vertebrate sex chromosome or gene (Figures 1 and 3). Although DMRT1 and SOX3 are present in distantly related vertebrates, it is difficult to see how both could be ancestral. And although DMRT1 is implicated in sex determination in widely divergent species, fish and frog DMRT1 homologs lie on non-homologous chromosomes. DMRT1 is not sex-linked in therian mammals, snakes, at least one turtle, one dragon lizard, two frog, and sixteen fish species (Figure 3 and Table 2). It is thus difficult to identify an ancestral vertebrate sex chromosome.

Yet it does appear that particular chromosomes are repeatedly used as sex chromosomes in fish (Figure 1 and Table 2). We identified the locations of genes on 16 known sex chromosomes in the genomes of the threespine stickleback (Gasterosteus aculeatus), green-spotted pufferfish (Tetraodon nigroviridis), Japanese medaka (O. latipes), and zebrafish (Danio rerio), in order to assign the sex chromosomes to the proposed ancestral teleost and vertebrate protokaryotypes [67]. The protokaryotype is the inferred karyotype of the common ancestor; that is, the karyotype of the common ancestor of teleosts or the karyotype of the common ancestor of vertebrates. For example, the teleost protokaryotype chromosome 3/vertebrate protokaryotype chromosome 5 appears as an XY sex chromosome in the ninespine stickleback [17,68] and the tiger pufferfish [69], and the same chromosome appears as a ZW sex chromosome in Lake Malawi cichlids [18,19] and the medaka species O. hubbsi [15]. Likewise, the teleost protokaryotype chromosome 7 has been independently used three times as a sex chromosome (Table 2).

Perhaps, then, the repeated appearance of homologous genes or chromosomes results from their independent reuse during sex-chromosome evolution. We shall next look at the question of whether there are limited options in the types of genes that can become sex-determination triggers or in the genomic regions that can become sex chromosomes.

Are particular genes better sex-determination triggers?

The second hypothesis set out at the beginning of this article suggests that the repeated use of the same genes for sex determination arises because they are best at doing this job. The best case for repeated use can be made for DMRT1, because its independent duplication, spawning novel systems (a male-dominant DMY in Japanese medaka and a female-dominant DM-W in X. laevis), implies that this gene is particularly suitable for such a role. Likewise, the involvement of SOX3 homologs in XY and ZW systems in mammals and a frog suggests that this gene, too, makes a good sex-determining gene. What properties could make DMRT1 and SOX3 particularly adept at sex determination? These genes have two features in common. First, they encode transcription factors, and so could regulate the expression of other genes in the pathway. Second, they are related to genes that are already part of the conserved sex-determination cascade, so may be particularly favored to become master sex-determination genes [70].

SOX (SRY)- and DMRT1-related genes are clearly not the only genes that can act as triggers for sex determination. As mentioned earlier, several rodent species have lost the Y chromosome and thus lost SRY. In these species, therefore, there must be genes other than SRY that determine sex [52-54]. Testing candidate genes, including DMRT1, failed to find linkage to sex in these species, or in most other species examined (Figure 3 and Table 2). However, current efforts to identify the master sex-determining genes in a variety of snake, lizard, frog and fish species will clarify common features of vertebrate sex-determination genes, and permit rigorous testing of hypotheses about the types of genes that may play this important developmental role.

Do particular chromosomes make better sex chromosomes?

If vertebrate sex chromosomes are not identical by virtue of their descent from a common ancestor, how might we account for the repeated appearance of homologous sex chromosomes in diverse vertebrate lineages? Some chromosomes are surely selected because they already contain genes that are particularly well suited to a sex-determination role, such as DMRT1 or SOX3. Other sex chromosomes may be defined by the acquisition of either new mutations or transposed genes, such as copies of DMRT1. Are particular chromosomal regions more prone to evolve other sex-determining loci? Are particular genome regions favored for such transposition events?

One compelling hypothesis that addresses these questions proposes that an autosome containing gene with sexually antagonistic effects (that is, beneficial in one sex and detrimental in the other) would be subject to selection for the spread of a linked sex-determination locus [71]. This could occur by the evolution of a novel sex-determining locus, as appears to be the case in multiple Oryzias species [47]. It could also occur through the transposition of an existing sex-determination locus onto another chromosome; this seems to have happened in salmonids, where markers tightly linked to the sex-determination locus are found on different chromosomes [72]. Recent studies in Lake Malawi cichlids link a pigmentation trait predicted to be under sexually antagonistic selection to a transition between an XY and a ZW sex-chromosome system [18,19]. Some rock-dwelling female cichlids have an orange-blotch color phenotype that provides camouflage but might be a disadvantage to males because it disrupts their breeding colors [19]. Consistent with the sexual antagonism theory, the gene encoding the orange-blotch phenotype lies on a W sex chromosome that is epistatic to the existing XY sex-chromosome system in Lake Malawi cichlids [18,19].

