We're living in an "omic" world. Some of these "omes," such as the genome and the proteome, are familiar; others, less so. Now the metabonome, one of the newest omes in name if not in reality, has joined the pantheon of global biological measurements.

	METHOD OF CHOICE NMR is the most popular method for metabonomics experiments. The NMR facility at Imperial College is shown here. COURTESY OF JEREMY NICHOLSON

The "omics" suffix has come to signify the measurement of the entire complement of a given level of biological molecules and information. Therefore, genomics measures the entire genetic makeup of an organism, while proteomics measures all the proteins expressed under given conditions. Metabonomics is no different. As the name might imply, metabonomics is defined as measurement of the complete metabolic response of an organism to an environmental stimulus or genetic modification. Some people use the term metabolomics to refer to metabonomics at the level of a single cell type, rather than a larger system.

The omics can provide information for basic biological research and for pharmaceutical and clinical applications. One of the challenges is integrating the information from the various omics, something that really is only beginning. The goal of a meeting held last month in San Francisco by the California Separation Science Society was just such an integration. In the process, the organizers coined yet another word--systeomics--which was defined as the integration of genomics, proteomics, and metabonomics.

Despite the goal of integration, scientists appear to be sticking with their favorite ome. Most speakers concentrated on one of the areas without addressing how to fit the three together.

Metabonomics may be the most recently named of the omics, but it's one of the oldest. In fact, metabonomics harkens back to old-fashioned biochemistry, with its emphasis on metabolism, the sum of the processes to acquire and use energy in an organism, to biosynthesize cellular components, and to catabolize wastes.

"We've been doing toxicological and disease diagnostics based on metabolic profiling for more than 20 years. That's before genomics or proteomics raised their ugly heads," Jeremy Nicholson, professor of biological chemistry at Imperial College of Science, Technology & Medicine in London, told C&EN.

Nicholson believes that metabonomics is "more closely related to things in the clinical world" than either genomics or proteomics, owing to the fact that metabonomic signatures reflect both genetic information and environmental influences.

John Lindon, another professor of biological chemistry at Imperial College, agrees. "Genomics and proteomics are in 'omics world.' They're not in the real world," Lindon said. "What you're trying to do is relate changes in gene expression or changes in protein level with some real-world endpoints that relate to a disease or toxic episode."

Adelbert Roscher, professor of biochemical genetics on the medical faculty at the University of Munich, said that metabolite profiling "measures the real outcome of potential changes suggested by genomics and proteomics."

	Image is available in ( Adobe PDF format only)
	BIG PICTURE Metabonomics offers the opportunity to find patterns and changes in the entire metabolism, represented here by the metabolic pathways chart designed by Donald E. Nicholson, retired from the University of Leeds. ©INTERNATIONAL UNION OF BIOCHEMISTRY & MOLECULAR BIOLOGY

THE VALUE OF genomic and proteomic measurements, Nicholson believes, is "considerably more limited than most people think." For example, changes in gene and protein expression needn't result in an "endpoint change." That is, the change in one gene or protein could be compensated elsewhere, resulting in no net change. "That's always the big problem with genes and proteins," Nicholson said. "Their up or down regulation can be part of the overall homeostatic or corrective process of the cell, not necessarily part of the pathology."

Nicholson suspects that most diseases have a metabolic signature at some level. The challenge is finding that signature. Finding the right matrix is important, whether it be urine, blood, cerebrospinal fluid, or solid tissue.

"Urine carries information on almost everything, because [the kidney is] your ultimate excretory organ, where homeostasis is maintained," Nicholson said. "There's a tremendous amount of information that can be obtained from urine, if you can analyze all the thousands of metabolites that are in there."

Metabonomics experiments are carried out by analyzing biological fluids or tissue extracts with techniques--such as nuclear magnetic resonance spectroscopy, mass spectrometry, or infrared spectroscopy--that provide many data points simultaneously. Even intact tissue samples taken during biopsies can be analyzed, using the NMR technique known as magic angle spinning.

The metabonomic profile is dominated by molecules smaller than 1,000 daltons. That molecular weight range "incorporates pretty much all energy pathways, all catabolic pathways, and many biosynthetic pathways," Nicholson told C&EN.

Nicholson and his colleagues focus on NMR measurements. The subtle differences in NMR spectra are practically impossible to identify just by visual inspection. Data mining and statistical techniques must be used to pull out what Nicholson calls "latent diagnostic information."

METABONOMICS is proving useful in drug development efforts, especially for toxicity testing. Six pharmaceutical companies are participating with the scientists at Imperial College in a group called COMET, the Consortium on Metabonomic Toxicology.

