At barely six years old, CRISPR gene editing has gone through its share of controversies, including a headline-grabbing patent dispute and vigorous debates over the ethics of editing human embryos. More recently, worries over the fundamental safety of the DNA editing tool have cast doubt on its potential to offer precise, permanent fixes for genetic diseases.
Normally, safety concerns raised in preclinical studies conducted in mice or human cell lines don’t garner such rapid and widespread media coverage, let alone double-digit percent drops in company stocks. But that’s the recurring scenario faced by the CRISPR field’s three leading players, Crispr Therapeutics, Editas Medicine, and Intellia Therapeutics.
The most recent incident happened on June 11, when a pair of papers published in Nature Medicine showed that two kinds of cell lines are often impervious to CRISPR gene editing unless a protein called p53 is broken or missing. Since cells lacking p53 are prone to become cancerous, the papers warned that CRISPR-edited cells should be screened for functional p53 before using them in human therapies. The three companies aren’t developing therapies related to the cells tested in these papers, but once a press release and the first news story made the cancer connection to CRISPR, the trio lost hundreds of millions of dollars in combined worth.
“To me, these papers essentially were a technical cautionary tale,” says Thomas Barnes, senior vice president of innovative sciences at Intellia. “But what I read was, ‘CRISPR causes cancer,’ ” in the news.
“The market is reacting not much differently than it did for gene therapy 25 years ago,” says Jean-François Formela, a partner at Atlas Venture and an Intellia board member.
Cancer has always been a hypothetical concern for any therapy that breaks DNA, which is how the traditional form of CRISPR that will first be tested in humans works. An enzyme called Cas9 snips across DNA’s double-stranded helix at a particular location specified by a guide RNA. Scientists design guide RNAs to minimize off-target, unintended cutting, but there is always a risk that extra cutting could occur. “One off-target cut does not equal cancer,” Barnes says. But as the number of inadvertent snips increases, the odds of breaking an important anticancer gene, such as the gene for p53, increases.
One of the two studies, conducted by Bernhard Schmierer and his colleagues at the Karolinska Institute, began when the team noticed that CRISPR gene editing worked more often in cancer cells than normal human cells grown in their labs. “The finding was more or less accidental,” Schmierer says.
Hunting for an explanation, the team studied cells normally found in the human eye, retinal pigment epithelial cells, but genetically engineered for immortalization, a common practice that lets cells replicate indefinitely for laboratory studies. The researchers found that CRISPR was terrible at editing these cells unless they lacked functional p53 (Nat. Med. 2018, DOI: 10.1038/s41591-018-0049-z).
The protein p53 is well known for its role in initiating DNA repair in broken genes, earning it the moniker “guardian of the genome.” Schmierer reasoned that since p53 prevented CRISPR from working efficiently in most of the group’s cells, scientists may unwittingly use gene-edited, but p53-deficient, cells in their experiments and therapies. It wouldn’t guarantee cancer, but it certainly looked like a ticking time bomb.
A team of Novartis scientists led by Ajamete Kaykas published the second study, which reached the same conclusion about p53, only this time in human pluripotent stem cells (Nat. Med. 2018, DOI: 10.1038/s41591-018-0050-6). Kaykas declines to comment on what the results could mean for CRISPR gene editing in other cell types, but he says Novartis is not currently developing any CRISPR-based therapies with pluripotent stem cells.
Others also downplay the importance of the results. “We are not working on those cells, so the story ends there,” Intellia’s Barnes says. Sam Kulkarni, CEO of Crispr Therapeutics, concurs. “I can’t think of an application today where this is relevant, but it could be in the future,” he says.
And some worry about drawing conclusions from a pair of papers in a quickly evolving field. “We have all been looking for the possibility of cancer,” says Jacob Corn, who studies CRISPR at the University of California, Berkeley. So far, he hasn’t spotted any hints of it, and he doesn’t think these new studies are a warning for therapies. His own lab has seen efficient editing in cells with functional p53 and poor editing in cells lacking p53—the opposite results of the new papers. “So p53 can be a predictor but is probably not the only predictor. DNA repair is complicated.”
At this point, Kulkarni is becoming used to the short-lived hypes of CRISPR safety concerns. “We’ve had one of these pop up every six months or so.”
It started in May 2017, when a Nature Methods paper suggested that Cas9 was making extraneous cuts throughout the genomes of laboratory mice, meaning that CRISPR was far less precise than many claimed (Nat. Methods2017, DOI: 10.1038/nmeth.4293). Within months, Editas, Intellia, and several academic scientists fired back with their own papers questioning the results. The original report was retracted online on March 30 when additional reports showed that the alleged off-target cuts were actually due to the natural genetic variation of the mice (bioRxiv 2018, DOI: 10.1101/154450).
In January, Matthew Porteus, a stem cell biologist at Stanford University, posted a paper on the preprint server bioRxiv indicating that some humans may have a preexisting immunity to Cas9, which is derived from bacteria (bioRxiv 2018, DOI: 10.1101/243345). The paper led to media reports suggesting CRISPR might not work in humans and that CRISPR stocks could be worthless.
Again, the CRISPR companies said their experiments suggested something different. For instance, Intellia’s most advanced program will use CRISPR to edit genes in liver cells. The nightmare scenario is that the immune system recognizes Cas9 as foreign and mounts a rapidly fatal attack on the entire liver. “You can lay out the theoretical question,” Barnes says, but Intellia has done the experiment in mice, rats, and nonhuman primates. “And that doesn’t happen at all,” he says.
Sangamo Therapeutics, a company that uses a different gene-editing system, called zinc finger nucleases, also saw its stock drop upon the cancer scare news. “We get bundled in with CRISPR with bad things but rarely get bundled in with them when they are the best thing since sliced bread,” CEO Sandy Macrae says.
Like CRISPR, zinc finger editing makes double-strand breaks in DNA. The company has already tested its zinc fingers in 104 people with HIV and recently used them in four people with a rare disease called Hunter syndrome. There are no signs of ill effects thus far.
“At least initially, genome editing should only be done when the burden of the diseases balances this as-yet-unknown risk,” Macrae says. For serious diseases such as HIV or cancer, it could be difficult to parse out what is a natural progression of the disease or an unintended side effect of the therapy.
The CRISPR scares are likely far from over. “There is definitely an academic and journalistic incentive right now such that this process is likely to repeat itself in the future,” Stanford’s Porteus says. For example, unregulated clinical trials using CRISPR are underway in China, and already Crispr Therapeutics’ planned trial in the U.S. was temporarily delayed by the U.S. Food & Drug Administration without explanation. The public’s fears of editing the germ line to pass down alterations to future generations—something that gene-editing companies are staunchly opposed to—will likely continue to make CRISPR a subject of controversy.
The trio of public CRISPR companies will face intense scrutiny as their first clinical trials using CRISPR roll out over the next year. “A lot of these esoteric, early-stage-science, hypothetical papers might get people excited for 24 hours,” Atlas Venture’s Formela says. “But at the end of the day, what really matters is the clinical data.”