Understanding the key factors that govern the rate of protein aggregation is of immense interest since protein aggregation is associated with a number of neurodegenerative diseases. Previous experimental and theoretical studies have revealed that the hydrophobicity, charge, and population of the fibril-prone monomeric state control the fibril formation rate. Because the fibril structures consist of cross beta sheets, it is widely believed that those sequences that have a high beta content (β) in the monomeric state should have high aggregation rates as the monomer can serve as a template for fibril growth. However, this important fact has never been explicitly proven, motivating us to carry out this study. Using replica exchange molecular dynamics simulation with implicit water, we have computed β of 19 mutations of amyloid beta peptide of 42 residues (Aβ42) for which the aggregation rate κ has been measured experimentally. We have found that κ depends on β in such a way that the higher the propensity to aggregation, the higher the beta content in the monomeric state. Thus, we have solved a long-standing problem of the dependence of fibril formation time of the β-structure on a quantitative level.

Alzheimer's disease is presumed to be caused by the formation of intracellular plaques of amyloid β (Aβ) peptides inside neurons. The most abundant Aβ forms are Aβ40 and Aβ42 comprising, respectively, 40 and 42 residues. Recent experiments showed that the triple Gly33Val-Val36Pro-Gly38Val (VPV) mutation causes Aβ42 to become "super-Aβ42" with elevated aggregation rates and toxicity. Upon VPV mutation, oligomerization pathways of Aβ40 become similar to those of the Aβ42 wild type. It was hypothesized that the super behavior of Aβ42 occurs due to an enhanced content of the β-turn and β-hairpin, centered at residues 36-37, and the similarity in oligomerization pathways of Aβ40-VPV and Aβ42-WT comes from the increased β-turn population. As this is based on simulation of the truncated fragments, this hypothesis may not be valid for the full-length case, motivating us to perform all-atom molecular dynamics simulations for full-length Aβ sequences. We showed that the results obtained for truncated peptides fall short in explaining the similarity of self-assembly pathways of Aβ40-VPV and Aβ42-WT. Instead, we propose that the similarity is due to not only increased β-turn population but also due to the elevated β-structure of the entire sequence. Similar to VPV, the Gly33Val-Val36Asn-Gly38Leu mutation enhances the β-structure and the C-terminal β-turn making the behavior of Aβ40 similar to that of Aβ42-WT.