[PSI+] is a prion isoform of the yeast release factor Sup35. In some assays, the cytosolic chaperones Ssa1 and Ssb1/2 of the Hsp70 family were previously shown to exhibit “pro-[PSI+]” and “anti-[PSI+]” effects, respectively. Here, it is demonstrated for the first time that excess Ssa1 increases de novo formation of [PSI+] and that pro-[PSI+] effects of Ssa1 are shared by all other Ssa proteins. Experiments with chimeric constructs show that the peptide-binding domain is a major determinant of differences in the effects of Ssa and Ssb proteins on [PSI+]. Surprisingly, overproduction of either chaperone increases loss of [PSI+] when Sup35 is simultaneously overproduced. Excess Ssa increases both the average size of prion polymers and the proportion of monomeric Sup35 protein. Both in vivo and in vitro experiments uncover direct physical interactions between Sup35 and Hsp70 proteins. The proposed model postulates that Ssa stimulates prion formation and polymer growth by stabilizing misfolded proteins, which serve as substrates for prion conversion. In the case of very large prion aggregates, further increase in size may lead to the loss of prion activity. In contrast, Ssb either stimulates refolding into nonprion conformation or targets misfolded proteins for degradation, in this way counteracting prion formation and propagation.
PRIONS are infectious or heritable agents transmitted at the protein level. Prions cause neurodegenerative diseases, such as “mad cow” and Creutzfeldt-Jacob diseases, in mammals and humans (for review, see Prusiner et al. 1998). Prion and nonprion (cellular) isoforms of one and the same protein can have identical amino acid sequences but differ from each other in conformation and ability to form aggregated multimolecular structures. Mammalian prions are usually aggregation-prone proteins, capable of generating fiber-like polymers of ordered structure, called amyloids. This resembles noninfectious amyloidoses and inclusion body disorders, such as Alzheimer's or Huntington's diseases. The model of “nucleated polymerization” (for review, see Lansbury and Caughey 1995) proposes that prion reproduction occurs via shearing of the large aggregates into oligomeric “seeds” that grow by immobilizing the newly synthesized protein molecules. An alternative model of “template assistance” (for review, see Harrison et al. 1997) views prion seeds as misfolded monomers that stimulate misfolding of the newly synthesized or partially unfolded polypeptides of the same amino acid sequence.
Several proteins of different structures and functions can form prions in yeast, as concluded first by Wickner (1994) on the basis of genetic criteria and confirmed by subsequent genetic and biochemical experiments (for reviews, see Chernoff 2001, 2004). Known yeast amyloidogenic prion proteins include [PSI+] (Cox 1965), a prion isoform of the translation termination factor Sup35 (eRF3); [URE3] (Lacroute 1971), a prion isoform of Ure2, the regulatory protein in nitrogen metabolism; and [PIN+] (Derkatch et al. 1997), or [RNQ+] (Sondheimer and Lindquist 2000; Derkatch et al. 2001), a prion form of Rnq1, a protein of an unknown function. Yeast prions provide a molecular mechanism for protein-based inheritance by controlling phenotypic traits inherited in a non-Mendelian fashion (for reviews, see Chernoff 2001, 2004).
As prion formation and propagation is apparently modulated at the level of three-dimensional and/or quarternary structure of a protein, chaperone proteins play an important role in these processes. The yeast prion [PSI+] is eliminated by overproduction or inactivation of the chaperone Hsp104 (Chernoff et al. 1995). As Hsp104 is involved in disaggregation of heat-damaged proteins (Parsell et al. 1994; Glover and Lindquist 1998), it has been hypothesized that moderate levels of Hsp104 generate new prion seeds via shearing of the preexisting prion aggregates, while high levels of Hsp104 eliminate prion aggregates by converting them into nonprion monomers (Paushkin et al. 1996). In agreement with this model, Hsp104 overproduction leads to accumulation of soluble Sup35 protein (Paushkin et al. 1996), while Hsp104 depletion results in uncontrolled growth of Sup35 aggregates in the [PSI+] cells (Wegrzyn et al. 2001). At least one group has been able to demonstrate that excess Hsp104 can disassemble Sup35 amyloid fibers in vitro (Shorter and Lindquist 2004). At the same time, excess Hsp104 promotes de novo fiber formation from the soluble Sup35 protein in vitro, a role yet to be confirmed in vivo. Hsp104 is also required for propagation of [PIN+] (Derkatch et al. 1997) and [URE3] (Moriyama et al. 2000), but in contrast to [PSI+], these prions are not cured by Hsp104 overproduction.
Increased levels of Ssa1, a chaperone of the Hsp70 family, antagonize [PSI+] curing and solubilization of Sup35 aggregates by excess Hsp104 (Newnam et al. 1999). At least in some [PSI+] variants, excess Ssa1 also increases phenotypic expression of [PSI+], detected as readthrough of nonsense codons (nonsense suppression) as a result of a translational termination defect (Newnam et al. 1999). The Ssa subfamily of the Hsp70 proteins, one of the major heat-shock-inducible cytosolic chaperone subfamilies in yeast, includes Ssa1 and three additional members, Ssa2, -3, and -4. Presence of at least one Ssa protein is required for vegetative growth (Werner-Washburne et al. 1989). Ssa1 and Ssa2 are closely related to each other and somewhat diverged from Ssa3 and Ssa4. Each Ssa protein consists of three domains, namely ATPase domain, peptide-binding domain, and variable C-terminal domain (see James et al. 1997). Ssa2 is constitutively expressed at high levels, and Ssa1 is moderately expressed during vegetative growth and induced during stresses, whereas Ssa3 and Ssa4 are strictly stress inducible (Werner-Washburne et al. 1989).
Although Ssa's requirement for growth complicates the analysis of the essentiality of Ssa for [PSI+], ssa1 and ssa2 mutations that antagonize [PSI+] (Jung et al. 2000; Jones and Masison 2003) and [URE3] (Roberts et al. 2004), respectively, were identified. Phenotypes of mutants were most severe in the absence of the other Ssa protein, which is normally expressed in vegetative cells. Effects of the ssa1 mutations on growth did not correlate with their effects on [PSI+], suggesting that Ssa's role in [PSI+] propagation is not a simple consequence of its major cellular function and may reflect a certain specificity of Ssa's interaction with prion proteins.
Another cytosolic Hsp70 subfamily, Ssb, is composed of two nearly identical proteins, Ssb1 and Ssb2, which are not essential for viability and not inducible by heat shock (Nelson et al. 1992). In strong contrast to Ssa, Ssb consistently acts as a [PSI+] antagonist. Excess Ssb increases [PSI+] curing by overexpressed Hsp104 (Chernoff et al. 1999), inhibits [PSI+]-mediated suppression in certain [PSI+] isolates (Chernoff et al. 1999), and causes loss of [PSI+] upon prolonged incubation in certain genotypic backgrounds (Kushnirov et al. 2000b; Chacinska et al. 2001). Deletion of both SSB1 and SSB2 genes decreases efficiency of [PSI+] curing by excess Hsp104 and increases the frequency of the spontaneous [PSI+] formation in [psi−] cells (Chernoff et al. 1999).
