THOUGH THE EXPERIMENT INVOLVED STRENUOUS EXERCISE, the implications went far beyond the defining health benefits of working up a sweat. Robert Gerszten, a cardiologist at Massachusetts General Hospital, led a team of researchers that recruited two groups of several dozen middle-aged men. Those in the first group were fit and healthy; those in the second had been referred for testing because of shortness of breath or suspected heart disease. The men in both groups were asked to work out for 10 minutes on a treadmill or stationary bicycle at an accelerated pace, and their blood was drawn three times—right before and after the workout and then an hour later.

Gerszten’s team had undertaken the study because scientists have little knowledge about many of the chemical mechanisms involved in physical exertion. When they analyzed the blood samples, they chose to focus on metabolites, molecular by-products of the hundreds of thousands of chemical reactions always going on in every human cell. Identifying metabolites can be exceedingly difficult because they come in diverse and often unstable chemical forms, sometimes undergoing transformations even as they’re being isolated. Earlier exercise studies had measured only a handful of metabolites, mostly related to amino acids. But recent improvements in techniques to analyze such metabolites as lactate, pyruvate and glutamine—known to be involved in how the body burns fats, sugars and amino acids during exercise—helped facilitate Gerszten’s monitoring of more than 200 substances. Twenty or so proved particularly interesting, with levels that changed significantly during and after the 10-minute workouts.

The concentration of glycerol, a metabolite released into the bloodstream when fat is burned, rose sharply, whereas levels of allantoin, an indicator related to a cell’s ability to regenerate, went down. What’s more, while there was a boost in glycerol for all test subjects, the spike was much larger for those in the fitter group at their exercise peak, suggesting that there was something about their physiological makeup that helped them break down fats more quickly than did their less fit counterparts. In leaner subjects, levels of niacinamide, a metabolite involved in regulating insulin or blood sugar levels, increased more than twice as much as it did in heavier individuals.

Those results, and a wealth of other information Gerszten’s group is continuing to explore, are helping expand the frontiers of exercise physiology. But more than that, the study represents a major advance in metabolomics, the study of metabolites’ role in health and disease. Like genomics, which focuses on genes, and proteomics, which considers the proteins that genes encode (produce), metabolomics attempts to increase what we know about human physiology. And by focusing on the by-products of metabolic processes—which produce such substances as sugars, amino acids, lipids and fatty acids—metabolomics may relate most directly to how the body functions.

Involved in everything from digestion to waste elimination and temperature regulation, metabolites provide here-and-now evidence about conditions ranging from Alzheimer’s to breast cancer and Huntington’s disease. “Genes give the blueprint of a house, while metabolites give you a minute-to-minute snapshot of what’s actually going on in the house,” Gerszten says. “We can’t take pictures of every brick, but we can look at hundreds and hundreds of them through metabolomics.”

There are, in fact, thousands and thousands of bricks. Scientists involved in the Human Metabolome Project have so far identified the chemical composition of about 3,000 metabolites likely to be essential for growth and development, and 5,000 more compounds have been detected. The total number of metabolites might exceed 100,000. The payoff, though still years away, should be substantial. “I think that maybe 2,000 to 3,000 metabolites are going to explain most of human biology,” says Arthur Castle, program director of metabolomics initiatives at the National Institute of Diabetes and Digestive and Kidney Diseases at the National Institutes of Health.


BY SOME RECKONINGS, METABOLOMICS IS A VERY OLD SCIENCE. The ancient Greeks believed human fluids and tissues could hold clues to disease, and physicians during the Middle Ages consulted “urine wheels”—diagnostic charts of colors, smells and tastes. In 1905 the American chemist Otto Knut Olof Folin published a list of compounds in urine that could help distinguish healthy and insane patients, and in the 1960s Nobel laureate Linus Pauling collected human breath samples in test tubes. Though he lacked the technology to prove that chemicals in the breath could provide clues about disease, Pauling was among several mid-century scientists who worked to develop methods to test small numbers of metabolites in humans. A few decades later, Jeremy Nicholson, a professor of biological chemistry at Imperial College London, explored a technology, nuclear magnetic resonance, that could identify large numbers of metabolites simultaneously. Another method, mass spectrometry, soon also proved useful for finding metabolites.

