Increased Expression of Hsp40 Chaperones, Transcriptional Factors, and Ribosomal Protein Rpp0 Can Cure Yeast Prions*

The Sup35 (eRF3) translation termination factor of Saccharomyces cerevisiae can undergo a prion-like conformational conversion, thus resulting in the [PSI +] nonsense-suppressor determinant.In vivo this process depends critically on the chaperone Hsp104, whose lack or overexpression can cure [PSI +]. The use of artificial prion [PSI + PS] based on a hybrid Sup35PS with prion domain from the yeast Pichia methanolicaallowed us to uncover three more chaperones, Ssb1, Ssa1, and Ydj1, whose overexpression can cure prion determinants. Here, we used the [PSI + PS] to search a multicopy yeast genomic library for novel factors able to cure prions. It was found that overexpression of the Hsp40 family chaperones Sis1 and Ynl077w, chaperone Sti1, transcriptional factors Sfl1 and Ssn8, and acidic ribosomal protein Rpp0 can interfere with propagation and manifestation of [PSI + PS] in a prion strain-specific manner. Some of these factors also affected the manifestation and propagation of conventional [PSI +]. Excess of Sfl1, Ssn8, and Rpp0 influenced at least one of the tested chaperone-specific promoters,SSA4, HSP104, and model promoters, with either the heat shock or stress response elements. Thus, the induction of chaperone expression by these proteins could explain their prion-curing effects.

There are several proteins in the yeast Saccharomyces cerevisiae, which, similar to mammalian prions, can undergo autocatalytic conformational rearrangement. This process can last stably for many cellular generations, resulting in some cases in heritable phenotypes (1). The most studied of the yeast prion proteins is the translation termination factor Sup35 (eRF3), which is related to the [PSI ϩ ] determinant. The prion state of Sup35 is characterized by its aggregation and partial inactivation, which causes the nonsense-suppressor [PSI ϩ ] phenotype (2,3). Different [PSI ϩ ] isolates ("strains") may vary in the efficiency of suppression, which presumably reflects different structures of Sup35 aggregates. The [PSI ϩ ] strains with strong suppression ("strong") are highly stable, the strains with weak suppression ("weak") show decreased mitotic stability (4,5).
Studies of the yeast prions confirmed that they reproduce the basic properties of their mammalian prototype and also show profound similarities to another related phenomenon, amyloids. In particular, purified Sup35 formed amyloid fibers in vitro (6). Therefore, Sup35 could be used as a convenient model for studying the basic features of both the prion and amyloid phenomena, and it is likely that the cellular factors interfering with the propagation of yeast prions are similar to those that can interfere with the prion or amyloid formation in animals.
Although in vitro Sup35 alone is sufficient for its conformational rearrangement, in vivo some cellular proteins can participate in this process and modulate its efficiency. The most important of them is the Hsp104 chaperone, which is strictly required for the [PSI ϩ ] propagation, although its excess interferes with [PSI ϩ ] (7). The chaperones of the Hsp70 family showed weaker effects; the overexpression of Ssb1 cured weak [PSI ϩ ] strains (8,9), whereas overproduction of Ssa1 interfered with the [PSI ϩ ] curing by excess Hsp104 (10). Mutational alterations of Hsp104 and Ssa1 could also cause [PSI ϩ ] elimination (7,11). Construction of heterologous [PSI ϩ ] ([PSI ϩ PS ]) provided a sensitive instrument for identification of novel factors involved in the prion propagation. [PSI ϩ PS ] is based on a hybrid Sup35, in which the native prion domain was replaced with its analog from the yeast Pichia methanolica (12). Compared with [PSI ϩ ], [PSI ϩ PS ] showed increased sensitivity to overexpression of the Hsp70 chaperones Ssa1 and Ssb1 and the Hsp40 chaperone Ydj1 but decreased sensitivity to the excess Hsp104 (9). The [PSI ϩ PS ] curing by chaperones showed remarkable prion strain specificity. The chaperones ranked differently by their curing efficiency for each prion strain. This finding supports the existence of different prion structures corresponding to prion strains and differential sensitivity of these structures to curing mechanisms associated with different chaperones. Thus, the sets of chaperones able to cure a given prion may differ for different prions and prion strains.
