A recent paper, published in Cell, discusses how to anticipate the molecular events that may arise from genome editing and strategies for preventing off-target effects. These developments will be critical to appraise the benefit or risk associated with genome editing.
Genome Editing
Genome editing has produced unprecedented applications in research, healthcare and agriculture. This week, The Nobel Prize in Chemistry 2020 was awarded to Professor Emmanuelle Charpentier and Dr Jennifer A. Doudna for their work on the technology of genome editing.
Whilst this is an extremely exciting time for genome editing, the technology remains unpredictable, both on and away from target loci. This has serious implications for the safety of this technique for therapeutic genome editing.
Generally, genome editing is based on either the CRISPR/CRISPR-associated (Cas) system, zinc finger nucleases, or transcription activator-like effector nucleases. These molecules act as molecular scissors, by inducing a double-stranded cut in a specific DNA target sequence. The cell can the naturally repair this gap in the DNA, which scientist can manipulate to make changes or ‘edits’ in that part of the genome. For precise gene editing, single- or double-stranded DNA templates are delivered with these nucleases, to repair the break with a specific sequence by homology directed repair (HDR) or non-homologous end joining (NHEJ).
Genome Editing Produces Off-Target Effects
The safety of genome-editing technologies is just as crucial as their efficiency. In all fields of application, there are risks of undesired genetic changes that can be triggered by genome editing. These off-target effects are caused by interactions of these enzymes with DNA and DNA repair pathways. They can take many forms: single nucleotide variations, large or complex genomic rearrangements, chromosomal translocation or even loss of one or both arms of a chromosome.
As the use of genome editing increases, more unexpected and negative outcomes of its application are being discovered. For example, gene conversion can occur when DNA is transferred from one genomic location to another by HR. This can occur well beyond the targeted region, representing risks for clinical application. However, this process went unnoticed until recently.
Strategies for Validation
Incomplete characterisation of editing effects can lead to inaccurate or irreproducible research. Thus, a full evaluation of these outcomes is imperative to assess the benefit-risk ratios for this technology. To predict these outcomes, technological investments will be required. These may include technologies that:
- Amplify and sequence target sites, as well as chromosomally linked off-target sites
- Quantify number copies of deleted segments or capture on target duplications/ectopic integrations
- Allow more elaborate assays to inform on potential larger-scale chromosomal rearrangements.
- Where required, extend analysis to the whole genome to predict or identify off-target sites
The strategy for validation will depend on the genome-editing application. For example, some may simply aim to identify the presence of specific variants, whilst others may seek to capture complexity, or ascertain an entire genetic make-up. These suggested technologies allow sufficient scope to customise the complexity of the assay for specific contexts of utilisation.
Scanning the genome
Genome-editing nucleases are targeted to cut DNA at specific sequences in the genome. However, this can often result in changes in other similar sequences across the genome. Therefore, genome wide examination is required to identify these off-target effects.
Copy counting of deleted segments can indicate if a DNA segment has been rearranged rather than simply removed. Similarly, digital PCR (dPCR) is a straightforward assay for these measurements. On a larger genomic scale, comparative genomic hybridisation (CGH) arrays and FISH enable the entire genome to be surveyed. This can identify large sequence alterations that occur away from the nuclease site. Whole genome sequencing would also allow for broad and unbiased capture of genome-editing outcomes. However, this technology is expensive and may not capture complex genetic materials.
Importantly, no single technology can capture all of the unexpected sequence changes which arise from genome editing. Each technique presents its own biases and limitations, but methods are continuously evolving to address these challenges and capture rarer events of sequence modification. Overall, the characterisation of these events is vital to ensure the reliability of this technology and suitability for its intended use.
Preventing the Damage
Beyond understanding these off-target effects, preventing their occurrence is even more desirable. During the early development of CRISPR/Cas9 editing, computational designs of guide RNA were made to predict and model mutagenesis patterns and their potential off-target effects. This also helped to improve the efficiency of these guide RNAs. The expression of Cas9 in specific cell types and its targeted delivery may also reduce the risk of DNA damage. Introducing inactive ribonucleoproteins (RNPs) to off-target sites has also been suggested. These would compete with the gene editing nucleases and prevent off-target effects.
For gene editing events where a DNA template is used, HDR should be promoted above non-homologous end joining to ensure quality is maintained. This could be achieved through the co-delivery of HDR effectors, or pharmacological intervention using small-molecule compounds. Furthermore, repair templates should be delivered at low quantities, to reduce the risk of ectopic insertion across the genome. However, this may affect the overall efficiency of the genome-editing attempt so balance is required.
All of these techniques have been shown to increase the frequently of desired outcomes, but none of them guarantees it. Novel genome editing tools represent further progress towards controlling genome editing outcomes, although we are not yet capable of precision editing.
Concluding Remarks
Existing genome-editing sequences are error-prone and not entirely predictable. However, they are also incredibly powerful. This technology has the power to transform healthcare, improve crop resistance and yields and even livestock. However, it is of the upmost importance that potential collateral genetic damage is predicted and identified, to allow the full potential of this technology to be reached.