What we know about the fundamental laws of inheritance began to take shape in a monastery garden in Moravia in the middle of the 19th century, when Gregor Mendel patiently cross-bred pea plants over the course of several years, separated the progeny according to their distinct traits, and figured out the mathematical foundations of modern genetics. Since the rediscovery of Mendel’s work a century ago, the vocabulary of Mendelian inheritance—dominant genes, recessive genes, and ultimately our own era’s notion of disease genes—has colored every biological conversation about genetics. The message boils down to a single premise: your unique mix of physiological traits and disease risks (collectively known as your phenotype) can be read in the precise sequence of chemical bases, or letters, in your DNA (your genotype).
But what if—except in the cases of some rare single-gene disorders like Tay-Sachs disease—the premise ignores a significant portion of inheritance? What if the DNA sequence of an individual explains only part of the story of his or her inherited diseases and traits, and we need to know the DNA sequences of parents and perhaps even grandparents to understand what is truly going on? Before the Human Genome Project and the era of widespread DNA sequencing, those questions would have seemed ridiculous to researchers convinced they knew better. But modern genomics has run into a Mendelian wall.
Large-scale genomic studies over the past five years or so have mainly failed to turn up common genes that play a major role in complex human maladies. More than three dozen specific genetic variants have been associated with type 2 diabetes, for example, but together, they have been found to explain about 10 percent of the disease’s heritability—the proportion of variation in any given trait that can be explained by genetics rather than by environmental influences. Results have been similar for heart disease, schizophrenia, high blood pressure, and other common maladies: the mystery has become known as the “missing heritability” problem. Francis Collins, director of the National Institutes of Health, has sometimes made grudging reference to the “dark matter of the genome”—an analogy to the vast quantities of invisible mass in the universe that astrophysicists have inferred but have struggled for decades to find.
Joseph H. Nadeau has been on a quest to uncover mechanisms that might account for the missing components of heritability. And he is finding previously unsuspected modes of inheritance almost everywhere he looks.
Nadeau, who until recently was chair of genetics at Case Western Reserve University in Cleveland and is now director of research and academic affairs at the Institute for Systems Biology in Seattle, has done studies showing that certain traits in mice are influenced by specific stretches of variant DNA that appeared on their parents’ or grandparents’ chromosomes but do not appear on their own. “Transgenerational” genetics, as he calls these unusual patterns of inheritance, fit partly under the umbrella of traditional epigenetics—the idea that chemical changes wrought by environmental exposures and experiences can modify DNA in ways that either muffle a normally vocal gene or restore the voice of a gene that had been silenced. Researchers have begun to find that these changes are heritable even though they alter only the pattern of gene expression, not the actual genetic code. Yet it’s both more disconcerting and more profound to suggest, as he does, that genes an ancestor carried but didn’t pass down can influence traits and diseases in subsequent generations.
Consider the results of an experiment Nadeau and his colleague Vicki R. Nelson published last August. They created an inbred strain of mice and then compared two sets of females that were genetically identical except for one small difference: one set had a father whose Y chromosome came from another strain of mouse and contained a different set of genetic variants. That shouldn’t have affected the daughter mice at all, because females don’t inherit the Y chromosome. But the presence of that uninherited DNA in the previous generation exerted a profound effect on many of the more than 100 traits tested in the two sets of female offspring, whose own DNA was exactly the same. These results, Nelson and Nadeau concluded, suggest that “transgenerational genetic effects rival conventional genetics in frequency and strength.”
In a separate but similarly unsettling line of experiments, Nadeau and his collaborators are finding that the impact of any given gene depends on all the other genes surrounding it. Nadeau is hardly the only scientist to identify these complex gene-gene interactions, but he and his colleagues have created a unique set of genetically engineered mice that is giving them and other scientists unprecedentedly precise tools for dissecting these “situational genetics” to show how the variants in a gene’s molecular neighborhood affect the way it behaves.
Findings like these, taken together, could shed light on the missing-heritability problem, but at the cost of upending the dominance of traditional Mendelian ideas about how inheritance works. Sitting on the outside deck of the Institute for Systems Biology one recent afternoon, munching on a sandwich as seaplanes descended toward the skyline of Seattle, Nadeau recalled giving a talk about all this at a conference several years ago and discovering afterward that a prominent Ivy League geneticist in attendance, whom he declined to name, simply couldn’t get the heretical ideas out of his head. “He came up to me after the talk,” Nadeau recalled, “and said, ‘This can’t be true in humans.’ I ran into him at breakfast the next day and he said, ‘This can’t be true in humans.’ And then when the meeting was over, I ran into him at the airport, and he came up to me and said, ‘This can’t be true in humans.’ ” Or as another leading genome scientist once told Nadeau at a meeting in Europe, “If transgenerational effects happen in humans, we’re screwed.”