[URE3] prion was cured by overproduced Ydj1 but was insensitive to excess Hsp104 (13). Recent studies of [RNQ ϩ ] showed that it depends on another chaperone of the Hsp40 family, Sis1, which has not been found among the factors curing [PSI ϩ ] (14). Deletion of the inessential glycine-and phenylalanine-rich region of Sis1 abolished propagation of this prion. It should be noted that all mentioned chaperones are considered to be related functionally. Hsp70 and Hsp40 represent eukaryotic homologues of the bacterial chaperone system DnaK-DnaJ, and Ydj1 and Sis1 are presumed to be partners for Ssa1 (15). In vitro, Ssa1 and Ydj1 cooperated with Hsp104 to reactivate the aggregates of heat-denatured luciferase, with each of the chaperones being essential for this process (16).
To find additional factors that can influence the prion propagation in yeast and presumably in other organisms we decided to use the potential of the [PSI ϩ PS ] system to perform a sys-  (12). Yeast were grown at 30°C in rich (YPD) or synthetic (SC) media. Standard genetic methods were used (17). DNA transformation of lithium acetate-treated yeast cells was performed as described (18). To determine the [PSI ϩ ] loss, four transformants with each plasmid were grown on 5-fluoroorotic acid medium (17) to induce the plasmid loss and then cloned on YPD medium. The loss of [PSI ϩ ] or [PSI ϩ PS ] was scored as a ratio of redand white-colored colonies. At least 100 clones of each transformant were counted (300 or more for cases of less than 1% loss frequency). To test the invasive growth, cells were grown on YPD for 3 days and washed off the agar surface, and the plates were incubated for 1 more day.
Centromeric YCp111-SSA4pro-lacZ plasmid was obtained in two steps. The SSA4 promoter was amplified by PCR with oligonucleotides AGGTCAATCCGGTAGATGTG and GGCGTCGTTCTATTACCTTG and cloned into the SmaI site of YIp366 (23) immediately upstream of the lacZ gene. Then, the 4.2-kb BsrGI fragment of this plasmid containing the SSA4 promoter-lacZ fusion was inserted into the Acc65I site of YCplac111 (22). Integrative SUP35 promoter-lacZ fusion construct was made by cloning the 1.2-kb PvuII-PstI fragment of pEMBLyex-SUP35 (24) into the SmaI and PstI sites of YIp366. For genomic integration, this plasmid was cut with SnaBI. The pUKC1600 plasmid containing the lacZ gene under the control of HSP104 promoter (25) was a kind gift of F. Ness. The 4.5-kb NruI-HindIII fragment of pUKC1600 containing the HSP104-lacZ construct was inserted into the EheI-and HindIII-digested vector YCplac111 to generate the YCp111-HSP104-lacZ plasmid. The integrative HSE 1 -lacZ and STRE-lacZ fusion constructs based on pLS9 (26) were a kind gift of H. Ruis. The 2.3-kb PstI-Eco47III fragments of each construct encompassing the promoters and a part of lacZ were ligated with the 7.4-kb PstI-Eco47III fragment of the YCp111-SSA4-lacZ plasmid, resulting in the centromeric YCp111-HSE-lacZ and YCp111-STRE-lacZ plasmids used here.
