Precision tools for DNA editing
February 23, 2011 § 4 Comments
Do you long for an easier way to manipulate genomes? Better methods may be on their way. In some organisms, such as yeast, it’s relatively easy to introduce or remove specific genes. In others, it’s tedious and difficult. Recently, new methods of “genome editing” have been emerging that promise to broaden our ability to manipulate specific genes. These methods rely on the ability to create a double-strand break at a specific location within a genome; once you’ve managed to introduce a break, you can either let it be repaired and look for mutants in which the repair was performed incorrectly, leading to disruption of the targeted gene. Alternatively, you can introduce DNA that includes an area of homology to the broken area and look for recombination events, allowing gene repair, gene addition, or specific gene tagging. But how do you specifically introduce a double-stranded break into the genome just where you want it?
In the last decade or so, it’s become possible to create engineered nucleases that use zinc fingers to target specific genomic sequences. An individual zinc finger binds specifically to a three-base-pair DNA sequence, and by stringing several different fingers together you can make a protein that will recognize a much longer sequence. Once you’ve assembled 8 or 9 zinc fingers, the recognition sequence is long enough that there’s a pretty good chance there will be only one such sequence in the genome. This approach has been used in quite a large number of model organisms that don’t otherwise have excellent genetic tools, and it seems to work quite well; there are even clinical trials of various applications of the nucleases, for example using them to knock out one of the HIV co-receptors in T cells. But, the tricky bit is getting the zinc fingers to target the sequence you’re interested in. There’s a fairly large set of zinc fingers that target specific 3-base-pair sequences, but it’s not a perfect list.
Pam Silver pointed out a recent paper that takes a new approach, using a DNA-binding motif from the TALE proteins (transcriptional-activator-like effector proteins) of Xanthomonas bacteria (Miller et al. A TALE nuclease architecture for efficient genome editing. Nat. Gen. 29 143-148). The tale of the TALEs [sorry, I was trying not to do that] is interesting. The Xanthomonas bacteria are plant pathogens, and they use the TALEs to activate specific genes in their hosts to maximize the susceptibility of the host to infection. Unlike the zinc fingers, the DNA-binding domains of these proteins appear to bind specifically to just one base pair — so in principle you only need 4 different kinds to be able to target anything you want in the genome. (Mind you, you’ll need to string a lot of domains together). The specificity of the ~34 amino acid binding domain is determined by just two amino acids, making it very easy to change which base pair an individual domain binds to. One can imagine that a plant being targeted by a TALE-expressing pathogen would be under strong selection for mutations that change the TALE target sequence; perhaps the reason that this particular DNA-binding structure evolved was to make it easy for the pathogen to respond by changing the TALE to match the new sequence. [My mind is slightly boggled by the structural implications of a series of 34-amino acid domains that each recognizes a single base pair — especially since TALEs apparently work as dimers. I’ll be fascinated to see the structure once it comes along.] In any case, the modularity and editability of the TALE DNA-binding domain is a boon to would-be gene engineers.
Events have been moving rapidly in this field: the TALE code was deciphered just over a year ago by two groups simultaneously, and the first brief report of specific TALE-based nucleases came about 6 months ago; this paper now shows that such nucleases can be used to modify endogenous human genes.
Miller et al. first identified a natural 13-repeat TALE protein that binds highly selectively to its target, which they then used as a framework for creating new TALEs. They replaced the original 13 DNA-binding-domain repeats with 18 repeats predicted to target an 18-bp section of the promoter for an endogenous human gene, NTF3 (neurotrophin-3). To check that the new protein indeed bound to the desired target, they hooked it up to a transcriptional activator, VP16, and showed that the chimeric protein could activate NTF3 transcription. Then they replaced the VP16 domain with the Fok1 nuclease, which has been used extensively for zinc-finger-targeted editing, and showed they could get sequence-specific cleavage. There are many factors to optimize here: the length of the linker between DNA-binding domain and effector domain matters, and the optimal linker for cleavage is not the same as the optimal linker for transcription. Fok1 also needs to be dimerized to work, so to get cleavage you need two TALE nucleases, each binding to a different site on the DNA, one to the left of the desired cleavage site and one to the right. The spacing between these two binding sites also matters for efficient cleavage.
Once you have an appropriately targeted pair of nucleases, biology will do much of the rest of the work for you. A double-stranded break in DNA sets off DNA repair mechanisms such as non-homologous end joining; in some fraction of the treated cells, small errors in repairing the break will lead to small deletions or insertions that can destroy gene function. When Miller et al. tested the NTF3-targeting TALE in K562 cells, they found that up to 9% of the treated cells carried mutations. To show that the approach could be generalized, they ran through the whole program again with a set of TALEs targeted to the human CCR5 gene, a co-receptor for HIV. By testing various distances between the Fok1 targeting sites, they found that they could get over 20% modification of the CCR5 gene. And we’re not limited to non-homologous end joining; homologous recombination can be triggered too, giving gene replacement efficiencies of up to 16%.
Is TALE targeting a substitute for zinc finger targeting? It could be; there’s a lot more work to do before we really know. The modification efficiencies reported in this paper for TALE nucleases are comparable to the efficiencies for many zinc finger nucleases, though they’re not as good as the very best zinc finger nucleases. On the other hand, it’s early days; we’re just beginning to explore the potential for the TALE nucleases, and the simplicity of the recognition code is quite enticing. Miller et al. do flag some potential complications: the TALE domains are bigger than the equivalent zinc finger domains, perhaps making them harder to deliver, for example. But clearly TALE nucleases can work as precision gene editing tools, and clearly they’re worth further exploration.
Miller JC, Tan S, Qiao G, Barlow KA, Wang J, Xia DF, Meng X, Paschon DE, Leung E, Hinkley SJ, Dulay GP, Hua KL, Ankoudinova I, Cost GJ, Urnov FD, Zhang HS, Holmes MC, Zhang L, Gregory PD, & Rebar EJ (2011). A TALE nuclease architecture for efficient genome editing. Nature biotechnology, 29 (2), 143-8 PMID: 21179091