Of sex and determination: marking 25 years of Randy, the sex-reversed mouse

One of the most fundamental processes in biology is the development of two sexes. Many great minds, from Aristotle to Darwin, struggled with the question of how the two sexes arise. However, by 1959 it was known that in eutherian mammals a gene or genes on the Y chromosome initiates testis development in the embryo and that the testes then produce all the hormones needed to make a male. In the absence of the Y chromosome, the embryonic gonads develop instead as ovaries. It was hoped that finding the relevant gene(s), termed testis-determining factor (TDF in humans; Tdy in mice), would eventually lead to a complete understanding of the processes needed to generate a testis, providing a paradigm for elucidating the morphogenesis of other organ systems. In subsequent decades, TDF/Tdy was narrowed down to a small region on the short arm of both the human and mouse Y by analysing XX males carrying a small Y fragment. The overlap between this fragment and the Y chromosome interval deleted in XY female patients delimited TDF (Pritchard et al., 1987; Page et al., 1987), enabling the search for the gene or genes defining that locus to begin.

The TDF gene was predicted to have five key properties. First, it should be conserved amongst mammals. Second, it should be located on the Y chromosome in all mammals and within the smallest known sex-determining fragment of the Y chromosome in mice and humans. Third, its product should most likely act cell-autonomously and control the activity of other genes (Burgoyne et al., 1988). Fourth, it should be expressed in the somatic component of the developing testis at the time of sex determination (Burgoyne et al., 1988). Finally, gene mutations should lead to failure of testis development and to XY females, whether human patients or mice.

Two research teams were strong contenders in the race to identify TDF. In London, UK, Peter Goodfellow's human molecular genetics group teamed up with Robin Lovell-Badge's mouse developmental biology group to bring complementary expertise to the problem. In Boston, David Page's group led the race within the USA. In December 1987, the Boston group published a strong candidate for TDF, the gene encoding the zinc-finger protein ZFY (Page et al., 1987). ZFY mapped to the smallest region of the human Y chromosome found so far in XX males, was missing in XY females and encoded a transcription factor that could act cell-autonomously; but over the following two years, evidence increasingly indicated that it was, in fact, the wrong gene. Its marsupial homologue was found not to map to the Y chromosome (Sinclair et al., 1988). ZFY was expressed ubiquitously in humans, which did not fit well with the biology of TDF, while its mouse homologues were neither expressed in somatic cells of the fetal testis (Koopman et al., 1989) nor mutated in XY female mice known to carry a heritable mutation mapping to the Y chromosome (Tdy^m1; Lovell-Badge and Robertson, 1990; Gubbay et al., 1990). Crucially, four variably masculinised XX patients with testicular tissue were found to possess Y-derived markers but lack ZFY (Palmer et al., 1989).

Fortunately, however, these same four patients held the key to identifying a new candidate, as they had in common a mere 35 kb of Y chromosome DNA. The Goodfellow group cut this region into small fragments and tested each on Southern blots containing male and female DNA from human, mice and cattle. Only one human Y-fragment was identified that showed male-specific bands in human and several other mammalian genomes (Sinclair et al., 1990), as well as in XY and XX male mice [where the latter carried a small region of the Y chromosome short arm termed Sxr(b) (Laval et al., 1995)], but not XX or XY^Tdym1 female mice (Gubbay et al., 1990). As unique sequences conserved between the Y chromosomes of different mammalian species are very rare, these results implied that this human Y chromosome fragment probably contained TDF/Tdy. Sequencing revealed an open reading frame coding for a single exon gene, cautiously named sex-determining region on the Y chromosome (SRY in humans, Sry in mice: Sinclair et al., 1990; Gubbay et al., 1990).

