In a competition organized by the Association of Biomolecular Research Facilities, tandem mass spectrometry coupled with online liquid chromatography emerged as the most successful technique for correctly identifying low-abundance proteins, according to David Arnott, a mass spectrometrist in the protein chemistry department at Genentech in South San Francisco, Calif.
The results of the study, which Arnott presented at the ABRF meeting earlier this month in Austin, Texas, showed that compared to groups using MALDI-TOF or batch nanospray LC/MS, those employing online LC/MS/MS analysis accounted for all seven of the entries that correctly identified all five of the proteins present in the sample, out of a total of 41 entrants.
The ABRF proteomics research group refereed a similar competition in 1999, but in the most recent study the group lowered the concentrations of proteins in the sample, in an attempt to test the sensitivity and resolving power of the more advanced mass spectrometry technology available today. “We were interested in seeing how well laboratories would do at identifying all the components of that mixture, and we wanted to see what techniques they used, and if any of those techniques were more successful than others,” Arnott said.
To do this, the ABRF group solicited volunteers to participate in the study, and sent them a sample containing five predigested proteins; two at a concentration of about two picomolar, and three others at a concentration of about 200 femtomolar. The group sent samples to more than 100 protein mass spectrometry labs, and received responses from 41, some of which submitted results using more than one mass spectrometry technique. “The good news is that almost all the labs — about 89 percent — correctly identified the two major components of the sample,” said Arnott.
Of the seven labs that correctly identified the five proteins, five employed electrospray Q-TOF mass spectrometers from Micromass or ABI/MDS Sciex, and two used Thermo Finnigan LCQ ion trap mass spectrometers, Arnott said. Of the eight groups that identified four of the five proteins correctly, four used LCQ ion traps, three used Q-TOFs, and one a MALDI-TOF mass spectrometer.
Arnott said the choice of mass spectrometry software broke down roughly into two camps: those with Micromass Q-TOFs tended to use MASCOT software for searching and matching mass spectra with peptide sequence data, and those with LCQ ion traps relied on SEQUEST. “It didn’t really seem to matter which software package they used,” Arnott said, “if they were doing LC/MS/MS they tended to get great results.”
The reasons for the incorrect identifications of some of the proteins were unclear, Arnott added. The group did not ask the respondents about their level of expertise in applying mass spectrometry to protein identification, he said, “so some of the people who got wrong answers may have been new to this.” In addition, Arnott said incorrect results could have been due to poor data, or a group applying its search algorithm incorrectly. Nor did the study rely on a completely random sample, he said, because not every lab that requested a sample returned its results.
Arnott noted that the groups differed in what criteria they used to decide whether a search result was correct. In at least one case, he said, two groups with similar tandem mass spectrometry data on just one or two peptides came to different conclusions as to the validity of their protein identification. While one group claimed to have identified the protein based on the data, “another group listed the same protein as a tentative identification, on what looks like pretty similar data,” he said.
“The whole issue of what the criteria are for believing your search result is a big issue and maybe one that needs to be addressed in the future,” he added. “There’s still some subjective judgments going on here.”
A full report on the study will be available at www.abrf.org within a week, Arnott said.