[PSI+] aggregate enlargement in rnq1 nonprion domain mutants, leading to a loss of prion in yeast

Authors


  • Communicated by: Yuichi Iino

nak@ims.u-tokyo.ac.jp

Abstract

[PIN+] is the prion form of the Rnq1 protein of unknown function in Saccharomyces cerevisiae. A glutamine/asparagine (Q/N)-rich C-terminal domain is necessary for the propagation of [PIN+], whereas the N-terminal region is non-Q/N-rich and considered the nonprion domain. Here, we isolated numerous single-amino-acid mutations in Rnq1, phenotypically similar to Rnq1Δ100, which inhibit [PSI+] propagation in the [PIN+] state, but not in the [pin] state, when overproduced. The dynamics of the prion aggregates was analyzed by semi-denaturing detergent-agarose gel electrophoresis and fluorescence correlation spectroscopy. The results indicated that [PSI+] aggregates were enlarged in mother cells and, instead, not apparently transmitted into daughter cells. Under these conditions, the activity of Hsp104, a known prion disaggregase, was not affected when monitored for the thermotolerance of the rnq1 mutants. These [PSI+]-inhibitory rnq1 mutations did not affect [PIN+] propagation itself when over-expressed from a strong promoter, but instead destabilized [PIN+] when expressed from the weak authentic RNQ1 promoter. The majority of these mutated residues are mapped to the surface, and on one side, of contiguous α-helices of the nonprion domain of Rnq1, suggesting its involvement in interactions with a prion or a factor necessary for prion development.

Introduction

Prions are transmissible agents caused by self-propagating conformational changes in proteins (Prusiner 1982) that propagate by capturing soluble proteins and converting them into infectious aggregated form (Prusiner 2001). According to the ‘protein only’ hypothesis (Prusiner 1982), the prion protein (PrP) is the sole agent responsible for causing numerous infectious diseases including scrapie (sheep), bovine spongiform encephalopathy (BSE, cow) and chronic wasting (deer and elk) as well as kuru and Creutzfeld–Jacob disease (humans). In fungi, prions have also been characterized as non-Mendelian inheritable elements, such as [PSI+], [URE3], [PIN+], [SWI+], [MCA], [OCT+], [MOT3+] and [GAR+] in Saccharomyces cerevisiae and [Het-s] in Podospora anserina (Wickner 1994; Coustou et al. 1997; Sondheimer & Lindquist 2000; Du et al. 2008; Alberti et al. 2009; Brown & Lindquist 2009; Nemecek et al. 2009; Patel et al. 2009). Molecular and genetic studies of these fungal prions have greatly facilitated the elucidation of the molecular basis for prion conversion and propagation as well as the general criteria for prionogenicity in a protein’s primary structure.

[PSI+] is the prion form of Sup35, which is the polypeptide release factor eRF3 that is essential for terminating protein synthesis at stop codons (Stansfield et al. 1995; Zhouravleva et al. 1995; for a review see Ehrenberg et al. 2007). When Sup35 is in the [PSI+] state, ribosomes often fail to release polypeptides at stop codons, causing a non-Mendelian trait easily detected by nonsense suppression (Liebman & Sherman 1979; Patino et al. 1996; Paushkin et al. 1996). The [PSI+]-mediated suppression of auxotrophic markers ade1-14 or ade2-1 (nonsense alleles) has been widely used as nonsense suppression can be easily detectable by a color assay. [PSI+] allows cells to grow on synthetic medium lacking adenine and prevents the buildup of adenine metabolites that would cause [psi] cells (soluble Sup35, no nonsense suppression) to turn red on rich media. In contrast to [PSI+] selection as Ade+ or white colonies, a convenient genetic tool for [psi] selection had long been unavailable. Recently, we have developed the chromosomal ura3-197 (non-sense mutation) marker, which works as a powerful tool to select for the [psi] state, in which yeast cells become viable in the presence of 5-fluoroorotic acid (5-FOA) that is toxic to the [PSI+] cells (Kurahashi & Nakamura 2007; Kurahashi et al. 2008). Using this tool, we have isolated several mutations or effectors that eliminate [PSI+], such as chromosomal mutations in the HSP104 gene (Kurahashi & Nakamura 2007), a G-protein γ subunit homologue, Gpg1 (Ishiwata et al. 2009), and an N-terminal deletion mutant of Rnq1, Rnq1Δ100 (Kurahashi et al. 2008).

Rnq1 is a protein of unknown function and is one of several known yeast proteins containing a Q/N-rich prion domain, hence named so for rich in asparagine (N) and glutamine (Q) (Sondheimer & Lindquist 2000). Rnq1 forms the prion [PIN+] for [PSI+] inducibility (Sondheimer & Lindquist 2000; Derkatch et al. 2001; Osherovich & Weissman 2001) because [PIN+] is required for efficient de novo induction of [PSI+] (Derkatch et al. 1997), but not for [PSI+] propagation (Derkatch et al. 2000). Although it is known that several other yeast Q/N-rich proteins can be attributed to the Pin+ phenotype (Derkatch et al. 2001), [PIN+], also known as [RNQ+], always refers to the prion form of Rnq1 in this article. Two models, ‘seeding’ and ‘titration,’ have been proposed to explain how heterologous prions, e.g. [PIN+], facilitate the de novo appearance of [PSI+]. According to the seeding model, a heterologous pre-existing protein in the prion conformation templates the conversion of Sup35 into its prion form, which then proceeds to seed its own rapid and separate aggregation. Importantly, [PIN+] also facilitates the de novo appearance of the prion [URE3] and generally promotes polyglutamine (polyQ) aggregation and toxicity (Osherovich & Weissman 2001; Bradley et al. 2002; Meriin et al. 2002). Therefore, the seeding model predicts that [PIN+] aggregates provide a ‘friendly’ nidus on which the first seeds of a heterologous prion or polyQ amyloid can form (Derkatch et al. 2004; Vitrenko et al. 2007). The alternative titration model postulates that pre-existing heterologous prions or prion-like aggregates capture and inactivate an inhibitor that prevents the conversion of Sup35 into a prion (Derkatch et al. 2001; Osherovich & Weissman 2001). Whereas several observations favor the seeding model, a definitive experiment has neither proved nor disproved either of the two models.

