Back

Massively parallel quantification of mutational impact on IAPP amyloidformation

Badia, M.; Batlle, C.; Bolognesi, B.

2025-05-23 biophysics
10.1101/2025.05.19.654797 bioRxiv
Show abstract

Amyloid fibrils formed by the islet amyloid polypeptide (IAPP) cause pancreatic beta-cell damage, resulting in reduced insulin secretion and Type 2 diabetes (T2D). Variations in the primary amino acid sequence of IAPP can influence its aggregation rate and animals expressing IAPP variants that do not form amyloids, do not develop T2D. Conversely, specific single amino-acid changes in IAPP are enough to accelerate its aggregation rate. Understanding how mutations impact IAPP aggregation can help gain mechanistic understanding into the process of pathogenic amyloid formation of this peptide and preventively identify mutations that may contribute to the risk of developing T2D. Here, we employ deep mutational scanning to measure the ability to nucleate amyloids for 1663 IAPP variants, including substitutions, insertions, truncations and deletions and identify variants that increase amyloid formation in all mutation classes. Our results point at a continuous stretch of residues (15-32) which likely is structured in IAPP amyloids and that matches the core of the early aggregated species formed by IAPP in vitro. Inside this region, mutations have a more drastic effect in the 21-27 NNFGAIL segment, suggesting tighter structural constraints for this stretch in IAPP amyloids. Finally, by comparing this mutational atlas to that of another amyloid, Amyloid beta (A{beta}42), the peptide that aggregates in Alzheimers Disease, we find that the effects of mutations that slow down nucleation correlate between the two amyloids, but that when it comes to mutations that accelerate nucleation one single amyloid dataset cannot be used to predict mutational effects in the other.

Published in Nature Communications (predicted rank #1) · training set

Matching journals

The top 1 journal accounts for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.