Gene Disruptions-The disruption cassette for SSN8 was constructed as follows. The 2.3-kb PstI-EcoRI fragment of YEplac195-SSN8 was inserted into the corresponding polylinker sites of pBC KSϩ (Stratagene), forming the pBC-SSN8 plasmid. Then, the 1.2-kb PvuII-SmaI fragment of pJJ244 (27), containing the disruption marker URA3, was inserted inside the coding region of the SSN8 gene by BglII and SphI sites blunted with Klenow, forming the ssn8::URA3 cassette. The KpnI fragment from this cassette was used to disrupt the genomic SSN8 locus. Gene disruption was verified by Southern hybridization (19). For this, a 2.3-kb KpnI probe from the ssn8::URA3 cassette was hybridized with genomic DNA double digested with StuI and ScaI. For the SFL1 disruption, a 1.2-kb PvuII-SmaI fragment of pJJ244 was inserted into the YEplac195-SFL1 plasmid in place of the internal BsaBI fragment of SFL1, forming the sfl1::URA3 cassette. The NheI-SpeI fragment from this cassette was used to disrupt the genomic SFL1 locus. The disrup-tion was confirmed by Southern hybridization of 2.5 kb of NheI-SpeI probe from sfl1::URA3 cassette with genomic DNA digested with EcoRV.
Preparation and Analysis of Yeast Cell Lysates-Yeast cultures were grown in liquid YPD medium or in a medium selective for plasmid marker to an A 600 of 1.5. The cells were harvested, washed in water, and lysed by glass beads in buffer A (25 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM MgCl 2 , 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride). Cell debris was removed by centrifugation at 15,000 ϫ g for 10 min. To equalize the protein concentration in samples, the concentration was determined according to Ref. 28, where required.
To analyze the distribution of Rpp0, cells were treated with 0.1 mg/ml cycloheximide for 15 min before harvesting and then broken by glass beads in Buffer A, supplemented with 0.2 mg/ml cycloheximide and 10 mM MgCl 2 . The extracts were fractionated by centrifugation through a 7-40% sucrose gradient. Distribution of ribosomal RNA in the fractions was analyzed using 0.8% agarose electrophoresis with ethidium bromide. The presence of Rpp0 was revealed by Western blotting using a polyclonal antibody against Rpp0 (kind gift of J.-P. G. Ballesta). For Western blotting, proteins were separated on SDSpolyacrylamide gels according to Ref. 29 and electrophoretically transferred to nitrocellulose sheets. Bound antibody was detected using the Amersham Biosciences ECL system.
␤-Galactosidase Activity Assay-␤-Galactosidase activity was measured as described (30) except that the cells were permeabilized by subjecting them to two freeze/thawing cycles using liquid nitrogen and a water bath at 37°C.
Purification of Sfl1 and Mobility Shift Assay-The PvuII-HindIII fragment, encoding Sfl1 without the first 29 amino acids, was inserted into the Ecl136II (SacI) and HindIII sites of pET-30a (Novagen) to generate pET30a-SFL1 encoding His 6 -Sfl1 fusion. E. coli strain BL21 (DE3) pLysE (Novagen) transformed with this plasmid was grown to an A 600 of 0.5 at 30°C and then Sfl1 expression was induced by addition of isopropyl-␤-D-thiogalactopyranoside to 1 mM for 20 h at 17°C. Cells were harvested by centrifugation for 10 min at 2500 ϫ g, resuspended in 3 ml of TBS buffer (150 mM NaCl, 25 mM Tris-HCl, pH 7.8) with 0.2% Triton X-100, and subjected to two freeze/thawing cycles. Then DNase I was added to a final concentration of 150 g/ml, and phenylmethylsulfonyl fluoride was added to a final concentration of 1 mM. The cell suspension was incubated on ice for 20 min and then centrifuged at 12,000 ϫ g for 10 min at 4°C to remove cell debris. The His 6 -Sfl1 protein was purified from the supernatant using 0.3 ml of TALON resin (CLONTECH) according to the instructions from the manufacturer. The purity of the protein was confirmed by electrophoresis with Coomassie Blue staining.