The Rnq1 protein is composed of a non-Q/N-rich N-terminus (amino acids 1–152) and a Q/N-rich C-terminus (amino acids 153–405). It is widely accepted that the prion-forming ability of prion proteins resides within the Q/N-rich regions (Masison et al. 1997; DePace et al. 1998; Sondheimer & Lindquist 2000; Patel & Liebman 2007). The C-terminus of Rnq1 was shown to be necessary for the propagation of the prion aggregate (Sondheimer & Lindquist 2000). However, a deletion mutant of N-terminal nonprion domain (amino acids 1–100), Rnq1Δ100, shows several intriguing properties. First, Rnq1Δ100 is capable of forming and transmitting the [PIN+] prion in accordance with the aforementioned findings (Kurahashi et al. 2009). Second, [RNQ1Δ100+], the prion form of Rnq1Δ100 without full-length Rnq1, is selfish in that it eliminates a heterologous prion in S. cerevisiae (Kurahashi et al. 2009). Third, over-expression of Rnq1Δ100 inhibits the maintenance of not only [PSI+] but also [URE3] and Huntingtin’s polyglutamine (polyQ) aggregate in a [PIN+] background, but not in a [pin] background (Kurahashi et al. 2008). Rnq1Δ100, however, does not eliminate [PIN+] itself. Together, these observations suggest that Rnq1-Rnq1Δ100 co-aggregates in the [PIN+] state interact with other transmissible and nontransmissible amyloids to destabilize and eliminate their amyloid form. Furthermore, we recently reported N-terminal mis-sense rnq1 mutations that are defective in the stable [PIN+] propagation (Shibata et al. 2009).

To evaluate the Rnq1Δ100 phenotype comprehensively and to clarify the functional significance of the N-terminal nonprion domain of Rnq1 in detail, we isolated single-amino-acid rnq1 mutations whose over-expression is inhibitory to the propagation of [PSI+] in the [PIN+] state. Collectively, we found 28 rnq1 alleles in the N-terminal non-Q/N-rich region and no single allele within the C-terminal Q/N-rich region. The N-terminal region of Rnq1 contains five predicted α-helices, and these mutations frequently locate to the surface and on one side of these helices.

Results

Phenotypes shared between Rnq1Δ100 and Rnq1-L94A mutant proteins

Similar to other yeast prions, Rnq1-GFP fusion proteins form foci and SDS-stable polymers in [PIN+] cells, whereas they are cytoplasmically dispersed and do not form SDS-stable polymers in [pin] cells (Sondheimer & Lindquist 2000; Bagriantsev & Liebman 2004). In contrast, Rnq1Δ100 has a strong activity to self-aggregate or co-aggregate with Rnq1, independent of the [PIN+] or [pin] state, and the [pin] Rnq1Δ100 aggregates are SDS unstable (Kurahashi et al. 2008). These Rnq1Δ100 properties are similar to those reported with the Rnq1-L94A mutant protein, which has drawn considerable attention in the Rnq1 study. In [pin] cells, Rnq1-L94A exhibits a higher propensity than wild-type Rnq1 to coalesce into foci and forms high-molecular-weight aggregates that are SDS-unstable (Douglas et al. 2008).

Owing to these shared properties between Rnq1Δ100 and Rnq1-L94A, we then asked whether over-expression of Rnq1-L94A eliminates [PSI+] and [URE3] prions in [PIN+] cells. The presence or absence of [PSI+] was monitored by the color assay based on the ade1-14 allele as described earlier, i.e. white in [PSI+] and red in [psi] (Fig. 1A). Two independent [PSI+] strains were examined, one, NPK294 ([PSI+] [PIN+]) and its isogenic [pin] strain NPK299 and the other, NPK50 ([PSI+] [pin]) and its isogenic [PIN+] strain NPK300. Upon transformation with an empty plasmid or a wild-type Rnq1 plasmid, these strains remained white (Fig. 1C,D). However, two [PSI+] [PIN+] strains (NPK294 and NPK300), but not [PSI+] [pin] strains (NPK299 and NPK50), turned red upon transformation with plasmids overproducing Rnq1Δ100 or Rnq1-L94A (Fig. 1E,F).

Figure 1.

 [PSI+] elimination by Rnq1-L94A and Rnq1Δ100. (A) Color assay of [PSI+] and [URE3] strains. [PSI+] (NPK294), its isogenic [psi] (NPK200), [URE3] (NPK302) and its isogenic [ure-o] (NPK301) cells were grown on YPD, and their colors were monitored as control. (B–F) The strains indicated in (B) were transformed with empty plasmid, pRS413ADHp (C), or its derivatives carrying wild-type RNQ1 (D), rnq1Δ100 (E), or rnq1-L94A (F). The resulting transformants were selected on SC-his plates and then passaged onto YPD plates for color assay.

