NEW YORK (GenomeWeb) – In a systematic analysis of the effects of CRISPR genome editing at nearly 1,500 target sites, researchers from the Francis Crick Institute found that editing precision varies considerably among genomic sites, with some targets showing one highly preferred indel and others displaying numerous infrequent indels.
As they wrote in a new study in Molecular Cell, the investigators found that editing precision correlates with editing efficiency, and that precise targets and editing outcomes can be predicted based on certain simple rules "that mainly depend on the fourth nucleotide upstream of the protospacer adjacent motif (PAM)." The team also found that indel profiles can be influenced by chromatin features.
"Until now, editing genes with CRISPR has involved a lot of guesswork, frustration, and trial and error," senior author Paola Scaffidi said in a statement. "The effects of CRISPR were thought to be unpredictable and seemingly random, but by analyzing hundreds of edits we were shocked to find that there are actually simple, predictable patterns behind it all. This will fundamentally change the way we use CRISPR, allowing us to study gene function with greater precision and significantly accelerating our science."
To characterize general patterns of indels induced by RNA-guided Cas9 nucleases (RGNs), the researchers selected 1,491 sites across 450 nuclear genes, targeting them with multiple sgRNAs. They selected at least three sites for each gene, spacing the target regions along the genes and using sgRNAs with high predicted activity. In total, 1,248 sites showed detectable indels, ranging from one to 188 per target, with a median of 32.
The team observed several general editing patterns. Single-nucleotide indels were the most frequent type of indel for the majority of targets, with 44 percent and 26 percent of targets showing one-nucleotide insertions or deletions, respectively, as the most common indel, the researchers noted. There were also some sites that showed a preference for longer deletions, up to 41 nucleotides.
Also, the investigators saw that CRISPR-induced indels often resulted in frameshift alterations. On average, 80.1 percent of indels induced at a given site disrupted the gene coding frame, and 81 percent of all detected indels resulted in a frameshift. However, three sites showed strong preference for in-frame indels, which suggested to the team that it may be difficult to successfully induce gene knockout in certain cases.
Next, a hierarchical clustering analysis identified four groups of targets showing similar indel patterns. The researchers divided these targets into sites that preferentially showed small insertions, small deletions, long deletions, or that had no clear preference.
They also observed that sgRNA activity was highly variable, and that the indel count did not correlate with abundance of sgRNAs in the pools, suggesting that sgRNA activity is intrinsically variable.
"The observation that different targets display distinct preferences for certain indel types prompted us to examine the degree of editing precision across sites," the authors wrote. "This analysis revealed a large range of editing precision, with some targets displaying up to 79 distinct, infrequent indels and others showing one dominant indel and only a few additional ones. Overall, we found that for approximately one-fifth of the targets, there is at least a 50 percent chance of inducing a specific indel, but the majority of sites are more unpredictable."
On average, they added, the most common indel frequency for a given site was 34.1 percent, and the median number of observed distinct indels was 12.
Further analyses indicated that editing precision correlates with editing efficiency, indel type, and indel size, and that precise targets exhibit primarily homology-associated insertions and deletions. The researchers also observed that certain DNA sequences have a significant influence on editing precision. Specifically, the fourth nucleotide upstream from the PAM sequence had the strongest influence on editing precision as a single nucleotide. The second, third, and fifth nucleotides upstream from the PAM also had an effect, although it was weaker than that of the fourth.
"We refer to these combined nucleotides as the 'precision core' of a target site," the authors wrote.
They concluded that editing precision is site-specific and can be predicted, which has important implications. "Practically, knowing what editing outcome is likely to occur at a given site maximizes the chance of having a desired sequence alteration, for both clinical and research applications," the investigators said. "Moreover, given the extreme reproducibility of indel patterns, the selection of a precise target combined with experimental validation in model systems could considerably increase safety in clinical applications."