That is to say, discovering that his findings apply to humans would decouple a person’s DNA sequence from her or his traits, calling into question much of the work scientists have done to find the genetic sources of complex diseases and develop drugs that target them. At a time when companies are analyzing customers’ DNA for a fee, these ideas would make the results much more difficult to interpret medically and much more complicated to assess when trying to make a diagnosis or calculate disease risk.
Eric J. Topol, who heads genomic research at the Scripps Research Institute in La Jolla, California, agrees that genomics has suddenly gotten a lot more complicated. “There’s a lot of non-Mendelian stuff going on,” he says, “and there’s a lot that we’re going to have to sort out that doesn’t have anything to do with the DNA sequence.”
In 2009, a group of researchers based in the Netherlands published a stunning study on the genetics of human height—stunning because it failed to find much of a genetic component in one of the most obvious of inherited human traits. The group analyzed 54 recently identified genetic locations that statistical analysis suggested were the main contributors to height and discovered that all of them together accounted for only 4 to 6 percent of the height variance in thousands of subjects.
The “missing heritability” in the height study typifies many recent research reports in which large-scale genetic screens, known as genome-wide association studies, have identified a multitude of genes (or at least genetic neighborhoods) that are statistically associated with a biological trait like height or a disease like obesity, yet account for mystifyingly little of its propensity to run in families. What is interesting about Nadeau’s findings is that even though they diminish the significance of single genes and the DNA sequences of individuals, the research preserves—and in some ways increases—the significance of family history, since even the genetic variants that parents and grandparents don’t pass down through DNA seem to influence the traits of their children or grandchildren.
Nadeau, who is silver-haired and cheerful, has been investigating what he sometimes calls “funky” genetic results ever since sophisticated sequencing technologies became available about 10 years ago. Some of those results have been hinted at by traditional epigenetics, which has begun to trace changes that are transmitted from one generation to the next in animals even though the basic DNA sequence remains the same. (For example, researchers have found that rats whose cognitive performance was improved through environmental factors can pass those improvements down to offspring.) But where that field has typically focused on chemical modifications of DNA, Nadeau’s work expands the notion of epigenetics to include genetic effects that may be transmitted by different molecular players: ribonucleic acids (or RNAs), which exert powerful regulatory effects on DNA.
Key evidence for Nadeau’s general views on unconventional modes of inheritance grew out of a dissertation project that one of his students began around 2001. In the long tradition of misguided doctoral advice, everyone told Man-Yee Lam that her project was boring, derivative, and hardly worth doing; for five or six years, nothing in her results suggested otherwise. The focus of the project was testicular germ-cell tumors. It didn’t become clear until much later that the experiment represented the first rigorous demonstration of a transgenerational effect, showing that genetic variations in a parent—even though they were not passed along to offspring—could dramatically change disease risks in those offspring.
Lam set out to see if she could identify interactions between several “modifier” genes—interactions that would increase susceptibility to testicular cancer in mice. She found lots of these interactions (some quite strong), completed her thesis, and graduated. Then, when the group started to write up the results for publication, they noticed something very peculiar: the effects had also occurred in some of the control animals bred from the same original population. In other words, males that had been bred so as not to inherit the disease mutations still had a statistically significant increase in their risk for testicular cancer, simply because the parents possessed a particular genetic variant. The results suggested that there could be patches of DNA in parents that affected the traits of children, even if the children did not inherit this bit of parental DNA.
Even before publication in 2007, Nadeau began describing the findings—to decidedly mixed reviews. He says, “If they were geneticists, there were all sorts of technical [objections] or ‘It’s not fair to talk about this in public. This is just too complicating, too—it’s too everything!’ One even said, ‘Are you trying to ruin genetics?’ ”
Nadeau isn’t trying to ruin genetics, of course, but the other main focus of his research, involving gene-gene interactions in genetically engineered mice, also challenges the assumptions of modern Mendelians. Whereas conventional genomic studies assume that a number of individual genes contribute independently to complex diseases, Nadeau’s group has been investigating how genes can work in concert to produce illness or, surprisingly, suppress it. Certain genetic variants neutralize other disease genes, so that a person’s susceptibility to disease may depend more on the combined effect of all the genes in the background than on the disease genes in the foreground.
If this phenomenon is widespread, it holds significant implications for medicine. While enormous resources are routinely devoted to the search for disease genes, the research on gene-gene interactions—mostly in mice but increasingly in humans—suggests it may be at least as productive to identify protective and neutralizing genetic variants that counteract the effects of pathological variants. Understanding the biology of these protective variants could offer new routes to disease prevention and treatment. The mechanisms through which they exert their effects could even form the basis for new drugs.