[URE3] is the prion form of Ure2 (Wickner 1994), which is a regulator of nitrogen metabolism (Magasanik & Kaiser 2002). The [URE3] prion was monitored using test strains developed by Reed Wickner and colleagues (Brachmann et al. 2005), in which active Ure2 (i.e. in the [ure-o] state) negatively regulates the DAL5 promoter fusion to the ADE2 gene, producing red colonies on YPD, whereas the inactive Ure2 aggregate (i.e. in the [URE3] state) fails to turn off DAL5-ADE2 expression, producing white colonies on YPD (Fig. 1A). In this assay, NPK302 ([URE3] [pin]) and its isogenic [PIN+] strain NPK435 cells remained white upon transformation with the empty plasmid or the wild-type Rnq1 plasmid (Fig. 1C,D). However, the [URE3] [PIN+] strain (NPK435), but not [URE3] [pin] strains (NPK302), turned red upon transformation with plasmids overproducing Rnq1Δ100 and Rnq1-L94A (Fig. 1E,F). These findings indicate that over-expression of Rnq1-L94A is inhibitory to both [PSI+] and [URE3] propagation in the same manner as Rnq1Δ100.

[PIN+]-destabilizing mutations in Rnq1 interfere with heterologous prion propagation

During the course of separate studies, we became aware of another phenotype of Rnq1-L94A. When Rnq1-L94A was expressed from the authentic RNQ1 promoter in rnq1Δ deletion strain at levels equivalent to wild-type Rnq1, cells became defective in stable [PIN+] propagation (Shibata et al. 2009). This observation let us to speculate that an rnq1 mutant partially defective in the [PIN+] propagation might be generally inhibitory to heterologous prion propagation when over-expressed. This possibility was preliminarily monitored using nine such rnq1 alleles in our collection that are defective in the stable [PIN+] propagation, i.e. S12P, S15P, V23A, E43K, V53A, L91P, L123P, F146S and N397D (Shibata et al. 2009). A comprehensive selection of rnq1 alleles that destabilize [PIN+] was made through random mutagenesis of the whole RNQ1 sequence (Shibata et al. 2009). These were cloned into the expression plasmid pRS413ADHp, transformed into NPK50 cells ([PSI+] [pin]) or NPK294 cells ([PSI+] [PIN+]) and expressed from the strong ADH promoter. As expected, all these rnq1 mutants interfered with [PSI+] propagation except for N397D in [PIN+] cells; though, two alleles L123P and F146S inhibited [PSI+] less efficiently than the others (Fig. 2). Furthermore, all mutants failed to interfere with [PSI+] propagation in [pin] cells. It is remarkable that the nine rnq1 mutations including L94A are mapped within the N-terminal nonprion domain of Rnq1 (Fig. 3B).

Figure 2.

 Over-expression of [PIN+]-destabilizing rnq1 mutations eliminates [PSI+], similarly to Rnq1Δ100, in [PIN+] cells. NPK50 ([PSI+] [pin]) and NPK294 ([PSI+] [PIN+]) cells were transformed with plasmid pRS413ADHp (empty vector, vec.) and its derivatives carrying wild-type (WT) or the indicated rnq1 alleles. Transformants were selected on SC-his plates and then passaged onto YPD plates. Enlarged views from the NPK294 transformations are shown under the original plates.

Figure 3.

 Mapping the [PSI+]-inhibitory rnq1 mutations. (A) Localization of rnq1 mutations to the N-terminal domain of Rnq1. Schematic diagram of the non-Q/N-rich and Q/N-rich regions of the Rnq1 protein is shown. Numbers represent amino acid positions, and positions of amino acid substitutions are indicated. (B) Localized distribution of rnq1 mutations in the α helices. Secondary protein structure prediction of non-Q/N-rich region in the Rnq1 protein generated from the PSIPRED Protein Structure Prediction Server [http://bioinf.cs.ucl.ac.uk/psipred] (Jones 1999) is shown. Circular cylinders and lines indicate α-helices and coils, respectively. The [PSI+]-inhibitory rnq1 mutations isolated in this study are shown in chromatic colors (red and green); red indicates the mutations newly discovered in this work; green and black indicate the mutations that have been previously characterized as [PIN+] destabilizers by our group (Shibata et al. 2009). Over-expression of these mutations, with the exception of the C-terminal N397D, is inhibitory to the maintenance [PSI+]. (C) Secondary protein structure predictions of the non-Q/N-rich region in the Rnq1 protein generated by PSIPRED, DSC (King & Sternberg 1996), MLRC (Guermeur et al. 1999), PHD (Rost & Sander 1993) and Predator (Frishman & Argos 1996). rnq1 mutations are denoted in red. Secondary structure elements are shown as magenta ‘H’ (α helices), cyan ‘E’ (extended/β strands) and black ‘c’ (random coil).

Isolation of [PSI+]-eliminating rnq1 alleles

Next, we performed selection of rnq1 mutations that block [PSI+] propagation. Strain NPK294 ([PSI+] [PIN+]) was transformed with a PCR-mutated rnq1 plasmid (pRS413ADHp-rnq1: see Experimental procedures, marked with HIS3), and His+ transformants were first selected on synthetic complete (SC)-his medium. Then, His+ colonies were replica-plated onto YPD plates for color selection. Approximately 1000 clones were screened and 45 ‘red’ mutants were isolated. The resulting [psi] mutants were characterized by DNA sequencing. Of the 45 mutant clones examined (including those carrying multiple mutations in rnq1), we finally defined 23 distinct single-amino-acid substitutions that eliminate [PSI+]. Of these, four mutations overlapped with those isolated previously as [PIN+]-destabilizing mutations (Shibata et al. 2009), and the other 19 mutations were novel. Consistent with the [PSI+]-eliminating ability of Rnq1Δ100, these 23 rnq1 mutants eliminate [PSI+] in [PIN+] cells, but not in [pin] cells (data not shown). Interestingly, all these rnq1 mutations are localized within the N-terminal nonprion domain of Rnq1 (Fig. 3A,B).

