Insights into amyloid structural formation and assembly through computational approaches

Amyloids are long, insoluble ordered fibers. Due to their insolubility, to date the determination of an amyloid structure with an atomic scale resolution has proven to be a difficult task. Under such circumstances, computational approaches are a preferred option, providing the means to build likely models, test their stabilities and figure out the chemistry of their prevailing interactions. Computational models can be validated by targeted experiments, such as introducing mutations and testing for amyloid formation. Computations further provide vehicles for the comprehension of the mechanisms of amyloid seed formation and oligomer toxicity. Nevertheless, computations face an immense hurdle, the outcome of the time scales involved in amyloid formation and the immense sizes of the systems. In an attempt to overcome these, we adopt a strategy that encompasses (1) bioinformatics studies of native proteins containing β-sheet structures; (2) simulations of shorter peptides; and finally (3) construction of potential oligomeric models and tests of their stabilities. The results are correlated with experimental data where available. Here, we describe the computational methods in simple terms and present an overview of the results. The systems derive from amyloidogenic, disease-related proteins, including gelsolin, β₂-microglobulin, and peptides derived from the prion, Alzheimer's Aβ, IAPP and human calcitonin. Ultimately, obtaining molecular structures should facilitate efforts to therapy and drug design.