ABOVE: Modified from © Istock.com, VCHAL

For a person with two X chromosomes, full expression of the genes on both could mean a disastrous double dose of their protein products, interfering with the delicate balance of protein expression and interactions all over the body. But cells have a simple solution: turn off one X chromosome and crumple the extra genes into a quieted mass of DNA called a Barr body. A long noncoding RNA known as XIST (pronounced “exist”), which is expressed from the “inactive” X itself, plays a key role in this process. Acting only on the chromosome it’s transcribed from, XIST coats the DNA, turning it into silent heterochromatin.

Except there’s a problem. Certain genes just won’t stay mum.

Some Barr body genes, including XIST itself, actively control the silencing of other stretches of the X. 

Scientists had long known that some genes on X chromosomes aren’t...

Scientists had long known that some genes on X chromosomes aren’t subject to silencing, but the list of these “escapees” has grown over the past two decades. Somehow, about a quarter of the approximately 1,150 genes on the purportedly silenced human X speak out: they avoid coating by XIST. Some of these escapees are expressed just a bit, others quite a lot. Some appear to be beneficial, possibly helping to protect XX individuals from diseases such as cancer. Other X genes cause damage when expressed from both chromosomes, possibly predisposing women to autoimmune disorders such as multiple sclerosis or lupus. 

Recent clues suggest that the signals to activate a Barr body gene lie in the DNA sequence and proteins that attach to it. Some Barr body genes, including XIST itself, actively control the silencing of other stretches of the X. Such research into the nature of X escape has major implications for understanding biological differences between the sexes, says Sundeep Kalantry, a geneticist at the University of Michigan Medical School. Scientists have long pointed to sex hormones and genes on the Y chromosome as exclusive explanations for those differences. “I think this is being questioned by many groups,” Kalantry says. “The concept is out there that genes that escape X inactivation are influencing female-specific physiology.” 

Identifying the escapees

English geneticist Mary Lyon, who originally proposed X-chromo-some inactivation in 1961 while working at the UK’s Medical Research Council in Harwell, suspected early on that certain genes wouldn’t be shut off. At least 29 genes reside in the two pseudoautosomal regions (PAR1 and PAR2) shared by the X and the Y chromosomes. These regions at the tips of the sex chromosomes typically stay active even in the Barr body, so that XX and XY individuals get the same doses of those genes. 

Studies on non-PAR escapees began to trickle out in the 1970s and ’80s, but the extent of the phenomenon wasn’t known, and low expression of these escapees was presumed to have a minimal effect on biology. Only in the past 15 years have researchers begun to appreciate the importance of escape, in terms of the number of genes, their significant expression levels, and their physiological relevance. Genes on the inactivated X rarely reach 100 percent of active-chromosome expression levels; “escape” is usually defined as achieving RNA levels of at least 10 percent of the expression seen in the active X. But it turns out that even a small amount of expression from the Barr body can make a biological difference.

The concept is out there
that genes that escape X inactivation are influencing female-specific physiology.

 —Sundeep Kalantry, University of Michigan Medical School

The first real catalog of escapees, published in 2005, was astonishing for its size, recalls Carolyn Brown, a molecular geneticist at the University of British Columbia in Vancouver who was not involved in the project. Researchers from the Penn State College of Medicine and Duke University forced active and inactive human X chromosomes into rodent cells, creating a panel of hybrid cell lines. Then they looked for mRNAs derived from those human Xs. Ignoring the PARs, these researchers found that while 458 of the genes they measured were silenced on the inactive X in most or all of the cell cultures, 154 others—a whopping 25 percent—were expressed in at least some cells. (This would later turn out to be a much higher rate than in mouse X chromosomes, in which 3–5 percent of Barr body genes escape.) 

The diversity of escapees was also readily apparent, with genes for mitochondrial enzymes and blood cell surface proteins among the many on the list. Many escaped genes are transcriptional regulators, says Brown, giving them ample opportunity to influence biology. 

In a 2017 survey of escapees in human tissues, computational biologist Taru Tukiainen, then a postdoc at Massachusetts General Hospital and the Broad Institute, confirmed the escape rates seen in rodent-human hybrid cells and described remarkable variability in expression levels of individual escapees. 

In an analysis of transcriptomes from 940 individual blood cells from four different women, she assessed the activity of 165 X-linked genes—those expressed at high enough levels to be identified by single-cell RNA sequencing—including several PAR genes. She found that only 129 were reliably silenced. Others were silenced only in some of the women, or in some of their cells. 

For example, the PAR1 gene ZBED1 was expressed by the inactive X in all four women, making up 20–50 percent of total ZBED1 mRNAs. The known escapee MSL3, which encodes a nuclear protein thought to be involved in chromatin organization and regulation of transcription, was expressed equally from both X chromosomes in one woman, but transcripts from the inactive X only made up 10–20 percent of MSL3 RNAs in two other women.

To interrogate whether escape might contribute to sex differences, Tukiainen analyzed transcriptomes from men and women who participated in the National Institutes of Health’s Genotype-Tissue Expression (GTEx) Project. Her dataset covered more than 5,500 transcriptomes of 29 different tissues from 449 people. Of the 82 known escapee genes she analyzed, 74 percent showed considerable differences in expression between men and women in at least one of the 29 tissue types. 

Tukiainen also came across something surprising in the GTEx database. Normally, X inactivation happens randomly to one X chromosome in each cell, so most XX individuals are a mosaic of cells with one or the other X shut down. One of the project’s tissue donors didn’t seem to be a mosaic. “This one person really stood out,” recalls Tukiainen. Across all her tissues, “the active X chromosome was always the same.” 

Tukiainen checked her data several times to confirm it wasn’t an artifact. Such skewing of the mosaic can happen naturally in blood, especially as women age and some cell lineages die out, but this was an extreme example, in a woman who died of asphyxiation at the age of 21. And it was a lucky break for Tukiainen’s research. She could easily identify escapees by looking for genes that expressed the alleles of both X chromosomes, to compare expression patterns in 16 different tissues. 

In this woman, 23 percent of 186 genes assessed in the GTEx dataset escaped. Of those, 43 percent escaped in a majority of tissues, while 11 escaped in only one tissue. “The thing that really amazed me is how heterogeneous X inactivation escape really is,” says Tukiainen, who now runs a lab at the University of Helsinki’s Institute for Molecular Medicine Finland.

Genetic clues to why X genes escape

Escapee lists provide important clues to what Brown says is now the big question in the field: How do these genes avoid or reverse silencing?

Researchers suspect there are genetic markers that help maintain silencing or promote escape, and patterns in the sequences of known escapees suggest what the markers might be. For example, says Brown, sequences rich in long, repeating sequences called LINEs tend to be silenced, while those with shorter repeat elements (SINEs) are more likely to escape. In addition, geneticist Christine Disteche and colleagues at the University of Washington reported in a 2015 paper that a transcription factor called CCCTC-binding factor (CTCF) clustered around escapees and their promoters, leading the researchers to propose that the factor might promote escape.

In addition to sequence-gazing, scientists use genetic manipulation experiments to move known escapees and normally silent sequences around the X chromosome, in search of factors that the genes carry with them. For example, molecular biologist Laura Carrel and colleagues at the Penn State College of Medicine relocated the mouse escapee Kdm5c (which also evades silencing in humans), along with flanking genes that are normally inactivated, to parts of the mouse X chromosome that are normally silenced. Kdm5c persisted in escaping. The flanking genes, meanwhile, maintained their silence.9 The individual genes appeared to rely on local sequences that block or favor escape, rather than depending on their position on the X chromosome to control expression.

Further experiments pointed to genetic sequences that act as barriers between expressed and silenced regions. In a different cell line with an out-of-place sequence that included Kdm5c, the DNA was truncated, deleting everything after Kdm5c including the downstream, normally inactivated gene. Kdm5c still escaped, but so did three genes downstream of its insertion site. That is, once one gene found its voice, it passed on the ability. Normally, scientists presume, some barrier between genes prevents this.

X Escape Patterns in Humans

There are about 1,150 known genes on human X chromosomes. Genes in the pseudoautosomal regions (PARs) at an X chromosome’s tips pair with corresponding genes on the Y chromosome in XY cells and are expressed from both X chromosomes in XX cells. Outside of these regions, hundreds of genes are inactivated to avoid giving XX cells a double dose of genes that XY cells have in only one copy. However, dozens of X-chromosome genes escape this silencing.


CTCF might serve as one such barrier when it binds between escapees and silenced genes, as Disteche’s group reported in a 2005 paper. Indeed, CTCF binds just upstream of Kdm5c. Brown’s group and collaborators found evidence for another candidate escape marker when analyzing human datasets: binding sites for the YY1 transcription factor were common near escapees.

The escape and boundary signals appear to be similar in mouse and human DNA, Brown’s group found in experiments mirroring Carrel’s. The team isolated a snippet of human X DNA containing three genes: the escapee RPS4X, the normally silenced ERCC6L, and the variably expressed CITED1. They placed this segment in a mouse X, at a silenced region, and the human genes’ expression patterns held. The results reinforced Brown’s belief that there must be pro-escape signals in the escapee genes—signals that mouse cells can read. 