While an “antiprion” effect of Ssb fits perfectly with the role of chaperones as antiaggregation devices, the “proprion” role of Ssa1 comes as a surprise. Indeed, Ssa was previously thought to assist Hsp104 in disaggregation and refolding of the heat-damaged aggregated proteins (Glover and Lindquist 1998). Moreover, in the case of some atypical [PSI+] derivatives, excess Ssa1 becomes antagonistic to [PSI+] (Borchsenius et al. 2001; Kryndushkin et al. 2002). Excess Ssa1 also impairs propagation of another yeast prion, [URE3] (Schwimmer and Masison 2002). Most surprisingly, overproduced Ssa2 protein, which is 97% identical to Ssa1, does not exhibit any antagonizing effect on [URE3]. This shows that effects of Ssa1 cannot be automatically extrapolated to other members of the subfamily. To our knowledge, Ssa1 is the only Ssa protein whose overproduction was previously reported to influence [PSI+].
Here, we performed a detailed study on the effects of various members of the Ssa subfamily on [PSI+] formation and maintenance and on Sup35 protein aggregation. Our results show that Ssa proteins not only protect [PSI+] from the curing effect of excess Hsp104, but also facilitate de novo induction of the [PSI+] prion in [psi−] cells. Biochemical assays confirm physical interactions between Sup35 and Ssa and detect alterations in the average size and abundance of Sup35 polymers in [PSI+] cells overproducing Ssa. The pro-[PSI+] effect of Ssa disappears if its peptide-binding domain is substituted with the corresponding domain of Ssb. Remarkably, the pro-[PSI+] Ssa proteins turn into [PSI+] antagonists if Sup35 levels in [PSI+] cells are increased. These data demonstrate that the consequences of Ssa effects on yeast prions are determined by levels of prion proteins in the cell and/or by physical parameters of prion aggregates.
MATERIALS AND METHODS
Saccharomyces cerevisiae strains used in this study (see Table 1) belong to two isogenic sets described previously, namely 74-D694 and GT81-1C. All strains contain the [PSI+]-suppressible marker ade1-14 (UGA), used as a reporter for [PSI+]. Variants of [PSI+] were generated in both genotypes. For example, OT55 is a “weak” [PSI+] variant ([PSI+]w), which exhibits less efficient suppression of the ade1-14 reporter and is less stable in mitosis in comparison to the isogenic “strong” [PSI+] variant ([PSI+]s) OT56. GT202 is a [PSI+] strain of the GT81 series, designated as [PSI+]*, that was obtained in the ssb1/2Δ background and is characterized by decreased suppression in the presence of Ssb (Chernoff et al. 1999). All [PSI+] strains also contain another prion, [PIN+], which is, at least in this strain series, identical to [RNQ+], a prion isoform of the Rnq1 protein (see Derkatch et al. 1997; Sondheimer and Lindquist 2000; Derkatch et al. 2001). Among [psi−] derivatives, both [PIN+] and [pin−] versions were generated. For example, GT409 is a derivative of GT81-1C that was cured of [PSI+] by guanidine hydrochloride (GuHCl; see Chernoff et al. 2002) and has also been shown to have lost [PIN+]. Presence or absence of [PIN+] was controlled genetically by its ability to facilitate [PSI+] induction by overproduced Sup35 (Derkatch et al. 1997; Chernoff et al. 2002). In the [psi− PIN+] and [psi− pin−] strains used in this study (OT60, GT159, GT17, and GT409), presence or absence of [PIN+] was also confirmed biochemically by a differential centrifugation assay, performed as described previously (Meriin et al. 2002).
S. cerevisiae-Escherichia coli shuttle plasmids:
Shuttle vectors used in this study are listed in Table 2. Centromeric (CEN) vectors are usually present at one or several copies per cell, while 2μ DNA-based (episomal) vectors are present in a large number of copies. Some constructs contained the SUP35 or HSP genes placed under control of the galactose-inducible promoter (PGAL). Sup35 protein consists of three major domains: the N-proximal, or prion-forming, domain (Sup35N); the middle domain (Sup35M); and the C-proximal release factor domain (Sup35C). In addition to the constructs with the complete SUP35 gene, we also used constructs with only the SUP35N region and those missing only the very C terminus of SUP35C after the SalI site (SUP35ΔS). These constructs are not capable of compensating for the Sup35 function in termination, but they can still induce de novo formation of [PSI+] in the [psi− PIN+] cells (for review, see Chernoff 2001; Chernoff et al. 2002).
Plasmids pC211 and pN2, containing the SSA1 and SSA2 genes, respectively, under control of the PSSA2 promoter (Schwimmer and Masison 2002), were kindly provided by D. Masison. Two sets of plasmids encoding Ssa-Ssb chimeric proteins, in which the three functional domains (the ATPase domain, peptide-binding domain, and C-terminal variable domain) of SSA1 and SSB1 are systematically swapped (James et al. 1997), were kindly provided by E. Craig. In one set, designated the pTEF-A/B series, the chimeras are under the control of the strong constitutive PTEF1 promoter, while in another set, designated the pRS316-K-A/B series, they are under the Hsp70 promoter that corresponds to the gene from which the N-terminal (ATPase) domain originates (PSSA1 or PSSB1, respectively). Chimeric constructs are usually designated by characters corresponding to the gene from which the domains originate, in order from the N terminus to the C terminus. For example, AAB is a construct containing the ATPase and peptide-binding domains of SSA1 (A) in conjunction with the C-terminal variable domain of SSB1 (B). Both sets also contained control SSA1 (AAA) and SSB1 (BBB) genes expressed from either the PTEF1 promoter (pTEF-SSA1 and pTEF-SSB1, respectively) or their own promoters (pRS316K-SSA1 and pRS316K-SSB1, respectively).
