Replicating amyloids, called prions, are responsible for transmissible neurodegenerative diseases in mammals and some heritable phenotypes in fungi. The transmission of prions between species is usually inhibited, being highly sensitive to small differences in amino acid sequence of the prion-forming proteins. To understand the molecular basis of this prion interspecies barrier, we studied the transmission of the [PSI+] prion state from Sup35 of Saccharomyces cerevisiae to hybrid Sup35 proteins with prion-forming domains from four other closely related Saccharomyces species. Whereas all the hybrid Sup35 proteins could adopt a prion form in S. cerevisiae, they could not readily acquire the prion form from the [PSI+] prion of S. cerevisiae. Expression of the hybrid Sup35 proteins in S. cerevisiae [PSI+] cells often resulted in frequent loss of the native [PSI+] prion. Furthermore, all hybrid Sup35 proteins showed different patterns of interaction with the native [PSI+] prion in terms of co-polymerization, acquisition of the prion state, and induced prion loss, all of which were also dependent on the [PSI+] variant. The observed loss of S. cerevisiae [PSI+] can be related to inhibition of prion polymerization of S. cerevisiae Sup35 and formation of a non-heritable form of amyloid. We have therefore identified two distinct molecular origins of prion transmission barriers between closely sequence-related prion proteins: first, the inability of heterologous proteins to co-aggregate with host prion polymers, and second, acquisition by these proteins of a non-heritable amyloid fold.

In yeast, fragmentation of amyloid polymers by the Hsp104 chaperone allows them to propagate as prions. The prion-forming domain of the yeast Sup35 protein is rich in glutamine, asparagine, tyrosine, and glycine residues, which may define its prion properties. Long polyglutamine stretches can also drive amyloid polymerization in yeast, but these polymers are unable to propagate because of poor fragmentation and exist through constant seeding with the Rnq1 prion polymers. We proposed that fragmentation of polyglutamine amyloids may be improved by incorporation of hydrophobic amino acid residues into polyglutamine stretches. To investigate this, we constructed sets of polyglutamine with or without tyrosine stretches fused to the non-prion domains of Sup35. Polymerization of these chimeras started rapidly, and its efficiency increased with stretch size. Polymerization of proteins with polyglutamine stretches shorter than 70 residues required Rnq1 prion seeds. Proteins with longer stretches polymerized independently of Rnq1 and thus could propagate. The presence of tyrosines within polyglutamine stretches dramatically enhanced polymer fragmentation and allowed polymer propagation in the absence of Rnq1 and, in some cases, of Hsp104.