Amyloid Diseases of Yeast: Prions are Proteins acting as Genes

Abstract

The unusual genetic properties of the non-chromosomal genetic elements, [URE3] and [PSI+] led to their being identified as prions (infectious proteins) of Ure2p and Sup35p, respectively. Ure2p and Sup35p, and now several other proteins, can form amyloid, a linear ordered polymer of protein monomers, with a part of each molecule, the prion domain, forming the core of this beta-sheet structure. Amyloid filaments passed to a new cell seed the conversion of the normal form of the protein to the same amyloid form. The cell’s phenotype is affected, usually from the deficiency of the normal form of the protein. Solid-state NMR studies indicate that the yeast prion amyloids are in-register parallel β-sheet structures, in which each residue (e.g. asparagine 35) forms a row along the filament long axis. The favorable interactions possible for aligned identical hydrophilic and hydrophobic residues are believed to be the mechanism for propagation of amyloid conformation. Thus, just as DNA mediates inheritance by templating its own sequence, these proteins act as genes by templating their conformation. Distinct isolates of a given prion have different biological properties, presumably determined by differences between the amyloid structures. Many lines of evidence indicate that the S. cerevisiae prions are pathological, disease agents, although the example of the [Het-s] prion of Podospora anserina shows that a prion can have beneficial aspects.

Summary:

>Certain yeast proteins can form self-propagating amyloid (a fibrous beta-sheet rich aggregate) that produces cellular defects, largely by inactivating the amyloid-forming protein.

>The self-propagation of the amyloids makes these infectious and heritable, so these proteins act as genes by templating their conformation, just as DNA templates its sequence.

>Chaperones play an important part in prion generation and propagation, and aggregate-collecting mechanisms also impinge on prion processes.

Introduction

Infectious elements in yeast generally appear as non-chromosomal genetic elements. Most differences between mated strains segregates 2+: 2- in meiosis, the pattern of a difference in a single chromosomal gene. However, if one parent in a genetic cross carries one of the yeast viruses and the other does not, all of the meiotic progeny will have the virus, a pattern called 4+ : 0 segregation. Yeast viruses (and prions) do not exit one cell into the environment and then enter another as is the case for human viruses or prions. [URE3] [1] and [PSI+] [2] are two long - known yeast non-chromosomal genetic elements whose basis was unknown. The similarity of the phenotype of [URE3] to that of mutants in the ure2 chromosomal gene, and the fact that URE2 is necessary for the propagation of [URE3] [3], first led us to suspect that [URE3] was actually a prion form of the Ure2 protein, an altered form of the Ure2p with the ability to catalyze the conversion of the normal form into the same altered (inactive) form [4]. Ure2p is a regulator of nitrogen catabolism, turning off the genes encoding enzymes and transporters needed for using poor nitrogen sources if a good nitrogen source was present in the medium [5, 6]. If [URE3] were indeed a prion, we reasoned that overproduction of Ure2p should increase the frequency with which the [URE3] non-chromosomal genetic element would arise, and this too proved to be true [4]. Various treatments can induce very frequent mutation of the mitochondrial DNA or curing of the yeast viruses: growth in the presence of guanidine [7] or ethidium bromide [8], for example, induces mutation in mitochondrial DNA and its loss. But while these mitochondrial DNA mutations or loss are irreversible, curing a prion should be reversible, since the prion-forming protein is still being made in the cell and should be able (rarely) to again convert to the prion form. Thus, reversible curability (not curability itself) is expected to be a trait of a prion [4]. Indeed, [URE3] can be cured by growth in the presence of low concentrations of guanidine, but, unlike guanidine induced mitochondrial DNA curing, that of [URE3] is reversible (at low frequency), again indicating it is a prion [4].

Interestingly, these three properties had already been shown for [PSI+] and the SUP35 gene [9–11], and we concluded that [PSI+] was likewise a prion of Sup35p [4]. Sup35p is a subunit of the translation termination factor, a protein whose activity is, unlike that of Ure2p, essential for growth of S. cerevisiae [12, 13].

