The children were tiny, often resisting their parents' entreaties to eat. Their skin stretched over jutting cheekbones; toddlers might weigh what a healthy 6-month-old would. Back in the mid-1970s, when Madeleine Harbison was in medical school, her mentor, a pediatric endocrinologist at Massachusetts General Hospital in Boston, urged her to consider what was stunting their growth.
"There's something very special about this group of children that we don't understand," the mentor, John Crawford, told her. "I want you to figure out what it is, and figure out how to make them grow."
Deepening the mystery for Harbison was that some of the babies afflicted with this syndrome, Silver-Russell, had an unaffected identical twin. Silver-Russell syndrome had a genetic component, but the twins suggested that it wasn't a conventional hereditary disease, as identical twins carry carbon copies of each other's DNA. So what was it?
Harbison finished her training in pediatric endocrinology. Crawford, then near retirement, began directing his Silver-Russell patients to her office on New York City's Upper East Side, a leafy walk from Central Park. She worked alone, sometimes corresponding with Crawford, prescribing growth hormones, appetite stimulants, and other medications. "Treating these children is like witch work, and I'm the witch," she says now. That was especially true in those early days as her therapeutic blueprint took shape.
Harbison was unaware that an ocean away, a doctor in Paris had his own medical mystery: children who were the mirror image of those with Silver-Russell. Babies with Beckwith-Wiedemann syndrome were born unusually large, often with oversized body parts. They were also vulnerable to childhood cancers of the kidney and liver. Like Silver-Russell, this syndrome operated outside the rules of Mendelian disorders like cystic fibrosis, hemophilia, Tay-Sachs, and others that result when parents harbor an abnormal genetic sequence and pass it along to their offspring.
It would be years before scientists could read the instruction manuals for Silver-Russell and Beckwith-Wiedemann. Driven by advances in DNA analysis, they found that these two syndromes, along with several others, are imprinting disorders. They arise in a unique subset of genes in which, after conception, the DNA in the embryo that came from the mother is expressed differently than the DNA from the father.
The last several years have seen imprinting disorders emerge from the shadows, and with them a deeper appreciation for the human genome's ability to modulate gene expression in the earliest stages of development. After decades of anecdotes and experimentation, Harbison's treatment for Silver-Russell is finally being tested in a clinical trial. An international registry is enrolling hundreds of patients with Beckwith-Wiedemann and banking their tissue for further study. In the lab, scientists are probing the genetics of their patients and trying to link those findings to a child's health. Delving into imprinting is also yielding insights into the scaffolding of the genome, and the ways in which parents put a stamp on gene expression, influencing health and disease in their children.
Studying these rare diseases "can open a new way to understand the imprinting phenomenon, to see how, in the beginning of the development of the embryo, the embryo answers to stimuli" that regulate how its genes behave, says Giovanni Battista Ferrero, a pediatrician at the University of Turin in Italy. "That is one of the next big challenges in science."
Origins of imprinting
The origins of imprinting probably date back about 150 million years to early mammals. Imprinting, many scientists believe, reflects competition between a mother's interests and a father's when it comes to gestating the offspring. A mother wants a fetus that doesn't grow too big, so she can survive the pregnancy. A father wants the opposite: a fetus that becomes a strapping baby and, later, a strapping adult who hoards resources and spreads his genes to new progeny.
Essentially, imprinting means that in some places along the human genome—about 100 genes in all—the way DNA behaves depends on which parent passes it to the offspring. "The DNA you inherit from your mother and your father are not quite the same" in these imprinted genes, explains Andrew Sharp, a geneticist at the Icahn School of Medicine at Mount Sinai in New York City. "Even though they may have identical sequences," the way the DNA is expressed is different.
The jockeying plays out at a molecular level in all of us soon after conception. Some of the genes in sperm and egg cells have chemicals called methyl molecules that attach to them, a process called methylation; these molecules can either activate or silence a gene when the sperm and egg DNA unite in an embryo.
In some cases, the mother's copy of the gene is activated, and the father's silenced. In others the opposite is true. The function of each of the dozens of human imprinted genes isn't yet known, but many appear to guide metabolism and growth prior to birth.
When imprinting goes awry—and researchers don't understand yet why that happens—the outcome can be health problems in the baby. One leap forward came in 1991, when the first imprinted genes were reported in mice, says Marisa Bartolomei, one of the co-discoverers, who was a postdoctoral fellow at the time and is now at the University of Pennsylvania. Back then, she says, hunting for imprinted genes "was renegade science."
As imprinted rodent genes were uncovered, geneticists wondered whether imprinting could explain the puzzle of two apparently unrelated diseases in people, Prader-Willi and Angelman syndromes. Each affects at most one in every 15,000 babies, and children with the syndromes are nothing alike: Those with Prader-Willi are short, have delayed puberty, and eat excessively, whereas those with Angelman have severe developmental delays, and often have epilepsy and a small head size.
