A Yeast Prion, Mod5, Promotes Acquired Drug Resistance and Cell Survival Under Environmental Stress

See allHide authors and affiliations

Science  20 Apr 2012:
Vol. 336, Issue 6079, pp. 355-359
DOI: 10.1126/science.1219491


Prion conversion from a soluble protein to an aggregated state may be involved in the cellular adaptation of yeast to the environment. However, it remains unclear whether and how cells actively use prion conversion to acquire a fitness advantage in response to environmental stress. We identified Mod5, a yeast transfer RNA isopentenyltransferase lacking glutamine/asparagine-rich domains, as a yeast prion protein and found that its prion conversion in yeast regulated the sterol biosynthetic pathway for acquired cellular resistance against antifungal agents. Furthermore, selective pressure by antifungal drugs on yeast facilitated the de novo appearance of Mod5 prion states for cell survival. Thus, phenotypic changes caused by active prion conversion under environmental selection may contribute to cellular adaptation in living organisms.

Prion phenomena have been observed in yeast and filamentous fungi (1, 2), and fungal prion proteins share common characteristics with mammalian prion protein. Prion inheritance is caused by the propagation of self-perpetuating and infectious prion particles composed of β sheet–rich fibrillar aggregates called amyloid (35). All of the yeast prion proteins identified thus far contain aggregation-prone Gln/Asn-rich domains that are critical for the formation of self-propagating amyloid. A number of Gln/Asn-rich proteins in yeast have the potential to behave as prions (6), implying that yeast might use prion conversion to regulate some cellular functions in vivo. Prion states acquire previously unrecognized genetic traits (7, 8) and affect cellular functions such as transcriptional regulation (3, 9), though they may represent disease states (4). Induction of the prion state [PSI+] resulting from aggregation of Sup35 may be linked to a survival advantage under the selective pressure of environmental stressors (10), suggesting that prion conversion might help an organism adapt to environmental stress (11). However, our understanding of whether and how prion conversion responds to environmental stress for cell survival and if specific mechanisms exist that mediate such adaptive processes is limited.

To address these questions, we attempted to identify yeast prions. We performed a genome-wide screen for PIN (inducible to [PSI+]) factors whose aggregation facilitates the de novo appearance of [PSI+] (12). Among the known PIN factors (12), several proteins behave as yeast prions (1316). We used a [q] yeast strain, a nonprion form of yeast expressing a Q62-Sup35 chimera in which a Gln/Asn-rich domain (residues 1 to 40) in Sup35 was replaced with 62 glutamine repeats (Q62) (17), and searched for QIN (inducible to the [Q+] prion state) factors (18). Because expanded polyglutamine readily forms amyloids, we reasoned that both Gln/Asn-rich and non–Gln/Asn-rich proteins might represent QIN factors. Among the QIN factors we identified (fig. S1 and table S1), we focused on Mod5, a tRNA isopentenyltransferase that catalyzes the transfer of an isopentenyl group to A37 in the anticodon loop (19), because Mod5 did not contain Gln/Asn-rich or repeat domains but acted as both a QIN and PIN factor (fig. S1).

Prion aggregates demonstrate infectivity through their self-propagating amyloid forms (20, 21). We first investigated whether Mod5 forms amyloid fibers in vitro. By multiple criteria, Mod5 aggregates formed amyloid-like fibrillar aggregates (Fig. 1, A to D, and fig. S2). Because Mod5 does not contain Gln/Asn-rich domains, we searched for a core aggregation region in Mod5 amyloids. Limited proteolysis of Mod5 amyloids with proteinase K and following mass analysis allowed us to identify the core of Mod5 amyloids (Fig. 1E), which was predicted to be aggregation-prone by the TANGO algorithm (fig. S3) (22). Soluble Mod5 was fully digested under the same conditions. Furthermore, deletion of the amyloid core region abolished the ability of Mod5 to act as a PIN factor (fig. S1C) and greatly decreased the reactivity of Mod5 amyloids to thioflavin T (Fig. 1B). The aggregates of Mod5 were self-propagating, as the addition of preformed Mod5 fibers to soluble Mod5 substantially accelerated the aggregation of soluble Mod5 (Fig. 1F). Furthermore, Mod5 amyloid seeds facilitated polymerization of soluble Sup35NM in vitro (Fig. 1G), indicating cross-seeding between Mod5 and Sup35NM. This result is consistent with the ability of Mod5 as a PIN factor (fig. S1) and sequestration of intrinsically disordered proteins into cross–β sheets (23). Thus, Mod5 forms self-propagating amyloids in vitro, despite the lack of Gln/Asn-rich domains.

