This story originally ran on March 25.
By Tony Fong
SACRAMENTO, Calif. — With phosphorylation becoming a main area of interest among proteomics researchers, especially among those doing work with pharmaceutical implications, the Association of Biomolecular Resource Facilities’ Proteomics Standards Research Group this year conducted its study on evaluating the community’s ability to perform phosphoproteomics experiments.
The study’s results were presented here at ABRF’s annual conference and were among several made by the association’s research groups. Along with sPRG, the Proteomics Research Group, the Protein Sequencing Research Group, and the Glyco Proteomics Research Group presented their study results, which will be available online in the coming weeks.
Below are summaries of the results.
For its study this year, sPRG, whose mission is to develop and encourage proteomics standards, decided to “provide each participating lab an opportunity to evaluate its capabilities and approaches with regard to detecting phosphopeptides and identifying sites of phosphorylation,” the group said in a poster presenting its results.
More specifically, the goal of the study was to develop a “readily available” phosphopeptide standard that would be useful for researchers to learn how to detect phosphopeptides and identify phosphorylation sites, James Farmar assistant director of the mass spectrometry core at University of Virginia Health System and chair of sPRG, said during his presentation.
In addition, the study set out to provide an opportunity to evaluate new and current techniques of detection and identification, and identify best practices and techniques for detection and identification.
The decision to focus on phosphorylation was based on its growing importance, particularly within the pharmaceutical industry, which is putting more of its R&D emphasis into studying how phosphorylation, especially tyrosine phosphorylation, affects protein kinases, and subsequently how that could result in new therapeutics.
According to sPRG, proteomics labs are often asked to determine phosphorylation sites, and while strategies for identifying phosphopeptides abound, for many labs the task represents a “formidable challenge.”
For its study, the group developed a sample composed of 23 tryptic peptides that were previously used in earlier sPRG studies: two from its 2003 study, which also examined phosphorylation site mapping, and 21 from last year’s proposed study, which was supposed to develop a standard for use in assessing a lab’s capabilities to perform absolute quantitative analysis but that was never done.
The phosphopeptides were present in the sample at 5 moles. Of the 23 phosphopeptides, 14 had single phosphorylation sites, five had double sites, three had three sites, and one had four sites.
A total of 43 datasets were submitted. According to sPRG, the results were mixed, though Farmar said “the quiet take-home message [is that] the ability of the community to find these peptides has improved.”
Nonetheless, in its poster, sPRG said that none of the reporting labs were able to identify all 23 phosphopeptides that were in the sample, suggesting that the methods of phosphopeptide enrichment have different capabilities depending on the number of phosphosites in a peptide.
The group concluded that despite improvements in sample preparation technologies, identifying multiple phosphopeptides remains a challenge.
“It is important to note that there was an insufficient number of responses using any of the individual technologies to afford a statistically significant measure of the ability of any method to ‘get the correct answer,’” sPRG said.
The Proteomics Research Group chose to focus its study on the ability of labs to address unanticipated problems in proteomics experiments.
According to the group, the study sought to document approaches that are useful in identifying proteins in simple mixtures; identifying low-abundance proteins in the mixture; and addressing unforeseen problems such as 15N-labeled proteins, altered expression constructs, and unanticipated contaminants “modeling as close as possible to a realistic sample,” according to the study’s poster.
[ pagebreak ]
In addition, David Friedman, PRG chairman and associate director of the proteomics lab in the Mass Spectrometry Research Center of the Vanderbilt-Ingram Cancer Center, said during the group's presentation at the ABRF conference that PRG sought this year to increase the number of researchers who participate in the study, a figure that has declined in recent years, especially among ABRF members.
This year, he said, PRG tried to remedy the fall-off by broadening the scope of the study. They also designed the study to make it more in line with real-life experiments, and they made it easier by extending the deadline and simplifying the submission process.
The result was that the number of requests for the 2010 study sample increased by more than 50 percent to 96, though only 47 labs returned any results, and only 26 of those included data, Friedman said.
Of the study itself, PRG said analysis of protein complexes isolated by co-immunoprecipitation or affinity enrichment is a common experiment in proteomics labs, but such analysis is complicated by issues such as sample reproducibility and the presence of non-binding partners.
For the PRG study, participants were asked to identify and characterize components of a protein complex, including 15N-labeled proteins and proteins with differently processed N-termini.
Three different samples were created, each with different challenges. In one sample of an active protein complex for ubiquitination of beta-catenin, the challenge was to identify all six proteins and the two different forms of beta-catenin in the sample.
In another sample of an inactive protein complex containing two 15N-labeled proteins, including a bacterial contaminant, YodA, participants were asked to identify the 15N-labeled proteins.
And in the third sample, researchers were tasked with identifying the 15N-labeled protein that has “restored biological activity,” Siah1, in an active complex with two 15N-labeled proteins.