[PSI+]-eliminating rnq1 mutants impaired [PIN+] propagation in the normal expression level

Of the newly isolated 19 alleles in rnq1, four representative mutations, S8P, A29T, M101K and L138Q, were expressed from the authentic RNQ1 promoter and examined for their effects on [PIN+] propagation, using the plasmid shuffle and a Rnq1Δ100-based color assay. Rnq1Δ100 eliminates the [PSI+] prion in the [PIN+] state but not in the [pin] state. This allows us to find loss-of-[PIN+] rnq1 mutants as white, [PSI+] colonies (Shibata et al. 2009). The parental strain NS42 was [PSI+] [PIN+] ade1-14, its chromosomal RNQ1 gene was deleted by rnq1::KanMX (rnq1Δ), and Rnq1 expression was achieved by pRS416RNQ1p-RNQ1 (URA3 marker) using the authentic promoter of RNQ1. Then, this strain was transformed with pRS415RNQ1p derivatives carrying the S8P, A29T, M101K and L138Q mutant rnq1s (marked with LEU2). The resulting Ura+ Leu+ transformants were selected in SC-ura-leu liquid and incubated for 18 h and then cultured in SC+ura-leu liquid for 3 days. These cells were transformed with pRS413ADHp-rnq1Δ100 (ARS/CEN, HIS3 marker) and grown on SC-leu-his+5-FOA plates. Leu+ His+ 5-FOA-resistant colonies (i.e. doubly transformed with pRS415RNQ1p-rnq1 and pRS413ADHp-rnq1Δ100 but not with pRS416RNQ1p-RNQ1) were passaged onto YPD plates for color assay (Fig. 4A). As shown in Fig. 4B,C, upon substitution of rnq1 mutations, transformants with each individual mutation remained completely white at a frequency that varies among the mutations, indicating that S8P, A29T, M101K and L138Q mutant Rnq1s are partially defective, not completely, in stable [PIN+] propagation. These are consistent with the nine Rnq1 mutants including Rnq1-L94A that are defective in the stable maintenance of [PIN+] (Shibata et al. 2009). Importantly, the [PIN+]-destabilizing phenotype of rnq1 mutants was compensated for by over-expression of these mutant Rnq1s from the ADH promoter (Shibata et al. 2009). These results indicate that the same repertoire of nonprion domain alterations in Rnq1 exhibits pleiotropic effects on prion propagation, i.e. destabilization of [PIN+] and inhibition of [PSI+].

Figure 4.

 [PSI+]-inhibitory rnq1 mutations are defective in stable [PIN+] propagation. (A) Schematic of assessment for [PIN+] propagation. A wild-type RNQ1 plasmid in strain NS42 ([PSI+] [PIN+] rnq1::KanMX) was replaced with mutant rnq1 plasmids by plasmid shuffle. These cells were further transformed with a Rnq1Δ100 plasmid that is able to eliminate [PSI+] prion when cells are in [PIN+] but not in [pin] state. Therefore, red colonies (i.e. [psi]) appeared in [PIN+] transformants on YPD plate, whereas white colonies (i.e. [PSI+]) remained white in [pin] transformants. (B) Plasmid shuffled cells containing plasmid pRS415 (empty vector, vec.) or its derivatives carrying wild-type (WT) or mutant rnq1 alleles were transformed with pRS413ADHp-rnq1Δ100, and 10 independent double transformants were randomly chosen and plated on YPD. (C) Semi-quantitative analysis of rnq1 mutants defective in [PIN+] propagation. The frequency of loss-of-[PSI+] transformants is shown as the mean of three independent experiments shown in (A) with standard deviations.

[PSI+] aggregate enlargement detected by the semi-denaturing gel electrophoresis

To investigate the molecular details of how rnq1 mutations affect [PSI+] propagation, we examined the dynamics of [PSI+] aggregates by semi-denaturing detergent-agarose gel electrophoresis (SDD-AGE) and fluorescence correlation spectroscopy (FCS). For these experiments, we used strains with GFP integrated in the endogenous SUP35 ORF (Kawai-Noma et al. 2009). Three representative mutations (A29T, L94A, M101K) were expressed from the GAL1 promoter in a [PSI+] [PIN+] strain. Cells were harvested 0, 24 and 48 h after the galactose-responsive induction of rnq1 alleles, and their lysates were subjected to SDD-AGE, followed by Western blot analysis. As shown in Fig. 5A, the average size of detergent-resistant Sup35-GFP ([PSI+]) aggregates increased after the expression of rnq1 mutations. The amount of these aggregates tended to decrease upon longer incubation, and instead the amount of monomers increased, in accordance with the [PSI+] elimination phenotype (Fig. 5A,B). Similar dynamics of Sup35-GFP ([PSI+]) aggregates was observed in a strain expressing Rnq1Δ100 (Fig. 5A,B), suggesting a common molecular mechanism of [PSI+] elimination between Rnq1Δ100 and the rnq1 mutations isolated in this study. Apparent difference in the total amounts of Sup35-GFP proteins in the SDD-AGE image might be attributable to the poor sensitivity of the monomer bands. In general, small molecules like monomeric proteins are detected relatively less quantitatively in SDD-AGE. We confirmed by SDS–PAGE using anti-Sup35 antibody that no significant alteration occurred in the cellular abundance of Sup35-GFP, suggesting that mutant Rnq1 proteins do not affect the expression level of Sup35-GFP (Fig. 5A). Decrease in the amount of Sup35-GFP polymers occurred reproducibly in the [PSI+] control at 48 h, but the reason is currently unknown. The same lysates were subjected to the FCS analysis to evaluate the Sup35-GFP ([PSI+]) aggregate size in a quantitative manner.

Figure 5.

