Reverse-engineering amyloid strains with generative protein design

Authors:
Preprint
Publisher website
Google Scholar
Abstract
Amyloid fibrils are intrinsically polymorphic protein assemblies that form distinct structural strains linked to diverse biological and pathological outcomes. Yet, the principles governing how sequence encodes diverse fibril architectures, and the extent to which a given fold constrains underlying amino-acid sequence compatibility, remain poorly understood. Here, we apply generative protein design to directly interrogate the sequence–structure relationship of defined fibril architectures, using α-synuclein (αS), a protein known to form highly polymorphic amyloid fibrils, as a model system. Sampling sequence space under structural constraints reveals a continuous compatibility manifold in which diverse sequences encode a common amyloid architecture. De novo designed sequences assemble into fibrils, often with enhanced aggregation efficiency relative to αS. A subset exhibits strain-like behaviour, including similar morphologies, efficient cross-templating, and induction of αS cellular propagation, thereby functionally validating structural compatibility with the native fibril fold. Energetic analysis shows that stability is achieved through distinct but compensatory interactions, supporting a non-unique mapping between sequence and structure. Together, our results define a continuous and constrained compatibility landscape underlying amyloid strains, providing a framework for understanding the determinants of polymorphism and establishing generative protein design as a strategy to access this space, interrogate amyloid sequence–structure relationships, and engineer fibrillar protein assemblies and functional biomaterials.
Funding Information
  • University of Texas Southwestern Medical Center (Thomas O. Hicks Scholar in Medical Research awrd)

This publication has 90 references indexed in Scilit: