ZURICH, SWITZERLAND—The children living in SOS Children's Villages orphanages in Pakistan have had a rough start in life. Many have lost their fathers, which in conservative Pakistani society can effectively mean losing their mothers, too: Destitute widows often struggle to find enough work to support their families and may have to give up their children.
The orphanages, in Multan, Lahore, and Islamabad, provide shelter and health care and send kids to local schools, trying to provide "the best possible support," says University of Zurich (UZH) physician and neuroscientist Ali Jawaid. "But despite that, these children experience symptoms similar to PTSD [post-traumatic stress disorder]," including anxiety and depression.
Beyond these psychological burdens, Jawaid wonders about a potential hidden consequence of the children's experience. He has set up a study with the orphanages to probe the disturbing possibility that the emotional trauma of separation from their parents also triggers subtle biological alterations—changes so lasting that the children might even pass them to their own offspring.
That idea would have been laughed at 20 years ago. But today the hypothesis that an individual's experience might alter the cells and behavior of their children and grandchildren has become widely accepted. In animals, exposure to stress, cold, or high-fat diets has been shown to trigger metabolic changes in later generations. And small studies in humans exposed to traumatic conditions—among them the children of Holocaust survivors—suggest subtle biological and health changes in their children.
The implications are profound. If our experiences can have consequences that reverberate to our children or our children's children, that's a powerful argument against everything from smoking to immigration policies that split families. "This is really scary stuff. If what your grandmother and grandfather were exposed to is going to change your disease risk, the things we're doing today that we thought were erased are affecting our great-great-grandchildren," says Michael Skinner, a biologist at Washington State University in Pullman.
Skinner's own research in animals suggests changes to the epigenome, a swirl of biological factors that affect how genes are expressed, can be passed down through multiple generations. If trauma can trigger such epigenetic changes in people, the alterations could serve as biomarkers to identify individuals at greater risk for mental illness or other health problems—and as targets for interventions that might reverse that legacy.
But proving that emotional trauma, as distinct from physical stress, can be passed on to subsequent generations in people is a challenge. "The difficulty … is being able to disentangle what comes through social inheritance—which must be massive—and what doesn't," says neuroscientist Johannes Bohacek of ETH Zurich. "The jury is still out on humans."
Some of the field's biggest names also worry that the idea could have dangerous consequences. Rachel Yehuda of the Icahn School of Medicine at Mount Sinai in New York City studied the children of 40 Holocaust survivors and found lower baseline levels of the stress hormone cortisol as well as a distinctive pattern of DNA methylation, an epigenetic marker. But in a paper last year, she said it would be "premature" to conclude that trauma causes heritable changes, adding that hyped media coverage could promote a misleading narrative of hopelessness, suggesting that one generation's trauma permanently scars later generations.
"There's a lot of overinterpretation of initial results," says Columbia University biologist Katherine Crocker, who studies nongenetic inheritance in crickets. "What is out there in the public mind about epigenetics probably can never be proved."
To investigate, Jawaid is collecting blood and saliva samples from the Pakistani orphans and from classmates who live with parents. As a researcher in the lab of Isabelle Mansuy of UZH and ETH Zurich, he hopes to learn whether the trauma of loss and forced separation has left identifiable marks at the cellular level. But to really prove transgenerational inheritance, he'd have to study the orphans for years—until they have children of their own. That's why Mansuy herself has turned to mice.
One recent afternoon, Mansuy donned a fresh lab coat and blue sanitary booties and gently cracked the door of a darkened room at her lab at UZH. A powerful smell—something like dog chow mixed with animal musk—wafted out on a gust of warm air. Inside were hundreds of mice in 40 breeding cages. "We keep it dark during the day to preserve their circadian rhythm when we work with them," Mansuy says in a hushed voice. "This is our 31st cohort."
The idea Mansuy is exploring—that not all inherited characteristics are rooted in DNA—dates back more than half a century. Tantalizing early results came from maize, in which plants with identical DNA had variations in traits such as kernel color that persisted for hundreds of generations. The work was initially controversial, as geneticists saw it as a revival of the non-Darwinian ideas of 19th century scientist Jean-Baptiste Lamarck.
But experiments in many organisms suggested epigenetic inheritance was real. In simple creatures like Caenorhabditis elegans worms, researchers found that genes turned off once by altering the RNA they produced remained silenced for 80 generations or more. Some examples were even more dramatic: Water fleas exposed to the scent of a predator have offspring with spiky, armored heads. And in mice, researchers including Skinner found that parents exposed to altered diets, low temperatures, or toxins had descendants with behavioral changes and weight gain.
Epidemiological studies of people have revealed similar patterns. One of the best-known cases is the Dutch hunger winter, a famine that gripped the Netherlands in the closing months of World War II. The children of women pregnant during the food shortages died earlier than peers born just before, and had higher rates of obesity, diabetes, and schizophrenia. Studies of other groups suggested the children of parents who had starved early in life—even in the womb—had more heart disease. And a look last year at historical records showed the sons of Civil War soldiers who had spent time as prisoners of war (POWs) were more likely to die early than the sons of their fellow veterans. (The researchers controlled for socioeconomic status and maternal health.)
