Global detection of DNA repair outcomes induced by CRISPR-Cas9.

Aim of the study and some background information
The authors have developed an in silico analysis method called PEM-Q, which evaluates data from genome-wide examinations of genome-edited cells in such a way that all changes in the DNA induced by CRISPR/Cas can be detected. This includes both changes to the target sequence and changes to off-target regions. The authors divide the results of the cell’s own repair mechanisms into two categories: either the DNA double-strand break (DSB) is reclosed or two different DSBs are connected to one another, also referred to as translocations. If a DSB is reclosed, either the original state is restored or insertions and deletions occur.

Small deletions and small insertions often have the effect that certain target genes are knocked out – this is the most common intention of researchers using gene scissors. Rarely they intend to cause large insertions or deletions. However, large deletions and insertions can also occur unintentionally. Large deletions can affect the DNA sequence and expression of neighboring genes of the actual target gene. In the case of large insertions, DNA fragments that are present in the cell (e.g. the vector to introduce the gene scissors DNA into cells) are inadvertently incorporated into the DNA DSB.

Translocations restructure the DNA sequence and can occur on the same chromosome or on different chromosomes. For translocations to occur, several DSBs have to be present in a cell at the same time and be in close proximity. Several DSBs can arise, for example, through multiplexing, i.e. during the simultaneous modification of several different target sequences, but also through off-target activities of the gene scissors or naturally occurring DSBs.

Results
The authors evaluate several data sets from various CRISPR/Cas9 experiments in human and mouse cells. In the experiments, CRISPR/Cas9 was directed to different target regions of the genome to induce alterations (i.e. SDN-1 applications). The results of the repair of the DNA double-strand break were evaluated with the aid of genome-wide sequencing and the PEM-Q analysis method. They repeatedly come to the conclusion that deletions occur most frequently at the target site, followed by insertions and, to a lesser extent, translocations. In direct comparison, there are slightly fewer translocations in the cells of mice than in human cells.

Deletions
The authors further investigate what type of deletions occur. Mainly small deletions occur, larger deletions to a lesser extent (between 5 and 15%). Interestingly, there are similarities of the DNA-sequence in the genome where large deletions occur: small, repetitive DNA sequences can often be found at the ends of the large deletions. Such DNA sequences are also known as microhomologies. This suggests that a repair mechanism known as MMEJ (microhomology-mediated end joining) is involved in the formation of large deletions during CRISPR/Cas-induced DNA-DSB repair, which has already been suspected in other scientific studies.
In this study, the authors also use cells in which the NHEJ repair mechanism was inactivated and show that this increases the proportion of large deletions during CRISPR/Cas-induced DNA-DSB repair. Most likely, the repair mechanism MMEJ is compensating the absence of the NHEJ repair mechanism.

Insertions
The study also examines what type of deletions occur and how they are distributed. Smaller insertions can very often be found at the target site. Frequently, only a single additional base pair is incorporated. The larger insertions occur at various parts of the genome and are often DNA fragments of the transport vector, which are integrated in the genome. A vector is used during a CRISPR/Cas experiment to introduce the genetic information of the gene scissors into the cells. This is achieved in this study by adeno-associated viruses (AAV). Such viruses no longer contain infectious genes and act as transport vehicles to introduce exogenous DNA into mammalian cells. These transport vehicles are also known as viral vectors; they are often used in basic research. The size of the DNA fragments unintentionally integrated into the genome varies in this study between 1–20 bases (i.e. letters) of the DNA for smaller insertions and 20– approx. 120 bases of DNA for larger insertions. Interestingly, a specific part of the vector DNA is frequently integrated into the genome. If this specific DNA sequence is not present in the vector, there are fewer unwanted DNA insertions. Apparently the structure of this genetic element favors incorporation into the genome. In clinical applications, the introduction of the gene scissor DNA with transport vehicles should be avoided in order to prevent such large insertions.

Translocations
The scientists also investigated how often and where translocations occur in the genome after a CRISPR/Cas9-induced DNA DSB. Most of the translocations can be found on the same chromosome as the corresponding target sequence of the gene scissors. However, translocations also occur on other chromosomes. The scientists investigated (predicted) off-target regions that are similar to the target sequence. It is very likely that the gene scissors will cut and cause an alteration at these DNA regions. In their analysis, the authors show that translocations can indeed occur at off-target regions.

Further results
The scientists also re-evaluated experiments using especially precise Cas variants (e.g. high-fidelity variants, e.g. eCas9, HF1 and FeCas9) in order to reduce the appearance of off-target effects. The repair outcomes had a similar distribution in comparison to normal CRISPR/Cas9 repair outcomes. Thus, unintended changes, such as large deletions, large insertions and translocations, also occur with high fidelity variants of the gene scissors.
In addition, the authors inactivated the repair mechanism NHEJ and showed that, after induction of the DNA DSB by CRISPR/Cas9, there is a general increase in large deletions, large insertions and DSBs within gene regions. Several studies have already found that inactivating the NHEJ repair mechanism could increase the efficiency of the HDR repair mechanism in SDN-2 and SDN-3 applications of CRISPR/Cas. However, the results of this study show that inactivating the cell’s own repair mechanism NHEJ increases the occurrence of unwanted changes in the genome.