Fig. 1

Formation and structural analysis of Mod5 amyloid in vitro. (A) Transmission electron microscopy images of typical Mod5 amyloid. Scale bar, 100 nm. (B) Thioflavin T (ThT) fluorescence of Mod5, huntingtin-exon1 Q62 (Q62), Mod5Δcore (Δ199-207) amyloids, or soluble Mod5. (C) Binding of congo red to buffer alone, huntingtin-exon1 Q62 (Q62) amyloid, Mod5 amyloid, or soluble Mod5. (D) Circular dichroism spectra of soluble Mod5 (dark gray) and amyloid (light gray). (E) Mass spectrum of the Mod5 amyloid digested by proteinase K. Peptide regions corresponding to mass peaks are shown (arrows). (Inset) A full spectrum and no obvious peak was observed at >3400 mass/charge ratio (m/z). (F) Aggregation of Mod5 in the absence (black) or presence of Mod5 amyloid seeds [10% (light gray), 20% (dark gray) (mol/mol)] was monitored at 25°C. (G) Mod5 amyloid cross-seeds Sup35NM. Amyloid formation of Sup35NM in the absence (black) or presence of Mod5 amyloid seeds [10% (light gray), 20% (dark gray) (mol/mol)] was monitored by ThT fluorescence at 37°C. Error bars denote SD. *P < 0.01, based on an independent t test (n = 3 experiments) in comparison with soluble Mod5.

Next, we employed a range of assays based on color phenotypes using the [PSI+] system (18) and found that Mod5 has the potential to undergo a heritable conformational switch to the prion state by forming aggregates (fig. S4). Thus, we investigated whether a prion state could be induced by the aggregation of endogenous Mod5. Because a double knockout of Mod5 and Trm1, which encodes tRNA methyltransferase, shows sensitivity to 5-fluorouracil (5-FU) (24), we used Δtrm1 strains throughout this study unless otherwise indicated and introduced pure Mod5 amyloids (fig. S5) into Δtrm1 strains by a protein infection protocol (25). We used a Δtrm1 diploid strain with homozygous deletion of TRM1 to avoid the accidental isolation of recessive chromosomal mutants. We examined 480 infectants for their sensitivity to 5-FU, and 10 colonies showed such sensitivity. The sensitivity of 6 out of the 10 colonies was reversed by the transient treatment with 3 mM guanidine hydrochloride (GdnHCl) (Fig. 2A), an inhibitor of the Hsp104 chaperone (26). Disruption of the HSP104 gene also reversed it (Fig. 2B). This phenotypic reversal by elimination of Hsp104 is a common characteristic of yeast prions (9); we hereafter refer to this state as [MOD+]. Overexpression of Hsp104 in [MOD+] yeast also partially restored its sensitivity to 5-FU (fig. S6). These results established that the [MOD+] prion state propagated in an Hsp104-dependent manner. In addition, the [MOD+] state was mitotically stable because [MOD+] yeast showed sensitivity to 5-FU after many passages. Next, we investigated whether endogenous Mod5 undergoes conformational changes in [MOD+] yeast. We prepared cell lysates and separated them into supernatant and pellet fractions by centrifugation (27). In contrast to soluble Mod5 in [mod] and [MOD+]Δhsp104 yeast, Mod5 in [MOD+] yeast was observed in the pellet fraction (Fig. 2C). We examined the cellular localization of Mod5–green fluorescent protein (GFP) with or without mild overexpression of Mod5-GFP (18). In both cases, [mod] and [MOD+hsp104 cells displayed diffusible Mod5-GFP throughout the cytoplasm, whereas [MOD+] cells exhibited multicytoplasmic Mod5-GFP aggregates that were not colocalized to either mitochondria or nucleus (Fig. 2D and fig. S7). Like other yeast prions, Mod5 aggregates from [MOD+] yeast were resistant to SDS (Fig. 2E) (28). Thus, Mod5 is in an altered, aggregated conformational state in [MOD+] yeast, compared with the soluble and diffusible Mod5 in [mod] yeast. The ectopic expression of Mod5 in Δmod5 but not [MOD+] yeast restored the sensitivity to 5-FU, probably because ectopically expressed Mod5 was sequestered into preexisting Mod5 aggregates (Fig. 2F). [MOD+] cells expressing ectopic Trm1 could grow on 5-FU plates (Fig. 2F), indicating specific recruitment of Mod5 monomer into Mod5 aggregates. Thus, the [MOD+] state is caused by self-propagating Mod5 aggregates in vivo.

