Prions represent an unusual structural form of a protein that is ‘infectious’. In mammals, prions are associated with fatal neurodegenerative diseases such as CJD (Creutzfeldt–Jakob disease), while in fungi they act as novel epigenetic regulators of phenotype. Even though most of the human prion diseases arise spontaneously, we still know remarkably little about how infectious prions form de novo. The [PSI+] prion of the yeast Saccharomyces cerevisiae provides a highly tractable model in which to explore the underlying mechanism of de novo prion formation, in particular identifying key cis- and trans-acting factors. Most significantly, the de novo formation of [PSI+] requires the presence of a second prion called [PIN+], which is typically the prion form of Rnq1p, a protein rich in glutamine and aspartic acid residues. The molecular mechanism by which the [PIN+] prion facilitates de novo [PSI+] formation is not fully established, but most probably involves some form of cross-seeding. A number of other cellular factors, in particular chaperones of the Hsp70 (heat-shock protein 70) family, are known to modify the frequency of de novo prion formation in yeast.
- [PSI+] prion
- ubiquitin–proteasome system
To explain why the causative agent of sheep scrapie was resistant to treatments that inactivate nucleic acids, such as UV radiation and formaldehyde, Griffith  invoked the presence of an infectious protein. This idea was further elaborated on by Prusiner and co-workers, who introduced the term prion, a ‘small proteinaceous infectious particle which resists inactivation by procedures that modify nucleic acids’ [2,3]. The prion protein in scrapie and related diseases in other mammals [e.g. bovine spongiform encephalopathy and human CJD (Creutzfeldt–Jakob disease)] is designated PrP: while PrPc denotes the normal soluble form, PrPSc denotes the infectious ‘scrapie’ form. In 1994, Prusiner's ‘protein-only’ prion hypothesis was proposed by Reed Wickner to account for the unusual genetic behaviour of [URE3], a non-Mendelian trait of yeast Saccharomyces cerevisiae that modified the ability of cells to utilize poor nitrogen sources . Subsequently, several other fungal prions have been described and their unusual properties validated experimentally, namely [PSI+], [HET-s] and [RNQ+] .
Mammalian prion diseases such as CJD either can arise de novo (i.e. sporadic form) or can result from infection (i.e. acquired form) or can result from the inheritance of a mutant allele of the PrP gene (i.e. familial form), which leads to a destabilization of the structure of PrP. The most predominant form of CJD is the sporadic form, which accounts for over 80% of the verified cases of CJD in the U.K. (National Creutzfeldt–Jakob Disease Surveillance Unit; http://www.cjd.ed.ac.uk). For PrPSc to form de novo, it is assumed that the natively folded protein adopts an alternative conformational state, either by spontaneously misfolding or as a consequence of an interaction with other cellular factors, leading to the formation of the PrPSc ‘isoform’ that can act as a template for the conversion of PrPc molecules into the infectious PrPSc conformer. While soluble forms of PrP and the various fungal prion proteins can spontaneously polymerize to form amyloid fibrils in vitro in the absence of added cellular factors, the mechanism of de novo formation in vivo is poorly understood in molecular terms. A recent study has suggested that the formation of a PrP dimer is sufficient to trigger de novo formation of PrPSc . Alternatively, several molecules of the misfolded protein may spontaneously form a catalytically active oligomer, as has been suggested for the yeast [PSI+] prion .
Yeast prions can form de novo in the absence of transmission of pre-existing prion seeds (propagons) from another cell, but the frequency of such a spontaneous appearance is of the order of 1 in 105 (i.e. 10−5). This frequency can be elevated by several orders of magnitude by overexpression of the corresponding prion protein [4,9]. As with de novo PrPSc formation, the assumption is that spontaneous formation of yeast prions arises as a consequence of a random protein misfolding event, an event whose occurrence is greatly elevated when the protein folding machinery is swamped by excess of the protein or impaired by mutation. Yet, as has become clear from studies we review below, the level of spontaneous formation of a yeast prion is regulated by both cellular and epigenetic factors.
The yeast [PSI+] prion
Studies on the [PSI+] prion have provided considerable insights into cellular factors that modulate the de novo formation of prions. [PSI+] was first identified in 1965 by Brian Cox  as a non-Mendelian factor that regulated the efficiency of nonsense suppression. Some 30 years later, [PSI+] was shown to be the prion form of the Sup35p protein [11,12]. Sup35p [also known as eRF3 (eukaryotic release factor 3)] facilitates translation termination through its interaction with Sup45p (eRF1), leading to polypeptide chain release [13,14].
