When the gene-editing system known as CRISPR is used, DNA breaks occur fairly reliably, but not DNA repairs—specifically, the repairs that are supposed to insert synthetic donor DNA. But repairs may be more likely to “stick,” say Johns Hopkins scientists, if donor DNA is designed properly.
The scientists suggest that homology-directed repair, the usual mechanism for knockin of new genetic material, is more efficient if a few design rules are followed. In a nutshell, the rules are:
- Use donor DNA that’s linear, not circular.
- Keep homology arms (sticky single-stranded overhangs) fairly short, about 35 nucleotides in length.
To arrive at these rules, a Johns Hopkins team led by Geraldine Seydoux, Ph.D., inserted various combinations of donor DNA into human embryonic kidney cells, known for their ability to grow well and for their frequent use in cancer research. The scientists used donor DNA with a gene that codes for a fluorescent protein, which glows green in the cell's nuclear membrane when the gene insertion is successful.
Johns Hopkins research associate Alexandre Paix, Ph.D., found that linear DNA fragments function very well as donors and are two to five times more efficient than circular DNAs (known as plasmids) in human cells. “Linear DNA is very easy to prepare in the laboratory, using PCR,” says Paix, referring to PCR tools that are used to amplify DNA.
Paix also tested various lengths of homology arms. He found that homology arms of optimal length were much shorter than those scientists typically use. Specifically, it was found that homology arms of 33 to 38 nucleotides in length were as successful as those with 518 nucleotides, yielding 10% to 20% successful edits under optimal conditions. In contrast, when the scientists tested homology arms of 15 and 16 nucleotides in length, the insertion success rates dropped by half. They repeated these results in three different locations in the human genome.
They scientists also found that the newly inserted sequence, not counting the homology arms, can be up to 1000 nucleotides in length.
Additional details about the new CRISPR rules appeared November 28 in the Proceedings of the National Academy of Sciences, in an article entitled “Precision Genome Editing Using Synthesis-Dependent Repair of Cas9-Induced DNA Breaks.” The article describes how the Johns Hopkins scientists experimentally determined how donor DNA should be constructed to maximize the recovery of edits without cloning or selection.
“We find that repair is local, polarity sensitive, and prone to template switching, characteristics that are consistent with gene conversion by synthesis-dependent strand annealing,” wrote the article’s authors. “Our findings enable rational design of synthetic donor DNAs for efficient genome editing.”
The team achieved success rates of 10% to 50% with inserts 57 to 993 nucleotides in length. Shorter sequences were more successfully inserted than longer ones. For example, new sequences that were 57, 714, and 993 nucleotides long were successfully inserted 45.4%, 23.5%, and 17.9% of the time, respectively.
Beyond 1000 nucleotides, new inserts with 1122 and 2229 nucleotides had little success—about 0.5% of the time. “At that size, it becomes very difficult to introduce the quantity of donor DNA needed for editing,” said Seydoux. “Cells tend to 'choke' on so much DNA.”
Finally, the team also found that the success rate of editing peaks when the new sequence is positioned within 30 nucleotides from the CRISPR cut site. “Beyond 30 nucleotides, the insertion is not workable,” says Seydoux.
“These parameters should accommodate most genes that scientists are seeking to edit,” Seydoux added. “In fact, most experiments involve editing only two to three nucleotides close to the CRISPR cut site.”
The research team also tested whether the same approach could work in mouse embryos. Using a PCR fragment with 36-nucleotide homology arms, the team successfully inserted a 739-nucleotide-long sequence coding for a fluorescent protein into 27 of 87 (31%) mouse embryos.
CRISPR, which stands for clustered regularly interspaced short palindromic repeats, has gained popularity among scientists in the last five years as a tool to efficiently cut DNA. It was adapted for use in mammalian cells from a natural viral defense process in bacterial cells that involves creating lethal cuts in viral DNA. Essentially, the tool is a streamlined set of molecular “scissors.”
The prevailing belief, among scientists, is that cells repair DNA breaks by inserting a random set of nucleotides, the chemical building blocks of DNA. This usually destroys any gene that's located at the spot where the DNA is broken.
It's also well known to scientists that, occasionally, cells use a different source—a sequence from another piece of DNA, or “donor” DNA—to seal the break in DNA.
Yet scientists regarded using donor DNA as an inefficient way to repair the genome, assuming that it required long homology arms, especially when inserting a long DNA sequence and single-stranded or circular DNA, which are difficult to prepare in long sizes.
“We were surprised to find that cells will readily copy sequences from foreign DNA to repair DNA breaks, as long as the foreign DNAs are linear,” noted Seydoux. “By studying how foreign DNA fragments are copied during the repair process, we came up with some simple rules to make genome editing as efficient as possible, optimize the tool, and do so with confidence.”
The finding that PCR fragments containing edits up to 1 kb require only 35-bp homology sequences to initiate repair of Cas9-induced double-stranded breaks emerged from the experiments with human embryonic kidney cells and mouse embryos.
Seydoux's research team is already using the repair rules to study DNA in Caenorhabditis elegans, a roundworm model organism species, and the researchers are studying whether the repair rules apply to other types of human cells. Before the guidelines are widely adopted, Seydoux advises that they should be tested in more human cell types and other organisms.