The amyloid-β (Aβ) peptide is an intrinsically disordered protein whose self-association into toxic oligomers underlies Alzheimer's disease. Because of its dynamic and heterogeneous nature, identifying the conformational states that nucleate aggregation remains a central challenge. In this work, we introduce a chemically interpretable descriptor of amyloidogenic propensity derived from self-docking analyses of conformational ensembles generated through temperature-replica exchange molecular dynamics (T-REMD) using different and complementary force fields. This descriptor classifies individual conformers within the generated ensembles according to their intrinsic aggregation tendency, enabling the identification of metastable, aggregation-prone states. The resulting ensembles reproduce experimental observables, and their classification based on amyloidogenic propensity provides a consistent structural basis for the rationalization and study of these metastable conformers. As a test, we demonstrate that the molecular chaperone DNAJB6, experimentally known to bind amyloidogenic conformations, preferentially interacts with aggregation-prone conformers, thus supporting both the proposed protocol and the consistency of the classification scheme. More broadly, this framework outlines a potentially generalizable strategy to identify metastable states in intrinsically disordered proteins as prospective pharmacological targets to help develop drugs or biomolecules capable of inhibiting the early stages of their aggregation.
Bini et al. (Tue,) studied this question.