Yet in both syndromes, the genetic flaw—a bit of missing DNA on chromosome 15—looked the same. As geneticists studied these children, they learned something remarkable. "When you had the deletion on the father's chromosome, you had Prader-Willi, and when on the mother's, you had Angelman," says Karen Temple, a medical geneticist at the University of Southampton in the United Kingdom. Maybe the deleted region contained imprinted genes, and thus had different effects depending on how the genes were expressed. But genomic tools were rudimentary at the time, and scientists couldn't confirm the idea.
In 1993, three researchers spanning the globe—Anthony Reeve at the University of Otago in Dunedin, New Zealand, Rolf Ohlsson at the Karolinska Institute in Stockholm, and Andrew Feinberg at Johns Hopkins University in Baltimore, Maryland—independently discovered the first imprinted gene in humans. Feinberg was treating a baby boy who had Beckwith-Wiedemann and developed Wilms tumor, a childhood kidney cancer that's commonly associated with it. Most patients with Beckwith-Wiedemann have no family history of the syndrome, but this child had an aunt and grandmother with features of it, suggesting he might be one of the rare inherited cases. Patterns of inheritance often provide clues for geneticists trying to home in on a gene. If they could find an affected gene in the boy's family, it might lead researchers to the causes of more common, sporadic cases of Beckwith-Wiedemann, too.
Feinberg focused on a region on chromosome 11, because a handful of affected patients had abnormalities there, such as DNA that was repeated or deleted. One gene in that DNA stretch especially piqued his interest: IGF-2, which controls growth and was imprinted in mice. Was it also imprinted in people?
Ultimately, Feinberg found that the answer was yes. A healthy person has two copies of the IGF-2 gene, one from each parent; the copy from the father is naturally expressed and the one from the mother is silenced. But in Beckwith-Wiedemann, children may receive two copies that look like their father's chromosome 11—with IGF-2 and extra "grow" signals along for the ride. Alternately, disrupted imprinting can cause a blunting of "don't grow" signals. Regardless of the mechanism, the outcome is overgrowth in the baby.
A decade later, a pediatric endocrinologist in Paris, Irène Netchine at Pierre and Marie Curie University, working with her mentor, Yves Le Bouc, unraveled the mystery of Silver-Russell. Le Bouc was the Parisian doctor probing Beckwith-Wiedemann in those early days as Harbison's career was just getting started. While studying Beckwith-Wiedemann, Le Bouc and Netchine wondered what would happen if the gene expression patterns were flipped. If Beckwith-Wiedemann could arise when both copies of IGF-2 were expressed, what would happen if imprinting went wrong in the opposite direction, and both copies were inactivated? With their colleagues, they published the answer in Nature Genetics in 2005: It was Silver-Russell syndrome.
The disorders were already known to be "clinical mirrors" of each other, with one causing overgrowth and the other undergrowth, Netchine says. Now, they knew that the molecular mechanism was a "mirror" as well.
Advances in genomic technology
As genomic technology has advanced, so has our understanding of imprinting disorders. In the early 1990s, researchers were often only able to find glaring chromosome deletions and were stuck trying to infer how the deletions affected the workings of cells. Now, they can easily and cheaply measure gene imprinting through the chemical changes responsible for it. That's because when a methyl molecule is anchored to DNA, that DNA has altered properties, including a different molecular weight and a different sensitivity to certain enzymes.
Scientists are also exploring the consequences of an unusual feature of imprinting disorders like Beckwith-Wiedemann: Only some of the body's cells are abnormal, a phenomenon known as mosaicism. That's because the imprinting error usually happens after conception, when the blastocyst has already begun dividing. It's only in the last few years that genomic technology has been able to pluck 1% or 2% of abnormally imprinted cells out of a patient's sample—a level sufficient for symptoms to take hold.
This was on vivid display for Jennifer Kalish, a pediatric geneticist at the Children's Hospital of Philadelphia and the University of Pennsylvania, in the patient who got her hooked on Beckwith-Wiedemann 6 years ago as a medical resident: a toddler whose left side was a shade bigger than her right.
Despite this clue, the little girl had almost none of the classic Beckwith-Wiedemann features, like an oversized tongue that requires surgery to reduce it, low blood sugar, or an enlarged liver and kidneys. Except for the subtle body size discrepancy, she looked perfectly healthy. Kalish ran a blood test: That was normal, too, with chromosome 11—the site of the abnormality in Beckwith-Wiedemann—registering as it should.