Fig. 2

Isolation and characterization of [MOD+] prion states caused by aggregation of Mod5. (A) GdnHCl-reversible sensitivity of [MOD+] to 5-FU. Cultures of [mod], Δmod5, and [MOD+] diploids before and after the transient treatment with GdnHCl (3 mM) were spotted on yeast extract, peptone, and dextrose (YPD) in the absence (left) or presence (right) of 5-FU (15 μg/ml). (B) Elimination of Hsp104 restored the sensitivity of [MOD+] yeast to 5-FU. Cultures of [mod ], [MOD+], and [MOD+hsp104 diploids were spotted on YPD in the absence (left) or presence (right) of 5-FU (15 μg/ml). (C) Sedimentation analysis of [mod], [MOD+], and [MOD+hsp104 yeast. Lysates of the yeast strains were separated into supernatant (Sup) and pellet fractions, and endogenous Mod5-GFP was detected by immunoblotting with an antibody to GFP. (D) Localization of Mod5 in [mod], [MOD+], and [MOD+hsp104 yeast cells. Fluorescence images of [mod] (left), [MOD+] (center), and [MOD+hsp104 (right) cells mildly overexpressing Mod5-GFP are shown. Fluorescent foci appeared in the cytoplasm of 55% of [MOD+] cells, 12% of [mod] cells, and 9% of [MOD+hsp104 cells (n > 100 cells). Arrows and arrowheads show Mod5 aggregates and nuclei, respectively. Similar results were obtained in the yeast strains without overexpression of Mod5-GFP [68% of [MOD+] cells, 12% of [mod] cells and 10% of [MOD+hsp104 cells (n > 100 cells)] (fig. S7). Scale bar, 5 μm. (E) Detection of SDS-resistant Mod5 aggregates by semi-denaturating detergent agarose gel electrophoresis in the lysates of [mod] and [MOD+] cells that overexpress Mod5-GFP. (F) Ectopic overexpression of Mod5 in [MOD+] yeast did not restore the sensitivity to 5-FU, whereas that of Trm1 recovered it. Fivefold serial dilutions of yeast cells were spotted in the sensitivity assay.

We investigated dominant inheritance of [MOD+] yeast upon mating. To isolate [MOD+] haploids, we introduced lysates of [MOD+] diploids into [mod] haploid cells by protein infection and assayed infectants for both their sensitivity to 5-FU and phenotypic reversion by GdnHCl (Fig. 3A). Five colonies out of 360 infectants were sensitive to 5-FU in a GdnHCl-reversible manner and were mitotically stable, indicating that they represent [MOD+] states, whereas such colonies were not isolated from the same number of cells infected by [mod] yeast lysates (Fig. 3A and table S2). A [MOD+] haploid was crossed with a [mod] haploid, and the resulting diploid showed sensitivity to 5-FU (Fig. 3B), indicating that the [MOD+] state is dominantly inherited. Next, we disrupted the MOD5 gene in a [MOD+] haploid, crossed it with [mod] yeast and examined the sensitivity to 5-FU. The diploid recovered the ability to grow on 5-FU plates, indicating that the transient loss of Mod5 in [MOD+] yeast eliminated the [MOD+] state (Fig. 3B). Thus, the continuous expression of Mod5 is necessary for propagation of [MOD+], and Mod5 is the protein determinant of [MOD+]. Next, we explored non-Mendelian inheritance of [MOD+], but the tetrad analysis of [MOD+] diploids was unsuccessful (18). Thus, we examined cytoplasmic inheritance of [MOD+] by cytoduction. We crossed a kar1-1 Δtrm1 yeast with the [MOD+] or [mod] diploids that had been converted to Mat a/a diploids (18) and tested cytoductants for growth on 5-FU plates to identify [MOD+] colonies. About half of the cytoductants from [MOD+] showed sensitivity to 5-FU (47%), whereas those from [mod] did not (0%) (Fig. 3C). Thus, [MOD+] is a cytoplasmically inherited genetic trait.

