Posttranslational modifications have been called the next frontier in proteomics, and numerous papers have been written about the best ways to study them. In particular, phosphorylation has been the focus of many researchers’ attention, given its crucial role in cellular signaling processes. But as it turns out, talking about phosphorylation site mapping is easier than actually finding the sites.
Last fall, the Association of Biomolecular Research Facilities’ (ABRF) proteomics research group asked scientists to back up words with deeds, sending out standardized samples containing phosphopeptides to 106 labs around the world for analysis. The result: Only one laboratory identified both phosphopeptides in the sample as well as their phosphorylation sites, and only two others determined both phosphorylation sites correctly.
One of the aims of the study, which was conducted anonymously, was to establish realistic expectations for phosphorylation site analysis without radioactivity. “Any of us who manage a facility have been confronted by a biologist waving a paper at them saying ‘This person can do this, why can’t you?’” said Thomas Neubert, who runs the protein analysis facility at New York University and presented the results of the study at last week’s ABRF annual meeting in Denver.
Besides asking for identifications, the research group asked scientists how they prepared their sample, what separation methods, instruments, and search algorithms they used, and how many years of experience the operator had who performed the analysis.
The sample that ABRF provided contained a tryptic digest of two proteins - bovine protein disulfide isomerase (PDI) at 5 pmols, and BSA at 200 fmols. Mixed in with those were two synthetic peptides at 1 pmol level, which were both derived from PDI, and each of which had one phosphorylated residue.
Whether the task was too tough or researchers too busy, only about half of the labs that had requested a sample actually returned any data. A number of labs submitted several sets of results, resulting in a total of 67 analyses. But only 14 of the labs even attempted to map the phosphorylation sites.
Ken Standing’s group from the University of Manitoba in Canada, the only group whose identity was revealed at the meeting, was the sole laboratory that successfully identified both peptides and their phosphorylation sites. Their recipe for success: offline HPLC separation followed by MALDI-MS/MS, using a prototype of ABI’s QSTAR instrument. “We decided deliberately not to use any fancy enrichment technique,” Standing said in his talk.
Two other labs got both phosphorylation sites right, but correctly identified only one of the phosphopeptides. But this was probably not their fault: the proteomics research group acknowledged that the second peptide might have been cleaved by residual trypsin in a number of samples during shipment or processing. In their analyses, both these labs bet on LC-MS/MS, using a Thermo Finnigan LCQ Deca instrument.
A mere fifth of the analyses correctly identified at least one of the phosphopeptides, and one fifth put down phosphopeptide sequences that were wrong. Labs that felt confident enough to submit their results on phosphorylation sites seemed to have reason to do so: Of those who tried, the majority - 70 percent — correctly assigned at least one of the two sites of phosphorylation.
Notably, immobilized metal affinity chromatography (IMAC), one of the most popular strategies for phosphopeptide enrichment, failed miserably in the study: only one of the thirteen analyses that used it identified one of the phosphopeptides present in the sample.
Posttranslational analysis, it seems, still has a long way to go. “We need to implement robust technologies that can get all phosphorylation sites, not just the low-hanging fruit,” said Neubert.
A poster with the detailed results of the study will be up at the ABRF website within the next few weeks.