The [PSI+] genetic determinant of Saccharomyces cerevisiae decreases the efficiency of translation termination and may be revealed by suppression of nonsense mutations. It exhibits unusual genetic properties, which include a non-Mendelian mode of inheritance, an ability to disappear in the presence of low concentrations of guanidine hydrochloride (GuHCl) and to reappear upon overexpression of the Sup35 protein (Sup35p) (for review, see Cox et al., 1988; Wickner et al., 1995). Study of the biochemical properties of Sup35p showed that it aggregates in the [PSI+] strains, while being soluble in the strains lacking this determinant (Patino et al., 1996; Paushkin et al., 1996). These properties are best explained by a hypothesis that [PSI+] is a phenotypic manifestation of the prion-like state of the Sup35 protein (Wickner, 1994).
Prions are infectious agents responsible for a group of diseases typified by sheep scrapie, bovine spongiform encephalopathy and human Creutzfeld–Jacob disease (for review, see Horwich and Weissman, 1997; Prusiner et al., 1998). The contemporary concept suggests that they represent a conformationally altered form (PrPSc) of normal host-encoded protein (PrPC), which has acquired an ability to convert PrPC into this altered prion form.
Several models of prion conversion have been proposed. The heterodimer (template assistance) model assumes that the prion, existing as a monomer, can form heterodimer complexes with normal molecules catalysing their prion rearrangement. The two molecules then dissociate and the cycle repeats. According to this model, the aggregation is a secondary process, non-essential for the prion conversion (Cohen et al., 1994). The nucleated polymerization model considers prions as regular polymers, or one-dimensional crystals that serve as nuclei for further polymerization of the protein. The conformational rearrangement of monomers may either occur spontaneously and then be fixed in polymer structure (Jarrett and Lansbury, 1993) or, alternatively, be directly catalysed by polymer at the moment of accretion (Horwich and Weissman, 1997). As a modification of the latter model, it was proposed that the prion conformation is adopted concomitantly with assembly, via molten oligomeric intermediates (Serio et al., 2000).
Sup35p is a yeast member of the eRF3 family of translation termination factors (Zhouravleva et al., 1995; Stansfield et al., 1995) and is composed of the amino-terminal region and carboxy-terminal (C) domain of 253 and 432 amino acids, respectively (Kushnirov et al., 1988; Ter-Avanesyan et al., 1993, 1994). The evolutionarily conserved C domain of Sup35p is responsible for its function in translation termination and is essential for cell viability, while the N-terminal region is neither conserved nor essential. This region may be further subdivided into the middle (M) domain, of unknown function, and the N-terminal (N) domain of 123 amino acids, required for [PSI+] maintenance (Figure 1). The N domain plays a key role in the [PSI+] phenomenon, being required for [PSI+] propagation in vivo and solely responsible for Sup35p prion conversion and oligomerization into amyloid-like fibrils in vitro (Paushkin et al., 1997a, b; Glover et al., 1997; King et al., 1997).
One of the most intriguing properties of prions is the existence of their different strains. In mammals, different prion strains are defined by specific incubation times, distribution of vacuolar lesions and patterns of PrPSc accumulation (for review, see Prusiner et al., 1998). In yeast they can be revealed by differences in the suppressor efficiency and mitotic stability of independently isolated [PSI+] determinants (Derkatch et al., 1996). It was shown that at the molecular level mammalian prion strain differences correlated with stable variations in the prion protein structure (Bessen et al., 1995; Safar et al., 1998; Caughey et al., 1998). However, although some difference in the rate of Sup35p aggregation in cells with various [PSI+] was revealed (Zhou et al., 1999), the phenomenon of [PSI+] variability is poorly characterized.
In this work we demonstrate that the relative frequencies of appearance of [PSI+] with weak and strong suppressor phenotype depend on the SUP35 allele used to induce their generation. We further characterize the [PSI+] strains difference and describe [PSI+] with weak suppressor phenotype that can spontaneously convert into strong suppressor determinants.
