Chaperone Proteins Select and Maintain [PIN+] Prion Conformations in Saccharomyces cerevisiae

Background: Prion proteins adopt different conformations, known as variants, each with a distinct phenotype. Results: Deletion of specific chaperone genes (HSC82, AHA1, CPR6, CPR7, SBA1, TAH1, SSE1) alters established [PIN+] variants in S. cerevisiae. Conclusion: Chaperone proteins have a role in determining prion variants. Significance: Chaperone activity helps to regulate cell prion phenotype. Prions are proteins that can adopt different infectious conformations known as “strains” or “variants,” each with a distinct, epigenetically inheritable phenotype. Mechanisms by which prion variants are determined remain unclear. Here we use the Saccharomyces cerevisiae prion Rnq1p/[PIN+] as a model to investigate the effects of chaperone proteins upon prion variant determination. We show that deletion of specific chaperone genes alters [PIN+] variant phenotypes, including [PSI+] induction efficiency, Rnq1p aggregate morphology/size and variant dominance. Mating assays demonstrate that gene deletion-induced phenotypic changes are stably inherited in a non-Mendelian manner even after restoration of the deleted gene, confirming that they are due to a bona fide change in the [PIN+] variant. Together, our results demonstrate a role for chaperones in regulating the prion variant complement of a cell.

The mammalian prion protein, PrP, was originally identified as the causative agent of a group of neurodegenerative disorders collectively known as the transmissible spongiform encephalopathies (1). When the prion-determining domain (PrD) of PrP misfolds, PrP can switch from a non-infectious conformation (PrP C ) to a conformation prone to forming selfpropagating, ␤-sheet-rich, amyloid polymers (PrP Sc ) (1,2). These amyloids spread by catalyzing the transformation of PrP C into PrP Sc , eventually forming large aggregates (for review, see Ref. 3).
Although genetic polymorphisms of the PrP gene do affect its disease pathology (4,5), distinct sets of symptoms arising from genetically identical PrP Sc , called strains, have been described (6,7). Protease treatment of different strains of PrP Sc aggregates revealed that they differed by the size and composition of their amyloid core region, suggesting that it is specific conformations of the amyloid that give rise to PrP Sc strains (8).
Prions have also been characterized in non-mammalian model organisms, including bakers' yeast, Saccharomyces cerevisiae, where they have been demonstrated to occur frequently in the wild (9 -13). Yeast prions can also form different strains, called variants (14 -16). Studies have shown that different prion conformations distinguish each prion variant (17)(18)(19).
The prion protein Rnq1p can adopt different variants, each with a distinct phenotype. Although the function of its nonprion conformation remains unknown, the prion conformation of Rnq1p, [PIN ϩ ], acts to help another yeast prion, Sup35p/ [PSI ϩ ], adopt its own prion state (20,21). Rnq1p has a priondetermining domain between amino acids 153 and 405, with a non-prion domain N terminus (11 (16). Another [PIN ϩ ] variant-linked phenotype is observed when GFP-tagged Rnq1p is overproduced in vivo (22). [PIN ϩ ] high strains form multiple Rnq1-GFP foci per cell, whereas [PIN ϩ ] medium and [PIN ϩ ] low strains generally form only a single Rnq1-GFP focus. Another phenotypic difference between [PIN ϩ ] variants is the size and stability of their amyloid aggregates. [PIN ϩ ] high contains less stable aggregates that break down into smaller subparticles when heated, as opposed to [PIN ϩ ] low or [PIN ϩ ] medium aggregates, which remain stable when heated (16,23,24). When two [PIN ϩ ] variants are introduced into the same cell, either through mating or cytoduction, the diploid and all haploid progeny adopt the phenotype of the dominant variant.
[PIN ϩ ] high is dominant over [PIN ϩ ] medium , which is in turn dominant over [PIN ϩ ] low (16). These multiple, distinct phenotypes make [PIN ϩ ] an ideal model system with which to study the etiology of prion variants.