Mobility shift assays were performed as described (31). To create labeled HSE and STRE DNA fragments, oligonucleotides 5Ј-gtccttcta-GAAGCTTC and 5Ј-ctgtccccTTACGTAA, respectively, were self-annealed by the complementary region (capital letters), and the noncomplementary region (lowercase letters) was filled in with Klenow in the presence of [ 32 P]␣-dATP. The His 6 -Sfl1 protein or yeast lysates were incubated with 1 ng (1 ϫ 10 5 cpm) of the labeled HSE or STRE fragments for 20 min at room temperature in 25 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 7 mM MgCl 2 , 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 1 mg of poly(dI-dC) nonspecific competitor DNA. Protein⅐DNA complexes were resolved by electrophoresis in a 4.5% non-denaturing polyacrylamide gel at 70 V in 0.5ϫ TBE buffer (89 mM Tris base, 89 mM boric acid, 2 mM EDTA) for 1.5 h. The gel was dried on filter paper (1 h, 80°C) and exposed to x-ray film overnight at Ϫ70°C.

A Screen for Prion-eliminating Factors-
The strain 5V-H19 and its derivative, PS-5V-H19, were used in this work. Both strains carry the ade2-1 UAA nonsense mutation and the SUQ5 (SUP16) UAA nonsense suppressor but differ from each other by the SUP35 gene.  (Fig. 1A). These plasmids were sequenced from the ends of genomic inserts, and the inserts were identified by comparison of the sequences with the yeast genome at www.ncbi.nlm.nih.gov/blast/. Because the inserts usually contained several genes, the genes affecting [PSI ϩ PS ] were identified by deletion analysis. The following genes were found: SIS1 (six clones, two independent plasmids), YNL077w (three clones, one independent plasmid), STI1 (one clone, one independent plasmid), SFL1 (two clones, one independent plasmid), SSN8 (three clones, one independent plasmid), RPP0 (three clones, one independent plasmid), SUP35 (six clones, three independent plasmid). The SUP35 gene product, Sup35, causes antisuppression, but not [PSI ϩ PS ] loss, because Sup35 does not aggregate in [PSI ϩ PS ] cells and does not interfere with Sup35PS prion aggregation (12). The YNL077w gene encodes a previously uncharacterized protein of the Hsp40 (DnaJ) family designated hereafter as Apj1 (anti-prion DnaJ). The other genes encode Hsp40 chaperone Sis1, chaperone Sti1, transcriptional factors Sfl1 and Ssn8, and acidic ribosomal protein Rpp0.
The effects of overproduction of these proteins, antisuppression and prion loss, were characterized for the independently isolated strains of hybrid and conventional [PSI ϩ ] (Table I). These effects did not correlate. For [PSI ϩ PS-1 ], only excess Sis1 showed significant antisuppression, whereas excess Rpp0 caused the highest prion loss but low antisuppression. It should be noted that, as it was observed previously (9), the relative efficiency of chaperones in prion curing varied depending on the [PSI ϩ ] strain and the origin of Sup35 prion domain. The antisuppressor effect of excess Sis1 was mediated by an increase in the levels of soluble Sup35PS (Fig. 1C), which indicates that Sis1 interfered with the prion conversion.
Disruptions of the SFL1 and SSN8 Genes-Although all previously described prion-curing factors represent chaperones, the Sfl1 and Ssn8 proteins belong to a different class, being involved in transcriptional regulation. Sfl1 is a homologue of Hsf1, a key transcriptional factor controlling the heat shock response. Ssn8 represents a human cyclin C homologue that belongs to the transcriptional mediator complex (32). Ssn8 and Sfl1 interact physically and cooperate in regulation of the SUC2 gene expression (33). To further characterize the effects of Sfl1 and Ssn8, the chromosomal SFL1 and SSN8 genes were disrupted by insertion of the URA3 gene in the strain PS-5V-H19 [PSI ϩ PS-1 ]. The clones with disrupted SSN8 showed markedly reduced growth rate. This growth reduction was [PSI ϩ PS ]dependent, because the growth rate returned to normal after passages on GuHCl-containing medium or after transformation with the centromeric pRG416-SUP35C plasmid encoding Sup35 without the prion domain. The growth inhibition could be because of accelerated prion formation. In this case we would observe decreased levels of soluble Sup35PS in the PS-5V-H19 [PSI ϩ PS ] ssn8::URA3 strain compared with [PSI ϩ PS ]. However, we could not confirm this opportunity, because the soluble Sup35PS levels in these strains were extremely low and difficult to compare. On the other hand, the SSN8 disruption affected translation, because in the PS-5V-H19 [psi Ϫ ] background it showed 2-fold increased levels of nonsense readthrough compared with [psi Ϫ ] (data not shown). Such an increase did not allow the suppression of ade2-1 and did not affect growth of the [psi Ϫ ] strain, but it is likely that, in combination with [PSI ϩ PS ], which itself is a strong suppressor, it could be detrimental. In addition, the cells of both sfl1::URA3 and ssn8::URA3 disruptants were distinguished by increased flocculence and invasive growth (Fig. 2). These phenotypes were observed earlier for the sfl1 disruptants (34), and the flocculence was observed in ssn8 mutants (35), whereas the invasive growth of the ssn8 disruptants is a novel finding.
