The Proteomics Research Group of the Association of Biomolecular Resource Facilities has published a paper in the online edition of Proteomics presenting the results of its 2008, 2009, and 2010 research studies.
Designed to help proteomics researchers in academia, the clinic, and industry assess their capabilities as well as learn about various methods and technologies available to them, the PRG studies presented participants with challenges including the qualitative characterization of proteins, the relative quantitation of proteins in serum, and qualitative analysis of proteins in a contaminated sample.
One overarching conclusion from the studies is that "proteomics works," said David Friedman, co-associate director of Vanderbilt University's Mass Spectrometry Research Center and one of the paper's authors. "People can do these studies."
Beyond that, he told ProteoMonitor, the studies demonstrate the need for researchers to use multiple analytical strategies when examining proteomic samples, as well as the role of emerging tools and technologies in making complex methods accessible to a broader range of labs.
In particular, Friedman said, the 2008 study, which had participants identify the major proteins in two samples and report any qualitative differences between them, highlighted the dangers of relying solely on peptide-based approaches. While both samples contained several domains of the sRAGE protein, the second sample also included two truncated variants of that protein, which researchers would be unlikely to detect using just bottom-up LC-MS/MS.
"We didn't give any indication as to [the truncated variants] in the samples, and people that generally took just a standard shotgun-based LC-MS/MS approach could clearly see that both proteins were in there but would have missed that one of them was there in different bits as opposed to intact," Friedman said. "You have to do some type of complementary technique – either run it on an SDS gel and see that there are multiple bands, or do an intact MALDI or something to tell you that there's more than just one protein in there."
One group, he noted, started with a shotgun approach, which led them to report no difference between the two samples. They followed that, however, with an intact mass experiment that revealed that there were multiple forms of the sRAGE protein in the second sample. Reinterrogating the shotgun data looking specifically for truncation endpoints, they were able to identify all of the unique truncated peptides.
Of the 57 responding participants in the 2008 study, 21 provided at least some information on the truncated proteins, with eight finding all the truncation termini in the sample. While using multiple methods of analysis improved identifications, there was little indication that any particular method was more useful than another, Friedman said, noting, "it was more the overall approach to the experiment that mattered."
The 2010 study similarly tested labs' abilities to identify unanticipated sample components. In this case, the E. coli protein YodA was intentionally introduced into a sample in addition to the target protein Siah1. Of the 47 participants, nine were able to determine the presence of this contaminant.
"These studies aren't designed to identify the best protocol for a specific task," Friedman said. "They're designed more to educate, and open [researchers'] minds and eyes. These are the types of samples that come into core facilities all the time. So one takeaway message is that if you take in someone's sample and you only do a little bit [of analysis], you're only going to get answers for what you did. There may be some things that you overlooked."
The studies also serve to expose researchers to emerging methods, giving them a chance to try their hand at new techniques in a controlled environment. The 2010, study, for instance, also required participants to search MS/MS data from 15N-labeled peptides – a demand not all algorithms are able to accommodate.
"Normally, if you didn't [work with 15N-labeled samples] and you got a 15N-labeled sample in your lab, you wouldn't know what to do with it," Friedman says. "So there's the educational value of doing the study in a nice controlled environment. It's almost like buying a standard from Sigma and saying, 'Oh, I see. I could have done it that way. That's something to keep in mind.'"
According to the Proteomics paper, at the time of the 2010 study, five of the 47 participants analyzed 15N-labeled samples, with another 11 saying that they planned to offer the service in the near future.
The 2009 edition dealt with quantitation of known target proteins, which occasioned focus on another proteomic technique that has grown in use in recent years — selected-reaction and multiple-reaction monitoring mass spectrometry (PM 02/12/2009).
The study required researchers to report the relative quantities of four specified proteins in three different mixtures. The three mixtures were presented as duplicates – a total of six samples – which the researchers were tasked with matching.
While investigators successfully used techniques like LC-MS/MS data-dependent acquisition and LC-MS/MS targeting peptide precursors only, MRM/SRM mass spec was the most popular option, with eight of the twenty-seven participants using it.
In the paper, the authors suggested that if MRM software tools like Skyline — developed in the lab of University of Washington researcher and paper co-author Mike MacCoss — had been more widely used at the time, the study likely "would have demonstrated even more effectiveness for the SRM technique."
"A lot of people probably would have benefitted from having more sophisticated software tools for predicting MRM transitions," Friedman said. "If we had repeated this study the next year, the results probably would have been different, and more people would have had an easier time doing it."
"It's not a competition," he said. "It's more a set of well-controlled tools to let people assess their own ability relative to other people and to help them assess their own standards and their own good laboratory practices."
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.