Plasmids constructed in the course of this study (see Table 2) were generated as follows. To construct pmCUP1-SUP35-HA, which produces an HA-tagged Sup35 protein for coimmunoprecipitation experiments, the 2.3-kb BamHI-SacI fragment of the plasmid p315-Sp-SUP35-HA3, kindly provided by J. Weissman, was cut off and ligated with the large (5.3-kb) BamHI-SacI fragment of pmCUP1-sGFP (Serio et al. 1999). The resulting plasmid contains the SUP35 derivative with an HA-tag between the M and C regions, placed under the control of the copper-inducible (PCUP1) promoter. To construct pRS424-SSB1, the 2.4-kb BamHI-SacI fragment of YCp50-SSB1 (Ohba 1997) containing the SSB1 gene under its own promoter was inserted into pRS424 (Sikorski and Hieter 1989) cut with the same enzymes. The S. cerevisiae SSA3 ORF was cloned from the genome of S. cerevisiae strain S288C by PCR with direct (5′-GGCCGTCGACCGGATAGAATAGGTACTAAACGCTACA) and reverse (5′- GGCCAAGCTTTCATCATGGATAGATTACCCGC) primers, containing extensions with the SalI and HindIII sites, respectively (underlined). The PCR fragment was digested with SalI and HindIII, inserted into the plasmid pRS316GAL, and cut with SalI and HindIII, thus generating pRS316GAL-SSA3. To construct pYCL1-GAL-SSA3, the 2.8-kb PGAL-SSA3 cassette was cut from pR316-GAL-SSA3 with KpnI and HindIII and inserted into pYCL1 digested with the same enzymes. The 1.9-kb SalI and HindIII cassette, containing SSA3 ORF from pRS316GAL-SSA3, was inserted into pUK21 with the same enzymes, generating pUK21-SSA3. Plasmid pUK21-SSA3 was then digested with NotI and XhoI, and the SSA3 ORF was moved to pTEF-SSA4 (see below) digested with the same enzymes, thus replacing SSA4. Sequencing of the PCR-generated SSA3 insert, performed at The Nevada Genomics Center (University of Nevada, Reno), has revealed only one same-sense T-to-C substitution at position 852, changing the TCT (serine) codon into codon TCC with the same coding capacity. Functionality of the PCR-amplified SSA3 clone was confirmed by its ability to partly compensate for the temperature-sensitive phenotype of the yeast strain, lacking SSA1, SSA2, and SSA3 genes. Plasmids pTEF-SSA4 and pYCL1-GAL-SSA4, expressing the SSA4 gene from the strong constitutive PTEF1 promoter and the galactose-inducible PGAL promoter, respectively, were constructed by first inserting the 3-kb BamHI-SphI fragment with the SSA4 ORF from the plasmid YEpGALl-SSA4, kindly provided by E. Craig, into E. coli vector pUK21, thus generating pUK21-SSA4. Then the same 3-kb fragment was either cut from pUK21-SSA4 with XbaI and XhoI and ligated with the 5.1-kb XbaI-XhoI fragment of pTEF-SSA1, thus replacing the SSA1 ORF with the SSA4 ORF, or cut from pUK21-SSA4 with BamHI and SpeI and ligated with the 6.4-kb BamHI-SpeI fragment of the plasmid pYCL1-GAL-SSA3, thus replacing the SSA3 ORF with the SSA4 ORF. Western analysis with Ssa3/4 specific antibody proved that Ssa3 and Ssa4 overexpressor plasmids constructed in this study produce an excess of Ssa3/4 reactive material in comparison to the isogenic control strains grown under the same conditions (data not shown).
E. coli expression plasmids:
To construct the plasmid pET-20b-SUP35NM-(His)6, the 0.75-kb SUP35NM fragment was PCR amplified from CEN-GAL-SUP35 with direct (5′-CGGCCATATGTCGGATTCAAACCAAGGC) and reverse (5′-ACACTCGAGATCGTTAACAACTTCGTCATC) primers containing extensions with NdeI and XhoI sites (underlined), respectively. The PCR product was cut with NdeI and XhoI and inserted into the vector pET-20b(+), cut with the same enzymes. The resulting construct contains the first 251 codons of Sup35 fused in frame to the 6 His codons at the C terminus.
Genetic and microbiological procedures:
Assays for [PSI+] formation and curing:
The presence of [PSI+] was monitored by its ability to suppress the reporter allele ade1-14 (UGA). In the absence of [PSI+], ade1-14 confers red color and inability to grow on −Ade medium to the yeast cells, while its suppression by [PSI+] results in pink or white color and Ade+ phenotype. Weak and strong variants of [PSI+], such as OT55 and OT56, differ from each other by the efficiency of suppression, which results in different rates of growth on −Ade medium and different intensities of color on complete medium (YPD). Assays for [PSI+] induction by overproduced Sup35 (or Sup35N) and for [PSI+] inhibition and curing by overproduced Hsp104, as well as for [PSI+] curing by GuHCl, were performed in accordance with the previously described procedures (see Chernoff et al. 2002).
Antibodies and protein analysis:
Increase in total Hsp104, Sup35, Ssa, or Ssb levels in the presence of overexpressor constructs was confirmed by Western analysis as described previously (Chernoff et al. 1995, 1999; Derkatch et al. 1996; Newnam et al. 1999). Antibodies specific to Hsp104 and Rnq1 were kindly provided by S. Lindquist. Antibodies specific to total Ssa, Ssa3/4, and Ssb were generously provided by E. Craig. Antibodies specific to Ydj1 and Sis1 were kindly provided by D. Cyr. Rabbit polyclonal antibodies to Sup35NM were obtained as described previously (Wegrzyn et al. 2001). Rabbit polyclonal antibody to Ade2 was raised by Cocalico to purified Ade2 protein, kindly provided by V. Alenin. Rabbit polyclonal antibodies to Sup35C were the generous gift of D. Bedwell. HA-specific antibody 12CA5 was from Maine Biotechnology Services. In all experiments, Western blots were probed with appropriate secondary antibodies from Sigma (St. Louis) and developed according to the Amersham (Buckinghamshire, UK) ECL detection system protocols using 250 mm luminol, 90 mm coumaric acid, 0.03% H2O2, and 0.1 m Tris, pH 8.5. Yeast cell lysates were prepared as previously described (Newnam et al. 1999). Densitometry was performed on X-ray films by using the Kodak 1D program on a Gel Logic 200 imaging system. Several exposures were checked to confirm that measurements remain within the linear range of the assay sensitivity.
Analysis of Sup35 aggregation:
For all experiments, protein extracts were prepared from exponential cultures grown in synthetic media selective for the appropriate plasmid(s). Proteins were isolated as described previously (Paushkin et al. 1996; Chernoff et al. 2002), and protein concentrations were determined by Bradford assay according to Bio-Rad (Richmond, CA) protocol.
“Gel entry” assay (Kryndushkin et al. 2003) was used to detect the low-molecular-weight Sup35 fraction that is capable of entering the polyacrylamide gel without denaturation. Total lysates (20 μg) were cleared of cell debris by centrifugation at 900 × g for 10 min, incubated in the sample buffer (25 mm Tris-HCl, pH 6.8, 10% glycerol, 5% 2-mercaptoethanol, 2% SDS) for 5 min at 37° (nondenaturing conditions) or 100° (denaturing conditions), and applied to a 10% polyacrylamide gel containing a stacking gel of 1.8% agarose (in 1× TAE, 0.1% SDS). After electrophoresis, the proteins from the stacking and separating gel were transferred to a nitrocellulose membrane for further analysis.
Semidenaturing detergent-agarose gel electrophoresis (SDD-AGE ) was performed according to Kryndushkin et al. (2003) to investigate the distribution of the prion polymers by size. Total lysates, cleared of cell debris by centrifugation at 10,000 × g for 2 min, were incubated in sample buffer (0.5× TAE, 2% SDS, 5% glycerol) for 5 min at 37°, loaded on 1.8% agarose gels in 1× TAE, 0.1% SDS, and run at 80 V, followed by vacuum transfer to a PVDF membrane. Membranes were reacted to the Sup35C antibody. According to our observations, Sup35NM antibody was not as effective in detection of the polymerized Sup35 protein as Sup35C antibody, possibly due to decreased accessibility of the Sup35N domains, apparently forming the axis of a polymer.
