Discovery of novel anti-cyanobacterial allelochemicals by multi-conformational QSAR approach

Microcystis aeruginosa causes cyanobacterial harmful algal blooms (cHABs) in various freshwater environments. Due to global climate change, the cHABs have even spread to estuaries and coasts. Plant-derived flavones have been reported as allelochemicals that efficiently inhibit the growth of M. aeruginosa. Quantitative structure-activity relationship (QSAR) was applied to investigate the factors affecting the M. aeruginosa inhibitory activity of flavones, and to discover novel allelochemicals against M. aeruginosa. We constructed 2D and 3D-QSAR models based on the half maximum inhibitory concentration (IC₅₀) of 22 flavones against M. aeruginosa, using molecular descriptors from multiple stable conformations. Both models showed satisfactory performances (2D-QSAR: r²=0.899, q²=0.596, r_test²=0.801; 3D-QSAR: r²=0.810, q²=0.516, r_test²=0.897). The 2D-QSAR model indicates that the anti-cyanobacterial activity is positively correlated with minimum and maximum surface electrostatic potential, and negatively correlated with polarity index and polar surface area. Through the 3D-QSAR approach, electronegative hydroxyl groups in 5- and 4^'-position were favorable for the anti-cyanobacterial activity. In addition, we selected six untested flavones that fit the "activity-favorable" pattern of the visualized 3D-QSAR model. Five of the external flavones exhibited significant cyanobacterial inhibitory ability at their predicted IC₅₀ by the 3D-QSAR model. In particular, diosmetin achieved an inhibition rate of 70.50±4.74%, which was much higher than expected. The flavones screened by the 3D-QSAR model are novel cyanobacterial inhibitors and should be fully exploited to mitigate cHABs.