In silico investigation of FOXM1 binding and novel inhibitors in epithelial ovarian cancer

Objective: Using TCGA database, we had demonstrated that aberrantly activated Forkhead box M1 (FOXM1) correlates to worse overall survival in a subgroup of platinum resistant patients. Application of thiostrepton, a natural thiazole antibiotics that inhibits FOXM1 transcription activity in the clinic is hampered by difficulties in synthesis, degradation potential, and solubility. In this study, we aim to identify potential FOXM1 small molecule inhibitors to develop a new class of therapeutic agents to address the challenges in treating chemotherapy resistant EOC. Methods: We used in silico screening of compounds against a solved structure of FOXM1 and subsequently to derive a list of possible compounds that could inhibit FOXM1. Three compounds were tested for in vitro cytotoxicity and FOXM1 expression level was confirmed by RT-PCR and Western blot in EOC cell lines. Results: The FOXM1 structure obtained from 3G73 represented the DNA binding region of FOXM1 and possessed the winged helix fold representative of the Forkhead family of enzymes with two wings in direct contact with DNA. For ease of representation, we described both wings as a dimer and a single wing as a monomer. From this structure, we hypothesized two main models of how thiostrepton binding to FOXM1 could possibly curtail its transcriptional activity. In the first model thiostrepton could bind either of the wings or both wings and prevent association to DNA. In the second model thiostrepton bind the FOXM1/DNA complex and weaken association of FOXM1 to DNA. Subsequently, small molecular inhibitors could also use either of the models to inhibit transcription. To account for both models, the NCI diversity set was screened against the FOXM1 dimer: DNA complex (39 hits), dimer (11 hits) and monomer (14 hits). Those hits were further classified by chemical structure, biological function and chemical similarities to known molecules that target FOXM1. In cellular cytotoxicity assays, N-phenylphenanthren-9-amine (related to hit #225) successfully showed cytotoxicity to all three cell lines with IC50 around 1 mu M, and downregulate FOXM1 and transcription of its downstream molecules such as CCNB1. Conclusion: By a combination of in silico screening coupled to cellular cytotoxicity studies, we have taken the first step towards identifying potential inhibitors of FOXM1 that can replace thiostrepton. (C) 2015 Elsevier Ltd. All rights reserved.