Materials and methods
Strains and media
The S. cerevisiae strains used were 5V-H19 (MATaade2-1 SUQ5 can1-100 leu2-3,112 ura3-52 [psi−]), c10B-H49 (MATα ade2-1 SUQ5 lys1-1 his3-11,15 leu1 kar1-1 cyhr [psi−]) (Ter-Avanesyan et al., 1994; Kochneva-Pervukhova et al., 1998), 1B-H67 (MATα ade2-1 SUQ5 ura3-52 lys1-1 his3-11,15 leu1 leu2-3,112 kar1-1 cyhr [psi–]) (this work). The sup35-C allele of the strain 1–5V-H19 encodes a truncated Sup35 protein lacking amino acids 1–253 and causes dominant antisuppression and inability to propagate [PSI+] (Kochneva-Pervukhova et al., 1998). This strain contained [PIN+] determinant, which allows induction of [PSI+] by overexpression of Sup35p (Derkatch et al., 1997). We used standard rich (YPD) and synthetic (SC) media for yeast (Sherman et al., 1986). Non-fermentable media contained glycerol (24 ml/l) as the sole carbon source. All solid media contained 2.5% (w/v) agar. Yeast cells were grown at 30°C.
DNA manipulations were performed using standard protocols (Sambrook et al., 1989). A series of pEMBL-yex4-based plasmids (Cesareny and Murray, 1987) containing either the complete SUP35 gene or its 3′-deletion alleles (Figure 1) has been described previously (Ter-Avanesyan et al., 1993). The pUKC815 plasmid carrying the PGK–lacZ gene fusion, and pUKC817, which is a pUKC815 derivative carrying an in-frame TAA termination codon at the junction of the PGK and lacZ genes, were described by Stansfield et al. (1995). The transformants with all the plasmids used were selected on uracil omission medium.
Standard methods of yeast genetics were used (Sherman et al., 1986). Nutrition markers were scored by growth on synthetic complete (SC) media lacking specific amino acids or nucleic acids bases. DNA transformation of yeast cells was performed as described (Gietz et al., 1995). Non-suppressive petites of the strains c10B-H49, 5V-H19 and 1B-H67 were obtained by ethidium bromide treatment (Goldring et al., 1970). Yeast strains were cured, when necessary, of the [PSI+] determinant by growth on YPD medium supplemented with 3 mM guanidine hydrochloride (GuHCl test) (Tuite et al., 1981). The [psi−] colonies of ade2-1 SUQ5 carrying strains were identified by their red colour and adenine requirement because the weak serine-inserting tRNA suppressor SUQ5 cannot suppress the ade2-1 ochre mutation in the absence of the [PSI+] determinant (Cox, 1965). The mitotic stability of different [PSI+] isolates was determined as the percentage of [PSI+] cells in their individual colonies grown on YPD medium.
The transfer of [PSI+] in ‘cytoduction’ experiments was performed as described (Ter-Avanesyan et al., 1994; Kochneva-Pervukhova et al., 1998). The recipient strain in such crosses was [psi−] [rho−], while the donor was [rho+]. In addition, one of the crossed strains carries the kar1-1 mutation that blocks karyogamy (Conde and Fink, 1976). The strains were mixed together on the surface of a YPD plate, incubated overnight, and then replica-plated to appropriate selective medium containing glycerol as the sole carbon source. Respiratory-competent ([rho+]) colonies of the recipient strain were scored as cytoductants and tested for the ability to grow on adenine omission medium (Ter-Avanesyan et al., 1994). To quantify the induction of [PSI+], transformants of the 1-5V-H19 [psi−] [PIN+] strain with multicopy plasmids carrying different SUP35 alleles were crossed with the c10B-H19 kar1-1 [psi−] [rho−] tester strain carrying the cyhr mutation. Cytoductants were selected on glycerol medium containing 3 µg/ml cycloheximide and the frequency of [PSI+] induction was estimated as described by Kochneva-Pervukhova et al. (1998).
