|Home | About | Journals | Submit | Contact Us | Français|
Prions in the yeast Saccharomyces cerevisiae show a surprising degree of interdependence. Specifically, the rate of appearance of the [PSI+] prion, which is thought to be an important mechanism to respond to changing environmental conditions, is greatly increased by another prion, [RNQ+]. While the domains of the Rnq1 protein important for formation of the [RNQ+] prion have been defined, the specific residues required remain unknown. Furthermore, residues in Rnq1p that mediate the interaction between [PSI+] and [RNQ+] are unknown. To identify residues important for prion protein interactions, we created a mutant library of Rnq1p clones in the context of a chimera that serves as proxy for [RNQ+] aggregates. Several of the mutant Rnq1p proteins showed structural differences in the aggregates they formed, as revealed by SDD-AGE. Additionally, several of the mutants showed a striking defect in the ability to promote [PSI+] induction. These data indicate that the mutants formed strain variants of [RNQ+]. By dissecting the mutations in the isolated clones, we found five single mutations that caused [PSI+] induction defects, S223P, F184S, Q239R, N297S, and Q298R. These are the first specific mutations characterized in Rnq1p that alter [PSI+] induction. Additionally, we have identified a region important for the propagation of certain strain variants of [RNQ+]. Deletion of this region (amino acids 284–317) affected propagation of the high variant but not medium or low [RNQ+] strain variants. Furthermore, when the low [RNQ+] strain variant was propagated by Δ284–317, [PSI+] induction was greatly increased. These data suggest that this region is important in defining the structure of the [RNQ+] strain variants. These data are consistent with a model of [PSI+] induction caused by physical interactions between Rnq1p and Sup35p.
Prions are caused by transmissible protein aggregates. In mammals, the PrP protein forms a prion that is the causative agent of the transmissible spongiform encephalopathies (TSEs) such as scrapie in sheep, bovine spongiform encephalopathy (mad-cow disease) and Kuru in humans.1 Prions also exist in fungal species such as the budding yeast Saccharomyces cerevisiae. First discovered as a system of non-Mendelian phenotypic inheritance, prions in yeast do not cause cellular toxicity, but do share some characteristics with mammalian prions.2–4 Yeast prions provide an excellent model for studying both the intrinsic properties of the proteins that enable prion propagation and the cellular factors that facilitate prion formation and maintenance. In the non-prion state, these proteins are soluble and able to perform their molecular function in the cell. Transition into a prion state causes the proteins to aggregate, which generally causes a loss-of-function phenotype.3 The frequency of the transition to the prion state can be increased by overexpression of the prion-forming protein. Prion protein aggregates are transmitted between cells through the cytoplasm in both mitosis and meiosis. Due to the self-propagating nature of the prion structure, the associated phenotypes are always dominant. Maintenance of the prion phenotype is dependent on the gene encoding the prion protein; deletion of that gene, or its prion forming domain (PFD) abrogates the cognate prion phenotype.
Two of the most well-studied prions in yeast are [PSI+] and [RNQ+].5,6 [PSI+] is caused by aggregation of the yeast eukaryotic release factor (eRF3), Sup35p.7,8 When soluble and functional in [psi−] cells, Sup35p facilitates the faithful release of polypeptide chains from the ribosome at stop codons. In its prion form, Sup35p aggregates and the recognition of stop codons becomes inefficient, thereby resulting in global nonsense suppression. [PSI+] can be monitored using alleles in biosynthetic pathways that contain premature stop codons that are only translated in the [PSI+] state. The [PSI+] state has been proposed to be beneficial due to the read-through of naturally occurring stop codons which enables translation of otherwise silent regions of the genome.9,10 As a result of nonsense suppression, isogenic [PSI+] and [psi−] strains show differences in their ability to survive in various growth conditions.9,11 Thus, the induction and removal (curing) of [PSI+] may be a regulatory mechanism. In addition, the ability to sample new protein products due to the conversion to the [PSI+] state may be adaptive and evolutionarily advantageous.9–12 Interestingly, the de novo appearance of [PSI+] is more efficient in the presence of the [RNQ+] prion, which is caused by the aggregation of a protein of unknown function, Rnq1p.13–15 The reason for this interdependence is not fully understood but may relate to the ability of [PSI+] to serve as a mechanism to react to shifting environmental conditions. Several [RNQ+] yeast strains have been found in nature and these are appear to be competent to form [PSI+],16,17 indicating that [RNQ+] may provide an extra level of control on the induction of [PSI+].
