Mass spectroscopy and nuclear magnetic resonance go head-to-head in a battle for dominance in the world of proteomics.
Different sects of proteomics researchers have long identified themselves by their favored techniques: mass spectroscopists preach the scripture of sensitivity and speed, while NMR spectroscopists seek the revelations of reproducibility and resolution. Being scientists, though, the closest these two camps come to a religious war is an occasional inside joke.
Indeed, the choice of nuclear magnetic resonance (NMR) or mass spectrometry (MS) for protein analysis depends mostly on the type of question one wants to answer. NMR, with its ability to reveal detailed information about a protein or metabolite—even to the level of 3D atomic structure—is ideal for learning a lot about a few targets. MS, with its extraordinary sensitivity and the ability to resolve thousands of peptides simultaneously, is the perfect tool for proteome-wide studies, where the goal is to learn a little about a lot of targets.
In the past few years, researchers and gear-makers have refined both methods, with an eye toward addressing each technique’s weaknesses. Better software and hardware have improved the reproducibility of MS experiments somewhat, while super-cooled detectors have boosted the sensitivity of NMR. Nonetheless, researchers on both sides of the fence agree that with the latest technical innovations, and the convergence of different “omic” fields, the two techniques have become much more complementary than competitive.
Merge at metabolomics
Even with the latest advances, “NMR is not particularly sensitive. If you want to go after the less abundant [molecules] you have to do it by mass spec,” says John Markley, PhD, professor of biochemistry at the University of Wisconsin in Madison. In accordance with nature’s tendency to frustrate scientists at every turn, the less abundant molecules are often the most interesting.
Meanwhile, mass spectrometry’s legendary sensitivity comes with its own baggage. “Mass spec has some disadvantages, in that it’s sometimes difficult to identify what you’re looking at,” says Markley. He adds that mass spectroscopists also have a hard time quantifying the molecules they see, while NMR spectroscopists can often determine a protein or metabolite’s concentration within a few percent—if they can detect it.
To address this conundrum, researchers are increasingly combining NMR and MS in proteomic and metabolomic studies, using the strengths of each to compensate for the shortcomings of the other equipment manufacturers are also simplifying this process. At the Pittcon meeting in New Orleans in March, for example, Bruker BioSpin of Billerica, Mass., introduced its “Complete Molecular Confidence” system.
Combining both NMR and mass spectrometers, and automating the analysis with integrated software, the system lets researchers construct a complete profile of the sizes and structures of molecules in a sample. That allows comparatively small groups of researchers to take advantage of both techniques, without having to build multi-disciplinary teams and assemble expensive suites of equipment.
For the scientists pushing the field’s envelope, however, an off-the-shelf solution won’t work. “We have a big array of NMR and mass spec instrumentation,” says Jeremy Nicholson, PhD, chair of the department of biological chemistry at Imperial College in London. As one of the largest metabolomics centers in the world, Nicholson’s group has about one half-dozen large NMR spectrometers, and a similar number of mass spectrometry systems. The team develops its own algorithms and software to analyze the torrents of data from these systems.
Much of that analytical firepower is aimed at the enormously complex problem of gut microbe metabolomics. “The average human has about 1.5 kilograms of gut microbes inside them. That has a huge contribution to metabolic regulation, and in a huge variety of diseases you actually get disorders of gut microbes that complicate the metabolic profile,” says Nicholson. In order to dissect the vast tangle of interactions in that system, he and his colleagues combine not only MS and NMR, but also proteomic, metabolomic, and other -omic data into multi-dimensional analyses.
Researchers with seemingly simpler experimental challenges are also starting to combine techniques. For example, Gerhard Wagner, PhD, professor of biological chemistry at Harvard Medical School in Boston, initially set out to find small-molecule inhibitors of gene-regulatory proteins. He and his colleagues found some good inhibitors of their target molecules, but needed to know what other cellular targets those inhibitors might hit. That’s where the trouble started.
“Testing them on all the possible [protein] complexes is impossible, so we thought about looking at the entirety of the state of a cell using … analysis of metabolites,” says Wagner. In a typical experiment, the researchers grow cultured cells in the presence or absence of an experimental inhibitor, then lyse the cells and analyze their metabolites by both NMR and MS. The combined data highlight the metabolic pathways that the inhibitor affects, drastically narrowing the search for its specific protein targets.
