Thermodynamic signatures for hexapeptides with propensity for phase separation and amyloid formation

Abstract

Twenty different amino acids can arrange in different sequences to form proteins. These proteins can either have a definite structure and perform a definite function, or they can be intrinsically disordered proteins (IDPs) that perform a variety of functions by adopting different conformations. The IDPs are also an important constituent of membraneless organelles that are formed by phase separation of proteins. Some IDPs can also aggregate and form amyloids. The occurrence of these amyloids in vivo is a hallmark of various neurodegenerative disorders. Phase separation and amyloid formation are collective behaviour properties, which may be encoded by the primary structure (sequence) of the protein. In this thesis, we investigate if there is an incipient signature of these collective behaviour properties in the potential energy landscape of monomers and dimers of IDPs. The landscapes of short hexapeptide sequences are explored here. Some of these hexapeptides are chosen to encode aromatic–aromatic and cation–aromatic interactions that are known to drive phase separation, whereas amyloid-forming and control hexapeptides are chosen from the existing databases. The potential energy landscape framework is used to estimate two properties: heat capacity (Cv) using the harmonic superposition approximation, and frustration of the landscape. The frustration arises in the landscape when low-energy structures are separated by barriers that are large relative to the thermal energy at the temperature of interest. Overall, we find that these peptides show low-temperature features (peaks or inflection points) in Cv. For phase-separating proteins, the structures with alternative side chain interactions contribute to this feature. More features may occur for peptides that contain tyrosine and arginine, which are better at promoting phase separation compared to phenylalanine and lysine with similar functional groups. The energy landscape may be more frustrated for peptides containing tyrosine and arginine. The context-dependent nature of these trends are also discussed. For monomers and dimers of amyloid-forming peptides, the structures with variable backbone conformations contribute to low-temperature Cv features whereas control peptides lack such low-energy conformations. Realising the importance of cooperative interactions in the collective behaviour properties, a further attempt is made to estimate Cv for oligomers. However, this investigation first required building an interface between the potential energy land-scape exploration softwares and large-scale atomic/molecular massively parallel simulator (LAMMPS), which can be used to implement various coarse-grained potentials. The interface was built and used to explore the dimer landscapes of hexapeptides. The melting peak in Cv occurs at higher temperatures for peptides containing residues with better phase separation propensity.