Why we need a reactome

A powerful new tool to assess the functionality of the active proteins in any given cell -- the so-called reactome -- has been called into question. How would this recently developed "reactome array," described in a study published last October in Science, advance the field of functional genomics, and if it should fail, what would the ramifications be?

A biochip
Image: Flickr,
linkurl:Argonne Laboratory;http://www.flickr.com/photos/argonne/3397932229/

When Nobel Laureate linkurl:Richard Roberts,;http://nobelprize.org/nobel_prizes/medicine/laureates/1993/roberts-autobio.html currently the chief scientific officer of New England Biolabs, first read about the new technique, "it seemed too good to be true," he said. "It promised to do so much" -- uncover the functional identities of countless mystery genes -- "and seemed to do it very effectively". While current methods allow scientists to do this just one protein at a time, the reactome array would be able to look at hundreds of genes simultaneously. The array is comprised of nearly 2,500 metabolites and other substrate compounds, each tethered to a glass slide with a specially designed linker and coupled with a Cy3 dye label. The dye, which is initially inactive, becomes activated when an enzyme acts on a substrate; this reaction gives off a fluorescent signal and provides a quantitative measure of enzyme activity. Additionally, because the substrate linkers physically trap reactive enzymes, proteins can be isolated and identified using mass spectrometry after the array returns its results, allowing the array to analyze the contents of entire cells or tissues all at once. Not long after the publication of the new methodology, however, scientists began raising concerns about the dye-labeled metabolites that are key to the array. According to Science editor-in-chief Bruce Alberts, "a number of scientists" contacted Science with concerns about the paper. After receiving these queries, Alberts wrote to the two corresponding authors, Manuel Ferrer; of the linkurl:CSIC;http://www.csic.es/index.do Institute of Catalysis in Madrid and linkurl:Peter Golyshin;http://biology.bangor.ac.uk/people/staff/025123 of Bangor University in the UK. However, when the authors' replies did not resolve the "inconsistencies" in the paper and supplementary data, Alberts decided to post an 'Editorial Expression of Concern' that appeared in the journal last week (January 8). "We believe that it is our responsibility to alert our readers to such concerns, so that time and resources are not spent on independent efforts to replicate or build upon the work until the issues are resolved," Alberts wrote in an email to The Scientist. The CSIC has launched a formal investigation to verify the reactome array's validity, upon Alberts' request that they look into the matter further "by examining the original data and lab notes." "We don't know what happened," CSIC cell biologist Pere Puigdoènech told Nature News last week, "but we expect that the committee will be able to deliver a solid report within a couple of months." Additionally, at least two scientists, both in collaboration with Ferrer, have taken up the challenge. Microbiologist and bioinformatician linkurl:Frank Oliver Glöckner;http://www.mpi-bremen.de/en/Frank_Oliver_Gloeckner.html of the Max Planck Institute for Marine Microbiology in Bremen, Germany, recently received the first set of results back from Ferrer, who ran a sample from the marine microorganism Rhodopirellula baltica through the array. While Glöckner and his team are mainly interested in genes of unknown function, they are first using these data to confirm the technique's accuracy by comparing its results to the functions of more well understood genes. If the known functions of those genes match up with the array's results, they will then begin to explore the 300 genes of unknown function that the array returned on its first run -- an impressively high number, Glöckner said. While R. baltica has more than 3,000 genes of unknown function, "we couldn't have expected all of the genes [the first time]," he explained, since they did their experiment with just one sample under one condition. Additionally, Roberts, who is interested in using the technique to annotate the entire genome of the bacterium Helicobacter pylori, is conducting a blind experiment of his own. He recently sent 10 enzymes to Ferrer to run the array and return its results -- enzymes whose functions are known to New England Biolabs, but not to Ferrer. "From everything I've seen so far, it looks pretty good," Roberts said. "At the moment, I have no reason to think it's fraudulent in any way." Ferrer himself is standing behind the technique, though he conceded to Nature News that there was an error in the schematic in the paper depicting the methodology, as well as a few "small errors" in the supplementary information, all of which have now been fixed, he said. The new array comes at a key point in microbiology's history -- just as the sequence data is beginning to pile up. Of the thousands of new sequences, the function of only about half of those genes can be inferred from the databases. For the rest, we have " no clue what they are doing," Glöckner said. If it proves successful, the reactome array could help close the gap between the quickly advancing sequence data and the functional annotation that is lagging further and further behind, and provide an accurate and comprehensive way to study cellular metabolism. In addition, an accurate array could also lead to the discovery of "interesting reactions for biotechnological applications," Glöckner added. "If everything works as promised, it's fantastic," Roberts said. "It's a massive breakthrough."
**__Related stories:__***linkurl:Bring Me Your Genomes;http://www.the-scientist.com/article/display/15531/
[6th June 2005]*linkurl:What's Next for Bioinformatics?;http://www.the-scientist.com/article/display/15483/
[23rd May 2005]*linkurl:Reactome Explores Biological Processes;http://www.the-scientist.com/article/display/14875/
[2nd August 2004]