To conduct his experiments, Nadeau and his collaborator, genomic pioneer Eric Lander, engineered 22 substrains of a commonly studied mouse strain called Black 6 by systematically replacing a different chromosome in each mouse with the corresponding chromosome from another strain, known as A/J. The idea of all this mixing and matching was to create a highly controlled system for studying gene-gene interactions, in part to determine how much a given gene contributes to the heritability of a disease or trait. By dropping in a “foreign” chromosome while holding everything else constant, the researchers could calculate the influence of each newly introduced gene. As Nadeau and his colleagues inserted one chromosome after another against the otherwise stable background and then measured the genetic effects, they discovered that the extent to which any gene affected the heritability of a given trait was dramatically larger than what more conventional genomic studies would have predicted. The implication is that the potency—and, Nadeau would discover, the action—of disease genes must change with the context created by other genes on other chromosomes.
To illustrate how complicated this idea is, Nadeau hops out of his chair and rushes over to the whiteboard in his office, where he quickly sketches out how these “completely crazy” context-dependent effects can act even within a single chromosome. The experiments focus on a genetic variant they have identified on chromosome 6 in the A/J mice that completely protects the animal against obesity. When they drop the chromosome into Black 6 mice, they too are protected against obesity. But it’s not that simple. When researchers stitch a bit of the DNA from the A/J strain into a large section of chromosome 6 in the Black 6 mice, the resulting mice are obese. When they trim away some of the Black 6 DNA and replace it with more A/J DNA, the resistance gene becomes active and the mice are lean. But when they add even more A/J DNA to this hybrid chromosome, the resistance gene turns off again. This on-off genetics continues even when the researchers add or subtract extremely small portions of chromosome 6. In fact, no matter how small the patch of DNA, nibbling away at it alternately confers and erases resistance to obesity. The reason is not known, but the larger message is that the effect of any variant seems to depend on its genetic surroundings. “We see that effect all the time,” Nadeau says. “All the time! Everywhere, in every trait we look at.”
Nadeau’s group has also begun using these genetically engineered mice to explore transgenerational effects related to obesity. In research published several months ago, David Buchner, a researcher at Case Western Reserve, showed that one of the strains, which possesses a genetic variant that confers resistance to obesity, can pass this resistance to offspring that don’t inherit the variant. The presence of the resistance gene in the paternal line of ancestry—either in the father or in the grandfather—was sufficient to inhibit diet-induced obesity and reduce appetite in mice that were otherwise genetically predisposed to getting fat.
Could humans also experience non-Mendelian forms of inheritance, particularly the complex gene-gene interplay that Nadeau keeps finding in mice?
Several years ago, Eric Topol launched a systematic attempt to study the genetics of elderly people who were in particularly good health. The researchers sought out subjects who met a series of stringent criteria: they had to be 80 or older, free of chronic diseases, and not taking any long-term medications.
Topol initially suspected a Mendelian explanation for their medical good fortune: he figured that they’d managed to avoid inheriting variant genes, or alleles, known to be associated with disease. Nadeau thought otherwise. He predicted, in fact, that people in the study would possess disease-related mutant genes like everyone else; what conferred their unusual health, he suspected, was the complex gene-gene interactions he’d seen in mice, where certain genetic variants in the background could neutralize the effects of pathological mutations. “The original premise—and Eric and I had a little bet on this—is that when they sequenced them, they would be free of disease-causing genes,” Nadeau recalls. “My argument was, they’ve got the same load of disease-causing mutations as anybody else, but they also have variants that suppress those diseases.”
The study is still going on, and it turns out, as Nadeau predicted, that hundreds of the test subjects possess just as many disease-causing genes as members of the control group, which in this case consists of people who died more than a decade ago. According to conventional Mendelian genetics, people who harbor these “risk alleles” should be more susceptible to disease. And indeed, conventional genetic testing would point to a heightened risk for diseases they never developed. But Topol’s results indicate that you can’t gauge the impact of any given disease variant unless you know what other variants are in the background, potentially including some that either modify disease genes or protect against them. So Nadeau and Topol have advocated a systematic search for “modifier genes” and “protective alleles” that neutralize the deleterious effects of the disease-associated variants that everyone else has been looking for.
It may sound like a dramatic break, but Nadeau says these exceptions to Mendelian patterns should come as no surprise. “Mendel picked the traits where he would get simple genetics,” he explains. “What Mendel said is true. But it’s not the whole truth.”
Stephen S. Hall is a New York-based writer whose recent books include Wisdom: From Philosophy to Neuroscience and Size Matters, which explains the genetics and biology of height.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.