Back

Structural evolution of a yeast amyloid in vivo is shaped by chaperones

Wang, Z.; Weetman, S. L.; Altenhuber, B.; Murzin, A. G.; Gilbert, J. D.; Zenezini Chiozzi, R. G.; Koloteva-Levine, N.; Saibil, H.; Collinge, J.; Tuite, M. F.; Xue, W.-F.; Zhang, W.

2026-08-11 molecular biology
10.64898/2026.08.10.743537 bioRxiv
Show abstract

Cryo-electron microscopy (cryo-EM) studies of amyloid fibrils have revealed endpoint structures of disease-relevant filaments and polymorphic intermediates formed during in vitro assembly of prion-like proteins. However, how transmissible prion or prion-like amyloids evolve during de novo formation and maturation in living cells remains unknown. Here, using the yeast prion [PSI+] as a model, we isolated Sup35NM amyloid fibrils from successive stages of [PSI+] maturation in Saccharomyces cerevisiae and characterised their near-atomic structures and population-level structural diversity by combining cryo-EM and atomic force microscopy. We show that intermediate and mature states differ in predominant fibril structure and the regions of the Sup35 sequence incorporated into the core, and that structural diversity decreases during maturation. Curing of [PSI+] at the mature state by guanidine hydrochloride (GdnHCl), which selectively inhibits ATPase activity of the chaperone Hsp104, restored both the predominant intermediate amyloid structure and the broader structural diversity characteristic of the intermediate state. In addition, Hsp104, Ssa1 (Hsp70) and Sis1 (Hsp40) associate differently with fibrils from the two states. Together, these findings provide direct structural evidence for amyloid evolution in vivo and support a chaperone-mediated mechanism of conformer selection within a polymorphic amyloid population.

Matching journals

The top 4 journals account 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.