Common biotech wisdom holds that a new age of medicine is around the corner and that it will be characterized by broad use of metabolomics to detect and diagnose disease. Similar statements have been made about genomics as well, but metabolomics has some advantages that will make the sentiment true sooner rather than later.
Last January, David Wishart, Ph.D., professor of biological sciences and computer sciences, University of Alberta, and his team completed the first draft of the human metabolome. They compiled about 100,000 pages of all known metabolites, their normal concentrations, where they are found, and any linkages to disease. “I think this will have a more immediate effect on the practice of medicine than the human genome project,” says Dr. Wishart.
Currently, any given clinical test may rely on about a dozen metabolites to detect or diagnose disease, even though thousands have been identified. The goal, Dr. Wishart explains, is to expand the number of metabolites charted for any diagnostic test to 50 or 100. This will provide a far more detailed view of a patient’s health status. That’s already happening, according to Dr. Wishart—newborns are screened for metabolic disorders, and scientists are looking at pneumonia to determine whether it's viral or bacterial.
Use in Drug Development
Besides it’s effect on the way we diagnose disease, metabolomics is certainly changing the way we develop drugs. Recently launched Stemina Biomarker Discovery (www.stemina.com) is combining metabolomics and stem cell research to discover and validate small molecule biomarkers for high-throughput drug screening and disease diagnostics in an all-human system.
“This is a great drug discovery and diagnostic tool,” states Elizabeth L.R. Donley, co-founder and CEO. Stemina is building upon work conduced by co-founder and CSO, Gabriela G. Cezar, Ph.D., at the University of Wisconsin. The goal is to develop biomarkers of toxicity and disease using human embryonic stem cells and derived cell types, such as neural and heart cells, as cell-based assays for a variety of diseases.
Current work focuses on identifying markers for birth defects caused by certain drugs—specifically, neuro-developmental disorders such as spina bifida and autism. The work is at its early stages, but “we’ve identified a panel of biomarkers resulting from metabolites secreted by human embryonic cells in response to exposure to known birth defect-causing drugs,” says Donley, who spoke at the Cambridge Healthtech Molecular Medicine Tri-Conference in San Francisco in late February. “We are working with serum samples from humans with neuro-developmental disabilities to identify these same biomarkers. This may be then developed from a drug screening tool into a diagnostic tool.”
Stemina also is developing its approach for cancer. So far, “the cancer research has identified a panel of biomarkers predictive of response to radiation therapy in gliomas, thus determining who may best respond to radiation therapy and allowing oncologists to monitor patients during treatment,” Donley continues. Next, Stemina plans to work on predicting radiation response in other types of cancers.
Los Angeles-based SiDMAP (www. sidmap.com), with a platform developed at UCLA, is also actively using metabolomics to aid in drug development. Laszlo G. Boros, M.D., co-founder and chief scientific advisor, explains that a nonradioactive isotope—a neutron-enriched carbon in glucose or fatty acid, in this case—allows SiDMAP to analyze several biochemical pathways simultaneously. As the tracer substrate is metabolized, the label transfers to many metabolites, allowing researchers to follow its movement in nearly 240 biochemical pathways and reactions.
“The TCA cycle, or DNA or RNA, has certain metabolites associated with it,” Dr. Boros elaborates. “The presence or absence of heavy carbons tells us about energy flow and metabolite synthesis patterns in that pathway.”
Role in Diagnostics
SiDMAP’s technology, which was described at the Molecular Medicine Tri-Conference, is used primarily for mechanism-of-action studies and determining toxicity and tumorigenesis, but it also applies to diagnostics, Dr. Boros says. The technology is being used in humans by administering labeled glucose mixed in water or a fatty acid pill to patients. After an hour or two, a breath test or a finger-prick blood sample is obtained and analyzed to identify biomarkers for a particular disease. A similar method was used to identify Helicobacter pylori bacterium via a specific 13C-labeled substrate, which is associated with more than 90% of peptic ulcers. It resulted in a Nobel Prize for Barry J. Marshall and J. Robin Warren in 2005.
Many of the advances in metabolomics are due to corresponding advances in detection and analysis equipment. Thermo Fisher Scientific (www.thermofisher.com) and Advion Biosciences (www.advion.com), reportedly, are developing more sensitive detection methods for metabolites, and Bio-Rad Laboratories (www.biorad.com) has developed what it considers the holy grail for NMR-based metabolomics.
Thermo Fisher integrates “high-speed, high-performance liquid chromatography; high mass accuracy mass spectrometry; and automated data acquisition and analysis to increase throughput, the sensitivity of detecting metabolic changes in complex samples, and ease of data interpretation,” according to Anne Ferguson, Ph.D., strategic marketing manager. She says the result enables detection of two to 10-fold changes in compound abundance in complex components of body fluids.
The goal, she elaborates, is to “robustly identify subtle differences in metabolic profiles that will promote our understanding of disease and drug safety.”
Thermo Fisher’s Accela™ high-speed chromatography system reduces processing time. “If you consider a small 20-patient study, using normal chromatography, sample processing may take four eight-hour days. The high-speed system may conservatively reduce that time by half,” Dr. Ferguson says, while maintaining high-resolution and retention time reproducibility.