The field’s eventual name—metabolomics—was an obvious nod to genomics, the focus of the Human Genome Project, which in 2003 concluded its landmark effort to identify the 20,000 to 25,000 human genes and determine the sequence of the estimated 3 billion chemical base pairs that make up human DNA. Scientists thought that by studying that genetic map they would unlock the secrets of human biology, and the genomics revolution, though proceeding more slowly than many enthusiasts had predicted, has indeed led to breakthroughs in understanding and treating disease. But it turns out that the relationship between genes and the other building blocks of the body—including the RNA that transcribes genetic instructions from DNA, the proteins that are then created and the metabolites that finally result—is far more nuanced than once believed. There’s a lot of “cross talk” among these different elements, says Gary Siuzdak, a senior director for the Center for Metabolomics and Mass Spectrometry at the Scripps Research Institute in La Jolla, Calif., and the hope is that studying metabolites—farthest downstream in this cascade of processes—will reveal clues about what’s happening in a cell that would be difficult or impossible to observe by looking at genes or proteins.

The Human Metabolome Project, launched in Canada in 2004, has the daunting mandate to identify, quantify, catalogue and store all known information about metabolites. Since the project began, the pace of research to unravel metabolomic mysteries has accelerated on many fronts, with funding from major pharmaceutical companies and the National Institutes of Health, which allocated an estimated $50 million in 2010, says Castle. “Metabolomics is allowing us to achieve insights about what’s occurring in the disease process that aren’t being learned through the other ‘omics’ sciences,” Castle says.

AMONG EARLY ADVANCES ARE IDEAS ABOUT HOW TO GAUGE when particular diseases are taking hold. With prostate cancer, for example, today’s chief diagnostic tool is a blood test that measures a protein, prostate-specific antigen, or PSA. Produced by the prostate gland, PSA is present in small quantities in the blood of healthy men, but elevated readings may indicate cancer. Yet PSA test scores often aren’t conclusive and may be followed up by an invasive tissue biopsy. Another disadvantage is that PSA results—and even a subsequent positive biopsy—won’t necessarily show the cancer’s stage and aggressiveness. Patients with slow-growing cancer may not need treatment, yet because the disease can be deadly, physicians and patients frequently opt for prostatectomies or other procedures, all of which carry a risk of incontinence and impotence.

Could measuring metabolites do a better job? A University of Michigan study analyzed 1,126 distinct molecules in 262 samples of tissue, blood and urine from men who were either healthy or who had early-stage prostate cancer or the full-blown disease. The researchers identified 10 substances that were present with increasing frequency as cancer became more advanced; the amino acid sarcosine particularly stood out. Levels of the little-known sarcosine were elevated in about 80% of metastatic prostate cancer samples, in 40% of early-stage cases—and in none of the cancer-free study participants. Further analysis has suggested that sarcosine plays a direct role in making prostate cancer more aggressive and that drugs blocking its activity might improve treatment. It’s also possible that sarcosine could be developed as a more precise screening tool than PSA.

Other experiments have identified metabolites apparently involved in type 2 diabetes. Testing blood glucose, itself a metabolite, has long been a way to monitor the disease, in which either the body doesn’t produce enough insulin or cells don’t respond to the hormone. Insulin is essential for the body to utilize glucose, a form of sugar that generates energy; it normally stimulates the transport of glucose from blood into cells and acts as a key that opens the cell to allow glucose to enter. When the body produces insufficient levels of insulin, that key doesn’t work, and dangerously high levels of glucose build up in the bloodstream. Left untreated, high blood sugar can lead to such complications as blindness, nerve damage and kidney damage.

But glucose is far from the only metabolite related to diabetes, and Christopher Newgard, director of the Sarah W. Stedman Nutrition & Metabolism Center at Duke University, and his research group have found a metabolite associated with protein metabolism (rather than with glucose metabolism) that could help explain how obese people develop type 2 diabetes. The researchers recruited 74 healthy obese people and 67 healthy lean people, who fasted overnight. The researchers then used mass spectrometry to measure hundreds of metabolites from blood and urine samples taken from both groups. The obese participants had elevated levels of leucine, isoleucine and valine—known collectively as branched-chain amino acids, or BCAAs—as well as a cluster of six small metabolites that were produced when the BCAA metabolites broke down. “The overabundance of these metabolites led us to investigate how they might be associated with insulin resistance,” Newgard says.

