CRISPR-Cas9 is a genome manipulation method that allows us to modify a specific sequence in the genome of different organisms. It is used to systematically introduce double-strand DNA breaks at a selected target sequence in the genome and relies on cell's own repair mechanisms to correct them according to the DNA template that we introduce into the cell. In the yeast Saccharomyces cerevisiae, double-strand DNA breaks can be repaired in two ways. The first one is homologous recombination (HR), which is considered to be dominant over non-homologous end joining (NHEJ). NHEJ often results in indels and nucleotide substitutions at the site of the double-strand break.
In order to develop a method to easily and efficiently replace a single nucleotide in the yeast genome with minimal off-target effects, we tried to characterise as good as possible the unwanted effects that arise when a single nucleotide is replaced with CRISPR-Cas9. We characterized the impact of the deletion of the NHEJ-promoting gene NEJ1, the type of DNA template used for double-strand break repair and the impact of system components loss after the CRISPR-Cas9 activity on the efficiency of the system itself and the formation of indels.
Indels and other off-target effects were characterised on the level of the individual target locus (IXR1) and on the whole genome level. For this, we used next-generation sequencing methods, namely amplicon sequencing and whole genome sequencing. Using amplicon sequencing we characterised the effect of CRISPR-Cas9 on the sequence of the Cas9 target site and in the proximity of the DNA double-strand break if the cell did not contain the matrix for HR. We also used whole genome sequencing to characterise indels and nucleotide substitutions that occurred elsewhere in the genome and were not included in the amplicon sequencing analysis.
We found that after CRISPR-Cas9 action, indels almost always appear in the vicinity of the target site on the genome if the appropriate HR template is not provided, and that their appearance is most likely not random.
|