But it’s still difficult to predict from sequence alone if a gene will escape or not. For every potential marker, there are genes that buck the trend, remaining silent despite suspected escape signals. “We haven’t come up with a rule,” says Brown. “It’s also possible that there’s some unique combination of elements, so not just a single element, but multiple ones.”

To find the rules, she’s now dissecting RPS4X, taking out suspect elements one by one to learn how they influence escape. She’s also attempting to assemble a wholly artificial escapee, starting with a promoter and adding candidate escape elements such as YY1 binding sites. Then, Brown says, “we can actually test whether or not they’re going to favor escape.”

How X genes mediate silencing

Some Barr body genes control X inactivation itself. XIST, first described in 1991 by Brown, is one; the long noncoding RNA transcribed from the gene is still considered the primary factor in X inactivation. But it might not be the only involved gene located on the inactive X chromosome; Kalantry recently discovered evidence for another.

To study the function of Xist (the murine version of the gene) in mouse embryos, Kalantry took advantage of the gene’s own regulatory system. If RNA polymerase reads Xist in the antisense direction, it produces a transcript called Tsix (Xist backwards), somehow blocking production of Xist RNA. Kalantry used a strain with a mutation that cut short Tsix transcription, giving Xist full rein. This triggered the inactivation of both X chromosomes in cells of female embryos and stem cell lines. Not surprisingly, this killed those cells. 

But the Tsix mutation did not effectively shut down the sole X chromosome in XY cells. Kalantry reasoned that some factor on the Y chromosome might block X inactivation. That hypothesis proved false when he looked at embryonic stem cells with only one X and no Y, what’s called an XO genotype, and saw the same pattern: the X chromosome remained active

Kalantry then hypothesized that something in the XX cells permitted inactivation. These cells already contained one properly inactivated X, plus the active one Kalantry was trying to repress. Could some escapee from the inactive X be enabling inactivation of one or both chromosomes?

In a preprint posted on bioRxiv in 2017, Kalantry and colleagues described a candidate for this silencer: Kdm5c, the escapee from Carrel’s studies. It encodes a transcription factor that can turn genes on or off by removing methyl groups from the local histones. That fits what Kalantry believes the KDM5C protein does: turn Xist on, but other X genes off.

When the researchers knocked out Kdm5c in mouse embryos, females showed little to no X inactivation, and subsequently died. With female cells that had only one copy of Kdm5c, inactivation took place, but some normally silenced genes on the inactive X began to speak up. Engineering XY embryonic stem cells to overexpress the gene resulted in the inactivation of their only X, and they subsequently died. Thus, in both male and female embryos, full X inactivation appears to require more than one gene’s worth of KDM5C.

To follow KDM5C in the cell during X inactivation, the team watched it using immunofluorescence. As Xist RNA coated the X chromosome, KDM5C started to show up as well. Once the Barr body was formed, most KDM5C disappeared, but further experiments showed that a bit of the protein persisted around promoters and other regulatory elements.

Kalantry posits that KDM5C could regulate inactivation as follows: Early in embryogenesis, before inactivation, XX embryos make a high dose of KDM5C, expressed from both chromosomes. At this concentration, the transcription factor activates Xist, while helping to silence other genes on one X chromosome. After the Barr body is formed, Kdm5c continues to escape, maintaining that silence. Now, he’s looking to see if the actions of KDM5C are similar in other species. 

Escapee function for good and ill

Besides seeking to better understand the process of X inactivation, researchers studying escape want to know what all these unsilenced genes are doing in the cell, and whether they matter for human health. Escapee expression, which would create protein level differences between XX and XY individuals, may explain why some conditions are more prevalent in one sex or the other.

Multiple sclerosis (MS), a disease in which the immune system attacks the insulating myelin around nerve fibers, is about three times more common in women than men. Sex hormones seem to explain some of the increased risk in women, but not all of it. Rhonda Voskuhl, director of the University of California, Los Angeles (UCLA), MS program, wondered if the XX or XY genotype might contribute.

Plotting an Escape

The silencing of the one X chromosome in XX cells is mediated by XIST, a long noncoding RNA that is randomly transcribed from only one X early in development. It coats the DNA and shuts down gene expression on that X. For genes to escape, researchers hypothesize, certain sequences on the X, so-called escape elements, attract proteins that help nearby genes evade silencing. In addition, sequences known as boundary elements and their associated proteins seem to act as divisions between active and quiet regions. The identities of these regulators haven’t been conclusively pinned down, but there are several suspects.

