Ten years ago, scientists discovered that some people are naturally missing working copies of a gene known as PCSK9. The consequences of the mutation were extraordinary. These people, including a Texas fitness instructor, a woman from Zimbabwe, and a 49-year-old Frenchman, had almost no bad cholesterol in their blood. Otherwise, they were perfectly normal.
Drug companies pounced on the clue. To lower cholesterol, they would also try to block PCSK9. Now two separate drugs that disable the gene’s activity are nearing FDA approval. People taking the medications have seen their cholesterol levels plummet dramatically, sometimes by 75 percent.
Regeneron Pharmaceuticals, the company behind one of these drugs, now says it’s building a giant database of human DNA information in what amounts to a large-scale, systematic search for the next PCSK9. At a new genetic research facility that Regeneron completed last month in Tarrytown, New York, the company says it has begun a five-year effort to sequence the genomes of as many as 100,000 volunteers recruited from East Coast hospitals and identify rare genetic outliers among them.
So far, 13,000 people’s DNA has been partly decoded, and the company is using software to search for deleted genes. By checking against the volunteers’ medical records, or by calling them in for intensive testing, the company expects to learn if these missing genes cause illness or, perhaps, also have unusual protective effects.
People missing a particular gene are known as “knockouts” after a kind of laboratory mouse that’s been genetically engineered to lack a gene. Knocking out a gene from mice is a widely used technique that permits scientists to judge a gene’s function by its absence.
Rapidly expanding databases of human genomes mean researchers can now find knockout people instead. To scientists, that’s going to be a valuable shortcut to determine what human genes do. What effect does missing a gene have on a person’s body? To drug companies, these individuals promise living, breathing answers to some of the biggest questions they face, like whether their drugs will actually work, and whether blocking a given gene would be safe to do, or instead cause problems. “It’s a huge emphasis for us because these are incredibly informative natural human experiments,” says Aris Baras, director of R&D initiatives for the company.
By and large, it’s not good to be missing a gene. Such defects are the cause of diseases like cystic fibrosis and muscular dystrophy. Over time, scientists say, evolution has been working against these errors, which accounts for why they are relatively rare. Yet missing a gene can sometimes provide powerful protection against disease. People missing the SLC30A8 gene are half as likely to get diabetes as people with it. Those without working copies of a gene called CCR5 can’t get infected with HIV.
“This is going to be the major model of human disease research going forward,” predicts Sekar Kathiresan, a cardiovascular specialist at Massachusetts General Hospital who advises Regeneron. (This month Kathiresan reported another gene that, when missing, greatly cuts a person’s risk of heart disease.) Drug companies bet millions on drugs that cure animals or work in a lab dish. But 90 percent of drugs tested in human studies show no beneficial effect, or prove toxic. It’s guesswork at a staggering scale. “Animals don’t predict people. The concept now is to leverage the successes of the human genome, and develop medicines to mimic them,” he says.
Most large-scale genetic research is a search for the causes of disease, not the nature of health. But in 2008, Daniel MacArthur, a computational geneticist now at the Massachusetts General Hospital, became interested in how frequently genes are completely dysfunctional in healthy people. Along with collaborators, he scrutinized the genomes of 185 people.
MacArthur’s analysis, completed in 2012, found that each of us has, on average, one entirely defective copy of about 80 genes, and another 20 genes for which neither copy works. In other words, everyone’s genome is a little dysfunctional. (Most genes are present in matching pairs—one inherited from your mother, and one from your father.)
At Regeneron, Baras says they have confirmed MacArthur’s estimates in 8,000 people they’ve sequenced. “We are all using common algorithms, and we’re seeing the same kinds of statistics,” he says. Not all the missing genes are interesting. Many code for trivial traits, like one of hundreds of odor receptors, or crop up so frequently that they’re unlikely to be important. Baras says Regeneron is setting those aside and focusing instead on rare cases where people are missing the same genes its drugs are designed to block or interfere with.
Already, two-thirds of the company’s R&D projects have human experiments to match them. “That is a very interesting and exciting statistic for us. Usually we find one or two people, but there are cases where we already have hundreds of individuals,” Baras says.
As its next step, Regeneron plans to call some of these individuals in for detailed medical exams, to determine what’s different about them. In the case of PCSK9, the difference was much lower cholesterol. But that wasn’t the most important finding. Once enough examples of people missing that gene were found, researchers were able to compare them to tens of thousands of controls who did have the gene, and prove that their risk of dying from a heart attack was sharply lower. “Here you have definitive protection against cardiovascular disease, and it’s safe. This is the type of evidence that gives people a lot of confidence,” says Baras. “It’s a huge point. It tells you lifelong deficiency in this gene is good.”