The mechanism by which [RNQ+] promotes [PSI+] induction is not well elucidated. The gain-of-function of Rnq1p aggregates in the [RNQ+] state is an attractive model since the deletion of RNQ1 has no obvious phenotype, nor does it enhance the induction of [PSI+]. Some data suggest that a direct interaction between prion aggregates of Rnq1p and soluble Sup35p facilitates [PSI+] induction. Fluorescently-labeled Sup35p colocalizes with Rnq1p aggregates early in [PSI+] formation but not once [PSI+] is fully established.18 Once formed, the aggregates appear to be largely separate.19 Purified recombinant Rnq1p forms fibers that are able to seed in vitro amyloid formation of Sup35NM, albeit inefficiently.18,20 These data point to a templating model where Rnq1p aggregates transiently interact with monomeric Sup35p to nucleate the formation of Sup35p aggregates and generate [PSI+].14 This is also consistent with in vivo data, as the initiation of [PSI+] is enhanced by the transient overexpression of Sup35p in [RNQ+] cells, but once [PSI+] is formed, RNQ1 can be deleted and [PSI+] continues to propagate.14
The phenotypic effects associated with individual yeast prions show a wide range of severity, often due to a phenomenon referred to as prion strain variants. This is also true in mammalian prions, as strains of the infectious forms of PrP exist and are correlated with differences in age of onset, severity of disease, and pathology.21 Strain variants are structurally distinct conformations of the prion-forming protein, each of which is self-propagating.22–26 Strain variants of [PSI+] manifest as different strengths of nonsense suppression that typically correlate to the amount of soluble Sup35p.27–30 Strain variants of [RNQ+] result in differing propensities to induce [PSI+].13,31 Under semi-denaturing conditions, prion protein aggregates from different [PSI+] and [RNQ+] strain variants show a change in electrophoretic mobility.19,32 Additionally, recent biophysical studies have shown that the stable amyloid core of fibers of the prion-forming domain of Sup35p differs among strain variants produced in vitro.24–26,33,34
The propagation of yeast prions is dependent on the chaperone machinery of the cell. Deletion of the molecular disaggregase Hsp104p eliminates all known yeast prions.15,35–37 Transient inactivation of Hsp104p by growth on media containing millimolar amounts of guanidine hydrochloride (GdHCl) also cures yeast prions.15,37–40 Interestingly, the overexpression of Hsp104p cures [PSI+], but not other yeast prions.15,35,41,42 Prions are also sensitive to changes in the levels of various Hsp104p co-chaperones such as members of the Hsp70 and Hsp40 families.43,44 For example, Sis1p, an Hsp40 family member that interacts with the Hsp70 Ssa1p, is required for the maintenance of [RNQ+].45–48
Prion formation in yeast is driven by PFDs. These domains are typically modular and as such, can transfer the prion-forming ability to unrelated proteins when expressed as fusions.7,15,49 These fusion proteins form aggregates that retain prion character and can be used to study the properties of the original prion-forming proteins. The PFD of prion-forming proteins in S. cerevisiae generally have high glutamine (Q) and asparagine (N) content. These residues may be important for prion formation as a screen for mutant alleles of Sup35p that were unable to propagate [PSI+] found mutations in single Q or N residues that were generally mutated to a charged amino acid such as arginine.50 In contrast, the PFDs of both the mammalian PrP and the Het-s prion protein from the fungal species Podospora anserina are not Q/N-rich. Interestingly, the Q/N-rich region of Sup35p can be replaced with either the PFD of Het-s or an expansion of a set of oligopeptide repeats from PrP.51,52 These fusion proteins have a greatly reduced Q/N character but continue to propagate as a prion in S. cerevisiae. This indicates that high Q/N content is not an absolute requirement for prion propagation in budding yeast.
The proposed PFD of Rnq1p consists of the Q/N-rich C-terminus (amino acids 153–405) that is broken up into a several different motifs with repeating sequences.53,54 These include a set of QG repeats at the start of the prion-forming domain and several QN tracts separated by hydrophobic regions that are predicted to form alpha-helices. Several naturally-occurring Rnq1p alleles have been found with deletions in either the QG region or the QN tracts that may be responsible for a defect in the propagation of [RNQ+].17 A Rnq1p-GFP fusion has been used to determine which domains are important for prion formation by deletion mapping.53 Deletion of the C-terminal 64 amino acids reduces the ability of the fusion protein to propagate the [RNQ+] prion, while deletion of another 100 amino acids completely abrogated the ability of Rnq1p-GFP to maintain [RNQ+]. Additionally, the N-terminus of Rnq1p was also demonstrated to play a role in [RNQ+] propagation, as deletion of the first 133 amino acids reduces the ability of the Rnq1p-GFP fusion to transmit [RNQ+].53 Despite extensive deletion mapping of the Rnq1p PFD, very little is known about internal regions or individual residues that are important for the maintenance of [RNQ+] or the ability of [RNQ+] to promote [PSI+] induction.
Here, we describe mutations within Rnq1p that alter the ability of the [RNQ+] prion to propagate and to induce [PSI+]. Using a chimeric protein reporter, we identified mutants by phenotypic changes. These mutants have subtle effects on [RNQ+] propagation but several reduce [RNQ+]-mediated [PSI+] induction. Additionally, we have found a region of [RNQ+] that is responsible for the propagation of certain strain variants. Together, the data are consistent with a model that posits a direct physical interaction between Rnq1p aggregates and Sup35p that mediates the induction of [PSI+].
In order to better understand the primary structure requirements for [RNQ+] prion maintenance, we set out to find mutants in Rnq1p that were unable to propagate the [RNQ+] prion. Since the function of Rnq1p is unknown, there is no phenotype to use in a screen to identify alleles that do not propagate [RNQ+]. Therefore, we created a fusion protein, called RRP ([RNQ+] reporter protein), that enables the [RNQ+] status of a cell to be monitored phenotypically. RRP is composed of the proposed PFD of Rnq1p (amino acids 153–405) and the M and C domains of Sup35p (amino acids 124–685). The C-terminus of Sup35p is responsible for the translation termination function.55,56 Aggregation of Sup35p or its functional equivalent, RRP, in the prion form presumably reduces its ability to function and creates the nonsense suppression phenotype associated with [PSI+].7,8 In a strain carrying the ade1–14 allele, which contains a premature stop codon, cells cannot complete the adenine biosynthesis pathway due to efficient translation termination. These cells are red due to the accumulation of a biosynthetic intermediate of the adenine biosynthesis pathway.15 Isogenic cells with aggregated Sup35p or RRP, however, are light pink due to the production of Ade1p by nonsense suppression.