But the human studies faced an obvious objection: The trauma could have been transmitted through parenting rather than epigenetics. Something about the POW experience, for example, might have made those veterans poor fathers, to the detriment of their sons' lives. The psychological impact of growing up with a parent who starved as a child or survived the Holocaust could itself be enough to shape a child's behavior. Answering that objection is where mouse models come in.
Mansuy began in 2001 by designing a mouse intervention that re-creates some aspects of childhood trauma. She separates mouse mothers from their pups at unpredictable intervals and further disrupts parenting by confining the mothers in tubes or dropping them in water, both stressful experiences for mice. When the mothers return to the cage and their pups, they're frantic and distracted. They often ignore the pups, compounding the stress of the separation on their offspring.
Mansuy says the mice's suffering has a purpose. "We're applying a paradigm that is inspired by human conditions," she says. "We're doing it to gain understanding for better child health."
Unsurprisingly, the pups of stressed mothers displayed altered behavior as adults. But to Mansuy's surprise, the behavioral changes persisted in the offspring's offspring. Initially, she thought this could be a result of the offspring's own behavior: Mice traumatized as pups could have been bad parents, replicating the neglect they experienced in childhood. Thus they might simply be passing on a behavioral legacy—the same lasting psychological effect that might explain such findings in humans.
To rule out that possibility, Mansuy studied only the male line, breeding untraumatized, "naïve" female mice with traumatized males, and then removing males from the mother's cage so that their behavior did not impact their offspring. After weaning, she raised the mice in mixed groups to prevent litter mates from reinforcing each other's behaviors.
Her lab repeated the procedure, sometimes going out six generations. "It worked immediately," she says of the protocol. "We could see that there were symptoms [in descendants] that were similar to the animals that were themselves separated." Descendants of stressed fathers displayed more risk-taking behavior, like exploring exposed areas of a platform suspended off the ground. When dropped in water, they "gave up" and stopped swimming sooner than control mice, an indicator of depressivelike behavior in mice.
Mansuy is "definitely a pioneer," says Romain Barrès, a molecular biologist at the University of Copenhagen. Other researchers have developed conceptually similar models, for example giving male mice altered diets or exposing them to nicotine and tracing metabolic and behavioral changes out for generations.
"If you're asking, ‘Does the experience of the parent influence the process of development?’ the answer is yes," says epigenetics researcher Michael Meaney at McGill University in Montreal, Canada, whose own studies have shown that differences in maternal care can have epigenetic effects on brain development. "Isabelle and others have documented the degree to which the experience of the parent can be passed on. The question [is] how."
Three massive freezers down the hall from Mansuy's office are filled with samples of mouse blood, liver, milk, microbiome, and other tissues. These serve as a −80°C archive of more than 10 years of data. Mansuy estimates she's collected behavioral data and tissue samples from thousands of mice altogether.
She hopes the biological markers of trauma are hidden in those freezers, waiting to be revealed. Many of the early mammalian epigenetics studies focused on DNA methylation, which "tags" DNA with methyl groups that switch genes off. But those changes seemed unlikely to be directly inherited: In mammals, methylation is mostly erased when egg and sperm come together to form an embryo.
Mansuy and others still think methylation could have some role. But they are also zeroing in on tiny information-rich molecules called small noncoding RNAs (sncRNAs). Most RNA is copied from DNA, and then acts as a messenger to instruct the cell's ribosomes to produce specific proteins. But cells also contain short strands of RNA that don't produce proteins. Instead, these noncoding RNAs piggyback on the messenger RNAs, interfering with or amplifying their function, thus causing more or less of certain proteins to be produced.
Mansuy and others think stress may influence sncRNAs, along with the many other biochemical changes it causes, from higher levels of hormones like cortisol to inflammation. They have focused on the sncRNAs in sperm, which may be especially vulnerable to stress during the weeks that newly formed sperm spend maturing in a twisting tube on top of the testes. Later, when sperm and egg come together, altered sncRNAs could modify the production of proteins at the very beginning of development in a way that ripples through the millions and millions of cell divisions that follow. "Hosts of signals happen as those cells become a zygote," says epigeneticist Tracy Bale at the University of Maryland in Baltimore. "If dad brings small noncoding RNAs that have an effect on mom's RNAs, that can change the trajectory of embryo development."
Bale found evidence that trauma can affect sncRNAs in sperm—and that the effects might be transmitted to offspring. She stressed mice during adolescence by barraging them for weeks at unpredictable intervals, with things like fox odors, loud noises, and bright light. Then, she examined the sncRNAs in their sperm and offspring. She found differences in nine types of sncRNAs, including one that regulates SIRT1, a gene that affects metabolism and cell growth.
