Molecular dynamics studies of intrinsically disordered peptides and proteins

A tremendous amount of evidence has accumulated in regards to the importance of intrinsically disordered proteins (IDPs) in the functioning of the cell and their role in human disease. However, understanding and modelling the physics of such proteins is one of the remaining challenges for the biophysics field. IDPs can present in a variety of forms, including flexible and extended structures, compact molten-globules, or mixtures of the two. Furthermore, many proteins which have regions with well-defined native states can have segments which are unfolded and disordered under physiological conditions. This thesis is an exploration of the physics of such IDPs, and the computational methodologies available for their study. The unfolded regions of intrinsically disordered proteins have long been described using the random coil model, which has been shown to successfully predict global properties such as the radius of gyration and intrinsic viscosities of IDPs and denatured proteins, alike. However, the two main axioms of the random coil model in regards to protein dynamics, (i) the ability of amino acid residues to sample the entire sterically allowed Ramachandran space, and (ii) the isolated pair hypothesis, which states that the conformations of residues are unaffected by nearest neighbour interactions, have been challenged through various lines of evidence. First, amino acid residues each have unique restrictions to their Ramachandran space. Second, many residues tend to have a strong bias for the pPII and beta-strand conformations. Third, the conformations of residues in protein sequences are strongly affected by nearest neighbour interactions. Part of this thesis explores the underlying causes of the distinct Ramachandran spaces of amino acid residues. In a recent experimental study of the thermodynamics of the pPII-beta equilibria of amino acid residues in GxG host-guest peptide systems (G: glycine, x: guest residue), a nearly exact enthalpy-entropy compensation at ~300 K was revealed, suggesting a common mechanism for the intrinsic conformational preferences of amino acid residues. Motivated by these results and a number of studies linking water dynamics to the strong preference for the pPII conformation over the beta-strand conformation, a rigorous molecular dynamics (MD) study with explicit water molecules of 15 GxG peptides, along with trialanine and alanine dipeptide was performed. Several hydration properties were quantified, including a novel description of water orientation near protein surfaces, and correlated with experimental pPII propensities obtained from infrared (IR), Raman, vibrational circular dichroism (VCD), and NMR spectroscopy studies. Results revealed that the distributions of water orientations are more disordered in the beta-strand conformation than in the pPII conformation, in agreement with the entropic and enthalpic stabilization of the two conformations, respectively. Furthermore, the pPII to beta hydration differences and the solvent accessible surface areas of the Cbeta group correlate with experimental pPII propensities. These results suggest that the formation of a clathrate-like hydrogen bond network around side chain groups of residues might be stabilizing the pPII conformation, and that the intrinsic pPII propensity of amino acid residues represents their side chain's ability to act as a template for such a water structure. The amyloid beta-protein (AB) is a protein involved in Alzheimer's disease (AD). It is known for aggregating into toxic oligomers, which lead to cell death in the AD brain. There are two predominant isoforms of AB, AB1-40 and AB1-42 , of which the latter is more strongly linked to AD, forms oligomers that are more toxic to cell cultures, and aggregates faster. Chapter 4 of this thesis focuses on the N-terminally truncated amyloid beta-protein isoforms, AB3-40 , AB3-42 , AB11-40 , and AB11-42 , which have been shown to exist in comparable amounts to AB1-40 and AB1-42 , in addition to having toxicity and aggregation qualities which can potentially be exacerbating factors in AD. To study the folding and oligomerization of these peptides, a discrete molecular dynamics (DMD) method combined with a four-bead protein model with the DMD4B-HYDRA force field, was used to derive the oligomer size distributions, secondary, tertiary, and quaternary structures, free energy landscapes, and assembly pathways. Results corroborate a range of experimental findings which have shown that N-terminal truncations increase the aggregation rates of AB, in addition to revealing similarities between the structures of AB3-4X (X=0, 2) and AB1-42 , suggesting a common mechanism of toxicity. The DMD4B-HYDRA approach combines DMD with the four-bead protein model and implicit solvent force field, in which amino-acid specific interactions have been implemented based on the Kyte-Doolittle hydropathy scale. It has been successfully applied to the folding and oligomerization of a variety of proteins such as Abeta, stefin B, and the non-glycosylated domains of pig gastric mucin. In the project presented in Chapter 5, the DMD4B-HYDRA force field is characterized in terms of its applicability as a sampling technique within a multi-scale approach for structure determination. For this purpose, a simple 16-residue long polyalanine peptide chain, in addition to five proteins with known native structures selected from the Protein Data Bank, are simulated, and structurally analyzed. Results reveal that DMD4B-HYDRA can be used as a good approximation of the molten-globule states of proteins and can in combination with all-atom MD be employed as an effective method to study IDPs.