Skip to main content
Premium Trial:

Request an Annual Quote

Researchers Develop Improved Method for Modeling Protein Complexes via Crosslinking MS


NEW YORK (GenomeWeb) – A team led by researchers at the Pasteur Institute and European Molecular Biology Laboratory's Structural and Computational Biology Unit have devised a new approach to modeling structural proteomics data that takes into account the conformational heterogeneity of protein complexes.

In a paper published this week in Nature Methods, the authors introduced an automated modeling approach capable of integrating conflicting crosslinking mass spec data to reflect the various conformations a protein complex may take, allowing researchers to better capture the actual complexity and variability of protein complex structures, said EMBL researcher Martin Beck, an author on the study.

Crosslinking mass spec has emerged in recent years as a useful tool for generating structural information, particularly in the case of protein complexes for which the large amounts of sample required for conventional x-ray crystallography is not available.

The technique uses molecules to crosslink proteins at different points, with mass spec analyses then used to generate structural information based on the position of these linkages.

Such experiments, however, can run into the issue that protein complexes often exist in a number of different conformations, which can result in seemingly conflicting data.

"The crosslinking reaction happens in solution, and in solution almost all molecules exist in overlapping states of different conformations," Beck told GenomeWeb. "And if you have complexes with mobile domains, then you can easily get conflicting data in the sense that you have a domain that in conformation A and conformation B is in two different positions, and so you get crosslinks from both [conformations]."

Typically in crosslinking mass spec experiments, "you end up with one conformation that models the data the best, and that is where you stop," Beck said.

With the new approach, on the other hand, "you model a conformational space as an ensemble of conformations, and you ask which ones are more likely, which ones are more populated, and what is less likely, and that hasn't been achieved in crosslinking mass spec before," he said.

Pasteur Institute researcher Michael Nilges, senior author on the paper, previously tackled a similar problem with using NMR data for elucidating the structures of small proteins, and, Beck said, here Nilges and his co-authors adopted that approach to crosslinking mass spec data.

In the Nature Methods paper, the researchers used the approach to model the 17-subunit yeast RNA polymerase III, finding that their structure matched well with recent cryoelectron microscopy structures and, in fact, identified characteristics of the structure not identified by cryo-EM including localization of the C-terminal tail of the C31 subunit, which is thought to be key to initiation of transcription.

Their model also indicated the possibility of conformational states not yet observed for the complex and which could be important in preinitiation or transcription termination.

Beck and colleagues in his lab focus largely on modeling the nuclear pore complex and use crosslinking mass spec extensively. They haven't yet applied the new modeling method to study of that complex, however, he said, noting that it is considerably more complicated than the RNA polymerase III complex.

"It's a completely new method, so we sort of used an almost ideal sample to show that it works," he said. "We haven't yet used multistate modeling [for the nuclear pore] because the data we have in terms of [that complex] are not as comprehensive as what we have [on RNA polymerase III], but in the future, of course, we would love to do this, and we are working on generating better datasets."

Crosslinking mass spec is typically used for the study of individual complexes in isolation, but, Beck noted, lately researchers have begun exploring use of the technique for studying endogenous protein complexes in vivo. For instance, he said, Utrecht University researcher Albert Heck last year published a paper in Nature Methods on such a workflow that combined crosslinking mass spec via sequential collision-induced dissociation and electron-transfer dissociation acquisition and a search engine, XlinkX, specifically designed for crosslinking MS work. In that paper, Heck and his colleagues detected 2,179 unique crosslinks in HeLA cell lysates, allowing them to observe among other things, dynamic interactions of ribosomes in contact with different elongation factors.

Crosslinking mass spec "is on its way to becoming a normal proteomic method with which you can probe cells," Beck said.

Indeed, as Swiss Federal Institute of Technology Zurich researcher Ruedi Aebersold, in whose lab Beck was formerly a post-doc, told GenomeWeb in a recent interview, while crosslinking mass spec have proved a useful approach for structural characterization of protein complexes in isolation, he believed its greatest impact would ultimately come as a tool for better understanding the behavior of complexes in vivo.

"What we are particularly interested in doing is going into the cell and doing affinity purification [of a protein], and then asking what is the composition and structure of the complex around a particular protein under different cell conditions," Aebersold said. "These complexes are highly dynamic in their composition and are affected by mutations and the binding of factors and so on, so we see it as a means to characterize complexes in more detail than is currently possible."

The method remains a challenging one, though, both for looking at complexes in vitro and in vivo, Beck noted.

"The complication with crosslinking MS is the complexity challenge," he said, citing a variety of issues such experiments face.

For instance, he noted, certain amino acids are well exposed to each other and so crosslink easily, while others are more difficult to crosslink. This can result in a situation where you have a much lower amount of some crosslinks than others, making those peptides difficult to detect.

Additionally, you could have an amino acid that can be easily crosslinked to a number of other residues, which further increases the complexity of analysis.

"The crosslinked proteome is much more complex than the normal proteome, and so that just makes the mass spectrometry very challenging," Beck said. "So, you need a huge amount of spectra, a huge dynamic range to reasonably sample [the crosslinks], and then you also end up with a computational challenge."