We replaced the endogenous SUP35 locus in a [psi−] [RNQ+] strain with RRP. The integrants expressing RRP were light pink (Figure 1A) and this nonsense suppression phenotype was curable by transient growth on GdHCl (data not shown). We performed genetic tests to determine whether the nonsense suppression phenotype was dependent on the presence of wild type Rnq1p and the [RNQ+] prion. First, we generated a diploid that was homozygous for SUP35 replacement by RRP and heterozygous for a deletion of RNQ1. The diploid was light pink and the sporulation of this diploid yielded tetrads with a 2:2 segregation pattern of the light pink to red phenotypes (Figure 1A). Western blot analysis of the progeny showed that the light pink cells expressed wild type Rnq1p while the red cells did not (Figure 1A). The red cells do not maintain the [RNQ+] prion because of the RNQ1 deletion and also did not maintain the RRP-dependent nonsense suppression. Second, deletion of RNQ1 in a light pink haploid strain expressing RRP eliminated the nonsense suppression phenotype, turning the strain red (data not shown). These data suggest that RRP does not maintain the prion phenotype independently of [RNQ+]. Third, integration of RRP into a [rnq−] strain produced only red cells (data not shown). As such, we conclude that the function of RRP in translation termination is indicative of the status of the [RNQ+] prion in the cell, and presumably RRP co-aggregates with Rnq1p in the [RNQ+] state to produce the nonsense suppression phenotype (Figure 1B). A similar Rnq1p-Sup35p fusion construct, has previously been reported15 but it lacks a 6 amino acid sequence in the C-terminus of Rnq1p (NNGNQN) and forms a prion independent of the [RNQ+] status of the cells.
A library of autonomously replicating plasmids containing mutations in the RNQ1 PFD of RRP was created and transformed into a [RNQ+] strain carrying wild type RRP on the chromosome, which is light pink. Transformants were first plated on SD-ura medium to select for the mutant plasmids, then scraped and plated onto rich media (YPD) to assess the color phenotype. The rationale for this screen is as follows: mutants that are unable to efficiently join the existing Rnq1p aggregates remain in a soluble state and are therefore able to terminate translation efficiently (antisuppressors). Such RRP mutants would generate colonies containing red sectors (Figure 1B). Red sectors indicate candidate mutants that are defective in joining the Rnq1p aggregates. The light pink sectors are produced when the strain loses the plasmid but remains [RNQ+]. Approximately 20,000 mutant clones were screened and 25 mutants were found that had a reproducible red sectoring phenotype. The clones were sequenced and these mutants are listed in Table 1. Even though single mutants were present within the library, the vast majority of the library clones with antisuppression phenotypes contained multiple mutations. While this complicated the initial analysis, some striking features of the mutants were readily apparent.
Almost all of the RRP mutant clones contained at least one mutated glutamine or asparagine residue which was usually changed to a charged amino acid such as lysine or arginine. This is consistent with the results of a similar screen performed with Sup35p that was conducted in search of mutants unable to support [PSI+] propagation.50 Two of the mutants containing single point mutations, N307D and Q317R, fit this pattern.
Several non-glutamine or asparagine residues were altered multiple times, for example, S182, F214, and S345 (Figure 1C). Additionally, the other identified single mutation that suppressed the nonsense suppression phenotype was S223P. Several clones had small deletions in the N-terminal QG repeats, likely due to primer slippage. Finally, seventeen of the 33 amino acids between residues 284 and 317 were mutated in at least one clone. This suggests the importance of this region in the formation or maintenance of the [RNQ+] prion.
All of the RRP mutations found were generated using a reporter of the [RNQ+] prion status of the cell. Therefore, we next wanted to test whether the identified mutations had an effect on the propagation of [RNQ+] when cloned into full length Rnq1p. All mutations found in a mutant RRP plasmid were cloned en masse into native RNQ1. During the cloning process, the deletions found in the QG section were removed, with the one exception being Mut31. Wild type RNQ1 in a [RNQ+] strain was replaced by the mutant alleles using a plasmid-shuffle method. Western blot analysis indicated that the steady state levels of the Rnq1p mutants were similar to that of wild type Rnq1p expressed from either the chromosome or a control plasmid (data not shown). The solubility of mutant Rnq1p protein was then assayed by high speed ultracentrifugation to separate the soluble and insoluble material in the lysates. The distribution of the Rnq1p protein was then assessed by western blot. In lysates from [RNQ+] cells, Rnq1p fractionates to the insoluble fraction, but in [rnq−] cell lysates, Rnq1p is found in the soluble fraction. Surprisingly, the mutated Rnq1p fractionated exclusively to the insoluble fraction in all but one of the mutants (Figure 2A and data not shown). Despite a clear effect on the nonsense suppression of RRP, the majority of the mutants appear to be competent to maintain the [RNQ+] prion as the only copy of RNQ1 by this biochemical assay. Only Mut72 showed the presence of soluble Rnq1p though the majority of it appeared to be in the insoluble fraction after the plasmid shuffle immediately followed by six generations of growth in liquid medium (Figure 2B). After 36 generations in liquid medium, almost all of Mut72 had shifted from the insoluble fraction to the soluble fraction, indicating that Mut72 loses its ability to maintain [RNQ+] over time. The insoluble protein from the mutants was not due to non-specific aggregation. After growth of strains expressing mutant Rnq1p on GdHCl, the protein was found to be shifted into the soluble fraction (data not shown). Furthermore, following a plasmid shuffle through a [rnq−] strain, the mutants proteins segregated into the soluble fraction (data not shown).
