Characterization of Aβ Monomers through the Convergence of Ensemble Properties among Simulations with Multiple Force Fields

Amyloid beta (A beta) monomers represent a base state in the pathways of aggregation that result in the fibrils and oligomers implicated in the pathogenesis of Alzheimers disease (AD). The structural properties of these intrinsically disordered peptides remain unclear despite extensive experimental and computational investigations. Further, there are mutations within A beta that change the way the peptide aggregates and are known to cause familial AD (FAD). Here, we analyze the ensembles of different isoforms (A beta 42 and A beta 40) and mutants (E22 Delta, D23N, E22K, E22G, and A2T in A beta 40) of A beta generated with all-atom replica exchange molecular dynamics (REMD) simulations on the mu s/replica time scale. These were run using three different force field/water model combinations: OPLS-AA/L and TIP3P (OPLS), AMBER99sb-ILDN and TIP4P-Ew (ILDN), as well as CHARMM22* and TIP3SP (CHARMM). Despite fundamental changes in simulation parameters, we find that the resulting ensembles demonstrate a strong convergence in structural properties. In particular, antiparallel contacts between L17-A21 and A30-L34 are prevalent in ensembles of A beta 40, directly forming beta sheets in the OPLS and ILDN combinations. A21-A30 commonly forms an interceding region that rarely interacts with the rest of the peptide. Further, A beta 42 contributes new beta hairpin motifs involving V40-I41 in both OPLS and ILDN. However, the structural flexibility of the central region and the electrostatic interactions that characterize it are notably different between the different conditions. Further, for OPLS, each of the FAD mutations disrupts central bend character and increases the polymorphism of antiparallel contacts across the central region. However, the studied mutations in the ILDN set primarily encourage more global contacts involving the N-terminus and the central region, and promote the formation of new beta topologies that may seed different aggregates involved in disease phenotypes. These differences aside, the large degree of agreement between simulation sets across multiple force fields provides a generalizable characterization of A beta that is also consistent with experimental data and models.