Genome editing enables precise DNA sequence changes to be engineered within chromatin of living cells. This has transformed the way in which we produce genetically modified cells, plants, livestock, and laboratory animals, and contributed to a renaissance in human gene therapy. Simply stated, genome editing involves targeting a double-strand break (DSB) to a specific site in the genome using synthetic endonucleases. DSBs are then repaired by the cell to produce “edits,” comprising either small deletions and insertions or predetermined changes copied seamlessly from a “donor” DNA molecule.
Several different types of synthetic endonuclease have been used to achieve targeted DSBs. Zinc-finger and TAL effector-based nucleases (TALENs) rely on DNA sequence recognition systems that evolved to function in eukaryotic chromatin. However, in recent years, CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats)-Cas9 has emerged as the most popular approach. CRISPR systems exist naturally in many prokaryotic species as a form of immune defense system, recognizing and degrading nucleic acids of foreign origin. The sequence specificity of Cas9 complexes can be reprogrammed by changing a 20 nucleotide region of a short guide RNA molecule, which forms Watson-Crick base pairs with the genomic target.
Given the natural substrates of CRISPR-systems in prokaryotes, their ability to function in eukaryotic chromatin is remarkable. The eukaryotic genome is compacted by wrapping around nucleosomes, which have previously been shown to impede CRISPR activity in vitro. Individual nucleosomes can be modified post-translationally to affect their mobility and biochemical properties. Similarly, CpG dinucleotides in genomic DNA can be modified through the addition of methyl groups. These chromatin modifications are deposited at different levels in different regions of the genome and serve to control the association of diffusible regulators involved in transcription, DNA replication, and repair. The aim of our study was to investigate how these different chromatin states influence the ability of CRISPR-Cas9 to edit the mammalian genome.
To do this, we made use of a natural epigenetic process called genomic imprinting. Imprinting leads to the transcriptional silencing of one of two alleles of a gene in every nucleus, which occurs due to allele-specific modifications to chromatin. Although imprinting evolved in mammals to regulate transcription, we asked whether it could be “repurposed” to measure how modified chromatin impacts the frequency and spectrum of genome edits. Within imprinted regions, two identical target sites for the same CRISPR-Cas9 endonuclease exist in the same cell nucleus but are packaged into very different chromatin states. Consequently, any differences in the frequency or type of genome edits that occur between alleles of imprinted genes must be due to epigenetic properties of the chromatin.
In initial experiments where genome editing was allowed to proceed for 4 days, we found only very subtle differences in mutation frequency between transcriptionally active versus repressed alleles of imprinted gene promoters. However, when exposure to CRISPR was short (< 24 hours), or when the level of CRISPR expressed in cells was low, mutation frequencies in compacted chromatin were dramatically reduced. This occurred at least partly through the inhibition of Cas9 binding on the repressed allele. Together, these results showed that the inhibitory effect of transcriptionally repressed chromatin on genome editing is inversely proportional to the level of exposure to CRISPR-Cas9.
Using the same experimental system, we also asked whether the outcome of DNA repair following CRISPR cleavage was influenced by the chromatin state. We found that neither the size of deletions nor the ratio of deletions and insertions to precise sequence changes introduced from oligonucleotide donor molecules was significantly different between the two alleles of imprinted genes. Although our assay could not detect perfect break repair that did not result in mutations, it nonetheless suggested that the types of mutation that can be obtained by CRISPR-Cas9 genome editing is not dependent on chromatin state. This was surprising because prior studies using restriction enzymes to induce DSBs have suggested that different DSB repair pathways operate at different efficiency in different chromatin environments. However, the binding of Cas9 to specific DNA sites involves unwinding of the double helix and remodeling of the local chromatin environment. We speculate that this chromatin remodeling activity may clear pre-existing epigenetic modifications, making them irrelevant to the DNA repair process. Further work is now required to determine whether this hypothesis is true, and if so, whether the pre-existing epigenetic state recovers after the repair is complete.
Overall, our study suggests that chromatin modifications are an important determinant of genome editing outcomes when the level of exposure to CRISPR-Cas9 is limiting. This would be especially true where this technology is being used in a therapeutic setting, where minimal exposure is desirable to avoid unwanted mutations at off-target sites.
These findings are described in the article entitled Heterochromatin delays CRISPR-Cas9 mutagenesis but does not influence the outcome of mutagenic DNA repair, recently published in the journal PLOS Biology.