Differentiation and prediction of the in vitro antifungal activity of Brazilian propolis against Fonsecaea pedrosoi using chemometrics and diffuse reflectance infrared Fourier Transform spectroscopy

Abstract

The in vitro antifungal activity of ethanolic extracts of propolis was evaluated at five different concentrations in twelve Fonsecaea pedrosoi strains using agar dilution. Eight propolis extracts were collected from towns located in the central region of the state of Rio Grande do Sul, Brazil. Diffuse reflectance infrared Fourier Transform spectroscopy (DRIFTS) was employed to analyze the spectra of lyophilized propolis extracts. The use of chemometric tools allowed for the development of models with more relevant factors and with fewer variables. Partial least squares (PLS) regression was applied, allowing for the development of PLS-DRIFTS models to predict antifungal activity of all fungal strains with root mean square error of cross validation (RMSECV) lower to 5.00 mm2 and coefficient correlation of cross-validation (R2) upper to 0.99 with five to eight latent variables at 4000-2400 and 2200-600 cm-1. Concluding, DRIFTS models associated with chemometrics are powerful screening tools of propolis samples with fungistatic properties against Fonsecaea pedrosoi.