Widespread prion-based control of growth and differentiation strategies in Saccharomyces cerevisiae

SUMMARY

Theory and experiment suggest that organisms would benefit from pre-adaptation to future stressors based on reproducible environmental fluctuations experienced by their ancestors. Yet mechanisms driving pre-adaptation remain enigmatic. We report that the [SMAUG⁺] prion allows yeast to anticipate nutrient repletion after periods of starvation, providing a strong selective advantage. By transforming the landscape of post-transcriptional gene expression, [SMAUG⁺] regulates the decision between two broad growth and survival strategies: mitotic proliferation or meiotic differentiation into a stress-resistant state. [SMAUG⁺] is common in laboratory yeast strains, where standard propagation practice produces regular cycles of nutrient scarcity followed by repletion. Distinct [SMAUG⁺] variants are also widespread in wild yeast isolates from multiple niches, establishing that prion polymorphs can be utilized in natural populations. Our data provide a striking example of how protein-based epigenetic switches, hidden in plain sight, can establish a transgenerational memory that integrates adaptive prediction into developmental decisions.

Graphical Abstract

INTRODUCTION

Living systems endure constantly changing environments (Bell and Collins, 2008). To cope with this variability, organisms sense fluctuations in their surroundings and respond by altering their growth and survival strategies. In nature, however, environmental fluctuations are often correlated, occurring in a predictable sequence. In theory, the capacity to anticipate and pre-adapt to an upcoming stress would offer significant advantages (Tagkopoulos et al., 2008). Laboratory evolution experiments in which precise sequences of environmental stresses recur over hundreds of cycles can generate subpopulations of cells that ‘pre-adapt’ to future stressors. Continued propagation in static conditions leads to the loss of adaptive prediction, suggesting that inappropriate pre-adaptation incurs a fitness cost (Mitchell et al., 2009). The degree to which this phenomenon occurs in nature, and the underlying mechanisms involved, remain largely unexplored.

In nutrient-rich environments, the budding yeast Saccharomyces cerevisiae replicates rapidly via mitotic division. In this organism and many others, stress (e.g. nutrient starvation) triggers a developmental program, meiosis, that re-assorts and diversifies genetic material (Marston and Amon, 2004; McDonald et al., 2016; Miller et al., 2013; Neiman, 2011). This culminates with the formation of haploid spores, which are resistant to many stresses (Calahan et al., 2011; Coluccio et al., 2008; Smits et al., 2001). If a stressful condition is long-lasting, commitment to this program provides a strong advantage. If the stress proves transient, it would be preferable to delay meiotic commitment, and quickly resume mitotic growth after the stress ceases. Hardwiring of either strategy (via genetic mutation) allows progeny to benefit from the experience of their ancestors, but could prove maladaptive if future environmental fluctuations differ. By contrast, the ability to reversibly switch between developmental strategies could allow a population to thrive in a dynamic environment, expanding its niche. In theory, a mechanism that permits both strategies — a mode of inheritance that is simultaneously heritable and reversible — would provide significant adaptive value (Jablonka et al., 1995; King and Masel, 2007; Suzuki et al., 2012).

The unusual folding landscapes of prion proteins provide one such mechanism. Prions are stable, heritable, self-templating protein conformations that can serve as a robust mechanism for transgenerational epigenetics. In contrast to most other epigenetic states, prions are meiotically-stable, allowing for true transgenerational memory (Harvey et al., 2018). They are a reversible and lost spontaneously more frequently (~10⁻⁵ per cell division; reviewed in: Halfmann and Lindquist, 2010; Harvey et al., 2018; Rando and Verstrepen, 2007) than mutations in DNA (~10⁻⁸ per cell division; Lang and Murray, 2008). Although their adaptive value has been controversial (Wickner et al., 2011), prions can be beneficial in many stresses (Harvey et al., 2018; Jarosz et al., 2014a; Jarosz et al., 2014b), leading to the proposal that prions could provide adaptive value as bet-hedging devices in fluctuating environments (Halfmann et al., 2010; King and Masel, 2007; Lancaster et al., 2010). Modeling suggests that this property would have been sufficient to drive evolutionary retention of several prions (Jarosz et al., 2014a; Jarosz et al., 2014b; King and Masel, 2007; Simons, 2009). Regulated, but non-infectious, protein aggregation is also emerging as an important contributor to cell physiology in processes including metabolism (Saad et al., 2017), aging (Saarikangas and Barral, 2015; Schlissel et al., 2017), pheromone response (Caudron and Barral, 2013), translation (Franzmann et al., 2018), and meiosis (Berchowitz et al., 2015).

