Molecular dynamics

MD methodology and aims in inSPIN

Molecular dynamics (MD) simulations have developed over the last decade to become the method of choice for simulating dynamical properties of proteins, often with respect to properties that are hard to study by conventional experimental techniques. In MD simulations Newtons law of motions are solved iteratively by applying molecular mechanics force fields to describe the potential energy function. However, the MD-methodology is limited to the nanosecond timescale exploring local dynamical properties as global conformational changes are rarely observed. Fortunately, recent development of software has made microsecond time-scale simulations possible on dedicated hardware; in particular the Desmond program,. Similarly, we apply various MD-biasing techniques to speed up MD-simulations closing the temporal gap between real systems and simulations, e.g.steered MD, targeted MD, accelerated MD, importance sampling MD, metadynamics, and the use of coarse-grained (CG) force fields. All of these methods are still in their infancy, and more studies have to be carried out to assess their reliability in real systems; this is one of the goals of the research projects within inSPIN, to challenge and exploit current state-of-the art biasing MD technologies on real systems. 
From the MD simulations, we will be able to provide information about the structure, function and dynamics of proteins, which is much sought after by researchers in the fields of structural biology and protein dynamics and function. This is the other goal of the MD-studies in inSPIN projects; to get a deeper insight of the properties of selected important proteins of relevance to inSPIN, as described below.

MD competence lab – infrastructure

We are large users at the linux-cluster facility, grendel, at the Center for Scientific Computing, Aarhus, www.cscaa.dk. We have comprehensive knowledge of using state-of-the art MD software programs for biomolecular simulations. This includes, NAMD, AMBER, CHARMM, GROMACS and Desmond. Locally, we have several high-end Linux-PC’s loaded with RAM and hard-disk as well as being equipped with dual monitors for easier analysis of the generated MD trajectories. We similarly have a 3D-monitor and a 3D-projector, which enables us to better visualize the interactions of interest.

Research Projects in inSPIN using MD-methodology

The use of MD simulation methods in inSPIN research projects relates so far to two protein classes, amyloid fibrils and antimicrobial peptides. As a very interesting side-project, we also study different models of the biological cell membrane.

MD simulations of amyloid protofibrils

Numerous experiments have been carried out investigating the structure and function of the native amyloid proteins and the triggers for their conversion into non-native structures from which aggregation can occur. Likewise, a lot of work has gone into analyzing the final aggregated fibrils using various experimental techniques. Using coarse-grained molecular dynamics simulations we study the steps in-between the initial and final stages of the aggregation process. Analyses of the aggregation patterns and the formation of mature fibrils, can be related to lower resolution atomic force microscopy studies. Modifications to the coarse-grained force field and methodology are necessary to simulate this complex process, and thus method development is part of the on-going effort to follow the process of fibril formation.

Initial system of small protofibrils with randomized orientations
Aggregated protofibrils after 200ns simulation

MD simulations of wild type and mutants

A variety of proteins have been shown to be cause amyloid diseases such as corneal dystrophies, Alzheimer's and Creutzfeldt-Jakob disease. Some native proteins have been revealed to spontaneously form amyloid fibrils, but more importantly mutations have been identified as dominant players in the misfolding process causing amyloid fibril formation. 
The effect of mutations has been studied through classical molecular dynamics. Both the native and the disease causing mutation have been studied in by long simulations in explicit solvent and the differences in dynamics and stability have been explored.

The MD system used in calculations. A water box surrounding the entire protein.
WT protein tertiary structure stabilized by hydrophobic core.
Difference in structure between wt protein (gray) and protein containing fatal mutation (green)

MD docking of small molecules to an NFGAILS protofibril

Amyloid diseases are characterized by the deposition of amyloid fibrils in different types of tissues in the body. Examples of amyloid diseases are Alzheimers disease, Parkinsons disease and Type II diabetes mellitus(T2DM). Each disease is associated with a specific fibrillating protein and for T2DM this protein is human Islet Amyloid Polypeptide(hIAPP).
The fibrillating core of hIAPP can be stripped down to ten residues or less, the structure of which has been determined by NMR in inSPIN with the seven residues, NFGAILS, labeled with NMR active nuclei. 

NMR-structure of the seven-residue fibril NFGAILS with acetyl and aminomethyl capping of the C- and N-terminal ends.

Different small molecules are used as imaging agents and some has potential as inhibitors of fibrillation. Studying the binding of some of these small molecules to the protofibril by molecular dynamics methods can help determine the important interactions between the fibril and the molecules, and aid in the development for new inhibitors and imaging agents.

One protofibril and two identical small molecules, and each setup is started three times with different starting positions. Each protofibril consists of two layers of ?-sheet with ten seven-residue peptides in each layer

Bicelles

Bicelle, top and side view. Red lipids: DHPC, Blue lipids: DMPC, 476 lipids, [DMPC]/[DHPC]3.2

A bicelle is lipid system consisting of two different kinds of lipids, one lipid with short aliphatic tails, and one with a longer aliphatic tail, for example DHPC and DMPC, respectively. Under certain conditions the lipid mixture will form disk like structures. These structures will most likely have the short lipids placed in the rim and the long lipids in the interior of the disk, thereby mimicking a part of a larger lipid bilayer. 
Bicelles are most commonly used in both solid and liquid state Nuclear Magnetic Resonance (NMR) experiments as a membrane bilayer mimic, in order to obtain structural data of membrane embedded peptides, which structures are inherently hard to solve without some kind of detergent. It is believed that using bicelles will give a more native like environment for the peptide under investigation, since it has a true bilayer part opposed to, for example, an SDS micelle.
We use both Coarse Grain (CG) and All-Atom (AA) Molecular Dynamics (MD) simulations to investigate both the structural properties of these systems and also how they interact with peptides. Simulations of the bicelles are carried out using the Gromacs simulation software. The CG force field we use is the MARTINI force field, and for the AA MD simulations we use the GROMOS96 force field supplied with Gromacs.

Antimicrobial peptides

Antimicrobial peptides are short peptides excreted by large range of organism and animal as part of their immune response against unwanted invading bacteria. 
The antimicrobial peptides (AMPs) exert the antimicrobial effect by inserting into the (plasma/periplasmic) membrane forming pores, disrupting the transport of essential molecules in and out of the bacteria. This will eventually lead to the demise of the bacteria. It is known that there are several hundred different kinds of peptides exerting an antimicrobial effect.
We study the interaction between certain AMPs that are thought act positively as an antibiotic and different lipid membranes and model systems of lipid membranes. These AMPs are either naturally occurring or synthesized peptides, and are thought to be able to be developed into a new kind of antibiotic.
For this purpose, both coarse grained and all-atoms Molecular dynamics simulations are used to shed light on biophysical data obtained from the experimental groups in inSPIN. All-atom simulations allow us to investigate with atomistic precision the amino acids and the lipid membrane. It also allows us to help interpret data sets obtained from Nuclear Magnetic Resonance (NMR) studies of similar systems. When using a coarse grained description of the system of interest, the atomistic detail of the simulated system is lost, but a large gain in the simulate timescale. This is useful in for investigating the properties of the peptide-lipid interaction.

Contact

Prof. PhD. Birgit Schiøtt, birgit@chem.au.dk

www.chem.au.dk/~biomodelling/

Biomodelling group