 Effect of over-expressed rnq1 mutant proteins on [PSI+] aggregate formation. (A) SDD-AGE analysis of [PSI+] aggregates upon induction of rnq1 mutants. The [PSI+] [PIN+] strain (NPK620) was transformed with pRS413GAL1p-RNQ1s carrying the indicated mutations. Transformants were harvested at the indicated time points after induction of rnq1 alleles by addition of galactose, and lysates were examined by SDD–AGE analysis using anti-Sup35 antibody (upper) and SDS–PAGE analysis using anti-Sup35, anti-Rnq1 and anti-Pgk1 antibodies, respectively (lower). P and M denote polymer and monomer fractions of Sup35-GFP. Asterisks indicate probable degradations of Sup35-GFP or Rnq1s. (B) Semi-quantitative estimation of the degree of [PSI+] elimination. At least fifty colonies from 3 independent transformants were grown on YPD plates, and the frequency of red colonies was scored and represented with standard deviations.

Asymmetric distribution of [PSI+] aggregates in single mother/daughter pair

Next, we measured the size of the diffused Sup35-GFP aggregates in the rnq1 cells by exploiting FCS. FCS is a technique to determine the diffusion coefficients of fluorescence molecules by calculating the autocorrelation function in a microscopic detection volume under 10−15 L (1 femtoliter) defined by a tightly focused laser beam and pinhole, providing us with an estimation of the size of aggregates. Taguchi and colleagues have shown that this method is applicable to budding yeast for monitoring the distinct diffusion times of fluorescent Sup35-GFP aggregates in [PSI+] and [psi] living cells in a noninvasive manner (Kawai-Noma et al. 2006).

First, the same lysates used in the SDD-AGE analysis were subjected to the FCS analysis. Consistent with the aforementioned result, the FCS results indicated that the average size of Sup35-GFP ([PSI+]) aggregates was increased at 24 h after induction of rnq1Δ100, and then the amount of aggregates was dramatically reduced and instead the amount of monomers was increased at 48 h after rnq1Δ100 induction (Fig. 6A). This dramatic decrease of Sup35-GFP aggregate ratio was also detected for the rnq1 A29T mutation (data not shown). However, relative ratios of Sup35-GFP aggregates to monomers for rnq1 L94A and M101K mutations did not alter appreciably during 48-h incubation, although larger aggregates were occasionally detected (data not shown). These results suggest that rnq1 mutations are prone to enlarging Sup35-GFP ([PSI+]) aggregates, resulting in the decrease in the number of [PSI+] seeds that leads to a loss of prion. This seemingly resembles the loss of [PSI+] prion with GuHCl treatment that inhibits Hsp104, a protein-remodeling factor that disassembles denatured protein aggregates including prions (Kryndushkin et al. 2003; Kawai-Noma et al. 2010).

Figure 6.

 Diffusional properties of Sup35-GFP in [PSI+] cells upon expression of Rnq1Δ100. (A) Representative normalized fluorescence autocorrelation functions of Sup35-GFP in lysates from [psi] cells (green, showing monomer fraction of 90% with 38 μm2/s and aggregate form of 10% with 2.5 μm2/s) and [PSI+] cells harvested before rnq1Δ100 induction (black, aggregate fraction of 82% with 2.3 μm2/s), 24 h (blue, aggregate fraction of 80% with 1.2 μm2/s) and 48 h (red, aggregate fraction of 33% with 2.9 μm2/s) after rnq1Δ100 induction. Fitting results by a two-component model are depicted by solid lines. (B) Traces of average fluorescence intensities (count per second; cps) for Sup35-GFP in a single [PSI+] cell pair of mother (M, red) and daughter (D, black) 48 h after rnq1Δ100 induction and traces of fluorescence intensities for Sup35-GFP in a single [psi] cell (green) and a single [PSI+] cell (blue) without rnq1Δ100. Insert: The fluorescent image shows the merge of confocal and light microscopic images of the mother and daughter cell pair used for FCS measurement. The white circles show the positions of the FCS measurement. Bar: 2 μm. (C) Normalized fluorescence autocorrelation functions of Sup35-GFP in a single [PSI+] cell pair of mother (red) and daughter (black) 48 h after rnq1Δ100 induction and in a single [psi] cell (green) and a single [PSI+] cell (blue) without rnq1Δ100 induction. Each autocorrelation functions were respectively calculated from the fluorescence intensity fluctuations shown in (B). Fitting results of functions by a two-component model are depicted by solid lines.

For more detailed understanding of the size change of Sup35-GFP aggregates and the process of loss of [PSI+], we analyzed the dynamics of Sup35-GFP in single living cells that express rnq1Δ100, as described by Taguchi and colleagues (Kawai-Noma et al. 2009). Single cell analysis showed that cells can be categorized into three diffusional types in terms of mother and daughter cell states 48 h after expression of rnq1Δ100: the first type shows that both mother and daughter cells exhibit slow intracellular diffusion of Sup35-GFP, showing [PSI+] state; the second type shows much slower diffusion in mother cell but fast diffusion in daughter cell; and the third shows fast diffusion both in mother and in daughter cells, showing [psi] state. Fig. 6B,C show the second type of FCS measurement on a representative mother and daughter cell pair under rnq1Δ100 over-expression. Strikingly, the mother cell had freely diffusing Sup35 aggregates with high fluorescent intensity over average intensity and the high broad peaks directly reflect the right-shifted correlation function in Fig. 6B (red), which had much larger diffusional component than [PSI+] cells without expression of rnq1Δ100 as a control (Fig. 6B,C). Furthermore, the number of aggregates in the mother cell (= 5, N is the average of total number of fluorescent molecules in the detection volume defined in eqn 2, see Experimental procedures) was smaller than that in the control cells with similar average fluorescent intensity (= 10 for [PSI+], = 25 for [psi]). In contrast, the daughter cell only had stationary fluctuation of fluorescent intensity without high peaks of fluorescence and had fast diffusion similar to that of [psi] cells (Fig. 6B,C). It should be noted that immobilized large foci, which are a hallmark for [PSI+], and rapid photobleach, which can result from immobilized aggregates, were not detected. This result from the single cell analysis may show that the size of the diffusing and enlarged aggregates observed in the mother cell with expression of rnq1Δ100 is directly related to the physical size limitation of the aggregate for transmission from mother to daughter cells, accounting for [PSI+] elimination that occurred at this time point (Kawai-Noma et al. 2009).

