Search

link to homepage

Institute for Advanced Simulations (IAS)

Navigation and service


Current Project

Translational neuro-medicine via multiscale simulation and neuroinformatics

Neurological disorders constitute currently one of the greatest threats to human health. Disorders such as dementia, epilepsy, neurodegenerative diseases (such as Alzheimer’s and Parkinson’s diseases) and brain tumors bear an increasingly large burden for the patients and their families. For instance, Alzheimer’s is a fatal disease affecting 44 million patients worldwide, yet no cure is available. In this context, developing ligands imaging functional and dysfunctional brain states is imperative. These ligands most often target membrane proteins. Unfortunately, high-throughput structure-based ligand screening is greatly hampered by limitations in predicting affinities. These cannot be overcome by improvements of standard approaches (such as molecular docking), as the latter do not use rigorous approaches from physics (in particular statistical mechanics). When high-resolution structures of the target are available, a way of overcoming these difficulties is to combine docking procedures with extensive all-atom molecular dynamics (MD) simulations. Instead, when there is no tridimensional structural information of the target of interest, the only way to proceed is to predict protein structures, usually by bioinformatics techniques based on structural templates. Unfortunately, in several cases the sequence identity between target and template(s) is so low (<20%) that the orientation of the modeled side chains is very likely to be incorrect. This may introduce a strong bias in the prediction of the interaction between the ligand and the protein, hampering a correct description of the ligand/receptor binding event.

To solve this issue, the team from Juelich of the KeyLab developed a hybrid molecular dynamic approach in which atomistic (MM) simulations are combined with coarse grained (CG) simulations in a hybrid multiscale approach (MM/CG). Within this scheme only the binding cavity is represented with full atomistic detail (MM region) while the rest of the protein is represented with a coarse grain resolution (CG region). An interface region I is defined between MM and CG regions to bridge the two different resolution models (close-up in Fig. 1). Five walls around the GPCR are used to mimic the presence of lipid bilayer and prevent water evaporation from the atomistic region: the planar walls (φ1,2) are located at the height of the membrane lipids head and prevent water penetration through the membrane, two hemispheric walls (φ3,4) capping the extracellular and cytoplasmic ends of the protein represent the boundary of the water droplet, and the surface φ5 represents the membrane embedding the protein. Boundary potentials, defined as functions of the distance from the corresponding walls, are added to the MM/CG potential energy function. This scheme allows the description of the atomistic details of the ligand/receptor interactions without introducing unnecessary bias coming from potentially wrong orientations of the side chains far from the binding site.

MM/CG GPCRFigure 1 - MM/CG GPCR

We propose to develop an efficient, Monte Carlo based version of the MM/GC code and implement it in a Web-portal – called NDD (i.e. NeuroDrug Design) – dedicated to estimate the potency of ligands in an intuitive, scientifically sound and computationally efficient manner. Our in-house multiscale physics-based tools will be optimized for neuroreceptors and integrated in the NDD web portal in combination with modelling and ligand docking techniques.

Using this tool, and building on the expertise on all of the members of the team, the KeyLab plans to identify radioligands for the diagnostics of neuroinflammation in the brain, including those caused by Alzheimer’s and Parkinson’s diseases.

PETFigure 2 - A. This PET scanners permit imaging of radioligands or other radiotracers in awake rats [Nature Methods 8, 301–303 (2011)]. B. Schematic representation of the proposed protocol.

Specifically, we will focus on ligands targeting the human translocator membrane protein TSPO. This is a key biomarker for the diagnostics of inflammation in the brain. Expression of TSPO in the mitochondria is indeed strongly up-regulated in areas of brain injury and in neuro-inflammatory conditions, including those caused by Alzheimer’s and Parkinson’s diseases. Increased expression levels of TSPO can be monitored by positron emission tomography. Hence, radiolabeled ligands may sensitively recognize lesions and active disease processes of the brain. Moreover, they can be used to develop synthetic ligands that act as both diagnostic and therapeutic tools. Initially, we will start using standard simulation tools to study the receptor-ligand complexes, based on docking and all-atom molecular dynamics. Then, we will perform in silico high-throughput structure-based ligand screening to detect potential lead candidates. The characterization of the binding properties of the ligands will be achieved with the NDD server, relying on multiscale techniques to speed up the simulation protocol.
In the last three years, we will use NDD to detect promising radioligand candidates, along with drug leads targeting neuroreceptors, in collaboration with pharmaceutical industries in Germany, such as the Grünenthal Pharmaceutical company in Aachen. By the end of the six years we plan to create a spin-off in Vietnam for high-throughput virtual screening of ligands, interacting with pharmaceutical industries worldwide.
This project adheres to and supports the PoFIII program from Juelich “Decoding the human brain”


Servicemeu

Homepage