Saccharomyces cerevisiae: a sexy yeast with a prion problem.

Yeast prions are infectious proteins that spread exclusively by mating. The frequency of prions in the wild therefore largely reflects the rate of spread by mating counterbalanced by prion growth slowing effects in the host. We recently showed that the frequency of outcross mating is about 1% of mitotic doublings with 23-46% of total matings being outcrosses. These findings imply that even the mildest forms of the [PSI+], [URE3] and [PIN+] prions impart > 1% growth/survival detriment on their hosts. Our estimate of outcrossing suggests that Saccharomyces cerevisiae is far more sexual than previously thought and would therefore be more responsive to the adaptive effects of natural selection compared with a strictly asexual yeast. Further, given its large effective population size, a growth/survival detriment of > 1% for yeast prions should strongly select against prion-infected strains in wild populations of Saccharomyces cerevisiae.

Prions are known to cause a variety of fatal neurodegenerative diseases in humans, livestock and wildlife. A number of prion proteins have been characterized in the budding yeast Saccharomyces cerevisiae, but their effects on yeast fitness remain controversial. The most commonly studied variants of the yeast prions, [PSI+], [URE3] and [PIN+] appear relatively harmless, though the identification of toxic and lethal variants of these prions indicates that laboratory growth conditions artificially select for mild prion variants. Some claim that prions are beneficial for yeast,- and have been naturally selected to promote evolvability. However, our recent work demonstrated that even the mildest forms of [PSI+], [URE3] and [PIN+] are deleterious diseases that slow cell growth or impair survival by greater than 1%. We suggest that the detrimental nature of infectious prions hinders their spread in the wild and contributes to their rarity in natural yeast isolates.

While budding yeast predominantly reproduce mitotically, nutrient deprivation induces meiosis in diploids, with each cell producing four haploid spores. Spores are either mating type a or α, and mating partners must have different mating types. Therefore mating can occur between spores with opposite mating types within an ascus (intratetrad) or with a spore originating from a different diploid parent (outcrossing). Alternatively, if germinated spores do not mate immediately, cells can undergo a mating-type switch (homothallism) which produces a mixture of a and α cells from the same clone that can subsequently mate. Outcrossing is the predominant mode of prion transmission because with homothallism and intratetrad mating, spores within a single tetrad will likely be all infected or all uninfected.

The frequency of prion infections in the wild is affected by the rate of transmission, spontaneous loss or gain of the prion state and the fitness effects that prions impart on their host (Fig. 1). If yeast prions provide a benefit for their hosts as suggested by others, then natural selection would act to increase their frequency in natural populations. Based on this assumption, and the fact that they spread easily through mating, we would expect prions to be common in nature. However, surveys from wild and industrial Saccharomyces isolates found that most yeast prions are extremely rare in natural populations.,- A recent study found [PSI+] in < 1% and [PIN+] in ~6% of 700 wild isolates surveyed. Furthermore, attempts to detect prions in several other yeast species (Saccharomyces paradoxus, Saccharomyces bayanus, Naumovozyma castellii, Saccharomyces dairensis, Saccharomyces exiguus, Saccharomyces unisporus, Saccharomyces kluyveri and Pichia methanolica) were unsuccessful.

Figure 1. The equilibrium frequency of prions in wild populations is determined by spread, growth slowing effects and loss/gain of the prion state.

