As a new study reveals that hundreds of unintended mutations were induced in mice by a genome editing technique, GMWatch asks what the implications are for the safety of genome-edited food products
Update: The new study by Schaefer et al. was retracted in April 2018.
A new study published in Nature Methods has found that the genome editing technology CRISPR introduced hundreds of unintended mutations into the genome of mice.
In the study, the researchers sequenced the entire genome of mice that had undergone CRISPR gene editing to correct a genetic defect. They looked for all mutations, including those that only altered a single nucleotide (DNA base unit).
They found that the genomes of two independent gene therapy recipients had sustained more than 1,500 single-nucleotide mutations and more than 100 larger deletions and insertions. None of these DNA mutations were predicted by the computer algorithms (software packages) that are widely used by researchers to screen the genome (the total DNA base unit sequence) of an organism to look for potential off-target effects.
While this study was conducted in the arena of gene therapy, it has clear implications for the regulation of food plants and animals derived from CRISPR and other genome editing techniques.
Regulatory agencies across the world are currently engaged in a debate about how to assess genome-edited products for safety. Many GMO proponents are proposing “light-touch” regulation or even no regulation at all, based on the assumption that the outcome genome editing techniques like CRISPR are precise, predictable, and therefore safe.
The new study shows that this assumption is false. So how should these products be regulated?
One suggestion that has been put forward is to require whole genome sequencing of gene-edited organisms to be conducted and submitted to biosafety authorities.
But this raises a further question: if the whole genome sequence does not show any mutations or off-target effects, other than those intended, should we be reassured?
We asked Dr Michael Antoniou to comment. Dr Antoniou is a London-based molecular geneticist who uses genetic engineering techniques, including genome editing, to develop gene therapies.
Dr Antoniou says:
I agree that the whole genome sequences of gene-edited organisms must be submitted to biosafety authorities. And if the whole genome sequence did not show any additional mutations/off-target effects other than those intended, this would be somewhat reassuring.
However, it is highly unlikely that this will ever be the case. It is a matter of how many off-target mutations there are, rather than a matter of their total absence. The technology is not perfect. It will in future become less prone to off-target effects as it is refined procedurally, but it is extremely unlikely that it will ever arrive at a point where only the intended change will result.
In addition, the application of genome editing technologies in an agricultural context will involve plant tissue culture, which has its own inherent mutagenic properties. So no matter how precise the genome editing becomes, there will always be a large spectrum of tissue culture-induced mutations present.
Many of the genome editing-induced off-target mutations, as well as those induced by the tissue culture, will no doubt be benign in terms of effects on gene function. However, many will not be benign and their effects can carry through to the final marketed product, whether it be plant or animal.
There is an additional feature that makes genome editing more likely to bring about off-target gene damage resulting in a disturbed function. This stems from the fact that the off-target effects will not be random, but will take place at sites within other genes that are similar in DNA base unit sequence to where the intended change has taken place.
This does not exclude the recent and unsurprising finding from the recent study in mice, which found that off-target effects from CRISPR can occur at sites within the genome whose DNA base unit sequence is markedly dissimilar from the targeted site. The researchers found that the large numbers of off-target mutations caused by CRISPR in mice could not be predicted by the usual computer algorithms.
Thus not only is it necessary to conduct whole genome sequencing to identify all off-target mutations from CRISPR-based genome editing, but it is also essential to ascertain the effects of these unintended changes on global patterns of gene function. Therefore one needs to follow up the whole genome sequencing with other molecular profiling analyses or “omics”: transcriptomics — gene expression profiling, proteomics — protein composition profiling, metabolomics — profiling of metabolites, and miR-omics – microRNA profiling.
In addition, it is important to acknowledge that the targeted intended change in a given gene may also have unintended effects. For example, the total disruption or modification of an enzyme function can lead to unexpected or unpredictable biochemical side-reactions that can markedly alter the composition of an organism, such as a food crop.
The compositional alterations in food products produced with genome editing techniques will not be fully revealed by the molecular profiling methods due to the current inherent limitations of these techniques. So it is still necessary to conduct long-term toxicity studies in established animal model systems.
In the absence of these analyses, to claim that genome editing is precise and predictable is based on faith rather than science.
We conclude from Dr Antoniou’s explanation that the products of genome editing techniques should be at least as stringently regulated as the products of old-style GM techniques. In fact, regulation for all types of GM products should be overhauled to include “omics” molecular analyses and long-term animal feeding studies. Such analyses and studies are not currently demanded by any regulatory authority anywhere in the world.
1. Schaefer KA, Wu W-H, Colgan DF, Tsang SH, Bassuk AG, Mahajan VB. Unexpected mutations after CRISPR-Cas9 editing in vivo. Nat Methods. 2017;14(6):547-548. doi:10.1038/nmeth.4293.
Report by Claire Robinson