Molecular dynamics simulation and thermodynamic modeling of the self-assembly of the triterpenoids asiatic acid and madecassic acid in aqueous solution

The self-assembly behavior of the triterpenoids asiatic acid (AA) and madecassic acid (MA), both widely studied bioactive phytochemicals that are similar in structure to bile salts, were investigated in aqueous solution through atomistic-level molecular dynamics (MD) simulation. AA and MA molecules initially distributed randomly in solution were observed to aggregate into micelles during 75 ns of MD simulation. A "hydrophobic contact criterion" was developed to identify micellar aggregates from the computer simulation results. From the computer simulation data, the aggregation number of AA and MA micelles, the monomer concentration, the principal moments of the micelle radius of gyration tensor, the one-dimensional growth exhibited by AA and MA micelles as the aggregation number increases, the level of internal ordering within AA and MA micelles (quantified using two different orientational order parameters), the local environment of atoms within AA and MA in the micellar environment, and the total, hydrophilic, and hydrophobic solvent accessible surface areas of the AA and MA micelles were each evaluated. The MD simulations conducted provide insights into the self-assembly behavior of structurally complex, nontraditional surfactants in aqueous solution. Motivated by the high computational cost required to obtain an accurate estimate of the critical micelle concentrations (CMCs) of AA and MA from evaluation of the average monomer concentration present in the AA and MA simulation cells, a modified computer simulation/molecular-thermodynamic model (referred to as the MCS-MT model) was formulated to quantify the free-energy change associated with optimal AA and MA micelle formation in order to predict the CMCs of AA and MA. The predicted CMC of AA was found to be 59 microM, compared with the experimentally measured CMC of 17 microM, and the predicted CMC of MA was found to be 96 microM, compared with the experimentally measured CMC of 62 microM. The AA and MA CMCs predicted using the MCS-MT model are much more accurate than the CMCs inferred from the monomer concentrations of AA and MA present in the simulation cells after micelle self-assembly (2390 microM and 11,300 microM, respectively). The theoretical modeling results obtained for AA and MA indicate that, by combining computer simulation inputs with molecular-thermodynamic models of surfactant self-assembly, reasonably accurate estimates of surfactant CMCs can be obtained with a fraction of the computational expense that would be required by using computer simulations alone.