Scientists at CRISPR Therapeutics have compared three commonly used methods to detect off-target gene editing activity of the CRISPR-Cas9 machinery and found them to be equally adept, although each had its own strengths and weaknesses.
The report – “Evaluation of Homology-Independent CRISPR-Cas9 Off-Target Assessment Methods” – is published today in The CRISPR Journal. “The biggest takeaway is that this study validates the use of CRISPR-Cas9 as a gene editing approach for the development of clinical therapeutics,” says senior author Andrew Kernytsky, PhD.
A bench-to-bedside translation for the Nobel Prize-winning CRISPR-Cas9 genome editing technology is fast becoming a reality. Earlier this month, CRISPR Therapeutics reported positive preliminary results in clinical trials treating patients with sickle-cell disease and beta-thalassemia. Two other genome editing biotechs, Editas Medicine and Intelia Therapeutics, have also initiated CRISPR-based clinical trials.
This progress is due in part to the development of methods for detecting and minimizing unintentional off-target edits. The specificity of targeted CRISPR-Cas9 genome editing rests on guide RNAs, which serve the dual function of recognizing the desired DNA target and guiding the Cas9 nuclease to it. But if the guide RNA recognizes a genomic site with a similar sequence to the target, off-target and potentially deleterious mutations can occur.
Potential off-target sites are typically identified using a combination of homology-dependent and homology-independent (bioinformatic) approaches followed by deep sequencing to confirm if CRISPR-Cas editing activity occurs at these sites.
Homology-dependent approaches computationally suggest off-target sites based on the degree of similarity between the target DNA sequence and the hybridizing section of the guide RNA, together with the presence of an adjacent PAM sequence. Sites that are a close match to the target stand a greater chance of getting erroneously edited and are said to have a shorter “editing distance”. Homology-independent approaches empirically nominate off-target sites based on genome-wide assays, which are then confirmed using next generation sequencing (NGS).
In the new research paper, Kernytsky and colleagues at CRISPR Therapeutics performed a comprehensive evaluation of three homology-independent off-target assessment methods, benchmarked by sequencing 75,000 homology-nominated sites.
“A concern that led us to this study was that one or more of the methods would fail to detect certain off-target sites,” Kernytsky told GEN.
The authors treated HEK293T cells with Cas9 and eight different guide RNAs to compare the performance of the cell-based assay GUIDE-seq, and two biochemical assays, CIRCLE-seq and SITE-seq, to bioinformatically nominated sites. Cell-based assays take advantage of the cell’s internal repair machinery to mark the editing sites. These tagged sites are then enriched and sequenced to detect edits.
“The major difference between the two assay types,” the authors note, “is that cell-based assays nominate sites in accessible regions of the genome in a cell type-specific manner, whereas biochemical assays potentially nominate cleavable sites irrespective of chromatin accessibility.”
The relative abilities of cell-based and biochemical assays to identify true sequence-confirmed off-targets have not been systematically investigated at this scale. Understanding the relative performance of these assays will aid researchers in their choice of methods and experimental design.
“It is exciting that guides have been used without detectible off-target cutting, which is why this study focused on testing the three assays using eight guide RNAs that have been previously published to have off-target events. This provided a more informative comparison than using clinical guide RNAs that lacked off-target events,” says T.J. Cradick, PhD, co-senior author on the study. (Cradick, formerly with CRISPR Therapeutics, is now with a gene editing start-up, Excision BioTherapeutics.)
“The three methods performed similarly in nominating sequence-confirmed off-target sites, but with large differences in the total number of sites nominated,” the authors note. “When combined with homology-dependent nomination methods and confirmation by sequencing, all three off-target nomination methods provide a comprehensive assessment of off-target activity.”
“None of the three methods shows a systematic bias against detecting one type of off-target site versus another,” says Kernytsky.
While all three genome-wide methods nominated most of the sequence-confirmed edited sites, they differed vastly in the number of false-positives––sites suggested as potential off-target sites that are found to be unedited upon hybrid capture sequencing.
“In this sense, the cellular assay GUIDE-seq has the highest precision and CIRCLE-seq has the lowest,” the authors note.
Going further, the investigators correlated the nomination signal with editing rate in cells. “Nomination assays would have greater utility beyond off-target site nomination if the read counts they produced were to reflect editing in cells quantitatively. This would enable prioritization of gRNAs based on the inferred frequency of editing at off-target sites,” the researchers note.
GUIDE-seq read counts correlated highly with the indel mutation frequency difference determined by sequencing, with on-target sites having the highest GUIDE-seq read counts. In the case of CIRCLE-seq, this correlation was lower and for SITE-seq there was no correlation between read counts and observed indel mutation frequency difference between treated and control samples.
GUIDE-seq’s low false-positive rate and the high correlation of its signal with observed editing make it a good choice for nominating off-target sites for ex vivo CRISPR-Cas therapies.
Based on this comprehensive study, the authors offer several practical suggestions in guiding the assessment of CRISPR-Cas9 off-target effects. These include the prioritization of candidate guide RNAs using computational off-target and on-target site prediction methods; identification of potential off-target sites using both homology-dependent computational methods and homology-independent empirical methods; and sequencing at a depth of more than 5,000x to increase the sensitivity of detecting low-frequency editing activity.