Assay for in vitro protein interactions:
Maximum production of the His-tagged Sup35NM protein, which is toxic to E. coli, could be achieved in the strain HMS174 (pLysS) purchased from Novagen and containing the recA mutation (preventing plasmid rearrangements) and the pLysS plasmid (inhibiting basal expression of pET-based constructs in the absence of inducer). E. coli cultures were grown in Luria broth medium with 100 μg/ml ampicillin and 75 μg/μl chloramphenicol, which is selective for both pET and pLysS plasmids, at 37° to OD550 0.6–0.8. After 4 hr induction of the pET-based construct by 1 mm IPTG, cells were precipitated, resuspended in cold 1× binding buffer (5 mm imidazole, 0.5 m NaCl, 20 mm Tris-HCl, pH 7.9) with 0.1% IGEPAL, and destroyed at power 40 on a Cole-Parmer ultrasonic homogenizer (4710 series). Cell debris was removed by centrifugation at 16,000 × g for 15 min at 4°, and supernatants were applied to the Novagen Ni2+ resin, charged according to the company protocol. After 1 hr of incubation at 4°, resin was washed with 1× binding buffer, followed by 1× wash buffer (60 mm imidazole, 0.5 mm NaCl, 20 mm Tris-HCl, pH 7.9) to remove unbound protein. Resin treated with the extracts of the control E. coli culture containing the empty vector pET-20b(+) was always used as a control.
To prepare yeast extracts, cells from late exponential yeast cultures (OD600 1.0–5.0) were precipitated, washed in 1× binding buffer without imidazole, resuspended in 500 μl of 1× wash buffer with 0.9 mm phenylmethylsulfonyl fluoride (PMSF), and destroyed with 300 μl of sterile acid-washed glass beads by vortexing. Lysates were cleared by 1 min centrifugation at 2000 × g, and aliquots were run on SDS-PAGE followed by reaction to the corresponding antibodies to detect the proteins of interest. Half of each remaining lysate was loaded onto the control resin prepared from pET-20b(+) extract, and the other half was loaded onto resin prepared from pET-20b-Sup35NM-(His)6 extract, as described above. The resin was incubated with yeast lysates for 1 hr at 4° and then washed with 10 vol of 1× binding buffer, followed by 6 vol of 1× wash buffer to remove any unbound proteins. Bound proteins were then eluted with 6 vol of 1× elution buffer (1 m imidazole, 0.5 m NaCl, 20 mm Tris-HCl, pH 7.9), and eluates were analyzed by SDS-PAGE and Western blot, followed by reaction to the corresponding antibodies.
Co-immunoprecipitation assay for detection of in vivo protein interactions:
Yeast cultures were grown to OD600 of 1.0 in synthetic media. Cells were collected, washed, suspended in lysis buffer (50 mm Tris-HCl, pH 7.5, 200 mm NaCl, 1% Triton X-100, 0.1% SDS, 1 mm EDTA, 1 mm PMSF, 15 mm N-ethylmalemide, Roche protease inhibitor cocktail), and broken by agitation with glass beads. Lysates were cleared by centrifugation at 10,000 × g and anti-HA, anti-Ade2, or anti-Sup35C antibody was added to the supernatant. After 2 hr of incubation at 4°, immobilized protein A (purchased from Invitrogen, San Diego) has been added. After 1 hr of incubation at 4°, protein A beads were washed six times in lysis buffer and boiled in 2× SDS-PAGE sample buffer for 5 min. Aliquots were run on SDS-PAGE and analyzed by Western blotting followed by reaction to specific antibodies.
Excess Ssa1 protein increases de novo [PSI+] induction by excess Sup35:
Transient overproduction of either full-length Sup35 protein or its fragments containing the prion-forming domain, Sup35N, induces de novo formation of [PSI+] in the strains containing another prion, [PIN+] (Chernoff et al. 1993; Derkatch et al. 1996, 1997). To check the effect of excess Ssa1 protein on de novo [PSI+] formation, we have transformed both control [psi− PIN+] strains, OT60 and GT159 (see Table 1), and their derivatives containing the PGAL-SSA1 plasmid, with the PGAL-SUP35N construct. Overproduction of both Sup35N and Ssa1 was induced simultaneously on galactose medium. Cells were then shifted to the −Ade/glucose medium, where the PGAL promoter is repressed. Growth on −Ade, resulting from suppression of the ade1-14 (UGA) reporter, was indicative of [PSI+] induction (see Chernoff et al. 2002). Simultaneous overproduction of excess Ssa1 increased de novo [PSI+] induction at ∼10-fold after 48 hr of incubation in the presence of galactose (Figure 1, A and B). A total of 14 independent Ade+ derivatives, induced in the presence of excess Ssa1, were all shown to be curable by GuHCl, an agent eliminating [PSI+]. This confirms that the Ade+ phenotype was due to induction of the [PSI+] prion. Both episomal and centromeric plasmids expressing SSA1 from its own promoter, as well as the centromeric constructs expressing SSA1 from strong constitutive promoters, PSSA2 or PTEF1, also increased [PSI+] induction in combination with PGAL-SUP35N or PGAL-SUP35 plasmids (data not shown). A total of 14 independent [PSI+] colonies, induced by excess Sup35N in the presence of the construct expressing SSA1 from its own promoter, were all shown to be capable of growing on −Ade after the loss of SSA1 plasmid, confirming that increased frequency of observable Ade+ colonies was not due to better detection of the weak [PSI+] derivatives in the presence of excess Ssa1. No [PSI+] induction by PGAL-SUP35N or PGAL-SUP35 was observed in the isogenic [psi− pin−] strains independently of the presence or absence of excess Ssa1 (data not shown). This shows that excess Ssa1 does not bypass the requirement of [PIN+] for de novo [PSI+] induction by overproduced Sup35.
Excess Ssa1 or Ssb1 protein increases [PSI+] loss in the yeast cells overproducing Sup35:
Overproduction of the Sup35 protein in [PSI+] cells inhibits growth and increases size of the Sup35 aggregates (for review, see Chernoff et al. 2002). We have observed that overproduction of Sup35 in a [PSI+] background is also accompanied with the loss of [PSI+] in a fraction of the cells, occurring in a time-dependent fashion (Figure 1, C and D). Possibly, Sup35 overproduction leads to accumulation of the large Sup35 “dead-end” aggregates, which are insensitive to the “shearing” effect of Hsp104 and incapable of further promoting prion proliferation. In the presence of excess Sup35, newly generated [psi−] cells would have a growth advantage over [PSI+] cells, undoubtedly contributing to their accumulation over time in a Sup35-overproducing culture. To a certain extent, Sup35 overproduction recapitulates the effects of the Hsp104 depletion (Wegrzyn et al. 2001) or deletion of the 22–69 region of Sup35 (Borchsenius et al. 2001) on [PSI+].
