Massively parallel quantification of mutational impact on IAPP amyloid formation.

Amyloid fibrils formed by the islet amyloid polypeptide cause pancreatic beta-cell damage, resulting in reduced insulin secretion and type 2 diabetes. Changes in the amino acid sequence of this peptide can influence its aggregation rate, and animals expressing variants that do not form amyloids do not develop type 2 diabetes. Conversely, specific single amino acid changes can accelerate the aggregation rate of this peptide. Here, we employ deep mutational scanning to measure the ability of 1916 islet amyloid polypeptide variants, including substitutions, insertions, truncations and deletions, to nucleate amyloids. Our results identify a continuous stretch of residues from 15 to 32 that is particularly sensitive to mutation. This region, which is likely structured in amyloids, matches the core of the early aggregated species formed by this peptide in vitro. Within this region, mutations in residues 21 to 27 have a substantial effect, suggesting tighter structural constraints. Finally, we compare the mutational atlas of the islet amyloid polypeptide to that of amyloid beta - the peptide that aggregates in Alzheimer's disease - and find that mutations that slow down nucleation correlate between the two amyloids, but mutations that accelerate nucleation in one amyloid cannot be used to predict mutational effects in the other.