Skip to Main content Skip to Navigation

Chémoinformatique intégrative pour guider la conception des médicaments : application à la re-conception d’un inhibiteur clinique de kinase protéinique

Abstract : Despite years of intensive research and development, cancer remains one of the leading causes of death worldwide. Chemotherapy is the most commonly used treatment for cancer, as surgery and radiation therapy are often not effective in treating cancer at every location where it spreads. However,drug resistance of cancer cells to chemotherapeutic agents and/or reduction in effectiveness of a drug is the leading cause of failure of chemotherapy. Drugs are developed to bind efficiently to a given therapeutic target, called the primary target. Unfortunately, drug treatments can suffer from bindingto a secondary target that perturbs drug activity and/or impacts its metabolism. The main aim of the PhD project was to develop an integrative chemoinformatics approach to optimize drug design by studying not only the primary target but also putative secondary effects at the atomic level in order to compute more accurate binding modes and to derive better affinity estimates.The presented drug design project aims for an improved inhibitor of the serine/threonine kinase mutant BRAFV600E with simultaneous loss of binding to the secondary target PXR. Focus is on the study of both, protein kinase BRAF and nuclear receptor PXR, which is involved in regulation of xenobiotic metabolism. A machine learning tool is first developed on the well studied nuclear receptor ERα due to large amounts of experimental data, and subsequently similarly generated for BRAFV600E. Despite its recognized importance in drug metabolism, we are still lacking sufficient structural information and affinity measurements to develop machine learning models for PXR. So, an alternative approach that relies on molecular dynamics combined with the Molecular Mechanics Poisson-Boltzmann Surface Area method is employed in order to obtain a precise estimation of ligand affinities. Finally, diverse computational tools are applied to design new derivatives of the initial drug, which is too rapidly metabolized in many patients resulting in resistance and cancer relapse. The properties of the new compounds prevent activation of metabolizing enzymes that are degrading the original drug. This is expected to provide a new drug-candidate with much better pharmacokintics properties and enhanced efficacy.This thesis comprises a complete drug design pipeline and presents an integrated strategy that includes modeling, in silico design and synthesis, virtual screening, affinity predictions, in vitro tests and X-ray crystallography. The main focus is on the computational part that comprises complementary approaches from the drug’s and from the proteins’ point of view.
Document type :
Complete list of metadatas

Cited literature [420 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Thursday, February 20, 2020 - 12:59:13 PM
Last modification on : Tuesday, September 8, 2020 - 4:18:53 AM
Long-term archiving on: : Thursday, May 21, 2020 - 2:30:48 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02485659, version 1



Melanie Schneider. Chémoinformatique intégrative pour guider la conception des médicaments : application à la re-conception d’un inhibiteur clinique de kinase protéinique. Agricultural sciences. Université Montpellier, 2019. English. ⟨NNT : 2019MONTT057⟩. ⟨tel-02485659⟩



Record views


Files downloads