We have checked the effects of overproduced Ssa1 and Ssb1 chaperones on [PSI+] loss in cultures overproducing Sup35 (Figure 1, C and D). As expected, overproduction of Ssb1, known to behave as a [PSI+] antagonist (Chernoff et al. 1999; Kushnirov et al. 2000b; Chacinska et al. 2001), significantly increased [PSI+] loss under these conditions. Surprisingly, overproduction of Ssa1, which usually behaved as a “[PSI+] helper,” also increased [PSI+] loss in the presence of excess Sup35. In contrast to cultures overproducing Sup35 protein alone, co-overproducers of Sup35 and Ssa1 (or of Sup35 and Ssb1) accumulated significant fractions of [psi−] cells even at relatively short periods of incubation under the induced conditions (Figure 1D). “Co-curing” with excess Ssa1 was also observed in cases when only the prion-forming domain of Sup35 (Sup35N) was overproduced, instead of the complete Sup35 protein (data not shown). The co-curing effect of excess Ssa1 was more profound in the weak [PSI+] strain OT55 (see Figure 1E) than in the strong [PSI+] strain OT56, despite the fact that increased levels of Sup35 are less toxic to the weak [PSI+] variants than to the strong [PSI+] variants. This suggests that the co-curing effect of excess Ssa1 cannot be explained simply by variations in cellular toxicity of excess Sup35. Apparently, excess Ssa1 becomes antagonistic to [PSI+] when levels of the Sup35 protein (and, consequently, size of the Sup35 aggregates) are increased in the [PSI+] background.
Other members of the Ssa subfamily reproduce Ssa1 effects on [PSI+]:
Next, we checked constructs overexpressing other members of the Ssa subfamily, specifically Ssa2, Ssa3, and Ssa4, for their effects on [PSI+]-mediated suppression (Figure 2, A–C), [PSI+] inhibition and curing by excess Hsp104 (Figure 2, D–F), de novo [PSI+] induction by excess Sup35 (Figure 2, G–I), and co-curing of [PSI+] in combination with excess Sup35 (data not shown). In all assays, the other members of the Ssa subfamily exhibited effects on [PSI+] in the same direction as did Ssa1. These data confirm that major patterns of Ssa effects on [PSI+] are conserved within the Ssa subfamily.
Effects of Hsp70 proteins on [PSI+] are not due to alterations of the Hsp104 levels:
Ssa1, and possibly Ssa2, were shown to downregulate transcription of some heat-shock protein genes (Stone and Craig 1990), while depletion of both Ssa1 and Ssa2 induces Ssa3 (Boorstein and Craig 1990a), Ssa4 (Boorstein and Craig 1990b), and Hsp104 (Sanchez et al. 1993). One could suggest that Hsp70 proteins influence [PSI+] via modulating expression of Hsp104 that is known to play a crucial role in propagation of the yeast prions. However, we previously failed to detect any observable effect of excess Ssa1 or Ssb1 or of Ssb depletion on Hsp104 levels in our strains (Chernoff et al. 1999; Newnam et al. 1999). Here, we confirm that neither overexpression of SSB1 from PTEF1 promoter nor overexpression of SSA1, SSA3, or SSA4 from either PTEF1 or PGAL promoters influences Hsp104 levels in the yeast strains used in this study (Figure 3). We have also detected no effect of excess Ssa on levels of Ssb and vice versa (data not shown). Neither excess Ssa nor excess Ssb influenced levels of the Hsp40 chaperones Ydj1 and Sis1 (data not shown). Taken together, these data rule out a possibility of overproduced Ssa or Ssb proteins modulating prion propagation via alterations in levels of any other Hsp's known to influence [PSI+].
Role of various domains of the Hsp70 proteins in the differential effects of Ssa and Ssb on [PSI+]:
Hsp70 proteins of the Ssb subfamily, in contrast to Ssa, consistently behave as [PSI+] antagonists (Chernoff et al. 1999; Kushnirov et al. 2000b; Chacinska et al. 2001). Hsp70 proteins consist of three domains: (1) the N-proximal ATPase domain, (2) the middle peptide-binding domain, and (3) the C-proximal variable domain. A series of Ssa-Ssb chimeric constructs, containing domains of Ssa and Ssb origin in all possible combinations, has been prepared in E. Craig's laboratory and used to demonstrate that any two domains of Ssb, even in combination with the third domain of Ssa, are sufficient to compensate for at least some phenotypic defects caused by lack of Ssb (James et al. 1997; Pfund et al. 2001). We used these constructs to investigate the role of various domains of the Hsp70 proteins in their differential effects on [PSI+].
First, we checked all Ssa-Ssb chimeric constructs for their effects on prion curing by excess Hsp104 in strong [PSI+] strain GT81-1C (Figure 4A) and found that any construct with PTEF1 promoter containing Ssa's peptide-binding domain in combination with at least one other domain of Ssa origin (i.e., AAA, AAB, and BAA) was able to counteract [PSI+] curing by excess Hsp104, even if the third domain came from Ssb. In contrast, all constructs containing a peptide-binding domain of Ssb (ABA, ABB, and BBA) increased the curing effect of excess Hsp104, similar to the complete Ssb protein (BBB). The construct containing a peptide-binding domain of Ssa in combination with two other domains of Ssb (BAB) exhibited no observable effect. Thus, differential effects of Hsp70 proteins on [PSI+] curing by excess Hsp104 primarily depended on the peptide-binding domain, although the proprion effect of Ssa's peptide-binding domain could be overcome by the simultaneous presence of two other domains of Ssb origin.
Next, we checked chimeric Ssa-Ssb constructs for their ability to increase nonsense suppression (see Newnam et al. 1999) in the weak [PSI+] strain OT55 (Figure 4B). Only a construct containing all three domains of Ssa origin (AAA) was able to do so. In contrast, all constructs containing a peptide-binding domain of Ssb (ABA, ABB, and BBA) antagonized nonsense suppression, similar to the complete Ssb protein (BBB). Remarkably, none of the constructs containing a peptide-binding domain of Ssa exhibited such an inhibitory effect on suppression. Thus, presence of the peptide-binding domain of Ssb origin was both necessary and sufficient for the inhibition of [PSI+]-mediated suppression in this assay.
Next, we checked effects of the chimeric constructs on de novo [PSI+] induction by overproduced Sup35N in the [psi− PIN+] background (data not shown). Of the whole set, only the AAA construct was able to increase [PSI+] induction, confirming that the presence of all three Ssa domains is necessary for such an effect.
Some [PSI+] isolates, obtained in the ssb1/2Δ background and designated here as [PSI+]*, exhibit decreased nonsense suppression in the presence of wild-type Ssb protein (Chernoff et al. 1999). We have checked effects of the chimeric Ssa-Ssb constructs on one such isolate, GT202. In this strain, only constructs bearing the peptide-binding domain and at least one other domain of Ssb (ABB and BBA) were able to reproduce the inhibitory effect of Ssb protein on nonsense suppression (Figure 4C).
