Abstract

Aggregation of beta-amyloid (ABeta) peptides into oligomers and fibrils is implicated in the pathogenesis of Alzheimer's disease. Two broad hypotheses exist concerning the process of aggregation. One hypothesis suggests that an autonomously folding core domain, possibly ABeta 21-30, initiates folding that allows flanking hydrophobic residues to associate and form beta-sheet in fibrils. The alternative hypothesis posits that interactions between hydrophobic residues flanking the flexible, hydrophilic core domain (ABeta 21-30) cause polymorphic associations that propagate self-association into polymorphic oligomers and fibrils. To compare these hypotheses, we examined several internal amyloid beta core fragments - ABeta 21-30, 16-34, and 13-38 - as well as cyclic and Cysteine-modified variants by NMR spectroscopy and other techniques. We found no evidence that ABeta 21-30 constitutes an autonomously folding core domain or has the ability to promote oligomerization and fibrillization of ABeta monomers. Conversely, extended peptides - ABeta 16-34 and 13-38 - possess a beta-turn-like structure and self-aggregate into soluble oligomers and fibrils. Furthermore, their Cys variants favor interactions between sidechain residues known to interact in ABeta fibrils. Specifically, in Cys-ABeta 16-34, fibrils form pari passu with the formation of disulfide bonds. Finally, sequential NMR of ABeta 16-34 and Cys-ABeta 16-34 demonstrate non-uniform signal decay, indicating a failure to adopt a single structure and the formation of polymorphic aggregated species (i.e. oligomers and fibrils) favored by the hydrophobic effect. Thus, the defining feature of amyloid beta - or any amyloid - is its incompetence in assuming a unique three-dimensional fold beyond the most rudimentary architectures, e.g., parallel, in-register beta-sheets.