Choosing Sex

In many animals, sexual characteristics are quite plastic—even in adult life. In some species of fish, all it takes is a glance, or lack thereof, to cause an adult female to change her sex and become male.  When the dominant male goes out of sight from the school, one of the females will undergo a sex change, taking on the coloration and behavior of the alpha male, and transition from making eggs to making sperm instead. A more subtle example is a species of mole that maintains “ovotestes” in adult life, changing from female to male characteristics and back again, depending on the season and whether it's more advantageous to be submissive or to produce high levels of testosterone and exhibit aggressive behavior.² 

What accounts for the remarkable sexual plasticity seen in many animals? Perhaps it is the inherent plasticity of the gonad. For most developmental processes, there is only one possible outcome. For example, a kidney primordium can only make a kidney, and a lung primordium can only make a lung. In contrast, the gonad can develop into either a testis or an ovary.  This choice, “sex determination,” occurs during fetal life and is stable thereafter, but in other animals like some fish, this choice may be reconsidered later.

Another striking difference between sex determination and other developmental processes is that the genes that control most developmental mechanisms are tightly conserved across the animal kingdom. However, the mechanisms controlling sex determination seem to vary wildly across the animal kingdom. In some animals, the sex of offspring depends on population density, whereas in others, it depends on temperature. Humans develop inside the uterus, where they are (for the most part) protected from the vagaries of the environment. They use a genetic mechanism to determine sex based on their X and Y chromosomes. No unifying mechanism has been found that controls sex determination in all vertebrates, yet it seems impossible that such an essential process isn't tightly conserved at some level.

When I began research in my own lab, it seemed to me that insight into this problem might come from a better understanding of how sex determination occurs at the level of the cell biology of organ development. How do the cells of the gonad decide to form a testis or ovary, and how do the different mechanisms of sex determination seen across the animal kingdom regulate this process?  Recent work from my lab and many others suggests that there may be a common underlying mechanism after all.

For me, the story began in 1991, with the discovery of the gene that governs sex determination in mammals. I remember it as an eventful week in Robin Lovell-Badge’s lab at the National Institute for Medical Research in London, where I was a postdoc.  We had journalists and photographers putting us in "busy scientist" poses and film crews grilling us about the details of our work. We had taken our candidate gene, the mouse Sry gene on the Y chromosome, and inserted it into the genome of an XX (female) mouse embryo, turning it into a male. In a play on that experiment, one newspaper sported a cartoon of a sex-reversing Minnie Mouse. 

The bipotential gonad
In the early stages of development, the gonad has the potential to become either a testis or ovary. Both male (mesonephric) and female (Mullerian) ductal systems are initially present. Once the fate decision is made, the gonad either descends or rises in the body and the ductal systems of the opposite sex degenerate.

In the reciprocal experiment, we showed that removal of the Sry gene from the Y chromosome of male embryos caused genetically XY male animals to develop as females.³ In collaboration with the Peter Goodfellow lab at the Imperial Cancer Research Fund (ICRF) in London (which worked on the human SRY gene), we paid a visit to the London Zoo to collect DNA samples from male and female horses, chimps, rabbits, pigs, cattle, and tigers. We found that all these animals carried the SRY gene on their Y chromosome, reflecting the wide conservation of this mechanism of sex determination in mammals.⁴

With the Sry work behind me, I started my own lab at Duke University in 1993. While other groups coming out of the Lovell-Badge and Goodfellow labs continued to characterize the Sry gene and other genes immediately downstream, I wanted to study the earliest cellular mechanisms that trigger the decision to develop a testis or an ovary, at the point when the SRY transcription factor is expressed in the gonad.

The first challenge was to set up a system where I could study the gonads as they developed in a controlled environment. It was no small task to figure out the right conditions to keep embryonic mouse gonads viable in a dish for several days while they made their fate decision.