Double deletion of both Ssb-encoding genes, ssb1/2Δ, causes sensitivity to some translational inhibitors, such as paromomycin and hygromycin (Nelson et al. 1992), and to GuHCl (Chernoff et al. 1999). We checked the chimeric Ssa-Ssb constructs for their ability to compensate for these defects. As in the case of [PSI+]* suppression, only constructs containing the peptide-binding domain of Ssb in combination with at least one other domain of Ssb origin (BBA, ABB, and BBB) were capable of compensating for sensitivity to paromomycin and GuHCl in our strains' background, while only complete Ssb (BBB) compensated for hygromycin sensitivity (Figure 4D). These results differ from previous observations by Craig's group (James et al. 1997; Pfund et al. 2001), possibly due to differences in genotypic backgrounds. Remarkably, compensation of the paromomycin and GuHCl sensitivity phenotypes by the ABB construct was stronger in the [psi−] strain than in the isogenic [PSI+] strain (not shown).
Effects of excess Ssa on the size of prion polymers:
Strong [PSI+] variants exhibiting more efficient nonsense suppression (Derkatch et al. 1996) are also characterized by decreased amount of Sup35 protein in the low-molecular-weight (monomeric) fraction (Zhou et al. 1999) and decreased average size of prion polymers (Kryndushkin et al. 2003) in comparison with the isogenic weak [PSI+] variants. Apparently, smaller Sup35 polymers are more efficient in immobilizing the monomeric Sup35 protein molecules. Despite the fact that Ssa overproduction in weak [PSI+] variants increased nonsense suppression up to a level comparable with the strong [PSI+] variants (Newnam et al. 1999), we found that extracts of both strong (Figure 5A) and weak (not shown) [PSI+] strains overproducing Ssa1 contained more monomeric Sup35 protein capable of entering the SDS-PAGE gel under nondenaturing conditions in comparison with extracts of the same strains not containing the Ssa1-encoding plasmid. Likewise, the SDD-AGE procedure (see Kryndushkin et al. 2003) detected a reproducible increase in the abundance of the monomeric Sup35 fraction in the [PSI+] strains overproducing Ssa1, accompanied by a slight but reproducible increase in the average size of the Sup35 polymers in the presence of excess Ssa1 (Figure 5B). These data confirm that effects of excess Ssa1 on [PSI+]-mediated nonsense suppression are not caused by the same mechanism that is responsible for the differences in suppression between the weak and strong [PSI+] variants.
Hsp proteins physically interact with Sup35 in vitro and in vivo:
Existing models of the Hsp's effects propose direct physical interactions between the Hsp and prion proteins (for reviews, see Chernoff 2001, 2004). Ssa and Ssb are chaperones with a broad spectrum of action that are supposed to interact with many various proteins. However, their ability to physically interact with prion proteins has never been specifically investigated. We have developed an in vitro assay for protein-protein interactions that involve the Sup35NM region. For this purpose, the Sup35NM fragment with the C-terminal poly-His tag (Sup35NM-His) was expressed in E. coli and immobilized on a Ni2+ resin. By using an immobilized Sup35NM-His as a bait, we were able to pull down the full-size Sup35 protein from the extracts of [PSI+] cells (data not shown). As Sup35 molecules have been shown to interact with each other in the [PSI+] extracts (Ter-Avanesyan et al. 1994), this result confirms that our assay is capable of detecting biologically meaningful interactions. Thus, we applied the same approach to studying interactions between Sup35NM and Hsp. Both Ssa and Ssb, but not Hsp104, were pulled down from yeast extracts by Sup35NM-His bait (Figure 6A). This is the first direct evidence of in vitro interactions between Hsp70 proteins and Sup35.
One could argue that Sup35NM-His protein produced in the heterologous (E. coli) system is not properly folded, so that Hsp70 chaperones recognize a misfolded species rather than Sup35 per se. To address this possibility, we performed co-immunoprecipitation experiments aimed at identifying in vivo interactions between Sup35 and Hsp70 (Figure 6B). Both Ssa and Ssb (but not Hsp104) were co-immunoprecipitated with HA-tagged Sup35 from the yeast cells expressing Sup35-HA. In contrast, none of these chaperones was precipitated at a comparable level with the antibody specific to yeast protein Ade2 (Figure 6B). Binding to Sup35-HA was not an artifact of HA tagging, as Sup35C-specific antibody also precipitated Ssa together with Sup35 from the extracts of yeast cells containing nontagged Sup35 (not shown). Thus, Ssa and Ssb chaperones certainly interact with Sup35 in vivo more tightly than with just any randomly chosen protein from the yeast cell.
Ssa proteins are [PSI+] helpers:
Previously, it was shown that excess Ssa1 protein antagonizes the [PSI+]-curing effect of excess Hsp104 and stimulates nonsense suppression by [PSI+] (Newnam et al. 1999) and that mutations in the SSA1 gene impair [PSI+] maintenance, most severely in the absence of wild-type Ssa1/2 protein (Jung et al. 2000; Jones and Masison 2003). Here, we demonstrate for the first time that excess Ssa1 also facilitates de novo formation of the [PSI+] prion in the [psi−] cells. We also show that other members of the Ssa subfamily (Ssa2, Ssa3, and Ssa4) exhibit the same effects on [PSI+] as does Ssa1. Thus, differential effects of various Ssa proteins on prion propagation, previously reported for [URE3] (Schwimmer and Masison 2002), are not confirmed in the case of [PSI+]. These data firmly establish Ssa proteins as [PSI+] helpers, assisting in prion formation and propagation.
Effects of excess Ssa on prions depend on the size and physical parameters of prion aggregates:
In contrast to its effect on conventional [PSI+] isolates, excess Ssa is antagonistic to some atypical derivatives of [PSI+]. These include the heterologous prion [PSI+]PS (Kryndushkin et al. 2002), formed by a chimeric Sup35 protein with a highly diverged prion-forming domain from the distantly related yeast Pichia methanolica (Chernoff et al. 2000; Kushnirov et al. 2000a; Santoso et al. 2000), and the [PSI+]Δ22/69 prion, formed by a deletion derivative of the Sup35 protein and defective in aggregate “shearing” and seed production (Borchsenius et al. 2001). Likewise, excess Ssa also increases [PSI+] curing by the dominant negative mutant of Hsp104, which also causes accumulation of the abnormally large Sup35 aggregates due to impairment of the Hsp104 function (Wegrzyn et al. 2001). Here, we show that excess Ssa also causes increased loss of [PSI+] when [PSI+] cells simultaneously overproduce Sup35.
