Screening for anti-inflammatory activity of Baccharis species using in silico models

Abstract

Natural products have always played a vital role in drug discovery. Asteraceae species have showed potent anti-inflammatory activity against cyclooxygenase-1 (COX-1) and 5-lipoxygenase (5-LOX) enzymes [1]. The genus Baccharis has more than 400 species and some of them present anti-inflammatory activity. This study aimed at screening extracts of Baccharis for their anti-inflammatory potential using UHPLC-HRMS and in silico models. EtOH:H2O (7:3 v/v) extracts of 223 Baccharis species and 33 other Asteraceae were prepared, their fingerprints were obtained and processed by MZmine. SIMCA-P was used for multivariate analysis and Partial Least Squares Discriminant Analysis (PLS-DA) was used to build prediction models. The model was based on extracts of 14 Asteraceae species that previously presented dual inhibition against COX and LOX and the remaining species presented one or no inhibition. The PLS model was build with two-thirds for training and one-third for testing. As a result, a matrix was obtained with 4,758 variables from the negative mode of ionization. After validation, the PLS-DA model showed statistical significance (p < 0.05, 95% confidence level) and coefficients of determination (R2) and prediction (Q2) of 0.99 and 0.65, respectively. The model was able to sort 50% correctly to Baccharis species, where a total of 100 species can present dual inhibition. Among the tested species, B. boliviensis, B. subalata and B. incarum can be cited for their dual inhibition for the first time. Therefore, the purposed model was able to filter the species with anti-inflammatory potential, and this way to avoid waste of time and money.