NEW YORK (GenomeWeb) – Researchers from the Karolinska Institute have devised a new workflow for identifying multiple peptides from single MS/MS spectra in mass spec experiments.
The technique, detailed in a paper published this week in Molecular & Cellular Proteomics, significantly increases the efficiency of peptide identification, the authors wrote. Using the technique, they managed to identify an average of 1.24 peptides per MS/MS spectra, allowing them to identify more than 5,000 proteins in a single-dimension, two-hour LC-MS/MS shotgun analysis of human HeLa cell lysate.
In developing the approach, which they have termed DeMix, the Karolinska team sought to turn what has traditionally been considered a challenge of shotgun mass spec – the overlap of multiple peptides in single MS/MS isolation windows – into an advantage.
Traditionally, researchers have tried to limit such overlap as the generation of product ions from different precursors within a single MS/MS window made peptide IDs more difficult, increasing the possibility of false identifications. As the authors noted, many standard workflows essentially ignore the multiple precursors represented by such "chimeric" MS/MS spectra, focusing solely on matching a single peptide to a particular spectrum.
The Karolinska researchers, on the other hand, aimed to deconvolute these chimeric spectra to enable identification of the multiple peptides each represents, thereby allowing for more efficient peptide identification and, ultimately, deeper proteome coverage.
Key to this effort, said Roman Zubarev, a Karolinska researcher and leader of the study, is the improved performance – in particular, the improved mass accuracy – of recent mass spec instrumentation. He and his colleagues performed their analysis on a Thermo Fisher Scientific Q Exactive machine.
However, Zubarev told ProteoMonitor that he believed the limited adoption to date of such an approach has been due to a "mental rather than technical" issue.
"We believe that our approach could have been introduced earlier, with such instruments as [Thermo Fisher's] Orbitrap Velos or a range of high-end Q-TOFs," he said. However, "it was hard – and still is for some – for the data-dependent acquisition crowd to accept that multiple peptides can be reliably identified from a single MS/MS spectrum."
In fact, Zubarev said, he and his team were not the first to present such an approach. Methods such as the Percolator algorithm developed by the lab of University of Washington researcher Michael MacCoss and the MixDB method developed by the University of California, San Diego allow for deconvolution of chimeric MS/MS spectra.
The Karolinska team, however, is the first to have managed in practice to assign more than one peptide per MS/MS spectra at the level of a whole dataset in a DDA mass spec experiment, Zubarev said.
Among the key differences in the Karolinska approach compared to past efforts is the use of "cloned" spectra. That is, when potential co-fragmenting peptides are detected, the MS/MS spectrum is cloned and the original precursor is replaced with the new candidate precursor.
This approach, Zubarev said, streamlines data processing because it does not require adjusting the list of MS/MS fragments.
"Other methods of deconvolution actively change the list of MS/MS fragments – say, [by] removing all assigned peaks after each round of peptide sequence assignment – which interferes with the statistical model used for [false discovery rate] estimation," he said.
In the MCP paper, the researchers found that the extra identifications generally came from peptides that were not targeted via DDA, likely because of low abundance. In fact, the average abundance of the peptides added via the cloning approach was roughly 2.5 times lower than the primary peptides identified. According to Zubarev, the technique allowed the researchers to go deeper into the proteome by half an order of magnitude, on average, compared to traditional methods.
Given these gains, Zubarev said he considered the seeming reluctance within the DDA community to pursue such analyses somewhat confusing – particularly given the dramatic growth in recent years of data-independent acquisition mass spec methods like Swath, which, similarly, require deconvoluting complex spectra containing evidence of multiple peptides.
In DDA mass spec, the instrument performs an initial scan of precursor ions entering the instrument and selects a sampling of those ions for fragmentation and generation of MS/MS spectra. Because instruments can't scan quickly enough to acquire all the precursors entering at a given moment, many ions – particularly low-abundance ions – are never selected for MS/MS fragmentation and so are not detected.
In DIA, on the other hand, the mass spec selects broad m/z windows and fragments all precursors in that window, allowing the machine to collect MS/MS spectra on all ions in a sample.
In the MCP paper, however, the researchers took advantage of the fact that in DDA, due to the overlap of multiple peptides in a single isolation window, the MS/MS spectra of the selected peptides often contain evidence of additional peptides beyond those specifically selected by the instrument for fragmentation.
In a sense, then, DDA analyses like DeMix close the space somewhat between DDA and DIA, as both approaches are trying to match spectral information within a given isolation window to multiple peptides. And, Zubarev suggested, this similarity between the two along with the broad acceptance of DIA, bolsters the case for the adoption of DeMix-style methods.
"In Swath one uses a huge – up to 20 m/z units – window for MS/MS, while we use only 4 m/z units," he said. "So the degree of multiplexing and thus the burden of proof [that the required deconvolution can be achieved] is much lighter for us compared to Swath or similar DIA methods."