Determination of the nonsense suppression efficiency
The UAA nonsense suppression levels were determined as ratio of β-galactosidase activity in cells transformed with pUKC817 plasmid to that in cells with pUKC815, as described previously (Stansfield et al., 1995).
Preparation, fractionation and analysis of yeast cell lysates
Yeast cultures were grown in liquid YPD medium or in a medium selective for plasmid marker to an OD600 of 1.0 (exponential phase) or, when necessary, to OD600 of 3.5 (stationary phase). The cells were harvested, washed in water and lysed by vortexing with glass beads in buffer A (25 mM Tris–HCl, pH 7.4, 100 mM NaCl, 5 mM MgCl2, 1 mM dithiothreitol) containing 1 mM phenylmethylsulphonyl fluoride (PMSF) to limit proteolysis degradation. Cell debris was removed by centrifugation at 15 000×g for 10 min. To analyse the size distribution of Sup35p by differential centrifugation, the lysates were underlaid with 1 ml 30% sucrose pads made in buffer A and centrifuged in a Beckman SW50 rotor at 45 000 r.p.m. at 4°C for 30 min. The resulting supernatants, pellets and intermediate fractions were analysed by Western blotting. To obtain sedimented material containing Sup35pPSI+, the lysates of 5V-H19 [PSI+] strain were loaded onto a sucrose layer (1 ml, 30%) and centrifuged at 200 000×g for 30 min at 4°C. The sedimented material was resuspended in buffer A for further use in conversion reactions, performed as described previously (Paushkin et al., 1997a). A ratio of Sup35pPSI+ to Sup35NMppsi− in mixtures was 1:4 in all cases. Protein samples were separated on a 10–15% SDS–polyacrylamide gel, according to Laemmli (1970), and electrophoretically transferred to nitrocellulose sheets (Towbin et al., 1979). Western blots were probed with polyclonal rabbit anti-Sup35p antibody or antibody against the Sup35N2p fragment and developed using the Amersham ECL system. Estimation of relative amount of soluble Sup35p in lysates of different [PSI+] isolates was performed as described by Zhou et al. (1999), with minor modifications.
Factors influencing the spectrum of [PSI+] strains generated de novo
The overexpression of Sup35p or its N-terminal part induces the de novo appearance of the [PSI+] determinant (Chernoff et al., 1993; Derkatch et al., 1996). This procedure generates various forms (‘strains’) of [PSI+] determinant with different nonsense suppressor efficiency and mitotic stability (Derkatch et al.1996). The C-terminal truncation of Sup35p greatly increases the frequency of [PSI+] induction (Kochneva-Pervukhova et al., 1998). Here, we observed that such Sup35p truncation also influences properties of the generated [PSI+] determinants. For the [PSI+] generation, the multicopy plasmids encoding different C-terminally truncated Sup35p variants were expressed in the strain 1–5V-H19 [psi−] [PIN+] encoding Sup35Cp, which lacks the prionogenic N-terminal region and thus cannot convert into the prion form. The use of this strain provides two advantages over the wild-type strain for the quantitative analysis of the [PSI+] generation. First, the [PSI+] induction in transformants of the 1-5V-H19 strain should be solely defined by the prion properties of the plasmid-encoded C-terminally truncated Sup35p variant. Second, this strain allows avoidance of counterselection of the appearing [PSI+] cells because, in contrast to the strains wild-type for SUP35, the combination of [PSI+] with inducing plasmids is not detrimental in the strains carryingthe SUP35-C allele (Dagkesamanskaya and Ter-Avanesyan, 1991; Paushkin et al., 1997b).
