Metabolomics: Small-Molecule 'Omics

Courtesy of SRI International METABOLITES GO 'OMIC': Metabolomics researchers seek to catalog and quantify the myriad small molecules found in biological fluids. Such molecules stem from biological pathways, whose complexity and interrelationships are neatly captured in this depiction of SRI International's BioCyc database. The dots represent individual metabolites and the lines represent enzymatic reactions that interconvert the metabolites. The circle surrounding the pathways depicts the cell membrane, with embedded transporters. Every cellular component has an 'ome.' There is a genome, a proteome, a transcriptome, and a glycome to catalog DNA, protein, RNA, and sugars, respectively. Not to be outdone, the low-molecular-weight compounds that carry out much of the cell's business--lipids, carbohydrates, vitamins, and hormones, among others--now have an 'omics' to call their own, too. They are the focus of a small but growing field called metabolomics. "Biology encompasses genes, proteins, and metabolites. Any biologist without an agenda to promote will tell you that each of these are important," says Chris Becker, director of chemistry at Mountain View, Calif.-based SurroMed. The global study of these metabolites goes by the name metabolomics. Another term, metabonomics, refers specifically to studies of the changes in the metabolic profiles of higher organisms in response to stress. Metabolomics researchers generally concentrate on biofluids, including blood serum, urine, and spinal fluid. These samples already are routinely tested for metabolites, such as cholesterol and glucose, which are used as indicators of health.1 What makes metabolomics new is the attempt to tabulate and quantify all the small molecules within a sample to find new markers for disease or drug toxicity, or indicators of nutritional status.1-3 If the nascent metabolomics industry is to succeed at this goal, researchers will need to supplement existing experimental tools--chiefly nuclear magnetic resonance (NMR) and mass spectrometry (MS)--with new computational ones that can expedite and improve data analysis. That's because, like other 'omics,' the challenge in metabolomics is not so much in data collection as data analysis. SIGNS OF DISEASE One obvious use for metabolomics is searching for new disease markers.1,2 People have long recognized that levels of blood metabolites such as lipids or byproducts of cellular reactions can indicate disease, says Becker. "What's happening now that's new is that people are consistently trying to expand the horizon of what they can see." So metabolomicists scan the thousands of metabolites present in a sample, looking for ones that correlate either positively or negatively with disease, with the goal of identifying new prognostic indicators. SurroMed has started an internal project looking for biomarkers for rheumatoid arthritis. "We are finding molecules that are differentially expressed," Becker says. But SurroMed does not limit itself to metabolites; the company also searches for protein biomarkers. "We take an open-minded view," he says. "It could be a small molecule or it could be a protein that we use to identify a relevant population subset." In addition to internal projects, SurroMed works with several companies to look for disease biomarkers in neurodegenerative diseases, tumors, and autism, as well as markers of drug efficacy and safety in clinical and preclinical samples. Paradigm Genetics, located in Research Triangle Park, NC, and Beyond Genomics in Waltham, Mass., also are working both internally and in partnership with outside companies to identify metabolites and find disease biomarkers. Such information can then be licensed to other companies looking for quick ways to screen for disease risk. Paradigm recently began collaborating with researchers at Duke University Medical Center to verify drug targets and identify new biomarkers for several illnesses, starting with cardiovascular disease. Likewise, Beyond Genomics has agreements with several pharmaceutical companies to discover biomarkers for cardiovascular disease, Alzheimer disease, and cancer. Internally, the company is searching for biomarkers of atherosclerosis in a mouse model. MINIMIZING DRUG DUDS Specific metabolites are already used to pinpoint drugs with liver or kidney toxicity; for example, succinate, glycine, and dimethylamine in the blood indicate kidney damage.1 With the ability to broadly survey biological samples, it may be possible to find additional biomarkers of drug toxicity, with the long-term goal of eliminating the pharmaceutical duds from development pipelines. Paradigm Genetics is developing a database of toxicities as part of a predictive toxicology program. "The goal is to be able to know that seeing metabolites appearing at a certain time tells you something about what is happening in the body," says John E. Hamer, Paradigm's chief scientific officer and vice president of business development. Metabolomics can also highlight groups of people who respond poorly to a given drug. Researchers at Sacramento, Calif.