Pop quiz: What's the best molecular indicator of the status of a cell? It isn't the genome, which remains more or less unchanged during growth, development, and disease. Nor is it the proteome or the transcriptome; though protein abundance and mRNA expression do serve as molecular bellwethers of the cellular condition, these biomolecules turn over far too slowly.
When it comes to cellular status, the molecular canary-in-the-mineshaft that researchers should look to is small molecules—or metabolites—says Renato Zenobi, professor of chemistry and applied biosciences at ETH Zurich. “The metabolome is arguably the most sensitive measure of a cellular phenotype,” he and Matthias Heinemann of the University of Groningen, the Netherlands, wrote in 2010 (1). Whereas proteins and mRNA turn over in minutes to hours, Zenobi says, the metabolome “changes on the time scale of seconds.”
To measure those changes, researchers typically use nuclear magnetic resonance (NMR) or mass spectrometry (MS), both of which, generally speaking, assess the bulk abundance of metabolites in populations of cells. Increasingly, though, such population-based approaches are proving unsatisfactory.
Jonathan Sweedler, the James R. Eiszner Family Chair in Chemistry at the University of Illinois, Urbana-Champaign, studies neurotransmitters and neuropeptides. According to him, some neurons express molecules at high abundance, while others don't express them at all. Homogenizing brain tissue and analyzing it directly would produce data representing an average over all the cells in a sample. Interrogating cells one by one, on the other hand, leads to a very different picture.
Today, Sweedler can focus on those individual cells, thanks to a technique called MS imaging (MSI) that couples the exquisite sensitivity and mass accuracy of MS with the cellular (or nearly cellular) spatial resolution of imaging. MSI is not without issues; not all molecules ionize equally well, for instance, and mass spectrometers struggle to capture the less-abundant molecules in a sample. But the approach is quickly evolving, and researchers can now apply MSI to living samples in the lab or clinic, as well as in the 3-D molecular reconstruction of biological samples, among other things.
While several forms of MSI exist, all share certain key features: A tissue sample is repeatedly interrogated by an MS ionization source point by point, resulting in a multidimensional dataset in which each x-y coordinate (or pixel) is associated with a mass spectrum. Those spectra may themselves contain hundreds or even thousands of peaks, each of which can be represented graphically by a distinct color.
Richard Caprioli, the Stanley Cohen Professor of Biochemistry at the Vanderbilt University School of Medicine, likens this kind of representation to the color channels in a digital image. Just as every pixel in a digital picture can be expressed in terms of red, green, and blue channels, biological pixels in an MSI experiment can be represented in terms of discrete molecular masses.
“Let's say you are looking at a piece of tissue and there are hundreds or thousands of compounds that you can ablate with the laser,” Caprioli explains. “We can take a whole spectrum and pick out one signal and ask how that signal changes over the thousands of pixels that we make in the image.”
Of course, the technique isn't limited to a single channel; MSI data can theoretically contain thousands of discrete channels. Caprioli's team is working to identify multi-component molecular signatures of disease—collections of six or eight MS peaks whose presence or absence correlates with disease progression and prognosis. His team has already published signatures for lung, skin, and brain tumors, and is now turning its attention to prostate cancer.
Caprioli favors MALDI-based MSI, in which a tissue section is placed on a MALDI target plate, coated with a matrix material, and placed inside the mass spectrometer under vacuum. There, an ultraviolet laser raster is performed across the sample, collecting spectra at each user-defined pixel.
Two variables govern resolution in MALDI-MS: how the matrix is deposited on the sample, and the spot size of the MALDI laser. Several years ago, Caprioli's team was imaging with 50- to 80-micron pixel resolution (five cell diameters); now he says they routinely focus down to 20 microns, and sometimes even attain 2 microns (subcellular).