• Home /
  • Blog /
  • Off-target editing with CRISPR/Cas9 : The evidence

Off-target editing with CRISPR/Cas9 : The evidence

Jun 23, 2017 10:16:36 AM No Comments

What more do we know since last year?

As more and more papers are published using models generated by CRISPR/Cas9 editing, and new and exciting applications for the CRISPR/Cas system continue to be invented, the potential for off-target editing continues to be discussed. We published an article on our blog a year ago which explains the potential for off-target editing with CRISPR/Cas9 and summarised some of the literature on this topic, and thought this was a good time for an update.


One article which has generated discussion in the Horizon offices was a recent letter to the editor in Nature Methods (Schaefer et al., 2017). The authors reported an unexpected finding from their study when using CRISPR/Cas9 editing to correct a mutation which causes blindness in mice. The genomes of two mice which had been injected with an sgRNA plasmid as zygotes, and one control mouse which had not been injected, were sequenced. The authors report a higher mutation rate, and a majority of common mutations, in the two mice which had received the sgRNA injection compared to an uninjected control mouse, which they suggest is evidence that these mutations were caused by CRISPR/Cas9. Although we think extrapolation of these findings from one sgRNA in such a small sample of mice, particularly as the control mouse did not receive a sham injection, to CRISPR/Cas9 more generally would be unwise, we are always keen to see studies investigating off-target editing and think it is important more work is carried out in this area.
Another study which included assessment of off-target editing by CRISPR/Cas9 in vivo was published in Scientific Reports (Nakajima et al., 2016). This was also a small study, where whole exome sequencing was performed on four knock in mice derived from CRISPR/Cas9 edited founders and one control. This study identified far fewer mutations than the Nature Methods letter, and their frequency was similar in the control and knock in mice, suggesting they were unrelated to the CRISPR/Cas9 editing.
Scientists at Horizon have also investigated off-target editing in our HAP1 cell lines. Ten sgRNAs were chosen with a range of off-target scores, and the top ten predicted off-target sites for each of these were sequenced. Off-target editing was not detected in any of these predicted sites (sequences were obtained for 82 of the 100 sites). While this does not preclude off target editing at other locations, the lack of editing in the top ten sites closest resembling the target is consistent with off-target editing being a rare event.

What's on the horizon?

Looking to the future it is possible that non-editing forms of gene regulation such as CRISPRi and CRISPRa (the subject of a recent Horizon blog) will be used in some cases instead of CRISPR/Cas9 editing. CRISPRi has been shown to be very sensitive to mismatches in sequence (Gilbert et al., 2014) and gene expression analysis has shown very few changes in off-target gene expression (Gilbert et al., 2013). Another option being investigated is the use of engineered Cas9 or alternative RNA guides nucleases, such as Cfp1 which has been shown to have fewer off-target effects than Cas9 (Kim et al., 2016).

Controlling for off-target effects

Much of the focus on off-target editing by CRISPR/Cas9 relates to its potential use in the clinic, where even very rare unintended mutations could be detrimental. Experimentally these very rare events are less of a concern. However, to be on the safe side we believe there are two important and simple ways to minimise the potential of off-target editing to affect your experiments. Firstly, is the use of the most up-to-date sgRNA design algorithms and selection of guides with low predicted off-target potential. Secondly, to mitigate against the possible effects of off-target editing in CRISPR/Cas9 engineered cell lines, is to use two different sgRNAs to generate two clones. The use of two different sgRNAs means that even if there is some off-target editing this will occur at different sites, and so you can have confidence that any common phenotypes between the two lines are due to mutations in your target gene.
Horizon KO cell lines are available with multiple clones, produced with independent sgRNAs, so these easy access cell models enable controls for identifying and validating your gene target. 
For more information:
 Learn more
#Gene editing