We tested the ability of the mutants to transmit as a prion. Cells carrying the mutant protein in the insoluble form were mated to cells with wild type Rnq1p in the [rnq−] state. The resulting diploids were dissected and the [RNQ+] status of haploids carrying only wild type RNQ1 was assessed by sedimentation assay. All of the mutants except Mut72 were able to convert wild type Rnq1p into the insoluble [RNQ+] state (data not shown). Mut72 likely failed to maintain a prion through the mating procedure and therefore failed to transmit. These results indicate that the other mutant proteins still propagate as a prion.
We were surprised to find that the all of the mutants but Mut72 failed to show any defect in [RNQ+] prion propagation by the biochemical solubility assay. Therefore, we turned to other methods to assess the structural integrity of [RNQ+] propagated by the mutants. To further examine the structure of the mutant [RNQ+] prion protein aggregates, we performed semi-denaturing detergent agarose gel electrophoresis (SDD-AGE)32 on any mutant that shared a mutation with another mutant found in the screen. This technique resolves differences in high molecular weight protein aggregates and has been used to show differences in prion strain variants of both [RNQ+] and [PSI+].19,32,57 Variants of [RNQ+] can be distinguished by the stability of protein aggregates following incubation in SDS at different temperatures. Most Rnq1 protein aggregates in [RNQ+] strain variants remain stable at moderate temperatures. However, the high [RNQ+] variant previously characterized 31 breaks down into a presumably monomeric form upon exposure to increased temperature.19 Our [RNQ+] plasmid shuffle strains were made in strains that carry this thermolabile high [RNQ+] variant. Therefore, we asked if the aggregated mutants were still susceptible to thermal breakdown. Following incubation in sample buffer at either room temperature or 65°C, we performed the SDD-AGE analysis on the mutants (Figure 3).
Lysate from a [rnq−] strain showed little high molecular weight species, with nearly all of the monomeric Rnq1p located at the bottom of the blot. Lysate from a wild type [RNQ+] strain incubated at room temperature showed most of the protein migrating as a high molecular weight smear and little monomeric protein. Following treatment of the protein lysate at 65°C, however, the aggregates began to break down and a low molecular weight, monomeric Rnq1p band appeared. Many of the mutants tested harbor Rnq1p aggregates that show a pattern of SDS and temperature resistance similar to wild type. However, Rnq1p aggregates of Mut62, Mut83 and Mut84 did not break down upon heat treatment, indicating a more stable aggregate structure. Interestingly, two mutants, Mut27 and Mut72, showed large amounts of monomeric protein even without heat treatment. This is not surprising for Mut72, which is unable to stably maintain the [RNQ+] prion but likely indicates an alternate structure of Mut27 which maintains [RNQ+] mitotically. Other mutants, Mut33, Mut75, and Mut82, showed more subtle changes in the aggregate properties when subjected to heat treatment.
Next, we quantified the differences in temperature sensitivity of the aggregates seen in the SDD-AGE. For the thermosensitive mutant, Mut27, we subjected lysates to various temperatures and analyzed the amount of soluble protein by western blot. Rnq1 protein solubilized by heat treatment enters the gel while aggregated protein does not. Following SDS-PAGE, the protein was transferred to PVDF and probed with anti-Rnq1p antibody (Figure 3B, upper). Aggregates of Mut27 are sensitive to temperatures as low as 55° C while wild type Rnq1p aggregates shows no solubilization until 65° C. The quantification of three replicates is shown (Figure 3B, lower).
For the mutants that were more thermostable than wild type, we treated lysates at 65° C and then centrifuged them through a 20% sucrose solution. The aggregates entered the cushion while the solubilized protein did not. The amount of Rnq1 protein that did not enter the sucrose cushion was analyzed by western blot (Figure 3C). Both Mut83 and Mut84 showed significantly less soluble protein than wild type in this assay. These results, coupled with the SDD-AGE, establish that at least some of the mutants have different physical properties than wild type protein when in the prion state. These data suggest that the mutations are changing the structure of the aggregates, at least as long as the mutations themselves are present.
As some of the mutants had apparently altered prion structures, we assessed whether those structures influenced the interaction between [RNQ+] and [PSI+]. The presence of the [RNQ+] prion has been shown to greatly enhance the appearance of the [PSI+] when Sup35p is overexpressed.13,14,41 Moreover, strain variants of [RNQ+], which presumably propagate different prion structures, permit different efficiencies of [PSI+] induction.31 We wanted to determine if the [RNQ+] prions propagated by the mutant Rnq1 proteins were also capable of stimulating the induction of [PSI+]. Red [psi−] [RNQ+] cells expressing the Rnq1p mutants as the only copy of Rnq1p were transformed with a plasmid that expresses Sup35p. Transformants were grown overnight in selective medium and plated on YPD. [PSI+] induction was scored by counting colonies with light pink sectors and dividing that by total number of colonies. Fifteen of the 25 mutant clones showed a significant decrease in [PSI+] induction as compared to wild type Rnq1p in [RNQ+] cells (Figure 4 and Table 1). These same mutants showed a defect in [PSI+] induction when expressed in a strain harboring the medium [RNQ+] variant as well, indicating the defect caused by the mutants was not specific to the high [RNQ+] variant (data not shown).