To identify statistically significant differences in the metabolome that correlate to disease, Sieve™ differential expression software was used to compare the chromatographic and spectral information from sets of samples, letting users focus only on the meaningful data, thus yielding more detailed structural analyses.
The LTQ Orbitrap™ hybrid mass spectrometer, another integral component, delivers tandem mass spectrometry and MSn data, offering subfemtomole sensitivity, mass accuracy that is independent of intensity, and many ionization modes. Coupling this with the Mass Frontier™ software provides the ability to “confidently determine the identity and structure of metabolites,” Dr. Ferguson says.
“This is the first time we have used the combination of Sieve software to identify differentially expressed metabolites and Mass Frontier software to determine their structure. We are planning to seamlessly integrate the two software packages,” adds Dr. Ferguson, who detailed steps to taking an integrated approach to metabolomics during a presentation at last month’s International Symposium on Environmental Metabolomics at the University of Louisville in Kentucky.
At Advion Biosciences, Jack Henion, Ph.D., CSO and chairman, is championing an automated nanoelectrospray that can be coupled with most mass spectrometers for metabolomic studies. The TriVersa™ Nanomate, he says, is adept at detecting and identifying chemicals for qualitative and quantitative applications.
The benefits for metabolomic studies include reduced sample consumption, improved detection sensitivity, and enhanced data quality and information for proteomic studies, which lead to gains in unknown compound characterization.
This is particularly important when dealing with complex samples. “There are high levels of albumin in blood, for example, and other important biological compounds that are trillions of times lower,” points out Dr. Henion, who also spoke at the Louisville meeting. Fractionation simplifies detection, identification, and analysis of those less abundant substances by infusing the sample continuously to the mass spectrometer and averaging the signal until sufficient spectra are acquired to accurately measure mass and gain compound structure information. With signal averaging, “noise is averaged to zero, while analyte information increases,” he adds. With a dynamic range that may exceed 10,000:1, automated nanoelectrospray mass spectrometry exhibits “greater than 10 times more sensitivity than LC/MS/MS,” according to Advion.
Tests comparing automated nanoelectrospray with LC/MS showed a fourfold to fivefold throughput increase and, typically, a 100-fold reduction in required analyte quantities, so picograms of compound may be sufficient to achieve accurate measurements.
Bio-Rad Laboratories (www.bio-rad.com) developed a way to overlay all known metabolites onto scatterplots of raw spectrometer data. “This feature doesn’t exist anywhere else in the marketplace,” reports Gregory Banik, Ph.D., the Bio-Rad informatics division’s general manager who is speaking at this week’s Society of Biomolecular Sciences Conference in Montreal.
Reaching that point involved work with overlap density heatmaps and an improved binning method called IntelliBucket™; both of which are incorporated into version 7.8 of Bio-Rad’s KnowItAll® Informatics System. Overlay density heatmaps look at different spectra or chromatographs, showing areas of overlap and uniqueness and allowing users to hide common or uncommon aspects to reveal pseudospectra for analysis.
By combining overlap density heatmap consensus spectra with variable binning—the IntelliBucket™ approach—Bio-Rad was able to remove the least common data peaks and use the resulting pseudospectrum to set base widths of the bins.
Dr. Banik showed a data set with 23 sets of raw NMR data, each from a different plant sample, in which genes were knocked out, over-expressed, or wild type. They were exposed to salt water or fresh water and analyzed to identify any differences in the metabolites. “When we changed the overlay detection levels and removed some of the purple peaks to focus on those with more overlap densities, the 65,536 data points were reduced to 837 bins,” he says, and subjected to principle components analysis.
In the resulting scores plot, the samples with over-expressed genes were clearly separated from the knock-out and wild-type samples. Samples exposed to saline also showed subtle differences. Thus, the IntelliBucket “helped reveal changes in levels and types of metabolites that are present,” Dr. Banik says. Removing the bins and doing the analysis on all 65,536 points took slightly longer, showed a marked difference in placement, and eliminated much of the separation.
“The results are quite astounding,” he says. “We can see a statistical difference between knock-out, wild-type, and over-expressed plants.”
The question then is to identify those metabolites. Dr. Banik used a Loadings plot to show spectral regions of the plant samples—mixtures of many metabolites—that are positively or negatively correlated with the placement in the original PCA Scores plot. The Loadings information was then used to project a database of known metabolites onto the original Scores plot, identifying specific metabolites that could be implicated in the observed spectral differences between samples, which allowed relevant information to be pulled out immediately.
The entire process, from raw data through the final metabolite correlations (plotted with names) and access to KEGG (Kyoto Encyclopedia of Genes and Genomes) data (for purine metabolism, in this case), took a matter of minutes. Competing methods, Dr. Banik says, may take hours to days. “This is a huge step forward, and we got here with the use of binning, bucketing, and data projection.”
The next step is to develop a list of putative metabolites likely to be involved that is rank ordered by the number of implicated metabolites that co-occur in that collection.
By summer, Bio-Rad will have also launched the KnowItAllU for academic researchers, which Dr. Banik says is the largest collection of spectra in the world. It will be accessible through any web browser and will include a database of the NMR spectra of pure metabolites.