Science Editor

Zinc finger nucleases – which edit genes by inducing double-stranded DNA breaks at specific sites in the genome, then repair those same breaks with template DNA provided by the zinc finger nuclease complex itself – are clearly more specific than vector-based gene therapy, which delivers its therapeutic payload into patients DNA at essentially random places.

But how much more specific? To date, there has not been a method that allowed scientists to identify all DNA sequences that zinc finger nucleases cut.

Two papers published this week described two different ways of looking at the entire genome to see where specific zinc finger nucleases bind.

Philip Gregory, chief scientific officer at Richmond, Calif.-based Sangamo Biosciences and co-author on one of the papers, was sanguine about what the results published in the two papers meant in practice. "These two papers confirm what we already thought: that these nucleases are highly specific," he told BioWorld Today.

In their paper, which appeared in the Aug. 7, 2011, online edition of Nature Biotechnology, Gregory and his co-authors used a virus to tag all sites where two zinc finger nucleases cleave the genome. "We let the cell tell us where the nuclease cleaves," Gregory said. The team then sequenced those cleavage sites to see whether they could predict what sorts of DNA sequences would lead to off-target binding.

The main finding of the paper was, essentially, that zinc finger nucleases work as advertised: "The vast majority of all events" – in his team's paper, just almost 95 percent of them – "happen precisely at the intended target," Gregory said.

That contrasts sharply with vector-based gene therapy, where vector integration into the genome is pretty much random and so "the introduction of the therapeutic cassette automatically comes at the price of an unknown integration event," he said.

Furthermore, he added, even in those cases where the nucleases cut where they are not supposed to, they do so in predictable ways – and understanding where they cut gives clues to how to avoid having them go for sites where integration could cause real problems.

Another paper, which appeared in the Aug. 7, 2011, issue of Nature Methods, used a combination of methods to look at two different zinc finger nucleases. In this second paper, the authors looked at CCR5-224, which is a zinc finger nuclease that Sangamo has worked on in the past. Gregory stressed, however, "that was not the clinical lead," which has not been published. The authors of the Nature Methods paper also looked at VF2468, which targets the human VEGF gene.

The Nature Methods paper makes zinc finger nucleases sound a bit like desperate daters, grasping – in their zinc fingers – whatever genetic match looks just about right and then refusing to let go: "Ironically, they are designed to bind too well," senior author David Liu, who is professor of chemistry and chemical biology at Harvard University, told BioWorld Today. And that, in turn, "gives them . . . tolerance for imperfections" that leads to off-target binding.

Using DNA libraries, Liu and his team identified a very large number of sites that could be cleaved by the two zinc finger nucleases his team looked at; a number of those sites were validated in cell culture studies, though the team cautioned that their finding "remains to be tested in other relevant cell types."

In their paper, Liu (a Howard Hughes medical investigator) and his team suggested specific ways in which future zinc finger nucleases could be improved – by making their binding less strong than it could be.

In a nutshell, Liu said, the optimal way to develop a zinc finger nuclease is to first find sequences that the nuclease will bind optimally – and then put imperfections into that sequence, though he stressed that the latter would have to be done "thoughtfully, not randomly."

Liu also noted that his team's findings are not meant to suggest that he thinks zinc finger nucleases should be abandoned. They are imprecise, yes, but to say that they are disappointingly imprecise would be "a superficial conclusion to draw from our paper," he said.

"Overall they are still fairly precise," he added, "and I don't think anyone expected them to be perfect. . . . I still think they are very promising agents for manipulating genomes."

Gregory, in turn, noted drily that though the papers marked the first opportunity to look at genomewide effects of specific zinc finger nucleases, they do not mark the first time it has occurred to anyone to think about their possible off-target effects.

"It only takes one adverse event to really put a crimp in our style," Gregory said. "And so we've been really, really careful" to design therapeutics that avoid such an adverse event.

The work now published in Nature Biotechnology is only one piece of that effort – and, he added "gives us tremendous confidence" in the specificity of the approach.