The presence of multiple mutations within most of the original [RNQ+] mutants made the interpretation of their individual effects challenging. Therefore, we cloned single mutations from the mutants into wild type Rnq1p. We chose to assess the individual mutations from Mut72 due to its inability to propagate the [RNQ+] prion. Further, we assessed all mutations that appeared in more than one RRP plasmid isolated from the screen by creating single mutants in RNQ1 (Figure 1C and Table 1). When the same residue was mutated to multiple amino acids (as in the case of S345G and S345R), only one of the mutations was tested. Plasmids carrying RNQ1 with each single mutation were then used to replace wild type RNQ1 in a [RNQ+] strain by plasmid shuffle. We then tested the [PSI+] induction capability of the single mutants to identify which residues were responsible for the defects in [PSI+] induction. The majority of single mutations tested showed no significant defect in [PSI+] induction (data not shown). However, F184S, Q239R, N297S, and Q298R showed strong reductions in their ability to induce [PSI+], indicating that they may be primarily responsible for the defects seen in the originally isolated mutants (Figure 5A).
Additionally, we tested the solubility of the single mutant Rnq1 proteins. The single mutants that had defects in [PSI+] induction all partitioned into the insoluble fraction (Figure 5B). Additionally, none of the other single mutations altered the solubility of Rnq1p in the [RNQ+] strains as they separated into the pellet fraction (data not shown). None of the single mutants from Mut72 recapitulated the instability of the mutant prion as all remained in the insoluble fraction (data not shown). The insoluble Rnq1p single mutants were curable by GdHCl treatment and did not aggregate when shuffled through a [rnq−] strain (data not shown).
We also analyzed the structural properties of the four single mutants with [PSI+] induction defects by SDD-AGE. All of the aggregates were of similar size to wild type Rnq1p in the [RNQ+] state (Figure 5C). The single mutants also did not show a difference from wild type in SDS resistance at moderate temperatures as they were all partially broken down at 65°C but not at room temperature. These results indicate that the single mutants alone are not responsible for the alterations in thermal sensitivity seen when multiple mutations are present.
Next, we wanted to know if the single mutants that showed [PSI+] induction defects were able to pass those defects onto wild type Rnq1p. To answer this question, we mated the strains containing RRP mutants in the [RNQ+] state to a [rnq−] strain expressing wild type Rnq1p. The resulting diploids were sporulated and the haploid progeny showed a 2:2 inheritance of wild type RNQ1 to the RNQ1 deletion. Progeny with wild type RNQ1 that did not maintain the mutant Rnq1p plasmid were assayed for [RNQ+] by Rnq1p solubility. In all cases the single mutants converted wild-type, soluble Rnq1p into its insoluble [RNQ+] form (data not shown), indicating that they maintain and transmit a prion. Additionally, for all mutants tested (all the multiple mutants, F184S, Q239R, and Q298R), [PSI+] induction from the converted wild type [RNQ+] was equivalent to the wild type parent (data not shown). This indicates that the mutants do not irrevocably alter the structure of the [RNQ+] prion nor create a new stably propagating strain variant. The high Q/N content of the Rnq1 protein and the large putative PFD, together with our data that demonstrate that the majority of single mutants tested do not affect [RNQ+], suggest that the propagation of the [RNQ+] prion is difficult to disrupt.
Seventeen of the 33 amino acids between positions 284 and 314, which we refer to as a mutational hotspot, were found to be mutated in one or more of the recovered RRP mutants (Figure 6A). We generated a deletion of this entire region to create an allele called rnq1Δhot. We then replaced the wild type copy of RNQ1 in various [RNQ+] strain variants with the rnq1Δhot allele by plasmid shuffle. We next tested [PSI+] induction with the rnq1Δhot allele in the different [RNQ+] strain variant backgrounds. The deletion showed a markedly decreased ability to induce [PSI+] in the high [RNQ+] variant (Figure 6B). However, no such defect was observed in either the medium [RNQ+] or the low [RNQ+] variant (Figure 6B). Strikingly, the low variant seemed to have increased [PSI+] induction with the rnq1Δhot allele.
We also examined the solubility of the deletion proteins in lysates from the [RNQ+] strains. The Rnq1p encoded by the rnq1Δhotallele remained insoluble in the low and medium [RNQ+] strain variants (Figure 6C) and this property was reversible by GdHCl treatment (data not shown). Interestingly, a large amount of soluble protein was seen in the high [RNQ+] strain variant expressing rnq1Δhot (Figure 6C). This indicates that amino acids 284–317 are important for propagation of the high [RNQ+] variant and for the interaction of high and low [RNQ+] variants with Sup35p, but not for the propagation of the medium [RNQ+] strain variant.
Given the strain-specific alterations engendered by the rnq1Δhot allele, we wanted to ask if there were changes in the structure of the mutant prions by assessing the SDS and thermal sensitivity of the aggregates. We performed SDD-AGE on the [RNQ+] variants carrying this allele and compared them to wild type variant aggregates (Figure 6D). The aggregates in the strains carrying the rnq1Δhot allele showed a broader range of aggregate sizes than the wild type Rnq1p aggregates in the corresponding strain variants. Additionally, the monomeric form of the Rnq1Δhot protein migrates as a doublet, which may be a consequence of an interaction with another protein, an SDS-stable oligomer, or a degradation product. Finally, the Rnq1Δhot protein aggregates showed an increased sensitivity to temperature relative to the corresponding wild type protein. These data strongly point to a change in the structure of the prion protein aggregates propagated by the rnq1Δhot allele.
