Application of Asymmetric Model in Analysis of Fluorescence Spectra of Biologically Important Molecules

Abstract

Having a valid mathematical model for structureless emission band shapes is important when deconvoluting fluorescence spectra of complex molecules. We propose a new asymmetric model for emission spectra of five organic molecules containing aromatic ring: catechol, coniferyl alcohol, hydroquinone, phenylalanine and tryptophan. For each molecule, a series of emission spectra, varying in excitation wavelength, were fitted with the new model as well as with two other analytical expressions: log-normal, described previously in the literature, and sigmoid-exponential. Their deconvolution into two, three and four Gaussian components was also performed, in order to estimate the number of symmetric components needed to obtain a better fitting quality than that of the asymmetric models. Four subtypes of the new model, as well as the log-normal one, did not differ significantly in their fitting errors, while the sigmoid-exponential model showed a significantly worse fit. Spectra of two mixtures: hydroquinone-coniferyl alcohol and hydroquinone-tryptophan were deconvoluted into two asymmetric and four Gaussian components. Positions of asymmetric components of mixtures matched those of separate molecules, while Gaussian did not. Component analysis of a polymer molecule, lignin, was also performed. In this more complex case asymmetric and Gaussian components also grouped in alternating positions.

Structural Differences Between Lignin Model Polymers Synthesized from Various Monomers

“…Emission spectra of each sample were obtained by excitation at different wavelengths, starting from the excitation maximum at 360 nm, with a 5 nm-step. These were deconvoluted into a varying number of lognormal components, from three to seven, using the lognormal method [14,15,18,19] in order to find the optimal number of components for emission spectral analysis. Nonlinear fitting of all fluorescence spectra was performed using the Nelder-Mead simplex algorithm implemented in Matlab 6.5.…”

“…Therefore, different model compounds are used in order to reveal spectroscopic properties of lignin [10][11][12][13]. Further, by using fluorescence spectroscopy and appropriate mathematical methods for deconvolution of emission spectra into individual components, it is possible to get better insight into the structural characteristics of the molecule, namely to estimate the number and nature of fluorophores [14,15]. FT-IR and Raman spectroscopy are appropriate complementary methods in structural studies of complex molecules that provide data on individual bond types within the molecules.…”

Comparison of macromolecular interactions in the cell walls of hardwood, softwood and maize by fluorescence and FTIR spectroscopy, differential polarization laser scanning microscopy and X-ray diffraction

“…As result of spectral deconvolution, an approximation of the probability density (APD) for component positions on the wavelength-axis was obtained, where an APD maximum corresponds to the position of a cell wall spectral component (Fig. 6), as shown in the previous studies of mathematical analysis of fluorescence spectra (Djikanović et al 2012a,b;Donaldson et al 2010;Kalauzi et al 2007;Radotić et al 2006). The spectral components are designated I -V (Fig.…”

“…The spectral components obtained as result of spectral deconvolution in the form of an approximation of the probability density -APD (Djikanović et al 2012a,b;Donaldson et al 2010;Kalauzi et al 2007;Radotić et al 2006), are designated I -V (Fig. 6).…”

“…For each sample we found an optimal log-normal number of components (three for maize, four for maple and five for spruce). Attempts with more components did not reduce error significantly (Kalauzi et al 2007;Radotić et al 2006). …”