Meiosis is one of the most complex decision-making processes in which S. cerevisiae engages, its efficiency varies substantially across this species (Mortimer, 2000). Common laboratory strains are notoriously poor sporulators (Gerke et al., 2006), whereas some wild strains sporulate far more readily (Liti et al., 2009; Warringer et al., 2011). Some genetic determinants of this variability have been characterized (Ben-Ari et al., 2006; Gerke et al., 2009). But what epigenetic factors contribute, if any, remain unknown.

Here we investigate the advantages of a heritable epigenetic element that promotes mitotic proliferation and delays meiosis in S. cerevisiae. We report that [SMAUG⁺], a prion formed by the S. cerevisiae homolog of the RNA-binding protein Smaug (encoded by the VTS1 gene), controls this cellular decision. Modeling and experiment suggest that [SMAUG⁺] has adaptive value in fluctuating environments as an adaptive prediction strategy. [SMAUG⁺] is ubiquitous in laboratory yeast strains, which have been exposed to repeated cycles of starvation followed by nutrient repletion, where it re-shapes the post-transcriptional landscape underlying meiotic commitment. Distinct [SMAUG⁺] variants are widespread in wild yeast from diverse ecological niches, establishing that structural polymorphism within this prion is common in nature. Our observations provide a striking example of protein-based epigenetic inheritance hidden in plain sight, profoundly altering growth and differentiation strategies.

RESULTS

Modeling tradeoffs between mitotic and meiotic proliferation strategies

We began by modeling tradeoffs between mitotic proliferation and meiotic differentiation. In nutrient-rich conditions, S. cerevisiae proliferates rapidly via mitotic cell division. However, when nutrients are scarce, this diploid organism commits to sporulation, a developmental program that encompasses meiosis (Chu et al., 1998; Marston and Amon, 2004; Miller et al., 2013). In addition to genetic diversification, meiosis creates a differentiated stress-resistant cell state, a haploid spore (Calahan et al., 2011). If conditions remain stressful, facile commitment to sporulation would provide a strong survival advantage, but it could inflict a strong disadvantage if a stress is transient.

We simulated these scenarios in silico (Figure 1A), considering two populations of cells, ‘proliferators’ that commit to meiosis late and ‘sporulators’ that commit early. We assigned a growth advantage (μ) to efficient proliferation in nutrient-rich conditions and imposed a penalty (1–z) on individuals that did not rapidly sporulate in the stressful environment (Figure 1A). We assumed that both populations could achieve the same maximum sporulation efficiency (see Method Details).

Figure 1. Tradeoffs between meiotic and mitotic proliferation in fluctuating environments.

(A) (Left) Schema of the model framework. Mitotic proliferation is represented by logarithmic growth as a function of t₁, the period of nutrient abundance. Meiotic proliferation is represented by a simple Hill function of t₂, which captures differences in the rate of sporulation, but maintains the same maximum. (Right) Schema of the sporulator (S) and proliferator (P) developmental strategies. (B) (Left) Framework for altering the environmental parameters, duration of abundance and starvation, of the model. (Right) Example simulation showing the relationship between log₁₀(P/S) and nutrient starvation (t₂). Highlighted are the t₂ values where small changes would alter which developmental strategy is more successful. (C) Parameter space of simulated populations of proliferators relative to sporulators in fluctuating environments. For each plot, there are two independent variables: the period of starvation (t₂) and the title variable (changed at intervals; right bar). For all plots, the simulation was run for N = 10 cycles.

We iteratively simulated the fate of mixed populations of proliferators and sporulators through multiple cycles of feast and famine, varying intrinsic (e.g., growth and death rates) and extrinsic (e.g. duration of stress) parameters (Figure 1A). Over a wide range, centring on values that have been experimentally observed in S. cerevisiae (see SI for further discussion).

These simulations revealed clear ‘niches’ in which it was more beneficial to be either a proliferator (i.e., log₁₀(P/S) > 0) or a sporulator (i.e., log₁₀(P/S) < 0) (Figure 1B, 1C and S1A). For example, we regularly observed non-monotonic behavior when varying the period of nutrient starvation. When times of scarcity were brief, proliferators flourished. Conversely, as times of scarcity lengthened, sporulators benefited. However, this advantage diminished and ultimately disappeared with lengthened periods of scarcity. The general patterns we observed were robust to changes in the initial ratio of proliferators to sporulators (Figure S1B). Critically, slight perturbations to the duration of environmental stress, as little as an hour, could switch which strategy would be more successful (Figure 1C).

