Deconvolution of ion mobility mass spectrometry arrival time distributions using a genetic algorithm approach: Application to α1-antitrypsin peptide binding

Abstract

Ion mobility mass spectrometry (IM-MS) is a fast and sample-efficient method for analysing the gas phase conformation of proteins and protein complexes. Subjecting proteins to increased collision energies prior to ion mobility separation can directly probe their unfolding behaviour. Recent work in the field has utilised this approach to evaluate the effect of small ligand binding upon protein stability, and to screen compounds for drug discovery. Its general applicability for high-throughput screening will, however, depend upon new analytical methods to make the approach scalable. Here we describe a fully automated program, called Benthesikyme, for summarising the ion mobility results from such experiments. The program automatically creates collision induced unfolding (CIU) fingerprints and summary plots that capture the increase in collision cross section and the increase in conformational flexibility of proteins during unfolding. We also describe a program, based on a genetic algorithm, for the deconvolution of arrival time distributions from the CIU data. This multicomponent analysis method was developed to require as little user input as possible. Aside from the IM-MS data, the only input required is an estimate of the number of conformational families to be fitted to the data. In cases where the appropriate number of conformational families is unclear, the automated procedure means it is straightforward to repeat the analysis for several values and optimize the quality of the fit. We have employed our new methodology to study the effects of peptide binding to α1-antitrypsin, an abundant human plasma protein whose misfolding exemplifies a group of conformational diseases termed the serpinopathies. Our analysis shows that interaction with the peptide stabilises the protein and reduces its conformational flexibility. The previously unresolved patterns of unfolding detected by the deconvolution algorithm will allow us to set up a fully automated screen for new ligand molecules with similar properties.