Work in Progress

Mass spectroscopy (MS) alone or combined with liquid chromatography is being used as an analytic tool in proteomics. The disadvantages of isotopic labeling methods include expense, alterations in proteins that may occur due to the chemistry involved, and limits to the number of proteins and variables that can be examined at a time. Label-free MS methods are being developed to overcome these disadvantages. In theory, label-free MS should allow measurement of unlimited numbers of proteins in a sample in a wide range of states. However, the vast amounts of data that result require software for analysis that may not be available for desired applications.

Critical Mass

Michael MacCoss, Assistant Professor, Department of Genome Science, University of Washington, Seattle, WA, has spent the last few years developing label-free MS methods. He was skeptical at first, but has been pleased with the results. He believes that adding labels late in an experiment increases the noise, and proteomic data are already noisy. “Label-free methods are currently better at measuring large changes, where the noise doesn't matter as much,” he observes. Stable isotope labeling allows measurement of more subtle changes in protein levels. The addition of stable isotopes as early as possible may help. “Our philosophy,” he says, “is if you can't add a label early, you might as well do without it.” However, he feels that both label-based and label-free methods are complementary. His group had put substantial effort in developing not only methodology, but also analytic software, for both.

“We've had to write a lot of software to handle the data. I think that's been underappreciated,” he says. Half of his laboratory is devoted to software development to handle the huge amount of data the other half generates. Companies that manufacture MS equipment do not necessarily know what scientists will be using it for. MacCoss says that although users are determining the utility, the software lags behind the hardware.

“A lot of people rely on freely available software,” he notes. “Some decide on hardware based on the software that's available, which may lead them to choose equipment that's less than ideal [for their needs].” His group wanted the highest quality hardware and is figuring out how best to analyze the data. “A critical mass is needed to do proteomics. You need good cell biologists and protein biochemists, people good at instrumentation, and experts in data analysis.”

MacCoss' group is working with the model organisms C. elegans and S. cerevisiae to eliminate some of the method's noise. With a strain of defined genotype, differences due to mutations or environmental stress are more detectable. Most quantitative experiments involve changing a few conditions. “The technology is not robust enough,” he acknowledges, “to detect differences among human clinical samples given the biologic noise, genetic diversity, and environmental factors that affect the proteome.”

“There is a huge push to detect human biomarkers, but it's not straight-forward to know what you need to measure,” MacCoss says. He believes that although biomarkers may be identified using label-free MS, clinical tests are likely to involve ELISA or protein chip-based methods. “The main focus in my lab is developing technology to improve throughput and analysis by increasing the speed, accuracy, precision, and dynamic range,” he says.

No Single Platform

Michael Washburn, Director of Proteomics Center, Stowers Institute for Medical Research, Kansas City, MO, says that although using label-based MS on human clinical or even animal samples is very expensive and challenging, he is still convincing people of its potential. “Now we are asking what's the best approach instead of asking if it's possible at all,” he observes. “It will take some time for the majority of people to be convinced that this is an approach they can use. A healthy skepticism is always good in science.”

Washburn agrees with MacCoss on the complementarity of label-free and label-based methods. “Proteomics will not have a single technological platform,” he predicts. His group is currently using the multidimensional protein identification technology MudPIT for proteomic analysis. MudPIT is a chromatographic method in which a complex peptide mixture from a sample is loaded on a biphasic microcapillary HPLC column that is placed directly in line with a tandem mass spectrometer. They are applying label-free analyses, based on normalized spectral counting, to allow for variation in replicate MudPIT analyses. Washburn thinks that MS is probably more quantitative than people believe. He is working with collaborators focusing on transcriptional complexes, particularly involving methyl transferases and histone deacetylases. One caveat for working with large protein complexes is that the there will be a “different picture, depending on what protein you use to purify it,” he notes. “We are doing basic biologic analyses at this time, not being translational yet. Will we cure cancer anytime soon doing this? Probably not.” He concludes that this methodology will allow his group to look at multiprotein complexes in a more informed fashion. When histone deacetylases are inhibited, the mechanism of action of some cancer chemotherapeutics on the market or in development and the impact on the 3-D structure of the protein complexes involved are not known. The effect of some drugs may be as much on the stoichiometry and assembly of protein complexes as it is on direct enzyme inhibition. Washburn says “we are focused on convincing ourselves this approach is generally applicable.”