Public Projects Gear Up to Chart the Protein Landscape

See allHide authors and affiliations

Science  21 Nov 2003:
Vol. 302, Issue 5649, pp. 1316-1318
DOI: 10.1126/science.302.5649.1316

Researchers in industry as well as those in the public sector seek the protein equivalent of the human genome sequence

MONTREAL, CANADA—With the human gene sequence now in hand, researchers have moved on to a new goal: identifying all the body's proteins. The task is massive. Not only does each of the body's 252 cell types harbor its own complement of proteins, but their expression patterns also vary with age, nutrition, health, and disease.

These difficulties haven't deterred researchers who want to determine the body's complete set of proteins—the proteome, as it's called. Buoyed by hopes that their efforts will identify proteins that can serve as both markers for disease and targets for new drug therapies, the pharmaceutical industry jumped into the research a few years ago, investing hundreds of millions of dollars in high-speed protein-tracking technology (Science, 7 December 2001, p. 2074).

More recently, the public funding agencies have gotten into the act, launching their own large-scale proteomics projects. They say they had little choice, as the corporate push has left university researchers out in the cold. “Companies don't open their resources for academics,” says Marius Ueffing, a proteomics researcher at the German Society for Proteome Research and the Institute of Human Genetics in Munich.

To fill that gap, four separate public international proteomics initiatives have been launched over the past year and a half. Three of them, spearheaded by researchers in the United States, Germany, and China, are aimed at tracking down all the proteins in human blood plasma, the brain, and liver. A fourth effort seeks to create antibodies against thousands of human proteins, a resource that should help researchers devise other protein-tracking tools.

Additional proteome projects are looming, with kidney, muscle, heart, and saliva among the possible targets. “We are being bombarded by groups that want to have this or that initiative,” says Samir Hanash, president of the Human Proteome Organization (HUPO), an international coordinating group. Once researchers have collected snapshots of the ever-changing proteomes of different tissues and cells, they hope to assemble them into a kind of full-length movie showing the ebb and flow of proteins in the body.

Speed boost.

An international effort to churn out 10,000 antibodies to human proteins could enable the development of protein-tracking technology, such as the antibody microarrays produced by this machine.


In addition to these major efforts, research groups are also pursuing numerous smaller efforts, such as tracking down all the proteins in subcellular structures—including the mitochondria and Golgi—as well as in various microbes (see sidebar). “It's a very, very hot field,” says Fuchu He, director of China's Beijing Institute of Radiation Medicine and head of HUPO's liver proteome project.

Initial plans and results from many of these projects were on display at HUPO's 2nd Annual World Congress held here 8 to 11 October. Just how they will unfold is uncertain. “We're still early on and testing the waters,” Hanash says.

Indeed, large-scale programs are likely to be far more difficult to pull off with proteins than with genes. The chemistry is more variable, for one. Some proteins reside in watery environments such as blood, for example, whereas others hide out in the fatty membranes surrounding cells. Proteins range considerably in size, from 5000 daltons to 1 million or more. They differ in their electrical charges. And most challenging of all, most proteins exist only in vanishingly small quantities. “This is a nightmare analytically,” says Thomas Conrads, a biochemist at the National Cancer Institute at Frederick (NCI-Frederick) in Maryland.

The equipment is more demanding, too. Whereas genome researchers could concentrate their resources on a single technology, sequencing machines, that's not possible in proteomics. “No method right now is able to analyze a complete single proteome,” says Thierry Rabilloud, a proteomics expert at the French Atomic Energy Commission in Grenoble. As a result, proteomics researchers must make hard choices about what to go after. “You have to reduce the proteome to manageable units and define achievable goals,” Hanash says.

That's where HUPO has stepped in. Launched in 2001, this loosely knit federation of proteome researchers doesn't dole out any research funds itself. That money comes from traditional biomedical funding agencies in participating countries. But HUPO helps set priorities, coordinate research, set standards for handling and processing samples, and arrange for the use of common bioinformatics tools to ensure that researchers can directly compare their results.

HUPO didn't waste any time picking favorites. The organization quickly targeted blood plasma as its first priority, aiming to discover blood-borne proteins indicative of disease. The $1 million pilot project was launched in April 2002 and currently consists of researchers from 47 labs around the globe, including 28 in the United States.

For now, HUPO's Plasma Proteome Project (PPP) is focused on comparing the strengths and weaknesses of different protein-hunting technologies, such as two-dimensional gel electrophoresis and liquid chromatography for separating proteins, and various versions of mass spectrometry for identifying them.

In July, PPP's leaders mailed out standardized plasma samples to participating labs and asked each to use its technique of choice to separate and identify as many proteins as possible. HUPO leaders plan to set recommendations about which technologies are most appropriate for tracking down different subsets of plasma proteins. Initial results, which started coming in last month, suggest that the task won't yield simple answers, however.