Sfl1 Binds the Heat Shock Element-Transcriptional response to heat shock is mediated by binding of Hsf1 to the HSE typical of heat-inducible promoters (36). The similarity of Sfl1 to Hsf1 suggests that Sfl1 could also bind to HSE and direct the expression of heat shock-related proteins. Such a possibility was studied by mobility shift assay. HSE was synthesized as an oligonucleotide duplex of 25 base pairs. Purified Sfl1 was obtained via E. coli expression as described under "Experimental Procedures." Labeled HSE was incubated with Sfl1 or yeast lysates differing in the levels of Sfl1 and run on a gel (Fig. 3). HSE showed an efficient binding to pure Sfl1, which suggests that Sfl1 can prime the HSE-dependent transcription. HSE also bound some components of yeast lysates, and the amount of such complexes correlated with the Sfl1 levels. This suggests that Sfl1 was a component of these complexes, together with some other proteins, because the mobility of the complexes was lower than that of the pure Sfl1⅐HSE complex. The amount of the HSE⅐protein complexes in the lysates lacking Sfl1 was reduced greatly respective to the wild-type lysates. This suggests that under the experimental conditions Sfl1 was a major protein bound to HSE. In contrast, pure Sfl1 did not bind to STRE, another DNA element typical of many heat-and stressinducible promoters (36). Nevertheless, the amount of protein bound to STRE in the lysates depended notably on the Sfl1 levels, being the lowest in the wild-type lysates. This suggests that although Sfl1 does not bind STRE, it exerts a significant indirect influence on the STRE-driven transcription (Fig. 3).
Overexpressed Rpp0 Is Found in Ribosome-free Fractions-Rpp0 represents one of the five so-called acidic ribosomal proteins that form a stalk at the large subunit of a yeast ribosome. Among them, Rpp0 plays a key role, because it is essential for viability and mediates binding of four other acidic ribosomal proteins to the ribosome. Earlier, it was observed that the increased transcription of RPP0 did not lead to increased levels of Rpp0 and its appearance free from ribosomes (37). Such a conclusion would exclude a possibility of any effects of multicopy RPP0. Therefore, we studied the levels of Rpp0 and its subcellular distribution. The immunoblot analysis of Rpp0 revealed an ϳ3-fold excess of it in the total extracts of cells with Labeled HSE and STRE DNA fragments were incubated with either purified His 6 -Sfl1 or with yeast lysates prepared from the PS-5V-H19 [psi Ϫ ] cells with the multicopy SFL1 plasmid (mSFL1), Yeplac195 vector (Control), or SFL1 disruption (⌬SFL1) and analyzed on a 4.5% polyacrylamide gel. The positions of HSE and STRE complexes with His 6 -Sfl1 or lysate proteins are indicated by arrows. Ϫ, no His 6 -Sfl1, ϩ, His 6 -Sfl1 added.

