Advances in predicting ligand binding poses and affinities through multiscale molecular simulations

RWTH Publications (RWTH Aachen)

Abstract

In drug design campaigns, simulation approaches including Molecular Dynamics (MD) and enhanced sampling simulations are frequently-used techniques to predict the ligand’s binding poses and affinities. These are often initial, very important steps in structure-based drug design. Unfortunately, there are drug targets that may challenge conventional MD approaches, including proteins with polymetallic active sites and highly flexible polyelectrolytes (such as DNAs, RNAs). Here, I have applied in-silico methods to two of such targets, of importance for neurodegenerative diseases. The first target is one of the three human NEET proteins, called human mitoNEET protein. Human mitoNEET protein contains iron-sulfur [2Fe-2S] clusters and very shallow, solvent-exposed surfaces for ligand binding. By employing a parameterized force-field for metal ions in combination with an enhanced sampling approach - Localized Volume-based Metadynamics, developed by our lab, I predicted multiple binding poses and affinity for one ligand - furosemide (4-chloro-2-(furan-2-ylmethylamino)-5-sulfamoylbenzoic acid), compatible with experiment. The second one is the mutated messenger RNA with CAG triplet repeat expansion that also lacks well-defined ligand-binding pockets. This RNA is a target against Huntington’s disease. Here, I applied a multiscale Quantum Mechanics/Molecular Mechanics (QM/MM) MD approach using the MiMiC interface, also developed mostly in our lab - as a refinement tool for the complex structure. I focused on the GCAGCAGCUUCGGCAGCAGC RNA in complex with the ligand DB213 (1-N',4-N'-bis[3-(dimethylamino)propyl]benzene-1,4-dicarboximidamide). In contrast to force field-based MD, QM/MM MD simulations improved the pose emerging from Nuclear Overhauser Effect Spectroscopy (NOESY) experiments (which provide the structural information of the biomolecules in solution from the change in NMR resonance intensities of nuclei that are in spatial proximity to other nuclei affected by the radio frequency perturbation). This pose is also consistent with NMR chemical shifts measured by our collaborator Prof. Martin Tollinger at University of Innsbruck, Austria. The QM/MM MD approach can also capture polarization effects, which are absent in standard force fields. Finally, I studied one of the major targets for pharmaceutical intervention against the Sars-CoV-2 virus, the main protease (Mpro). My simulations provided insights into dynamics of the protein/ligand complexes and hence suggested that this homodimeric enzyme functions with negative cooperativity, which is essential for the virus resistance and adaptability. In conclusion, the computational protocols in my thesis could now be applied to other ligands, possibly helping develop new therapeutic agents.

Authors 1

1. Gia Linh Hoang corresponding Aachen

   RWTH Aachen University · Forschungszentrum Jülich

   Affiliation as printed

   RWTH Aachen

   Forschungszentrum Jülich