Mutations in a prion protein can affect its amyloid structure and, through that, the type of variant it adopts (25,26). Still, variants can arise from prion proteins with identical sequences (14 -16, 27), suggesting that other cellular factors may influence the variant conformation that a given prion will adopt. Chaperone proteins are strong candidates for such variant regulating factors. Chaperones are known to affect the conformation of a wide array of client proteins (for review, see Ref. 28) and have been implicated in other aspects of prion biology, including the de novo formation, propagation, and curing of prions (for review, see Ref. 29 -31). Additionally, changes to chaperone activity and levels have been shown to affect prion variants. Yeast strains expressing N-and C-terminal truncations of the primary stress response transcriptional regulator, Hsf1p, which were shown to increase Hsp104p and decrease Hsp90 levels, respectively, preferentially formed specific [PSI ϩ ] variants upon de novo induction (32). Likewise, strains over-or underexpressing SSE1, which is important to Hsp70 activity, gave rise to specific [PSI ϩ ] variants when induced (33). To date, only one genetic mutation has been reported to lead to a change in a pre-existing variant. Sondheimer et al. (27) demonstrated that deletion of the Sis1p G/F domain altered Rnq1-GFP aggregation pattern in a manner stably propagated even after reintroduction of wild-type Sis1p.
Here, we report the findings of our investigations into the actions of chaperone proteins upon already established [PIN ϩ ] variants. We found that disruption of several chaperone genes gives rise to shifts in [PIN ϩ ] variant-linked phenotypes. Genetic analysis showed that the phenotypic shifts are inherited in a non-Mendelian manner, confirming that a bona fide change in [PIN ϩ ] variant was achieved. Our findings provide evidence that chaperones can affect established prion variants and highlight a potential role for chaperones in regulating prion-linked phenotypes through their modulation of prion variants.

EXPERIMENTAL PROCEDURES
Yeast Strains, Culture, and Genetic Manipulation-S. cerevisiae strains used in this study are listed in supplemental Table  S1. All yeast strains were cultured at 30°C with the exception of diploids undergoing sporulation, which were cultured at 25°C. Media were as follows: YEPD (1% yeast extract, 2% peptone, 2% glucose); CSM (0.67% yeast nitrogen base without amino acids, 2% glucose, 1 ϫ Complete supplement mixture (Bio 101, Vista, CA)); CSM auxotrophic marker growth medium (same as CSM but containing 1 ϫ CSM minus the appropriate auxotrophic selection (ϪADE, ϪHIS, ϪLEU, ϪURA, ϪLYS-URA, ϪTRP-URA)); CSM auxotrophic marker induction medium (same as CSM but with 2% galactose in place of 2% glucose). For solid media, the same recipes were used with 2% agar added. Deletions were made using a HIS3 cassette as described (34) and confirmed by PCR. Isogenic MAT␣ strains were made using YCpGAL::HO and mating-type switching (35). Mating, diploid sporulation, and dissection of haploids were performed as previously described with modified sporulation medium (0.3% potassium acetate, 0.02% raffinose) (36). Haploid progeny of dissections were tested for the presence of a gene deletion by culture on CSM-HIS plates.
Yeast Plasmids and Cloning-Plasmids used in this study were constructed as follows; pYES2.0-SUP35NM-GFP was constructed by ligating the sequence encoding enhanced GFP (37) in-frame with the 3Ј-end of base pairs 1-762 of the SUP35 gene into pYES2.0 (Invitrogen). pYES2.0-GFP was similarly constructed with only the GFP-encoding sequence. pYES2.0-Rnq1-GFP was constructed by amplifying sequence coding for Rnq1p C-terminally tagged with GFP from genomic DNA of a commercially available GFP-tagged library strain (Invitrogen) and ligating it into pYES2.0. pGREG535-Rnq1 was constructed by amplifying RNQ1 and inserting it into pGREG535 according to the Drag & Drop protocol (38). Plasmids were verified by sequencing. Escherichia coli and yeast were transformed with plasmid using standard chemical transformation protocols (34,39).
