NEW YORK (GenomeWeb) – Researchers from the European Molecular Biology Laboratory-European Bioinformatics Institute and the University of Oxford have devised a periodic table of protein complexes.
The table, which the researchers presented in a paper published this month in Science, provides a classification system that covers roughly 92 percent of known protein complex structures while also enabling predictions of new structures likely to be observed in the future.
While not aimed primarily at proteomics research, the table could prove a useful resource for the field in that it provides structural information that can be added to protein-protein interaction experiments, noted Joseph Marsh, an EMBL researcher and author on the paper.
From the perspective of protein-protein interaction work, "the most practical benefit is that with those sort of nonstructural techniques you can find out what [proteins] are interacting with each other, and, now, methods are becoming more prominent where you can start to figure out the stoichiometry of the complex — how many copies of each protein exist within the complex," he said.
"Part of what our table can do is tell you the most likely way that they are arranged within the complex. It's more trying to get at the structural basis of the interactions," he added. "So you could combine it with [for instance] crosslinking mass spec and homology modeling for trying to predict the actual structure of a complex. That's where it could be really useful."
Marsh and his colleagues developed the table by analyzing the assembly and disassembly pathways for a number of protein complexes. For this, they used literature searches to collect data on the assembly and disassembly pathways of previously studied complexes. They also analyzed large collections of protein complex structural data taken from resources including the Protein Data Bank.
The work was very much enabled by "the huge amount of structural data we have now," Marsh said. "There are over 10,000 structures in the Protein Data Bank, so we can start to do these really large scale systematic analyses of existing structures."
Also key, he noted, was mass spec work by the lab of Oxford researcher and paper co-author Carol Robinson and her colleagues, who use mass spec to experimentally characterize the assembly and disassembly pathways of protein complexes.
In these experiments, the researchers subject complexes to increasingly destabilizing treatments in an effort to break them apart step by step. By characterizing the components after each round of destabilization, they are able to piece together their destabilization pathways.
Because the complex disassembly and assembly pathways are typically reversible, this also reveals? the assembly pathway, Marsh said. "The experiments are generally done in a reversible manner where first you disassemble it and then you reassemble it from the free proteins and see, basically, that no off-pathway intermediates are formed."
In the Science paper, the researchers used mass spec to characterize the disassembly and assembly pathways of nine heteromers of varying quaternary structure, finding that in all cases they were able to characterize well-defined intermediate subcomplexes. They added this data to mass spec-based characterizations they had previously done of 16 other complexes — eight homomers and eight heteromers.
Using these collections of structural information, Marsh and his colleagues used principles of protein complex symmetry to determine the range of possible protein complex structures.
"An important part of [the table] is based on the sort of limited types of closed symmetry groups that proteins can adopt when they form complexes," he said. "So it kind of recapitulates what is known about [protein] symmetry, but puts it in a new assembly-centric framework."
He noted that while, in the past, much structural biology work has focused on homomeric complexes involving just one kind of protein, technological advances, particularly in the field of cryo-electron microscropy are enabling better characterization of large multiprotein complexes. The periodic table presented in the Science paper is, likewise, "more applicable to heteromeric complexes where you have [multiple]subunits coming together," Marsh said.
The table covers 92.5 percent of observed homomer complexes and 91.7 percent of heteromers, the authors noted, and, of the remaining 8 percent, roughly half of these exceptions are due to "basically either crystallographic artifacts or mistakes people have made when assigning the biological unit in the Protein Data Bank," Marsh said.
Beyond observed complexes, the table also predicts a number of structures that are likely to exist but haven't been seen yet, he said, adding that he expected that many of these predicted structures would be observed in coming years.