Molecular dynamics (MD) simulations play a vital role in amyloid aggregation study that is crucial to the therapeutic agent development for Alzheimer’s Disease. The accuracy of MD simulation results strongly depends on the force field (FF) employed. In this work, we assess the applicability of 18 widely used atomistic molecular mechanics force fields (FFs) on amyloid aggregation using Aβ_16–22and VQIVYK (PHF6) peptides which are essential for the aggregation of Alzheimer’s Aβ peptides and tau-protein. Although, both Aβ_16–22 and PHF6 peptides formed fibrils in vitro, the predominated forms of PHF6 fibrils are parallel β-sheets, while the Aβ_16–22fibrils are antiparallel β-sheets. We performed large-scale all-atom MD simulations in explicit water on dimer and oct-peptides systems for a total simulation time of 0.54 ms. We compared the FFs on the structural sampling of the amyloid peptides. Importantly, we benchmarked the FFs on the kinetics of the amyloid oligomer formation, which is rarely considered in previous FF comparisons. For the case of Aβ_16–22, our result showed that (i) the AMBER94, AMBER99, and AMBER12SB do not predict any β-sheets; (ii) AMBER96, GROMOS45a3, GROMOS53a5, GROMOS53a6, GROMOS43a1, GROMOS43a2, and GROMOS54a7, form β-sheets too rapidly; (iii) AMBER99-ILDN, AMBER14SB, CHARMM22*, CHARMM36, and CHARMM36m achieved a balanced performance in reproducing the experimental findings on structures and kinetics. For the case of PHF6, the result pointed out that PHF6 β-sheet structures sampled by AMBER99SB-disp, AMBER14SB, GROMOS54a7, and OPLSAA are in favor of the antiparallel β-sheets, while the predominant type of β-sheet structures sampled by CHARMM36m is parallel β-sheet. Additionally, our simulation results also revealed that the protein-water interaction was overestimated in AMBER99SB-disp, in contrast the protein-protein interaction was overestimated in GROMOS54a7, and OPLSAA. In summary, AMBER99-ILDN, AMBER14SB, CHARMM22* and CHARMM36 are good candidates for studying Aβ assembly, while CHARMM36m is currently most suitable among the evaluated FFs for studying the aggregation of both Aβ and Tau through MD simulations.

Viet Hoang Man

Faculty Research Instructor, University of Pittsburgh

Appearances