TABLE I Effects of the anti-prion factors
The effects were determined for the transformants of indicated [PSI ϩ ] strains overproducing described proteins. The strength of antisuppression was estimated by the color of yeast streaks on medium selective for plasmid and lacking adenine, where only the prion-containing cells can grow (Fig. 1B). The frequency of prion loss was determined as described under "Experimental Procedures." S.E. did not exceed 20% of the value (50% for values Ͻ1%). 0 denotes less than 0.1% prion loss. Sectored loss (see Fig. 1A) reflects a visual proportion of red sectors in colonies. ϩϩ, strong effect; ϩ, weak effect; Ϫ, no effect. Control, transformants with the YEplac195 vector. [ multicopy RPP0 (Fig. 4). The lysates were fractionated by centrifugation through a sucrose density gradient, and the ribosomal fractions were identified by the presence of ribosomal RNA. This allowed finding Rpp0 in subribosomal fractions, though not in a free monomeric form. The amount of ribosomefree Rpp0 was rather low in a control lysate and much higher in the one overexpressing Rpp0. Therefore, the multicopy expression of RPP0 led to the appearance of Rpp0 in the form of relatively large ribosome-free complexes. This result does not strictly contradict the earlier data (37) because of the use of different expression systems. These authors (37) used a singlecopy GAL1 promoter, which presumably should provide a lower Rpp0 expression level than multicopy RPP0. Alterations of the Sfl1, Ssn8, and Rpp0 Levels Affect Expression of Chaperones and Induce Heat Shock Response-In our experiments, the prion curing was probably because of the altered levels of proteins involved directly in the prion propagation. The studied factors could, themselves, either belong to such proteins or act indirectly by altering expression of the proteins of this group. The first opportunity appears likely for the chaperones Sis1, Apj1, and Sti1, whose function is to mediate the protein folding. The second suits best for the transcriptional factors Sfl1 and Ssn8, as well as for the ribosomal protein Rpp0. Although Rpp0 is not a transcriptional factor, its influence on gene expression may be expected, because the lack of acidic ribosomal proteins Rpp1A/B and Rpp2A/B, structurally and functionally related to Rpp0, altered the levels of many proteins, including chaperones. As a particular case of the indirect mechanism, the [PSI ϩ PS ] loss could be because of decreased expression of Sup35PS. Indeed, earlier we have observed the [PSI ϩ ] loss at decreased Sup35 levels. 2 However, here we found that the overexpression of Ssn8, Sfl1, and Rpp0 did not reduce the Sup35 levels (Table II and Western blotting  data; not shown). Furthermore, the overexpression of Rpp0 increased Sup35 levels about 2-fold, which should counteract with, rather than promote, the prion curing.
If Sfl1 and Ssn8 regulate expression of the chaperone genes, they would affect the levels of many chaperones, and the curing of prions will likely be because of a combination of changes in the levels of several chaperones, rather than of any single chaperone. It is known that expression of the chaperone-encoding genes usually is controlled by either one or combination of the two promoter elements, HSE and STRE (36). To characterize the effects of overproduced Sfl1, Ssn8, and Rpp0 on the chaperone expression, we determined their influence on the levels of lacZ expression under the control of model promoters containing either HSE or STRE. We also tested the activity of SSA4 and HSP104 promoters (Table II). SSA4 is an example of a gene expressed only under heat shock or stress conditions, and HSP104 is known for its unique and essential role in prion propagation. Excess Sfl1 increased the HSE-driven expression about 3-fold, which is consistent with its binding to the HSE element, and suggests that Sfl1 acts here as transcriptional activator, rather than repressor. Excess Sfl1 also increased the STRE-dependent expression. Excess Ssn8 did not affect the level of HSE-driven expression but repressed the STRE promoter 1.5-fold and the SSA4 promoter 2-fold. The lack of Ssn8, conversely, activated the STRE-and SSA4-controlled expression, consistent with its repressor function. Excess Rpp0 increased the HSE-and SSA4-driven expression to approximately the same levels as excess Sfl1.