The prion-like behaviour of Sup35p is controlled by its N-terminal region, which is also called the PrD (prion-forming domain), whereas its C-terminal region is necessary and sufficient for its role in translation termination . The N-terminal and C-terminal regions of Sup35p are separated by the charged M region that is dispensable for translation termination [15,16] but may contribute to the prion properties of the protein . When Sup35p forms prion aggregates, this gives rise to the [PSI+] phenotype, suggesting that the protein can no longer participate in translation termination, possibly due to inhibition of the formation of the Sup35p–Sup45p complex. This loss of function results in the low-level translation of nonsense codons (i.e. nonsense suppression), which can be used to identify the presence of [PSI+]  by exploiting the colour difference caused by suppression of the ade1-14 or ade2-1 nonsense alleles: [Psi−] cells form red colonies, while [PSI+] cells form white colonies.
The [PSI+] prion can be transmitted from mother to daughter cells with high efficiency, and spontaneous loss of [PSI+] from an actively growing culture is a rare event, typically occurring at a frequency of the order of 10−5–10−6. Some variants of the [PSI+] prion exist (the so-called ‘weak’ variants) that show a higher frequency of mitotic loss [18,19].
The efficiency with which [PSI+] (and other yeast prions) is maintained in growing cells lies in the action of the molecular chaperone Hsp104 (heat-shock protein 104) . New prion seeds (propagons) are thought to be generated by the disaggregation of the Sup35p prion polymers that continually form in [PSI+] cells. Such a disaggregation is mediated by the action of Hsp104 in conjunction with other chaperones of the Hsp70 (Ssa/Ssb) and Hsp40 (Ydj1p, Sis1p) families (see  for a review). High-frequency loss of [PSI+] from growing cells can result from inhibiting the activity of Hsp104 using millimolar concentrations of GdmCl (guanidinium chloride) [22,23] or by overexpression of wild-type Hsp104 . It is important to stress that both the [PSI+] and [psi−] states are stable and are lost at rates similar to those associated with nuclear gene mutation in yeast.
De novo formation of [PSI+]
For de novo formation of [PSI+], there is an absolute requirement for the cells to carry a second prion referred to as [PIN+]. In [PIN+] [psi−] cells, the frequency of appearance of [PSI+] is ∼10−5, whereas in an otherwise isogenic [pin−] [psi−] strain this frequency decreases to non-detectable levels (<10−8). Where [PSI+] cells do arise from a [pin−] strain, these usually have also converted into [PIN+] (N. Koloteva-Levine, G. Merritt and M.F. Tuite, unpublished work). The frequency of de novo formation of [PSI+] is dramatically elevated when either the full-length Sup35p or just the Sup35p-PrDM region is overexpressed even for comparatively short periods of time (Table 1).
[PIN+] was initially defined as a factor that induces [PSI+] formation de novo [18,24]. Subsequent studies revealed that [PIN+] is usually the [RNQ+] prion that consists of aggregates of Rnq1p [25–27]. Rnq1p is a protein of unknown function and the [RNQ+] prion is the only prion that has so far been detected in natural isolates of Saccharomyces cerevisiae . Several other heterologous protein aggregates can also facilitate de novo conversion of [PSI+], namely the [URE3] prion  and variants of the poly-glutamine-rich protein huntingtin . In addition, the overexpression of a number of other yeast glutamine/asparagine-rich proteins can facilitate de novo [PSI+] formation. These include the New1p protein that contains a putative PrD that can functionally substitute for the Sup35p–PrD in [PSI+] formation [27,30].
The data so far available are consistent with [PIN+] prion aggregates acting as imperfect templates on which Sup35p molecules misfold and assemble into the characteristic transmissible prion aggregates. This ‘cross-seeding’ interaction occurs only as the initial step during the de novo formation of [PSI+], and subsequently, the two proteins do not stably co-aggregate , with the exception of Sup35p from certain closely related yeast species . There is, however, evidence for their co-localization during the initial cross-seeding [29,33] and we have been able to detect interactions between soluble forms of Sup35p and Rnq1p in vivo using both yeast two-hybrid analysis and co-immunoprecipitation (N. Koloteva-Levine, G. Merritt and M.F. Tuite, unpublished work).
Other factors influencing the de novo formation of [PSI+]
A variety of proteins, other than those that have been proved to form prions, can also influence the de novo formation of [PSI+]. These include Sup45p, the native binding partner of Sup35p, members of the Hsp70 chaperone family and components of the UPS (ubiquitin–proteasome system) (Table 2).
Sup45p physically and functionally interacts with Sup35p to form a release factor complex essential for translation termination [13,14]. Sup35p interacts with the C-terminus of Sup45p , and any Sup35p molecule that has formed a complex with Sup45p is presumably not available for de novo conversion. Consistent with this hypothesis, Derkatch et al.  found that overexpression of the SUP45 gene significantly reduced the frequency of induced de novo formation of [PSI+], but did not affect its subsequent propagation. Sup45p could, by binding to Sup35p, stabilize its conformation enough to prevent imperfect seeding by [PIN+] but not to prevent polymerization once [PSI+] seeds have been established. Alternatively, Sup45p could directly interact with other cellular factors such as chaperones, or Rnq1p and/or [RNQ+].