The three-year project is based in the academic lab at Imperial College. Membership fees from the companies have funded the formation of a team there, paying for NMR spectroscopy and robotic sampling equipment and the salaries of four or five postdoctoral associates. The participating companies are Pfizer, Pharmacia, Hoffmann-La Roche, Novo Nordisk, Bristol-Myers Squibb, and Eli Lilly.

"The idea is to assess metabonomics in toxicology at the stage when compounds are being tested in vivo prior to the discovery-development interface," Lindon said. "People are looking for adverse effects to try to minimize attrition in the drug discovery program later on. Once you get a drug into humans, it's very expensive if it fails."

The COMET team is performing metabonomic experiments on a range of model compounds with well-known toxic effects. The NMR spectra are being collected in a database that can be interrogated using multivariate statistical models. All animal handling and dosing is done at the pharmaceutical companies, which send samples to the group at Imperial College.

"We've spent a long time making sure that the protocols we operate to are uniform across all the companies," Lindon said. "We hammered out all the NMR and data-processing protocols that we do at Imperial College so that they can be transferred back to the companies."

That work appears to have paid off. Lindon presented data at the San Francisco meeting showing that COMET has gotten good analytical reproducibility, for instance, when a company prepares a split sample that is analyzed at both Imperial College and the company. In addition, biological variations were small compared with the effects of the toxins. "We could still look at dose response and time responses of toxins, despite the natural biological variation," Lindon said.

Over the three-year period, COMET will collect tens of thousands of NMR spectra. "We can have a very good definition of what normality is for rat or mouse urine and blood serum," Lindon said. He told C&EN that there have been some "interesting subtleties" with distinctly nonnormal and bimodal concentration distributions for some metabolites. They are investigating the significance of these findings prior to publication.

Metabonomics can have an impact in the pharmaceutical industry all the way from the early stages of drug discovery to clinical trials, notes Donald Robertson, a scientist at Pfizer's Ann Arbor, Mich., facility. The technology helps researchers incorporate safety data early in the process, even before a lead compound is selected.

"We don't see this as a check box exercise, a criterion for moving a drug forward," Robertson told C&EN. "We see it as a tool we can use to bring rational selection early on in the process, hopefully reducing attrition due to safety that comes in later on down the pipeline."

A KEY ADVANTAGE of metabonomics measurements of urine is the noninvasiveness of the approach, Robertson said. A single animal can be followed through the entire course of the toxicity study. In addition, Robertson said, metabonomic analyses can be "piggybacked" on studies with pharmacological models.

Metabonomics studies do not focus on the metabolism of the drug compound itself. In fact, large amounts of the drug or its metabolites can complicate the NMR spectrum. "This technology is measuring the physiological response to either toxicity or efficacy of a compound," Robertson said. "It measures hundreds or thousands of endogenous biomolecules. We're looking for the pattern of change of those endogenous biomolecules that is the fingerprint of toxicity or efficacy."

The move toward "pattern biomarkers"--where changes in several molecules, rather than just one molecule, constitute the biomarker--represents a "new way of thinking," Robertson said. "Are people going to be intellectually satisfied with five things going up and five things going down being the biomarker? That's something we're trying to wrestle with."

Metabonomics will accelerate the drug development process, Robertson believes, but probably not for individual compounds. "We're looking at a technology that can eliminate problematic compounds much earlier in the process," he said. "We don't waste time bringing compounds forward into traditional safety studies in toxicology."

For example, Robertson said that Pfizer is using metabonomics to test for vasculitis (inflammation of blood vessels). "Vasculitis is a tough nut for us to crack in drug development, because the only way to assess it is by taking sections of animals' vessels and seeing it," Robertson said. "With metabonomics, we can pick this up noninvasively via the urine."

Stephen G. Oliver, a professor in the School of Biological Sciences at the University of Manchester, in England, uses metabolomics to uncover gene functions. Because there are many fewer metabolites than genes or proteins, they don't have the same type of direct link to the genome that messenger RNA and proteins do.

Oliver uses a method that he calls FANCY, for functional analysis by co-responses in yeast. In FANCY, genes of known function are used to uncover the function of unknown genes. It is based on the idea that if two mutants--one with a deleted gene of known function and the other with a deleted gene of unknown function--show similar responses in their metabolic profiles, the two genes must act on the same pathway.

Oliver demonstrated the principle with separate deletions of the yeast genes PFK26 and PFK27, which encode the same enzyme, 6-phosphofructo-2-kinase. He showed that the deletions resulted in similar changes in the metabolomic profile. Extending the method to genes with unknown functions requires cluster analysis of the data from a comprehensive method such as infrared spectroscopy, mass spectrometry, or NMR spectroscopy.