Robin Lovell-Badge and a former post-doc of his, Katarina Nordqvist, came to visit my new lab in 1995.  The three of us were very interested in testing an old idea that a population of cells from the mesonephros—a nearby tissue that is closely associated with the gonad at this stage—migrates into the gonad. We made a recombinant organ by combining a mesonephros carrying a beta-galactosidasegene that makes all cells blue, with a “white” unlabeled gonad, and cultured the two pieces together for several days. To our excitement, blue cells from the mesonephros migrated into the unlabeled gonads—but only into male XY gonads, never into XX female gonads.  Once in the male gonad, the mesonephric cells surrounded the Sry-expressing Sertoli cells and formed testicular cords, the first morphological change that signals a commitment to testis development.⁵

We could see that cells had migrated, but it wasn't clear whether that really mattered. One of my students, Christopher Tilmann, devised an experiment to test the importance of cell migration. He placed a membrane barrier between the cultured mesonephros and gonad, demonstrating that blocking the mesonephric cells from migrating prevented the early steps of testis development. We wondered what would happen if we induced migration into an XX gonad—could we make it develop more like a testis than an ovary?

After many failed efforts to test this idea, it finally dawned on me that we could make a "sandwich" organ culture. We would place a developing XX female gonad in between an XY male gonad on one side and a blue mesonephros on the other. We were excited to see that cells from the mesonephros crossed over the XX female gonad on their way to the XY male gonad. Along the way, these traveling cells induced the developing female gonad to activate some genes associated with male development and to form male-like structures resembling testis cords—all in the absence of the master Sry gene.⁶

The sandwich experiment
To test whether the cellular process of mesonephric cell migration is important for sex determination, we dissected the gonads from normal XX female and XY males and combined them with a mesonephros from a mouse expressing a blue marker (labeled here in dark gray). We could see blue mesonephric cells penetrate the XY male gonad, but never the XX female gonad (A). But when we sandwiched the female gonad in between the male gonad and the mesonephros (B), the XX female was not only penetrated by blue cells but also began developing male-like structures.

These experiments and others in the lab were gradually changing the way we viewed the problem of sex determination. Although SRY lies at the top of the sex determination cascade in mammals, it was becoming clear that the pathways downstream of SRY are critical in controlling testis morphogenesis, and without a testis, the embryo develops all female secondary sex characteristics.

By the late 90s we had identified several developmental processes essential for gonad development, but still lacked a clear picture of the genes that controlled it. Luck took a hand when David Ornitz at Washington University Medical School called to tell me about a mutant mouse. A post-doc in his lab, Jenny Colvin, had generated mice incapable of producing fibroblast growth factor 9 (FGF9). Mice lacking Fgf9 died at birth because their lungs could not form properly. However, Jenny noticed that all of the embryos developed as females. This was a very exciting finding because it suggested that Fgf9 was one of the genes that control developmental processes important for testis development. While SRY—a transcription factor—can only act on the cell that expresses the protein, FGF9 is a secreted protein and acts as a signaling molecule to nearby cells. It sounded like it might be just the sort of signal that controls proliferation or attracts the migration of cells from the mesonephros. 

Further work in my lab showed that during the bipotential stage of gonad development—before the critical fate decision—Fgf9 is expressed in both XX and XY gonads.  But after Sry is expressed, Fgf9 is strongly up-regulated in XY male gonads, and down-regulated in XX female gonads.  In XY male gonads that lacked Fgf9, testis development was completely blocked and some aspects of ovary development could be detected (see graphic below).

Since FGF9 is a secreted factor, we wondered what would happen if we added it to the culture medium for female gonads. To our delight, soluble FGF9 induced mesonephric cells to migrate into the XX female gonads, pushing their development toward the testis pathway.^7,8

All indicators were pointing to the idea that Fgf9 played an important role in testis development. But what controlled female development? To the great irritation of many female investigators in the field, female development had classically been referred to as the “default pathway”—suggesting a passive process. To most of us, this was not an attractive idea.

The first evidence for an active female pathway came in 1999, when Andy McMahon's group at Harvard generated a mouse incapable of producing WNT4. Like FGF9, WNT4 is a secreted signaling molecule that can affect cells at a distance. In mice lacking the Wnt4 gene, even those that were genetically XX female, gonads developed with some characteristics of testes. For example, XX gonads from these mutants showed patterns of cell migration similar to XY gonads and, later in development, produced testosterone.^9,10 This was particularly interesting because it was consistent with reported cases of genetically XX female humans who develop a testis in the complete absence of SRY.  One explanation suggested for these patients was that something had gone wrong with their active ovary-determining pathway—a pathway necessary to block testis development.

We found that, like Fgf9, Wnt4 is expressed in both sexes while the gonad is still bipotential, but it is up-regulated in XX gonads and down-regulated in XY gonads precisely at the time when the gonadal fate decision occurs—the opposite of Fgf9 expression.

