NEW YORK (GenomeWeb) – Human genetic variation can alter the efficacy of CRISPR targeting when it's used for therapeutic purposes, and increase the potential for off-target effects, according to the authors of a new study published yesterday in the Proceedings of the National Academy of Sciences.
"The sequence-based targeting for CRISPR systems necessitates consideration of the unique genomes for each patient targeted for therapy," senior author and Harvard University researcher Matthew Canver and his co-authors wrote. "We show using 7,444 whole-genome sequences that SNPs and indels can reduce on-target CRISPR activity and increase off-target potential when targeting therapeutically implicated loci; however, these occurrences are relatively rare. We further identify that differential allele frequencies among populations may result in population-specific alterations in CRISPR targeting specificity."
The researchers added that genetic variation should be considered in the design and evaluation of any CRISPR-based therapy in order to minimize the risk of side effects and maximize the potential for treatment success.
These findings are similar to those of the Broad Institute's David Scott and Feng Zhang, who published an analysis in Nature Medicine in July, in which they said that prescreening patients with whole-genome sequencing and then integrating that information with empirical methods for guiding RNA selection could help researchers design CRISPR-based therapeutics that are both efficacious and safe.
For their new study, Canver and his colleagues sought to evaluate whether genetic variants in individual should be considered for developing CRISPR-based therapies. They mined the literature for loci previously targeted by CRISPR-Cas9 for therapeutic purposes, and put together a list of 23 human and 7 viral genome targets. They also generated a list of guide RNAs (gRNAs) by designing all gRNAs targeting the indicated regions or using the gRNAs from previous studies. They ended up with 2,481 gRNAs targeting human DNA regions and 484 gRNAs targeting viral DNA regions. In addition, 128 nontargeting gRNAs were included as negative controls.
They first investigated whether SNPs altered the number of off-target sites in the genome, using 7,444 genomes from the 1000 Genomes Project, a subset of the gnomAD database, and a French-Canadian genome dataset. "SNPs can alter off-target sites by increasing or decreasing the number of mismatches between a genomic region and the gRNA sequence," the authors wrote. "In addition, SNPs can create or destroy [protospacer adjacent motifs] PAM sequences. Creation of PAM sequences may generate new loci for off-target cleavage while destruction of PAM sequences potentially removes loci for off-target cleavage."
The researchers also investigated whether variants such as indels could affect the gRNAs' on-target specificity. They selected the subset of gRNAs with an aggregate off-target score of 80 percent or less in the reference genome — 481 human-genome-targeting, 150 viral-genome-targeting, and 128 nontargeting gRNAs — and tested them against every haplotype in the 1000 Genomes dataset, including SNP and indel variants.
In total, they found that 9.7 percent of the gRNAs had null local on-target scores in at least one haplotype, and that the most common null haplotype was present in 22.6 percent of haploid genomes. Further, 69 percent of null haplotypes were due to an altered PAM sequence. "Overall, these results suggest that genetic variants in the on-target site can dramatically affect gRNA targeting specificity and efficiency, particularly if the variants are located within PAM sequences," the authors wrote.
Their analysis also suggested that indels are more likely to create more potent novel off-target sites than SNPs.
"Given the creation/alteration of off-target sites in noncoding sequence, WGS would likely be required for this analysis as opposed to whole-exome sequencing. Minimally, our results suggest that on-target sites should be investigated by conventional Sanger sequencing to assure maximal gRNA efficiency," the researchers concluded. "Taken together, our analysis suggests the necessity for preclinical studies to consider variants at the gDNA design stage and/or to validate more than one gRNA for clinical translation to increase the likelihood of providing safe, effective, and personalized therapeutic options for all patients regardless of genotype."