Drawing with Mass Spec

If a picture is worth a thousand words, what is its value when drawn with chemical information? Mass spectrometric imaging paints pictures in which the mass-to-charge ratios of the ions provide the "colors." It takes familiar mass spectra and turns them into pictures like those seen under a microscope. So, given a two-dimensional sample, one can tell exactly where different components of different masses are located. Especially in biological and medical applications, the technique combines the visual inspection long associated with fields such as pathology with detailed chemical information.

These pictures are created by ionizing the sample with a beam, moving the beam or the sample, and acquiring mass spectra at various spots on the sample. Pictures of the spatial distribution of specific components can be reconstructed by using characteristic mass values from the spectrum at each spot, or "pixel." Multiple pictures can be made by selecting different masses from the same set of spectra. Using smaller, more tightly focused beams achieves higher spatial resolution, but more pixels means that the image takes longer to obtain.

Most mass spectrometric imaging uses one of two methods to generate ions from the sample: matrix-assisted laser desorption ionization (MALDI) or secondary-ion mass spectrometry (SIMS). In MALDI, the sample is covered with a matrix, and a laser beam (usually ultraviolet) is used to desorb and ionize components. In SIMS, a beam of ions is shot at the sample, and the projectiles knock molecules from the surface and ionize them. The two methods are complementary: MALDI works well for relatively large molecules such as peptides and proteins, whereas SIMS is better for small molecules.

Mass spectrometric imaging can be used for medical and biological applications. MALDI imaging is used to study peptides and protein distributions and is therefore a natural fit for proteomics studies.

For example, using mass spectrometric imaging to look at human lung tumor biopsies, Richard M. Caprioli, a biochemistry professor and director of the Mass Spectrometry Research Center at Vanderbilt University, and his collaborators found that certain protein profiles can be used to assess survival, "whether a patient is going to die within months or live several years," he says [Lancet, 362, 433 (2003)]. MALDI imaging has revealed a class of patients previously unrecognized by pathologists, Caprioli says.

Jonathan V. Sweedler, a chemistry professor at the University of Illinois, Urbana-Champaign, is hoping to apply MALDI imaging to single cells. Using MALDI mass spectrometry on single cells, he has discovered neuropeptides in invertebrates. Now, he would like to incorporate the spatial information that MALDI imaging provides to make the job of isolating cells easier.

"We're trying to bring our success with single cells and small-sample mass spectrometry, which has enabled us to make a lot of discoveries, to the realm of automating cell isolation," Sweedler says. "We've discovered a lot of peptides with single-cell profiling, and we're just getting to the point that we're doing that with imaging."

Single-cell imaging may be pushing the envelope for MALDI imaging, but SIMS imaging works well at the single-cell level. Andrew G. Ewing and Nicholas Winograd, chemistry professors at Pennsylvania State University, have used SIMS imaging to study the cell membranes of the protozoan Tetrahymena thermophila during mating [Science, 305, 71 (2004)]. The membranes fuse, and a channel forms between two organisms to allow exchange of genetic material. Ewing and Winograd find that the composition of the membranes in the region where fusion pores are located is different from the composition in other areas of the membrane.

SOME PEOPLE have hypothesized that the cells synthesize the lipids to undergo conjugation, but Ewing doesn't buy it. Instead, he thinks "if you were to do a chemical analysis of the lipids in that cell, you would find that they didn't actually change; they just rearranged." Ewing believes that his group's experiment is the first to show such composition changes actually occurring.

Sweedler has started using SIMS in addition to MALDI imaging to take chemical pictures of neurons. For example, he has measured the subcellular distribution of the antioxidant vitamin E. Immunochemical methods can't work with molecules like vitamin E because that would require antibodies to vitamin E. "If any animal made an antibody to vitamin E, it would attack itself," he says. "That's where imaging comes in. There are certain types of molecules that are too ubiquitous, that are too hard to detect."

So far, mass spectrometric imaging has been restricted to a handful of labs, but pharmaceutical companies have begun to express interest. They are using MALDI imaging as a way to look at changes at the edges of diseased tissue after the application of a drug, Caprioli says. "You could load a person with a drug and it could be sky-high in their blood, but if it doesn't get to the site of injury, it isn't worth much," he explains. With this technology, "you could start to look at drug efficacy at the molecular level."

Likewise, pharmaceutical companies are interested in SIMS imaging. Winograd says he received calls from three drug companies within the space of six weeks asking him to use SIMS imaging with cluster-ion beams to investigate the diffusion of small molecules through skin. Winograd has also shown that SIMS imaging with cluster-ion beams can be used for another application of interest to drug companies: characterizing combinatorial libraries on polymer beads [J. Am. Chem. Soc., 126, 3902 (2004)]. "There's been a gradual increase in enthusiasm from pharmaceutical companies for SIMS imaging experiments, but with the talk about cluster beam sources, it's started to take off," he says.

