Multivariate Analysis of Deep Ultraviolet Resonance Raman Spectra: Secondary Structure Determination

Abstract

Studies pertaining to protein or peptide structure, dynamics or function fundamentally require a method for secondary structure measurement. There are four structurally sensitive protein vibrational modes in deep-ultraviolet resonance Raman (DUVRR) spectra, the amide I, II, III and S modes, making it a potentially powerful tool for studying secondary structure changes in even the most difficult peptide samples. Experimental studies reveal that the position and intensity of the four amide modes in UVRR spectra of proteins are correlated with the varying fractions of α-helix, β-sheet and unordered structures of proteins. This relationship has been applied to the prediction of protein secondary structure.Specifically, multivariate methods have been employed to decompose protein spectral data sets into their underlying pure secondary structure Raman spectra (PSSRS). Ideally, the resolved PSSRS could then be used for prediction of the secondary structure content of unknown proteins from their UVRR spectra. However, previous studies have not examined the performance of multivariate methods on prediction of the structural composition of an unknown test protein. A comparative study of the ability of classical least squares (CLS), principal component regression (PCR), partial least squares (PLS) and multivariate curve resolution- alternating least squares(MCR-ALS) to resolve the pure secondary structure Raman spectra (PSSRS) and predict the structural composition of unknown test proteins will be presented.