Effect of rnq1 mutations on expression and activity of Hsp104

Hsp104 is required for the propagation of [PSI+], [URE3] and [PIN+] prions because it breaks up amyloid filaments to generate prion seeds for efficient prion transmission (Chernoff et al. 1995; Paushkin et al. 1996; Ness et al. 2002; Shorter & Lindquist 2004). To examine whether the prion elimination by rnq1 muations is mediated through disabled functionality of Hsp104 or not, we monitored the cellular level of Hsp104. Western blotting using anti-Hsp104 antibody showed that over-expression of rnq1s slightly increased the cellular abundance of Hsp104 (Fig. 7A). This slight increase in the Hsp104 level is unlikely to be the cause of the elimination of [PSI+] or [URE3], because [URE3] cannot be eliminated by over-expression of Hsp104 but can rather be eliminated by over-expression of rnq1-L94A and rnq1Δ100 (Fig. 1). Furthermore, over-expression of Hsp104 is considered to reduce the size of Sup35 aggregates by its disaggregating activity. Then, the activity of Hsp104 was examined by monitoring thermotolerance because Hsp104 is a heat-inducible protein required for thermotolerance and acts to resolubilize protein aggregates generated by severe stress, thereby allowing yeast to survive an otherwise lethal stress (Parsell et al. 1994). To measure thermotolerance, transformants with plasmids carrying rnq1 mutations expressing from the constitutive strong ADH promoter were exposed to 37 °C for 60 min as a pre-treatment to induce the heat-shock response and subsequently incubated at 50 °C for 20 min. The extent of survival was determined through fivefold serial dilutions spotted to YPD plates at 30 °C (Fig. 7B). In contrast to the hsp104 strain that showed no thermotolerance, exogenous rnq1 alleles-bearing NPK200 strain showed no change in the thermotolerance and was comparable to the exogenous rnq1-free strains. These results showed that prion elimination by rnq1 mutations is not caused by reduced functionality of Hsp104.

Figure 7.

 Effect of overproduced rnq1 mutant proteins on expression and activity of Hsp104. (A) The cellular abundance of Hsp104. Equal amounts of cells used in Fig. 5A were examined by SDS–PAGE analysis using anti-Hsp104 antibody. (B) Thermotolerance of rnq1-overproducing strain. [psi] [pin] hsp104Δ strain (NPK372) or [psi] [PIN+] HSP104+ strain (NPK200) was transformed with pRS413ADHp-RNQ1s. The viability of heat-shocked cells (lower) was compared with untreated cells (upper) by fivefold serial dilutions on YPD media. Cells were incubated at 30 °C for 2 days.

[PSI+]-eliminating rnq1 alterations localize to the surface on one side of the α-helices of Rnq1

The PSIPRED secondary protein structure prediction (Jones 1999) indicates that the N-terminal domain of Rnq1 is rich in α-helical structures, designated α1 through α5 (Fig. 3). Of 25 residues affected by 28 rnq1 mutations isolated in this study and in the previous study (Shibata et al. 2009), 19 residues are mapped to these α-helices (Fig. 3B). Four other algorithms, DSC (King & Sternberg 1996), MLRC (Guermeur et al. 1999), PHD (Rost & Sander 1993) and Predator (Frishman & Argos 1996), predict more or less similar secondary structures (Fig. 3C). Interestingly, α1, α2 and α3 are amphiphilic in nature, with asymmetric distribution of nonpolar and hydrophilic amino acid residues around the helices, as shown in the helical wheel projection (Fig. 8). Importantly, 14 of 19 rnq1 residues are localized to one side of these α-helical surfaces, and the other five are proline substitutions mostly mapped on the other side of the α1 and α2 surfaces (Fig. 8). Proline substitutions are known to diminish a helical structure. Therefore, it is tempting to speculate that the α-helical surfaces are involved in protein–protein interactions and that the loss-of-activity substitutions would disable or alter these protein interactions.

Figure 8.

 Helical wheel projection of residues in α1-α5 of Rnq1. Residues 3–15 (helix α1), 19–30 (helix α2), 36–53 (helix α3), 87–96 (helix α4) and 138–145 (helix α5) of Rnq1 predicted by PSIPRED are shown in a helical wheel projection. Acidic residues are shown in pink, basic residues in blue, uncharged residues in green and nonpolar residues in yellow. Numbers of residues affected by loss-of-[PSI+] mutations are shown in red. Numbers in parentheses indicate the residues substituted to proline.