The rarity of yeast prions in the wild could reflect negative fitness consequences, high rates of prion loss, infrequent transmission or a combination of these. To quantify transmission rates in the wild, we used the naturally occurring 2 μ DNA plasmid as a model for infectious element spread among natural isolates. Like yeast prions, 2 μ plasmids are exclusively intracellular, and thus spread solely through outcross mating. We used the known rate of spontaneous loss and gain of 2 μ DNA, the known growth detriment of the plasmid (1−3%,) and its frequency in nature (38/70) to estimate outcrossing rates for yeast. Our model projected that yeast outcross once per 100 cellular divisions, an estimate that is much higher than previous reports for S. cerevisiae and S. paradoxus., Genomic-based methods utilizing recombination as a proxy for mating produced estimates of one outcross per 50,000 mitotic divisions for S. cerevisiae and one per 100,000 divisions for S. paradoxus. Recombination-based methods for estimating outcrossing rely heavily on the detectability of heterozygotes. In highly clonal populations, occasional outcrossed mating may not be detected with genetic methods because spatially proximate mating partners are highly related or because mitotic recombination has resulted in a loss of heterozygosity. Furthermore, the power to detect recombination (and therefore mating) is reduced for populations with low diversity and low rates of recombination, a potential problem for some yeast populations.

In addition to estimating the overall outcrossing frequency, we were able to use genetic heterozygosity as an indicator of the proportion of matings that are outcrossed (vs. intratetrad or homothallic). A previous study, that we earlier failed to cite, showed that wild S. cerevisiae isolates are largely heterozygous, and thus that mating is commonly of the outcross type, thereby, pre-confirming our findings in this regard. We used nucleotide sequences at RNQ1 (coding for Rnq1p, the protein that forms the [PIN+] prion) to estimate population heterozygosity because the high polymorphism at this locus facilitated the detection of heterozygotes. Our results indicated that 23–46% of matings are outcrossed, but this is a minimum as it does not take mitotic recombination into effect. Our findings are consistent with other experimentally derived mating frequencies. Murphy and Zehyl (2010) found that yeast grown in dense cultures outcross 10–25% of the time, while spores in physically adjacent asci outcross with frequencies ranging from 10–50%. Similar rates of outcrossing were reported for a geographically isolated population of S. cerevisiae in New Zealand, where ~20% outcrossing was observed among genetically subdivided lineages.

Our estimate of outcrossing frequency equates to the transmission rate for infectious prions in wild yeast. Using this information we were able to subsequently estimate the fitness effects of [PSI+], [URE3] and [PIN+], again assuming that the observed frequency of prions in the wild reflects transmission rate, spontaneous loss or gain of the prion state (derived experimentally for [PSI+] and for [URE3],) and any fitness effects afforded to the host (Fig. 1). With our estimate of 1% outcrossing and the observed prion frequencies of 1%, 1% and 16% for [PSI+], [URE3] and [PIN+], respectively, we showed that yeast prions slow cellular growth by at least 1% (Fig. 2).

Figure 2. Relative growth rate of [PRION+] vs. [prion-] cells for varying frequencies of mating, prion loss and equilibrium prion prevalence. For all calculations, the probability of spontaneous prion gain = 1.6 × 10−6. Experimentally determined values of each variable (prevalence = 0.01, mating = 0.01, loss = 0.00001) are indicated by *. Horizontal lines indicate relative growth rate = 1, where the fitness effects of the prion are neutral. Relative growth rate > 1 indicates a survival advantage for [PRION+] cells, while relative growth rate < 1 indicates growth detriment for [PRION+] cells.

Masel and Griswold (2009) used theoretical modeling to extrapolate the selective forces acting against [PSI+] and in agreement with our data found that [PSI+] is usually deleterious. Their results however, suggest that the detriment of [PSI+] is 100 times lower than our estimate of 1%. This difference is likely due to their assumption that an outcrossing frequency of 0.001% as estimated for S. paradoxus, was generalizeable for S. cerevisiae. Recent genomic surveys have shown that the population structure of S. paradoxus differs dramatically from S. cerevisiae, with the latter species showing extensive mixing of lineages among subpopulations. Thus, mating rates estimated for S. paradoxus may not hold true for S. cerevisiae, and likely resulted in a substantial underestimate of [PSI+] detriment with theoretical models. Indeed our own analyses showed that estimates of prion growth detriment are extremely sensitive to changes in mating frequency (Fig. 2).

