Multivariate analysis of FTIR and Raman spectra of lactobacilli isolated from kefir grains

P. MOBILI, C. ARAUJO-ANDRADE, A. LONDERO, C. FRAUSTO-REYES, J. R. MARTÍNEZ-MENDOZA, R. IVANOV-TZONCHEV, G. DE ANTONI AND A. GÓMEZ-ZAVAGLIA

Palermo, Italia

Congreso; 13th European Conference of Spectroscopy on Biological Molecules; 2009

ECSBM

Modern technical developments have made Raman and FTIR spectroscopy fast, powerful and easy-to-use analytical techniques, combining detailed spectral fingerprints, non-destructivity and minimal sample preparation. Over the years, both methodologies have been developed and utilized for microbiological applications. Several research groups have already worked out on the identification procedures for bacteria using FTIR and Raman spectroscopies [1]. For these purposes, spectroscopic information is considered as mathematical data on which chemometric techniques, such as Principal Component Analysis (PCA), Partial Least Square-Discriminant Analysis (PLS-DA) among others [2] are applied, allowing discrimination and classification of bacteria at the species and even at the strain level [3]. In this work, a combined approach of vibrational spectroscopic methodologies and multivariate analysis was used for the characterization of three species of heterofermentative lactobacilli isolated from kefir grains: L. kefir, L. parakefir and L. brevis [4,5]. The microorganisms were cultured and harvested in the stationary phase and further liophylized for the registration of the FTIR and Raman spectra. A second derivative algorithm was used to enhance and differentiate the FTIR features. The Raman spectra could be analysed without any further treatment. Cluster analysis enabled strains to be grouped according to their spectral diversity. Using principal component analysis (PCA), L. kefir and non-L. kefir strains could be clearly differentiated in the principal component space when the 1700-1500 cm-1 Raman range was analysed. No clear discrimination was found when the FTIR second derivative spectra were used for the analysis. To develop a classification rule, partial least squares discriminant analysis (PLS-DA) was carried out. This method allowed the discrimination and classification of the sixteen strains under study in two groups: L. kefir and non- L. kefir. The prediction model was better for the FTIR data. The novelty of this work resides from one side, in the use of two complementary methodologies (Raman and FTIR spectroscopy) for the discrimination of heterofermentative species isolated from kefir. From the other side, multivariate analysis was used for the first time in the classification and discrimination of L. kefir from the non- L. kefir ones.