Using an affinity purification technique combined with mass spectrometry and bioinformatics, researchers at Cellzome and the European Molecular Biology Laboratory have performed the first genome-wide screen for protein complexes in yeast, revealing 491 complexes that function in a modular manner.
The research gives a deeper understanding of how protein complexes work within disease pathways. Ultimately, it may help scientists to better select drug targets for diseases, said Gitte Neubauer, the vice president of research operations at Cellzome, the Heidelberg, Germany-based company that EMBL spun out five years ago.
"It is very advantageous if you know an entire disease pathway," said Neubauer. "By knowing all the players in the pathway, and which position they are in upstream or downstream, you can make the best choice for a drug target."
Cellzome currently has a collaboration with the Novartis Institutes of Biomedical Research to discover disease therapies in a broad sense. In addition, the company has a partnership with Johnson & Johnson to discover therapies for Alzheimer's disease.
Neubauer said Cellzome and EMBL did the work on yeast to understand "the principles of how proteins come together to form functional units." However, Cellzome moved away from yeast work in favor of human work a number of years ago, Neubauer said, and the company's main interest now is to exploit the technology first developed in yeast to map core disease pathways in humans.
"It is very advantageous if you know an entire disease pathway. By knowing all the players in the pathway, and which position they are in upstream or downstream, you can make the best choice for a drug target."
"The reason we are publishing this now is that we had moved onto human cells a long time ago, and now we can safely let the public know about our [yeast] work without really giving them a competitive advantage," said Neubauer.
Cellzome launched the large-scale project to find all the protein complexes in Saccharomyces cerevisiae about five years ago. In January 2002, Cellzome and EMBL researchers published a paper in Nature describing 232 yeast protein complexes, representing about 25 percent of the entire yeast genome (see ProteoMonitor 1/14/2002).
The current work, published in this week's online edition of Nature, is a completion of the 2002 research. Of the 491 protein complexes found, 257, or 52 percent, had never been previously described, according to the study's authors.
In addition to revealing more yeast protein complexes, the work reveals that protein complexes act in a modular fashion, with different complexes performing primary functions, as well as "plug-in" peripheral functions.
"Your car has a function to drive from A to B, but from your car you can plug in a lot of devices — a ski rack, your baby seat, a radio," explained Anne-Claude Gavin, a team leader at EMBL and the first author of the current study. "The same is true with proteins. You have a core group of proteins that have a function, and then you can attach on that core group plugs."
Knowing which proteins a particular protein interacts with goes a long way towards understanding that protein's function, Gavin said.
"It's like sociology. If you want to understand me, you need to understand to whom I'm married, where I work, and who are my kids," said Gavin. "If you want to understand the function of a protein, you need to understand with whom the protein interacts."
Aside from identifying which proteins interact, the current study used bioinformatics techniques to quantitatively characterize how likely two molecules are to bind to one another. The affinity measurements correlate with actual dissociation constants described in the literature, said Patrick Aloy, a staff scientist in the structural and bioinformatics group at EMBL.
"Most proteomics studies in the past have shown whether molecules interact or not in a 'yes/no' way," said Aloy. "The completeness of this data lets us see how likely any particular molecule is to bind to another. … Using these affinity measures, you can move from static models to simulating how a system behaves."
The next step after the current study is to look at how proteome complexes change when a cell is perturbed, said Gavin.
"What we've mapped so far is snapshots of the proteome," she said. "The next step is to map several different proteomes. The idea is to try to understand the dynamic nature of the complex — what gets recruited when you change the cell growth condition?"
In addition, the researchers said they'd like to look at the structural aspects of the protein complexes to try to understand why certain proteins recognize each other and form complexes, Gavin said.
In terms of human protein interactions, there is a lot of data that is unpublished, said Neubauer. However, Cellzome and EMBL did publish their work on the human Tumor Necrosis Factor signaling pathway in February 2004 in Nature Cell Biology.
"Before, you knew that if you stimulated cells with TNF alpha, they would release cytokines. You knew at a cellular level that if you do something to the cells, they would give a response," said Neubauer. "Now you can have a closer look at the molecular level. You can use your [pathway] compound library and screen against the cellular assay. You can take hits out of a screen and show that it has an inhibitory effect [on the cellular response.]"
The TNF work as well as the work on yeast supports Cellzome's philosophy that protein complexes represent cellular function, said Neubauer.
Neubauer said she could not disclose what potential drugs have resulted from Cellzome's collaboration with Novartis, which began in September 2004.
In addition to its collaborations with Novartis and Johnson & Johnson, Cellzome also concluded last year a deal with Bayer to use its chemical proteomics technology in identifying lead compounds.
— Tien-Shun Lee ([email protected])