Using the same approach, the investigators are also comparing cell lines with and without an oncoprotein, to identify pathways involved in carcinogenesis. The real application for a metabolomic cancer test is in humans, of course, but the variability of clinical samples challenges even NMR’s legendary reproducibility. “With urine or blood or serum from patients, if you get it in a state where the metabolism is ongoing, and you have to be very careful to do reproducible sample-preparation,” says Wagner, adding that “it makes a big difference if the sample is fresh or takes 20 minutes to get on ice.”
Regardless of the sample source, identifying a metabolite by NMR is much easier if the spectroscopist has a defined reference spectrum, showing the NMR peaks of that metabolite under specific buffer conditions. Fortunately, two major efforts are now building libraries of reference spectra for metabolomics.
In Canada, Genome Alberta and Genome Canada are supporting one project, which focuses on human clinical samples collected and analyzed under defined conditions. The resulting database is available freely online (https://www.hmdb.ca/). Meanwhile, the Madison Metabolomics Consortium (University of Wisconsin), the Human Metabolome Project (University of Alberta, Canada), and Bruker are assembling a complementary database of human, plant, and animal metabolites.
“We are collecting our data under conditions that are optimal for working with extracts, whereas they’re collecting under conditions that are suitable for working with biological fluids,” says Wisconsin’s Markley, who is leading the US-based effort. Like its Canadian counterpart, the Madison database is openly accessible (https://mmcd.nmrfam.wisc.edu). “It’s been very popular—over the past year, it had over 90,000 visitors,” says Markley, adding that the users hail from around the world and many have downloaded the entire database.
Taking it from the top
While the availability of the new databases is expanding the utility of NMR, mass spectrometry is undergoing its own expansion, in the form of “top-down proteomics.” Originally coined in the late-1990s, the term now common is MS labs, thanks to technological innovations and a rapidly-accumulating pile of interesting results.
Technically, a major limitation of MS in proteomics has been the need for tryptic digestion—that is, using the protease trypsin to break the proteins in a sample down into small peptides for analysis. The spectrometer can analyze the digested peptides rapidly, usually revealing enough information to identify the individual proteins in the proteome.
Unfortunately, this bottom-up approach destroys some of the most important proteomic information in the first step. “There’s not one form of a protein—different proteins have all sorts of different flavors,” says Neil Kelleher, PhD, associate professor of chemistry at the University of Illinois in Urbana. By rendering the post-translational modifications unreadable, tryptic digestion flattens these flavors.
In the top-down approach, researchers feed intact proteins into the mass spectrometer, using new techniques such as electron-capture dissociation and electron-transfer dissociation to fragment the proteins into peptides inside the spectrometer. Top-down proteomic MS can’t compete with the speed of the bottom-up approach, which is ideal for experiments that simply catalog the entire proteome of a sample. However, the top-down strategy may be the only way to study pathways governed by post-translational modifications.
“I would say the most clear example of why top-down is useful is in the area of histones and chromatin biology,” says Kelleher. In Illinois, Kelleher and his colleagues have focused on deciphering the histone code, which regulates gene expression by changing complex sets of modifications on the histones in chromatin. “So to decode it, you have to measure proteins intact and really understand how these modifications work together on the same molecule,” says Kelleher.
Though his focus makes him an unabashed fan of MS, Kelleher agrees that combining analytical techniques is the future, especially outside the basic research lab: “MS and NMR … both have been very successful in going into disease diagnosis and so forth.”
”The future is hybrid diagnostics, certainly if you’re trying to understand disease processes,” adds Nicholson. He explains that probing the vast number of parameters biologists can now measure requires top-down systems biology. In this view, diagnosis is simply a question of proper data analysis. “Disease to a systems biologist is just a change of phenotype, and a phenotype is something you can measure at the -omic level. That is the 21st century way of understanding disease,” says Nicholson.
About the Author
Originally trained as a microbiologist, Alan Dove has been writing about science and its interfaces with industry and government for more than a decade.
This article was published in Drug Discovery & Development magazine: Vol. 11, No. 4, April, 2008, pp. 40-42.
Filed Under: Genomics/Proteomics