There are a vast number of research groups using proteomics to try and find that elusive biomarker that will act as a precise indicator of a particular disease. One of the most popular approaches uses two-dimensional gel electrophoresis (2-DE) to separate the many proteins extracted from biological fluids or tissues. Proteins which are expressed in markedly different quantities between the control system and the disease condition are flagged as potential biomarkers then identified, often by mass spectrometry.

It is no good if several diseases all produce elevated or reduced levels of a certain protein. To be a valid biomarker, a candidate must be unique to one disease or condition. Indeed, some practitioners have declared that it is too dangerous to rely on just one protein as a biomarker and recommend the use of a panel as a more reliable indicator.

These concerns appeared to be backed up by a recent study by Jiri Petrak and colleagues that was published in 2008. They examined 186 separate 2-DE proteomics reports and found that many of the same proteins were differentially expressed regardless of the experiment, species or type of tissue examined. Keratins and peroxiredoxins were most frequently identified, at least one member of the groups being differentially expressed in about 40% of all experiments. The group speculated whether these common changes were a result of common cellular stress responses or were caused by the technical limitations of 2-DE.

Now, a second group has performed a similar comparison with similar results. Edwin Mariman, Ping Wang and Freek Bouwman from Maastricht University generally work with 2-DE followed by MALDI TOF MS for peptide analysis and protein identification. They, too, recognised the repeated occurrence of some proteins across many studies, regardless of the basis of the experiment.

To check out this phenomenon, they located 66 proteomics studies from six journals and via PUBMED that were published from 2000-2007 and used 2-DE and mass spectrometry, mostly with MALDI MS. The species were limited to human, rat, mouse, fruit fly and worm (C. elegans) because these have high quality gene/protein databases. The proteins were transferred to human orthology and those that were differentially expressed from each study were entered into a database and the common ones were identified.

In this case, there were a set of 44 proteins that were considered to be "generally detected." Only 13 of the studies overlapped with those of Petrak, due to the different selection criteria. Nevertheless, 17 of the 23 common proteins identified by Petrak in humans or rodents were also present in the Mariman list, with seven of them in the top ten.

Both reports grouped the proteins into families and, again, there was significant overlap. Fourteen of the Petrak protein families were also present in the set of 28 Mariman families, with seven among the top ten.

The common proteins found in the latest study are a function of 2-DE. The technique does not detect low- and high-molecular-weight proteins and is biased against hydrophobic and extremely acidic or basic proteins. So, it favours hydrophilic, abundant and medium-sized (10-100 kDa) proteins. In addition, by examining whole proteome studies, most of the proteins would tend to be cytosolic rather than membrane or secreted proteins. The analysis bears this out with many of the 44 proteins possessing most of these features.

Examination of the functions of the 44 differentially expressed proteins revealed some commonality. Many have been associated with the cellular stress proteome and overlapped with proteins differentially expressed under oxidative or hyperosmotic stress. Other proteins could also have altered status under cellular stress, including the actin and intermediate filament families.

The results seem to imply that the cellular stress response in all of these studies is responsible for the fact that many of the same proteins are differentially expressed. No matter what the study or the species, the same proteins are altered.

For that reason, the researchers declared that none of these proteins should be considered as biomarkers for any conditions apart from cellular stress. They strongly recommended that other potential biomarkers be sought to prevent false conclusions being drawn from proteomics studies.

Related links:

• Proteomics 2008, 8, 1744-1749: "Deja vu in proteomics. A hit parade of repeatedly identified differentially expressed proteins"

• Proteomics 2009, 9 (Article in Press): "Generally detected proteins in comparative proteomics - A matter of cellular stress response"

Article by Steve Down

The views represented in this article are solely those of the author and do not necessarily represent those of John Wiley and Sons, Ltd.