Beyond the Human Genome, Proteomics the Next Frontier
By Nuala Moran
VANCOUVER, British Columbia – The recent history of biology has been dominated by the Human Genome Project, but even as the first complete sequence was published 11 years ago, it was evident that translating that into something of use for human health and medicine would require a far greater undertaking - of mapping and understanding all the proteins in the human body.
The 20,244 genes in the genome somehow hold the instructions for 5 million protein forms in humans, making proteins the "ultimate transformers" able to splice and switch roles, Chris Overall, a researcher in metalloproteinase proteomics and systems biology at the University of British Columbia told attendees at the American Association for the Advancement of Science meeting Friday.
While it takes 20,000-odd genes to make a human, it takes almost the same number to make a fly or a worm. So the question is, "Can proteomics fill the gap between the genome and phenotypes?" Overall said.
Mechanisms such as proteolytic processing, slicing, nonconventional trafficking and protein-protein interactions generate information and new forms of protein that cannot be predicted from the genetic code, or from expression data. The information thus generated leads to phenotypic differences between the states of biological system, between individuals and between populations, and ultimately is reflected in human health and disease.
And, for the industry, is it the source of unexpected drug off-target effects.
Overall's own work in metalloproteinases has led to a shift in understanding how those enzymes control the immune response, and how subtle changes involving no more than four amino acids can lead to the innate immune response being blocked, resulting in chronic inflammation. Overall has applied that finding to understanding the mechanisms behind periodontal disease.
That such subtle changes can lead to a complete reversal of a protein's activity is bad news for the biotech and pharma industry. The finding underlies the reason so many metalloproteinase inhibitors have failed in development as drugs, Overall believes.
He described adding a protease inhibitor to cells and monitoring the proteomic changes that result. "You get levels of 30 other proteases and their inhibitors going up and down," he said.
That is clearly a problem. "MMPs have beneficial roles, so if any drug blocks that action you can end up with worse inflammation," Overall said.
Overall also has worked on "moonlighting" proteins, which take up a different function depending on their location. Those proteins change roles when proteases splice their ends, creating new "termini" that enable them to interact differently. "The problem with moonlighting proteins is that they can be a drug target in one [location] but performing an essential function in another," Overall said.
That raises the question of what the industry can do to confront the unfolding complexity of the proteome.
Overall suggests the most important thing is target validation to make sure the protein has an unequivocal role in disease. "There has to be very critical and in-depth basic-level research," he said, adding that the appropriate place for that to take place is in academia.
While the complexity of the proteome may be a headache for drug development, proteomics is beginning to deliver improvements for human health and medicine, believes Gil Omenn, of the Center for Computational Medicine and Bioinformatics at Michigan University, who is chair of the Human Proteome Project, the international collaboration begun in 2010 to deliver the complete map of human proteins in context.
"There is really active work leading to a flow of papers on proteomic variants in breast, prostate and lung cancer, for example," Omenn said.
Those and other biomarkers are beginning to drive the three components of personalized medicine: first getting the correct diagnosis sooner; then coming up with the correct prognosis and so avoiding unnecessary treatment – for example, predicting whether prostate tumors will become cancerous; and, thirdly, defining the most appropriate treatment.
The sequencing of the human genome is a fundamental and important piece of science, but it can't predict the phenotype.
"The fact is, much of heritability is down to the environment," Omenn said. "At present 180 SNPs [single nucleotide polymorphisms] have been associated with height, but this explains only 10 percent of the variation."
Even as the price of sequencing falls and more genomes are sequenced, the new data generated will only explain a small additional proportion of the variation seen. "Proteomics is the way to fill the gap," Omenn said.
Outside of the US
In the U.S. and Canada: +1-800-336-4474
Outside the U.S.: +44-203-684-1796
Hours: Monday - Friday, 8:00am - 6:00 pm EST