© n.r.fuller, sayo-art, llc

Binding sites for YY1 are common at escapee promoters. YY1 is a transcription factor that may work in part by influencing histone acetylation, which influences gene expression.

Binding sites for CTCF (CCCTC-binding factor) are enriched in escapees as well as at the boundaries between silenced and active regions. The mechanism by which this transcription factor and manager of chromosome looping supports escape is unclear.

Topologically associated domains (TADs) distinguish segments of the genome and typically contain genes that physically interact. Researchers haven’t yet identified the nucleic acid sequences and proteins that set TAD boundaries, but the domains may help delineate silenced and escaping regions on the X.

See full infographic: WEB | PDF

She partnered with UCLA geneticist Arthur Arnold to work with a set of mouse lines he’d created to disassociate the effects of sex chromosomes from those of gonads and sex hormones. The lines include XX animals with male gonads and hormones, and XY mice with female gonads and hormones. Together with male and female mice with the usual combinations of sex chromosomes, hormones, and gonads, these form the four core genotype models. Using them, researchers can control for the effects of sex hormones to elucidate the role of sex chromosomes alone. 

Voskuhl immunized the four core mice against a component of myelin so their immune systems would attack it, a standard method to create an MS-like condition. No matter their gonads, XX mice were more susceptible to developing the condition. In a 2019 paper, Voskuhl reported that T cells from XX mice expressed more of a particular escapee, the gene for the transcription factor KDM6A. Knocking out the Kdm6a gene diminished neuroinflammation and protected mice from developing MS-like symptoms. “This is a really nice gene to target for therapy,” says Voskuhl. “It’s a master switch for so many genes involved in autoimmunity.” 

We’re just taking the first step in really understanding the full complexity of this phenomenon.

 —Taru Tukiainen, Institute for Molecular Medicine Finland

In fact, the diabetes drug metformin is known to regulate KDM6A, blocking its demethylase activity to increase gene silencing across the genome. In an MS mouse model, metformin treatment minimized symptoms. Voskuhl plans to test the medicine further in mice and in people with MS, and says she thinks it might treat other autoimmune conditions, too.

KDM6A escape isn’t all bad news for women’s health. The gene is also a tumor suppressor.

Men are about 20 percent more likely to get cancer than women, and in some cancers, the rate in men is more than double that in women. Scientists had long assumed that was because men were more likely to smoke, drink, or toil in factories suffused with toxic chemicals. But recent epidemiology suggests those aren’t the only factors.

Several years ago, Andrew Lane, a physician-scientist at the Dana-Farber Cancer Institute in Boston, noticed—as others had—that certain tumor suppressors located on the X chromosome were more often mutated in male cancer patients than in female patients. In the case of KDM6A, for example, mutations are more common in men’s cancers than in women’s. 

At first, Lane couldn’t fathom how this could be so. With one X inactivated, shouldn’t one KDM6A mutation be sufficient to inactivate the tumor suppressor? The “Aha” moment came a year or two later, when he discovered Brown’s and Disteche’s work describing X escape on PubMed. Lane hadn’t heard of X escape before, but suddenly he wondered—might escapees be protecting women?

Inspired, he developed what he calls the Escape from X-Inactivation Tumor Suppressor, or EXITS, hypothesis: by expressing spare tumor suppressors from the inactive X, women may stave off cancer. To test the idea, Lane teamed up with cancer genomics experts at the Broad Institute of MIT and Harvard to analyze cancer mutations from 4,126 patients, with 21 different tumor types, from Broad data and the National Cancer Institute’s Cancer Genome Atlas. The researchers scanned all 23 chromosome pairs for genes that skewed toward more mutations in men than in women and got six hits, including KDM6A. Four were known tumor suppressor genes, and all were on the X chromosome. 

Of the six X-linked genes identified in Lane’s study, five had been seen escaping before. The sixth, a gene for a chromatin remodeler and tumor suppressor called ATRX, was a surprise. It had never been listed as an escapee. But Lane found hints, in GTEx data, that ATRX can escape in brain tissue. The results support the idea that having a second, expressed copy of these tumor suppressor genes can save women from developing cancer, whereas men, with only one X, can succumb to disease after a single mutation in those genes. 

Researchers predict that there are more escapees left to find, across different types of cells, tissues, individuals, and probably people of different ages. “We’re just taking the first step in really understanding the full complexity of this phenomenon,” Tukiainen says. “There is definitely a lot more to explore so we can complete the picture.”

Amber Dance is a freelance science journalist living in the Los Angeles area. Read her work or reach out at AmberLDance.com.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?