Assay for Nonsense Suppression-To quantify [PSI ϩ ] induction efficiency based upon nonsense suppression, Sup35NM-GFP or GFP alone was overproduced by culturing strains harboring pYES2.0-SUP35NM-GFP or pYES2.0-GFP, respectively, in liquid CSM medium for 24 h followed by subculturing in liquid CSM induction medium for 48 h. When testing primary deletions and diploids, cell cultures were normalized to an A 600 of 1.0, and when testing tetrads, cultures were normalized to an A 600 of 0.25. After normalization, 10-fold serial dilutions were made. 5 l of each dilution were spotted onto both CSM and CSMϪADE plates, and colony-forming units (cfu) were counted after 2 and 4 days, respectively. Percent [PSI ϩ ] induction efficiency was calculated as cfu (CSMϪADE)/cfu (CSM) ϫ 100. Four to six independent experiments were performed for each strain.
Assay for Plasmid Retention-To determine if a given strain retained pYES2.0-SUP35NM-GFP, induced strains were plated for single colonies on both CSM and CSMϪADE, and colonies from both platings were then replica-plated onto CSM and CSM-URA media. Percent plasmid retention was calculated as cfu (CSM-URA)/cfu (CSM) ϫ 100. Between 100 and 200 colonies were compared for each strain.
Assay for Prion Curing-To test induced [PSI ϩ ] strains for curability, colonies growing on CSMϪADE medium plates were streaked onto curing plates (YEPD ϩ 3 mM guanidine HCl) and allowed to grow at 30°C for 3 days. Putatively "cured" colonies were then selected based on their pigmentation, restreaked onto CSM and CSMϪADE plates, and allowed to grow at 30°C for 2 and 4 days, respectively. At least four colonies were tested for each strain.
Characterization of Rnq1-GFP Aggregates in Vivo-Strains carrying pYES2.0-Rnq1-GFP were cultured for 24 h in liquid CSM growth medium then subcultured in liquid CSM induction medium for 24 h. The number of Rnq1-GFP foci per cell was then quantified by acquiring random wide-field micrographs of cells using an Olympus IX-80 fluorescence microscope. The frequency of multiple foci was expressed as a percentage of all cells containing multiple Rnq1-GFP foci. 800 -1000 cells were counted and categorized over the course of 4 independent experiments.
Protein Analysis-Strains carrying pGREG535-Rnq1 were cultured for 24 h in liquid CSM growth medium, then subcultured in liquid CSM induction medium for 8 h. Protein samples were prepared and pelleted as described (24). The pellet fraction, enriched for insoluble HA-Rnq1p, was heated briefly at 55°C and analyzed by semi-denaturing detergent-agarose gel electrophoresis (SDD-AGE) 3 (24,40). Blots were probed with anti-HA antibody (F-7 Sc7392, Santa Cruz Biotechnology, Santa Cruz, CA) and detected with HRP-conjugated antimouse IgG (GE Healthcare).