The prion concept first arose as a leading hypothesis in studies of the uniformly lethal mammalian transmissible spongiform encephalopathies (TSEs)[14–16], but proving it was difficult because of the long incubation times for even infected rhodents. While the yeast prion systems can be non-lethal (but see below), show phenotypes unrelated to the TSE diseases and involve proteins with no sequence relation to the mammalian prion protein PrP, the yeast systems provided an important model in which it was possible to actually prove that proteins could be infectious elements, and quickly explore the mechanisms involved.

Most yeast prions are amyloid forms of a normally soluble protein.

Amyloid is a filamentous polymer of identical protein monomers that is characterized by a cross-beta structure (see next section). Amyloids play a prominent role in human degenerative diseases such as Alzheimer’s disease, Parkinson’s disease and amyotrophic lateral sclerosis. These neurodegenerative diseases, each quite common, are caused by deposits of amyloid, each of a specific protein or peptide, that cause damage to the brain. Type 2 (late onset) diabetes mellitus is associated with (though probably not caused by) deposits of amylin, a peptide that, like insulin, is made in the Islets of the pancreas. Senile amyloidosis is a very common disorder of the elderly due to deposition of amyloid filaments of transthyretin, a protein normally found in serum.

Perhaps with the many amyloidoses of humans it is not surprising that yeast has several amyloid diseases. A series of studies showed that Sup35p is aggregated in cells carrying the [PSI+] prion, that Ure2p is aggregated in [URE3] cells, and similarly for the other identified yeast prions (see Table 1) [17–19]. Recombinant Sup35p could form amyloid in vitro [20, 21] and introduction of this amyloid into yeast cells makes them prion positive [22, 23]. Similar results were later obtained for Ure2p and [URE3] [24, 25]. Thus these amyloid filaments are infectious.

Table 1.

Prions of S. cerevisiae and Podospora anserina.

Prion	Protein	Normal Function	Prion Phenotype	Ref.
[URE3]	Ure2p	Nitrogen regulation	Inappropriate derepression of nitrogen catabolism genes; slow growth	[4]
[PSI+]	Sup35p	Translation termination	Readthrough of translation stop codons	[4]
[PIN+]	Rnq1p	Unknown	Rare cross – seeding of [PSI+] or other prions	[79]
[SWI+]	Swi1p	Chromatin remodeling	Inability to utilize non-fermentable carbon source and slow growth	[80]
[OCT+]	Cyc8p	Transcription repression	Derepressed transcription; flocculence	[81]
[MOT+]	Mot3p	Repressor of genes for anaerobic growth	Inappropriate derepression of anaerobic growth genes	[82]
[ISP+]	Sfp1p	Transcription factor	Decreased translation read through	[83]
[MOD+]	Mod5p	tRNA isopentenyl - transferase	Slow growth; resistance to azole anti-fungal drugs	[84]
[Het-s]	HET-s	Heterokaryon incompatibility	Heterokaryon incompatibility in Podospora anserina (a functional prion)	[63]
[BETA]	Prb1p	Vacuolar protease B	normal sporulation, survival in stationary phase (a functional prion)	[85]

Prion variants.

A single prion protein sequence can form amyloids of different detailed structure resulting in different biological properties. These are called ‘prion variants’. Since prions are proteins acting as genes, prion variants can be thought of as different alleles of the protein gene. Each prion variant is (relatively) stable and propagates as cells divide. Prion variants were long known in mammalian prions, and were found in yeast prions soon after their discovery [25–27]. In yeast, variants are commonly distinguished based on whether their phenotype is ‘strong’ or ‘weak’, whether the variant is stably propagated or readily lost, response to excess or deficiency of certain chaperones, and transmission efficiency across interspecies or intraspecies barriers. The existence of variants is critical in understanding many aspects of prions, including structural studies, interactions of prions with other cellular components and the biological role of prions for yeast.

Prion domains, their structure and biological implications.

