Upper bound on the mutational burden imposed by a CRISPR-Cas9 gene drive element
Authors: Michael Overton; Sean Guy; Sergey Kryazhimskiy
Affiliation: University of California San Diego
Keywords: Experimental evolution
CRISPR-Cas9 gene drives (CCGDs) are poised to expand our ability to control wild populations, such as the eradication of disease vectors. However, Cas9-gRNA complexes can produce off-target dsDNA breaks. This raises the possibility that a CCGD element introduced into a genome that has not previously evolved in its presence could increase rates of mutations, genome instability, and loss of heterozygosity (LOH), with unpredictable long-term evolutionary consequences for the engineered species and the surrounding ecosystem.
To assess the potential evolutionary effects of a CCGD in a naive genome, we performed a large-scale mutation accumulation (MA) experiment in the yeast Saccharomyces cerevisiae. We introduced a CCGD construct into the genome of a diploid hybrid yeast strain, heterozygous at ~50,000 loci to detect LOH events at high resolution (Drive strain). We also created two control strains in the same genetic background, one carrying a neutral marker (WT control) and the other carrying the Cas9 gene alone without a gRNA (Cas9 control). We propagated at least 65 MA lines of each strain for 800 generations, and sequenced whole genomes of the founders and end-point clones.
Using these data, we estimated the rates of spontaneous mutations in the Drive, Cas9 control, and WT control strains to be 4.09×10–11, 5.05×10–11, and 4.68×10–11 per base-pair per generation, respectively, and the rates of LOH events to be 7.65×10–3, 8.27×10–3, and 7.93×10–3 per genome per generation, respectively, all of which were consistent with previous estimates. Neither de novo mutation nor LOH event rates were statistically distinguishable between the strains. Given the statistical power of our study, we estimate with 70% confidence that the presence of a CCGD element increases the genome-wide rates of LOH events and de novo mutations by less than 25% and 70%, respectively. Despite the lack of significant differences for the genome-wide rates, we found a significant difference in the distribution of LOH events across Chromosome V between the WT and Drive strains.
While it is important to examine the evolutionary consequences of CCGD elements in other species, our results suggest that these elements impose at most a weak and likely localized additional mutational burden to a naive genome, and that CCGD-based population control efforts are likely to be evolutionarily safe.