The [PSI+] phenotype is not manifested in the strain 1–5V-H19 due to the presence of non-prion Sup35Cp (Ter-Avanesyan et al., 1994). To monitor the [PSI+] generation, we transferred cytoplasm from the 1–5V-H19 transformants to a tester strain, c10B-H49 [psi−], using the ‘cytoduction’ procedure (see Materials and methods). The crosses of all tested transformants carrying plasmids with different 3′-deletion SUP35 alleles produced Ade+ cytoductants, which were shown to be [PSI+] by the GuHCl test. These [PSI+] clones differed by their suppressor efficiency and could be roughly divided into two classes: ‘strong’, with strong suppressor phenotype (white on YPD medium, and able to grow on adenine omission medium after 2 days of incubation) and ‘weak’, with weak suppressor phenotype (pink on YPD medium, and growing on adenine-free medium on the third or fourth day of incubation). Remarkably, the relative proportion of the weak and strong [PSI+] depended on the Sup35p variant used for their generation. The increased efficiency of [PSI+] induction by overexpression of short Sup35p was mainly due to the appearance of weak [PSI+] (Table 1).
Table 1. The frequency and properties of [PSI+] generated de novo depend on the structure of plasmid SUP35 alleles
Total number of [rho+] cytoductants
Number of [PSI+] cytoductants
Number of ‘strong’ [PSI+]
[PSI+] cytoductants (%)
‘Strong’ [PSI+] (%)
‘Strong’ [PSI+] among all [rho+] cytoductants (%)
The frequency and spectrum of [PSI+] induced in the strain 1–5V-H19 by multicopy plasmids carrying indicated SUP35 alleles were estimated as described in Results. In each case, where it was possible, 100 respiratory-competent Ade+ cytoductants were examined by the GuHCl test. All tested Ade+ cytoductants appeared due to [PSI+] acquisition. The [PSI+] induction data represent averages from three independent transformants. The standard deviation is indicated.
Strain-specific variation of [PSI+] properties
While we roughly divided the [PSI+] strains into ‘strong’ and ‘weak’ groups, it was of interest to study in more details their distribution by suppressor efficiency. For this, [PSI+] induced in the strain 1–5V-H19 by overexpression of Sup35NMp were transferred by cytoduction to the strain 1B-H67, bearing the suppressor-assaying plasmid pUKC817 (Stansfield et al., 1995). Cytoductants were selected on the glycerol medium supplemented with cycloheximide and checked for the ability to grow on adenine omission medium. Ade+ clones were taken for determination of nonsense codon readthrough. As a reference, the 1B-H67 [psi−] recipient tester strain carrying the pUKC815 plasmid was used. It should be noted that the strain 1B-H67 carries the SUQ5 ochre suppressor, which increases the level of UAA suppression. Among the 51 cytoductants tested, the efficiency of UAA suppression ranged from 5% to 65% (Figure 2). The distribution of [PSI+] by suppressor efficiency was normal, rather than bimodal. This argues against the existence of only two strains, but is consistent with high number of strains.
The [PSI+] isolates obtained also differed by their mitotic stability. [PSI+] isolates with strong suppressor phenotype were 100% stable during growth on YPD medium, while [PSI+] isolates with weak suppression segregated [psi−] subclones with a frequency of<0.1–4.6% (Table 2). This shows that the mitotic stability may be used for distinguishing the ‘weak’ [PSI+] strains. Remarkably, some ‘weak’ [PSI+] isolates (e.g. #81) showed a novel type of instability, being able to segregate not only [psi−], but also [PSI+] clones with a strong suppressor phenotype. This unusual ‘structural’ instability of ‘weak’ [PSI+] was a heritable trait, because it was reproduced in consecutive subclonings: pink clones always produced red, pink and white colonies. [PSI+] showing this type of instability appeared with a noticeable frequency: two of the 40 especially studied independently generated [PSI+] cytoductants belonged to this type.
Table 2. Mitotic stability of different [PSI+] strain
Type of [PSI+]
Number of clones
Number of [psi−] clones
% of [psi−] clones
Percentage of [PSI+] loss was calculated as described in Results. Data represent averages from three independent subclones of each [PSI+] isolate. The standard deviation is indicated.