The capacity to reversibly switch between strategies could then allow an organism to thrive in a wider range of environmental conditions (Lachmann and Jablonka, 1996; Lachmann and Jablonka, 1996; Lancaster et al., 2010; Lancaster and Masel, 2009). Moreover, utilizing a metastable memory of previous switching events to inform future decisions would advantageous when environmental fluctuations exhibit long-term temporal correlations (Kussell and Leibler, 2005). Together with these previous studies, our modeling suggests that an epigenetic switch enabling oscillation between a proliferator and sporulator identity that is heritable, but also reversible, would expand the niches in which S. cerevisiae could thrive.

A prion-based proliferative growth strategy in laboratory yeast strains

Protein-based forms of epigenetic inheritance, e.g. prions, are an example of a heritable phenotypic switch that would allow cells to benefit from the memory of their ancestors but also be reversed more readily than a mutation. We recently showed that prions are pervasive in nature and can influence growth in a wide variety of environments (Chakrabortee et al., 2016; Halfmann et al., 2012). Repeated cycles of starvation followed by nutrient repletion are inherent to husbandry of microbial cultures. Therefore, we hypothesized that laboratory cultivation might select for a reversible epigenetic factor of the type we had modeled above.

We investigated whether the S. cerevisiae laboratory strain BY4741, which is widely used in systematic genetic and cell biological studies of this organism (Brachmann et al., 1998; Costanzo et al., 2016; Giaever et al., 2002; Huh et al., 2003; Winzeler et al., 1999) harbors a prion that confers a mitotic growth advantage. To eliminate prion-like elements from this strain, we took advantage of their strong dependence on protein chaperone activity (Chernoff et al., 1995; Eaglestone et al., 2000; Masison et al., 2009; Shorter and Lindquist, 2004). Propagation of most prions requires robust Hsp70 activity. We therefore transformed derivatives of BY4741 with a plasmid-based dominant-negative allele of Hsp70 (Ssa1-K69M; Chakrabortee et al., 2016; Garcia et al., 2016; Lagaudriere-Gesbert et al., 2002), propagated these transformants for 75 generations, eliminated the plasmid, and then propagated them for an additional 25 generations to restore chaperone function. This regimen eliminates most amyloid and non-amyloid prions that have been tested (Chakrabortee et al., 2016). We hereafter refer to strains subjected to this regimen as ‘cured’.

We next examined how curing affected fitness, competing fluorescently-marked cured and uncured strains in co-culture and diluting into fresh medium every 48 hours. At each dilution, we determined the number of cured and uncured cells by flow cytometry. Cured BY4741 was reproducibly outcompeted by its uncured parent (Figure S2A). We also cured BY4743 (the diploid derivative of BY4741) and performed analogous competitions, observing an even stronger selection coefficient relative to its cured derivative (s = 0.9%; Figure 2A). As a frame of reference, these values are larger than the average fitness ascribed non-essential genes (Breslow et al., 2008), and natural genetic variants in this organism (Jakobson and Jarosz, 2019; Sharon et al., 2018).

Figure 2. The laboratory strain BY4743 contains a prion.

(A) Competitions between cured and uncured BY4743 were passaged every 48 hours (left) or every 24 hours (right). Each point represents the natural log of the number of cells expressing the mNeonGreen (Neon) fluorescent marker over the number of cells expressing the mKate2 (Kate) fluorescent marker (Wong et al., 2018; Zalatan et al., 2012), and the mean of three independent biological replicates. Filled dots represent data obtained when cured cells were marked with Neon and uncured cells were marked with mKate2. Unfilled dots represent the marker-swap experiment. Selection coefficients were calculated from a linear fit to these data (p < 0.0001 for all fitted lines; Chevin, 2011). Shaded area indicates the 95% confidence interval of the fitted line. Shown are the selection coefficients ± SEM of the reciprocal experiments. (B) (Top) Representative micrographs of cured and uncured BY4743 cells after 4 days of sporulation. White asterisks denote spores. Scale bar represents 10 μm. (Bottom) Sporulated fractions of cured and uncured BY4743 vs. time. Sporulation rate (fraction sporulated/hour): BY4743: 6.9 × 10⁻⁴; cured BY4743: 1.2 × 10⁻³. **: p < 0.005 by Mann–Whitney U test; n.s.: p > 0.05. (C) (Bottom) Sporulation efficiency of cytoductants with BY4741 and cured BY4741 as donor strains (p-value calculated by Mann-Whitney U test). (D) Sporulation efficiency of four sister progeny from diploid generated by crossing an uncured haploid crossed with a cured haploid P-values calculated by Sidak’s multiple comparison test. A cured BY4743 is shown as a control. In all panels in this figure, error bars represent SEM from the mean of three independent biological replicates.