The ability to search systematically for genetic outliers among healthy people is relatively new. Previously, scientists relied on rougher gene maps that didn’t give them access to letter-by-letter DNA information as a complete genome sequence does. That is changing as DNA sequencing becomes cheaper. This year, some labs purchased a new type of sequencing machine, called the X-10, which costs $10 million to buy, but can decode 50 genomes each day, or about two per hour, for a price of between $1,000 and $2,000 each.
That means more labs are betting on the law of large numbers. They include Human Longevity Inc., a startup created by entrepreneur J. Craig Venter, which wants to decipher one million people’s genomes by 2020. Another effort, known as the Resilience Project, is looking for one million healthy volunteers over age 40 so they can be checked for any of 125 disease genes that should have caused debilitating childhood illnesses, but for some reason didn’t. Even Google, which made its own leap into genomics this year, said it would start decoding the DNA of healthy people, although its objectives aren’t known.
“We couldn’t scan for these mutations before for a very simple reason. Before, we did not have the data,” says Kari Stefansson, CEO of DeCode Genetics, an Icelandic genetics company owned by the U.S. biotechnology giant Amgen. He says his company bought an X-10 and will have sequenced the genomes of 25,000 Icelanders by the end of next year, augmenting gene maps it previously built.
In October, researchers with his company said it was well on its way toward building its own knockout catalog. They said the company would invite 8,000 people it had identified to undergo “deep phenotyping,” or a barrage of measurements involving scales, rulers, brain scans, IQ tests, and blood measurements to assess 500 traits.
Even so, Stefansson is skeptical that human knockouts will turn out to be the easy path to blockbuster drugs that some are hoping for. “It’s not an argument without virtues, but it’s also an incredible simplification. Most of these scenarios are more complicated,” he says.
To MacArthur, knockouts are interesting because they’re a way to document the function of genes, good or bad. He created something he calls the “Human Knockout Project” that is helping to study populations in Finland and London and whose aim, he says, is to take every gene in the genome and find a person missing it. But that’s only a manner of speaking. He expects that for many genes, a knockout will never be found. That’s because the majority of genes are essential to life. Without them, you’d never have been born.
That turns out to be a question of urgent importance to husband and wife scientists Eric Minikel and Sonia Vallabh, who have been working alongside MacArthur at Massachusetts General Hospital. Vallabh’s mother died of fatal familial insomnia, an extraordinarily rare disease in which a misfolded protein builds up in the brain, causing dementia and early death. Vallabh has inherited the gene mutation that causes FFI, and has a 100 percent chance of developing the illness, unless some kind of treatment is developed.
Before her diagnosis three years ago, Minikel was an urban planner and Vallabh had gone to law school. But they switched careers and became scientists in order to try to cure Vallabh before she falls ill.
Vallabh’s mutation is the opposite of a knockout—it adds an unwanted function, causing her prion protein to fold in a way that it shouldn’t. This month she switched to another Boston laboratory to explore whether an advanced form of gene therapy, called genome editing, might allow her to eliminate the prion gene from her brain cells altogether.
But would doing so be dangerous? Knockout mice that have been genetically engineered to lack the prion gene seem to be mostly normal, but that’s no guarantee that the same is true of humans. For instance, the knockout surveys carried out by MacArthur’s lab have found more than 40 healthy people with mutations known to prove fatal to mice. Vallabh says she worries that if she were to succeed in eliminating her prion gene it could cause another disease, perhaps equally grave.
In the compressed time frame Vallabh faces—she has perhaps 20 years to cure herself—finding a living person without the prion gene would be one important clue. This year, she and Minikel carried out such a search across DNA sequences of more than 60,000 people as part of MacArthur’s Knockout Project.
They turned up three individuals missing one copy of the prion gene—but, so far, no one who is missing both copies.
Minikel says it may mean that people can’t live without the gene. Or it could be that their database isn’t yet big enough. The gene is small and therefore less likely to be affected by mutations. Working quickly with a pad and paper, with Vallabh looking over his shoulder, Minikel roughly estimated it might take a database of a billion people to know for sure.
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.
Driving companywide efficiencies with AI
Advanced AI and ML capabilities revolutionize how administrative and operations tasks are done.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.