The properties of [PSI+] do not depend on the host yeast strain
The prion concept for [PSI+] implies that the difference of [PSI+] strains should be independent of the host yeast strain. To check this, we studied whether the difference in properties of [PSI+] determinants isolated in one yeast strain is reproducible in another yeast strain. Three ‘weak’ and five ‘strong’ [PSI+] determinants listed in Table 2 were transferred by cytoduction from the strain c10B-H49 to 5V-H19 [psi−]. In all cases [PSI+] determinants retained their characteristic suppressor phenotype. The same was true for the structurally unstable [PSI+] isolate #81: the transfer of its cytoplasm to the 5V-H19 [psi−] recipient strain produced pink [PSI+], red [psi−], and white [PSI+] cytoductants, and the pink cytoductants were again able to segregate red and white colonies. Thus, independently isolated [PSI+] represented different strains of the [PSI+] determinant whose properties did not depend on the nuclear genetic background of the host yeast strain.
The suppressor efficiency of [PSI+] strains correlates with the properties of Sup35p
It was observed that yeast cells with [ETA+], which represents a ‘weak’ [PSI+] strain, have a higher proportion of soluble Sup35p than those with ‘strong’ [PSI+] (Zhou et al., 1999). A similar correlation was also found for the artificial [PSI+] based on the Sup35p prion domain from P. methanolica (Kushnirov et al., 2000). However, the reported proportion of soluble Sup35p in a lysate of ‘strong’ [PSI+] was very low: about 0.5% (Zhou et al., 1999). This is likely to be below the physiological range, since Sup35p is essential. By our data, the decrease of Sup35p below 5% of its wild-type levels causes growth inhibition and cell death (I. A. Valouev, unpublished). It is noteworthy that the experiments described above were performed with yeast cultures in stationary phase of growth, which could be an important factor increasing the Sup35p aggregation.
The use of cells harvested in the stationary phase was due to observation that this allows distinguishing more clearly [PSI+] and [psi−] cells by Sup35p aggregation than in the exponential phase. Therefore, the experiment was reproduced with the cells harvested in exponential phase, which gave remarkably different results: the soluble Sup35p of the strain with ‘strong’ [PSI+] constituted 6% and with ‘weak’ [PSI+] 25% of its amount in the [psi−] strain (Figure 3). It should be noted that the [psi−] reference level of Sup35p in this case was about three-fold higher than in stationary-phase cells.
The Sup35p from ‘weak’ and ‘strong’ [PSI+] strains was also distinguished by its ability to seed the prion conversion of Sup35ppsi−in vitro. Lysate of [psi−] transformant of the strain 5V-H19 with multicopy plasmid encoding Sup35NMp was mixed with sedimented material obtained from the lysates of 5V-H19 carrying either ‘strong’ or ‘weak’ [PSI+]. In both cases Sup35NMp was convertible to an aggregated (Figure 4) and protease-resistant (data not shown) form, but the amount of converted material was about two-fold lower when the reaction was seeded by the lysate of cells with ‘weak’ [PSI+]. No Sup35NMp aggregation was observed in a control experiment without addition of the Sup35pPSI+ seeds.
The protein-only concept for prions presumes that the differences between their strains can only result from conformational variations of prion protein. Indeed, the structural difference of PrPSc was demonstrated for different strains of mammalian prion (Caughey et al., 1998; Safar et al., 1998). Thus, the phenomenon of prion strains reveals the ability of prion proteins to exist not in just two, but in multiple alternative self-propagating conformations. Such an ability is interesting in itself, but it also gives an important insight into the debate about the models for prion conversion: the existence of strains is more compatible with variants of the nucleated polymerization model than with the heterodimer model. Indeed, it appears unlikely that many different prion conformations can stably exist and self-reproduce as monomers. In contrast, in the polymerization model the alternative conformations would be stabilized by intermolecular interactions and reproduced along the length of the prion polymer. The strain variation was observed for both the conventional [PSI+] of S. cerevisiae and artificial [PSI+] based on N-terminal domain of Sup35p from yeast P. methanolica (Derkatch et al., 1996; Kushnirov et al., 2000). This supports the idea that the ability to exist in multiple forms is a general property of prions, rather than a specific trait of the mammalian prions. However, the evidence for the existence of yeast prion strains is not as strong as for mammalian prions, for which it was shown that their strain-specific differences are preserved upon transmission from one animal to another. In this paper, we reinforce the evidence for the [PSI+] strain variation being related solely to Sup35p by showing that genetic properties of [PSI+] determinants independently isolated in one yeast strain are reproducible in another yeast strain.
