MCF10A cells were infected with a retrovirus encoding mouse Twist1 or control cDNA, and after allowing 14 days for EMT reprogramming, the cells were seeded into 384-well plates and screened against a chemical library of 20,000 chemical compounds including FDA-approved drugs, kinase inhibitors, chromatin modifiers, and a set of structurally diverse compounds (enamine) with unknown activity. The screen was performed in duplicate with 3 days of drug incubation at a concentration of 5–10 μM, depending on the compound. Quantification of viable cells was performed with Celltiter Glo (Promega). The toxic cardiac glycoside sanguinarine (Sigma-Aldrich, Darmstadt, Germany) was used as a positive control and vehicle (DMSO) as a negative control. Celltiter Glo signals were normalized per plate, and hit selection was based on a z-score threshold (z-score < −2.12) derived from the distribution of positive and negative controls. Jaccard indices were calculated from the structures of the hit compounds and were then clustered using Tanimoto similarity and distance.
PKC412 (Novartis), R406, AB1010 (masitinib), OSI-930, sunitinib, and MLN518 were obtained from Selleckchem. Go6976 and LY333531 were purchased from Tocris Bioscience. BAY61-6306 was obtained from Sigma-Aldrich. For the in vivo experiments, PKC412 was kindly provided by Novartis.
PKC412 (Novartis), R406, AB1010 (masitinib), OSI-930, sunitinib, and MLN518 were obtained from Selleckchem. Go6976 and LY333531 were purchased from Tocris Bioscience. BAY61-6306 was obtained from Sigma-Aldrich. For the in vivo experiments, PKC412 was kindly provided by Novartis.