Investigating androgen receptor antagonism by structural analyses, molecular dynamics simulations and support vector machines

AACR Annual Meeting-- Apr 14-18, 2007; Los Angeles, CA 4426 To date, it is well established that androgen receptor (AR) is over-expressed in vast majority of the hormone-refractory prostate cancer and remains active through mechanisms that are refractory to androgen deprivation therapy. It has been found that AR must bind its ligand to confer hormone-refractory growth, indicating that AR remains to be a key therapeutic target for advanced prostate cancer. Accumulating biochemical data indicate antagonist profiles of antiandrogens could be changed by point mutations in AR ligand-binding domain (AR-LBD). For example, hydroxyflutamide is an antagonist for wild type AR, but an agonist for the T877A mutant. Despite crystal structures of a series of agonist-like AR-LBD are available, detailed structural mechanism by which T877A mutation results in antagonist-agonist conversion of hydroxyflutamide is unknown. Fortunately, given the structural and functional conservation among nuclear receptors, structural studies of the other family members could shed light on the AR antagonism. In this work, we have investigated AR antagonism by combining structural analyses, molecular dynamics (MD) simulations and support vector machines (SVM) method. First, we performed comparative structural analyses of ERα-LBD/agonist and ERα-LBD/antagonist structures. Our analyses indicated that residues D351 and L354 are critical for the antagonistic activity of ERα ligand, which were termed ‘switch residues’; Second, based on crystal structures of AR-LBD in complex with agonists containing various chemical scaffolds, we have detected a ‘soft spot’ inside the wall of AR hormone-binding pocket, which could be exploited to design novel AR antagonists with enhanced specificity and affinity; Third, by MD simulations, we showed T877A mutation result in significant change in distance between E709 and L712, which are the corresponding AR residues of ERα D351 and L354, respectively. This suggests a possible structural mechanism by which the T877A mutation results in the antagonist-agonist conversion of hydroxyflutamide; Next, to gain insights about the correlation between ligand binding modes and the helix-12 positioning, we have trained the SVM model by a dataset composed of ER/agonist, ER/antagonist and AR/agonist structures. Molecular descriptors used in the SVM include electrostatic complementarity, shape complementarity, hydrophilic interactions, surface fit rate as well as interaction between ligand and switch residues. The SVM model was subsequently used to classify a test dataset into antagonist-like or agonist-like structures solely based on the molecular descriptors. The area under the receiver operating characteristic curve of our SVM model is 0.898. These studies allow us to propose a pharmacophore model for AR antagonists, shedding light on further development of novel antiandrogens.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.