Abstract
A pharmacophore model, Hypo1, was built based on 21 training-set indole compounds with varying levels of antiproliferative activity. Hypo1 possessed important chemical features required for the inhibitors and demonstrated good predictive ability for biological activity with high correlation coefficient of 0.96 and 0.89 for the training-set and test-set compounds, respectively. Further utilization of Hypo1 pharmacophore model to screen chemical database in silico led to the identification of four compounds with antiproliferative activity. Among these four compounds, 43 showed potent antiproliferative activity against various cancer cell lines with the strongest inhibition on the proliferation of MCF-7 cells (IC50 ＝ 25 nM). Further biological characterization revealed that 43 effectively inhibited tubulin polymerization and significantly induced cell cycle arrest in G2-M phase. To our knowledge, 43 is the most potent antiproliferative compound with antitubulin activity discovered by computer-aided drug design. The chemical novelty of 43 and its anticancer activities made this compound as an interesting hit worthy of further lead optimization.