Pushing Proteomics

Genomics is slowly but surely moving off center stage, replaced by proteomics. Though proteomics is a young field that hasn't fully found its stride, two new developments provide glimpses of the future. At the end of February, attendees of the Cambridge Healthtech Institute (CHI) Genome Tri-Conference 2002 in Santa Clara, Calif., got their first glimpse of the Protein Atlas of the Human Genome™. Developed by Abingdon, UK-based Confirmant Ltd.—a joint venture of Abingdon, UK-based Oxford GlycoSciences Ltd. and Marconi plc of London—the protein atlas is slated for full release in June. Explaining the purpose of the atlas, Jonathan Sheldon, Confirmant's chief technology officer and head of bioinformatics, compares it to existing protein databases. "If you look at the protein databases that are out there, they are largely hypothetical," he says. "Analysis at the genetic level also doesn't tell you much about alternative splicing or post-translational modification, or where the protein is in the cell." In contrast, Confirmant's atlas is created by sequencing proteins from normal and diseased samples using mass spectrometry, mapping them back to the genome with a computational algorithm, and then identifying the location of the exons on the genomic sequence. So far, the company has experimentally verified 7,000 human genes, but it should have closer to 10,000 by the time of its release, says Sheldon. "Our aim is to identify and experimentally verify all of the proteins [coded by] the human genome," says Sheldon. "We're trying to isolate them from a whole different variety of tissues and bodily fluids, both normal and diseased. We're also trying to identify post-translation modifications." Sheldon admits that the atlas is a work in progress, predicting that it will contain 30,000 genes by June 2003, and that the number will continue to rise. It is an ambitious project. "As with any informatics-oriented businesses, people have become a little desensitized and are always a little skeptical," says Michael Liebman, a former of colleague of Sheldon's, and now director of computational biology at the Abramson Family Cancer Research Institute at the University of Pennsylvania Cancer Center. "I think the first release will be a proof of concept and enable people to dabble with it and get some sense of its applicability.... It will be of value, but obviously not anywhere near the value of the longer-term releases." Nevertheless, the company asserts that the protein-centric and experimentally derived nature of the atlas differentiates it from other databases that are based on computational predictions and thus, in the company's opinion, suspect. Liebman agrees, citing Confirmant's experimental backing as a key advantage to the atlas. The concentration on experimental results means that proteins in the atlas are exactly as they appear in the cell—that is, they are the correct isoform and they bear the post-translational modifications that protein data derived from genetic information would not be able to predict. That information is crucial for screening drug candidates. "It's really important that you screen the most appropriate form of the protein for the disease you're looking at," says Sheldon. Confirmant also sees the protein atlas as useful for gene and protein microarray design. Because space is at a premium on arrays, it is important to apply gene sequences that actually encode proteins to get the maximum information out of an experiment. Again, because it is derived from experimental data rather than from gene prediction, Confirmant believes its data will provide a good source for array producers. Other applications could include identification of novel targets and biomarkers, positional cloning studies, and improved annotation of existing targets. The protein atlas and similar projects may well be poised to carry forward the promise of the genome. "I think [the protein atlas] is a great thing to do," says Michael Snyder, professor and chair of molecular, cellular, and developmental biology at Yale University. "If they only do it on humans, they'll be missing out on [a lot of information], but it's a reasonable place to start." Courtesy of Confirmant Ltd.Neo-Cartography: The strategy Confirmant adopted in creating their Protein Atlas of the Human Genome. For their part, Snyder and his colleagues have generated a bit of excitement over their large-scale analysis of subcellular protein localization in baker's yeast, Saccharomyces cerevisiae.1 Snyder's team used two techniques to tag the proteins—directed cloning of PCR-amplified open reading frames using a yeast tagging/expression vector, and random tagging by transposon mutagenesis—and then visualized protein localization using fluorescently labeled antibodies that recognize the tags. Only about 60% of the yeast proteins could be reliably tracked—for the rest, the team relied on predictions, drawing from expression profiles, protein interaction data, and other sources. The "localizome," as Snyder calls it, revealed that 47% of yeast proteins are cytoplasmic, 13% are mitochondrial, 13% are exocytic, and 27% are nuclear. The work is a significant departure from earlier proteomics efforts, which tended to be little more than lists of the proteins present in a cell or tissue under a given set of conditions, according to Jon Klein, professor of medicine at the University of Louisville School of Medicine. "It's not enough just to make a laundry list of proteins in a cell to understand their behavior. By identifying their localization, we learn a great deal about their function." And in proteomics, after all, that's the whole point. Jim Kling (jkling67@cs.com) is a freelance writer in Washington, DC. 1. A. Kumar et al., "Subcellular localization of the yeast proteome," Genes & Development, 16[6]:707-19, March 15, 2002.

Pushing Proteomics

Interested in reading more?

Become a Member of