Researchers from the Arabidopsis Interactome Mapping Consortium – a collection of more than 20 laboratories worldwide – this week released the first data from the group's four-year effort to map the protein-protein interactions of Arabidopsis thaliana.
The work, which was detailed in a paper published in the current issue of Science, identified roughly 6,200 reliable interactions between about 2,700 proteins, and provides plant scientists with a unique research resource, said Jeffery Dangl, a professor at the University of North Carolina and AIMC member.
"Like most interactome maps, it's a giant hypothesis generator," he told ProteoMonitor. "So anybody out there working on, for instance, root development as a way to increase biomass or flowering time to increase seed production – anyone working on any plant can immediately access this data and if you find interactions in Arabidopsis, you can infer that they're real in rice or tomatoes or zucchini or whatever you want."
To identify the interactions, the scientists, led by Harvard researcher Marc Vidal, used a yeast two-hybrid-based system, testing pairwise combinations of proteins encoded by roughly 8,000 open reading frames representing around 30 percent of the plant's predicted protein-coding genes.
They evaluated the quality of their hits against a positive reference set of 118 well-documented Arabidopsis protein-protein pairs and a random reference set of 146 random protein pairs.
These controls ensure the dataset contains a low level of false positives, something that has not always been the case with past interactome studies, said Geert De Jaeger, a researcher in VIB-Ghent University's department of plant systems biology, and who is not affiliated with the study.
"The first interactomes [researchers did] in yeast and C. elegans were done with much less controls," he told ProteoMonitor. "Marc Vidal, whose technology is behind this, has seriously optimized his [yeast two-hybrid] system, so he's really built a lot of control systems into his procedure, and he's really only selecting interactions that he can confirm in several experimental repeats."
This makes for a resource that is relatively small – covering only an estimated two percent of the plant's protein-protein interactions – but very robust, de Jaeger said. He noted that, while in the paper the researchers say the cited interactions can be trusted with roughly 80 percent confidence, "in reality I think it is even a bit higher."
De Jaeger added that roughly two-thirds of the proteins used in the project were ones about which very little was known, making the map a useful resource for initiating their study.
"For these proteins we have now a huge amount of interactions that can give you a starting place [for investigating] what these proteins of unknown function are doing," he said. "So this is, I think, really the main asset of this first version of the Arabidopsis interactome. We have now a huge amount of genes we know a bit more about."
He added that given the problem of false negatives in interactome research, combining the data with that obtained via other techniques like tandem-affinity purification could further flesh out the map.
"The major problem with interactomics is its false negative rates," he said. "So there are a huge number of false negatives that you don't pick up using one technique. So you only get a very small snapshot, but still it's very valuable."
In an accompanying paper in Science, researchers including Dangl demonstrated the potential usefulness of the interactome study, using Vidal's workflow to identify Arabidopsis proteins key in the plant's pathogen response.
The purpose of the study was to test a hypothesis about plant immune systems that Dangl and Sainsbury Laboratory researcher Jonathan Jones proposed 10 years ago: that all pathogen effectors should interact on a largely overlapping set of host defense proteins.
To that end, the scientists mapped interactions between Arabidopsis proteins and virulence effector proteins from the bacterium Pseudomonas syringae and the oomycete Hyaloperonospora arabidopsidis, finding that, as predicted, the effector proteins from both pathogens converged on a set of 165 plant proteins, despite the fact that the two organisms use very different approaches to infect plants and haven't shared a common ancestor in more than 2 billion years.
Given that the 8,000 Arabidopsis proteins they used represent roughly a third of the plant's proteome, one might expect that the 165 defense proteins might also represent around a third of the total proteins involved in directing its response to pathogens, Dangl said, noting that this finding "shrinks the haystack down from the 30,000 genes in most crop plants" to a number closer to 500 for scientists investigating plant defense machinery.
"If you're working on some disease in corn, corn has 32,000 proteins, so you might immediately take those 165 proteins [identified in the study] and look at them," he said. "We know that those genes are conserved, and so we would suggest that if the pathogen evolution is telling us what the important defense machinery is, then those 165 [pathogen] effector targets we described are certainly likely candidates to be required for proper defense response."
The study, De Jaeger said, is a good example of the sort of specific biological questions to which interactome research – including his own, which recently used interactomics to investigate plant cell division (PM 08/20/2010) – is being increasingly applied.
Traditionally, "the interactome community has worked a bit independent of biological questions," he said. "So you just had these huge sets of data that were published, and of course it's extremely valuable, but in a sense it's just a database. And the insights that people extracted from this data were more like evolutionary stuff, extremely fundamental knowledge with not that many applications linked to it."
More recently people have been using interactomics "to try to get a glimpse of biological problems," De Jaeger said. "And here you see a very nice example."
From a practical standpoint, the findings could be particularly relevant to efforts to engineer hardier crop plants – an important area of research given that worldwide between 30 and 40 percent of crop loss is due to pathogens.
"By focusing on yield, crop lines have been selected that yield a lot but have very poor immune systems," he said. "One of the things that genetic engineering is trying to do is build up the native immune system of the plant. This knowledge will be very important to developing new strategies" for doing that.
Dangl said that the researchers are now performing the experiment with two additional pathogens to see if the initial findings hold. He added that work to expand the number of proteins in the interactome map is ongoing, as well.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.