NEW YORK (GenomeWeb) – Researchers at the Salk Institute for Biological Studies have developed a protein interaction mapping method that allows for cheaper, more streamlined large-scale experiments.
The technique, which was described in a paper published this week in Nature Methods, uses the recombinase Cre along with next-generation sequencing to detect and identify protein-protein interactions in yeast two-hybrid experiments.
In a set of 10 screens done in Arabidopsis thaliana, the researchers were able to test 36 million different interactions and identify 8,577 interactions between 1,452 transcription factors, representing what they said was a three-fold jump in the number of known plant transcription factor interactions.
The study is the latest in a series of efforts refining the Cre-based process, said Shelly Trigg, first author on the paper and a graduate student in the lab of Salk researcher Joseph Ecker, senior author on the study.
The basic idea behind the method is the use of Cre to link the cDNA molecules coding for interacting proteins. These molecules can then be read out using next-generation sequencing, identifying interacting pairs.
One of the first such methods was published in a 2007 Nucleic Acids Research paper by researchers at the Roswell Park Cancer Institute. However, that effort, essentially a proof-of-principle study, was confined to a relatively small scale.
Last year, a team led by University of Toronto researchers published a paper in Molecular Systems Biology that built upon this notion of Cre-enabled recombination by incorporating specific DNA barcodes into each plasmid, which could be read out via sequencing to identify interactors.
The addition of barcodes, the MSB authors noted, helped get around the challenges inherent to PCR amplification of "templates varying widely in length and base composition."
However, Trigg said, incorporating these barcodes can be time consuming and expensive, particularly in the case of large-scale, multiplexed screens like she and her colleagues were undertaking. To streamline the process, the Salk researchers left out the barcoding step and directly sequenced the protein coding sequences of the interacting pairs instead.
In a cost analysis comparing conventional yeast two-hybrid screens to the barcoding method and the Salk team's approach, the researchers found that a conventional screen testing 30,000 bait proteins against 30,000 prey proteins would cost around $12,000 and take 10 days. The barcoding method would cost $18,000 and take eight days, while the Salk team's approach cost $860 and took six days.
Using their method, the researchers performed 10 screens of 1,956 Arabidopsis transcription factors and regulators in "all-against-all" fashion, identifying 8,577 interactions, 7,994 of them previously unreported. This, they noted, roughly tripled the 3,170 such interactions contained in the BioGRID protein-protein interaction database.
The researchers determined that at 10 screens they had detected around 50 percent of the interactions they could potentially detect, suggesting, Trigg said, that including additional screens could further expand the network.
"We just went with the 10 to try out the method and see what the data looked like," she said. "But once you generate your expression libraries for the screen, it's very easy to mix your libraries together, grow your cells, prepare them, and sequence them. So anyone could easily do multiple replicates, which gives you more chances to see [more interactions] than if you did, say, just two [screens]."
Trigg noted that the approach is also applicable to other model systems including mammalian ones.
"We are a plant biology lab and we had access to this great collection [of Arabidopsis transcription factors], but you could put whatever kind of collection you wanted to in this system," she said.
The study authors also suggested that the technique could incorporate "en massecloning strategies," that could enable "cDNA library-against-cDNA library screening," which they said would allow for "comparisons of unprecedentedly large-scale interactomes derived from different ecotypes, growth conditions, or tissue types."