Conrads, for example, described how the NCI-Frederick team had managed to sift out some 1444 plasma proteins using a standard technique—first separating out the high-abundance proteins, dividing the rest of the sample into numerous fractions, and then scanning each one using an electrospray mass spectrometer. That laborious effort ensures that researchers don't see only common proteins such as albumin, which constitutes up to 50% of the total protein in plasma.

Where's Waldo?

Most plasma proteins, including potential diagnostic markers and drug targets, are present in only tiny quantities.


But discarding the high-abundance proteins comes at a cost. When the NCI-Frederick team took a closer look at such proteins, they found that they readily bind a wide variety of low-abundance proteins, acting like molecular sponges. The researchers identified 341 different proteins and peptides bound to albumin alone. Other sets of proteins bind to other highly abundant proteins such as antibodies. “This was rather stunning to us,” Conrads says. “Each one of these proteins is binding different peptides. So there does seem to be some precise interaction.” Although it's still too early to be certain, Conrads says it might be possible to use common proteins to track down the low-abundance proteins and peptides that most groups are interested in as potential diagnostic markers.

PPP director Gilbert Omenn of the University of Michigan Health System in Ann Arbor says that other techniques will have their own strengths and weaknesses. It's still too early to make meaningful comparisons, he adds. He expects that the final results of the plasma analyses should be in by the end of the year.

PPP aims to move on to a full-scale plasma project, attempting to correlate changes in the abundance of select proteins with various diseases. There could be a snag, though. The project could cost about $50 million, Omenn says, and he hasn't yet identified a funding source. But he is confident that “there will be a major follow-up.”

HUPO's Human Brain Proteome Project (HBPP) is also up and running. Begun in April as a pilot project, HBPP builds on a brain proteome project begun in 2000 and backed by $17 million from the German ministry of research. The collaboration aims to sort out technology and standards issues, focusing at first on tracking down the proteins in the substantia nigra and hippocampus, the brain regions that degenerate in Parkinson's and Alzheimer's diseases.

Project co-leader Helmut Meyer, a protein chemist at the University of Bochum in Germany, says that HBPP researchers hope to find proteins that mark the early disease stages, because most damage to brain cells occurs before the first symptoms show up. If the researchers find these markers they can ask, “Are they already visible when people are 30 or 40 years old?” Meyer says. The HBPP team will also scan cerebral spinal fluid and blood plasma for proteins linked to brain diseases.

The early results reveal some tantalizing hints of what's to come. At the meeting, HBPP co-director Joachim Klose, a protein scientist at Humboldt University in Berlin, described a series of experiments with mice. Klose and colleagues tracked changes in the abundance of 250 brain proteins as the mice grew from embryos to aging adults. As expected, the researchers found that the overall amount of proteins remained essentially constant from a few days after birth until the animals died.

On the hunt.

Proteomics researchers hope that their quest to find thousands of novel proteins will turn up good candidates for new therapeutic drugs such as the so-called Src kinase, which has been implicated in some cancers.


Even so, the abundance of a large percentage of different proteins changed considerably during the animals' early growth. But the researchers were somewhat surprised to find that nearly 20% of brain proteins continued to change their abundance levels when the animals were in the final stages of life. The results, Klose says, suggest that changes in members of this protein subset could be linked to disease.

The Human Liver Proteome Project (HLPP), meanwhile, is backed by an initial round of $25 million in funding from the Chinese government for a 3-year pilot study to be completed in 2005. The effort was launched in May and is aimed at setting up the collaborations, standards, and procedures for tallying the thousands of proteins expressed in human liver cells. So far, 79 labs, 37 of them in China, have signed on to the liver proteome effort. The project's leader, Fuchu He, says he expects a full-scale production phase to follow from 2006 to 2010.

As with the other proteome efforts, the idea is to ultimately link liver-specific proteins to diseases such as hepatitis and liver cancer. According to He, the Chinese government has promised to kick in another $250 million if HLPP makes it into the production phase. This commitment, He says, stems from the fact that liver diseases kill hundreds of thousands of people in China each year.

Proteomics experts caution, however, that they can't link a change in protein expression to disease from just one or two samples. “If you want sound results, you will have to repeat it five or 10 times,” Klose says. That's not likely to be accomplished with the large proteome projects.

But confirming a protein linkage to disease should become much easier if HUPO's fourth initiative—to make a vast library of antibodies against human proteins—succeeds. Such antibodies could be used to track a particular protein in many people as a way of confirming its involvement in a disease. In addition, Omenn says, this project will help researchers in each of the other initiatives create protein microarrays and other tests for tracking the ebb and flow of thousands of proteins simultaneously in different tissues.

Both HBPP and HLPP have dedicated a significant portion of their early funds to antibody production. And a group led by Ueffing is applying to companies and the European Union for an initial round of $12.5 million in research funding on what it hopes will eventually become a $60 million initiative to raise antibodies against 10,000 human proteins. If researchers line up the money, the antibodies produced could make life easier for proteomics researchers.

Navigate This Article