Some have proposed that prions promote cell survival.,, This hypothesis is mainly based on claims that [PSI+] can be an advantage under some circumstance in some strains,, and is induced in the presence of certain cellular stressors. Furthermore, ure2 cells (void of Ure2p, the protein monomer of [URE3]) were more tolerant of certain alkali metals ions compared with wild type, even though [URE3] slows the growth of most strains on the usual laboratory medium. The reported resistance of [PSI+] cells to heat and high ethanol stress was not confirmed by a later report, and the benefits reported by the later report, were also not confirmed by others using identical strains. The reports of [PSI+] induction by stress conditions used a non-natural Sup35 prion domain known to have an elevated frequency of generating [PSI+]. In our recent work we tested whether the stress conditions reported to induce [PSI+] with the artificial prion domain would do so with the native Sup35p prion domain. We have been unable to observe toxicity-based [PSI+] induction, suggesting that cells are not adapted to form prions in response to stress.

Despite the lack of experimental evidence that [PSI+], [URE3] and [PIN+] are beneficial, an adaptive role for yeast prions is often proposed in the literature. Evolutionary conservation of yeast prion forming ability is often cited as evidence that they provide a fitness advantage.,, A few experimental studies have shown that overexpression of prion domains (domains necessary and sufficient for amyloidization) and occasionally full length proteins from various Ascomycetes will induce prion formation by that domain/protein in S. cerevisiae cells.,- These cross-species inductions are interpreted as indicators of evolutionary conservation of prion-forming ability in yeast. However, this inference fails to acknowledge the artificiality of induction conditions (overexpression of prion domains and the use of inter-species fusion proteins); such conditions would not be encountered by yeast in natural environments. Furthermore, S. cerevisiae proteins can be induced to form prions from other species, but prion forming ability has not been assessed in the majority of yeasts., In addition, the Ure2p proteins of Kluyveromyces lactis and Saccharomyces castellii, are unable to form [URE3]. The K. lactis report from Safadi et al. (2011) is particularly important since this yeast is closely related to S. cerevisiae and the authors showed that in the K. lactis host, the K. lactis Ure2p could not form [URE3]. Ure2p of Candida albicans can form the [URE3] prion in Candida glabrata, but Candida glabrata Ure2p is unable to form [URE3] in either its own cells or in S. cerevisiae cells., Thus, prion-forming ability of Ure2p is generally not conserved. Further, the existence of a phenomenon in many related species does not provide evidence that it benefits the organisms. Most vertebrates suffer from broken legs and arthritis, but this hardly qualifies these phenomena as helping the animal. Of course, even a broken leg can (rarely) be beneficial as discussed elsewhere.

Molecular studies of yeast prion proteins also claim that prion-forming domain sequences have been conserved by evolution. Jensen et al. (2001) reported evolutionary constraint for the Sup35 N and M-domains based on relatively low rates of non-synonymous compared with synonymous substitution for this region. Compared with a more recent study, the number of Sup35 polymorphisms reported by Jensen et al. (2001) is relatively low, likely due to a limited number and narrow range of strains analyzed. Additionally, functional constraint for Sup35 N and M domains cannot be solely attributed to prion formation, as these domains also play an active role in mRNA degradation. Likewise, the prion domain of Ure2p prevents degradation of the protein in vivoand may affect interactions with other proteins. Further, since prion-forming ability does not depend on sequence, but mostly on amino acid content, even if prion-forming ability were selected and conserved, it would not necessarily result in conservation of amino acid sequence.

Though prion-like domains can be found in a variety of budding yeast, this cannot be taken as direct evidence for evolutionary constraint of prions in fungi, particularly given their non-prion functions. Moreover, prions documented in wild yeast isolates are likely the most benign, and thus easily detected, variants. Lethal and toxic prions are generally not detectable in the laboratory under standard growth conditions because they abolish or severely inhibit cell growth. Still, even a 1% growth detriment for benign prions could have substantial selection effects in wild populations. Neutral theory of evolution predicts that alleles are deleterious when s > 1/(2N) (where s is the selection coefficient against an allele and N is the number of breeding individuals; calculated with N from, s > 10−8 would be deleterious in yeast). Their large population size would therefore make yeast sensitive to modest fitness effects from prions.