With this strong candidate in hand, it was relatively straightforward to test its properties against the predictions made for TDF. The first two were satisfied by Southern blotting, as described above. In those experiments, additional bands were observed that were not sex-specific – an early indication of the existence of an Sry-related gene family later named Sox. Although much of the SRY/Sry open reading frame was not well conserved, both human and mouse sequences included a region encoding a 79 amino acid domain with homology to a conserved DNA-binding motif present in the nuclear high-mobility group proteins HMG1 and HMG2, and in the eukaryotic transcription factor UBF (Jantzen et al., 1990). Shortly afterwards, a related ‘HMG domain’ was found in the transcription factors TCF1 and LEF1, further suggesting that SRY was itself a transcription factor (van de Wetering et al., 1991; Travis et al., 1991). RT-PCR experiments showed that Sry is expressed in XY mouse fetal gonads at 11.5 days post coitum – the time point immediately preceding the first morphological signs of testis differentiation (Gubbay et al., 1990). Indeed, Sry is expressed for only a brief period prior to sex determination in mouse gonads and, like sex determination itself, does not depend on the presence of germ cells (Koopman et al., 1990). Finally, de novo point mutations affecting the SRY HMG domain were detected in DNA from two human XY females, providing strong genetic evidence that the function of SRY is required for testis development and male sex determination (Berta et al., 1990). All the evidence pointed to SRY/Sry being the right gene. But how could the putative role of SRY/Sry be proved?

It was reasoned that the gene(s) representing Tdy should cause XX sex reversal if introduced as a transgene via pronuclear injection into mouse zygotes. A 14 kilobase (kb) genomic clone containing the mouse Sry open reading frame and about 8 kb of 5′ and 5 kb of 3′ flanking DNA was identified and mapped. This had been sequenced and characterised sufficiently to be confident that there was no additional ORF in the region, but there was no assurance that it contained all the necessary sequences required for correct quantitative, temporal and cell type-specific expression, or mRNA stability. Without adding a tissue-specific promoter or insulator sequences, the possibility that this small and naked genomic fragment might function to drive male development seemed remote, but it was worth a try.

Injected embryos were analysed at 14 days post coitum, some 36 h after the formation of testis cords, which, together with the formation of a male pattern of gonad vasculature, make it possible to visibly distinguish testes from ovaries. Chromosomal sex and the presence or absence of the transgene were verified by PCR and/or Southern blotting. Early experiments were disappointing. Then, in one experiment, two out of four XX transgenic embryos had testes indistinguishable from those of control XY embryos (Koopman et al., 1991). Another two transgenic XX fetuses were female and another four XX fetuses had low transgene copy numbers: six XX females, in which mosaicism and/or unfavourable integration had presumably precluded transgene function. The two sex-reversed embryos were cause for relief and an ambrosial sense of triumph, but there was little time to celebrate: there was much more work to do.

In subsequent experiments, mice were allowed to develop through to birth after pronuclear injection. Five of these were transgenic, including three XX. One of the latter looked to be an entirely normal male externally (Fig. 1), was very willing and able to mate with females, and so was given the nickname Randy. He had a normal male reproductive tract, but his testes were small and devoid of spermatogenesis, entirely as predicted for an XX male: two X chromosomes are incompatible with the proliferation of spermatogonia after birth and other Y chromosome genes are required to make sperm. Because the sex reversal in Randy was complete, this was the final proof that Sry was not only necessary, but also sufficient on its own to initiate male somatic development. Thus, SRY/Sry was functionally and genetically equivalent to TDF/Tdy. The paper was prepared with urgency and excitement (Fig. 2), reviewed positively, accepted on 5 April and published within five weeks as a feature article in Nature (Fig. 1).

By 1991, several transgenic mice had already hit the headlines. These included big mice that overexpressed rat growth hormone (Palmiter et al., 1982) and mice that developed specific tumours (Brinster et al., 1984). However, many hundreds of others flew below the radar and it was assumed that this one would not create much of a stir in the media, especially because Randy looked like a normal male mouse and there had already been media attention around the discovery of SRY/Sry the previous year. We could not have been more wrong!

The story made front-page news in the UK and spread from there around the world (Fig. 3). Reporters, photographers, TV and radio crews invaded the lab for weeks. The technical accomplishment, the achievement of a significant milestone in genetics and possibly some cute photos of mice, resonated with many. Some stories trumped up a relevance to cancer, others questioned the ethics or talked of a slippery slope, and yet others found the concept of ‘a sex-change for Mickey’ mildly amusing.

