To really figure out what genes really do, you gotta get in there and get your hands dirty. That usually means deleting or modifying genes, or other regulatory regions. In a recent “Research Highlight” article, researchers at Johns Hopkins reviewed the three main ways to do this, including the increasingly popular CRISPR/Cas system.
To engineer human and other mammalian genomes at targeted locations, it helps to cut the DNA in a double-strand break, which is then repaired by non-homologous end joining or homologous recombination. That’s when you can really tinker with the sequence.
Regular ol’ homologous recombination has its problems. It’s not that efficient, except in mouse cells. Getting a double-strand break in the right place is hard. If the break is in the right place, though, recombination is efficient.
So over the years scientists have developed three general ways to target double-strand breaks:
Zinc finger nucleases (ZFNs)
Researchers can design ZFNs to cut at certain places in the genome. Basically, they combine a zinc finger protein and a non-specific cleavage domain of a restriction enzyme (FokI). Changing amino acids in the ZF domain changes the specificity, and ZFNs usually recognize a three-base-pair site. That works fine, unless an aspartic acid is around. Then, the recognition sequence is thrown off. This makes it difficult to design the required ZF proteins with a general, modular approach.
Transcription-activator-like effector nucleases (TALENs)
TALENs also target specific genomic sites. To recognize DNA sequences, they use the central repeat domain of TAL effectors. Like ZFNs, TALENs also use FokI to cut the DNA. But TALENs are better than ZFNs because neighboring sequences don’t interfere with recognition.
CRISPR/Cas System
RNA-guided genome engineering based on the prokaryotic CRISPR/Cas system might be the best thing going, though. Bacteria and archaea use it kind of like an immune system to get rid of plasmids and nucleic acids from viruses. It also has good target specificity. But unlike the other two methods, you don’t have to engineer specific enzymes. All you need is the Cas9 enzyme for all the targets in the human genome. And the guide RNAs that you need are simpler and easier to make. Many groups have shown that this system works well in vitro—in fact, it works just as well as or even better than TALENs or ZFNs. The system could have advantages for high-throughput applications because you can multiplex it by using lots of guide RNAs at the same time.
“But the most important question, especially for clinical translation, is as follows: how specific are these methods?” say the researchers, who add that this is still an unanswered question.
We want to know who’s been using these approaches to look at regulatory regions in epigenetics studies, so if that’s you, drop us a line and tell us about your experience.
Get your hands dirty at Genome Biology, February 2013.