] by coexpression of a dominant, ATPase-negative allele of HSP104 were similar to those observed for GdnHCl-induced elimination. Based on these and other data, we propose a two-cycle model for "prionization" of Sup35p in [PSI ؉ ] cells: cycle A is the GdnHCl-sensitive (Hsp104-dependent) replication of the prion seeds, while cycle B is a GdnHCl-insensitive (Hsp104-independent) process that converts these seeds to pelletable aggregates.[PSI ϩ ] is a non-Mendelian epigenetic element originally identified in a genetic screen by its ability to increase the efficiency of tRNA-mediated nonsense suppression in Saccharomyces cerevisiae (6, 7). Based on genetic arguments, Wickner (47) was the first to suggest that the [PSI ϩ ] determinant was a self-replicating prion form of translation termination factor Sup35p (also called eRF3 [43,50]). In accordance with Prusiner's original prion hypothesis (33), Sup35p is able to exist in one of two different states: the normal soluble form, required for translation termination, and an "infectious" (i.e., transmissible) aggregated form, i.e., as a prion. ] cells, Sup35p is largely inactivated by its sequestration into high-molecular-weight complexes, thereby precluding it from efficiently performing its role as a translation termination factor.Studies on the conversion of mammalian prion protein PrP to transmissible prion PrP Sc form have generated a number of different models to explain how a prion protein determinant can be propagated and transmitted (see reference 2 for a review). A detailed in vitro analysis of yeast Sup35p self-directed aggregation led Serio et al. (39) to suggest that a process of nucleated conformational conversion of Sup35p most probably occurs in the establishment of the [PSI ϩ ] determinant. This mechanism involves both the nucleated-polymerization and the template assembly processes proposed for PrP (2), and Serio et al. (39) proposed that conformational conversion of the native Sup35p occurs by the seeding activity of a nucleated complex. These nuclei (or intermediate oligomers) would be formed by the conformational modification of the partially unfolded Sup35p protein. Once the intermediate polymers reach a critical concentration, assembly of large polymers can occur through a templating mechanism.Physicochemical analyses of Sup35p have shown that the N-terminal portion of the protein (the so-called N domain or prion-forming domain) contains the minimal region responsible for [PSI ϩ ] propagation in vivo and directs amyloid fiber formation in vitro (16,23,32). The Sup35p amyloid fibers have features in common with the amyloid proteins implicated in several human diseases (for reviews, see references 1, 11, 34, and 48). However, it remains to be established whether the * Corresponding author. Mailing address:

SummaryIn the yeast Saccharomyces cerevisiae, Sup35p (eRF3), a subunit of the translation termination complex, can take up a prion-like, self-propagating conformation giving rise to the non-

BackgroundPrions are infectious proteins propagating as self-perpetuating amyloid polymers. The [Het-s] prion of Podospora anserina is involved in a cell death process associated with non-self recognition. The prion forming domain (PFD) of HET-s adopts a β-solenoid amyloid structure characterized by the two fold repetition of an elementary triangular motif. [Het-s] induces cell death when interacting with HET-S, an allelic variant of HET-s. When templated by [Het-s], HET-S undergoes a trans-conformation, relocates to the cell membrane and induces toxicity.Methodology/Principal FindingsHere, comparing HET-s homologs from different species, we devise a consensus for the HET-s elementary triangular motif. We use this motif to screen genomic databases and find a match to the N-terminus of NWD2, a STAND protein, encoded by the gene immediately adjacent to het-S. STAND proteins are signal transducing ATPases which undergo ligand-induced oligomerisation. Homology modelling predicts that the NWD2 N-terminal region adopts a HET-s-like fold. We propose that upon NWD2 oligomerisation, these N-terminal extensions adopt the β-solenoid fold and template HET-S to adopt the amyloid fold and trigger toxicity. We extend this model to a putative prion, the σ infectious element in Nectria haematococca, because the s locus controlling propagation of σ also encodes a STAND protein and displays analogous features. Comparative genomic analyses indicate evolutionary conservation of these STAND/prion-like gene pairs, identify a number of novel prion candidates and define, in addition to the HET-s PFD motif, two distinct, novel putative PFD-like motifs.Conclusions/SignificanceWe suggest the existence, in the fungal kingdom, of a widespread and evolutionarily conserved mode of signal transduction based on the transmission of an amyloid-fold from a NOD-like STAND receptor protein to an effector protein.