Fusions between sex chromosomes and autosomes might also link a sex-determination locus to genes with sexually antagonistic effects [73]. Y-autosome fusions have occurred in at least 25 different fish species [74], including twice independently in sticklebacks [17,20]. One Y-autosome fusion has occurred exclusively within the Japan Sea threespine stickleback population [20], males of which exhibit a unique mating behavior that is a key component of behavioral isolation from a neighboring stickleback species. This behavior maps to the neo-X chromosome created by the fusion, suggesting that sexually antagonistic selection might have driven this rearrangement to fixation [20]. Similar evolutionary forces might have been involved when an ancient ZW system underwent translocation to an ancient chromosome 2, causing a switch in the identity of the ZW chromosome pair between snakes and birds [5,6].

In mammals, it might be that the addition of the XAR (or the YAR) to the ancient mammalian XY pair added genes to the X or Y for which there was sexual antagonism [1]. The X as well as the Y chromosome is replete with genes involved in reproduction, and many of these genes are expressed in gonads in birds. Translocations between sex chromosomes and autosomes are also common in marsupials, and fusion of sex chromosomes to autosomes containing sexually antagonistic genes might provide an explanation for the bizarre translocation complex in monotreme mammals, which would be expected to be deleterious because it makes segregation of sex chromosomes into sperm difficult [2].

If sexually antagonistic selection drives the rapid turnover of sex chromosomes in fishes and other vertebrates, autosomes harboring sexually antagonistic genes could be repeatedly selected to be sex chromosomes. Interestingly, the invading ZW system of Lake Malawi cichlids involves a chromosome that shares homology with the sex chromosomes of three distantly related fish species, Pungitius pungitius, Takifugu rubripes, and O. hubbsi (Table 2). This chromosome also carries a number of genes that function in pigmentation pathways (PAX7, WNT4A, MITF) as well as opsin genes involved in color vision across these species (Table 2). Linkage between male-specific pigmentation traits and sex chromosomes also occurs in poeciliid fishes such as the guppy (Poecilia reticulata) and the platyfish (Xiphophorus maculatus) [75]. In guppies, the male-specific pigment patterns are under both natural and sexual selection [75]. Thus, it is possible that sexual antagonism for bright color patterns might contribute to the rapid turnover of sex chromosomes in fishes.

The future of research on sex determination

The availability of sequence data, and of more sophisticated molecular and cytological techniques, will revolutionize comparative studies of sex chromosomes and sex-determining genes. We can now identify cryptic sex chromosomes in even the most exotic vertebrates, obtain and map molecular markers, and deduce homologies with better-known genomes. Promising systems are those in which there has been rampant turnover within a well-defined phylogeny (for example, dragon lizards, gekkos, sticklebacks and cichlids). Such studies should identify new sex-determining genes, and enable the classification of the types of genes that might be involved.

We can also take an unbiased genome-wide approach to ask whether particular autosomes are rich in sexually antagonistic genes that could be selected for linkage with a sex-determination locus, driving the fixation of new sex chromosomes. Recently, it has been suggested that sexually dimorphic or sex-biased gene-expression patterns might reflect sexually antagonistic selection [76,77]. Sex-biased gene expression appears to be common, being found in mice, birds, flies and worms [77]. So, by analyzing genome-wide patterns of sex-biased gene expression, we could determine whether particular chromosomes contain an excess of sexually antagonistic genes, as is the case for the more ancient sex chromosomes of flies, mammals and birds [77].

Evidence that genes with potential sexually antagonistic effects lie on nascent sex chromosomes would provide compelling evidence that sexually antagonistic selection plays an important role in driving vertebrate sex-chromosome evolution and turnover. An excess of sex-biased gene expression on homologous chromosomes across a number of vertebrate systems might also explain why the same autosomes have been independently selected as sex chromosomes several times during vertebrate evolution.

In conclusion, we have provided examples of deep homology in the ZW chromosomes of birds, a gekko and monotremes, and XY chromosomes in all therian mammals. However, we also have clear examples of the independent re-use of particularly handy genes (copies of DMRT1 and SOX3), and of fish chromosomes that have become sex chromosomes multiple times. Thus, the answer to the question we pose in the title is - both.