Molecular chaperones play a central role in the co- and post-translational folding of polypeptides and would therefore be expected to act as key cellular regulators of de novo prion formation in addition to their role in prion propagation . Although chaperones from the Hsp104, Hsp70 and Hsp40 families contribute to the mechanism of yeast prion propagation , only members of the Hsp70 family (namely Ssa1,2 and Ssb1,2) and Sse1p (Hsp110) have been implicated in de novo formation of prions.
The Ssa and Ssb chaperones play distinct but opposite roles in the control of de novo prion formation. For example, strains lacking the Ssb proteins (ΔSSB) exhibit up to 10-fold higher rate of de novo formation of [PSI+] , whereas overexpression of SSA1 and other members of the SSA family, but not SSB1, increases the frequency of [PSI+] de novo formation . These effects suggest that the Ssa chaperones might stabilize the ‘misfolded’ conformations of prion proteins and in so doing facilitate their conversion . The closely related Ssbs might function to remove or refold the critical misfolded conformers that form. The fundamental differences in the way Ssa and Ssb chaperones deal with such misfolded conformers can be attributed to their distinct peptide-binding domains .
Factors that affect Ssa/Ssb synthesis and/or function also affect the frequency of de novo prion formation. For example, elevations in the levels of Sse1p (a nucleotide-exchange factor for Hsp70) promotes de novo formation, while a ΔSSE1 strain shows a reduced frequency of appearance of [PSI+] . Mutational changes to the HSF (heat-shock factor; encoded by the HSF1 gene) also affect de novo formation , presumably by modulating the transcription of the various Hsp70/Hsp110-encoding genes.
While a number of studies have demonstrated a role for chaperones in the formation and propagation of [PSI+], none have yet identified the molecular species that chaperones interact with. In vitro formation of mature amyloid fibrils of Sup35p molecules is preceded by aggregation of these molecules into oligomeric species , and it is possible that rather than binding monomers of Sup35p, the Ssa/Ssb chaperones may interact with such oligomers, if they form in vivo. The different chaperones would interact preferentially with different oligomeric forms, thus altering the balance in favour of on- or off-pathway species and thus promote or inhibit de novo prion formation.
While functional Hsp104 is essential for the continued propagation of [PSI+], it remains to be established whether this chaperone also plays a role in de novo formation. In vitro studies have suggested that Hsp104 promotes the formation of Sup35p fibrils from soluble full-length Sup35p [40,41]. However, if Hsp104 function is inhibited by GdmCl during de novo prion formation induced by overexpression of Sup35p or Sup35p-PrDM, no reduction in the frequency of [PSI+] appearance is seen (Table 1), suggesting that Hsp104 may not be required in vivo at least for induced de novo conversion.
The UPS facilitates the removal of cellular proteins that are damaged, misfolded or otherwise defective but whose accumulation may be detrimental to the cell. This may also be so for alternative conformers of Sup35p that form during the de novo formation of [PSI+]. Creating defects in the yeast UPS can affect de novo formation of [PSI+]; for example, deletion of the UBC4 gene (which encodes the major ubiquitin-conjugating enzyme) increases the frequency of spontaneous formation of [PSI+] . However, as there is no evidence that Sup35p is ubiquitinated, this effect might be indirect, possibly by triggering the formation of large aggresome-like structures .
In conclusion, a number of cellular factors can modulate the frequency with which yeast prions form de novo. While these factors exist primarily to deal with misfolded and potentially toxic conformers in both stressed and non-stressed cells, they can also modulate the formation of novel epigenetic elements, one of which, namely the [PSI+] prion, has an effect on the ability of the yeast cell to combat a variety of environmental abuses [43,44].
Our studies on yeast prion formation were funded by the BBSRC (Biotechnology and Biological Sciences Research Council) and the Wellcome Trust.
British Yeast Group Meeting 2008: Independent Meeting held at National University of Ireland Maynooth, Maynooth, Co. Kildare, Ireland, 18–20 March 2008. Organized and Edited by Gary Jones (National University of Ireland Maynooth, Ireland).
Abbreviations: CJD, Creutzfeldt–Jacob disease; eRF3, eukaryotic release factor 3; GdmCl, guanidinium chloride; Hsp, heat-shock protein; PrD, prion-forming domain; PrP, prion protein; PrPC, normal soluble form of PrP; PrPSc, infections scrapie form of PrP; UPS, ubiquitin–proteasome system
- © The Authors Journal compilation © 2008 Biochemical Society