Our model predicts that a proliferative strategy would be more beneficial when periods of nutrient scarcity are shorter. To test this, we performed another competition, instead diluting the cultures every 24 hours. This resulted in a much stronger benefit for uncured cells (s = 1.8%, twofold stronger than with 48 hour dilutions; Figure 2A). Thus, even modest changes in the period of environmental fluctuations can have a profound impact on the selective advantage of this chaperone dependent epigenetic element.

We next asked whether chaperone curing would also increase meiotic differentiation into a stress resistant spore, as might be predicted by our model. Cured BY4743 sporulated 3.1-fold more efficiently after 5 days than its isogenic, uncured parent (Figure 2B). Over dozens of independent propagations, we never observed spontaneous reversion of this phenotype, establishing its robust heritability. We observed a similar, curable, repressed sporulation phenotype in isolates of this strain obtained from multiple laboratories and genetic stock centers and in other widely used laboratory strains such as W303 (Figure S3A), establishing that this behavior is widespread.

To determine whether this curable epigenetic element was prion-like, we tested a defining feature of this form of inheritance: cytoplasmic transmission. We mated BY4741 (donor) to a cured, petite karyogamy deficient (kar1Δ–15 rho⁻) recipient strain. This yielded heterokaryons that received cytoplasm, but no nuclear material, from BY4741 (Figure 2C). We obtained cytoductants by selecting for haploid buds with restored mitochondrial function and a nuclear-encoded auxotrophic marker unique to the recipient strain.

We next used these recipients as donors for a ‘reverse’ cytoduction (see Method Details) to obtain karyogamy competent cells with a mixed cytoplasm, mated these ‘reverse’ cytoductants to a cured laboratory strain of the opposite mating type, and examined the sporulation of the resulting diploids. All diploids that received the cytoplasm of uncured BY4741 sporulated poorly. Those that received the cytoplasm from cured BY4741 sporulated well (Figure 2C), establishing that the phenotype is transmitted cytoplasmically.

Finally, we examined the transmission of this trait in genetic crosses. Because they are based on heritable changes in protein conformation rather than DNA sequence, prion-based traits defy Mendelian patterns of inheritance and are passed to all meiotic progeny; in contrast, mutations are inherited by only half of the progeny (Harvey et al., 2018; Itakura et al., 2018; Lindquist et al., 1995; Wickner, 1994). We sporulated BY4743, dissected complete tetrads, and crossed all four haploid progeny to cured strains of the opposite mating type. Each of these diploid strains sporulated poorly (Figure 2D). That is, the low-sporulation phenotype showed a non-Mendelian pattern of inheritance. All four haploid progeny also exhibited a growth advantage in low glucose (Figure S2B). Based on the strong chaperone-dependence and the cytoplasmic, non-Mendelian patterns of inheritance of these phenotypes, we conclude that laboratory strains of S. cerevisiae harbor one or more prions that restrict sporulation and promote a proliferative growth strategy.

BY4743 harbors [SMAUG⁺]

In the accompanying manuscript (Chakravarty et al., submitted), we characterized [SMAUG⁺] as a prion that confers a proliferative advantage akin to the state we simulated above. We wondered whether a variant of [SMAUG⁺] exists in BY4743, promoting proliferation and restriction sporulation. Indeed, prions can exist as multiple stable variants with differing phenotypic strengths (Jarosz et al., 2014b; Tanaka et al., 2006; Toyama et al., 2007), a point to which we return to below.