We used a chimeric proxy for the [RNQ+] status of the cell to identify mutants with defects in [RNQ+] propagation. While the mutants clearly affected the nonsense suppression caused by the chimeric RRP, they failed to dramatically affect the solubility of Rnq1p when moved into the native protein. One possible reason for this seeming discrepancy is the addition of the N-terminus. It is clear that while the C-terminus of Rnq1p is sufficient for joining or forming a prion aggregate,15 the N-terminus contributes to propagation of [RNQ+]. Deletions within the N-terminus of Rnq1p cause [RNQ+] to be transmitted inefficiently.53 Furthermore, expression of Rnq1p with a deletion of the first 100 amino acids is able to cure [PSI+] in a [RNQ+] background, indicating that the amino terminus may play a role in prion-prion interactions.54 Finally, the amino terminus of Rnq1p has been shown to have a binding site for the chaperone Sis1p.58 Sis1p is essential for the propagation of [RNQ+] and mutations in the Sis1p binding site of Rnq1p alter the ability of a Rnq1p-GFP fusion to join pre-existing aggregates.46,47,58 In an attempt to ask whether a reduction in Sis1p binding to Rnq1p would alter the strain variant or [PSI+] induction, we generated Rnq1-L94A that was recently shown to drastically reduce Sis1p-Rnq1p interaction.58 Unfortunately, the Rnq1p-L94A mutant was unable to maintain the [RNQ+] prion state (unpublished data J.P.B. and H.T.). This suggests that drastic changes in Sis1p-Rnq1p interaction are not occurring in our Rnq1p mutants that maintain the prion state. Finally, our data also suggest that the N-terminus is able to compensate for mutations in the C-terminus that reduce or inhibit interactions with prion aggregates.
Several of the Rnq1p mutants altered the efficiency of [PSI+] induction caused by the overexpression of Sup35p. A change in [RNQ+]-mediated [PSI+] induction efficiency is generally thought to be associated with changes in the structure of the [RNQ+] prion. We examined the aggregates biochemically by analyzing their sensitivity to temperature and SDS. Several of the mutants showed an increased resistance to heat treatment, while others showed increased sensitivity. The results of the SDD-AGE show that at least some of the mutants are forming [RNQ+] structures that are different than wild type. However, they are unable to pass any structural differences onto wild type Rnq1p during the conversion of soluble wild type Rnq1p into an insoluble prion state, as the wild type protein propagates the original strain variant (Figure 5D and data not shown). This indicates that either the structural information that specifies strain variants is not altered in the mutant conformations despite alterations in the overall structure, or that wild-type protein cannot adopt the structures specified by the mutant proteins.
Previously, the primary sequence domains of the Rnq1p PRD and their role in [RNQ+] prion maintenance have been defined using deletion analysis.15,53,54 However, these deletion analyses provided no clues as to the specific residues required for [RNQ+] propagation or for its ability to regulate [PSI+] induction. We have defined five specific mutations that decrease the ability of [RNQ+] to induce [PSI+] but do not appear to significantly affect [RNQ+] propagation. The substitution of proline for a serine residue (S223P) likely induces large structural changes within the protein.
The other four mutations may shed more light on the interaction between Sup35p and Rnq1p and the induction of [PSI+] by [RNQ+]. Three of these mutations are in glutamine or asparagine residues (Q239R, Q298R, and N297S). Glutamine and asparagine have been shown to be important for the propagation of other yeast prions, as mutants in these residues of Sup35p can abrogate its ability to form [PSI+].50 Additionally, the crystal structure of an amyloid-forming peptide of Sup35p shows the side chains of the glutamine and asparagine residues forming hydrogen bonds between adjacent members of individual beta sheets.59 These hydrogen bonds are important for the positioning of residues within the sheets. The final mutation that affected [PSI+] propagation was in an aromatic residue, phenylalanine (F184S). In the same crystal structure, aromatic tyrosine residues showed stacking of their side chains.59 Given that both glutamine and tyrosine are important as structural components of some amyloids, the fact that mutation of these residues alters [PSI+] induction implies that the heterologous interactions between prion proteins result from a similar set of amino acid interactions. This is consistent with direct templating model of [RNQ+]-mediated [PSI+] induction.
The deletion of amino acids 284–317 revealed a region of Rnq1p that is required for maintenance of the high and low [RNQ+] strain variants but not for the medium strain variant. Interestingly, replacement of a shorter stretch of amino acids within this region, 292–298, with alanines had no effect on [RNQ+] propagation (data not shown). Deletions within the hotspot region have previously been shown to exist as naturally occurring polymorphisms of Rnq1p.17 While strains carrying deletions within this region of the protein were shown to be [rnq−], their ability to support [RNQ+] upon mating to a [RNQ+] strain was not tested. Previously, both large domains and specific residues of Sup35p that are required for specific strain variants of [PSI+] have been defined.34,60–63 This represents the first such discovery for the [RNQ+] prion. Additionally, [RNQ+] propagated by the Rnq1 Δhot protein showed similar levels of [PSI+] induction when derived from either medium or low [RNQ+], which may indicate that deletion of this region locks the mutant protein into a specific conformation when templated by those structures.
Here, we have defined residues in Rnq1p that are important for the induction of [PSI+]. These data are consistent with a direct physical interaction between Sup35p and [RNQ+] prion aggregates. As the formation of [PSI+] is much less efficient than the templating of additional Rnq1p molecules onto existing [RNQ+] aggregates, the interaction between Rnq1p in the [RNQ+] state and monomeric Sup35p may be much more sensitive to the mutations. This could explain the relatively modest effects of the mutants on [RNQ+] propagation. Using the relative sensitivity of [PSI+] induction, it may now be possible to find compensatory mutations within Sup35p that overcome the [PSI+] induction defects of the mutant [RNQ+] prions, thereby providing stronger evidence for a direct physical interaction. Understanding which residues are important for this interaction is an important step in understanding how [RNQ+] controls the induction of [PSI+] and how this translates to the regulation of prion formation by other epigenetic elements.
