Multiscale simulations of protein dynamics

Characteristic timescales associated with the function of biomolecules, like proteins, range from femtoseconds up to minutes, whereas their corresponding spatial extent ranges from few ̊A to μm when associating in large macromolecular complexes. Moreover, biomolecules are functional in a large variety of different physico-chemical conditions strongly dependent on pH, ionic strength, crowding agents, etc. This huge complexity is hard to be studied with an arbitrary level of resolution embracing all these spatial and temporal scales. Molecular simulation is a well established approach to gain mechanistic insights into the function of biomolecular systems, producing atomistically detailed models of in vitro and/or in vivo conditions. I present in this thesis two projects that aim at improving on the current limitations of multiscale molecular simulations, namely (i) the sampling of large systems, and (ii) the detailed representation and description of realistic physiological conditions. Addressing the first issue, I propose a new coarse-grained model for proteins to be used in molecular dynamics simulations. This coarse-grained model is based on a more accurate description of protein electrostatics, which accounts for dipolar contributions. The parameterization of this force field is based on force-matching methods and on the use of a particle swarm optimization heuristic algorithm. The obtained results are encouraging being structural and electrostatic properties accurately reproduced with the coarse-grained model for a variety of protein folds. Moreover, the parameterization procedure can be straightforwardly applied to any protein, and can be extended to a larger dataset to generate a fully transferable coarse-grained force field, to be applied to any protein and any large macromolecular assemblies for which long all-atom simulations are still a challenge. While the development and use of coarse-grained models are important to tackle the limited sampling of large systems, it is still important to use all-atom molecular dynamics simulation to investigate with high accuracy in physiological conditions protein dynamics. For this reason, I present here the results of state-of-art molecular dynamics simulations applied to study the influence of crowding agents on the internal dynamics of the protein ubiquitin. Their analysis allows to describe how ubiquitin dynamics is slaved by crowding agents. This work demonstrates that the description of protein dynamics should take into account its intrinsic multiscale nature. The development and applications of coarse-grained models permit to simulate proteins at low computational cost, the use of atomistic simulations allows to accurately describe proteins in absence and presence of crowding agents, and both of them permit to highlight the essence of protein dynamics.

Studying membrane-protein interplay by modeling realistic conditions

Martina Audagnotto

Realistic models of cellular environments have long captured the imagination of biologists and physical scientists. More than 20 years ago, Goodshellâs inspiring rendering of cellular environments provided the foundation for the idea of capturing the full complexity of biological cells. Since then, increasingly realistic simulations of subcellular models have appeared, showing that we are rapidly moving towards comprehensive, as well as physically and biologically accurate cellular models. The major challenge in developing such models is the necessity to cover a wide range of scales. Indeed, atomistic details of molecular processes are on the length scale of 0.1 nm, while cellular dimensions range from 300 nm (for the smallest bacterial cells) up to 100 ÎŒm (for large eukaryotic cells). The biological range also encompasses the internal dynamics of individual molecules (from ns to ms) as well as entire biological processes (ms to hours). Although an atomistic representation of an entire cell is conceivable, the cost for achieving biologically significant timescales with most advanced computers is very high or nearly impossible currently. Therefore, the combination of in silico approaches with in vivo and in vitro experimental data, allows gaining further insights into the biophysics of the cell. In this thesis, a multi-scale approach was applied to investigate the interplay between proteins and their respective biological environments. In particular, two specific systems were the objects of my Ph.D. project: (i) the amyloid precursor protein (APP) and its interactions at the synaptic plasma membrane, and (ii) the human acyl-protein thioesterase I (hAPT1) and its role in protein depalmitoylation cycle across the Golgi and plasma membranes. Combining atomistic and coarse-grained molecular dynamics (MD) simulations, the stability of the dimeric conformation of APP in a realistic model of the synaptic plasma membrane was investigated. Within this approach, the stable dimeric conformation of the APP as a function of lipid membrane composition was identified, highlighting the important role of lipids unsaturation in protein stability and function. In detail, this indicated not only the preferential dimerization motif of the protein, but also the possible recruiting mode of cholesterol, which has previously been experimentally observed to correlate with Alzheimerâs disease. Furthermore, the role of the synaptic plasma membrane and its ability to form raft microdomains was explored for the interactions of APP with Î³-secretase, which specific catalytic cleavage is responsible for the production of physiological or toxic amyloid Î² peptides, responsible for the Alzheimerâs disease onset. This same multi-scale approach was applied to understand the role of S-acylation in the trafficking and function of hAPT1. In particular, the combination of experimental techniques (i.e., cell biology, biochemistry, microscopy and structural biology) with molecular modeling and simulation disclosed the substrate/enzyme interaction mode, the enzyme/membrane interface, and the role for hAPT1 for a key post-translational modification as S-palmitoylation. Based on our findings, we provided for the first time an atomistic representation of the hAPT1 enzyme mechanism, suggesting that hAPT1 is active in the monomeric form. In conclusion, these studies demonstrate once more the importance of modeling the molecular details of the biological environment at the most accu

EPFL2017

Electrostatic-Consistent Coarse-Grained Potentials for Molecular Simulations of Proteins

Davide Alemani, Michele Cascella, Matteo Dal Peraro, Matteo Thomas Degiacomi, Enrico Spiga

We present a new generation of coarse-grained (CG) potentials that account for a simplified electrostatic description of soluble proteins. The treatment of permanent electrostatic dipoles of the backbone and polar side-chains allows to simulate proteins, preserving an excellent structural and dynamic agreement with respective reference structures and all-atom molecular dynamics simulations. Moreover, multiprotein complexes can be well described maintaining their molecular interfaces thanks to the ability of this scheme to better describe the actual electrostatics at a CG level of resolution. An efficient and robust heuristic algorithm based on particle swarm optimization is used for the derivation of CG parameters via a force-matching procedure. The ability of this protocol to deal with high dimensional search spaces suggests that the extension of this optimization procedure to larger data sets may lead to the generation of a fully transferable CG force field. At the present stage, these electrostatic-consistent CG potentials are easily and efficiently parametrized, show a good degree of transferability, and can be used to simulate soluble proteins or, more interestingly, large macromolecular assemblies for which long all-atom simulations may not be easily affordable.

Amer Chemical Soc2013

Studying membrane-protein interplay by modeling realistic conditions

Martina Audagnotto

EPFL2017