Although NMR spectroscopy is the most widespread technique for metabonomics, it is not the only method people are using. "Any sort of analytical technique that gives you a lot of simultaneous measurements, which should give direct atom-specific molecular information if biomarkers are to be identified," can be used for metabonomics, Lindon said. "We are using mass spectrometry for metabonomics, and so are other people. The two techniques are complementary."

Roscher agrees that both techniques have their place. "NMR is an excellent tool for initial discovery" of biomarkers, he told C&EN. "Once you have a marker discovered, it might be more applicable to clinical practice if you shift, as we have done, to tandem mass spectrometry."

ROSCHER HAS FOCUSED on mass spectrometry because of its cost-effectiveness, he said. He is involved with a program that screens all newborns in the German state of Bavaria for metabolic disorders. "We can analyze by one technology 30 disorders simultaneously," he said.

Roscher and colleagues selected mass spectrometry for its sensitivity and for the ease of sample preparation. They analyze blood samples that have been dried on filter paper, "which can be sent all over the country by a simple envelope to a central lab."

The sensitivity is particularly important, Roscher said, because the metabolites that they are looking for occur in low concentrations in the blood. "They are at much higher abundance in the urine, but urine is very hard to use for mass screening purposes."

Unlike in other metabolomics studies that are still looking for patterns, Roscher already knows what he is looking for. "We have been focusing on specific markers indicating known diseases that are preventable by early diagnosis."

Metabolomics is more cost-effective than genomic or proteomic tests, Roscher said. However, he emphasized that he is using metabolomics as a screening method.

"We first use metabolomics as a high-efficiency screening procedure and, if we see a pathological pattern, confirm by genomics or proteomics, which are much more expensive techniques and more time-consuming," he said. "We consider the genetic or genomic techniques as excellent confirmatory procedures."

David Grainger, an investigator in the department of medicine at the University of Cambridge, is using metabonomics to help diagnose and treat coronary artery disease. Like many others in the field of metabonomics, Grainger is also collaborating with the group at Imperial College.

A variety of metabolites are associated with coronary artery disease, including low-density and high-density lipoprotein, triglycerides, and glucose. "There's a large range of markers, but they aren't useful for making individual diagnoses," Grainger said.

Grainger decided to try a variety of high-data-density approaches to search for patterns associated with coronary artery disease. He started with one-dimensional NMR, because if that was good enough it wouldn't be "worth it to move to more complex methods," he said.

"To be useful for clinical diagnosis," Grainger said, "the spectrum must contain information unique to the individual and be temporally stable." He observed temporal variation that was "not surprising" because glucose and other sugar levels tend to be more variable than other metabolites. The spectra contained "stable interperson variation," which Grainger thought was unlikely to be the result of diet or the environment. Instead, he thought that the variation could be genetic, "so that genomics and metabonomics are accessing overlapping information."

To test that suspicion, Grainger performed a metabonomics study using twins. Modeling of the data indicated that the variance in metabolic signature was not genetic. "The metabolic signature may have been set early in life or even in utero," he said. Instead, the metabonomic and genetic variance (single-nucleotide polymorphisms, for example) are likely to be orthogonal, he said.

To see if metabonomics could successfully diagnose coronary artery disease, Grainger conducted a pilot study with 100 patients. All the patients had the same symptoms, and samples were taken as the patients were being prepared for angiography.

Before a principal-components analysis of the NMR spectra would actually work, signal correction was necessary to remove the largest source of variation unrelated to heart disease. Because they had used signal correction, it was especially important to validate the predictive capability of the method to make sure they weren't overfitting the data, Grainger noted.

Grainger is now preparing to take metabonomics into a larger clinical study called MAGICAD, for metabonomic and genomic investigation of coronary artery disease. He and his colleagues at Imperial College will be comparing metabonomics and genomics. "I suspect that the combination will be better than either by itself," he said. Recruitment of 3,650 patients is expected to be completed in mid-2004, with the analysis completed in early 2005.

These examples show that the field of metabonomics is indeed constructing a bridge between the omics world and the real world. It is finding its way into the clinic where it can help diagnose disease and into drug discovery and development where it can speed the overall drug development process.

Top

GO TO

PHARMA OVERVIEW
The pharmaceutical industry has recently suffered a downturn in productivity and investor goodwill, but a number of factors remain in its favor

DRUGMAKERS PLAY IN CONTRACT FIELD
Several big-name drug companies have a lesser known business serving their competitors

FDA WALKS A FINE LINE
Agency tries to find the right balance between thoroughness and efficiency in drug approval

Chemical & Engineering News
Copyright © 2002 American Chemical Society