Fast photochemical oxidation of proteins for comparing solvent-accessibility changes accompanying protein folding: data processing and application to barstar

Abstract

Mass spectrometry-based protein footprinting reveals regional and even amino-acid structural changes and fills the gap for many proteins and protein interactions that cannot be studied by X-ray crystallography or NMR spectroscopy. Hydroxyl radical-mediated labeling has proven to be particularly informative in this pursuit because many solvent-accessible residues can be labeled by OH in a protein or protein complex, thus providing more coverage than does specific amino-acid modifications. Finding all the OH-labeling sites requires LC/MS/MS analysis of a proteolyzed sample, but data processing is daunting without the help of automated software. We describe here a systematic means for achieving a comprehensive residue-resolved analysis of footprinting data in an efficient manner, utilizing software common to proteomics core laboratories. To demonstrate the method and the utility of OH-mediated labeling, we show that FPOP easily distinguishes the buried and exposed residues of barstar in its folded and unfolded states. This article is part of a Special Issue entitled: Mass spectrometry in structural biology.

Fig. 1

Peptide-resolved FPOP yields for ten tryptic peptides of barstar. White bars signify the unfolded state, and gray bars signify the folded state. Standard error bars were determined from triplicate labeling experiments for each state.

Fig. 2

Two views of the barstar crystal structure [55], with 21 residue sidechains shown in bond depiction. Red residues were significantly more modified in the unfolded state; green residues more labeled in the folded state. Blue residues, labeled in both states, do not show a statistical difference.

Fig. 3

Difference plot of all modified residues. The difference between unfolded and folded yields is divided by the maximum yield for each kind of residue. Error bars were determined by propagating the yield standard deviations from each state through this formula. Residues with values above the x-axis exhibit more labeling in the unfolded state.

Fig. 4

LC vs. high resolving power MS plots for the LC-MS/MS acquisitions of two complex samples. Plot A shows a 130 min acquisition for a proteomics sample from a biological source (unpublished data) [57]. Plot B shows an 85 min acquisition for a protein-footprinting sample, of FPOP-treated purified human apolipoprotein E3 [21].

Fig. 5

Workflow for protein footprinting, LC-MS/MS acquisition, and analysis. Gray boxes signify external software used in this study; blue boxes signify VBA software that we developed for protein footprinting.