About this time, we remembered a piece of evidence from organ-culture experiments done earlier in my lab suggesting that FGF9 could block expression of Wnt4. Could these two signaling pathways be acting antagonistically, staging the battle of the sexes in the gonad? Yuna Kim, another graduate student in my lab, planned a set of experiments to test this idea.

Other researchers had shown that the primary role of SRY is to up-regulate a closely related transcription factor, Sox9.  Various experiments showed that SOX9 is capable of substituting for SRY in activating testis development. The question was how WNT4 and FGF9 fit into the story. Yuna found that FGF9 and SOX9 reinforce each other's signaling to establish the testis pathway in XY gonads. She showed that when Fgf9 is eliminated, XY male gonads switch sex and activate ovarian genes.  But our most exciting finding was when she discovered that SOX9 and FGF9 are both up-regulated in an XX female gonad when Wnt4 is absent. This clearly showed how the male pathway could be activated in an XX genetic female, in the complete absence of the Sry gene—just as those human XX male patients had predicted.¹¹

Based on these experiments, we proposed a new model for mammalian sex determination. In both XX and XY primordial gonads, Fgf9, Sox9, and Wnt4 are all expressed simultaneously early in development, when the fate of the gonad is still undetermined. In an XX gonad, WNT4 dominates and turns off the testis pathway.  However, in an XY gonad, SOX9 and FGF9 get an extra boost from SRY, which allows them to dominate and repress WNT4.

The animal kingdom has many means of determining sex, from population density and behavioral cues in fish, to temperature in turtles, alligators and other reptiles, and hormonal influences in many egg-laying species.  Yet, surely a process as important as sex determination must be conserved at some level.

I and others have begun to suspect that although the primary gene controlling sex determination varies among species, perhaps what is conserved is an underlying pattern of antagonistic signals—such as the ones we've seen in mice with FGF9 and WNT4. This fundamental sex-determining mechanism could easily operate in response to a genetic switch (such as Sry in mammals) or to an environmental cue (such as temperature in turtles), as long as the initial decision is amplified and reinforced by downstream pathways that recruit all the cells of the gonad to one game plan.¹²

Competing signals
In the early days of gonad differentiation, both male female genes are turned on and in relative balance (A). In XY embryos, the Sry gene turns on in the gonad, causing the Sox9 and Fgf9 genes to reinforce each others' signaling and suppress the Wnt4 female pathway. In females, the absence of the Sry gene allows Wnt4 signaling to overpower the Sox9/Fgf9 feedback loop, turning on the female program of development.

In an effort to learn from another species, we began to work with red-eared slider turtles, which determine sex via temperature. When their eggs are incubated at 26 degrees Celsius, 100% become male but when incubated at 31 degrees, 100% become female. (At temperatures in between, mixed sex ratios occur.) We have begun to explore the cellular basis for the development of the testis and ovary in the turtle, and to search for similar control signals by returning to our organ-culture methods.

This work has led us to suspect that the antagonistic signaling system that we uncovered is just the tip of the iceberg—that we should be looking at the workings of the entire complex system of signals that underlie sex determination and gonad development rather than at single genes. We are very excited about a new project to do just that, employing many of the new techniques and computational skills of systems biology.

Our understanding of sexual development is evolving along with our ability to test and measure the process. We have only begun to clarify the early genetic and cellular processes that influence the initial stages of gonad differentiation. The subsequent effects of hormones, environment, and neurological wiring all have critical roles in the eventual identification of an individual as “male” or “female.”

In the face of this complexity, tests used by many athletic organizations for the presence of SRY as the sole means of classifying contestants as male or female seem very simplistic. Among other things, this assessment does not provide a category for qualified individuals who possess some combination of male and female characteristics. Yet these individuals also represent the spectrum of human abilities. In the case of Caster Semenya, it is a pity that her impressive achievements might be overshadowed by accusations that may simply stem from her misalignment with Western standards of beauty rather than from purposeful deception with respect to her sex.

Blanche Capel is a Professor in the Department of Cell Biology at Duke University Medical Center. She thanks the many former and current members of her lab for their wonderful work, especially Lindsey Barske and Jonah Cool, who have helped edit this article.

Correction (15 September 2009): This article originally identified Blanche Capel as an associate professor. She is actually a full professor. The Scientist regrets the error.