In 2017, researchers at Oregon Health and Science University came out with some big (if true) news. Led by a reproductive biologist named Shoukhrat Mitalipov, the scientists had used the Nobel Prize–winning molecular tool known as Crispr to fix a heart-condition-causing mutation in human embryos—a first in the US. A week later, the journal Nature published details of these boundary-pushing experiments. Up until that point, viable embryos had only been Crispr’d once before, in China. As WIRED reported at the time, Mitalipov’s team’s editing appeared to work surprisingly well. But one thing didn’t go as expected.
Crispr works by cleaving DNA apart at a specific location in the genome. Then it’s the cell’s job to repair the resulting double-stranded break. One way to make sure it does it right is to supply a bit of corrective DNA along with the Crispr components. But Mitalipov’s group reported that their embryos didn’t use the template they provided. The embryos had been created by fusing a healthy donor’s egg with a sperm that carried the mutation. But it turned out that the newly fertilized embryos borrowed the egg donor’s healthy copy to rebuild the gene, rather than the string of DNA the scientists had supplied it.
This was weird, but potentially really cool. It meant that early-stage embryos might have unique repair mechanisms other cells don’t that could be harnessed for gene editing. And that might make it easier to correct life-threatening mutations in embryos for which only one parent carries the genetic glitch. (The template-feeding technique is notoriously inefficient; cells don’t like being told what to do.)
Not every scientist bought this explanation, though. It kicked off a dispute within the Crispr community about what else might be going on, although only a few of the skeptics had the expertise and resources to investigate. One of them was Dieter Egli, a stem cell biologist at Columbia University. Editing viable human embryos isn’t illegal in the US. But it is controversial, and Congress won’t foot the bill for any such research. So scientists who want to pursue human embryo editing have to find private philanthropies to fund the work. Egli had backing from the New York Stem Cell Foundation and the Russell Berrie Foundation. His team started work almost immediately. Now, nearly three years later, he thinks they finally have a more plausible answer to what happens when you edit a human embryo. More often than not, Crispr doesn’t spur some novel repair mechanism. It makes the whole gene—sometimes the whole chromosome—disappear.
“Our results point to the extraordinary caution necessary before progressing this kind of research to the clinic,” says Egli, whose study was published Thursday in the journal Cell. He believes that had such data existed two years earlier it would have discouraged anyone from attempting to use Crispr to edit human embryos with the intention of starting a pregnancy. But in 2018, a Chinese scientist named He Jiankui did just that, claiming to have created the world’s first Crispr’d children and bestowing upon them a mutation that’s protective against HIV infection. The experiment was a catastrophic scandal—a failure of science, ethics, and regulation. He Jiankui is now serving a three-year prison sentence, but his work opened the door for others eager to advance the technology. Last year, a Russian scientist made public his plans to use Crispr to help deaf parents have children who won’t inherit a gene mutation that causes hearing impairment.
In the Cell paper, researchers led by Egli injected Crispr into human sperm from a donor with a blindness-causing mutation in a gene called EYS2 that resides on the long arm of chromosome 6. Once inside, Crispr made cuts at the site of the genetic glitch. The researchers didn’t add any new material to correct the sequence, because their goal wasn’t necessarily to fix the mutation. It was more to see how an embryo would repair the break if left to its own devices. Then they used the edited sperm to fertilize healthy eggs in the lab, creating 24 embryos. When they analyzed the genomes of the resulting embryos, they couldn’t detect the mutation in about half of them. On the surface, it looked like the edit had worked.
But then they looked closer, using DNA-screening techniques developed by employees of Genomic Prediction, a New Jersey startup that sells an embryo-selection tool to IVF clinics, to help parents pick the ones least likely to develop genetic disorders. (That includes missing or rearranged chromosomes, though Genomic Prediction is more famous for its founders’ forays into intelligence testing for embryos.) The company’s software counted snippets of DNA from both the maternal and paternal sides of chromosome 6, revealing that the mutation hadn’t been corrected. In fact, the genetic material contributed from the sperm had disappeared altogether.
Egli believes that the cut made by Crispr isn’t getting repaired at all, leaving a gap in the DNA. That fracture separates the long arm of the chromosome from its spindle—the fibers that pull chromosomes apart during cell division. “If it’s not attached to the spindle, then they can be lost,” says Egli. “Where exactly they go, we do not know yet.”