Because altered levels of Sfl1, Ssn8, and Rpp0 affected the expression of chaperones, they are also likely to affect the resistance to heat shock. To check this, we tested the ability of cells with increased or decreased levels of Sfl1, Ssn8, and Rpp0 to grow at 37°C. At this temperature, because of accumulation of misfolded proteins, yeast cells may become transiently arrested in the G 1 phase of the cell cycle via an Hsf1-dependent mechanism (38). Although the control PS-5V-H19 [psi Ϫ ] cells almost did not grow at 37°C, the presence of multicopy plasmids with SFL1 and RPP0, but not SSN8 gene, allowed them to grow at this temperature (Fig. 5). Even better growth was observed for derivatives of this strain disrupted for SSN8 or SFL1.

DISCUSSION
In this work, a search for the cellular factors interfering with the [PSI ϩ PS ] propagation was conducted. Although previously only chaperones (Hsp104, Ssb1, Ssa1, and Ydj1) were known as such factors, this search revealed, in addition to chaperones Sis1, Apj1, and Sti1, transcriptional factors Sfl1 and Ssn8, and acidic ribosomal protein Rpp0. We propose that all revealed anti-prion factors could act either by directly interfering with the Sup35PS prion conversion or by altering the levels of the proteins involved in this process. The former mechanism appears likely for the chaperones, whereas the latter appears likely for the transcriptional factors and the ribosomal protein.
Chaperones That Cure Yeast Prions-Two of the three newly found chaperones, Sis1 and Apj1 (Ynl077w) belong to the that the three considered Hsp40 proteins function similarly in prion curing. For example, they could cooperate with Hsp104 and Hsp70 in disassembly of prion aggregates (14). It is of interest to note that whereas the antisuppressor effect of Sis1 was much stronger than that of Ydj1, Sis1 is normally expressed at 10 -15-fold lower levels than Ydj1, which represents a major Hsp40 in yeast (39). Thus, Sis1 molecules are much more efficient in inhibiting prion formation.
No chaperones of the Hsp70 family were found in this screen. This is not surprising, because these chaperones are tightly autoregulated and allow a significantly lower increase of their expression than Hsp40 chaperones. For example, Ssa1 had a rather weak effect, and Ssb1 had no effect on [PSI ϩ PS-1 ] used in the screen (9). Another chaperone found in this screen is Sti1. This protein represents a component of the Hsp90⅐Hsp70 complex (40). Thus, it is related to both Hsp90, which did not influence the [PSI ϩ ] propagation, and Hsp70 (Ssa1 and Ssa2), which interfered with [PSI ϩ PS ] but reduced the [PSI ϩ ] curing by excess Hsp104 (10). Sti1 could interfere with prion formation either directly or by modulating the activity of Ssa chaperones at the functional level. Also, the overexpression of Sti1 was shown to increase the Ssa4 expression (41), and thus it could act indirectly by altering the expression of Ssa or other chaperones.
Non-chaperone Proteins Curing [PSI ϩ PS ] Affect the Expression of Stress-inducible Genes-One of these proteins, Sfl1, is a transcriptional factor structurally similar to the heat shock factor Hsf1, which plays a key role in the heat shock response. In contrast to Hsf1, the previously described effects of Sfl1 were related mainly to the cell wall biogenesis (42). Furthermore, Sfl1 is presumed to mediate the transcriptional repression rather than activation (33). Sfl1 is highly similar to Hsf1 in the region corresponding to its DNA binding domain, which suggests that Sfl1 could also bind to HSE and thus regulate chaperone expression. We confirmed this by finding that purified Sfl1 binds to HSE in vitro. Capacity of the cell lysates to bind HSE correlated with the Sfl1 levels and was rather low in the lysate lacking Sfl1. This suggests that Sfl1 was one of the major proteins bound to HSE under the given experimental conditions. The ability to bind to HSE was also shown for Skn7 (43) and could be expected of Mga1 and Hsm2 based on sequence similarity. 3 Apparently, these proteins and Sfl1 should compete with Hsf1 for binding to HSE, thus performing a complex regulation of HSE-dependent expression in response to different environmental challenges. Sfl1 did not bind to the STRE. Nevertheless, the amount of the STRE⅐protein complex in cell lysates notably depended on Sfl1, being increased both in lysate with excess Sfl1 and in lysate lacking it. This suggests that Sfl1 has a significant indirect influence on the STRE-controlled expression.
