Because of their ease of use and scalability, RNA-guided nucleases designed for gene editing have quickly become a hit among researchers since their introduction last year. However, a new study suggests that these enzymes cleave off-target DNA sites more often than competing methods, compromising the usefulness of these new gene-editing tools.
researchers identified one such programmable nuclease, called Cas9, and this year, several teams showed that the CRISPR system works in mammalian cells. Cas9 requires a guide sequence of just 20 nucleotides, making the CRISPR platform easier to use than other gene-editing methods, such as zinc finger nucleases (ZFNs) and transcription activator-like effector nucleases (TALENs).
But until now, no one had looked at the system’s potential to cause off-target genetic changes—a critical factor for both research and therapeutic applications.
To determine the frequency of off-target effects, researchers from the Massachusetts General Hospital and Harvard Medical School conducted an assay using enhanced green fluorescence protein (EGFP). The team designed a 20-nucleotide guide to match EGFP and then applied it to human cells containing the EGFP gene. A loss of fluorescence indicated a sequence capable of guiding Cas9 to the target DNA.
The researchers tested three guide RNAs, each targeting a different region of the EGFP gene. Many single mutations did not interfere with the sequence’s ability to guide Cas9. Even more surprisingly, many double mutations were tolerated.
That suggested a high level of off-target activity. “So we decided to do the experiment that people really care about… if you target a DNA sequence, what other DNA variants is it going to cut?” said J. Keith Joung, associate chief of pathology at Massachusetts General Hospital and Harvard Medical School and author of an article published in Nature Biotechnology this week describing the study.
The team designed guide RNAs for six human genes. For each target sequence, they identified other genomic sites that differed by 1–2 base pairs and a subset that differed by 3–6 base pairs. For each gene, they composed a list of sites most likely to be mistakenly targeted by Cas9. They then sequenced these 40 to 60 of those hot spots for each gene, to see if an insertion or deletion mutation had taken place.
Of the 6 guide RNAs, 4 produced at least 1 off-target mutation site. Erroneous insertions or deletions occurred at sites with 3, 4, and even 5 mismatches from the target sequences. The rate of mutagenesis in off-target sites was sometimes as high or higher than the target site.
“This was very different than what you see with zinc finger nucleases and TALENs, where the off-target mutagenic frequencies tend to be much lower than the on-target (frequencies),” said Joung.
The results suggest that the method needs further refinement to reduce off-target effects, but Joung is not giving up on the CRISPR system. “We’re still excited about the platform. To me, this directs our research and says we need to improve its specificity and (come up with) assays to assess its off-target effects.”
1. Fu, Y., J. A. Foden, C. Khayter, M. L. Maeder, D. Reyon, J. K. Joung, and J. D. Sander. 2013. High-frequency off-target mutagenesis induced by CRISPR-cas nucleases in human cells. Nat Biotech advance online publication(June).