Among the results: Nine participants were able to detect the appearance of beta-catenin in two different forms; nine also identified the 15N-labeled contaminant YodA; four identified the low-abundant component S100-6A and seven identified another S100 protein.
PRG said that a number of participants reported finding the study moderately difficult, which Friedman highlighted during his presentation. By comparison, during the past few studies some responses raised concerns that the study designs from PRG were interesting but too hard.
Formerly called the Edman Sequencing Research Group, the name was changed last year to the Protein Sequencing Research Group to reflect the decline in Edman sequencing and the use of new mass spec-based technologies for protein sequencing.
Last year’s study by the group looked at how Edman degradation compared to mass spec-based methods for N-terminal sequencing. This year, the group sought to continue the 2009 study by making the complex more difficult, Wendy Sandoval, a PSRG co-chair and scientific manager at Genentech, said at the group’s presentation.
This year’s study compared Edman degradation with mass-spec based methods for sequencing a monoclonal antibody.
The main goals were to determine what techniques are complementary and parallel to Edman sequencing, and to evaluate whether any techniques exist that could replace Edman degradation for N-terminal sequencing, Sandoval said.
“The variable regions for the heavy and light chains, which include the N-termini, are not recorded in the protein databases,” the group said in a poster describing the study. The challenge for participants, it said, was that the sample contained two N-termini, including one that was N-terminally blocked. They were asked to obtain terminal sequence.
[ pagebreak ]
The techniques used by participants included Edman degradation, enzymatic digestion, in-source decay, terminal labeling, and techniques that incorporated multiple methods.
The three most successful methods, PSRG determined, were Edman sequencing, bottom-up mass spectrometry using enzymatic digestion, and top-down mass spectrometry utilizing in-source decay.
For Edman sequencing, PSRG found that participants could directly determine the protein’s N-terminal sequence and identify all commonly occurring amino acids and some stable post-translational modifications.
Proteins with blocked N-termini could not be sequenced, though, unless the protein was deblocked, while the length of the sequencing read was affected by the efficiency of the instrument and size of the molecule, PSRG said.
Top-down mass spec-based sequencing methods found that “matrix-generated hydrogen radical mediated fragmentation of the intact protein in the ion source via laser,” according to the group’s poster.
In addition, PSRG said that “the entire ion series representing the termini may not be present and extrapolations of the [in-source decay] spectra were not sufficient to obtain the terminal sequence.” In-source decay also needed T3 sequencing to obtain the terminal sequence, the group added, though blocked termini are “not an issue” for in-source-decay techniques.
Bottom-up mass spec-based methods resulted in smaller fragments in random order that did not usually cover the entire protein sequence and may not include the terminal fragments. Such techniques also require multiple enzymes and “rely heavily on database homology and bioinformatics to assemble and fill in sequence gaps,” PSRG added.
While Edman degradation is losing cachet as a method for protein sequencing, members of PSRG were not ready to give up on the technology.
“Edman is not going away,” said John Smith, a PSRG co-chair and manager of the protein chemistry lab at the University of Texas Medical Branch. “I think it’s something that’s staying around that’s complementary to mass spec methods.”
The Glyco Proteomics Research Group was formed a year ago. For its first study it chose to assess the different approaches that have been developed for quantitating N-glycosylation in three similar glycoprotein samples.
In his presentation, gPRG Chair Ron Orlando, assistant professor of chemistry at the University of Georgia, said the objectives of the study were to identify the major N-linked glycans in the three samples; and to quantify the relative differences in the distribution of the N-linked glycans in the samples.
However, in creating their mixtures, the members of gPRG encountered a problem: It is impossible to produce a single glycoprotein with known changes in N-linked glycosylation. To overcome this, they came up with a strategy using a mixture of different glyoproteins with unique glycans. By changing the glycoprotein ratio, they changed the distribution of N-linked glycans.
The glycoproteins selected for this were ovalbumin; asialofetuin; alpha1-acid glycoprotein; NGNA; and NANA. The three samples sent to participants had different glycoprotein mixtures at different glycoprotein ratios.
Thirty-five samples were requested and 19 labs submitted data. According to Orlando, most labs readily detected glycans from each sample. There were, however, some issues as gPRG delved deeper into the results. While 11 labs were able to detect the NGNA glycoprotein, eight could not. And among the 11 that detected NGNA, seven said it was a very abundant glycan, Orlando said, prompting him to say “clearly something is wrong” when the results show such discrepancy.
Also, 18 of the labs identified fucosylation glycans correctly, despite those glycans being present at low levels. But 14 of them incorrectly said that this was core fucosylation.
There were similar problems in the participants' ability to detect differences in the glycan levels. For example, none of the labs could identify more than two of the four changes in glycan composition, Orlando reported. One lab identified two glycan changes, seven labs identified one glycan change, and 11 labs identified no glycan changes, “which I thought was pretty bad,” Orlando said.
“There is room for lots of improvement,” he added during his presentation. “We have not even gotten to the hard problems like quantitating individual glycans in isomeric mixtures.”