Skip to main content
Premium Trial:

Request an Annual Quote

New Sequencing Methods Reveal Off-Target Effects of CRISPR/Cas9


NEW YORK (GenomeWeb) – Two recently published sequencing methods demonstrate the ability to systematically reveal unintended consequences of CRISPR/Cas9 activity in the genome.

The methods could provide researchers with an accurate and unbiased tool to gauge the specificity of the gene editing technology as it moves closer toward clinical therapeutic use, according to experts.

The ability of the Cas9 nuclease to create a double-stranded break is the crux of CRISPR, but scientists have been reporting off-target effects and have to design their experiments to account for it.

"We're still in the process of understanding the specificity of Cas9 nuclease technology," said Fei Ann Ran, a CRISPR/Cas9 researcher at the Broad Institute who was not involved with either publication, told GenomeWeb. Understanding off-target effects is important for being able to confidently use the technology to modify genomic targets. "There isn't a gold standard method in the field," she said.

Along with a lentiviral vector-based approach, these two sequencing methods represent a new suite of tools to determine the effects of Cas9 nuclease activity.

Both of the recent papers describing the methods were published in December in Nature Biotechnology. In one, researchers from Massachusetts General Hospital, led by Shengdar Tsai and Zongli Zheng, described a method called genome-wide, unbiased identification of double-stranded breaks (DSBs) enabled by sequencing (GUIDE-seq).

Meantime, scientists at Boston Children's Hospital, led by Richard Frock and Jizazhi Hu, described an improved method of high-throughput, genome-wide translocation sequencing (HTGTS) that detects DSBs generated by engineered nucleases, which could include Cas9, TALENS, and zinc-fingered nucleases, as well as other processes.

Both methods detect DSBs in the genome, which typically manifest as indel mutations. When these occur in protein coding regions, they can lead to frame shift and early truncation of proteins during translation. They might also hit regulatory regions that could change how a gene is expressed.

There are existing techniques that allow researchers to use Cas9 with a reasonable degree of specificity, including the "double nicking" method developed by the Zhang and Church labs at the Broad Institute; truncated guide RNAs developed by Keith Joung, who is a senior author on the Tsai and Zheng paper; and FOKI-dCas9 fusions, developed by the labs of David Liu of Harvard University and Keith Joung of MGH, who is a senior author on the GUIDE-seq paper.

And there are several ways to measure off-target effects. For instance, one might use ChIP-seq to look at the DNA binding of Cas9, or assume that the off-target locations are homologous to the target sequence given by the guide RNA and use deep sequencing to find the off-target activity. But these methods are affected by a fundamental limitation: "You only find things where you look for them," Tsai said.

Ran explained that the new papers were exciting because they're among the first technological advances allowing researchers to look at genome-wide Cas9 specificity in an unbiased fashion.

The GUIDE-seq method works by tagging DSBs in the genome with a blunt, 34-basepair, double-stranded oligonucleotide by means of a non-homologous end-joining process. The foreign bait oligo features a phosphorthioate modification to protect it from exonuclease activity and to increase the efficiency of integration into the DSBs. Tsai said he was able to incorporate the tags into approximately one-third of DSBs.

Following DSB tagging, the researchers selectively amplified those sections of the genome with the use of primers that complement each of the strands in the oligonucleotide. This gave reads of the adjacent genomic sequence on both sides of the tag, necessary because the tag could be incorporated into the DSB in two directions.

The amplification step yields a sequencing library for high-throughput sequencing with the Illumina MiSeq instrument.

Looking for overlapping reads (because the tag could be incorporated in the DSB in two directions), the researchers used those as windows into DSBs, and mapped them down to the nucleotide level.

GUIDE-seq indicated that CRISPR/Cas9 guide RNAs could induce off-target activity at sites numbering in the range of zero to more than 150. GUIDE-seq could detect off-target activity that happened with a frequency of only about 0.1 percent. In many cases, the location of DSBs was not predicted by existing methods of determining off-target activity. The authors wrote that the GUIDE-seq method might still be improved by deeper sequencing.

Furthermore, Tsai and his colleagues used GUIDE-seq to show that truncated guide RNAs of about 17 or 18 nucleotides (as opposed to 100) could lead to a threefold reduction in off-target activity. The technique also revealed DSB hotspots that were determined to be unrelated to CRISPR/Cas9 at all.

Similarly, Frock et al. showed off-target activity by pointing out translocations using high-throughput sequencing combined with linear-amplification-mediated PCR.

"We can show that there are hundreds of translocations that occur," from CRISPR/Cas9 off-target activity, said senior author Frederick Alt. "We know where they go and what the potential damage is."

The proof-of-concept study looked at the off-target activity of four guide RNAs targeting the RAG1 gene, a proposed target for human gene correction therapy.

HTGTS detected 33 significant hotspots from two of the guide RNAs, but none from the other two. In addition, the researchers used HTGTS to show that the double nicking method, first described by Ran, reduced CRISPR/Cas9 off-target activity.

In a perspective piece in the same issue of Nature Biotechnology which summarizes the approaches, Richard Gabriel, Christof von Kalle, and Manfred Schmidt of the National Center for Tumor Diseases and German Cancer Research Center in Heidelberg, Germany, noted that sequences located on the same chromosome as the bait sequence will preferentially fuse to it, and therefore HTGTS "may overestimate on-target versus off-target ratios when using a specific bait sequence or overestimate the frequency of off-target double-strand breaks that are located on the same chromosome as the bait sequence."

What HTGTS definitely shows, however, is that off-target effects of CRISPR/Cas9 can be more than trivial.

"One thing people have suspected for a long time is that these nuclease cleavages can lead to translocations and larger structural rearrangements that are more extensive than indel mutations," Ran said.

This highlights the importance of understanding the rules behind off-target activity before CRISPR/Cas9 is pushed into use as a clinical therapy.

"You can imagine that if you're treating large populations of cells, billions, you'd want to understand low-level off-target effects," Tsai, one of the authors of the GUIDE-seq paper, told GenomeWeb. "If you hit tumor suppressor genes, you could cause deleterious effects."

Ran said that understanding specificity of CRISPR/Cas9 can also help predict cleavage efficiency for a given on- or off-target. "We might gain some additional understanding of what the rules governing Cas9 specificity might be. In the future, the hope is that we can use this information to help us understand the rules governing Cas9 specificity more accurately," she said.

The ideal situation, both Ran and Tsai said, is a tool that could predict the off-targets for any given target.

"This becomes very important for potential clinical applications, where both efficiency and specificity matter a great deal," Ran said.