NEW YORK, Jan. 10—Proteomics researchers at Cellzome and MDS Proteomics have made substantial advances in decoding the yeast proteome, using accelerated techniques that rapidly identify protein function and characterize multidimensional protein complexes. The findings come in two papers published in the Jan. 10 issue of Nature, released today.
Both teams used techniques that pair genetically engineered tags with sensitive automated mass spectrometry to probe multi-protein complexes in cells of Saccharomyces cerevisae. Both teams also say that they are already applying these methods to human cells, and suggest that this research shows that the ultimate goals of human proteomics—understanding protein functions and interactions in human cells, and applying this knowledge to drug development—may be attainable.
The teams’ techniques are similar: researchers use genetically modified DNA to tag “bait” yeast proteins. The tags are used as “retrieval hooks” to recapture the protein once it has formed a functional complex, and its protein partners are then identified through mass spectrometry.
In one report, researchers from Toronto’s MDS Proteomics paired with scientists at the University of Toronto and Toronto’s Samuel Lunenfeld Research Institute at Mount Sinai Hospital. The authors say their technique, which they call high-throughput mass spectrometric protein complex identification, allowed them to capture and identify three times as many protein complexes as the yeast two-hybrid system.
Converting about 10 percent of the yeast genome into bait proteins, the MDS Proteomics team detected 3,617 protein interactions. The team identified more than 25 percent of the yeast proteome, including more than 500 new proteins, in a matter of weeks.
They specifically focused on proteins involved in the DNA damage response, a process implicated in the initiation and progression of cancer, and were able to characterize much of the protein network related to that response.
MDS Proteomics president and CEO Frank Gleeson said in a statement that the company “is capable of determining a focused map of a functional and regulatory proteome in a human cell in less than a year.” The company also said their researchers and affiliates will be able to identify 1,000 new drug targets during the next five years.
The Cellzome paper, featured on the cover of the journal, uses a slightly different method: introducing tagged gene sequences directly into yeast cells through homologous recombination. That approach, according to Cellzome bioinformatics and mass spectrometry vice president Gitte Neubauer, is slower but more accurate than the overexpression method used by MDS Proteomics. Because the proteins are expressed directly in yeast cells under near-normal physiological conditions, the resulting interaction data is more specific, she said.
Cellzome used techniques developed at the European Molecular Biology Laboratory to characterize the function and interactions of 1,440 yeast proteins in 232 multiprotein complexes. The team was able to identify potential new cellular roles for 344 proteins, including 231 with no previously known functional annotation.
The analysis “provides an outline of the eukaryotic proteome as a network of protein complexes at a level of organization beyond binary interactions,” the study authors write. “This higher-order map contains fundamental biological information and offers the context for a more reasoned and informed approach to drug discovery.”
Cellzome’s data is publicly available at http://yeast.cellzome.com.
Both companies anticipate that this knowledge will be quickly harnessed in the drug discovery process.
In a related article in Nature, proteomic researchers Anuj Kumar and Michael Synder in Yale University’s Department of Molecular, Cellular and Development Biology caution that despite these impressive results, characterizing the proteome—even in yeast—still remains a daunting task.
These two teams have collectively identified as many as 11,000 protein associations, and their work indicates that a yeast proteome project is at least feasible. But given the multiplicity of protein interactions and physiological changes during the life cycle, Kumar and Synder write, “the characterization of all remaining interactions will almost certainly be labour intensive.”
Neubauer said that Cellzome had not yet determined how much the two teams’ data overlapped.