Unlike the genome, the repertoire of proteins expressed in a given cell depends on a host of factors, including whether the cell is sick or healthy, its location, and age.
Proteomic pioneer Large Scale Biology Corp. (LSB) was one of the first companies to attempt to map the proteome. The company describes its Human Protein Index (HPI) as "the protein equivalent of the Human Genome Project." The first version of HPI was announced in early 2001; the database aims to catalogue every protein expressed in the human body.
In June 2001, LSB rival Oxford Glycosciences (OGS) teamed up with telecommunications company Marconi to launch Confirmant Ltd., aimed at developing a protein atlas of human proteins with their variants. The database already contains information on about 7,000 genes and all of their encoded proteins.
Celera briefly launched a proteome project, but subsequently backed out. "There ain't no such thing as a proteome," then-president J. Craig Venter told The Wall Street Journal in April 2001.
Since the discovery by Myriad Genetics of the breast cancer gene BRCA1 in 1994, a maze of proteins that interact with the gene product has been identified.
Scientists want more than mere expression data. Proteins act in concert, so researchers need to go beyond the study of single proteins and look at protein-protein interactions to fully understand cellular biology. With this in mind, last year Myriad Genetics Inc., in collaboration with Hitachi Ltd. and Oracle Corp., launched a project aimed at producing "a proprietary database of all human protein interactions, all biochemical pathways, and a comprehensive catalog of purified proteins by 2004."
Early this year, two public-private consortiums published the results of large-scale yeast protein-protein interaction studies in Nature that highlighted a new generation of tools for such studies. In one study, the University of Toronto and MDS Proteomics Inc. generated data on more than 1,500 distinct interacting proteins. Those data will go into the publicly accessible Biomolecular Interaction Network Database. In the other study, Germany's Cellzome collaborated with the European Molecular Biology Laboratory in Heidelberg to identify 589 proteins and 232 distinct complexes by studying 1,739 genes.
But although they may dazzle with their breadth and complexity, such projects are difficult to justify in commercial terms. "This was a proof-of-concept study to establish that we have a robust technology platform," says Anne-Claude Gavin, director of biochemistry at Cellzome. The German company continues to map the yeast proteome, but the focus is increasingly on finding and studying human proteins that match the most interesting yeast proteins its work has uncovered. About 90 percent of the complexes mapped by Cellzome contain proteins of unknown function, and some of these could be important new drug targets.
"These maps will generally have impact, especially when you look at known drug targets that are not currently addressable," says Hanno Langen, director of proteomics at Roche Group in Basel, Switzerland. "And this technology is much more trustworthy than the yeast two-hybrid [assay, the standard approach for detecting protein interactions]."
But in general, groups like Langen's have their priorities clearly set. "The large-scale data are interesting, but nonetheless you are drowning in data and you have to evaluate it," he says. "I see more interest in focused applications, where you have a protein and you want to know everything about it."
So proteomics is increasingly moving from a stand-alone effort to becoming part of a palette of genomic technology offerings. CuraGen Corp. was a pioneer of this concept, and its pipeline may be a prime test ground of its validity (see May Bio·IT World, page 46).
Meanwhile, just as they rose to popularity together, OGS and LSB are simultaneously hitting hard times.
LSB recently announced a restructuring and downsizing. Proteomics pioneers — and father and son — Norman and Leigh Anderson have left the company, although they will apparently remain involved as consultants.
OGS is reeling from a June FDA letter saying its lead drug — Vevesca (OGT 918) — cannot be approved without additional trials. Although Vevesca, which targets Gaucher disease, is not one of the company's proteomics-derived products, its approval and the accompanying revenue would have made a big difference in the company's outlook.
With the business climate as tenuous as it is, and tens, possibly hundreds, of thousands of proteins still to be mapped, it may be up to public efforts like the Human Proteome Project and the world of academia to sustain and generate the large-scale projects for now.
Back to The Proteomics Odyssey