The broader significance of the work undoubtedly lies in having finally provided an answer to one of our most fundamental and long-standing questions – why am I male or female? That a single gene makes all the difference was a surprise to many. Scientifically, the discovery of SRY/Sry helped to open the field of sex determination. It provided an entry point into the genetics and embryology underlying testis development, explained cases of sex reversal, and provided an important new diagnostic. It also spurred research into ovary development – the other half of the story of gonadal sex determination.

Proving the testis-determining role of SRY/Sry was expected to open the gate to discovering the next gene in the pathway, and the one after, such that a comprehensive picture of how sex determination works at the molecular genetic level would be ‘quickly’ gained. Things did not work out quite that way. SRY/Sry proved to be extremely difficult to work with in almost every way imaginable. The gene is small, intronless and, being from the Y chromosome, embedded in a poorly conserved and highly repetitive DNA sequence (Gubbay et al., 1992). It is expressed in a small number of cells in an inaccessible fetal tissue for (in mice) only a brief period (Albrecht and Eicher, 2001; Sekido et al., 2004; Wilhelm et al., 2005). No specific antibody for SRY was produced for over a decade (Bradford et al., 2007). Even initial efforts to define its mRNA transcript literally had us going around in circles (Capel et al., 1993). Immortal cell lines expressing SRY/Sry bore questionable resemblance to fetal gonadal cells and primary gonadal cells placed into culture quickly lost SRY/Sry expression (Capel et al., 1996; Beverdam et al., 2003). The SRY binding site is small and degenerate (Harley et al., 1992), hence putative target sites are present throughout the genome. In short, standard techniques for discovering upstream or downstream genes were essentially unworkable for SRY/Sry (Larney et al., 2015).

Despite this, we now have a basic understanding of the SRY protein's cellular and molecular functions. SRY binds specifically to DNA in one or more enhancers of the target gene Sox9, in order to upregulate Sox9 transcription (Sekido and Lovell-Badge, 2008). This occurs in just one cell lineage – the bipotential supporting cell precursors that give rise to granulosa cells in the ovary or Sertoli cells in the testis. SOX9, in turn, activates the expression of other genes required for Sertoli cell differentiation and function, including factors that instruct the other cell lineages within or adjacent to the gonad to follow a male differentiation pathway (reviewed by Svingen and Koopman, 2013). Importantly, SOX9 also promotes its own expression, so that continued Sry expression is not required to sustain testis differentiation or maintain testis phenotype. At the same time, SOX9 and/or its direct or indirect targets repress genes associated with ovarian development. In summary, Sry acts as a trigger and Sox9 an effector in driving testis development (reviewed by Canning and Lovell-Badge, 2002; Kashimada and Koopman, 2010). Not surprisingly, transgenic expression of Sox9 during XX mouse fetal gonad development is as effective as Sry in inducing testis development and maleness (Vidal et al., 2001).

As with all transcription factors, a threshold level of SRY/Sry expression is required to activate Sox9. It is also clear that SRY/Sry must act within a specified time window, otherwise it ‘misses the boat’; transgenic mice in which Sry is expressed later than normal via an inducible promoter do not switch on Sox9 and do not make testes (Hiramatsu et al., 2009). Because the timing and levels of SRY expression are integral to its sex-determining function, considerable effort has focused on understanding the regulation of SRY/Sry transcription. In mice, Sry expression is thought to be activated by some combination of the core transcription factors steroidogenic factor 1 (SF1), Wilms tumor 1 (WT1) and GATA binding protein 4 (GATA4), although whether and how these factors act specifically on Sry regulatory sequences are matters that remain to be fully explained experimentally (reviewed by Larney et al., 2014). A signal transduction pathway involving GADD45γ, MAP3K4 and p38 MAPK (MAPK14) is also crucial (Warr et al., 2012), although the extracellular ligands and their source have yet to be identified. SRY/Sry is a rapidly evolving gene, so the specific details may not be the same in all mammals.