Only a part of each prion protein actually forms the amyloid structure, and this part is sufficient to transmit the prion trait [20, 24, 28, 29]. These ‘prion domains’ have been the focus of structural studies. The prion domains of Ure2p and Sup35p are the N-terminal ~70 residues and ~124 residues, respectively, although the size of the prion domain varies somewhat with prion variant (see below for discussion of prion variants). One particularly clear demonstration of the role of the prion domain is in Ure2p, whose C-terminal part is essential for its role in regulating nitrogen assimilation, and which has a glutathione peroxidase activity [30]. Remarkably, this activity is unaffected by amyloid formation, showing that the C-terminal domain is unchanged by amyloid formation, including its homodimer status [30]. The definition of amyloid includes the cross-beta structure, meaning that the filaments are rich in beta sheet, and that the beta strands run perpendicular to the long axis of the filament. Within this definition are several possible architectures, depending on the relation of the beta strands to eachother:

1) Antiparallel beta sheets have adjacent beta strands running in opposite directions. This is the most common type of beta sheet in monomeric proteins.

2) In parallel beta sheets, the peptide chains are oriented in the same direction. If a parallel beta sheet is in-register, each residue is aligned with the same residue in the molecule before and after it in the filament (see Fig. 1). Each molecule occupies a single 4.7 angstrom layer along the long axis of the filament. In-register parallel beta sheets are the most common architecture of pathologic amyloids (reviewed by [31]), and current evidence supports this form for the infectious amyloids of the prion domains of Sup35p, Ure2p and Rnq1p [32–34].

Figure 1.

The in-register parallel architecture of yeast prion amyloid can explain templating of conformation. A. The unstructured prion domain of molecules newly joining the end of the amyloid filament are directed to assume the same conformation as molecules already in the filament by the favorable interactions between identical side chains that can only occur if the molecules are in-register [77]. B. Three interactions of an amino acid residue are shown. The black lines show the main chain hydrogen bonds of the β-sheet. The black double-headed arrow indicates the intermolecular side-chain interactions between identical residues (emphasized in A). Also shown are the interactions between non-identical side chains in the plane perpendicular to the long axis of the filaments. These interactions have been studied by X-ray crystallography of short peptides [78].

3) In a beta helix, each molecule occupies more than one layer along the long axis of the filament. Each molecule forms a helix of two or more turns. This architecture has been realized in the [Het-s] prion of Podospora anserina [35, 36].

The parallel in-register structure is maintained because of favorable interactions between identical amino acid side chains. Hydrophilic side chains can form a line of hydrogen bonds along the filament long axis and hydrophobic residues can favorably interact, but only if they are aligned. Aligned charged residues would repel eachother, but there are very few charged residues in these prion domains. These favorable interactions stabilize the in-register parallel structure, and direct a new monomer joining the ends of the filaments to assume the same conformation as that of the other monomers in the filament. Like the hydrogen bonding of complementary bases in DNA allows templating of sequence, these interactions of identical side chains allow templating of conformation.

Biology of yeast prions.

Prions arise stochastically, and not in response to specific environmental cues (an apparent exception will be discussed below). Thus, it would seem counterproductive to inactivate Ure2p at a random time, even though its normal function is to become inactive when good nitrogen sources are lacking in the medium, the appropriate time for this to happen. Similarly, allowing translational readthrough of all mRNAs by inativation of Sup35p does not seem to be a useful way to regulate gene expression. Nonetheless, there have been suggestions that the [PSI+] prion, and even [URE3], are beneficial to yeast. A report that [PSI+] made cells resistant to heat or high ethanol concentrations [37] was not reproduced in a subsequent study [38], and the advantages of [PSI+] reported in that study were not reproduced (using the same strains) in a further study [39]. A report of specific induction of [PSI+] generation by certain unfavorable conditions [40] was, again, not reproducible [41], but even in the original report, [PSI+] was usually unfavorable for growth under the conditions reported to induce its appearance [40], suggesting that this is not an adaptive response.

Even deadly viruses are found in natural populations because infectious agents can spread and outstrip the damage they do to their hosts. An advantageous infectious agent would surely be nearly universal in its distribution because infectivity and benefit to the host would be working together, instead of opposition. Thus, an infectious agent that is rare in the wild must be detrimental to its host. To examine the issue of whether yeast prions are adaptive or detrimental, we surveyed 70 wild yeast strains, and found that none carried either [URE3] or [PSI+], although each of the known selfish DNA/RNA plasmids and viruses of yeast were found in varying proportions of the wild isolates [42]. This implied that these prions, even their mildest variants, must be detrimental to their hosts. Another study confirmed the rarity of [PSI+] in the wild, but did not examine [URE3] [43]. This study did report frequent phenotypes of wild strains that were affected by growth on guanidine, and inferred that these phenotypes were due to unknown prions [43]. However, this report did not examine whether the phenotype was a result of guanidine’s known mutagenic effect on the mitochondrial DNA [7]. We were able to quantitate the detriment of acquiring a prion by comparing the frequency of the prions to that of 2 micron DNA, a selfish plasmid known to slow cell growth 1–3% [44–46], but nevertheless found in over half of wild isolates [42]. We find that the mildest forms of the [PSI+], [URE3] and [PIN+] prions must have a >1% detrimental effect on growth and/or survival to account for their limited occurence in the wild [41].

