Major CRISPR errors were discovered by chance

A recent study found CRISPR-Cas9 gene editing may be less precise than previously thought.

A study released last week suggests that the gene editing technology CRISPR-Cas9 may not be as precise as previously thought. Researchers found that in addition to small errors already known to be part of the gene editing process, CRISPR-Cas9 can cause large deletions and even rearrangements in genes around the target site.

The researchers looked at errors in mouse and human cells in lab conditions. In some cases, deletions were as large as several thousand DNA bases, enough to potentially alter the function of a cell. Whether the findings will affect CRISPR’s utility for clinical applications is not yet known. But researchers do expect increased scrutiny of the technology, and the scope of gene editing errors, going forward.

We spoke with one of the study’s authors, Michael Kosicki, to learn more about the discovery and its implications.

ResearchGate: What motivated this study?

Michael Kosicki: It was really serendipity. A control experiment for another study gave an unexpected result, and we decided to investigate. Initial results made it clear we were looking at something very exciting indeed, with implications for gene editing. So we put the other project on hold and investigated further.

RG: Can you describe the errors you found?

Kosicki: “Error” is a relative term – small errors in DNA repair is what we often want in a gene editing experiment. That's one way we can permanently turn specific genes off and study their function. But that's only good in moderation. What we have discovered are large errors: loss of thousands of DNA letters, inversions, and complex rearrangements. Some of them are quite tricky to detect by standard means. They may potentially affect the expression of other genes.

RG: How did you discover these large errors?

Kosicki: We edited cells using Cas9 in places no one usually does: non-coding parts of genes called introns. One would expect edits in this area to have no effect on gene expression. But we saw one! We investigated the affected cells in more detail and were surprised to see some of them lacked DNA at the targeted locus completely. Curious!

We tweaked the standard tools, which extended our field of vision. Only then did we begin to see that the size of the “errors” was why we missed them at first. The simplicity of the system we initially studied was the key. Even though the frequency of “large errors” was substantial, it could have been (and mostly was) overlooked under less optimal circumstances.

RG: Were you surprised by the scope of these errors?

Kosicki: Yes and no. Every time we looked with a slightly different tool, we would find something new. On the other hand, the signs were there before. We have been making observations that didn't quite add up, but we attributed them to technical problems. So our discoveries made things fall into place.

RG: How does this compare to previous findings about CRISPR errors?

Kosicki: We weren't the first to report similar events. Both anecdotal and systematic studies of edited mouse embryos revealed some of them. However, no similar systematic study of different therapeutically relevant cells has been done before. Also, the sheer size and variability of the events reported is a novel finding.

Importantly, what we discovered are not “off-target” effects caused by an additional cut at a mismatched target site elsewhere in the genome. The research community has made a tremendous effort to develop methods that detect and limit this kind of damage. Unexpected “on-target” events we detected flew under the radar, but could potentially have similar consequences to those caused by “off-target” damage.

RG: How do you think this new information will affect CRISPR-Cas9 research going forward?

Kosicki: We hope it will be a call to increased scrutiny, further research into DNA repair mechanisms, and alternative ways of genome editing. We cannot say all therapeutically relevant cells are affected, but it certainly won't hurt to double check. As always, some degree of risk is acceptable when attempting life-saving therapies, but it needs to be known very explicitly.

RG: What can the research community do to continue gaining a better understanding of CRISPR and its risks?

Kosicki: I’ll take this opportunity to plug for freedom in research. Just like the discovery of CRISPR/Cas9 system itself, this research was made possible by the freedom afforded to the scientists to tinker and snoop around.