NEW YORK — A team led by researchers at ETH Zurich has developed a mass spec-based approach for detecting protein structural and functional changes at proteome scale.
Detailed in a paper published last month in Cell, the approach uses limited proteolysis-mass spectrometry (LiP-MS) to look at structure and function across large numbers of proteins and with high spatial resolution, offering another dimension of protein data beyond the expression-based measurements that have traditionally dominated proteomics research.
"Measuring changes in protein levels has been more straightforward than measuring changes in protein structures on a global scale," said Paola Picotti, professor of systems biology at ETH Zurich and senior author on the study. However, she noted, protein structural changes, which are key to cellular processes, aren't necessarily reflected in the expression changes.
"A change in protein level and in protein structure could be, but are not necessarily, concomitant events," Picotti said. "A protein could change just in abundance or just in structure or both, upon a perturbation."
In fact, "entire pathways can be regulated by molecular events that do not require changes in protein levels," she added, citing phenomena like protein post-translational modifications, protein-metabolite interactions, and protein-protein interactions. "Classical expression profiling will not detect these types of changes, although these pathways are profoundly regulated."
"We see this as a new and far more comprehensive way to do functional proteomics analyses," she said.
Developed by Picotti and her colleagues several years ago, LiP-MS combines digestion with a broadly specific protease followed by a standard mass spec-based proteomics workflow to assess structural changes on a proteome-wide scale. The basic notion underlying the approach is that in the initial digestion step, the protease will cleave proteins only at sites that are accessible, left exposed by whatever structural conformation it happens to be in at the time of analysis. When a sample is treated with an agent like a drug or otherwise altered, the proteins that bind to this molecule will undergo a structural change, and this will be reflected in changes where the protease is able to cleave the protein.
By following this initial digestion step with standard trypsin digestion and mass spec analysis, researchers can compare the peptides generated in treated and untreated samples and, based on changes in the peptides produced, determine which proteins had their structures altered by the treatment in question.
Picotti has in the past used the approach to look at a variety of phenomena including protein-metabolite interactions in Escherichia coli and binding between proteins and nutrients in yeast. Her lab has licensed the technology to Swiss proteomics firm Biognosys, which offers the method for commercial use, primarily for drug development work.
In the recent Cell paper, the researchers used the technique not only to look across the proteome at what protein structural changes occurred during different cellular events but at the functional implications of those changes.
Because LiP-MS allows for parallel measurement of protein expression and structural changes, the researchers were able to assess the contributions of both, "thus capturing a broad range of functional alterations," Picotti said.
The study looked at the response in yeast to heat and osmotic stress, and how protein structures varied in E. coli grown in eight different carbon sources. They found that in these systems protein structural changes were far more abundant than expression changes. For instance, in the yeast experiment, around 1 percent of detected proteins showed abundance changes, whereas 23 percent and 11 percent showed structural changes in response to heat shock and osmotic shock, respectively. The E. coli work showed a similar pattern, with substantially more proteins exhibiting structural changes than expression changes due to changes in growth media.
Oliver Rinner, CEO and founder of Biognosys, noted that in the yeast work this was due in part to the fact that the measurements were done on a short timescale and that protein structural changes occur more quickly than do protein expression changes. Picotti noted, though, that the E. Coli work looked at changes on a longer timescale where protein expression changes would have enough time to make themselves apparent.
Rinner said that the paper, which he and Biognosys were not involved with, "opens up a new dimension" for proteomics by allowing for large-scale measurement of protein functional changes.
"Our focus [with LiP-MS at Biognosys] has been primarily in drug target deconvolution, and there you are trying to find very few targets, but the right targets," Rinner said. "This paper turns that around and tries to see the structural changes globally across the proteome."
As Picotti and her coauthors noted in the Cell paper, researchers are exploring several other tools including protein crosslinking mass spec and surface footprinting to similarly look at structural changes at proteome scale. She said that thus far LiP-MS appeared to offer substantially higher proteome coverage than the other techniques. She added, though, that unlike crosslinking mass spec, LiP-MS offered no information on interacting proteins and that while crosslinking and surface footprinting have been used in vivo, LiP-MS has only been applied to cell lysates as of yet.
In addition to identifying proteins that have undergone structural changes, the technique is able to identify what sites within the protein have been altered with a resolution of a few amino acids, which Picotti said allows researchers to use existing knowledge of protein structure and function to determine how the structural changes are impacting biological processes or to produce mutated forms of the proteins to explore hypotheses about the impact of these structural changes.
Rinner suggested that the LiP-MS approach combined with developments in modeling protein structure such as those recently achieved by DeepMind's AlphaFold protein prediction tool could in the future advance the ability of proteomics researchers to incorporate structural data into their analyses.
"We have so many protein structures already available and now with this publication by DeepMind there is the possibility that we will have virtually all the structures soon, at least for soluble proteins," he said. "So I see that the availability of [protein] structures will increase, and we have a dynamic tool to look at changes in structure, and all of this together can be the basis of a new way to look at functional biology."
Nearer term, Rinner suggested the approach could be useful for confirming and further fleshing out the impact of structural changes measured by other methods.
For instance, a researcher interested in signaling pathway activity might use both traditional phosphoproteomics and LiP-MS, which could pick up protein structural changes caused by protein phosphorylation, to look at patterns of phosphorylation within a particular pathway and changes in those patterns due to different perturbations.
Picotti said that she and her colleagues are also interested in using the approach for the sort of disease biomarker work that is commonly done using expression-based proteomics. She said that they had recently demonstrated that the approach could scale to biological samples from a human cohort and that they are applying the method to the study of Parkinson's disease.
In terms of further optimizing the approach, Picotti cited as a goal fleshing out the informatics tools available for structure-based analyses. She noted that her team's studies thus far "have relied on in-house scripts to automatically map global structural data to available protein structures and to identify altered functional sites on each protein" which they then analyzed using "established tools for protein network analysis."
She said that developing "robust and user-friendly pipelines to link mass spec and structural data will be crucial to support the broad application of the approach," as will be improving the resolution of their network analyses to handle data from single functional sites on a protein.
She added that she hopes to improve the sensitivity of the technique, as well, which would allow for better detection of structural changes in low-abundance proteins.