Résumé
It is generally accepted that spongiform encephalopathies result from the aggregation into amyloid of a ubiquitous protein, the so-called prion protein. As a consequence, the dynamics of amyloid formation should explain the characteristics of the prion diseases: infectivity as well as sporadic and genetic occurrence, long incubation time, species barriers and strain specificities. The success of this amyloid hypothesis is due to the good qualitative agreement of this hypothesis with the observations. However, a number of difficulties appeared when comparing quantitatively the in vitro experimental results with the theoretical models, suggesting that some differences should hide important discrepancies. We used well defined quantitative models to analyze the experimental results obtained by in vitro polymerization of the recombinant hamster prion protein. Although the dynamics of polymerization resembles a simple nucleus-dependent fibrillogenesis, neither the initial concentration dependence nor off-pathway hypothesis fit with experimental results. Furthermore, seeded polymerization starts after a long time delay suggesting the existence of a specific mechanism that takes place before nucleus formation. On the other hand, polymerization dynamics reveals a highly stochastic mechanism, the origin of which appears to be caused by nucleation heterogeneity. Moreover, the specific structures generated during nucleation are maintained during successive seeding although a clear improvement of the dynamics parameters (polymerization rate and lag time) is observed. We propose that an additional on-pathway reaction takes place before nucleation and it is responsible for the heterogeneity of structures produced during prion protein polymerization in vitro. These amyloid structures behave like prion strains. A model is proposed to explain the genesis of heterogeneity among prion amyloid.