ERROR 1
ERROR 1
ERROR 2
ERROR 2
ERROR 2
ERROR 2
ERROR 2
Password and Confirm password must match.
If you have an ACS member number, please enter it here so we can link this account to your membership. (optional)
ERROR 2
ACS values your privacy. By submitting your information, you are gaining access to C&EN and subscribing to our weekly newsletter. We use the information you provide to make your reading experience better, and we will never sell your data to third party members.
X-ray crystallography and nuclear magnetic resonance spectrometry (NMR) have long been the star players in structural biology. Now, mass spectrometry (MS) is finding its own place in the starting lineup, filling gaps left by other methods.
Crystallography provides atomic-resolution images of protein structures, but those images are snapshots of a single state frozen in time. And a protein must form crystals to get that picture, often not an easy thing to do.
NMR, in contrast, can capture dynamic information without crystals. But it requires high concentrations and doesn’t work well for aggregation-prone proteins. Plus, despite some success with large complexes, most conventional NMR is best suited to relatively small proteins.
MS avoids many of these problems. A protein need not crystallize, making MS suitable for such recalcitrant protein classes as membrane proteins and intrinsically disordered proteins. MS doesn’t have the same size and concentration restrictions as NMR. And MS can analyze complex mixtures of proteins, which is particularly important for complexes that can exist in multiple oligomeric states.
Even with these strengths, mass spectrometrists have struggled for more than a decade to gain acceptance of MS as a suitable tool for structural biology. One important question has been whether gas-phase structures have any relevance to solution-phase structures.
Oxford University chemistry professor Carol V. Robinson, a pioneer in applying mass spec to structural biology, still runs into skeptics. “Sometimes we get comments saying, ‘It’s just in the gas phase. Who’s going to believe that?’ That’s always a bit disappointing because I’ve been doing this so long.” Over the years, Robinson’s group and others have published many examples showing that a protein’s solution structure is preserved in the gas phase.
Brandon T. Ruotolo, whose group at the University of Michigan, Ann Arbor, studies multiprotein complex topology and stability using a combination of gas-phase ion-mobility separation and MS, suggests that MS is now at a stage similar to that of crystallography in the 1950s.
“Crystallography in the 1950s was struggling to assert itself as a method for structural biology. Crystallographers had to go through a lot of basic experiments to prove to people that the structure of a protein within a crystalline packing structure accurately represents what it might look like in a solution or in a cell,” Ruotolo says. Now, mass spectrometrists are “still trying to convince people that the models we come up with using our technology have relevance for how these things might look in solution,” Ruotolo says.
Nonetheless, MS is catching on. Mass spectrometrist Vicki H. Wysocki of the University of Arizona sees structural biology colleagues turning to mass spec as a frontline tool. “I have colleagues who years ago rarely looked at their complexes by mass spec. Now, it’s the first tool they use.”
Nobody is suggesting that MS will replace other structural methods. In fact, many researchers think the best understanding will come from combining information from multiple methods.
“No technique seems to work for everything,” says Justin L. P. Benesch, a mass spectrometrist at Oxford University. Modern structural biology “really is about trying to use as much information as you can glean from as many different techniques as possible.”
John R. Engen, a chemistry professor at Northeastern University, agrees. Mass spec is “most powerful when you combine it with other structural techniques,” he says. For example, mass spec can provide dynamic information that can flesh out otherwise static crystal structures. “It actually increases the value of the crystal structure,” Engen says. And in the interplay between the two techniques, interpretation of that dynamic mass spectral information becomes much easier with a known crystal structure.
“You get a beautiful view of electron density from protein crystallography,” says Gregory A. Petsko, a structural biologist at Brandeis University. “But then you have to interpret that electron density in terms of a chemical structure. That’s not always so obvious. Mass spec is great for that. It tells you what chemical species you’ve got in there.”
“Mass spec has become an indispensable tool in structural biology,” says David S. Eisenberg, a structural biologist at the University of California, Los Angeles. “We use it routinely to examine the homogeneity of protein samples and to learn the stoichiometry of complexes. It reveals posttranslational modifications of proteins that are hard to detect by other means.”