Yeast Two-hybrid Analysis-Yeast two-hybrid analysis was performed using the yeast strain HF7c and the plasmids pGAD424 (prey) and pGBT9 (bait) (Clontech, Mountain View, CA) following standard protocols (41). Genes encoding Rnq1p or chaperone proteins were ligated in-frame into pGAD424 and pGBT9 for expression. Interactions were scored by growth of cells on CSM-HIS medium after 4 days of incubation relative to growth on YEPD. At least three independent experiments were done for each tested pair of proteins.  (Fig. 1A). We measured the relative [PSI ϩ ] induction efficiencies of deletion strains using a nonsense suppression assay (Fig. 1B) as described under "Experimental Procedures." In brief, we overproduced Sup35NM-GFP using the galactose-driven expression vector pYES2.0-SUP35NM-GFP and quantified cfu on CSMϪADE medium plates relative to cfu on CSM medium plates. ADE-competent colonies were confirmed to be prion-linked because they lost ADE competence after treatment with guanidine HCl (supplemental Fig. S1). Also, the extent of [PSI ϩ ] induction in a strain being dependent on the levels of plasmid retention by that strain was excluded as a possibility, as both wild-type and deletion strains retained plasmid at levels between 90 and 95% whether or not the cells were ADE-competent (data not shown). We first quantified the localization of Rnq1-GFP in all the deletion strains affected in [PSI ϩ ] induction efficiency ( Fig. 2A). When GFP-tagged Rnq1p was overproduced from the galactose-driven plasmid pYES2.0-Rnq1-GFP, ϳ75% of focus-containing wild-type [PIN ϩ ] low and wild-type [PIN ϩ ] medium cells contained a single Rnq1-GFP focus, whereas ϳ25% contained more than one focus. ϳ50% of wild-type [PIN ϩ ] high focus-containing cells contained multiple foci, in agreement with previ-ous findings (22  A, Rnq1-GFP was overproduced in deletion strains of interest, and the frequency of multiple Rnq1-GFP foci in foci-containing cells was calculated as a percentage of total cells. Error bars represent S.E. Student's t tests were performed to compare the frequency of multiple Rnq1-GFP foci in cells containing foci with that of their parental wild-type strain. **, p Ͻ 0.01; ***, p Ͻ 0.001. B, the Rnq1p aggregate size in deletion strains of interest was compared by SDD-AGE. Samples were incubated at room temperature (top) or 55°C (bottom) before loading.

Chaperone-mediated Determination of [PIN ؉ ] Variants
We next characterized the sizes of Rnq1p amyloid subparticles in the deletion strains.  (23). Strains with a single focus display a relatively narrow range of large subparticles, whereas strains with multiple foci display subparticles ranging in size from monomer to as large as, or larger than, those found in strains with a single focus. Accordingly, we produced HA-tagged Rnq1p from the plasmid pGREG535-Rnq1 in the wild-type [PIN ϩ ] low , [PIN ϩ ] medium , and [PIN ϩ ] high strains and in these strains harboring a gene deletion. We then prepared samples enriched for proteins in an insoluble prion state and incubated them at room temperature and 55°C, as Rnq1p aggregates purified from strains with multiple Rnq1p foci, specifically [PIN ϩ ] high , have been shown to be more sensitive to changes in temperature and to degrade partially at moderately elevated temperatures, whereas other Rnq1p variant aggregates do not (24). Finally, we compared the sizes of the purified HA-Rnq1p amyloid aggregates using SDD-AGE (Fig. 2B) aggregates, also proved to be sensitive to temperature, degrading in part to a monomer at 55°C (Fig. 2B). It is important to note that the pattern of protein migration displayed by these partly degraded protein aggregates is easily distinguished from that produced by protein aggregates isolated from a [pin Ϫ ] strain. [pin Ϫ ] aggregates were monomeric and of low molecular weight, whereas partially degraded aggregates from [PIN ϩ ] strains still displayed species of high molecular weight. Our results suggest that deletion of specific chaperone genes leads to changes in the prion amyloid physical properties, thereby affecting its heat sensitivity.
Our results show a correlation between chaperone gene deletion strains displaying altered [PSI ϩ ] induction efficiency and those displaying changes in other [PIN ϩ ]-linked phenotypes. The tah1⌬ strains were exceptions. They were unaffected in the sizes of their HA-Rnq1p aggregates vis à vis the sizes of the aggregates in their corresponding wild-type [PIN ϩ ] variant background. Also of note is that although the sse1⌬ strain in the [PIN ϩ ] low background showed multiple Rnq1-GFP foci per cell, its aggregates were limited to a narrow range of large sizes.