Discussion

In this study, we isolated 23 distinct rnq1 mutations whose over-expression is deleterious to [PSI+] maintenance in [PIN+] cells, but not in [pin] cells, a phenotype similar to Rnq1Δ100. Of these 23, four mutations overlapped with those in ten mutations previously defined as [PIN+]-destabilizing mutations (Shibata et al. 2009) and the other 19 mutations were novel (Note that the previous rnq1 isolates by Shibata et al. have not been examined for the [PSI+] propagation). This raised a possibility that [PIN+]-destabilizing mutations might be generally deleterious to [PSI+] when over-expressed. This turned to be the case for the remaining five [PIN+]-destabilizing mutations except for the C-terminus mutation (see Fig. 2). Conversely, it was shown that four rnq1 alleles isolated as [PSI+]-inhibitory mutations are partially defective in stable [PIN+] propagation when expressed from its authentic (relatively weak) RNQ1 promoter, a phenotype similar to the [PIN+]-destabilizing rnq1 mutant (Fig. 4). Therefore, it is likely that these rnq1 mutations exert pleiotropic effects on [PIN+] and [PSI+], depending on the level of expression. Importantly, all the 28 rnq1 mutations including the known [PIN+]-destabilizing mutations are mapped within the N-terminal nonprion domain of Rnq1 at 25 amino acid positions.

It is widely accepted that the prion-forming ability of prion proteins resides within the Q/N-rich regions (Masison et al. 1997; DePace et al. 1998; Sondheimer & Lindquist 2000; Patel & Liebman 2007). Consistent with this paradigm, the C-terminus of Rnq1 is composed of multiple rich Q/N tracts and is known to be sufficient for joining or forming a prion aggregate (Sondheimer & Lindquist 2000). In contrast, the N-terminus is non-Q/N rich and is known to be dispensable for the prion-forming ability and is therefore referred to as the nonprion domain (Kurahashi et al. 2008). However, this study and the previous study (Shibata et al. 2009) indicate that N-terminus of Rnq1 is involved for proper functioning of the C-terminal prion domain to propagate [PIN+]. Not only the deletion of the N-terminal 100 amino acids of Rnq1, Rnq1Δ100, but also many single-amino-acid substitutions in the N-terminal domain exhibit the same phenotype (Kurahashi et al. 2008, 2009; Shibata et al. 2009).

The previous study indicated that [PIN+]-destabilizing mutations are defective in stable maintenance of [PIN+] when expressed from the authentic (relatively weak) RNQ1 promoter, whereas the activity was mostly compensated for by over-expression of mutant rnq1 proteins from the strong ADH promoter (Shibata et al. 2009). This indicated that these rnq1 mutants might not be complete loss-of-function mutants but retained some residual activity. However, the [PSI+]-eliminating activity of all these 28 rnq1 mutants was observable when they were expressed from the ADH promoter, but not from the RNQ1 promoter (data not shown). It has been reported that cells expressing Rnq1-L94A showed a severe reduction of de novo appearance of [PSI+] (Bardill et al. 2009). As they used the strong GPD promoter to express Rnq1-L94A in [PIN+] cells, the observed reduction is therefore likely due to [PSI+] elimination, rather than to a defect in [PSI+] induction.

How do rnq1 alterations inhibit the propagation of [PSI+]? Sup35 aggregates were enlarged upon over-expression of rnq1 mutations (Figs 5 and 6). Furthermore, we observed in the mother/daughter pair cell analysis that, during the eliminating process, mother had larger and fewer aggregates and daughter had no aggregates (Fig. 6B,C). These results indicated that cells might be first impaired in its disaggregating activity of prion, which then led to the decrease in the number of prion seeds and finally resulted in the prion loss. When the aggregates of Rnq1s are overproduced, the aggregates may titrate some disaggregating chaperone and then cause the instability of other prion propagations. One likely candidate of titrated chaperones is Hsp104, which directly disaggregates prions (Shorter & Lindquist 2004). However, over-expression of rnq1 mutations did not inhibit the thermotolerant activity of Hsp104 (Fig. 7B), suggesting that Hsp104 may not be a direct target for this titration model. Another candidate is Sis1, a member of Hsp40, which is considered to act as the specific J-protein partner of Ssa and collaborates with Hsp104 in the fragmentation of prion fibers. Cyr and collaborators have reported that Sis1-binding domain of Rnq1 is located in 82–106 amino acid residues and isolated a Sis1-binding defective mutation, L94A (Douglas et al. 2008). Indeed, L94A protein expressed from strong CUP1 promoter impaired interaction with Sis1 by immunoprecipitation (Douglas et al. 2008), although L94A did not impair the interaction under normal expression level in our procedures (Shibata et al. 2009). Furthermore, Rnq1-L91P that we previously isolated was within the 82–106 region and showed any defect in Sis1 interaction (Shibata et al. 2009), and we isolated numerous [PSI+]-destabilizing mutations out of the 82–106 region in this study. Furthermore, elimination of [PSI+] by depletion of Sis1 is significantly slow because [psi] cells emerge after approximately 30 generations (75 h; we postulated that one generation needs 2.5 h) (Higurashi et al. 2008). In the case of rnq1 mutations, [psi] cells appeared within 24 h. These observations suggest that Sis1 may not be a direct target for the titration by Rnq1 mutant proteins.

Although the mechanism remains to be investigated, it is important to emphasize that all 28 substitutions isolated in this study and in the previous study (Fig. 3B) are mapped to the N-terminal nonprion domain of Rnq1 and, regardless of the unbiased mutagenesis of the entire RNQ1 sequence, no single allele appeared in the C-terminus. These rnq1 substitutions are located on the surface of N-terminal α-helices, α-1 through α-5 (see Fig. 3B), and interestingly, all these mutations except for proline substitutions are localized to the surface of one side (see Fig. 8). Although the function of these α-helices is not known, it is reasonable to speculate that they are involved in a crucial interaction with Rnq1 itself or some other cofactors. The rnq1 mutations isolated in this study will prove useful for elucidating a putative binding partner to the nonprion domain of Rnq1 and the molecular mechanism underlying the prion elimination.

