Molecular Dynamics simulations of amyloidogenic proteins. Unfolding, misfolding and aggregation.

Proteins are the main bulding blocks of biological systems. Their structure and function have been extensively studied so far both by experiments (Nuclear Magnetic Resonance, X-ray crystallography, Mass Spectrometry, etc.) and modeling strategies (Molecular Dynamics and Monte Carlo simulations, Density Functional Theory. etc.). In vivo in general and in solution in particular, they mostly adopt different and unique secondary and tertiary configurations, owing to their conformational freedom. The route and mechanism by which a specific shape is formed, i.e. the folding, which is not reversible in many cases, is not fully understood for several protein models, nothwithstanding the fulgurant advances achieved in experimental and in silico techniques in the last decades. Under specific conditions (pH, temperature, concentration, etc.), such three-dimensional arrangement as well as the intra/inter-chains interactions can be lost, and species such as disordered or fibrilar aggregates involved in several known human pathologies may develop. In this thesis we probe the atomistic scale conformational dynamics of two amyloidogenic proteins, transthyretin and β2-microglobulin, using molecular dynamics simulations. We aim at understanding the major factors driving the misfolding and/or (un)folding of the latter specified proteins, which play a precursor and prominent role in neurodegrative deseases. To this end the dynamics and dissociation of wild-type and mutant transthyretin is simulated. In particular the behaviour of a triple mutant (designed by Prof. R. Berni and coworkers to be monomeric) is studied. It comes out that the mutation considerably shifts the tetramer-folded monomer equilibrium towards the monomer, making this triple mutant a useful tool for structural and dynamical studies. The interaction of β2-microglobulin with hydrophobic surfaces is studied by molecular dynamics and the thermodynamics of the process is addressed using end-point free energy calculations. The results rationalize experimental observation reported in the literature. Protein conformational dynamics and thermodynamics are currently experimentally probed by the backbone amide hydrogen exchange experiment (HDX). The Bluu-Tramp experiment developed by prof. Esposito and coworkers allows the measurement of free energy, enthalpy and entropy of exchange in a single experiment. A proper comparison between experimental and simulation data require modeling of the process at atomic detail. Hence, we analyze also this aspect and try to relate the amide hydrogen protection observed in NMR experiments to various microscopic properties of the protein structure computed in the simulations. Using free energy calculations we aim at reproducing also the temperature dependence of the process. Given the predominant role of protein association in most biological functions, we introduce a modeling approach to estimate the entropy loss upon complex formation, a contribution which is almost always neglected in many free energy calculation methodologies due to the high dimensionality of the degrees of freedom, and adequate theoretical methods. The approach is applied to the case proteins considered in this thesis and an exact and approximate estimation of the full rotational-translational entropy are obtained in the context of nearest neighbor-based entropy formulation. Overall, this thesis explores various aspects favouring the formation of misfolded and/or (un)folded protein species, ranging from dissociation of an homotetramer of transthyretin engineered in silico, through the interaction of β2-microglobulin with an hydrophobic surface model, to the backbone amide hydrogen exchange pattern of protection of the latter. Lastly and not the least, the thesis presents a computational methodology to address the roto-translational entropy loss upon complex formation of biomolecules.