Somewhat surprisingly, gonadal phenotype has emerged as relatively labile. The deletion of Foxl2 from adult granulosa cells allows Sox9 to be expressed, which then triggers transdifferentiation to Sertoli cells. This leads to gonadal sex reversal, with the ovaries exhibiting a structure and cell types resembling that of the testis (Uhlenhaut et al., 2009). Conversely, loss of Dmrt1 in the adult testis leads to upregulation of Foxl2 and repression of Sox9, and a transformation from testicular to ovarian cell types (Minkina et al., 2014). All of this suggests that there is no crucial role for chromatin modifiers to lock down the system. Could this flexibility and the need for continuous repression of the opposite pathway reflect an evolutionary hangover from early vertebrate ancestors that could change sex, as seen in some species of fish (Kobayashi et al., 2013)?

The discovery of SRY/Sry also encouraged a search for essential sex-determining genes in non-mammalian vertebrates, where it is absent, and consideration of how sex chromosomes and sex-determining genes arose during evolution. Mapping of Sox3, one of the original group of Sox genes co-discovered with Sry, and its human homologue SOX3, to the mouse and human X chromosomes, first suggested the prevailing model that Sry evolved from Sox3 (Stevanović et al., 1993). SOX3 does not seem to have any role in sex determination, but has conserved functions during gastrulation, in the development of the central nervous system, and in a few other tissues. It seems most likely therefore that Sox3 was originally autosomal, as it still is in prototherian mammals, and that a regulatory mutation led to expression of one copy in the early gonad, where its protein product promoted transcription of Sox9, which does appear to play a role in Sertoli cell differentiation in all vertebrates. This variant Sox3 became Sry, the chromosome it was on became the Y chromosome, and the chromosome retaining a normal Sox3 became the X chromosome. In support of this model, situations where SOX3 is mis-expressed in the early gonad can lead to XX male sex reversal in mice and probably in humans, perhaps mimicking events in evolution (Sutton et al., 2011).

Despite much progress from many labs, we still do not have a comprehensive picture of the regulatory networks leading to the differentiation, maturation and maintenance of either the testis or the ovary in mammals. Most causes of XY and XX gonadal dysgenesis, other than those involving SRY, remain unknown. As was the case with the original identification of SRY/Sry, co-ordinated efforts involving multiple species will continue to drive progress in filling these gaps. For example, transcriptomic studies in mice can identify candidate disorders of sex development (DSD) genes that can be tested for mutations in the DNA of DSD patients. Conversely, DSD patient screening can identify new candidate genes that can be characterised and functionally studied in mice. In this way, and through the application of increasingly sophisticated biochemical and genetic approaches, the field will continue to advance.

These days, it is hard to imagine the stir created by a single transgenic mouse experiment in 1991. Given the lack of current tools such as whole genome sequencing, internet-based bioinformatic tools and advanced gene-editing tools such as CRISPR/Cas9, it remains something of a miracle that within the space of a couple of years, a plausible TDF/Tdy candidate was identified and its function so definitively validated by the creation of Randy the mouse. While Randy did not leave any direct descendants, other transgenic mice made with Sry have been part of his legacy. For example, the Tdym1 mutation, where Sry is deleted from the Y chromosome, can be complemented by an Sry transgene on an autosome to give fertile XY males. The four types of offspring these would have produced when mated with XX females would make it possible to distinguish direct effects of X- and Y-linked genes from the effects of gonadal hormones on differences between the sexes (De Vries et al., 2002). This ‘four core genotype’ model has now been used to explore the basis of many sex-specific differences, from neuroanatomy to food preferences. Moreover, it has led to a more general appreciation of how commonly such differences occur and their clinical importance (Arnold, 2014). Sry could itself have roles outside the gonad. Indeed, it is expressed in the brain in mice and other mammals (Loke et al., 2015). This could place male-specific traits under selection, perhaps explaining aspects of rapid Sry evolution. Finally, the discovery of Sry also, importantly, led to the identification of other members of the Sox gene family, of which there are 20 in man and mouse (Schepers et al., 2002). These Sox genes, unlike Sry, are highly conserved in evolution and have proved essential for many cell fate decisions during embryogenesis. They are also clinically relevant for both congenital disease and cancer. Quite an impressive legacy from a sterile, albeit frisky, XX male mouse.

We gratefully acknowledge all our co-workers for their part in the Randy story. We thank Melissa Little for assistance in the preparation of the manuscript and Jérôme Collignon for permission to use the photograph in Fig. 2. We apologise to colleagues whose contributions we could not include due to space constraints.