Tailor-Made Mass Spec
Mass spec tinkerers describe their custom fixes for commercial hardware limitations.
Once upon a time, mass spectrometers were open-platform devices that could be tweaked as new applications arose. Today’s mass specs, though, are tightly engineered black boxes: sample in, data out.
“As the level of sophistication of software and components has improved, it’s almost impossible to crack open one of these instruments and make a change, because you don’t have the schematics for all the electronics, nor access to all the control codes,” says Joshua Coon, assistant professor of chemistry and biomolecular chemistry at the University of Wisconsin, Madison. As a result, the pool of researchers capable of building mass spec components from scratch has dwindled, says Richard Smith, who has been building mass spec hardware at the Pacific Northwest National Laboratory for nearly 30 years.
For most mass spectrometrists, turnkey systems do all they could want and more. Lack of customizability is a small price to pay for the speed, accuracy, and resolution such mass analyzers provide. Sometimes, though, commercial systems simply don’t cut it; some problems fall outside their carefully calibrated parameters. Fortunately, there are those with the savvy, skills, and wherewithal to build a custom mass spec. The Scientist spoke with four researchers who, confronted with a scientific problem commercial instruments couldn’t handle, set out to devise their own solution. Here’s what they said.
RESEARCHER: Alan Marshall, Robert O. Lawton Professor of Chemistry and Biochemistry, Florida State University, and Director, Ion Cyclotron Resonance Program, National High Magnetic Field Laboratory, Tallahassee
PROJECT: Petroleomics: characterizing the molecular composition of crude petroleum
PROBLEM: Petroleum contains some 30,000 molecular species, many with similar structures and nearly identical masses. Commercial machines aren’t sensitive enough to detect and distinguish them all.
SOLUTION: A pioneer of Fourier transform-ion cyclotron resonance (FT-ICR) mass spectrometers, Marshall knew exactly what it would take to crack the crude. He coupled a Thermo Fisher Scientific LTQ linear ion trap to a custom, 14.5 Tesla magnet (290,000 times the strength of the Earth’s magnetic field) to build one of the world’s most powerful hybrid FT-ICR mass specs.
Offering unsurpassed mass resolution and accuracy, FT-ICR instruments use the orbital rotation frequency of ions inside a magnetic field to measure their masses. The bigger the magnet, the better: Resolution and speed rise linearly with field strength, while mass accuracy and dynamic range jump exponentially. Marshall’s machine can resolve masses that differ by 0.003 Da—enough to distinguish C3 (mass 36.0000) from SH4 (mass 36.0034).
Yet it’s not just about size, says Marshall. Bruker Daltonics and Thermo Fisher Scientific already offer FT-ICR machines with magnets in that range—15T and 12T, respectively. Marshall’s machine also contains a wider range of options for ionization and fragmentation, improved detection geometries and ion lenses, and custom data analysis software.
Plus, custom systems have other, more intangible benefits, he says. “If it breaks, we can fix it. And, if anything new happens, we can implement it immediately, instead of waiting to buy one.”
CONSIDERATIONS: Homemade FT-ICR construction is not for the faint of heart, says Marshall, who has six full-time staff dedicated to his instrument: 4 PhDs, a technician, and a machinist. Instead, he says, “Talk to us, we’re a resource…[Scientists] can bring their own samples, see what the system can do, and if they’re interested, they can then buy their own system.”
RESEARCHER: R. Graham Cooks, Professor, Department of Chemistry, Purdue University, West Lafayette, Ind.
PROJECT: Miniaturizing mass spectrometry
PROBLEM: How do you make a benchtop instrument portable?
SOLUTION: Cooks has spent the better part of two decades putting ion trap mass spectrometers “on a weight-loss program.” His goal: to squeeze a mass spec into a cell phone, where it can be used, for instance, to monitor health or food safety.
The mass analyzer itself is not the limitation, says Cooks; the actual trap itself measures just a few cubic centimeters. It’s the surrounding hardware—the ion lenses, vacuum system, power supply, signal amplifiers, and so on—that make the system so big. As these interconnected pieces began to shrink, the team found, new engineering possibilities emerged.
“What we experienced is a kind of Moore’s law process,” Cooks says, noting that they have reduced the system’s physical size, power requirements, and mass every 18 months or so. “If you make the RF amplifier smaller, then you can make another element, like the power supply, smaller. So it’s a cyclical process, and we were astonished how smoothly the size of the instrument dropped off as we made changes to all of the components.” Starting with a mass spec weighing several hundred kilograms, Cooks’ team has in five generations stripped the device down by two orders of magnitude; the current Mini-11 weighs a petite 4 kg, including batteries.
There’s still room for improvement, Cooks says. Yet, the device has serious power. “We can do MS/MS, MS5, we can do solids, we can do solutions, we can do proteins by electrospray [ionization]. We don’t get resolution like Alan Marshall gets, but we can do the measurements, and we can do them outside, while we walk around with the instrument.”
