Efficient but unpredictable CRISPR

Continued technical developments in genome engineering have recently allowed the generation of conditional mouse models using long DNA templates. Lydia Teboul, author of a new study in BMC Biology, explains how these long templates can also be used to create point mutations away from the target site, but additional unwanted modifications render the validation of new mutants essential.

Historically, the field of genome engineering in laboratory animals has relied on the use of stem cells to target new mutations in mice. This all changed with genome editing tools that allow researchers to work directly in embryos and tissues. Starting from simple mutations, strategies are now evolving for the introduction of increasingly sophisticated genetic changes.

The basis of these current technical improvements is the CRISPR/Cas9 system, which is revolutionizing our ability to generate targeted mutations directly in the embryo, used in conjunction with single-stranded DNA donors. Whilst oligonucleotide donors (short DNA templates provided by the researcher) have been shown to facilitate editing in the early literature, the use of longer single-stranded DNA donors is a more recent addition to the genome editing toolbox. The applicability and success of these longer donors for the introduction of mutations into specific locations in the genome resulted in the method being called ‘Easi’.

Extending the range of modeled mutations

In our recent study, published in BMC Biology, my colleagues and I concur with these findings and extend them. We present the use of long single-stranded donors in a high-throughput setting for the generation of conditional knock-out mutants.

We also utilize long single-stranded templates to facilitate the introduction of point mutations remote from the recognition site of active Cas9/sgRNA complexes (which so far has not been possible). Thus, we have extended the range of point mutations that are achievable with this technology.

Such a strategy is particularly useful in the context of mutations identified by human genomic sequencing efforts, because it expands our ability to replicate human mutations in mice.

Unwanted effects are more frequent than previously thought

An increasing body of evidence is being compiled to indicate that model validation is the newest challenge for the genome editing community.

We also describe the unpredictability of the outcome of using long single-stranded donors. We illustrate that, alongside sequence-perfect, on-target integrations, the system can also produce an array of incorrect alleles. These include unintended point mutations, small or larger sequence rearrangements, some of them likely based on local micro-homology, and additional donor integrations.

These events are unpredictable by-products that must be excluded in the process of the validation of newly established mutant lines. Importantly, these unwanted events are much more frequent than the much-publicised potential off-target effects of CRISPR/Cas9 reagents.

Whilst our recent study focuses on the use of long single-stranded donors, we and others have previously shown unpredictable events arising from the use of CRISPR/Cas9 reagents alone or in combination with oligonucleotides. As such, an increasing body of evidence is being compiled to indicate that model validation is the newest challenge for the genome editing community. After all, the quality and reproducibility of research based on genome editing mutants depends entirely on the thorough characterization of the mutant in question.

Validating new mutant lines is essential

These observations by no means diminish the value of the system. They simply illustrate the importance of a comprehensive validation of new mutants, including sequencing of the locus and counting of the number of copies of donor sequence integrations. Whilst we await additional tools to ensure an increased accuracy in obtaining the desired outcome of genome editing, we must keep in mind that although unpredictable, the existing strategies remain efficient at generating sought-after mutants.

Going forwards, it is clear that those involved in the genome editing field have a number of hurdles to clear: 1) an awareness and understanding of the unpredictability of the system (mosaicism, single base changes, rearrangements, additional integrations) should be disseminated and emphasized to the research community; 2) broadly recognized standards for the validation and documentation of mutants should then be established; 3) developing strategies to enhance the chances that only the desired on-target result is produced remains the holy grail of genome editing.

Lydia Teboul

Lydia Teboul is a molecular biologist with a background in mouse embryology at the Mary Lyon Centre at the Medical Research Council Harwell Institute. She specialises in genome engineering and is currently managing a large-scale mouse mutant production programme employing genome editing.

Latest posts by Lydia Teboul (see all)

View the latest posts on the On Biology homepage