Indeed, MS is providing answers to structural biology questions that are difficult to approach with other methods. A key example is membrane proteins, which are notoriously difficult to crystallize.
For example, Robinson is using MS to study intact ATPases, membrane-associated molecular machines that use adenosine triphosphate hydrolysis to power rotary motors. Researchers have used cryoelectron microscopy (cryoEM) and X-ray crystallography to look at pieces of the complex, but they’ve been unable to see the whole structure in detail.
At first, Robinson just wanted to learn the subunit stoichiometry of these 600-kilodalton-plus complexes. With MS, she was able to provide evidence to settle a disagreement between cryoEM and crystallography about the stoichiometry of the complex’s ring. CryoEM suggested that the ring contained seven subunits, whereas crystallography suggested 10. Initially, Robinson thought both methods were wrong, because the mass spectrum seemed to suggest 11 subunits. But with an improved mass assignment algorithm, Robinson’s postdoc Nina Morgner realized that the complex contains 10 protein subunits and 10 tightly bound lipids (Science, DOI: 10.1126/science.1210148).
“I hadn’t expected to see the very specific and tight binding of lipids,” Robinson says. “When I started, I didn’t want to see lipids at all. But then I realized they were holding the key to how these things were functioning.”
In ATPase, the lipids form a plug that adapts the ring so that it can turn the machine’s rotor. Without seeing the proteins and lipids together, Robinson’s team wouldn’t have been able to figure that out, she says.
Mass spec can also help identify intermediates that other methods miss. Albert J. R. Heck, a professor at Utrecht University, in the Netherlands, studies the assembly of viral capsids using mass spec. Virologists hadn’t been able to figure out how a virus gets from its monomeric or dimeric building blocks to a complete capsid, a process so fast that virologists couldn’t find stable intermediates.
By tweaking various parameters and by combining ion mobility with MS, Heck and his coworkers found that the hepatitis B virus forms a hexamer on the way to its 180-subunit capsid. Before that, researchers had been able to see only monomers, dimers, and the intact capsid, because those high-abundance structures swamped the low-abundance intermediates. Ion-mobility spectrometry, which provides a shape-based separation, helped them show that the intermediates have a sheetlike structure (Nat. Chem., DOI: 10.1038/nchem.947).
In the case of the Triatoma virus, which infects the Chagas disease-causing Trypanosoma cruzi, its capsid consists of 60 copies each of three proteins. Using MS, Heck’s group found that the capsid is actually assembled from 12 building blocks that contain five copies of each of the proteins. In dissociation experiments, they could see two of the proteins but not the third one. The one that doesn’t dissociate at all is likely in the core of the building block, Heck says.
So far Heck has looked at relatively simple viruses that contain multiple copies of only one or a few types of capsid protein. Now he is starting to look at viruses that are more complicated such as the adenovirus, which has 12 different proteins in its capsid, some of which have 700 copies and some of which have only one.
MS can also reveal information about protein complexes that exist in many different stoichiometries at equilibrium. For example, the small heat shock protein αB-crystallin exists in about 40 states, each differing from the others by only a few percent in size. Such a complicated mixture isn’t a good candidate for conventional structural methods.
Benesch uses MS to figure out the oligomeric distribution and, by monitoring incubation with a heavier counterpart, how the oligomers interconvert. By coupling MS with ion mobility, he can also determine how the shapes change.
He manipulates conditions such as temperature and pH to mimic various stresses and then sees how the protein responds. “The protein becomes much more dynamic,” Benesch says. The monomeric subunits “hop around” faster between oligomers.
He finds that only certain geometries fit his ion mobility data, and these explain the ability of αB-crystallin to interconvert so easily. His data suggest, for example, that the 24-subunit oligomer forms an octahedron. Getting from the 24-mer to a 26-mer involves adding an edge—two subunits—to form an augmented triangular prism. Adding another edge yields a 28-mer gyrobifastigium, which consists of two triangular prisms joined at square faces with one prism rotated by 90° (Structure, DOI: 10.1016/j.str.2011.09.015).