Overproduction of Sup35 or its prion domain in the [PSI+] cells is known to result in formation of huge cytologically detectable Sup35 aggregates, usually not seen in [PSI+] cells with normal levels of Sup35 (for example, see Bailleul-Winslett et al. 2000; Chernoff et al. 2002). Recent biochemical data suggest that large polymers of Sup35 or its amyloid-forming region exhibit decreased ability to interact with Hsp104 both in vivo (Kryndushkin et al. 2003) and in vitro (Narayanan et al. 2003). Our preliminary data confirm that increased levels of Sup35 partially protect [PSI+] from the curing effect of excess Hsp104 (S. Müller, G. Newnam and Y. Chernoff, unpublished results). Remarkably, both [PSI+]PS and especially [PSI+]Δ22/69 prions are characterized by an abnormally large size of Sup35 aggregates and are relatively insensitive to the curing effect of excess Hsp104 (Kushnirov et al. 2000b; Borchsenius et al. 2001; Kryndushkin et al. 2003). Expression of the dominant negative mutant allele of Hsp104 also results in increased size of the Sup35 prion aggregates and interferes with the [PSI+]-curing effect of excess wild-type Hsp104 (Wegrzyn et al. 2001). These data strongly suggest that excess Ssa is deleterious for propagation of the large Hsp104-insensitive prion aggregates. Ability of excess Ssa1 to cure the [URE3] prion, also insensitive to excess Hsp104 (Schwimmer and Masison 2002), further confirms the existence of a reverse correlation between sensitivities of yeast prions to the curing effect of excess Ssa and excess Hsp104.
Mechanism of the Ssa effects on [PSI+]:
Our measurements detect a slight but reproducible effect of excess Ssa on the distribution of polymeric Sup35 in the extracts of prion-containing cells (see Figure 5). The average size of prion polymers is slightly but reproducibly increased in the presence of excess Ssa. This helps to explain most phenotypic effects of excess Ssa on [PSI+]. The aggregation-promoting effect of Ssa counteracts the disaggregating effect of Hsp104. Therefore, when Hsp104 is present at high levels, excess Ssa protects [PSI+] from curing. However, promotion of aggregation by excess Ssa could become detrimental if aggregates are already large and/or relatively insensitive to Hsp104. In such a scenario, excess Ssa would antagonize yeast prion propagation in the same way that depletion of Hsp104 does (Wegrzyn et al. 2001).
The simplest explanation for the aggregation-promoting effect of excess Ssa would be that Ssa downregulates production of Hsp104. However, neither our previous experiments (Newnam et al. 1999) nor new data (see above and Figure 3) have detected any observable alterations of the Hsp104 levels in our strains overproducing Ssa. Moreover, the antagonistic interactions between Ssa and Hsp104 in [PSI+] curing were also reproduced when both proteins were expressed from the galactose-inducible (PGAL) promoter, not influenced by Ssa (Newnam et al. 1999; this study, Figure 2, E and F). This rules out transcriptional regulation of HSP104 as a major component of Ssa's effects on prions.
It is possible that increased levels of Ssa inhibit Hsp104 activity or prevent Hsp104 from interacting with aggregated substrates. Although Ssa was shown to work together with Hsp104, rather than antagonizing it, during solubilization of the heat-damaged protein aggregates in vitro (Glover and Lindquist 1998), efficiency of this interaction could depend on the chaperone stoichiometry. Another possibility is that Hsp104 generates a misfolded intermediate, whose cellular fate depends on Ssa. The E. coli homolog of Hsp104, ClpB, is thought to initiate solubilization of heat-damaged proteins by increasing the exposure of hydrophobic sequences and preventing them from sticking to each other. Exposed hydrophobic regions are bound by bacterial counterparts of Hsp70, DnaK and Hsp40, DnaJ, which stabilize partially unfolded structures and promote further aggregate solubilization and refolding (Goloubinoff et al. 2000). It is possible that Hsp104's interaction with prion aggregates also leads to the exposure of protein regions involved in aggregate formation, while Ssa subsequently binds and stabilizes misfolded protein intermediates, generated by Hsp104, and prevents them from amorphous (nonprion) aggregation or degradation. However, in the presence of active prion “nuclei” such stabilization would actually increase the likelihood of these intermediates being turned back into prions. It is not clear whether these intermediates represent oligomers, misfolded monomers, or both.
The effect of excess Ssa on de novo prion formation could be explained by the same mechanism. It is logical to expect that Sup35 overproduction would lead to increased abundance of misfolded Sup35 polypeptides. Ssa could stabilize these misfolded polypeptides, protecting them from degradation and therefore increasing the possibility of their conversion into prions. It is worth noting that a stimulatory effect of excess Ssa on de novo [PSI+] formation was observed only in strains containing another Gln/Asn-rich prion, [PIN+] ([RNQ+]). [PIN+] stimulates de novo [PSI+] formation (Derkatch et al. 1997), presumably by providing initial polymerization “nuclei” (Derkatch et al. 2001).
Interestingly, neither in vitro nor in vivo assays employed in our work detected physical association between Sup35 and Hsp104 (Figure 6). Previously, in vitro interactions between Sup35 and Hsp104 were observed only under circumstances where Sup35 concentrations were high and therefore its aggregation was either observed or likely (Schirmer and Lindquist 1997; Inoue et al. 2004; Shorter and Lindquist 2004). Indeed, it has been reported (Narayanan et al. 2003) that Hsp104 interacts in vitro with oligomers but not with monomers of the Gln/Asn-rich fragment of Sup35N. On the other hand, larger polymers were not capable of binding Hsp104. Apparently, Hsp104 specifically recognizes the oligomeric isoform of Sup35, while Ssa (or Ssb) can interact with the Sup35 prion-forming region whenever it is exposed. In such a scenario, Hsp104 interacts only with a fraction of aggregated Sup35, and this interaction is necessarily of a transient nature as it constantly generates the monomeric material incapable of interacting with Hsp104. Obviously, such an interaction would be very hard to detect in the traditional biochemical assays.
In the case of the weak and strong variants of the S. cerevisiae [PSI+] prion (see Derkatch et al. 1996), a greater degree of nonsense suppression (that is, more severely affected termination of translation) usually correlates with smaller aggregate size, larger number of prion seeds, and decreased amount of the soluble (presumably monomeric) protein (Zhou et al. 1999; Kryndushkin et al. 2003). These effects are probably due to more efficient immobilization of small Sup35 molecules by more readily abundant polymeric “nuclei.” Polymeric (aggregated) Sup35 is apparently less functional than the soluble monomeric Sup35, due to poor access of the substrate (polysomes) to the aggregated protein. These observations were reproduced in our study. Weak [PSI+] variant OT55 was characterized by the larger average size of prion polymers and slightly (but reproducibly) increased proportion of the soluble Sup35 protein, in comparison to the strong [PSI+] variant OT56 (Figure 5B). Excess Ssa increased [PSI+]-mediated suppression in OT55 almost up to the level of OT56 (see Newnam et al. 1999), but this increase was not accompanied by a corresponding decrease of the polymer size and amount of the monomeric Sup35 protein. On the contrary, both the average polymeric size and the amount of low-molecular-weight protein, as estimated from nondenaturing PAGE and from SDD-AGE data, were increased in the extracts of the [PSI+] cells overproducing Ssa1 (Figure 5). Given that in the presence of excess Ssa an increase in amount of monomeric Sup35 was not accompanied by increased efficiency of translational termination, one might suggest that in this case most monomeric Sup35 protein remains nonfunctional or poorly functional. It is possible that the binding of partially misfolded Sup35 to Ssa prevents its access to the ribosomal machinery. In this way, it may further increase the defect of translation termination. In contrast, binding to Ssa apparently does not prevent Sup35 from interacting with prion seeds, possibly due to the fact that the chaperone complex including Ssa remains in close association with prion aggregates.