-based Lipomics Technologies examined the lipid metabolite profile of diabetic mice taking a drug to treat type II diabetes.4 The drug, rosiglitazone (Avandia), is known to improve the body's use of insulin and lower blood lipid levels. The animals' lipid profiles, however, revealed an increase in lipid synthesis and an accumulation of lipids in the liver, which had not previously been observed. "The beauty of metabolomics is that it integrates all the genetic and environmental influences and gives you a top-level view of everything that is going on," says Steven Watkins, president of Lipomics. The group also found that one particular mouse strain had a different metabolic profile in response to rosiglitazone than other mouse strains, data which could help clarify why some people have more liver toxicity in response to the drug than others. THE BEST TOOL NMR and MS dominate metabolomics research. "There is no one magic tool that can capture the diversity of composition and concentration which is present in a single sample," says Aram Adourian, senior director of advanced technologies at Beyond Genomics. "It really depends on the question you are asking. You need to have an array of tools available." The advantage of NMR is that it can quickly and accurately identify analytes in samples that can be tested without initial processing. The weakness of NMR--and the strength of MS--is in quantifying individual components within the sample. NMR also suffers from a poorer dynamic range, says Adourian, identifying only the more abundant metabolites. Mass spectrometry, in contrast, can quantify everything from the most populous to the vanishingly rare metabolites.1 Says SurroMed's Chris Becker, "I see mass spectrometry as the way to go because of its sensitivity." At Paradigm, John Hamer views MS as the more powerful metabolomics tool, though he says NMR is sometimes useful. "With the modularity of modern mass spectrometry, you can combine novel separation and detection technologies together and really build almost a Lego-block modularity," he says. Putting a sample through gas or liquid chromatography before the MS phase, for instance, fractionates the sample for simpler spectral analysis. Until recently, however, coupling a chromatographic separation to a mass spec run resulted in a highly sensitive analysis that could take as long as an hour per sample--too long for high-throughput work. "Depending upon the metabolites, you have this interesting analytical challenge," says Ian Jardine, president of San Jose, Calif.-based mass spectrometer manufacturer Thermo Finnigan. Metabolomics researchers, he says, need high sensitivity to quantify low-abundance compounds, coupled with the ability to identify those compounds, all in a high-throughput package. Thermo Finnigan recently introduced a new time-of-flight (TOF) mass spectrometer that analyzes samples in about six minutes. "That's still not ultrahigh-throughput," says Jardine, "but it's not bad." MINING METABOLITE DATA Some instrument manufacturers, including Thermo Finnigan, Billerica, Mass.-based Bruker Daltonics, and Foster City, Calif.-based Applied Biosystems, have designed specialized software for analyzing metabolomic data generated with their instruments. Software companies have entered the arena, too. GeneLinker™ Platinum software from Molecular Mining of Kingston, Ontario, for instance, integrates genomic, proteomic, and metabolomic data. Most metabolomics companies supplement what's available with in-house software for compound identification and quantification. Often, this software integrates numerous disparate datasets and data types to provide a global overview of the cellular state. "The real challenge is how you put the pieces of information together so you can see all that data in a coherent way," says Paradigm's Hamer. "How will this fit into an Oracle database to be queried?" Beyond Genomics, among others, has developed an integrated informatics suite to deal with this problem. BioSystematics™, says the company's Web site, is "a proprietary suite of novel algorithms, hardware, and software to translate sets of gene, protein, metabolite, and other experimental data into biochemical pathway and whole-system knowledge." Often, this software is not available for purchase; instead, the companies use it as incentive for potential clients and partners. Phenomenome, a Saskatoon, Canada-based metabolomics company, for instance, offers access to its MetArray platform for MS data analysis to clients and partners, as does Paradigm Genetics with MetaVantage, and Lipomics with Surveyor. Edmonton, Canada-based Chenomx, however, does offer its Eclipse software for metabolomic NMR analysis commercially. BLOOD, SWEAT, TEARS ... AND BREATH? Though they are all 'omics,' metabolomics could hardly be more different from genomics and proteomics. Where gene and protein researchers have well-annotated databases at their disposal, metabolomics researchers have traditionally had no such resources. And while genomicists and proteomicists have only a limited set of building blocks to study--four nucleotides and 20 amino acids, respectively--metabolomicists must contend with a limitless variety of carbohydrates, amino acids, lipids, and other small molecules. Private companies are largely developing their own data storehouses from internal research or through partnerships, but a few publicly available databases exist. One of the most widely used databases is maintained by the National Institute of Standards and Technology and includes mass spectra for thousands of human metabolites. Adding to these resources is a recently announced initiative by SRI International in Menlo Park, Calif., to develop a new human metabolite database called HumanCyc. The nonprofit research institute has already developed a database called MetaCyc, which contains metabolites and metabolic pathways