Skip to main content
Premium Trial:

Request an Annual Quote

CRISPR Researchers Find Cytosine Base Editors Generate Substantial Off-Target Effects

NEW YORK (GenomeWeb) – In a study published in Science late last week, researchers from the US, China, and Germany reported a new method they've developed to detect off-target mutations created from editing one blastomere of two-cell mouse embryos using either CRISPR-Cas9 or one of two base editing technologies.

This method — named GOTI (Genome-wide Off-target analysis by Two-cell embryo Injection) — found that off-target single nucleotide variants (SNVs) were rare in embryos edited by CRISPR-Cas9 or the adenine base editor 7.10 (ABE7.10), which both caused off-target effects with a frequency close to the spontaneous mutation rate. However, the researchers also found that the cytosine base editor 3 (BE3) induced SNVs with frequencies more than 20-fold higher than the spontaneous mutation rate.

The team began by injecting CRISPR-Cas9, BE3, or ABE7.10, along with Cre mRNA, into one blastomere of two-cell embryos derived from mice engineered to express the tdTomato fluorescent protein. The progeny cells of the edited and non-edited blastomeres were then sorted on tdTomato expression in the gene-edited cells at embryonic day 14.5, and the researchers performed whole-genome sequencing separately on the tdTomato-positive and tdTomato-negative cells. SNVs and indels were called by three algorithms in the tdTomato-positive sample, using the tdTomato-negative sample from the same embryo as the reference.

They validated the on-target efficiency of their approach in embryos at the eight-cell and embryonic day-14.5 stages by Sanger sequencing, and then performed WGS at an average depth of 47x on 46 samples from 23 embryonic day-14.5 embryos in order to further explore the on-target efficiency and potential genome-wide off-target effects.

"The SNVs detected in the Cre- or Cas9-treated samples were likely caused by spontaneous mutations during genome replication during development, since the number of variants was within the range of simulated spontaneous mutations and no sequence similarity was observed between the adjacent sequences of the identified SNVs and the target sites," the authors wrote. "Surprisingly, we found on average 283 SNVs/embryo in BE3-treated embryos, a level at least 20 times higher than that observed in Cre- or Cas9-treated embryos. By contrast, ABE7.10 generated on average 10 SNVs/embryo, with a frequency close to the spontaneous mutation rate."

The researchers further found that SNVs induced by BE3 were significantly enriched in transcribed regions, especially in genes with high expression, and that none of the off-target sites were shared by any of the BE3-treated embryos or overlapped with predicted off-target mutations.

"Besides, no similarity was observed between the off-target and on-target sequences," they added, "whereas the top predicted off-target sites showed high sequence similarity with BE3 on-target loci. Thus, the BE3 off-target SNVs were sgRNA-independent and likely caused by overexpression of APOBEC1."

The team also noted that numerous de novo SNVs induced by BE3 turned up in this analysis which had not been reported in previous studies. The investigators speculated that this could be because GOTI examines the cell population derived from a single gene-edited blastomere, whereas previous studies have used large pools of cells where editing is variable, resulting in loss of signal for random off-target effects. They concluded that off-target effects of base editors may be reduced by decreasing the DNA binding ability of APOBEC1 or using different versions of cytidine deaminase, and that GOTI could be useful in examining the off-target effects of various gene editing tools.

In a different study published today in Nature Biotechnology, researchers from South Korea showed that Cas9, BE3, and ABE7.10 often recognize different off-target sites. They also used a targeted sequencing method along with preassembled adenine base editor ribonucleoproteins, modified guide RNAs, and Sniper/Cas9 to reduce adenine base editor off-target activity in human cells.

In a previous study, these researchers used a modified Digenome-seq protocol to detect off-target effects from cytosine base editors in the human genome. For this study, they sought to profile the off-target activity of adenine base editors using the same protocol.

They began by analyzing whether the off-target activities of ABE7.10, BE3, and CRISPR-Cas9 were different from each other by testing a series of mismatched sgRNAs targeted to endogenous genomic loci, and found that base editors and Cas9 nucleases tolerated most of the single or double mismatches in the sgRNAs but were poorly active when combined with most of the sgRNAs having triple or quadruple mismatches.

They also noted that the tolerance of Cas9, ABE7.10, and BE3 for mismatched sgRNAs was often different from each other, suggesting that ABE7.10, BE3, and Cas9 could recognize separate sets of off-target sites in the human genome. The researchers concluded that a method was needed to determine the genome-wide specificity of adenine base editors in an unbiased manner.

After various analyses suggested that Digenome-seq could be used to comprehensively map genome-wide on-target and off-target sites of adenine base editing, the researchers moved on to determine ABE7.10 off-target sites in the human genome. They assigned a DNA cleavage score to each base pair position across the entire genome, allowing direct comparisons between adenine and cytosine base editors and Cas9, and found that adenine base editors in general are more specific than Cas9 but can recognize different sets of off-target sites.

They also unexpectedly found that ABE7.10 activities were correlated more strongly with Cas9 activities than with BE3 activities. Among six sites edited at relatively high frequencies with ABE7.10 but poorly with BE3, two sites had no cytosine within the BE3 editing window of several nucleotides.

In a subsequent experiment to minimize or avoid off-target effects from adenine base editing, the researchers tested three different methods that have proven to reduce off-target effects from Cas9 and BE3 editing: sgRNA modification, delivery of RNPs in lieu of plasmids, and use of an engineered Cas9 variant called Sniper-Cas9. They found that extended sgRNAs reduced off-target activity at almost every site without sacrificing on-target activity, while truncated sgRNAs reduced both on-target and off-target activity at most sites and exacerbated off-target effects at sites with mismatches at or near the 5' terminus.

Further, the use of Sniper ABE7.10 or delivery of ABE7.10 RNPs rather than plasmids improved the specificity of base editing, and combining Sniper ABE7.10 with modified sgRNAs further reduced off-target activity at many validated off-target sites.