CONSIDERATIONS: If you’d like to get your own Mini-11, you’re too late. Cooks and his team built 10 instruments, and all already spoken for. Yet “there is a market out there,” says Cooks, “and it will get picked up commercially, I’m sure.” Each hand-built unit costs $55,000, but with component costs of only about $1,000, commercial instruments will probably cost under $5,000, Cooks estimates. “This is like a handmade Rolls Royce,” he says. In the meantime, those looking to downsize their own ion trap can head back to school: Cooks, with Purdue colleague Zheng Ouyang, is developing an analytical chemistry instrumental analysis lab/lecture course based on this project. He hopes to offer it in Fall 2010.
RESEARCHER: Joshua Coon, Assistant Professor, Departments of Chemistry and Biomolecular Chemistry, University of Wisconsin, Madison
PROJECT: Sequencing the yeast proteome
PROBLEM: Typically, proteins are identified based on just a few representative peptides; thanks to variable efficiencies of extraction, ionization, and fragmentation, that’s all the mass spec sees. Broader peptide coverage would provide richer data.
SOLUTION: Coon’s lab decided to focus on fragmentation efficiency. They recently had helped Thermo Fisher Scientific incorporate a method called electron transfer dissociation (ETD) onto its popular Orbitrap mass spectrometers. ETD is a chemical fragmentation method that works especially well for sequencing multiply charged peptides. For some peptides, though, an older method called collision-activated dissociation (CAD) provides better coverage. Unfortunately, researchers typically must choose their fragmentation method in advance, not at run-time.
“[We said], what if we allow our mass spec, every time it goes to sequence a peptide, to use the method most likely to give a positive outcome,” Coon says. “If we can predict when to use ETD and when to use CAD, we’ll get more coverage.”
Coon’s student Graeme McAlister wrote software called Decision Tree to determine, on the fly, which fragmentation method to use for each peptide. “When you would previously run your sample all with ETD and then with CAD, you’d get x number of peptides,” Coon says. “If we used Decision Tree, in one run we’d get 90% or more of the identifications we would get in two runs, but in half the time and with half the sample.”
CONSIDERATIONS: Coon, who has a sponsored research agreement with Thermo Fisher Scientific, advises prospective tinkerers to first forge a relationship with their instrument’s manufacturer. After all, they have critical assets you won’t have otherwise: the manufacturing expertise, circuit diagrams, and software code necessary to interface with the machine.
For Decision Tree, that relationship provided an additional benefit: Thermo has now incorporated the software into its Orbitrap mass spec. “For us it’s great because, as an instrument developer, it’s one thing to come up with a new method, but it’s hard to get the new ideas into others’ hands. A manufacturer has to see something you’ve done and say, ‘yes, that’s good. We’ll make it available’.”
RESEARCHER: Richard Smith, Battelle Fellow and Chief Scientist, Environmental Molecular Sciences Laboratory and Biological Sciences Division, Pacific Northwest National Laboratory, Richland, Wash.
PROJECT: Biomarker discovery in blood plasma
PROBLEM: A given sample may contain thousands of proteins varying in concentration by up to 10 orders of magnitude. Typically the lowest abundance proteins are the most interesting, yet existing platforms lack the sensitivity and dynamic range to detect them.
SOLUTION: For Smith, it’s all about sensitivity and separation. “The more you can separate molecules before they get to the mass spectrometer, the more you are going to be able to see,” he says. Working with Agilent Technologies, Smith is interfacing two components to a standard time-of-flight (TOF) mass analyzer to do just that.
The first is an ion funnel, a device Smith himself designed that concentrates ions into tight beams, thereby improving sensitivity. The second, an ion mobility separation (IMS) system, provides additional separation between the liquid chromatrography system and the mass analyzer, like “gas-phase electrophoresis,” says Smith. “You make ions of the peptides or proteins, and in the presence of a gas and an electrical field, ions drift at a velocity that depends on size and mass.” Separation occurs in about 50 msec, says Smith; combined with the TOF’s 10,000 Hz scan speed, that translates to about 500 spectra per IMS run.
By combining these components with a TOF MS, Smith says he’s built a system (now in its third generation) whose performance “is significantly better than existing platforms,” both in terms of sensitivity and throughput. For instance, ion trap systems have a fundamental limit of about 1 million ions per second. “We have been able to make measurements where we create closer to one billion ions per second. This fundamentally increases the amount of information you can get, and that pays off when we are looking at these really complex proteomics measurements in greater coverage, broader dynamic range, and detection of lower abundance species.”
CONSIDERATIONS: You don’t have to be a hardware wizard to improve proteomics mass spectrometry, says Smith. Optimized software tools, separation strategies, and ion sources can all pay dividends in resulting data, he says. “There’s so much to be done in these really challenging realms of proteomics or metabolomics, that you can almost pick any part of the flow … and find something that could be improved significantly.” Two examples: improved front-end fractionation strategies and back-end data analysis. “There’s a spectrum of steps and tools that are used, and all are in play at the moment,” he says.