She then created RNA molecules with similar alterations and injected them into early-stage embryos. When those embryos grew to adults, they carried RNA alterations like those seen in the sperm. This second generation also had lower levels of corticosterone, the mouse equivalent of cortisol, after a stressful spell inside a tight tube. "If you do the same RNA changes, you produce offspring with the same phenotype," Bale says.
Mansuy found similar RNA changes in her male mice traumatized as pups. They had higher levels of specific sncRNAs, including miR-375, which plays a role in stress response. Mansuy is convinced those molecular changes account for some of the inherited behavioral traits she documented. In one experiment, her team injected RNA from traumatized male sperm into the fertilized eggs of untraumatized parents and saw the same behavioral changes in the resulting mice.
But although the cause, in the form of altered RNA, and the effect, in the form of altered behavior and physiology, are identifiable in mouse experiments, everything else remains maddeningly difficult to untangle, especially in people. "The field has come a long way in the last 5 years," Bale says. "But we don't know what's going on in humans because we don't have a controlled environment."
Still, mouse data in hand, Mansuy has been looking for similar epigenetic changes in people. She analyzed blood samples from Dutch soldiers, collected before and after deployment to Afghanistan between 2005 and 2008. And she's working with clinicians in Nice, France, to examine blood samples from survivors of a horrific 2015 terror attack.
Other researchers had found altered sncRNAs in the blood of the soldiers. In 2017, for example, Dutch researchers showed soldiers exposed to combat trauma had recognizable differences in dozens of sncRNA groups, some of them correlated with PTSD. But Mansuy couldn't find the same kinds of RNA changes that appeared in her lab's mice. That could be because the soldiers' samples were years old, or simply because mice and people are different, showing the limits of mouse models. But Mansuy hopes it means epigenetic changes are sensitive to the type of trauma and when it occurs in the life course. Mice can never perfectly replicate human suffering, but, she says, "the best approach" for research "is to select a population of humans who have gone through conditions which are as similar as possible to our model."
That's where the Pakistani orphans come in. The children's chaotic early years may have some similarities to what the mice in Mansuy's lab experience, she says, including unpredictable separation from their mothers.
Early results are promising. "We have overlapping findings with the mouse model," Jawaid says. In a preprint uploaded last month to bioRxiv, Mansuy and Jawaid documented changes in the levels of fatty acids in the orphans' blood and saliva that mimicked changes in the traumatized mice—as well as similar sncRNA alterations. The presence of similar biomarkers "suggests that comparable pathways are operating after trauma in mice and children," Mansuy says.
In a conceptually similar effort to go from mice to people, biologist Larry Feig at Tufts University in Boston exposed male mice to social stress by routinely changing their cage mates. Their sperm had altered levels of specific sncRNA groups—albeit different ones from those altered in Mansuy's mice—and their offspring were more anxious and less sociable than the offspring of unstressed parents.
Working with a sperm bank, Feig then looked for the same sncRNAs in human sperm. He also asked donors to fill out the Adverse Childhood Experience (ACE) questionnaire, which asks about abusive or dysfunctional family history. The higher the men's ACE score, the more likely they were to have sperm sncRNA profiles matching what Feig had seen in mice.
But this body of research hasn't convinced everyone. Geneticist John Greally at the Albert Einstein College of Medicine in New York City has been a vocal critic of the evidence for epigenetic inheritance of trauma, pointing at small sample sizes and an overreliance on epidemiological studies. For now, he says, "Mouse models are the way to go." He's not yet seen definitive experiments even in mice, he says. "I'd like to see us be more bold and brave and move from preliminary association studies to definitive studies—and be open to the idea that there may be nothing there."
In a darkened room down the hall from Mansuy's office, just outside the mouse breeding area, two cages stand side by side on a table. One is a standard lab mouse enclosure, not much bigger than a shoebox. Wood chip–strewn cages like this are where most lab mice, including most of Mansuy's animals, spend their lives.
Next to it, black-furred, pink-tailed mice scurry up and down in a luxury two-story mouse house, equipped with three running wheels and a miniature maze. Their environment is designed to stimulate their senses and engage more of their brains in play and exploration.
In 2016, Mansuy published evidence that traumatized mice raised in this enriched environment didn't pass the symptoms of trauma to their offspring. The limited data—Mansuy says her lab is now working on an expanded study—suggest life experience can be healing as well as hurtful at the molecular level. "Environmental enrichment at the right time could eventually help correct some of the alterations which are induced by trauma," Mansuy says.
This and a few other studies suggesting epigenetic change is reversible have the potential to change the narrative of doom around the topic, researchers say. "If it's epigenetic, it's responsive to the environment," says Feig, who more than a decade ago found similar effects on brain function across generations by giving mice play tubes, running wheels, toys, and larger cages. "That means negative environmental effects are likely reversible."
In public talks and interviews, Mansuy says she's careful not to promise too much. As confident as she is in her mouse model, she says, there's lots more work to be done. "I don't think the field is moving too fast," Mansuy says. "I think it's moving too slow."