Our finding that Saccharomyces cerevisiae is far more sexual than previously thought has far reaching implications for the evolutionary dynamics of this species. Sex increases genetic variability, promotes adaptation and allows natural selection to proceed more efficiently. Through recombination, sex can bring together multiple beneficial mutations in a single generation, whereas a purely clonal population would have to acquire successive beneficial mutations along a single lineage to incur the same fitness effects. The novel gene combinations created by sex increase fitness variation among individuals, which in turn increases the efficaciousness of natural selection. When grown in glucose limiting media at elevated temperatures (37°C), sexual yeast had higher relative fitness over 300 generations as compared with isogenic asexual yeast. The fitness advantage for sexual yeast was not observed in benign growth conditions (30°C in glucose limiting media), where both asexual and sexual yeast are sufficiently adapted, precluding fitness variation necessary for natural selection. In nature, wild yeast are likely to encounter extreme growth conditions such as freeze-thaw cycles, elevated temperatures, nutrient deprivation, etc. Thus in the wild we would expect sexual yeasts to adapt more efficiently than their asexual counterparts.

The loss or acquisition of the prion state is associated with phenotypic changes in the cell. For example, [PIN+]-infected cells gain the ability to induce other prions,, and [URE3]-infected cells inherit a defect in nitrogen catabolite repression. [PSI+]-infected cells have reduced translational termination, which allows some genes with internal stop codons and nonsense mutations to be translated., Despite speculation that the transient and heritable phenotypic changes observed in prion-infected laboratory strains could promote evolvability,, selection for evolvability has never been documented in yeast, and is generally not feasible for sexual populations. First of all, most novel mutations are deleterious. Indeed, the Saccharomyces cerevisiae genome as a whole is under purifying selection, an indication that there is pressure to reduce accumulation of deleterious alleles. Selection for evolvability is well documented in asexual organisms, because modifiers that increase mutation rate can hitchhike to higher frequencies when they are linked to a positively selected locus.- However, with 1% outcrossing, Saccharomyces cerevisiae is by and large sexual. We would expect recombination to break down associations between mutation modifiers and beneficial alleles which will rapidly result in the loss of modifiers from sexual populations. Thus selection for evolvability is unlikely to explain phenotypic variation associated with prions. A more straight forward explanation is that prion mediated disruption of intracellular protein levels changes the cellular environment.

Given the complexities of disease transmission in wild yeast populations, it is difficult to predict if negative selection is sufficient to counteract spread by mating and decrease prion prevalence over time. This is especially true for Saccharomyces cerevisiae, given ecologically-driven differences in mating propensity observed for isolates in different environments. Magwene et al. (2011) found that outcrossing rates ranged from 36–52% for domesticated yeast (e.g., vineyard, clinical strains) as compared with 2% for strains isolated in natural environments (e.g., tree bark, fruit). Domestication could increase contact rates among different yeast strains and create mating opportunities that would otherwise be rare for isolates in low-density natural environments. Our own data showed that 14/15 [PIN+] strains originated from domesticated environments. Additionally, 12/15 [PIN+] strains were heterozygous, a likely indication that prions behave as sexually transmitted diseases in yeast. Taken together, these findings suggest that prion prevalence for yeast in human-associated environments is higher than for undomesticated yeast due to increased sexual transmission in highly outcrossed, domesticated strains.

Our interpretation of the rare occurrence of yeast prions in nature is that they are deleterious, and this negative selection outpaces prion transmission in spite of substantial rates of outcrossing in Saccharomyces cerevisiae . Though they have been extensively studied in the laboratory, few studies address prion disease dynamics in natural populations. Ecological factors such as mating clearly influence prion transmission rates and it would be interesting to determine if there are additional environmental variables that can influence the spread of prions in the wild.