To investigate, we took advantage of the fact that prion propagation requires continuous expression of the causal protein, employing a genetic strategy to transiently eliminate Vts1 expression. First, we crossed BY4741 to vts1Δ cells and sporulated the ensuing heterozygous diploids. These diploids sporulated poorly (Figure S3B), consistent with the [SMAUG⁺] phenotype. We next dissected tetrads to obtain meiotic progeny, half wild-type and half vts1Δ. We mated these cells to the naive haploids that we had previously ‘cured’ of prions by transient Hsp70 inhibition (Figure 2B) and examined the sporulation of the resulting diploids. In diploids with the wild-type VTS1 allele, where Vts1 had been constantly expressed, we observed low sporulation. By contrast, diploids derived from spores with the vts1Δ deletion sporulated 5.8-fold more robustly (Figure 3A). Thus, the prion-like sporulation phenotype that is ubiquitous in laboratory yeast strains depends upon continuous expression of the Vts1 protein.

Figure 3. BY4743 harbors [SMAUG⁺].

Schema and sporulation efficiency of (A) the meiotic progeny of uncured haploid crossed to vts1Δ and of (B) the vts1Δ progeny from Figure 3A after the genetic restoration of VTS1. Bars represent means ± SEM from three independent biological replicates. (C) Growth rates of diploid and haploid cured, selectively cured, and uncured strains in low (0.08%) glucose. Means ± SEM from six independent biological replicates. (D) (Left) Schema of protein transformation. (Top) Growth rates of haploid protein transformants in low (0.08%) glucose. Bars represent means ± SEM from six independent biological replicates. (Bottom) Sporulation efficiency of selectively cured strains transformed with either Vts1 condensates or bovine serum albumin (BSA). Bars represent means ± SEM from four and ten independent biological replicates for the Vts1 and BSA transformants, respectively. All p-values in this figure were calculated by Mann-Whitney U test.

We extended this genetic crossing strategy restore Vts1 expression, sporulating and dissecting spores of these cured WT/vts1Δ diploids (Figure 3B) and isolating haploid progeny that contained the wild-type VTS1 allele. We then mated these cells to an isogenic cured strain, creating diploids isogenic to BY4743, but where one parent was cured of all prions by transient Hsp70 inhibition and the other parent was selectively cured of [SMAUG⁺] (but no other prion) by transient loss of Vts1 (denoted as [smaug⁻]). This selectively cured strain sporulated as efficiently as strains cured by transient chaperone inhibition (Figure 3B). Selective curing also eliminated any proliferative advantage that we observed in both haploid and diploid strains (Figure 3C). Thus, the difference in mitotic proliferation and sporulation between BY4743 and cured BY4743 is dependent on a product of the VTS1 gene.

Next, to establish that the low-sporulation phenotype is driven by the self-assembly of Vts1, we performed a protein transformation. We treated [smaug⁻] haploids with Zymolyase to generate spheroplasts that we transformed with condensates made from purified Vts1 protein (Chakravarty et al., submitted). After passaging successful transformants for ~125 generations (to eliminate by dilution any protein introduced during the transformation), we mated these strains to a cured partners and scored sporulation efficiency of the resulting diploids. Diploids derived from the Vts1 transformants sporulated poorly compared to control BSA transformants (Figure 3D). Vts1 transformants also exhibited a growth advantage in low glucose relative to controls transformed with BSA (Figure 3D). We conclude that widely used laboratory strains of S. cerevisiae harbor natural [SMAUG⁺] variants that reduce sporulation and favor a proliferative growth strategy.

Finally, we examined sporulation in the [SMAUG⁺] strain generated by transient exogenous overexpression (hereafter ‘induced’ [SMAUG⁺]; Chakrabortee et al., 2016). Induced [SMAUG⁺] diploids formed 4.9-fold fewer tetrads than BY4743 (Figure S3D), suggesting that it likely represents a stronger prion variant. Notably, increased VTS1 expression (from a GPD promoter) also reduced sporulation in cured BY4743 cells (Figure S3C), raising the possibility that natural [SMAUG⁺] might activate Vts1 function.

Natural [SMAUG⁺] activates RNA decay function

Vts1 binds to specific hairpin RNA loops known as Smaug recognition elements (SREs) and targets transcripts containing them for degradation (Aviv et al., 2006; She et al., 2017). In the accompanying paper (Chakravarty et al., submitted), we report that Vts1 can exist in a self-templating prion conformation that accelerates target degradation. To determine how the natural prion variant impacts Vts1 activity, we turned to a well-established reporter expressing GFP with three SREs in its 3’-untranslated region (GFP-SRE; Aviv et al., 2006). When expressed from a galactose-inducible promoter, we observed a higher level of GFP fluorescence in the cured and selectively cured BY4741 strains than in uncured strains (Figure 4A and 4B). We observed no difference among these strains when using a near-identical reporter in which the SRE elements were permuted to abolish Vts1 binding (Aviv et al., 2003; Figure 4A and 4B). Together, these data establish that the natural [SMAUG⁺] prion is also a hyperactive state of the protein.

