In late March 2018, Nature Methods officially retracted the original paper, agreeing with critics that the study’s authors lacked sufficient evidence to prove that the mutations they identified were indeed caused by off-target edits made by CRISPR-Cas9 and were not simply the natural result of genetic differences between inbred mice.
When the original paper was published in 2017, Seeker spoke with Vinit Mahajan, co-author of the CRISPR study and professor of ophthalmology at the Stanford University Medical Center. Also an eye surgeon, Mahajan was excited about CRISPR’s potential to use targeted gene-editing to reverse genetic conditions, even blindness.
In fact, Mahajan and colleagues had used the CRISPR-Cas9 to restore sight to three mice blinded by a genetic mutation. But when they compared the fully sequenced genome of their CRISPR-treated mice to the full genome of another mouse from the same inbred strain, they found more than 1,500 small mutations and 100 larger insertions and deletions far from the target editing site.
Mahajan didn’t believe the off-target edits meant that CRISPR was inherently dangerous, but said his paper was a “cautionary tale” to scientists and companies rushing to use CRISPR in human patients without fully understanding its risks.
RELATED: ‘No-Cut’ CRISPR Activates Genes Without Altering DNA
The response from the scientific community was swift, alleging that it was Mahajan and his co-authors who didn’t fully understand mouse genetics. A month after publishing the original paper, Nature Methods added an “Editorial Expression of Concern” based on multiple letters it had received casting doubt on the conclusions.
Lauryl Nutter co-wrote one of those letters. Nutter is the associate director of the Center for Phenogenomics where she works with researchers to create mouse models of human diseases. She also chairs the Cas9 working group for the International Mouse Phenotyping Consortium. Nutter told Seeker that the conclusions of the retracted paper, while not intentionally misleading, should be taken with “a pound of salt.”
The biggest mistake the authors made, said Nutter, was to assume that all mice from the same inbred strain are genetically identical. Inbred mouse strains are the result of hundreds of generations of brother-sister mating to produce mice that look exactly the same and respond predictably to experimental procedures. But that doesn’t mean that they are genetic clones.
“That’s actually a myth. If you look at subsequent generations of an inbred strain, they get 100 new mutations every generation,” said Nutter, adding that it’s not all that surprising for two unrelated mice from the same strain to exhibit 1,600 genetic differences. “It would actually be more surprising if they didn’t.”
So when the CRISPR researchers thought they were making an apples-to-apples comparison between the genomes of two genetically identical mice, they were actually making more of an apples-to-crabapples comparison. The genomes of the treated and untreated mice likely contained many hundreds of slight genetic differences before they were ever injected with CRISPR-Cas9.
But the critiques of the CRISPR paper didn’t stop there. Nutter said that there’s a wealth of biochemical research that’s been done on Cas9 — the protein that acts as CRISPR’s genetic “scissors” — to understand how it finds its targets, and that none of the alleged off-target edits were at sites that fit the established model.
“That alone should have made them step back and ask: What are we missing in our interpretation? Because this doesn’t look anything like it should,” said Nutter.
Other scientists from Harvard University, South Korea, Germany, and biotech companies Editas Medicine and Intellia Therapeutics, called out the study authors for not identifying and sequencing the genomes of the treated mice’s parents, a standard way of comparing genetic variability between generations.
Critics also pointed out that the two CRISPR-treated mice used in the study were much more genetically similar to start with than the untreated control mouse. In fact, said Nutter, the untreated control mouse exhibited several of its own unique mutations when compared to other mice in the same inbred strain.
“But you would never conclude that those differences were attributed to the lack of Cas9 treatment.”
RELATED: Human Immunity to Cas9 Could Threaten CRISPR’s Future
Explaining why the paper was ultimately retracted, Nature Methods admitted that it should have enlisted someone with more expertise in inbred mouse genetics during the peer-review process. And it advocated for many more studies that sequence whole mouse genomes. The only way for scientists to definitively calculate the potential off-target effects of CRISPR-Cas9 is if there’s solid data about the frequency of genetic variation between generations of inbred mice and their siblings.
While more than half of the study’s original authors disagreed with the retraction, all agreed that there needs to be a lot more whole genome sequencing to accurately distinguish naturally occurring mutations from gene-edited mistakes.
In an email to Retraction Watch, Mahajan cited at least one independent analysis that casts doubt on the premise that all the mutations were due to genetic drift, but said that he looked forward to other scientists and companies publishing more whole genome sequence data and analysis, adding, “We are confident that by working together in an open, transparent, and unbiased manner, the research community can find safe and effective ways to implement new therapies for our patients.”