Another transcription factor revealed by [PSI ϩ PS ] curing is Ssn8, which represents a subunit of the transcriptional mediator complex (32). Ssn8 represents a homologue of human cyclin C, and it degrades upon entry to meiosis and heat shock. Ssn8 acted as transcriptional repressor of SSA1 and presumably some other chaperone genes (44). Importantly, Ssn8 was shown to interact with Sfl1 both physically and functionally. These proteins coimmunoprecipitated from cell lysates, and both participated in the transcriptional repression of the SUC2 gene (33). Because Sfl1 and Ssn8 are involved in transcription, their excess is likely to affect the expression of many proteins.
To characterize the influence of the excess Sfl1 and Ssn8 on the overall chaperone expression, we used model promoter constructs with the lacZ reporter gene. The excess of Sfl1 activated the HSE promoter 3-fold, which agrees with the Sfl1 binding to this element and suggests that Sfl1 acts as activator, rather than repressor, for this element. It also induced the STRE-driven expression but only 1.5-fold. This also agrees with the observed increase of the amount of protein bound to this DNA fragment in cell lysate with increased Sfl1 levels. In contrast, the excess of Ssn8 acted only on the STRE promoter and decreased its activity 1.5-fold, whereas the lack of Ssn8 induced it 1.5-fold. These data are consistent with the proposed Ssn8 function as a repressor of transcription (45). Thus, both Sfl1 and Ssn8 should affect significantly the expression of many chaperones, which is likely to be a cause for their prioncuring effects. It is noteworthy that in contrast to the earlier data, Sfl1 and Ssn8 exerted essentially different influence on transcription. This may indicate that these proteins do not always cooperate in their action. Although the ribosomal protein Rpp0 is not supposed to participate in transcription, we observed that overproduced Rpp0 increased the HSE-and SSA4-driven lacZ expression 2.5-fold, which could be sufficient for the observed [PSI ϩ PS ] curing. At the same time, the STRE and HSP104 promoters showed little activation. This difference in the effects of Rpp0 was defined at the transcriptional rather than post-transcriptional level, because the translated message, lacZ, was the same in all cases. We suggest that the excess of Rpp0 may affect transcription by participating in it either directly or indirectly. The latter opportunity is related to the observation that the lack of acidic ribosomal proteins Rpp1A/B and Rpp2A/B, structurally and functionally related to Rpp0, significantly affected the expression of many proteins and chaperones in particular (46). It was proposed that the altered content of acidic proteins in ribosome changed its specificity toward different mRNAs. It appears unlikely that overproduction of Rpp0 would alter its content in ribosomes, because it is difficult to imagine that a ribosome can accommodate more than one molecule of this protein. However, the ribosome-free Rpp0 accumulated upon its overexpression may titrate a significant portion of Rpp1/2, thus creating a subpopulation of ribosomes that lack these proteins. This, in turn, may alter the synthesis of transcriptional factors that regulate the expression of chaperones.
Implications--The properties of yeast prions suggest that they may be used as a model for studying both prions and amyloids of higher eukaryotes (1). This implies that the factors similar to those found here may be active against human and animal prion and amyloid diseases. This is supported by the observation that overexpression of Hsp40 and Hsp70 chaperones in human cell lines interfered with aggregation of proteins with expanded polyglutamine tracts and amyloidogenic immunoglobulin light chains (47,48). In this work and previously (9) we observed that the efficiency of the prion curing factors varied greatly depending on the prion strain and the primary structure of the prion domain. This suggests that the use of screening systems based on different prions may allow finding more factors involved in the prion-like processes. On the other hand, this should complicate the prediction of anti-prion activity of a particular factor against a given prion or amyloid. The non-chaperone factors similar to those discovered in this work could be no less active against prions and amyloids than individual chaperones, because they allow activation of several chaperones simultaneously.