Quantitation of the global secondary structure of globular proteins by FTIR spectroscopy: Comparison with X-ray crystallographic structure

Fourier transform infrared spectroscopy (FTIR) is potentially a powerful tool for determining the global secondary structure of proteins in solution, providing the spectra are analyzed using a statistically and theoretically justified methodology. We have performed FTIR experiments on 14 globular proteins and two synthetic polypeptides whose X-ray crystal structures are known to exhibit varying types and amounts of secondary structures. Calculation of the component structural elements of the vibrational bands was accomplished using nonlinear regression analysis, by fitting both the amide I and amide II bands of the Fourier self-deconvoluted spectra, the second-derivative spectra, and the original spectra. The methodology was theoretically justified by comparing (via nonlinear regression analysis) the global secondary structure determined after deconvolving into component bands the vibrational amide I envelopes with the calculated structure determined by first principles from Ramachandran analysis of the X-ray crystallographic structure of 14 proteins from the Brookhaven protein data bank. Justification of the nonlinear regression analysis model with respect to experimental and instrumental considerations was achieved by the decomposition of all the bands of benzene and an aqueous solution of ammonium acetate into component bands while floating the Gaussian/Lorentzian character of the line shapes. The results for benzene yield all pure Lorentzian line shapes with no Gaussian character while the ammonium acetate spectra yielded all Gaussian line shapes with no Lorentzian character. In addition, all-protein spectra yielded pure Gaussian line shapes with no Lorentzian character. Finally, the model was statistically justified by recognizing random deviation patterns in the regression analysis from all fits and by the extra sum of squares F-test which uses the degrees of freedom and the root mean square values as a tool to determine the optimum number of component bands required for the nonlinear regression analysis. Results from this study demonstrate that the globular secondary structure calculated from the amide I envelope for these 14 proteins from FTIR is in excellent agreement with the values calculated from the X-ray crystallographic data using three-dimensional Ramachandran analysis, providing that the proper contribution from GLN and ASN side chains to the 1667 and 1650 cm −1 component bands has been taken into account. The standard deviation of the reg