Fig. 3

Dominant and cytoplasmic inheritance of [MOD+] genetic traits. (A) [MOD+] haploids isolated by protein infection show GdnHCl-reversible 5-FU sensitivity. [mod] and [MOD+] haploids before and after treatment with 3 mM GdnHCl were spotted on YPD in the presence of 5-FU (15 μg/ml). (B) [MOD+] is dominantly inherited, and the propagation of [MOD+] requires continuous expression of Mod5. [MOD+], Δmod5, or [MOD+mod5 strains were crossed with [mod] haploids. The sensitivity of the resulting diploids to 5-FU is shown. (C) [MOD+] is inherited by cytoduction. [mod] and [MOD+] strains (donor), a recipient, and a representative of cytoductants were spotted on YPD with 5-FU (20 μg/ml). Fivefold serial dilutions of yeast cells were spotted in the sensitivity assay.

Finally, we investigated physiological consequences of the prion conversion of Mod5. Mod5 catalyzes isopentenylation of tRNA by transferring dimethylallyl pyrophosphate (DMAPP) to tRNA A37; DMAPP is also a substrate for Erg20 in the sterol biosynthetic pathway (29). Thus, a decrease in the tRNA modification by less soluble (functional) Mod5 should boost the sterol synthesis. We found that [MOD+] cells contain lower levels of the tRNA modification [isopentenyladenosine (i6A)] and higher ergosterol levels than [mod] cells (Fig. 4, A and B, and fig. S8) (30, 31). [MOD+] yeast showed resistance to a microtubule inhibitor, nocodazole (Fig. 4C), as in the case of overexpression of Erg20 in [mod] yeast (fig. S9). Thus, the prion conversion to [MOD+] states stabilized microtubule structures. [MOD+] yeast also acquired resistance against antifungal agents such as fluconazole, ketoconazole, and clotrimazole that inhibit ergosterol biosynthesis (Fig. 4D), presumably because of the increased ergosterol levels. Δtrm1 was not responsible for the acquired antifungal resistance of [MOD+] because the rescue of Trm1 in [MOD+] yeast did not alter the antifungal resistance (Fig. 4D). To address whether the antifungal resistance of [MOD+] is linked to positive selection under environmental stress, we examined if the de novo appearance of [MOD+] prion states could be detected by culturing nonprion [mod] yeast with antifungal drugs. The [MOD+] prion state appeared in culture in the presence of antifungal agents, but not sodium chloride that causes general stress conditions (Fig. 4E) (10). The de novo appearance of [MOD+] states was also seen in a wild-type [mod] strain that expresses intact Mod5 and Trm1 (fig. S10), indicating that the GFP tag in Mod5 or the deletion of Trm1 was not responsible to the selective advantage of [MOD+]. The de novo appearance of prion states was selective for [MOD+] because the color of [MOD+] yeast remained red; hence [PSI+] states did not appear. Thus, whereas prions are known to be unstable when they first appear, antifungal drugs selectively allowed newly appearing [MOD+] prions to grow and stabilize. Next, we performed competition experiments to examine which [MOD+] or [mod] yeast grows preferentially in the absence or presence of antifungal drugs. In the presence of fluconazole, [MOD+] yeast showed a growth advantage (Fig. 4F), indicating positive selection of [MOD+] under the pressure of antifungal agents. In contrast, the fraction of [MOD+] cells was decreased in cultures without fluconazole. This reduction was not due to prion loss (18) but rather to differences in the doubling time of yeast between [MOD+] (150 min) and [mod] (124 min) because the distinct doubling time predicted the decrease in the fraction of [MOD+] (Fig. 4F). Furthermore, Δtrm1 did not contribute to the competitive growth of [MOD+] and [mod] strains (fig. S11). Thus, the [mod] nonprion yeast became dominant when both [MOD+] and [mod] yeasts were released from the pressure of antifungal agents. These results uncover a cellular mechanism in which a conformational switch of non–Gln/Asn-rich Mod5 from a soluble state to an aggregated form allows the yeast to adapt to the harmful environment of antifungal drugs by up-regulating ergosterol biosynthesis at the expense of tRNA modification (fig. S12). The dominance of the [MOD+] yeast due to its growth advantage was eventually lost when the cells were released from the selective pressure of antifungal drugs. Thus, yeast cells employ prion conversion only when necessary for cell survival.