Deletion-induced Phenotypes Are Inherited in a Non-Mendelian Manner-The [PIN ϩ ] variant-linked phenotypic changes we observed in strains deleted for chaperone genes could be due directly to the effects of gene deletion or could be due simply to changes in the cell chaperone complement and to the overproduction of tagged protein resulting from our methodology. For example, loss of a chaperone could impair the ability of the cell to deal with the overexpression of tagged Rnq1p, leading to formation of denatured, non-prion aggregates that are detectable in our assays as multiple Rnq1-GFP foci and/or low molecular weight aggregates in SDD-AGE. To eliminate this trivial explanation and to confirm that it is indeed deletion of the chaperone genes that gives rise to the observed [PIN ϩ ] variant shifts, we reintroduced a wild-type copy of a deleted chaperone gene by crossing a deletion strain with an isogenic wild-type strain. This wild-type strain was [pin Ϫ ] so as to avoid any convolution related to introducing dominant [PIN ϩ ] variants (16). If the effects of deletions were specific to the deletion of the gene, then restoring the chaperone gene would be expected to restore the original [PIN ϩ ] variant phenotype. If on the other hand a permanent change in the [PIN ϩ ] variant had occurred in the deletion strain, then the phenotypic changes should persist after reintroduction of the chaperone gene. Fig. 3 shows that, for the most part, [PSI ϩ ] induction efficiencies, the frequency of Rnq1-GFP foci, and the sizes of Rnq1p aggregates of diploid strains did not revert to the phenotypes of the original wild-type strains from which the deletion strains were derived. Instead, these diploid strains maintained the phenotypes of the deletion strains. The only exceptions were diploids arising from matings with sse1⌬ in the [PIN ϩ ] low background and with tah1⌬ in both the [PIN ϩ ] low and [PIN ϩ ] medium backgrounds. The increased frequency of Rnq1-GFP foci in the original haploid strains deleted for SSE1 or TAH1 was eliminated in the diploid strains, suggesting that these changes were [PIN ϩ ]-independent. It should be noted that under our experimental conditions we were unable to statistically distinguish [PIN ϩ ] low from [PIN ϩ ] medium based upon [PSI ϩ ] induction efficiencies (supplemental Fig. S2). As such, we could categorize our results only as being consistent with either [PIN ϩ ] low/medium or [PIN ϩ ] high levels.
There was the remote possibility that spurious mutations had been introduced into our deletion strains and had produced dominant effects mimicking an apparent variant switch. If this were the case, it would be expected that this trait would be co-inherited in a Mendelian manner as opposed to the non-Mendelian manner of a true change in variant. To eliminate the possibility of introduced spurious mutations, we sporulated and dissected our diploid strains, yielding wild-type and deletion-carrying haploid progeny. The presence of the chaperone gene deletion cassette (HIS3) was detected by growth on CSM-HIS medium and showed a 2:2 ratio in the progeny as expected (data not shown). We then measured [PSI ϩ ] induction efficiency of the dissected tetrads by nonsense suppression assay (Fig. 4). The specificity of this assay was demonstrated by mating wild-type strains carrying different [PIN ϩ ] variants with a strain deleted for HSP104, which is required for both [PSI ϩ ] and [PIN ϩ ] maintenance (20,42). The wild-type haploid progeny were inducible at levels similar to their parental strains, whereas the hsp104⌬ haploids were unable to be induced. When the chaperone gene deletion strains were mated with the wild-type [pin Ϫ ] strain, we found that the haploid progeny of 24 tetrads all maintained the same [PSI ϩ ] induction efficiency as the parental deletion strain regardless of the presence or absence of the chaperone gene of interest. These levels were consistent with either [PIN ϩ ] low/medium or [PIN ϩ ] high levels. Haploid progeny were found to retain pYES2.0-Sup35NM-GFP at a rate consistent with the parental strains, with no strain losing or retaining the plasmid at a markedly higher level than any other strain. Also, putative [PSI ϩ ] colonies that arose after [PSI ϩ ] induction were shown to be curable by growth on medium containing guanidine HCl (supplemental Fig. S3).
