Plasma was the star of the show in the “Human Proteome,” the second conference included in last week’s Peptalk Protein Information Week held in San Diego. While the usual debates over the use of various separation and depletion techniques for studying plasma dominated much of the discussion at the Cambridge Healthtech Institute-sponsored meeting (see PM 1-16-04), a couple of less familiar trends in plasma proteomics also emerged: Most notably there was an emphasis on narrowing down the large field of proteins in plasma by pre-selecting or targeting the proteins or peptides that are expected to be of interest. Some presenters referred to this method as “targeted proteomics.”
Leigh Anderson, CEO of the Plasma Proteome Institute, kicked off the conference by announcing that “[t]his is the time in history when actual discovery of plasma proteins by proteomics is possible.” Unlike during earlier years, when 2D gel limitations reduced the possibilities, multidimensional separations enable this type of discovery, he said. Also, while literature searches produced a list of plasma proteins that was heavily weighted toward proteins with signal sequences and extracellular proteins, newer LC-MS methods have helped better correlate the distribution of protein types found in the plasma with the distribution found in the whole proteome, Anderson said. The goal of plasma proteomics now, he said, is to make the detected plasma proteome “more representative of the entire human proteome and less of the plasma [proteome],” so that more productive biomarker and drug discovery work could be done.
But Anderson was less optimistic about translating this discovery work to the clinic. “The dark side is, the gap between research and diagnostics is widening,” he said. He cited as a major problem the discrepancy between the needs of clinical diagnostics versus the abilities of discovery proteomics: For example, the former involves large sample size, short time period, and low CV; the latter uses a small sample size, long time period, and higher CV. Anderson proposed filling the gap between the two with targeted proteomics. His version of this, called stable isotope standards with capture by anti-peptide antibody, or SISCAPA, involves using antibody columns to enrich for pre-selected peptides of interest after digestion of the proteins. Particular peptides are pre-selected based on discovery proteomics and genomics experiments. Anderson said his group was able to enrich for peptides of interest by greater than 100 fold, and to extend sensitivity by at least two orders of magnitude. The resulting smaller, more concentrated pool of proteins could then be more easily translated into diagnostics, according to Anderson.
Numerous conference attendees told ProteoMonitor that they rated Anderson’s talk as one of the “most interesting” of the conference.
Meanwhile, several other presenters promoted their own versions of targeted proteomics — though they called it by a variety of names. Markus Kalkum, assistant professor of immunology at the Beckman Research Institute of the City of Hope, presented a method called hypothesis-driven multi-stage mass spectrometry. This method involves predicting the m/z location of an expected peptide ion on an MS spectrum, and then specifically probing for signatures in the corresponding MS/MS spectrum for that m/z region. If the expected signatures are found in the MS/MS spectrum, that serves as an indication that the “parent ion” is present. In this way, even MS spectra that mostly consist of noise and have no distinguishable peaks — such as those produced by samples at concentrations of less than 1 femtomole — can produce information. Kalkum said this method had applications ranging from yeast mating studies to protein complex elucidation, and that it also could be applied to complex plasma samples to help in picking out low-abundance proteins that the researcher predicts will be present but can’t necessarily see in an initial MS spectrum.
Christie Hunter, applications scientist at Applied Biosystems, discussed another version of targeted proteomics using the precursor ion scan mode on ABI’s Q-TRAP. By looking in the spectrum peaks for a loss of molecular weight that corresponds to the mass of a particular post-translational modification, researchers can scan for expected phosphorylation or other PTM sites on a protein of interest. Hunter also discussed using the multiple reaction monitoring mode to look for protein biomarkers or expected synthetic peptides in serum.
David Hammond, executive director of the plasma derivatives department at the American Red Cross, described another sort of targeted proteomics that he said was “non-biased,” in that there is no pre-selection for what proteins to look for before the researcher begins looking. Instead, Hammond selects for proteins from un-fractionated serum by capturing the proteins on affinity beads containing ligands from a combinatorial library, and then subjecting them to various drugs or antigens in a type of blot format. Preliminary leads are selected based on binding and activity to the drugs or antigens, and are then characterized using MudPIT, LC/LC-MS/MS techniques, or direct sequencing. One application of this technique, which Hammond called FIoNA — for functional identification of novel activities — would be to analyze the immunoproteome, which several conference presenters said had under-appreciated scientific value. Hammond said that IgGs were ideally suited for separation by peptide libraries using the technology “because that’s their job — to find peptides.”
Keith Rose, chief scientific officer of GeneProt, related targeting to sample pooling — another technique that previously may not have been on the radar for many proteomics scientists but that was mentioned several times at Peptalk. Rose suggested that pooling large numbers of serum samples and other biofluids was useful in some cases to obtain larger starting amounts for experimentation or diagnostics, to dilute individual variation among patient samples, and to gather relatively large amounts of less common proteins in the plasma or serum, such as insulin. In this last case, the concentrations of low abundance proteins are still low, and that’s where the need for targeting comes in. “One can see femtomole amounts [of low abundant proteins] when one knows what one is looking for,” Rose said.
Several other presenters mentioned pooling approaches as well. Leigh Anderson discussed the pooling of plasma for mass screening of viruses in the general population. Anderson spoke in lieu of his father, Norman Anderson, senior scientific advisor at the Viral Defense Foundation, who was unable to attend the conference at the last minute. By pooling plasma that is left over from diagnostic labs from various populations, enriching for viruses by large scale centrifugation, and then either shotgun sequencing the viral genomes or performing MALDI-MS on the viral proteins, Anderson said one could screen for the prevalence of various known and unknown viruses in the population. “We can look at large populations to see what is going around,” Anderson said.