This is potentially a much bigger problem than the initial blindness-causing mutation. Massive amounts of missing or rearranged DNA might cause birth defects, cancer, or other health problems, if such embryos prove viable at all. “The outcome couldn’t be more different from correcting the mutation,” says Egli. “The loss of a chromosome is not compatible with normal development.”
Egli’s team’s experiments, and the safety concerns they raise, have already influenced the debate about whether scientists should use heritable human genome editing—that is, modifying the DNA of sperm, eggs, or embryos—to prevent genetic disease. The US bans any experiments involving establishing a pregnancy with an embryo that has been genetically modified. Seventy-five other countries have similar prohibitions on the books, according to a recent survey of global gene editing policies. No country explicitly permits heritable human genome editing, but many nations have no laws that address it at all.
A World Health Organization panel is working to establish coordinated global regulatory standards for governments to follow. Last July the WHO issued a statement urging countries to put an immediate stop to any experiments that would lead to the birth of altered humans. Last month a second committee, convened by the National Academies in the wake of the Crispr baby scandal, released a 225-page report describing how safe and ethically permissible heritable human genome editing might proceed. (TL;DR: not yet; not for a while; not for most diseases.) The report cited Egli’s team’s work—along with two other preprints describing unintended chromosomal modifications that at the time were not yet peer-reviewed—as evidence that the science is still too premature to move to clinical trials.
“It’s a clever study, well-performed, and the results are very compelling,” says Gaétan Burgio, a geneticist at the Australian National University, who was not involved in the research. He says it reinforces the fact that the technology is still not safe and will require huge improvements before anyone should try starting any pregnancies with edited embryos. “I think we are still miles away from translating this to the clinic,” says Burgio.
However, he thinks the science of how human embryos respond to Crispr’s patented cutting action is still far from settled. Egli’s team only looked at a single gene. And they didn’t supply a repair template. Developmental biologist Kathy Niakan and her team at the Francis Crick Institute in London had more success deleting a gene using Crispr, as they reported in a preprint this summer. However, in about a quarter of their embryos, Crispr also deleted large sections it wasn’t supposed to—sometimes erasing several thousand DNA letters. Mitalipov’s group also has a new preprint, doubling down on their 2017 claims. Their new experiments suggest that embryos borrow DNA templates from themselves to make repairs—a process called gene conversion—up to 40 percent of the time.
Egli still isn’t buying it. When he looks at the most recent paper out of Mitalipov’s lab, he sees chromosome loss, not gene conversion. “The issue is there’s no counting,” he says. To really prove the embryos are replacing the paternal mutation with a new sequence copied from the normal maternal gene, you’d have to show that the embryos have two copies of the genetic material that came from the egg. And the Mitalipov paper only proves the existence of one. “With that data, it's insufficient to conclude that gene conversion has occurred,” says Egli.
Paula Amato, an OB-GYN at Oregon Health and Science University who is a close collaborator of Mitalipov’s and a coauthor on both the 2017 paper and the most recent preprint, told WIRED via email that Egli’s work does not disprove their hypothesis; it only offers an alternative explanation for their results. In response to his contention that their data might be explained by chromosome loss, she noted that the genetic analyses Egli’s group used are subject to a unique problem. Sometimes, when you’re sequencing the DNA of a single cell plucked from an embryo, one or both copies of a gene don’t register. The phenomenon is known as allele dropout, which she thinks could be another explanation for the chromosome disappearance. She would have liked to have seen Egli’s team repeat their blindness gene experiments in stem cell clones which don’t suffer from this random disappearing act.
Amato added that her group is currently conducting additional studies to further validate their own results. She did not say whether the preprint her team released this summer has yet been accepted for publication.
“I would say that both groups agree, however, that heritable human genome editing is not ready for prime time and much more research is needed before this technology is applied clinically,” Amato wrote.
The other thing everyone agrees on is that settling the science of what happens when you inject Crispr into an early-stage embryo is of paramount importance. If the classic form of the molecular tool (which was used in all studies mentioned above) wreaks too much havoc on the genome, alternative approaches that don’t break the DNA cleanly in two might be preferable. These include base editing—which unzips the DNA enough to swap a single base pair for another—and prime editing, which does that, and more, with just a little nick on one side of the DNA double helix. Neither technique has been tested in humans.