This article was not written by Coon Laboratories and was originally posted on GenomeWeb here: https://www.genomeweb.com/proteomics-protein-research/multi-omic-yeast-mass-spec-study-provides-insights-mitochondrial-protein
In October 2013, University of Wisconsin-Madison researcher Josh Coon presented a mass spec-based method for profiling the yeast proteome in one hour.
The approach, which used Thermo Fisher Scientific’s Orbitrap Fusion instrument, represented a fourfold increase in speed compared to the field’s previous best efforts, potentially enabling proteomics researchers to significantly up their study sizes.
This week, roughly three years later, Coon and his UW-Madison colleagues published in Nature Biotechnology one of the first large-scale studies taking advantage of the technique — a combined proteomic, lipidomic, and metabolomic analysis of 174 yeast strains, each having knocked out a single gene linked to mitochondrial biology.
Having demonstrated the one-hour yeast method, Coon was looking for an appropriate application. Meanwhile, fellow UW-Madison researcher David Pagliarini, co-author of the Nature Biotechnology paper, was using yeast knockout strains to investigate the function of unannotated mitochondrial proteins.
In that work, Pagliarini and his team had noticed that knocking out genes in the same pathway resulted in similar proteome-wide changes.
“That was surprising to us because we thought it would only be very direct [protein] changes that would be in common,” Pagliarini told GenomeWeb. “But it was actually an entire [proteome-wide] signature. So we thought, what if that is actually a way to identify functions for proteins whose functions aren’t known?”
Essentially, the researchers hoped that by generating knockout yeast for large numbers of genes involved in mitochondrial function, they could use comparisons of the resulting proteome-wide changes to shed light on the unannotated genes. For instance, if a strain lacking an unannotated gene produced proteome-wide changes similar to a stain lacking a known gene, it suggested the two were involved in a particular process or pathway.
“There are hundreds of proteins in mitochondria [with] function[s] that aren’t well annotated, and it is hard to figure out what a protein does when its function isn’t known,” Pagliarini said. “And so this gives us a way to do that.”
The one-hour yeast proteome approach gave Pagliarini and his team a way to perform these experiments on a very large scale.
“Josh was saying, ‘Hey, we can make all these measurements. What can we do with it?'” he said. “So it sort of was just perfect timing.”
The UW-Madison researchers did more than 3,000 mass spec experiments, profiling the proteomes, metabolomes, and lipidomes of 174 single-gene deletion yeast strains in biological triplicate under two different conditions, fermentation and respiration.
Of the 174 genes they looked at, 124 had been characterized, while the remaining 50 were uncharacterized but known to have mitochondrial functions.
The analysis produced a large amount of data and a variety of opportunities for follow-up, Pagliarini said, noting that one angle he and his team had investigated was fleshing out the biosynthesis of coenzyme Q(CoQ).
Discovered at UW-Madison roughly 50 years ago, CoQ “is an essential molecule for mitochondrial ATP production and something that many mitochondrial disease patients are deficient in,” Pagliarini said. CoQ’s biosynthesis is poorly understood, however.
“This approach allowed us to implicate a number of new proteins in that pathway, including one that we proved to be a missing enzyme that produces the CoQ precursor,” he said. “So we were able to help complete that biosynthetic pathway by observing these [proteomic] similarities to other known enzymes in the pathway.”
He added that there are now a number of opportunities for additional follow-up. “There are orphan proteins in mitochondria that have never really been studied by anyone for which we now have predicted functions,” he said. “There are new sorts of systems-level analyses that we think we can do with the data. There is so much to do with it. We have analyzed it at a decent depth, but there is a lot more to be done.”
In total, Coon said, the project generated around 4 million individual molecular measurements, and integrating this amount of data in a usable format was no small task.
“We had a student who basically spent the last year-and-a-half just figuring out how to synthesize all of that,” he said. “This student designed an interactive website that basically allows one to quickly look at any given knockout strain and compare it to any other strain and look at any molecule, protein, metabolite, or lipid. And that tool was used during the development of the project for [Pagliarini’s] team to sift through all of that information and sort of winnow down to the leads described in the paper.”
The website is open to outside researchers interested in analyzing the UW-Madison team’s data, and the raw mass spec data from the experiment is available at the Chorus mass spec repository, Coon said.
Pagliarini said that he and Coon have begun similar investigations in human cell lines using CRISPR to generate knockout lines similar to the yeast experiment. The goal, Pagliarini said, is to look at human mitochondrial proteins that are not conserved in yeast. Of the 174 genes they looked at in the yeast experiment, 144 have human homologs.
Analyzing the human proteome is considerably more complicated than the yeast proteome, Coon noted. However, he said that he found the quality of the data in the yeast experiment very encouraging.
“From a technical perspective, I think it was a little bit of a different approach,” Coon said. “Lots of folks would advocate that if you wanted to go at a scale like this, you would have to use some targeted approach and limit yourself to some subset of proteins, or you would need to do some sort of labeling. We are interested in all those methods, but this very simple approach was very fast, very deep, and gave a really good quality of data.”
“If you look at the precision of protein measurements, the [coefficients of variation] were about 12 percent globally,” he said. “And I think that is better than we could have ever hoped for, no matter how we had done the measurements.”
Key to this level of precision were rigorous sample collection and preparation protocols, as well as highly stable liquid chromatography and high-end mass spectrometry instrumentation, he said.
Regarding the human cell line measurements, Coon said his lab was able to detect and quantify more than 8,000 proteins in under four hours with a level of precision comparable to that of the yeast experiments.