For that additional work, conducted in rats, Newgard’s researchers fed one group all the high-fat food they would eat, while two other groups got either low-fat chow or high-fat chow with BCAAs. The rats on the fat/BCAA diets didn’t eat as much or become as chubby as the rats on the high-fat diet, but they became just as insulin resistant. This suggests that an overabundance of BCAA and its metabolites in a cell may disrupt insulin regulation and glucose metabolism. Newgard notes that about 20% of the protein in the typical American diet is made up of BCAAs. The findings suggest that such foods may trigger insulin resistance in obese people. “There’s not only fat in that hamburger but plenty of protein,” Newgard says.


METABOLITES CAN ALSO HELP RESEARCHERS DECIPHER HOW BIOCHEMICAL NETWORKS of genes and proteins interact and contribute to disease, and a study published last year looked at a metabolic pathway related to a gene, IDH1, implicated in the formation and progression of brain cancer. A large, multi-institutional research team teased out metabolites from a glioblastoma cell line with mutated IDH1 and found high levels of a metabolite called 2-hydroglutarate (2HG). When the team went on to analyze freshly biopsied tumors from brain cancer patients, they found that every tumor harboring the IDH1 mutation had as much as a hundredfold elevation of 2HG. “This was completely unexpected,” says Joshua Rabinowitz, a Princeton chemist on the team. “This metabolite appears to be made specifically in these brain cancers.”

In further experiments, the team mapped the structural molecular defects of the mutated IDH1 enzyme that results in the production of 2HG. “In theory, at least, you could make a drug that targets the altered shape of this mutant enzyme. Such a drug might fight this cancer with minimal side effects,” Rabinowitz says.

Other therapies might arise from an extension of Gerszten’s study in which an MGH research team analyzed blood samples taken from nearly two dozen runners before and after they completed the 2006 Boston Marathon. The scientists found marked changes in several metabolites in the post-race samples, including elevations in glycerol of more than 1,000%, while metabolites such as alanine, glutamine and asparagine—amino acids used as fuel to maintain adequate glucose levels—were depleted. The researchers also discovered significant differences in metabolite levels between faster and slower runners.

The MGH team used these results to select metabolites for testing to see whether they played a functional role. Seeking to mimic the effects of exercise in the lab, the researchers applied a “cocktail” of five metabolite molecules—selected because their levels had spiked in the exercise study—to human muscle cell tissue. That resulted in a threefold increase in the production of a transcription factor called Nur77. Transcription factors affect the way genetic information is conveyed, and in earlier animal studies, a lack of Nur77 had been linked to obesity and diabetes. The group then repeated the experiment, applying the metabolite cocktail to muscle cells taken from mice that had been exercised to exhaustion. Nur77 was found to increase fivefold in the mice muscle cells—suggesting, among other possibilities, that metabolites resulting from vigorous exercise reduce the risk of obesity and diabetes.

SUCH WORK WITH METABOLITES COULD BE INVALUABLE in understanding human physiology and how healthy processes may go off track. But identifying human metabolites and drawing conclusions about their significance is complicated because metabolite concentrations may vary widely from person to person and even in a particular individual, depending on the time of day, what the person has eaten, what else is going on in the body and what environmental pollutants may be present. “We’re only beginning to catalogue the variability among different people,” says Gerszten.

The ability to identify and analyze the significance of thousands of additional metabolites awaits advances in technology and statistical analysis as well as the development of global research standards. “Different groups use different approaches, and that makes it hard to reproduce results,” Newgard says, though researchers are now working to create best practices and reporting standards.

Current limitations also mean that while metabolomic research is proceeding quickly across many fronts, there are still only a handful of metabolites used to aid clinical diagnoses. Standard blood work and urinalysis measure levels of compounds known to be important, and newborns are tested for a few dozen metabolites that help screen for conditions that cause mental retardation, brain damage and congenital hypothyroidism, among others. But rapid progress could come soon. “Metabolomics today is sort of where we were in genetics prior to the Human Genome Project—everyone was discovering genes and trying to figure out if this gene I’ve found is the same gene you found,” says NIH’s Castle. “But we’re doing things now in genomics that 10 years ago we thought were impossible. There’s no reason to expect that metabolomics won’t follow the same pattern.”