Cas9 Cuts and Consequences; Detecting, Predicting, and Mitigating CRISPR/Cas9 On‐ and Off‐Target Damage

Abstract

Large deletions and genomic re‐arrangements are increasingly recognized as common products of double‐strand break repair at Clustered Regularly Interspaced, Short Palindromic Repeats ‐ CRISPR associated protein 9 (CRISPR/Cas9) on‐target sites. Together with well‐known off‐target editing products from Cas9 target misrecognition, these are important limitations, that need to be addressed. Rigorous assessment of Cas9‐editing is necessary to ensure validity of observed phenotypes in Cas9‐edited cell‐lines and model organisms. Here the mechanisms of Cas9 specificity, and strategies to assess and mitigate unwanted effects of Cas9 editing are reviewed; covering guide‐RNA design, RNA modifications, Cas9 modifications, control of Cas9 activity; computational prediction for off‐targets, and experimental methods for detecting Cas9 cleavage. Although recognition of the prevalence of on‐ and off‐target effects of Cas9 editing has increased in recent years, broader uptake across the gene editing community will be important in determining the specificity of Cas9 across diverse applications and organisms.

crisprSQL: a novel database platform for CRISPR/Cas off-target cleavage assays

“…IDLV ( 16 ) and GUIDE-seq ( 6 ) rely on the non-homologous end joining-mediated integration of sequence markers into sites upon DSB. IDLV integration does not always happen at the exact DSB location, whilst GUIDE-seq is limited to blunt-end DSBs ( 29 ). Typical sensitivities range from 0.1% ( 6 , 15 ) to 1% ( 16 ).…”

“…SITE-Seq ( 22 ) has been shown to overestimate off-targets relative to those observed in a cellular context. CIRCLE-seq ( 20 ) shows high sensitivity through enrichment of Cas9-cleaved genomic DNA, which leads to a considerable level of background noise overshadowing true positive off-targets ( 29 ). DIG-seq ( 23 ) uses chromatin-associated DNA and can therefore assess the influence of chromatin on Cas9 cleavage.…”

Advances in identifying GM plants: toward the routine detection of ‘hidden’ and ‘new’ GMOs

“…The vast number of publications on improving the specificity and efficiency, and reducing the number of off-target or unintentional on-target modifications show the widespread accidental side effects of NBT techniques (Rezza et al, 2019). In most cases, we are unable to affirm that they are not scars from in vitro cultures and related techniques (Ahmad et al, 2020;Anderson et al, 2008;Cullot et al, 2019;Gauchier et al, 2020;Germini et al, 2018;Kawall et al, 2020;Khan et al, 2017;Naeem et al, 2020;Newman et al, 2020;Peng et al, 2016;Thomas et al, 2019a,b;Wienert et al, 2019). Our inability to master these techniques is such that it is still recommended to continue working with TALEN for some species rather than with CRISPR-Cas (Jain et al, 2021;Khalil, 2020).…”

“…The ability to detect those unintended changes also relies on representative and dependable genomes and epigenomes, distinguishing the 'core genome' and pangenomes (Hurgobin and Edwards, 2017). Even these reference genomes remain difficult to constitute due to the numerous defects of sequencing techniques such as the preparation of sequencing libraries, the somatic mutations, and chromosomal rearrangements, not to mention the PCR amplification errors and significant imperfections of software in assembling and comparing sequences (Bertheau, 2019;Bolukbasi et al, 2016;Ekblom and Wolf, 2014;Newman et al, 2020;Ravindran, 2018;Robin et al, 2016;Wienert et al, 2019). Building up genomes of quality is rare, even for human ones (Ballouz et al, 2019;Bertheau, 2019;Hart et al, 2014;Michael and VanBuren, 2020;Tello-Ruiz et al, 2021).…”

“…As Stella and Montoya (2016) remind us, it is technically difficult and costly to attempt detection by sequencing low-and medium-frequency off-targets. Moreover, off-target detection capabilities vary significantly for the same sgRNA, gene-editing technique and the use of biased or 'unbiased' detection methods (Martin et al, 2016;Newman et al, 2020). Unintended modifications also vary between clonal cells, while unintended changes are searched only in the cell populations, thus diluting the low-frequency unintended changes (Kimberland et al, 2018;Sentmanat et al, 2018).…”