All S. cerevisiae strains were derived from 74-D694. Low, medium, and high [RNQ+] variants were kindly provided by Dr. Susan Liebman. Standard culturing and manipulation techniques were used for both yeast and Escherichia coli64 YPD is a rich medium while SD is a synthetic medium usually lacking one or more amino acids to select for appropriate plasmids.
For cloning, the pRS series of vectors were used to express or integrate constructs into yeast.65,66 RRP as cloned into the yeast integrating vector pRS306 in three parts. The promoter of SUP35 was amplified with the following oligonucleotides 5′-CGGAATTCCTCGAGAAGATATCCATC-3′ and 5′-CCCGGATCCTGTTGCTAGTGGGCAGATATAG-3′, digested with EcoRI and BamHI and ligated into an EcoRI/BamHI fragment of pRS306. The MC domains of SUP35 (coding for amino acids 124–685) were amplified with the oligonucleotides 5′-GGGCCGCGGATGTCTTTGAACGACTTTCAAAAGC-3′ and 5′-GGGGAGCTCGTGATTGAAGGAGTTGAAACCTTGC-3′, digested with SacI and SacII and ligated into the promoter construct cut with the same enzymes. Finally, the C-terminus of RNQ1 (encoding the PFD domain- amino acids 153–405) was amplified with the following oligonucleotides 5′-CCCGGATCCATGCAAGGTCAGGGACAAGGTCAAGG-3′ and 5′-GGGCCGCGGGTAGCGGTTCTGGTTGCCGTTATTG-3′, cut with BamHI and SacII and ligated into a BamHI/SacII fragment of the construct carrying the promoter and SUP35 MC domain and promoter to generate pRS306-RRP.
A ScaI fragment containing the Cen/Ars4 sequence from pRS316 was cloned into plasmid pRS306-RRP to make a plasmid that ectopically expressed RMC. Libraries were made by 15 cycles of PCR using Taq polymerase with the oligos to amplify the PFD of RNQ1 listed above. PCR reactions contained either 125 μM or 40 μM MnCl2, but otherwise were done using the manufacturer’s conditions. To maintain complexity, each library consisted of 3 pools made from two different PCR reactions. Each pool contained approximately 4,000 clones (12,000 clones per library).
A XhoI/SpeI fragment containing RNQ1 from pRS315-RNQ167 was ligated into pRS313 or pRS316 to make pRS313-RNQ1 or pRS316-RNQ1 respectively. Mutants found in RRP were transferred en masse by the following procedure. The N-terminus of RNQ1 was amplified using the oligonucleotides 5′-GTCACGACGTTGTAAAACG-3′ and 5′-GCCAAAGACGCCAAAGCAGTAAAAGA-3′ with pRS313-RNQ1 as a template. The C-terminus of RNQ was amplified from the corresponding RRP mutant using the oligonucleotides 5′-TCTTTTACTGCTTTGGCGTCTTTGGC-3′ and 5′-CCCTACGTAAACAAAGGATAGAAGGCGAACTGAATCATCGTTCAGTAGCGGT TCTGGTTGCCGTTATTG-3′. The products of these reactions were used as a template for a third PCR reaction using the outer oligonucleotides. This product contained full length RNQ1 with the desired mutations and was digested with SpeI and SnaBI and ligated into a SpeI/SnaBI fragment of pRS313-RNQ1. Due to a random PCR-generated error, all RRP and RNQ construct from the original screen encoded a Y277C mutant in Rnq1p. Rnq1p containing this mutation behaved identically to WT in all in vivo assays (data not shown).
Single mutants were also cloned with bridge PCR. The N-terminus of RNQ1 was amplified with the oligonucleotide 5′-GGGGATATCATGGATACGGATAAGTTAATCTCAGAGG-3′ and an oligonucleotide specific to the mutant. The reverse complement of the mutant specific oligonucleotide was then used along with the oligonucleotide 5′-CCCGTCGACTCAGTAGCGGTTCTGGTTGCCG-3′ to amplify the C-terminus of RNQ1. The oligonucleotide specific to the mutant contained the desired mutation. The pieces were then used as templates for a PCR reaction with the non-mutant specific oligonucleotides used to generate each fragment. This produced full length RNQ1 carrying the desired mutation which was digested with EcoRV and SalI and ligated into pRS313 that contained the RNQ1 promoter on an EcoRV/EcoRI fragment and the RNQ1 terminator on a SalI/XhoI fragment.
Strains carrying RRP integrated at the SUP35 locus were generated by the pop-in pop-out method. Plasmid pRS306-RRP was linearized with MluI and transformed into 74-D694 strains carrying either the medium or high variant of [RNQ+]. Transformants were plated on SD-ura to select integrants. Integrants were then grown overnight in liquid YPD and loop-outs were selected by growth on plates containing 1 mg/ml 5-fluoroorotic acid (5-FOA). RRP expression was checked by western blotting with anti-Rnq1p antibody. Additionally, cells that carried the RRP allele turned pink on YPD. This process generated strains 74D-694-RRP(high) and 74D-694-RRP(med) respectively.
The RNQ1 plasmid shuffle strains were created by deleting RNQ1 using a kanMX4 cassette flanked by RNQ1 sequence. The cassette was generated as described previously68 using pFA6a-kanMX4 and 5′-GAACGTACATATAGCGATACAAACGTATAGCAAAGATCTGAAATGTCGTACG CTGCAGGTCGAC -3′ and 5′-CAAATACGTAAACAAAGGATAGAAGGCGAACTGAATCATCGTTCAATCGATG AATTCGAGCTCG-3′ as primers. The product was transformed into 74-D694 strain and deletions were selected on YPD with 200 μg/μl G418. Deletion of RNQ1 was confirmed by western blot. To make this deletion strain [RNQ+], it was first transformed with pRS316-RNQ1. This strain was then mated to 74-D694 cells of the opposite mating type carrying either high, medium or low [RNQ+].31 The resulting diploid was then sporulated and G418 resistant, URA+ haploid colonies were selected. The [RNQ+] status of these strains was then checked by sedimentation assay.
