To figure out how the atoms are connected in an unknown molecule, chemists have a suite of spectroscopic techniques at their disposal. But those techniques aren’t always well suited to proteins and other large, complicated biomolecules. Whereas the spectrum of a small molecule might consist of a few discrete, informative peaks, a typical biomolecular spectrum is an impenetrable forest of overlapping spectral lines that are nearly impossible to interpret.
Decades ago, NMR spectroscopists realized that they could obtain more informative spectra by plotting the spectral signal as a function of two frequency variables rather than one. The expansion to two dimensions not only makes the peaks easier to resolve by spreading them over a larger space, it yields information not present in a one-dimensional spectrum—for example, the off-diagonal peaks can represent signals from nearby parts of the molecule. Well-designed 2D NMR experiments can resolve the structures of full proteins (see Physics Today, October 2016, page 19).
Now Marina Edelson-Averbukh of Imperial College London and her colleagues have turned the 2D approach to a different analytical method: mass spectrometry. To take a mass spectrum, researchers ionize a sample of identical molecules and send them through an electric or magnetic field to measure their mass-to-charge ratio. The ionized molecules can be broken into fragments, whose masses then also show up in the mass spectrum and reveal bits of information about the molecular structure. For example, a peak at the mass of a carbon atom plus three hydrogen atoms is good evidence that those atoms were bound together in the original molecule. But the larger and more complicated the molecule, the less useful such insights become.
Edelson-Averbukh and colleagues’ approach derives additional information from several mass spectra of the same substance. Because of statistical fluctuations in the fragmentation patterns, those spectra aren’t quite identical. But when two fragments form in the same fragmentation process, their spectral signals rise and fall together, and one can plot those correlations in a 2D mass spectrum.
For example, in the toy-model system shown in the figure—fragmentation of the peptide chain PEPTIDE, where each letter represents an amino acid—the fragments PEP+ and TIDE+ always form together; the fragments PE+ and PT+ are likewise correlated. Those correlations show up as the dark spots in the 2D spectral plot.
Correlation-based 2D mass spectrometry is an old idea, but it’s been challenging to apply to biomolecules. Not only do individual fragmentation patterns fluctuate from spectrum to spectrum, the spectrum’s total ion count does too. As a result, every fragment appears to be correlated with every other fragment, whether they’re formed in the same process or not. The 2D PEPTIDE spectrum would therefore show peaks not just at the dark spots in the figure, but also at the positions of the green circles. It would be no simpler or more informative than a 1D spectrum.
Edelson-Averbukh and colleagues developed a new data analysis technique called self-correcting partial-covariance mapping to calculate the 2D spectrum from the fluctuations of each fragment signal and the total ion count. In experimental tests on real peptides—including several 16-amino-acid fragments and the complete protein cytochrome c—they found that their technique reliably eliminates the spurious correlations and leaves only the true ones. The experimental technique, which requires nothing more than a commercial mass spectrometer, is applicable to biomolecules of all types. (T. Driver et al., Phys. Rev. X 10, 041004, 2020; credit for thumbnail illustration of cytochrome c: Vossman, CC BY-SA 3.0.)