Phenolic acids are naturally occurring compounds that are known for their antioxidant and antiradical activity. We present experimental and theoretical studies on the antioxidant potential of the set of 22 phenolic acids with different models of hydroxylation and methoxylation of aromatic rings. Ferric reducing antioxidant power assay was used to evaluate this property. 2,3-dihydroxybenzoic acid was found to be the strongest antioxidant, while mono hydroxylated and methoxylated structures had the lowest activities. A comprehensive structure–activity investigation with density functional theory methods elucidated the influence of compounds topology, resonance stabilization, and intramolecular hydrogen bonding on the exhibited activity. The key factor was found to be a presence of two or more hydroxyl groups being located in ortho or para position to each other. Finally, the quantitative structure–activity relationship approach was used to build a multiple linear regression model describing the dependence of antioxidant activity on structure of compounds, using features exclusively related to their topology. Coefficients of determination for training set and for the test set equaled 0.9918 and 0.9993 respectively, and Q2 value for leave-one-out was 0.9716. In addition, the presented model was used to predict activities of phenolic acids that haven’t been tested here experimentally.

Due to an outbreak of COVID-19, the number of research papers devoted to in-silico drug discovery of potential antiviral drugs is increasing every day exponentially. Still, there is no specific drug to prevent or treat this novel coronavirus (SARS-CoV-2) disease. Thus, the screening for a potential remedy presents a global challenge for scientists. Up to date over a hundred crystallographic structures of SARS-CoV-2 M^pro have been deposited to Protein Data Bank. With many known proteins, the demand for a reliable target has become higher than ever, so as the choice of an efficient computational methods. Therefore, in this study comparative methods have been used for receptor-based virtual screening, targeting 9 selected structures of viral M^pro. Reliability analyses followed by re-docking of the specific co-crystallized ligand provided the best reproductivity for structures with PDB ID 6LU7, 6Y2G and 6Y2F. The influence of crystallographic water on an outcome of a virtual screening against selected targets was also investigated. Once the most reliable targets were selected, the library of easy purchasable natural compounds were retrieved from the MolPort database (10,305 compounds) and docked against the selected M^pro proteins. To ensure the efficiency of the selected compounds, binding energies for top-15 hit ligands were calculated using Molecular Mechanics as well as their absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties were predicted. Based on predicted binding energies and toxicities, top-5 compounds were selected and subjected to Molecular Dynamics simulation and found to be stable in complex to act as possible inhibitors for SARS-CoV-2. Communicated by Ramaswamy H. Sarma