Zinc finger proteins recognize specific DNA sequences and, when fused to DNA-cutting enzymes, precisely cut the genomes of higher organisms. Despite the importance for genome engineering, predicting DNA-binding specificities has remained elusive, making it difficult for researchers to engineer these small protein domains to specifically bind any site of interest.
To address this issue, Mona Singh and Marcus Noyes of Princeton University and their colleagues developed an experimental-computational approach to enrich and recover distinct groups of zinc fingers that bind common targets, as reported in Nucleic Acids Research (1).
The researchers first built several large, diverse zinc finger libraries with up to six variable amino acid positions by optimizing a PCR-based cassette mutagenesis method. They then derived a simple analytical formula to calculate the expected diversity of a library and used high-throughput sequencing to show that their libraries offered levels of diversity that approached the theoretical maximum.
“I had always assumed that the limitations in the zinc finger engineering field were due to the incomplete libraries employed,” Noyes said. “I spent a year and a half figuring out how to build large libraries using PCR without introducing bias. Once I established the technique, I was not surprised that the results were more diverse and comprehensive than previous approaches.”
The researchers used one of these libraries to select and recover thousands of zinc fingers that bind several three base pair targets of interest. To do so, they used the bacterial one-hybrid system—a method for identifying the sequence-specific target site of a DNA-binding domain—as well as Illumina sequencing and analysis to uncover the amino acid sequence binding a desired DNA site. Using computational approaches, they clustered the recovered zinc fingers to reveal several distinct classes of proteins.
“The data can be leveraged to build predictive models of zinc finger binding specificity and to assemble zinc finger proteins with desired binding specificities,” Singh said. “In the future, these techniques can be used to select new zinc finger domains to bind all possible targets.”
According to the authors, the high-throughput approach has the potential to greatly influence genome and protein engineering. “Scaling this approach up may provide a deep understanding of the zinc finger and potentially other protein domains, allowing for the prediction of naturally occurring proteins and the design of novel ones,” Noyes said.
(1) Persikov AV, Rowland EF, Oakes BL, Singh M, Noyes MB. 2014. Deep sequencing of large library selections allows computational discovery of diverse sets of zinc fingers that bind common targets. Nucleic Acids Res 42(3):1497-508. doi: 10.1093/nar/gkt1034. Epub 2013 Nov 7.