NEW YORK, April 27 - Myriad Genetics made a splash earlier this month when it announced a collaboration with Hitachi and Oracle to "map the human proteome" by 2004.
The scope of this ambitious task could vary widely, depending on how the companies want to define "map" and how thorough they want to be. But regardless of how Myriad and its collaborators define the outer limits of the project, proteomics experts are questioning how much scientific value there is in Myriad's basic approach.
These scientists say Myriad's two pronged attack on the proteome, an automated yeast two-hybrid strategy combined with MALDI-TOF mass spectroscopy analysis of protein complexes, won' t provide Myriad and its partners with a thorough understanding of how proteins play a role in disease.
"It's a major limitation, focusing entirely on interaction measurements," said Ruedi Aebersold, a proteomics expert at the Institute of Systems Biology in Seattle. "My point is that they should not disregard other areas [of protein research], such as levels of protein abundance and post-translational modifications."
Myriad hopes to use the yeast-two hybrid system as the first thrust of its attack, by expressing proteins in yeast cells using libraries of cDNA assembled from all the open reading frames in the human genome, and gauging whether the individual proteins interact with each other. If two proteins form a complex, a reporter gene is expressed, stimulating the growth of the yeast colony, which indicates that the the two proteins do interact.
Myriad plans to build up a map of protein interactions by linking these individual interactions into one gigantic web, starting with a library of proteins expressed in the brain, said Jay Boniface, the director of protein science and proteomics at Myriad.
With another thrust, the company plans to investigate multi-protein complexes by designing specific affinity tags that pull out large clusters of proteins from cell lysates, or dissolved cells, and then identify the amino acids that make up the proteins using MALDI-TOF spectroscopy.
The two approaches are complementary, because they cover both binary and multi-protein interactions, said Sudhir Sahasrabudhe, executive vice president of research at Myriad. More importantly, Sahasrabudhe said, they are fast, given Myriad's experience in automating the yeast two-hybrid technique.
But other scientists say that what is truly of value--at least to pharmaceutical companies--is an understanding of which proteins play important roles in certain diseases, and that Myriad's method may not necessarily lead to the discovery of new targets.
"Myriad's approach is entirely predicated on the assumption that if I understand how a protein interacts with other proteins, I understand what it does [in the body],'" said Joshua LaBaer, the director of the Institute of Proteomics at Harvard Medical School. "There's some truth to that. Figuring out who a protein interacts with is valuable, but it's not a done deal."
Predictably, protein scientists who specialize in studying protein expression levels, and how they change in diseased-versus-healthy cells or tissues, found fault with Myriad's overwhelming emphasis on protein interactions.
"Their technical approach is focused on the structural interactions between pairs of proteins, rather than changes in protein amounts that are the basis of disease and drug effects," said Leigh Anderson, CEO of Large Scale Proteomics, a Gaithersburg, Md.-based subsidiary of Large Scale Biology that is studying protein expression levels. "Hence the Myriad deal only covers one niche in the human proteome, and one that we would argue is clearly of secondary interest in comparison to expression."
But Myriad scientists responded that they already have an idea of which genes or proteins are associated with particular diseases, and that mapping out the interactions of these proteins will allow them to identify proteins more promising as drug targets.
Furthermore, said Paul Bartel, president of proteomics at Myriad, "you need to work with all the proteins because all the information provides context for the particular area you're working on."
But Bartel agreed that interaction measurements are just one piece of a larger puzzle. Biologists will have to apply their knowledge from other systems to make the interaction data useful, he said. "None of this works in a vacuum."At this point, Bartel added, the valuable element is speed, and how quickly Myriad can gather data to sell to its customers and use internally. "The technology is available to do it very rapidly, and because it's so important for drug discovery it makes sense to do it in a rapid genome-wide fashion."