In trying to understand the overall impact of prion formation, one must consider the entire range of variants that can be formed. This range is as yet only beginning to be explored. In isolation of any prions, it is axiomatic that a lethal prion would not be recovered unless some special measures were taken to detect it. Sup35p is essential for growth, but the N-terminal prion domain is not essential. Low level expression of the essential C-terminal part of Sup35 lacking the prion domain allowed us to detect the formation of lethal [PSI+] prions [47]. Indeed, lethal and near - lethal variants of [PSI+] comprised more than half of total isolates, showing that [PSI+] generation is not generally a favorable event.

The method by which we isolated lethal [PSI+] variants implied that they were lethal because nearly all of the Sup35p joined the amyloid, leaving an insufficient amount for translation termination. This is not the mechanism of lethality of the TSE diseases, since the PrP protein is completely non-essential. Rather, the prion form of PrP has some toxic effect. Ure2p is likewise non-essential and, depending on the strain background, deletion of URE2 may not even slow cell growth. Using such a strain, we found that many [URE3] isolates were extremely toxic, slowing cell growth dramatically [47]. While the mechanism of [URE3] toxicity remains unclear, these results again show that acquisition of a yeast prion, like mutation of a chromosomal gene, is not generally a beneficial event.

The prion domains of Ure2p and Sup35p are perhaps misnamed as they have normal functions. The Sup35p prion domain is involved in the mRNA turnover process, interacting with polyA binding protein and the polyA degrading enzymes. In the absence of the Sup35p ‘prion domain’, turnover of all mRNAs is much slower [48]. The prion domain of Ure2p is necessary for the stability of Ure2p in vivo, and so plays an important role in the molecule [49]. Thus, these domains are not conserved for the purpose of prion formation. In fact, the Ure2p’s of Kluyveromyces lactis or Candida glabrata that have an N-terminal “prion domain”-like region cannot, in fact, form prions, even in their native hosts [50, 51]. Prion-forming ability appears to be sporadically arising, rather than conserved.

Sequence comparisons of Ure2p and Sup35p show that the prion domains change much more rapidly than does the remainder of either molecule [52–56]. These accumulated changes result in barriers to transmission of the [URE3] and [PSI+] prion between species [52, 55]. Even within the genus Saccharomyces there are barriers to transmission of [URE3] [57] and [PSI+] [58] based on sequence differences in the prion domains. There are even polymorphs in the Sup35p prion domain of wild S. cerevisiae isolates [54, 59, 60] that result in intra-species barriers to transmission of [PSI+] [60]. Indeed, the rare wild isolates of [PSI+] are sensitive to these barriers [60]. Just as in humans, where polymorphism at residue 129 of PrP results in a barrier to prion transmission [61], it is likely that these Sup35p polymorphisms were selected because of the transmission barrier they provide [60]. This suggests that prion acquisition is unfavorable, not advantageous.

The above is not to contend that no prions can be beneficial, nor that no yeast prions can be beneficial. We were the first to hail [62] a beneficial prion, the [Het-s] prion of the filamentous fungus Podospora anserina, which has a role in heterokaryon incompatibility, a normal function of filamentous fungi [63]. [Het-s] is striking in that it is present in 80% of wild strains of the appropriate chromosomal genotype, and forms only a single prion variant, as expected for a functional prion. This contrasts with the properties of those yeast prions investigated to date.

Prion Clouds.

Prions, like viruses and plasmids, are expected to be able to segregate during growth if they are mixed to begin with, and to mutate. Work on mammalian prions has suggested that prions exist as a ‘cloud’, a mixture of variants in one animal or cell [64, 65], and classic experiments with mouse scrapie showed that prions can mutate (without change in the protein sequence) when subjected to selective pressure [66].

