Identification of FDA-Approved Oncology Drugs with Selective Potency in High-Risk Childhood Ependymoma.

Children with ependymoma (EPN) are cured in less than 50% of cases, with little improvement in outcome over the last several decades. Chemotherapy has not affected survival in EPN, due in part to a lack of preclinical models that has precluded comprehensive drug testing. We recently developed two human EPN cell lines harboring high-risk phenotypes which provided us with an opportunity to execute translational studies. EPN and other pediatric brain tumor cell lines were subject to a large-scale comparative drug screen of FDA-approved oncology drugs for rapid clinical application. The results of this in vitro study were combined with in silico prediction of drug sensitivity to identify EPN-selective compounds, which were validated by dose curve and time course modeling. Mechanisms of EPN-selective antitumor effect were further investigated using transcriptome and proteome analyses. We identified three classes of oncology drugs that showed EPN-selective antitumor effect, namely, (i) fluorinated pyrimidines (5-fluorouracil, carmofur, and floxuridine), (ii) retinoids (bexarotene, tretinoin and isotretinoin), and (iii) a subset of small-molecule multireceptor tyrosine kinase inhibitors (axitinib, imatinib, and pazopanib). Axitinib's antitumor mechanism in EPN cell lines involved inhibition of PDGFRα and PDGFRß and was associated with reduced mitosis-related gene expression and cellular senescence. The clinically available, EPN-selective oncology drugs identified by our study have the potential to critically inform design of upcoming clinical studies in EPN, in particular for those children with recurrent EPN who are in the greatest need of novel therapeutic approaches. Mol Cancer Ther; 17(9); 1984-94. ©2018 AACR.

Statistical Coupling Analysis-Guided Library Design for the Discovery of Mutant Luciferases.

Directed evolution has proven to be an invaluable tool for protein engineering; however, there is still a need for developing new approaches to continue to improve the efficiency and efficacy of these methods. Here, we demonstrate a new method for library design that applies a previously developed bioinformatic method, Statistical Coupling Analysis (SCA). SCA uses homologous enzymes to identify amino acid positions that are mutable and functionally important and engage in synergistic interactions between amino acids. We use SCA to guide a library of the protein luciferase and demonstrate that, in a single round of selection, we can identify luciferase mutants with several valuable properties. Specifically, we identify luciferase mutants that possess both red-shifted emission spectra and improved stability relative to those of the wild-type enzyme. We also identify luciferase mutants that possess a >50-fold change in specificity for modified luciferins. To understand the mutational origin of these improved mutants, we demonstrate the role of mutations at N229, S239, and G246 in altered function. These studies show that SCA can be used to guide library design and rapidly identify synergistic amino acid mutations from a small library.