From milestone studies, it is known that conversion of soluble peptide to insoluble amyloid often involves the production of a partially unfolded intermediate.
18–
20 This thermodynamically unfavorable state rapidly progresses to the stable amyloidogenic form and the kinetics of this transition have been studied in vitro using a range of biophysical methods including light scattering, size exclusion chromatography, fluorimetry and ultracentrifugation. The results of such studies propose a nucleation dependent polymerisation model to describe fibril formation, likening the process to crystallization.
21 A heterogenous nucleus (‘seed') or peptide micelle form above a critical ‘threshold’ concentration and fibrils nucleate within these, elongating by irreversibly binding monomers to their free ends.
22,
23 Fibril growth may be represented diagrammatically as a lag exponential growth curve where the phase is considerably shortened in the presence of seeds.
In the evolution of mature fibrils, several metastable intermediates have been identified and isolated.
22,
24–
26 These include very early species: dimers, trimers, tetramers (collectively known as oligomers
27), Aβ derived diffusible ligands or ADDLs
28 and the later bead-like structures up to 200 nm in length called protofibrils.
29 Alternatively, fibril formation may follow an ‘offset pathway’ without the production of fibrils but instead involving conversion of the intermediates into amorphous deposits.
The dynamics of protein folding/misfolding appear to play a key role in in vitro fibril formation and consequently several molecular mechanisms to explain amyloid formation have been proposed; including the polar zipper
30 and domain swapping models.
31,
32 It has been proposed that amyloid formation is a generic property of all peptides
33 since under denaturing conditions many normally globular, nondisease related proteins, have been shown to assemble to form fibrils.
34,
35 However, the propensity of a peptide to form amyloid is dependent on several factors including: polypeptide charge, sequence, hydrophobicity and secondary structure. Consequently several algorithms have been developed to predict the propensity and the rate at which different sequences will aggregate and which mutations will result in an increase or decrease of aggregation rate.
36–
40 These algorithms will prove useful in determining which regions of a polypeptide chain is specifically involved in fibril formation or the amyloid core structure.