Mass spectrometrists have a number of tricks in their playbook for acquiring structural information. For example, they can put energy into a system to make complexes fall apart into fragments that reveal information about the intact complex. The most common fragmentation method is collision-induced dissociation (CID), in which the complex crashes into gas molecules, which transfer energy and make the complex fall apart.
The problem with CID is that it provides somewhat limited information. It usually results in just the loss of an unfolded, highly charged monomer, yielding easily discernible monomer and (n–1)-mer peaks, where n is the total number of monomers. That pattern makes it easy to figure out the oligomeric state of the intact complex but doesn’t provide much other information about how the pieces fit together.
Another method, called surface-induced dissociation (SID), addresses CID’s problems. In this method, pioneered by Wysocki for more than a decade, complexes slam into a surface and break into pieces that give additional information about how they fit together.
For example, Wysocki used SID to figure out the architecture of the complex toyocamycin nitrile hydratase, a heterohexamer comprising two each of three different subunits (Anal. Chem., DOI: 10.1021/ac200452b). With CID, all she saw were two types of monomers and the associated pentamers. With SID, the same heterohexamer dissociated into trimers with one each of the three subunits.
“SID is giving us substructure we’ve never been able to see with CID,” Wysocki says. “It could have been various arrangements, but the only things that popped out were these trimers.”
With even more energy, the trimers dissociate and pop out the third kind of monomer. Because that monomer comes only from the trimer and not the hexamer, it might be more tightly bound or buried in the hexamer’s core, Wysocki says.
Wysocki thinks that SID fragments retain their compact, nativelike structure (Angew. Chem. Int. Ed., DOI: 10.1002/anie.201108700). Collision cross sections of SID fragments measured by ion mobility are approximately the same as calculated cross sections based on known crystal structures.
Another trick that mass spectrometrists use to obtain structural information is labeling proteins in various ways. These labeling strategies can help reveal dynamic changes in protein structure.
Foremost among these labeling methods is hydrogen-deuterium exchange (HDX). In this method, a protein is placed in D2O, and deuterium from the solvent exchanges with accessible hydrogen atoms in a protein’s amide backbone and some of the amino acid side chains. A hydrogen atom’s accessibility is dictated by both its location and its hydrogen-bonding state. Unprotected hydrogen atoms on the protein surface will undergo rapid exchange, whereas those that are buried or involved in holding together secondary structures such as α-helices and β-sheets will undergo sluggish HDX, says Lars Konermann, a professor of chemistry at the University of Western Ontario.
For unprotected hydrogens, the HDX reaction occurs quickly in both directions. That means that deuterium is just as quickly lost in the back-reaction. Researchers slow down the reverse reaction by cooling the solution and lowering the pH.
After the exchange reaction is quenched, proteins are digested into peptides just as they would be in a standard proteomics experiment, and the peptides are analyzed by MS. Every deuterium increases a peptide’s mass by a single mass unit, revealing how much exchange occurred.
Many people find the protocol difficult, but that needn’t be the case. “You have to work fast at zero degrees. You don’t have an hour to do chromatography at room temperature to make a beautiful separation because all of the label would be gone,” Northeastern’s Engen says. “People think that’s hard because many are used to proteomics experiments where they do overnight digestions at 37 degrees followed by a long separation gradient. You just have to get your mind into this mode: You have to go fast, and you have to be cold.”
Improvements in high-speed chromatography have made it possible to apply HDX to larger proteins (see page 16). Previously “you could only work with a 20-kDa protein,” Engen says. “Now, we can work with 350-kDa proteins.”
Other people, such as Michael C. Fitzgerald at Duke University, are turning to other labeling methods that work with conventional proteomics methods.
“We decided you need to be able to do the experiment that people always do—the bottom-up proteomics,” Fitzgerald says. Those methods “can resolve and detect hundreds and thousands of proteins in a mixture.”
