Mechanistic Insights into the Allosteric Inhibition of Androgen Receptors by Binding Function 3 Antagonists from an Integrated Molecular Modeling Study

Abstract

An androgen receptor (AR) is an intensively studied treatment target for castration-resistant prostate cancer that is irresponsive to conventional antiandrogen therapeutics. Binding function 3 (BF3) inhibitors with alternative modes of action have emerged as a promising approach to overcoming antiandrogen resistance. However, how these BF3 inhibitors modulate AR function remains elusive, hindering the development of BF3-targeting agents. Here, we performed an integrated computational study to interrogate the binding mechanism of several known BF3 inhibitors with ARs. Our results show that the inhibitory effect of the BF3 antagonists arises from their allosteric modulation of the activation function (AF2) site, which alters the dynamic coupling between the BF3 and AF2 sites as well as the AF2-coactivator (SRC2-3) interaction. Moreover, the per-residue binding energy analyses reveal the "anchor" role of the linker connecting the phenyl ring and benzimidazole/indole in these BF3 inhibitors. Furthermore, the allosteric driver-interacting residues are found to include both "positive", e.g., Phe673 and Asn833, and "negative" ones, e.g., Phe826, and the differential interactions with these residues provide an explanation why stronger binding does not necessarily result in higher inhibitory activities. Finally, our allosteric communication pathway analyses delineate how the allosteric signals triggered by BF3 binding are propagated to the AF2 pocket through multiple short-and/or long-ranged transmission pathways. Collectively, our combined computational study provides a comprehensive structural mechanism underlying how the selected set of BF3 inhibitors modulate AR function, which will help guide future development of BF3 antagonists.

Molecular View on the Impact of DHA Molecules on the Physical Properties of a Model Cell Membrane

“…In order to address the above issue, the three-component lipid bilayer DPPC/DUPC/CHOL (DPPC: dipalmitoyl-phosphatidylcholine, DUPC: dilinoleoyl-phosphatidylcholine, CHOL: cholesterol) and μs-scale coarse-grained molecular dynamics (MD) simulation were employed in the current work. The former has been widely used as a model system for lipid rafts to study the dynamics of lipids and proteins. − The latter is a powerful computational tool for revealing the molecular mechanisms of biological processes. − Besides, DHA molecules are generally negatively charged under physiological conditions and can also be protonated into neutral molecules with the help of certain proteins and local low pH. , Therefore, in this work, we performed a series of coarse-grained MD simulations to systematically study the effects of both anionic and neutral DHA molecules on the model cell membrane (lipid rafts). We firstly studied the self-assembly process of DHA molecules, and then the fusion process of DHA molecules with phase-separated lipid membranes.…”

Determination of Molecule Category of Ligands Targeting the Ligand-Binding Pocket of Nuclear Receptors with Structural Elucidation and Machine Learning

“…Although it is possible to design drugs targeting other binding sites of an NR with more accessible molecular property, for example, design of antagonists targeting the activation function 2 (AF2) , or the binding function 3 (BF3) , sites of the LBD and the surface of protein–DNA interaction of DBD , in androgen receptor (AR), currently most drugs are still designed to target the LBP of NRs, which may suffer from the basic question that whether the designed compound is an agonist or an antagonist because both types of the ligands bind to the same position of the LBP, whereas exhibiting different downstream physiological effects. To answer this question, numerous studies have been conducted to investigate the binding mechanisms of agonists and antagonists to the LBP of NRs, explaining that the difference of binding preference of the agonists and antagonists may help to stabilize different conformation of the H12 helix of an NR. ,− However, developing experimental methods for determining the category of the NR ligands is usually complicated and the proposed method may only work in certain systems, and it is also very time-consuming to determine the category of an NR ligand through investigating the conformational change of the H12 helix derived from theoretical approaches such as long-time MD simulation. − As a result, more efficient strategies should be devised for distinguishing between agonists and antagonists.…”

“…Thanks to the rapid development of computational hardware (e.g., GPU acceleration) and artificial intelligence (e.g., AlphaFold2), in silico approaches featured by high-throughput screening may provide alternative methods for accelerating the characterization of agonists and antagonists for NRs. ,− Although numerous computational attempts have been conducted in discriminating or understanding the binding preference of NR ligands, − ,− for instance, Ramaprasad et al have built a series of ligand-based machine-learning (ML) models for discriminating actives (agonists and antagonists) from inactives (chemical background or compounds with low activity on the corresponding NR) for a number of NRs, few of them can distinguish between agonists and antagonists using a unified model with structural elucidation. One possible reason may be that it is hard to establish a universal and effective model by integrating (structural) information of multiple NRs into a unified framework, such as understanding why one ligand works as an agonist in one NR target, whereas functioning as an antagonist in another.…”