Quantitative methods for structural characterization of proteins based on deep UV resonance Raman spectroscopy

Abstract

Here we report on novel quantitative approaches for protein structural characterization using deep UV resonance Raman (DUVRR) spectroscopy. Specifically, we propose a new method combining hydrogen–deuterium (HD) exchange and Bayesian source separation for extracting the DUVRR signatures of various structural elements of aggregated proteins including the cross-β core and unordered parts of amyloid fibrils. The proposed method is demonstrated using the set of DUVRR spectra of hen egg white lysozyme acquired at various stages of HD exchange. Prior information about the concentration matrix and the spectral features of the individual components was incorporated into the Bayesian equation to eliminate the ill-conditioning of the problem caused by 100% correlation of the concentration profiles of protonated and deuterated species. Secondary structure fractions obtained by partial least squares (PLS) and least squares support vector machines (LS-SVMs) were used as the initial guess for the Bayessian source separation. Advantages of the PLS and LS-SVMs methods over the classical least squares calibration (CLSC) are discussed and illustrated using the DUVRR data of the prion protein in its native and aggregated forms.