We observed that the suppressor efficiency of [PSI+] strains correlated with the extent of Sup35p aggregation. Similar observations were made earlier for conventional and ‘artificial’ [PSI+] strains (Zhou et al., 1999; Kushnirov et al., 2000). Such a correlation may be anticipated, since the stop codon readthrough should be inversely related to the levels of soluble Sup35p. This observation indicates that the [PSI+] phenotype variation is related to the Sup35p properties, rather than to any additional mutations in other genes. The extent of Sup35p aggregation should be proportional to the rate of its prion conversion, and indeed Sup35pPSI+ seeds from ‘strong’ strains caused more efficient prion conversion of Sup35ppsi–in vitro. It is also interesting that the level of soluble Sup35p greatly depended on the growth phase of [PSI+] cells, being much lower at the stationary phase. In the ‘strong’ [PSI+] cells, this level constituted, in comparison to [psi−] cells, 6% at the exponential phase and only 1% at stationary phase. This may be explained by a decreased rate of synthesis of soluble Sup35p in stationary cells. On the other hand, the demand for functional Sup35p in stationary cells is also likely to be decreased.
Genetic properties allowed us to roughly divide [PSI+] into two classes: ‘strong’, mitotically stable [PSI+] determinants with strong suppressor phenotype, and ‘weak’, mitotically unstable [PSI+] with weak suppressor phenotype. Quantitative analysis of the suppressor efficiency of independently obtained [PSI+] has shown that they represent a heterogeneous group and may differ from each other more than 10-fold. The [PSI+] distribution by suppressor efficiency suggested that the number of [PSI+] strains is greater than two and may be rather high, but the number of strains which may be reliably distinguished is restricted by the relatively low precision in determination of suppressor efficiency. Additional strains may be also revealed among ‘weak’ [PSI+], based on their mitotic stability, which differed greatly within this group. Furthermore, some [PSI+] of this class demonstrated a novel type of instability, being able to convert into ‘strong’ mitotically stable determinants. The molecular mechanism of such inter-strain conversion is not clear, but if the conversion occurs during the replication of ‘weak’ prion aggregates, it would feature a novel phenomenon, not anticipated by a basic prion theory, the prion replication ambiguity.
In agreement with the previously published data, the efficiency of [PSI+] induction correlated inversely with the length of the plasmid-encoded Sup35p N-terminal fragment (Kochneva-Pervukhova et al., 1998). The spectrum of appearing [PSI+] also depended on the length of the Sup35p inducer fragment. Longer Sup35p fragments induced a high proportion of ‘strong’ [PSI+], whereas shorter fragments predominantly caused generation of ‘weak’ [PSI+]. The dependence of [PSI+] spectrum on the length of inducing Sup35p fragment supports the idea that [PSI+] variation is solely related to structural variation of Sup35p prion aggregates. Such dependence, indicating that different Sup35p fragments preferentially fold into different prion structures, may be explained by assuming that the ‘weak’ prion conformations are those which poorly accommodate bulky complete Sup35p. For this reason, full-length Sup35p should be inefficient not only for the formation of ‘weak’ [PSI+] determinants, but also for their replication. It should be noted that the ‘strength’ of [PSI+] was determined using complete Sup35p. Thus, it is possible that the ‘weak’ Sup35p prion conformations are not ‘weak’ per se, but only with respect to their propagation by complete Sup35p.
This work was supported by grants from INTAS, the Russian Foundation for Basic Research (to M.D.T.-A.) and the Wellcome Trust (to V.V.K.).