Libraries with mutagenized RRP were transformed into strain 74D-694-RRP(high) and plated on SD-ura. The colonies were scraped off of the SD-ura plates, pooled and replated on YPD. Disruption of the nonsense suppression phenotype was assayed by red/white sectoring on YPD. Sectoring colonies were spotted onto SD-ura to confirm the presence of the plasmid. Spots on SD-ura were spotted back onto SD-ura, SD-ura-ade and 5-FOA. Plasmids were recovered from isolates that were unable to grow on SD-ura-ade and continued to sector on YPD. These plasmids were transformed back into 74D-694-RRP(high [RNQ+] variant) to confirm the sectoring phenotype was linked to the plasmid and then sequenced.
Yeast cells were lysed in buffer containing 100mM Tris-HCl pH 7, 200 mM NaCl, 1 mM ethylendiaminetetraacetic acid (EDTA), 5% glycerol, 0.5 mM dithiothreitol (DTT), 50 mM N-ethylmalemide (NEM), 3 mM phenylmethanesulphonylfluoride (PMSF) and complete Mini protease inhibitor cocktail (Roche). Lysis was performed by vortexing with glass beads. Following lysis, an equal volume of RIPA buffer (50 mM Tris-HCl pH 7, 200 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS) was added to the lysate and cell debris were cleared by a brief centrifugation. This cleared lysate is the total protein. Insoluble protein was pelleted by centrifugation at 80,000 RPM in a Beckman TLA-100 rotor for 30 minutes. The supernatant containing the soluble protein was removed and the pellet was resuspended in a 1:1 mix of lysis buffer and RIPA buffer. Total, supernatant and pellet fractions were subjected to SDS-PAGE, transferred to PVDF membranes and probed with an anti-Rnq1p antibody.
SDD-AGE was performed as previously described47. Briefly, lysates were prepared by vortexing with glass beads in 25 mM Tris pH 7.5, 100 mM NaCl, 10 mM MgCl2, 1 mM EDTA, complete Mini protease inhibitor cocktail, 3 mM PMSF, 0.5 mM DTT, 5 μg/mL pepstatin 50 mM NEM. Lysates were cleared by brief centrifugation. Approximately 40 μg of protein was incubated at either room temperature or 65°C in sample buffer containg 60 mM Tris-HCl, 5% glycerol, 2% SDS for seven minutes and run on a 1.5% Tris-glycine agarose gel. Proteins were transferred to PVDF membranes and probed with an anti-Rnq1p antibody.
The sensitivity of prion aggregates to temperature was determined in two ways. Lysates were incubated in SDS-PAGE sample buffer (50 mM Tris-Cl pH 6.8, 2% SDS, 10% glycerol, 100 mM DTT) at various temperatures for seven minutes. The lysate was subjected to SDS-PAGE. Solubilized protein entered the gel while aggregates were not able to enter the gel. The protein was separated by SDS-PAGE, transferred onto PVDF, and probed with an anti-Rnq1p antibody. Quantification of Rnq1p protein was performed using ImageJ and values were normalized to the amount of protein at the highest temperature (95° C). Alternatively, lysates were incubated at 65° C in SDD-AGE sample buffer and then layered onto a 20% (w/v) sucrose cushion. This was subjected to ultracentrifugation at 80,000 RPM in a Beckman TLA-100 rotor for 30 minutes. Soluble protein did not enter the sucrose cushion. The top layer, containing protein that did not enter the sucrose cushion was removed, subjected to SDS-PAGE, transferred to PVDF membranes and probed with either an anti-Rnq1p or anti-Pgk1p antibody. Rnq1p levels were normalized to Pgk1p levels and quantified using ImageJ.
Plasmid shuffled strains containing the Rnq1p mutants were transformed with pSUP256 and plated on selective media for both plasmids. Individual transformants were grown in selective media to OD600 ~ 1.6 and plated on YPD. After five days of growth, [PSI+] cells were counted as any cell with a light pink sector. Representative colonies were checked for curing on plates containing 3 mM GdHCl. The vast majority (>95%) of colonies with light pink sectors were curable on 3 mM GdHCl (data not shown). Overexpression of Sup35p in a [RNQ+] strain has also been shown to create non-heritable amyloids of Sup35p that cause nonsense suppression similar to [PSI+].69 Since this nonsense suppression is dependent on the overexpression of Sup35p, we selected cells with light pink sectors and spotted them onto medium containing 5-FOA. Cells that converted to [PSI+] remained light pink on 5-FOA medium, while cells with non-heritable amyloids reverted back to red. Experiments with both wild type Rnq1p and the 15 mutants revealed that on average, about 12% of the light pink sectors were the result of non-heritable amyloids while the rest were bona fide [PSI+] (data not shown). This proportion is in line with the frequency of non-heritable amyloid previously reported.69
The authors would like to thank J.R. Fisher for technical support and members of the True laboratory and Dr. Jeff Moore for helpful discussions and critical reading of this manuscript. We thank Dr. Susan Liebman for providing strains and Dr. Susan Lindquist for providing the anti-Rnq1p antibody. This research was supported by National Institutes of Health grant GM072228 (H.L.T.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.