A Test Bed for Budding Technologies

DELETION MUTANT HOME BREW

Capitalizing on the yeast genome's ease of manipulation, an international consortium of yeast geneticists constructed a set of gene-deletion mutants representing 96% of the annotated yeast open-reading frames (ORFs).¹ Thousands of genes (5,916) were deleted and tagged with two so-called molecular barcodes to permit analysis of multiple strains in a mixed culture. Consortium members then profiled the strains under six different conditions and found that 80% of the deletions, roughly 5,000 strains, did not lead to cell death. The work represented the first global functional genomics analysis for any organism.

"It's a landmark paper because it's the first time that every gene in a eukaryotic cell was deleted to create a complete set of mutants," says Charlie Boone of the University of Toronto. "It now allows us to study particular processes systematically, because we can just test every mutant for a specific phenotype."

The set has proven remarkably versatile. "It's been much more widely used than even I imagined," says Mark Johnston of Washington University, St. Louis, who is one of the paper's senior authors. Some, like Fields, use it to comprehensively screen chemicals or small-molecule drug candidates, or to study processes such as aging, DNA damage, and metabolism. Recently, two of the set's developers, Guri Giaever and Ronald Davis of the Stanford Genome Technology Center screened 10 small molecules against the complete set of deletion strains, identifying known and novel protein-molecule interactions.³ Others use it as a step toward large-scale genetic-interaction mapping. Boone crossed 132 mutations against a set of 5,000 viable strains and scored these double mutants for lethality, which indicates a level of interconnectivity between the deleted genes.⁴

Tim Hughes, also at the University of Toronto, used micro-arrays to assay several hundred yeast deletion strains for those involved in RNA processing.⁵ Hughes says that he might not have even done the research without the collection. "We certainly couldn't have done it as quickly, and there's some question whether we would have done it at all," Hughes says.

While developing the set was technically demanding, getting more than 20 laboratories to work towards the same goal was also a challenge, says Fields. "It was quite a logistical feat to pull it off and have that many different labs working to make all of the same type of deletion and do it comprehensively on the entire set of open-reading frames that are predicted," Fields says. That it worked at all reflects well on the yeast research community, he notes, as does the continued effort to improve on the collection. Three labs – Davis', Mike Snyder's at Yale University, and Jeff Boeke's at Johns Hopkins – are currently developing version 2.0 of the knockout collection, fixing secondary site mutations and incorporating ORFs from a reannotated genome. Second author Angela Chu, main curator of the collection, estimates that after strain verification, version 2.0 might be released publicly in mid-2005.

SITE DETECTION

The second hot paper moves from functional genomics to proteomics, specifically the identification of protein phosphorylation sites, which has long been a challenge. Many conventional efforts involve proteolytic digestion followed by mass spectrometry, but these fail to identify low-abundance proteins and are limited to detecting, at most, 20 sites from a whole-cell lysate. In 2002, Scott Ficarro and colleagues at the University of Virginia, Charlottesville, described a method to enrich for phosphopeptides prior to separation via immobilized metal anion chromatography (IMAC) and detection by mass spectrometry. They were able to detect 383 sites in a whole-cell lysate from Saccharomyces cerevisiae.²

"It's certainly the most exciting paper to come out in large-scale phosphorylation analysis in many years, and it's the paper that we judge all of the other manuscripts by," says Steve Gygi of Harvard University Medical School.

Donald Hunt, one of the paper's senior authors, first used IMAC to characterize phosphopeptides from a plant protein in 1981. As interest in proteomics grew, Hunt began developing LC/MS techniques for high-throughput peptide sequencing and ultimately decided to combine the two methods for large-scale analysis of phosphoproteins. He found, however, a large amount of nonspecific binding in highly complex mixtures. "Any peptide that contains a c-terminus and a couple other acidic amino acids will also bind [to the IMAC column], and they bind fairly tightly. And if they're present in large excess, then they prevent binding of phosphopeptides," Hunt says.

Hunt's solution: Convert the phosphate groups into methyl esters so that they stick preferentially to the columns. The work represented an order of magnitude increase in the ability to map protein phosphorylation on a global scale, says senior author Forest White of Massachusetts Institute of Technology.

"That then opened up the way for large-scale phosphopeptide analysis," says Matthias Mann of the University of Southern Denmark. Ficarro, for example, applied the same technique to the analysis of phosphorylated sites in human cells, but with less spectacular results.⁶⁷ Gygi comments that the two follow-up papers by Ficarro, which analyzed human phosphoproteins, surprisingly identified fewer sites – between 60 and 70.

Some groups that have tried to reproduce the results simply failed. Mann, who has experienced trouble with the technique, speculates that the difficulty may simply lie in the stickiness of phosphopeptides, which in addition to sticking to metal-based affinity columns, also can stick to the metal parts of the HPLC apparatus. But, counters Hunt, some labs may be performing the method incorrectly. Yates says that the enrichment technique used in the Ficarro paper tends to work differently in different hands. "My lab has never been particularly successful at using it, but I don't think that that necessarily means that Hunt's results are wrong," says Yates.

Other methods that take advantage of the negative charge or hydrophilicity of the phospho group also are being developed, says Mann, as well as those that modify the phospho group using chemical methods.⁸ Gygi points to other unpublished techniques as promising alternatives for large-scale analysis of phosphorylation sites. For example, Cell Signaling Technology of Beverly, Mass., is developing an immunoprecipitation-based method.

Other techniques, he says, have not panned out as large-scale analysis methods. Indeed, no other method has been shown to identify a similar number of phosphorylation sites in a complex mixture of proteins. And the Ficarro paper went further than most in terms of validation. "They synthesized a large number of those peptides and confirmed that the interpretations were actually correct. That's not something that people ordinarily do," says Yates.

Hunt's laboratory continues to refine the technique, and has since developed a new technique for collecting peptide mass spectra, which should facilitate characterization of phosphorylated and other posttranslationally modified peptides. "I think this new paper will revolutionize how people apply mass spectrometry to proteomics," Hunt says.

Aileen Constans can be contacted at aconstans@the-scientist.com.