When Zinc Fingers Miss the Mark

But it’s not a perfect system: sometimes the molecule may bind and clip a different, nearly identical DNA sequence, potentially killing cells.

To systematically characterize those off-target cleavage sites, Harvard chemical biologist David Liu and his colleagues tested two ZFNs against a library of 100 billion snippets of DNA, some of which appear in the human genome. Most often, the nucleases cleaved the target site. But they occasionally cut other similar sequences as well—including one gene associated with a cancer signaling pathway.

“A superficial interpretation of our paper might lead one to be pessimistic about zinc finger nucleases, but actually I’m optimistic,” Liu said. In addition to confirming that the number of off-target breaks decreased with lower concentrations of ZFNs, the researchers found that using ZFNs that bind less avidly to the target sequence seemed to have fewer unintentional breaks, he said. That suggests it may be possible to design ZFN therapies in a way that minimizes those off-target effects. The researchers published their results in Nature Methods.

In the second study, published in Nature Biotechnology, researchers dosed human leukemia cells with a ZFN which cuts the CCR-5 receptor. They identified the cut locations by transfecting the cells with tagged virus particles that bound to the broken ends of the DNA, and found that by and large, the ZFN bound to the target CCR-5 DNA. About 1 in 20,000 times, however, it cleaved a second receptor gene nearly identical in sequence, as well as a few other similar sequences even more rarely. But the researchers used an extremely high concentration of ZFN, and used a cell line that is very permissive of ZFN action, to see what the worst case scenario would be, said coauthor Phillip Gregory, chief scientific officer of Sangamo BioSciences. The low rate of off-target cutting even under these conditions “validates our expectations that the proteins would be tremendously specific,” and suggests that much lower medical doses applied in the clinic would almost never be expected to cause off-target cuts, he said.

The methods might one day be used in early drug development to pick candidates with the best specificity, Barbas said. It’s unclear, however, just how comprehensive the new ZFN tests are, he said. The tagged virus particle method, for example, may miss some off-target cleavage, because the viral tags may not bind to every single break. Furthermore, Barbas added, unlike DNA in a test tube, cellular DNA is tightly wound into chromatin, so many of the binding sites found by the test tube method might be shielded from ZFNs and never be cut in living cells.

So while the new tests may be key tools for early drug development, a complete picture will only come once a person’s entire genome sequence can be had for $1000, he said, when researchers can test people who receive ZFN treatments for every off-target break.

R. Gabriel, et. al, “An unbiased genome-wide analysis of zinc-finger nuclease specificity,” Nature Biotechnology, doi:10.1038/nbt.1948, 2011.

V. Pattanayak, et. al, “Revealing off-target cleavage specificities of zinc-finger nucleases by in vitro selection,” Nature Methods, doi:10.1038/nmeth.1670, 2011.