Then Kalish's young patient developed a rare adrenal gland tumor associated with the syndrome. After the tumor was surgically removed, Kalish asked her hospital's genomics lab to run a sophisticated test on chromosome 11 to see whether there was one copy of the chromosome that looked like it came from mom and one that looked like dad's, or two copies that were methylated like the father's. Seventy percent of the tumor cells carried that signature. A skin biopsy on the girl's larger leg found 5% of cells in that sample had the same abnormality, even though her blood sample was clean. The child, now a healthy 8-year-old, had the imprinting disease and had nearly slipped through the cracks. "Just by looking at a child outside," says Kalish, "you can't tell what's going on" under the skin.
The more scientists like Kalish probe the genomics of imprinting diseases, the more complicated the picture gets. In Beckwith-Wiedemann, it's the cancer risk—which is highest before age 8—that terrifies families the most. We now know that different imprinting defects on chromosome 11 can lead to the syndrome. Last summer, Italian and Dutch groups separately published papers suggesting that children are at wildly different risks for cancer depending on which defect they carry—with the risk ranging from less than 3% to about 25%—and argued that for lower-risk children, screening might not make sense.
But that question is far from settled. If screening drastically improves a child's outcome, then it becomes easier to justify, Kalish argues. Furthermore, even low-risk kids have a higher chance of certain cancers than those without the syndrome. One on Kalish's mind is 1.5-year-old Finn Miller, who was suspected of having Beckwith-Wiedemann while in utero and was born 6 weeks early weighing almost 4 kilograms. Genetic testing confirmed the diagnosis, but it also reassured his parents: Along with about half of Beckwith-Wiedemann patients, Finn was considered unlikely to get cancer. Then when he was 8 weeks old, doctors found a tumor on his liver. After surgery and chemotherapy he's a healthy toddler, but his story highlights one of the central mysteries of this disorder, Kalish says: Why do patients develop cancer, how are specific imprinting defects driving it, and can it be prevented? To tackle these questions, she recently opened an international registry to track Beckwith-Wiedemann patients and gather tissue samples. She has several hundred participants so far spanning 16 countries.
Silver-Russell, too, is a focus of study. For years Harbison, who works at Mount Sinai, has prescribed patients a mix of growth hormones and, off-label, a type of breast cancer drug called an aromatase inhibitor that she believes gives a stunted child several more years of growth. The aromatase inhibitor is now being tested in a clinical trial in France, led by Netchine, who has become a close collaborator and friend of Harbison's.
We've still got an awful lot to learn about what one generation gives to the next.
A second study, led by Le Bouc, aims to get at one of the biggest mysteries of imprinting disorders—their link to in vitro fertilization (IVF). About a decade ago, three different groups found that a disproportionate number of affected children, particularly with Beckwith-Wiedemann and Angelman syndromes, had been conceived through IVF. The connection makes sense: In IVF, the embryo is left to divide initially in a culture dish, and "without realizing it," Temple believes, "you take away some of these essential proteins or don't provide the environment [the embryo] needs to maintain these very crucial marks" governing gene expression. That said, the risk is thought to be modest; Le Bouc's team found that about 4% of children with imprinting disorders were conceived by IVF, compared with 1.3% in the general population.
The trial includes about 500 babies who were either conceived spontaneously or through IVF. The scientists are comparing various imprinted spots in the children's genomes and expect to report results later this year. But because there are only a couple hundred families in each group, it may be tough to hit on a clear message.
Although they're rare, both Beckwith-Wiedemann and Silver-Russell raise other broad questions. To Kalish, the patients who are diagnosed with the syndrome after they develop cancer, like the little girl whose adrenal tumor set off alarm bells, have her wondering whether the incidence of Beckwith-Wiedemann is higher than reported. And increasingly, scientists are finding links between cancer in the general population and aberrant imprinting.
At the Bellvitge Biomedical Research Institute in Barcelona, Spain, epigeneticist David Monk is looking into an altogether different question inspired by the tiny Silver-Russell children: how imprinting might affect fetal growth in babies without the syndrome. He has collected 352 placenta samples from extremely small or premature newborns from hospitals around the city. Intriguingly, DNA in many of the placenta samples has the same imprinting abnormality seen in Silver-Russell, but a dampened down version of it—perhaps a smaller stretch of the gene is affected, or gene expression is less abnormal. And just as intriguingly, the vast majority of these children are now perfectly healthy, suggesting that once the placenta and its abnormal imprints are removed, the baby can catch up on growth. Monk wonders how variations in imprinting regulate, or dysregulate, fetal growth. It's probably not a coincidence that variations in imprinted genes are linked to obesity, height, and type 2 diabetes, he says.
Temple is reaching even further back, to the first moments of development when imprinting occurs, and looking for clues as to what can make it go awry. She and her colleagues are studying whether the imprinting process is affected by genetic variations carried by a woman, specifically those that govern the protein content of her eggs. It's a legacy from the mother that may leave a lasting mark—quite literally—on her offspring.
"We've still got an awful lot to learn," she says, "about what one generation gives to the next."