The Rnq1p aggregates of wild-type and mutant haploid progeny derived from the mating of deletion strains that gave rise to [PIN ϩ ] variant-related phenotypic changes were characterized using SDD-AGE (Fig. 5). We found that the deletion-induced changes in aggregate size were stable even after tetrad dissection. Deletion strains that exhibited no change in aggregate size were found to be consistent with the wild-type strain in which they were made (data not shown). Taken  Rnq1p Interacts Physically with Chaperone Proteins-We performed a yeast two-hybrid analysis to investigate possible physical interactions between our chaperones of interest and Rnq1p (Fig. 7). With growth on CSM-HIS medium as a reporter induction efficiency by nonsense suppression assay on ϪADE medium. Wildtype strains were mated with hsp104⌬ strains to demonstrate genetic specificity for this assay. Wild-type (ϩ) and deletion-carrying (Ϫ) haploid progeny are presented. Twenty-four tetrads were analyzed per diploid.

Chaperone-mediated Determination of [PIN ؉ ] Variants
of interaction, we detected a previously documented interaction between Rnq1p and Tah1p (43). We also identified two novel interactions: Cpr7p and Sba1p with Rnq1p. No interaction was detected between the remaining chaperones and Rnq1p. When we characterized the [PIN ϩ ] variant of our yeast two-hybrid strain (HF7c) by SDD-AGE and localization of Rnq1-GFP (supplemental Fig. S5), we found HF7c to be [pin Ϫ ], suggesting that the interactions we detected occur when Rnq1p is in its non-prion conformation.

DISCUSSION
Although recombinant S. cerevisiae prion proteins do not require cofactors to misfold into multiple infectious variants in vitro, they need the activity of different chaperone proteins to propagate stably in vivo. Most prominent among these chaperones are Hsp104p, Sis1p, and members of the Ssa subfamily (for review, see Refs. 29 -31). Together these chaperones facilitate the fragmentation of growing amyloid fibrils, thereby exposing more fibril growing ends and generating more infectious prion seeds. It has been proposed that conformational differences between variants alter their susceptibility to fragmentation and their rate of fibril growth and that equilibrium between these processes determines the stability of prion propagation and the strength of prion phenotype (18,44,45). For example, compared with [PSI ϩ ] weak , [PSI ϩ ] strong has a smaller amyloid core that fragments more easily, allowing for the generation of more growing ends and a higher rate of Sup35p incorporation into a greater number of prion seeds (18). These seeds in turn lead to increased nonsense suppression and more stable propagation of [PSI ϩ ]. Chaperones have also been implicated in variant determination, with alteration of Hsf1p or Sse1p activity affecting the de novo induction of the [PSI ϩ ] variant and truncation of Sis1p leading to stable changes in the established [PIN ϩ ] variant (27,32,33).
Our study provides further evidence that chaperones are important for the selection of prion variants. Deletion of AHA1  (46). However, such a scenario is unlikely in our case, because the changes we observed in [PSI ϩ ] induction in our chaperone gene deletion strains were also accompanied by changes in the size and localization pattern of Rnq1p, consistent with a shift in [PIN ϩ ] variant. Additionally, these phenotypic shifts persist after sporulation of diploids into wild-type and deletion-carrying haploid progeny upon reintroduction of the deleted chaperone gene. Together, our results demonstrate that the chaperone gene deletion-induced phenotypes we observed are due to stable shifts in the [PIN ϩ ] variant.
