Skip to main content
Premium Trial:

Request an Annual Quote

New Evaluation Method for CRISPR Base Editors Captures Novel Off-Target Effects

NEW YORK – Researchers at Peking University and Tsinghua University in China have developed a method to evaluate the editing specificity of cytosine base editors, or CBEs, by detecting CBE-induced off-target sites at the genome-wide level.

In a study published on Monday in Nature Methods, the researchers said the method, termed Detect-seq, uses chemical labeling and biotin pulldown to trace the editing intermediate deoxyuridine in order to reveal the CBE's editing sites across the genome, or editome. Using Detect-seq, they discovered unexpected edits outside the protospacer sequence and on the target strand. These off-target edits were prevalent, their occurrences exhibited cell-type dependency, and they could not be predicted based on the sequence of the guide RNA, the researchers said.

Further, they also found edits outside of the protospacer sequence and on the target strand near the on-target sites they tested, challenging the conventional knowledge that CBEs do not induce proximal off-target mutations.

"The cytosine base editor is one of the most promising genome editing tools to correct human pathogenic mutations, yet its genome-wide specificity must be thoroughly addressed before it is ready for clinical use," senior author Chengqi Yi said in an email. "Previous efforts examining CBE-induced DNA off-targets have revealed significant off-target effects, yet an unbiased method for specificity evaluation is still lacking."

Because of its selective chemical labeling and biotin pulldown procedures, Detect-seq is highly sensitive and readily distinguishes genuine off-target editing from SNVs, sequencing errors, and other similar events, Yi noted. It also detects random off-target events and Cas9-dependent off-targets, in addition to the out-of-protospacer and target-strand events.

"Our findings have multiple implications," Yi added. "We report a sensitive and unbiased method that clarifies potential confusions in off-target evaluation of CBE. Our study discovers out-of-protospacer and target-strand off-target editing, two novel types of off-targets that have been missed by previous methods. This finding adds to the current knowledge of CBE specificity. While more specific BEs remains to be developed, Detect-seq accurately reports the editing outcome of existing CBE tools, thereby enabling safety and risk assessment of potential therapeutic applications."

Base editors were created by tethering a base modification enzyme — rat APOBEC1 for a CBE or Escherichia coli TadA for an adenine base editor, or ABE — to a catalytically impaired Cas9 nuclease. CBEs use a deaminase to catalyze conversions from deoxycytidine to deoxyuridine, finally resulting in a transition to thymidine restricted within an editing window in the nontarget strand.

Detect-seq is based on chemical labeling and enrichment of deoxyuridine to trace the in vivo editing events in an unbiased manner. Specifically, the researchers hypothesized that when genomic DNA is purified from CBE-edited cells, deoxyuridine generated by a CBE in vivo can be recognized by uracil-DNA glycosylase, or UDG. Detect-seq uses an in vitro-reconstituted base-excision repair reaction to achieve specific labeling of deoxyuridine.

When they used Detect-seq to evaluate the off-target activity of the CBE BE4max in HEK293T or MCF7 cells, the researchers observed peaks with characteristic mutation patterns at the on-target sites. These features magnified the signal and could be readily distinguished from genomic background including SNVs and sequencing errors.

The researchers then developed a bioinformatic pipeline to identify Detect-seq patterns throughout the whole genome and searched for putative sgRNA binding sites, or pRBS, within the genomic loci identified by Detect-seq. They then systematically performed Detect-seq for CBE architectures without sgRNA, APOBEC1, and uracil-DNA glycosylase inhibitor (UGI) and found that Detect-seq signals for these pRBS-containing loci either dropped to background levels when sgRNAs or APOBEC1 were omitted, or responded to a varying degree to UGI deletion, indicating that these pRBS-containing loci were typical Cas9-dependent off-target sites.

Taken together, their analyses indicated that CBEs induced both Cas9-dependent and Cas9-independent off-target edits.

Unexpectedly, the researchers also observed evident Detect-seq signals outside the pRBSs. Targeted amplicon sequencing showed that they were genuine off-target mutations, and that these edits were dependent on the CBE complex. Some of these signals were also several or more than a hundred bases away from the pRBS, suggesting that CBEs can edit bases far away from the canonical editing window, the researchers said.

They also found that out-of-protospacer edits were prevalent, with nearly half of the typical Cas9-dependent off-target sites identified by Detect-seq being flanked by out-of-protospacer edits. One site verified by targeted amplicon sequencing showed an editing ratio of about 7.4 percent. This indicates that out-of-protospacer editing can lead to severe biological consequences, the researchers noted.

After they observed that out-of-protospacer and target-strand edits prevalently occurred near typical Cas9-dependent off-targets, the researchers interrogated the on-target sites for such edits. They were able to find both out-of-protospacer and target-strand edits for all the on-target sites they had previously tested. The edited cytosines could be located a few dozen bases away from the on-target sites, exhibiting low, but evident, editing ratios.

Unexpectedly, the researchers noted that base editors using Cas12a induced more off-target edits than base editors using Cas9, despite Cas12a's higher accuracy. They speculated that the higher binding sequence tolerance of Cas12a compared to Cas9, or the difference between the nuclease and base editor system, might have contribute to the discrepancy.

"At this moment, we do not know whether or not ABEs, [glycosylase base editors], [fusion ABEs/CBEs], etc. also possess such novel off-target edits," Yi added. "But our idea of off-target evaluation via capturing editing intermediates can certainly be applied to the base editors."