Recent data by Jones and Masison (2003) show that ssa1 mutations affecting [PSI+] maintenance may increase the “substrate-trapping” capability of Ssa protein. Likewise, alterations of the Hsp70 co-chaperones resulting in prolonged substrate binding further impaired [PSI+] propagation in the ssa1 mutant, while alterations promoting substrate release improved propagation of [PSI+] (Jones et al. 2004). If substrate binding capacity of Ssa is increased, release of the misfolded Sup35 protein bound by Ssa could be impaired. Consequently, its conversion back into a prion state becomes less likely. Moreover, if Ssa remains bound to Sup35 aggregates, it may prevent Hsp104 from accessing them. A likely result is the further growth of prion aggregates and decrease in the number of active seeds. Indeed, the first ssa1 mutation shown to affect [PSI+] propagation, ssa1-21, decreased the number of prion seeds as measured by the kinetics of GuHCl curing (Jung et al. 2000). Recent data on the [URE3] system (Roberts et al. 2004) also agree with the suggestion that increased substrate binding by Ssa counteracts prion propagation. These observations fit our model.
Molecular basis of the differential effects of the Ssa and Ssb proteins on prions:
In contrast to Ssa, the Ssb subfamily of Hsp70 proteins consistently behaves as a [PSI+] antagonist (Chernoff et al. 1999; Kushnirov et al. 2000b; Chacinska et al. 2001). Here, we confirm the anti-[PSI+] effect of excess Ssb by showing that it promotes [PSI+] loss when levels of Sup35 and, consequently, size of the Sup35 aggregates are increased. Most likely, Ssb acts by eliminating misfolded Sup35, which serves as a substrate for prion conversion. Shortage of such a substrate under the conditions where most of Sup35 is accumulated within huge aggregates and becomes inaccessible for disaggregating effect of Hsp104 makes the anti-[PSI+] effect of excess Ssb even more profound.
Our new data also show that differences between the Ssb and Ssa proteins with respect to their effects on [PSI+] are in significant part determined by differences in their peptide-binding domains. It therefore appears that the specificity of interactions between Hsp70 chaperones and other proteins is one of the primary determinants of Hsp70's effects on prions. The peptide-binding domain of Ssb is sufficient for the anti-[PSI+] effect even when it is combined with the ATPase and variable domains of Ssa origin. However, the presence of at least one more domain of Ssa origin in addition to the peptide-binding domain is required for the pro-[PSI+] effect (Figure 4A). Moreover, in some assays only complete Ssa aided [PSI+] at a detectable level (Figure 4B). This indicates that differences between Ssa and Ssb in regions other than the peptide-binding domain also contribute to the outcome of prion-chaperone interactions.
Ssb was implicated in the folding of nascent polypeptides (Nelson et al. 1992; Pfund et al. 1998) and in the targeting of misfolded polypeptides for ubiquitin-dependent degradation (Ohba 1997). An attractive possibility is that Ssa and Ssb proteins compete with each other for the misfolded Sup35 molecules produced as a result of either spontaneous misfolding or prion disaggregation initiated by Hsp104 (Figure 7). Ssa stabilizes these molecules in the misfolded state, making it more likely that they are turned into prions. In contrast, Ssb counteracts prion formation, as it catalyzes refolding of misfolded Sup35 molecules into their native state and/or targets them for degradation. Further experiments aimed at testing this model are currently underway.
Interestingly, when both Ssa and Ssb are overproduced in the weak [PSI+] strain OT55, suppression is increased up to the level that is comparable to the effect of excess Ssa alone (G. Newnam and Y. Chernoff, unpublished results). This indicates that a pro-[PSI+] effect of excess Ssa supercedes an anti-[PSI+] effect of excess Ssb. Such a result agrees with the observations that Ssa works on aggregated proteins together with Hsp104 (Glover and Lindquist 1998) and is colocalized to at least some Sup35-GFP clumps observed in the [PSI+] cells (R. Wegrzyn, L. Ozolins and Y. Chernoff, unpublished results). Possibly, physical association of Ssa with the prion-disaggregating machinery enables it to reach misfolded Sup35 molecules, generated in the process of disaggregation, ahead of Ssb.
The concerted effects of Hsp104 and Hsp70 chaperones provide an enzymatic machinery for the prion “life cycle” in the yeast cell. While Hsp104 is essential for the production of new prion seeds, Hsp70-Ssa promotes growth of these seeds into the larger aggregates by protecting the misfolded protein, a potential substrate for prion conversion, from refolding into the native state or from degradation. In this way, prions essentially feed on the cell stress-defense systems, using them for the purpose of propagation of the prion state. This provides a striking analogy with the selfish DNA-based genetic elements, such as viruses and transposons, that use the cell's replication machinery for their own proliferation. Moreover, our observation that variations in Ssa levels influence frequency of de novo prion formation provides a mechanism for environmental stresses to modulate generation of new heritable protein variations. Such a role for the stress defense systems in protein-based inheritance resembles the role of the DNA repair systems in fixation of DNA mutations induced by mutagens.
We are grateful to P. James and E. Craig for supplying us with the series of Ssa/Ssb chimeric constructs. We thank V. Alenin for the gift of Ade2 protein; D. Bedwell, E. Craig, D. Cyr, S. Lindquist, D. Masison, and J. Weissman for plasmids and antibodies; V. Kushnirov for useful comments regarding the SDD-AGE procedure; and L. Ozolins and Y. Yu for help in some experiments. This work was supported in part by National Institutes of Health grants to Y.O.C. (R01GM58763) and K.D.W. (R01GM30308). K.D.A. and P.A.W. were recipients of a Graduate Assistance in the Areas of National Need Predoctoral Fellowship from the Department of Education. K.B.W. was a recipient of an Undergraduate Summer Internship from the Howard Hughes Medical Research Institute.
↵ 1 These authors contributed equally to this article.
↵ 2 Present address: Center for Neurobiology and Behavior, Columbia University, New York, NY 10032.
↵ 3 Present address: Zentrum für Molekulare Biologie der Universität Heidelberg (ZMBH), D-69120 Heidelberg, Germany.
↵ 4 Present address: 8 rue des Renforts, 31000 Toulouse, France.
Communicating editor: S. Sandmeyer
- Received October 6, 2004.
- Accepted November 20, 2004.
- Genetics Society of America