Experimental procedures

Strains

Saccharomyces cerevisiae strains used in this study are as follows: NPK50 ([PSI+] [pin] MATa ade1-14 leu2 ura3 his3 trp1), NPK200 ([psi] [PIN+] isogenic with NPK50), NPK294 ([PSI+] [PIN+] isogenic with NPK50), NPK299 (rnq1::URA3 derivative of NK294), NPK300 ([PIN+] derivative of NPK50), NPK301 ([ure-o] [pin] MATa PD-ADE2 his3 leu2 trp1 kar1 PD-CAN1), NPK302 ([URE3] [pin] isogenic with NPK301), NPK435 ([PIN+] derivative of NPK302), NPK372 ([psi] [pin] hsp104::LEU2 derivative of NPK200) (Brachmann et al. 2005; Kurahashi et al. 2008), NS42 ([PSI+] [PIN+] MATa ade1-14 leu2 ura3 his3 trp1 rnq1::KanMX [pRS416RNQ1p-RNQ1 (ARS/CEN, URA3 marker)]) (Shibata et al. 2009), NPK620 ([PSI+] [PIN+] MATa ade1-14 leu2 ura3 his3 trp1 sup35::SUP35-GFP) and NPK635 ([psi] [PIN+] isogenic with NPK620) (NPK620 and NPK635 are derivative of G74-D694 [psi]), (Kawai-Noma et al. 2009).

Plasmids

Plasmids used are pRS400 series vectors (Stratagene). pRS416RNQ1p-RNQ1 (ARS/CEN, URA3 marker) or pRS415RNQ1p-RNQ1 (ARS/CEN, LEU2 marker) carrying RNQ1 from nucleotide positions −228 (counted from the translation start site) to 1653 were described previously (Shibata et al. 2009). pRS413ADHp-RNQ1s or pRS413GAL1p-RNQ1s carrying the ADH or GAL1 promoter, RNQ1 alleles and the CYC1 terminator was described previously (Kurahashi et al. 2009) or constructed in this study. The GAL1 promoter was PCR-amplified from yeast genetic DNA using a primer pair of P1 (5′-GAGCTCTTGAAGTACGGATTAGAAGCC-3′) and P2 (5′-GGATCCTATAGTTTTTTCTCCTTGACG-3′). The fragment was inserted into the SacI and the BamHI sites in pRS413.

Isolation of rnq1 mutants

The 1.5-kb fragment containing part of the ADH promoter, the coding region of RNQ1 and part of the CYC1 terminator was amplified by error-prone PCR using P3 (5′-TCAAGTATAAATAGACCTGC-3′) and P4 (5′-ATAACGTTCTTAATACTAAC-3′) primers and cotransformed into the NPK294 strain with a linearized, BamHI-XhoI, plasmid pRS413ADHp-RNQ1 (ARS/CEN, HIS3 marker) carrying the ADH promoter and the CYC1 terminator. The mutagenized rnq1 sequences were inserted in vivo into the vector by homologous recombination with the ADH promoter and the CYC1 terminator sequences. The resulting His+ transformants were grown on SC-his plates for 3 days and were replica-plated onto YPD plates to screen for red (i.e. [psi]) colonies. Plasmids were confirmed for reproducible phenotypes and characterized by DNA sequence analysis.

Protein analysis

SDS-polyacrylamide gel electrophoresis (SDS–PAGE) and SDD-AGE were carried out as described previously (Kryndushkin et al. 2003; Kurahashi & Nakamura 2007). The immunoblot experiments were performed using anti-Rnq1 antibody (Kurahashi & Nakamura 2007), anti-Sup35 antibody (Nakayashiki et al. 2001), anti-Hsp104 antibody (Affinity Bioreagents) and anti-Pgk1 antibody (Molecular Probes).

Thermotolerance

Experiments were performed as described previously (Tkach & Glover 2004). Briefly, cells in log phase in YPD at 30 °C were pre-incubated at 37 °C for 60 min to induce Hsp104 with the heat-shock response and then transferred to a 50 °C water bath for 20 min. Aliquots of cells were transferred to ice immediately and survival of cells in these samples was determined by titration on YPD media.

Fluorescence correlation spectroscopy

All of the FCS measurements were taken at 25 °C on LSM510 confocal microscope combined with a ConfoCor 2 (Zeiss), as described in our previous study (Kawai-Noma et al. 2006, 2009; Pack et al. 2006). Briefly, the fluorescence autocorrelation functions (FAF; G (τ)), from which the average diffusion time (τi), which is inversely proportional to the diffusion coefficient and the absolute number of fluorescent proteins in the detection volume are calculated, are obtained as follows:

image(1)

where I (t + τ) is the fluorescence intensity obtained by the single photon counting method in a detection volume at a delay time τ (brackets denote ensemble averages). The curve fitting for the multicomponent model is given by:

image(2)

where yi and τi are the fraction and the diffusion time of the component i, respectively, N is the total number of fluorescent molecules in the detection volume defined by the beam waist w0 and the axial radius z0, and s is the structure parameter representing the ratio of w0 and z0. Structure parameter was determined with standard Rh6G solution. The GFP fluorescence in living cells was excited with a minimal total power for enough signals to noise by adjusting the acousto-optical tunable filter. Five or ten sequential measurements of 10 s were made in a single cell. The effect of photobleaching on FCS analysis was minimized by lowering the excitation intensity and by selecting cells with low fluorescence. The FAFs in Fig. 6 were normalized by the number of particles, N, for the comparison of apparent differences in the diffusion of GFP-fused proteins from cell to cell.

Acknowledgements

We thank R. B. Wickner, H. Taguchi and S. Kawai-Noma for the gift of strains and/or helpful discussion and Colin Crist for critical reading of the manuscript and valuable comments. This work was supported in part by grants from The Ministry of Education, Sports, Culture, Science and Technology of Japan (MEXT).

Ancillary