Instead of conventional HDX, Fitzgerald uses covalent labeling reactions that are irreversible on the timescale of his experiments. For example, he focuses on the denaturant dependence of reactions such as oxidation of methionine and deuterium exchange of hydrogen in the histidine side chain. Unlike HDX of the amide backbone, HDX of the histidine side chain is a slow reaction with a half-life of about two days.
In the method, Fitzgerald uses denaturant to unfold the protein. The unfolding exposes previously buried sites, and those are the only sites whose labeling depends on the denaturant concentration. Because it is restricted to global unfolding reactions, the method reveals only those sites that were protected within the protein structure. “Methionine is attractive because a lot of methionine residues are buried in the hydrophobic core of a protein,” he says.
The denaturant experiment provides relatively coarse resolution, but that makes things easier from an experimental perspective. “We don’t have to get high peptide coverage of a protein,” Fitzgerald says. “We can identify just one methionine peptide, and that tells us about the whole domain from which it came.”
But methionine represents only about 2.5% of amino acids. Fitzgerald is trying to find similarly slow labeling reactions for other amino acids, which would allow him to get such information for additional proteins.
Konermann uses oxidative labeling of methionine, cysteine, and other residues to determine which parts of membrane proteins are exposed or protected. Solvent-exposed residues will be modified, whereas protected ones will not.
Konermann considers HDX and covalent labeling to be complementary methods. He used both methods to understand how glycerol facilitator, a member of the aquaporin family of channel proteins, transports glycerol and water across cell membranes (J. Mol. Biol., DOI: 10.1016/j.jmb.2011.12.052). Crystal structures have shown that glycerol interacts with specific binding sites, but those structures don’t explain why the glycerol doesn’t get stuck in the channel, Konermann says. His team found that the binding sites are actually the most dynamic parts of the protein. “There is enough structure to ensure specificity but enough dynamics to prevent glycerol from getting stuck,” he says.
Even with all these successes, sometimes the best use of MS is in partnership with other methods. Brian T. Chait, a mass spectrometrist at Rockefeller University, has long worked with structural and cell biologists to guide their use of techniques such as X-ray crystallography.
MS is used at every step of the process, Chait says. It’s used to find the best conditions for expressing and stabilizing proteins. It’s used to figure out the domains in a molecular machine. And it’s used to provide restraints for building models, with or without X-ray structures or cryoEM images.
Chait’s group has collaborated for more than a decade with cell biologist Michael P. Rout, also of Rockefeller, and computational structural biologist Andrej Sali of UC San Francisco, to figure out the workings of the nuclear pore complex. This 50-megadalton molecular machine, which is more than 10 times bigger than the ribosome, provides the only way into or out of the nucleus in eukaryotic cells. Despite its large size, the nuclear pore complex has many fewer types of proteins—only about 30—than the ribosome, which has about 80 different types of proteins.
Many of those parts occur in multiple copies in the nuclear pore complex. Because it’s so large, crystal structures have been solved only for pieces of it. Chait and his collaborators combine low-resolution techniques such as MS and cryoEM to build high-resolution models of pieces of the complex, some of which are pretty big themselves. With MS, they can find out how the pieces of a subcomplex associate with one another. Recently, they deduced a model for the structure of the 600-kDa, seven-protein complex Nup84, 16 copies of which form the outer ring of the yeast nuclear pore complex (J. Cell Biol., DOI: 10.1083/jcb.201109008).
Mass spec is getting better and better at looking at isolated protein complexes, but people ultimately want to see what these complexes look like in their native environment—the cell. Plenty of work will be needed before that’s possible though.
Nonetheless, mass spec is finding its place in the structural biology starting lineup. “Structural biology is changing to an era where you have a problem and you will study it using different techniques,” Heck says. “Mass spec will definitely be one of them.”
Join the conversation
Contact the reporter
Submit a Letter to the Editor for publication
Engage with us on X