Like other chaperone gene deletions, deletion of TAH1 or SSE1 in the [PIN ϩ ] low or [PIN ϩ ] medium backgrounds gave rise to changes in [PIN ϩ ] variant-dependent phenotypes. However, in contrast to what was observed for the other chaperone gene deletion strains, not all of the tested phenotypes were altered in TAH1 or SSE1 gene deletion strains, and of those phenotypes that were altered, not all were maintained after the wild-type gene was restored. This suggests that at least some of the changes observed in the tah1⌬ and sse1⌬ strains are not the result of shifts in the [PIN ϩ ] variant and that any putative var- Wild-type strains were mated with hsp104⌬ strains to demonstrate genetic specificity for this assay. Wild-type (ϩ) and deletion-carrying (Ϫ) haploid progeny are presented. The hsp104⌬ samples and wild-type controls were run on a continuous gel and exposed for a longer period. Twenty-four tetrads were analyzed per diploid.
iant shift that may have occurred in these deletion strains does not correspond to a previously characterized variant (16,22 (32).
We detected physical interactions of Rnq1p with Cpr7p, Sba1p, and Tah1p but not with the other chaperones. This suggests that some of these chaperones may act upon Rnq1p independently of Hsp90. Cpr7p, for example, has intrinsic proline isomerase activity (52,53). Rnq1p contains three proline residues: one in the N-terminal region, a region shown to affect prion propagation (54), and two in a putative loop region between the ␤-sheets of the amyloid core (55). Isomerization of these prolines could potentially affect Rnq1p conformation.
Additionally, as numerous physical interactions between these chaperones and other chaperone complexes have been documented, changes in the levels of these chaperones may affect how Hsp40s, Hsp70s, and/or Hsp104p interact with Rnq1p, resulting in variant change. Cpr7p, for example, has been shown to interact with Hsp104p in a manner that is not essential for its thermotolerance activity (56,57). Additionally, inhibition of Hsp90 ATPase activity has been shown to increase the levels of both Hsp104p and Hsp70 (58). The effect of SSE1 deletion that we report here also implicates Hsp70s in variant change, as SSE1 encodes an important Hsp70 nucleotide exchange factor (59). Fan et al. found that manipulation of Sse1p levels affected the [PSI ϩ ] variant (33), although our sse1⌬ strain did not display the same disposition toward an unstable weak [PSI ϩ ] variant. The difference between our and their results could be due to [PIN ϩ ] variant-specific effects, as Fan et al. (33) did not characterize the [PIN ϩ ] variant of their strain. It is also interesting that it has been reported that mutations in Sis1p give rise to an increase in Rnq1-GFP foci (27), similar to what we observed. It may be that our gene deletions indirectly impaired the activity of Sis1p and/or its association with Rnq1p.
How the loss of specific chaperones alters the levels and activities of other chaperones, in addition to their interaction with Rnq1p, will be an important avenue of future investigation to clarify the mechanisms underlying the variant changes that we observed. Chaperones could mediate [PIN ϩ ] variant changes by altering the conformation of Rnq1p. For example, chaperones could regulate the folding of monomeric Rnq1p in such a way as to predispose it to adopt a specific variant upon de novo formation or upon encountering prion seeds. More likely, because we observed changes to established variants, chaperones could work together to affect the conformation of existing amyloid polymers. For this to be effective, only the growing ends of polymers need be remodeled. Alternatively, if multiple or unstable variants are present at the same time in the cell, as has been shown to occur for [PSI ϩ ] (60), changes in the chaperone environment could alter the rates of amyloid polymer fragmentation and/or prion seed generation in a variant-specific manner. In this way, one variant could be selected over others.
In closing, we have demonstrated that altering the chaperone complement of a cell can alter existing prion variants without the introduction of exogenous prion material. By modulating existing prion variants, the cell can maintain prion seeds within the cell while mitigating potential negative effects of stronger prion phenotypes. Also, when environmental pressures demand, the existing prion variant could be quickly altered to provide a more advantageous phenotype to the cell. In light of recent findings reporting the prevalence of prions in wild strains of yeast as well as the apparent survival advantages that they bestow (13), prion variant regulation represents a powerful mechanism for modulating a cell response and adaptability to changing environmental conditions.