Fig. 4

Physiological roles of [MOD+] genetic traits. (A) Relative amounts of isopentenyladenosine (i6A) in tRNA of [mod ] and [MOD+] yeasts. (B) Relative ergosterol levels in [mod ] and [MOD+] yeasts. (C) The resistance of [MOD+] to nocodazole, a microtubule inhibitor. Fivefold serial dilutions of yeast cells were spotted on YPD plates with or without nocodazole (3 μg/ml). (D) Resistance of [MOD+] to antifungal agents was examined by a halo assay. [MOD+] and [mod ] cells (left panels) or those cells in which Trm1 is supplemented from a low-copy plasmid (right panels) were plated onto YPD plates. 10 μl of fluconazole (2 mg/ml), ketoconazole (100 μg/ml), and clotrimazole (50 μg/ml) were spotted onto a round paper filter. (E) Frequency of de novo appearance of [MOD+] by YPD culture with fluconazole (50 μg/ml), ketoconazole (10 μg/ml), or sodium chloride (0.5 M). (F) Growth advantage of [MOD+] in YPD culture with fluconazole. Yeast cells of [mod ] and [MOD+] strains were mixed at the 1:1 ratio and grown in YPD (black) or YPD containing fluconazole (50 μg/ml) (gray). A fraction of [MOD+] yeast was determined at each time point. Theoretical relative ratios of [MOD+] in YPD calculated from doubling times of [mod ] (124 min) and [MOD+] (150 min) strains are shown by a dotted line. Error bars denote SD. *P < 0.01, based on an independent t test (n = 3 experiments).

Acquisition of resistance to drugs including antifungal agents is a historical problem in medicine and agriculture. Recently, it has been shown that Hsp90 regulates the phenotype of antifungal resistance in pathogenic yeasts (32). This study suggests that active prion conversion in response to environmental selection may also be responsible for a wide spectrum of cellular adaptation and that cells may have evolved epigenetic prion conversion for fast on-demand adaptation in stressful environments to complement slower genetic adaptation processes and without the risk of generating deleterious mutations. In summary, our findings expand the definition of prion conversion beyond the disease state to a normal control mechanism for cellular fitness adaptation during environmental selection.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S13

Tables S1 and S2

References (3343)

References and Notes

  1. Materials, methods, and additional data are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank J. Weissman (Univ. of California, San Francisco) for providing plasmids and yeast stains, C. Yokoyama for critical reading of the manuscript, Y. Komi and M. Yoshizawa for help with the construction of plasmids and yeast strains and drug-sensitivity assays, Y. Ohhashi for advice on structural analyses of amyloids, Y. Nekooki for providing huntingtin-exon1 Q62 protein, N. Takahashi for help with the screen for QIN, and Y. Sakamaki for transmission electron microscopy analysis. DNA sequencing and mass spectrometry were performed at the RIKEN Brain Science Institute Research Resources Center facility. Funding was provided by Japan Science and Technology Agency PRESTO; grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Priority Area on Protein Society); the Next Program; and the Sumitomo Foundation and the Novartis Foundation (Japan) for the Promotion of Science (M.T.).
View Abstract

Stay Connected to Science

Navigate This Article