The pictures created with mass spectrometric imaging may have low resolution, but that is adequate for many applications. "We have a number of applications where a coarse image is just fine," Caprioli says. "We have other applications where you want very high resolution, so the smaller the pixels, the better."

The decision about resolution depends on the question being asked. For example, Caprioli often deals with relatively large tissue slices that may be several square centimeters. Because each pixel is obtained separately, the higher the resolution of the image, the more it costs in time and computer storage space. Caprioli believes that only the level of resolution needed to answer a question should be used, because resolution and sensitivity are often a trade-off.

One factor limiting the spatial resolution of MALDI imaging is the matrix itself. "You can get an instrument where you can focus the laser down to a few microns. You can get an instrument where you can move the sample stage a few microns or even submicrons," Sweedler says. "But can you image a hydrophilic peptide with submicron resolution in a brain slice? No, because adding the matrix causes the peptide to move around."

Sweedler also points out that higher spatial resolution often means bumping up against sensitivity issues. If a peptide or protein is evenly distributed across a sample, then a smaller probe will desorb less material for analysis than a larger probe. "If you're looking for a rare protein, imaging is going to be difficult unless it's clustered in very high concentration groups," he says. "Imaging mass spectrometry has really only been demonstrated for fairly abundant proteins."

SIMS has some weaknesses of its own. Most SIMS imaging is done with beams of gallium or indium ions, which can destroy the surface of samples.

To cause less damage, Winograd and Ewing have recently started using newer ion guns that produce clusters of fullerene ions. "The cluster beams deposit their energy closer to the surface and more effectively desorb the molecule than do other projectiles," Winograd says. "It's also more efficient, so there's less damage induced during that dramatic collision event."

So far, SIMS imaging has been restricted to surface imaging, but Winograd and Ewing are looking for ways to do depth profiling: to move imaging into the third dimension. One way is to just break open a frozen cell and expose regions at different depths. That is not always possible because it is difficult to control how the cell breaks up. The researchers would prefer to drill through the sample using the cluster beams.

"We're finding with carbon-60 as a projectile that the molecular information is retained," Winograd says. "The key issue for us now is, what kinds of materials does it work with? What is the balance between chemical damage and molecular specificity that we can retain?"

THE SPATIAL resolution for SIMS with ion-cluster beams is currently not as good as that with atomic beams, but Winograd says that's an artifact of the way the guns are designed and not a result of the larger size of the clusters themselves. He and Ewing are working to develop a second-generation cluster beam source with a 100-nm probe.

Most mass spectrometric imaging requires rastering--that is, moving the sample relative to the beam--taking mass spectra at every point and then reconstructing the image. Ron M. A. Heeren, a mass spectrometrist at the Institute for Atomic & Molecular Physics, in Amsterdam, is trying a different approach. He defocuses the laser, takes a single image of a wider area than a focused laser can target, and lets the detector do the work [Anal. Chem., 76, 5339 (2004)]. He calls the rastering method "microprobe mode" and the single-image method "microscope mode."

"We retain the spatial distribution of the ions we generate on the surface throughout the travel of the ions through the mass spectrometer," Heeren says. "Our position sensitivity comes from a two-dimensional detector, a camera."

The microscope mode increases the speed and spatial resolution of MALDI imaging, making it more applicable to the study of a larger area, Heeren says. Microscope-mode imaging is applicable to any of the ionization methods used in mass spectrometric imaging. It can even be used with infrared MALDI, which would usually have lower spatial resolution.

"Normally with infrared MALDI, you can focus the laser down to 50 or 100 µm and take one spectrum, which limits your spatial resolution to 50 µm or so," Heeren says. "We can just let the beam be 100 µm, but within that 100 µm, we get exactly the same spatial resolution as with ultraviolet MALDI."

Sample preparation is a major challenge for all types of mass spectrometric imaging for all sample sizes, from the single cell to the tissue slice.

Ewing and Winograd prepare samples with a cryogenic method known as freeze fracture. Frozen hydrated samples are broken open to reveal different regions of the cell [Anal. Chem., 69, 2225 (1997)]. This technique has two challenges. One is figuring out which particular region in the cell corresponds to the image. This can be addressed with selective fluorescent dyes [Anal. Chem., 74, 4020 (2002)]. Another challenge is preventing the formation of ice on the sample, which masks the signal. "We've put a lot of work into sample preparation, and we're still struggling with it," Ewing says.

Much work is also needed in data processing. "It's so fast, and you get so much data, you can easily be inundated," Caprioli says. "You could produce terabytes of data in no time at all." Caprioli is working with two companies to develop software for mass spectrometric imaging.

The instrumentation needs further improvements. But when equipment designed specifically for mass spectrometric imaging becomes commercially available, Caprioli believes, the popularity of the technique will soar.