We found that [PSI+] transmission to different polymorphs of Sup35p depends strongly on the prion variant examined [60]. A given [PSI+] clone can be a mixture of variants with different patterns of transmission, and simple mitotic growth results in segregation of clones with different transmission patterns [67]. However, extensive growth of a clone with one pattern results in production of clones, some of which have different patterns, indicating that there is variant mutation occurring as well [67]. The details of the experimental procedure were such that the changes were occurring without selection or interference with amyloid propagation. Interestingly, either ‘strong’ [PSI+] or ‘weak’ [PSI+] can have any of the four transmission phenotypes. Thus, yeast prions can exist as a cloud, a mixture of variants each of which is relatively stable, but that can interconvert over time under non-selective conditions. Further definition of the possible prion variant phenotypes and the mechanisms by which these phenotypes are produced will be an important area of future work.

Chaperones and Prions.

The involvement of chaperones, proteins that aid in the correct folding of other proteins, in prion phenomena began with Chernoff’s identification of Hsp104 as a gene whose overproduction could cure [PSI+], although [PSI+] was not then known to be a prion [68]. Either overproduction or deficiency of Hsp104 cures [PSI+] [69]. Hsp104 working together with Hsp70s and Hsp40s (chaperone families) break amyloid filaments producing new seeds, an essential step in prion propagation (reviewed by [70]). Other chaperones, co-chaperones and nucleotide exchange factors have also been found to play key roles in prion propagation (reviewed by [71])..

Btn2p and Cur1p and Prion aggregate collection.

A screen for proteins whose overproduction could cure the [URE3] prion produced two somewhat homologous proteins, Btn2p and Cur1p [72]. Deletion of both genes produced substantial effects on prion generation and increased prion stability, indicating that the normal levels of these proteins also affect prions [72]. In cells in the process of being cured of [URE3] by overproduced Btn2p, Ure2p and Btn2p were often co-localized in a single site, suggesting that sequestration of amyloid filaments might prevent one of the progeny cells from getting any seeds [72]. This was not observed in the case of curing of [URE3] by Cur1p. The involvement of the chaperone Hsp42 in aggregated protein sequestration was documented by Bukau’s group [73], and the interaction of Hsp42 with Btn2p (but not Cur1p) was shown by Malinovska et al. [74]. Several differences between Btn2p action and Cur1p action have been documented in spite of their sequence similarity. Mammalian cells have a centrosome-proximal structure at which aggregated proteins are collected [75]. In studying huntingin aggregation in yeast, an aggresome-like structure was also identified [76]. Further work will be needed to completely define the actions of these systems, and their relation to each other.

Perspectives.

In just under 20 years, the yeast prion field has reached a point where a great deal of information has accumulated and some of the central messages can be understood. However, there remain many very important areas that are largely unclear or controversial. The detailed structure of yeast prion variants has so far resisted all attempts at solution. What is the scope of prion variants, and how do they produce their many effects? The dramatic difference between a mild [PSI+] or [URE3] and a lethal form of the same prion suggests that there may be parallels in mammalian amyloidoses. While Alzheimer’s disease may be fatal, patients deceased from another cause are often found with extensive amyloidosis, but no brain damage. This is often taken as evidence that amyloid is not the toxic species, but it could well be that such patients have a mild amyloid variant, analogous to mild [PSI+] or mild [URE3] and a range of Abeta amyloid structures have been defined [31].

The functions of Btn2p and Cur1p and yeast aggresomes in dealing with aggregated proteins of various sorts remains to be elucidated. The many effects of elevated and depressed chaperones are far from understood. Yeast prions are largely pathogenic, but it remains possible that some functional yeast prions, like [Het-s] of Podospora anserina, will be found. Certainly the non-amyloid prion [BETA], the active form of vacuolar protease B, is beneficial to the cell. With an increasing number of prions identified in S. cerevisae, it seems likely that many prions that are not simply homologs of those found in this yeast will be found in other organisms if appropriate searches are made. It is perhaps most striking that proteins can act as genes - not often - but enough to make it interesting.