culled from the literature. Now SRI has developed an algorithm to predict metabolic pathways from a genomic sequence. The algorithm scans the sequence for the presence of gene homologs and uses that information to predict metabolic pathways based on pathways known to exist in other organisms. HumanCyc will include known metabolites and metabolic pathways from MetaCyc, along with those predicted by these new algorithms. "This database will emphasize information about metabolic enzymes--what reaction they catalyze and alternative substrate specificity, activators, inhibitors, and cofactors," says Peter Karp, director of the bioinformatics research group at SRI. The database will also include all the enzymes and metabolites in a given pathway, and will be able to work backwards to display all the reactions in which a given metabolite can participate. The database, being assembled by an undisclosed consortium of companies, will be freely available on the Web, but the accompanying data-analysis software will be licensed. Although HumanCyc will contain neither spectral data nor normal concentration values of the metabolites in different tissues, the database could greatly improve the amount of information that comes from each metabolic analysis, says Karp. "I think this could be of great value, because it would have a listing of all the known metabolites," he concludes. For a field still in its infancy, Paradigm's Hamer sees much potential in metabolomics. Future work could move beyond traditional biofluids like blood to include sweat, tears, or even breath, he says. "It's an exciting technology for the future." Amy Adams (amya@nasw.org) is a freelance writer in Mountain View, Calif. References 1. J.K. Nicholson et al., "Metabonomics: A platform for studying drug toxicity and gene function," Nat Rev Drug Discov, 1:153-60, 2002. 2. S.M. Watkins, J.B. German, "Metabolomics and biochemical profiling in drug discovery and development," Curr Opin Mol Ther, 4:224-8, 2002. 3. S.M. Watkins, J.B. German, "Toward the implementation of metabolomic assessments of human health and nutrition," Cur Opin Biotechnol, 13:512-6, 2002. 4. S.M. Watkins et al., "Lipid metabolome-wide effects of the PPAR* agonist rosiglitazone," J Lipid Res, 43:1809-17, 2002. NMR BOOST METABONOMICS Metabonomics is the study of how the metabolic profile of a complex biological system changes in response to stresses like disease, toxic exposure, or dietary change. Like metabolomics, metabonomics researchers often analyze biological fluids like blood or urine using nuclear magnetic resonance (NMR). "These are fluids that connect many cellular systems together, and therefore they're reporting on the integrated biochemical response of the whole organism," explains Jeremy Nicholson, Head of Biological Chemistry at the Imperial College of Science, London University. Though mass spectrometry is frequently used in metabonomics research, Nicholson says NMR has several advantages: It is non-destructive and can be used to study both solid tissue and liquid samples. Also, NMR requires minimal sample preparation and offers information on a wide range of chemical classes. NMR is the only technique that can provide information about where metabolites are located and what substances they are interacting with. "In the future, NMR is going to offer a lot more dynamic possibilities for metabolic analysis in terms of where the metabolites are what they are doing, which mass spec or any other technology will not get because you will have destroyed the sample and therefore all the information that's relevant," says Nicholson. Much of the current interest in metabonomics lies in its potential use in drug discovery and diagnostics; patterns of metabolites that appear in the NMR spectra of fluids from a patient can be indicative of a disease state or toxic response. Nicholson and coworkers, for example, recently used NMR to compare the metabolic profiles of serum or plasma from coronary heart disease patients to those of individuals with normal arteries.1 These scientists could distinguish the two groups based on differences in their NMR spectra. With further refinements of the technique, NMR could one day replace angiography as a diagnostic tool for artherosclerosis. "Based on our current models, we think you might be able to cut the use of angiography worldwide by 50%," says Nicholson. Replacing such a procedure with a simple blood test would not only place a smaller financial burden on the health care industry, but also put patients at ease. "About one in a thousand people who have angiography don't walk away from the table," says Nicholson. "It's a stressful, invasive procedure." Other areas that could benefit from metabonomics include toxicology, phenotyping, and nutritional research.2 But genomics-related fields like population genomics, functional genomics, and pharmacogenomics may see the biggest boost, says Nicholson. "It is the genomic applications of metabonomics that will unlock many hidden secrets of the genome," he says. Aileen Constans is a freelance writer in Pittman, NJ. References 1. J.T. Brindle et al., "Rapid and noninvasive diagnosis of the presence and severity of coronary artery disease using 1H-NMR based metabonomics," Nat Med, 12:1439-44, December 2002. 2. J.K. Nicholson et al., "Metabonomics: A platform for studying drug toxicity and gene function," Nat Rev Drug Disc, 1:153-61, 2002.

Metabolomics: Small-Molecule 'Omics

Interested in reading more?

Become a Member of