Figure 4. Natural [SMAUG⁺] hyperactivates Vts1.

(A) Representative micrographs of BY4741, cured BY4741, and selectively cured BY4741 expressing either a galactose-inducible GFP-SRE (SRE+) or a permuted GFP-SRE (sre−) construct. DIC and GFP channels are shown. Scale bar represents 5 μm. (B) Scatter plots of GFP fluorescence of BY4741, cured BY4741, and selectively cured BY4741 expressing either SRE+ or SRE− constructs as indicated (p < 0.005 by Sidak’s multiple comparison test). Bar denotes mean. N = 50–103 cells.

The endogenous [SMAUG⁺] gene expression program

The choice of mitotic or meiotic cell division is highly regulated in S. cerevisiae (Figure 5A). Because Vts1 is a post-transcriptional regulator of gene expression, we investigated the impact of [SMAUG⁺] on the meiotic transcriptome by performing mRNA-seq on natural [SMAUG⁺] cells and isogenic [smaug⁻] cells before and 14 hours after induction of sporulation. At the 14h timepoint no mature spores were yet visible in either culture, and mRNA abundance measurements from biological replicates clustered closely in principal component analysis (Figure S4A), establishing the robustness and reproducibility of the data.

Figure 5. MUM2 degradation suppresses sporulation in [SMAUG⁺] cells.

(A) Schema of exit from mitosis into meiosis (Neiman, 2011; Simchen, 2009). (B) Ratio of induction of all transcripts induced during sporulation in [smaug⁻] cells vs. [SMAUG⁺] cells (in gray). Purple indicates a subset of the transcripts with annotated function in meiosis (SGD). P-value < 0.0001 by bootstrap t-test (C) The [smaug⁻]/[SMAUG⁺] ratio of induction of meiotic transcripts that were significantly induced in [smaug−] cells. (Inset) Schema of the MUM2 transcript showing the location and sequence of the SRE hairpin loop. (D) Time course of MUM2 expression during early meiosis in the uncured and cured BY4743, measured by qRT-PCR, normalized against TAF10. p-values were determined by Mann-Whitney U test. Bars represent means ± SEM from three independent biological replicates. (E) Sporulation of strains expressing MUM2 under a strong promoter (NOP1; constituitive) and its native promoter (P-values were calculated by Mann-Whitney U test). Bars represent means ± SEM from three independent biological replicates. (F) Sporulation of cured and uncured BY4743 with either a wild-type or permutated SRE found in the 5’- UTR of MUM2. P-values were determined by Mann-Whitney U test. Bars represent means ± SEM from three independent biological replicates.

We benchmarked these data against previous studies of yeast meiosis (e.g. Chu et al., 1998). In both [smaug⁻] and [SMAUG⁺] cells, a large number of transcripts increased more than four-fold (Figure S4B). Many overlapped (279 messages, p-value = 1.25 × 10⁻²⁹⁸ by hypergeometric test, Table S1), especially those expressed early in meiosis (42 early transcripts vs. 17 late transcripts, p-value = 0.0032, Fisher’s exact test; Table S1). Early transcripts were also more inducible in [smaug⁻] cells than in [SMAUG⁺] cells (Table S1, p-value = 0.0004, bootstraped t-test).

We next examined quantitative differences in induction. Most differentially expressed messages were induced to a similar degree in both samples (Figure 5B). However, transcripts with annotated meiotic functions (Saccharomyces Genome Database [SGD]; yeastgenome.org; Cherry et al., 2012) were induced to a greater extent in the [smaug⁻] cells than in [SMAUG⁺] cells (Figure 5B). The affected transcripts included the master regulator of early steps of meiosis, IME1, as well as several other key players in early meiosis (e.g. SAE3, DMC1 and HED1; Table S1; Primig et al., 2000). Thus, [SMAUG⁺] exerts a strong repressive effect on the post-transcriptional landscape of meiosis.

Enhanced degradation of a specific transcript drives phenotype

We investigated if [SMAUG⁺]’s impact on meiosis was a consequence of altered regulation of its targets, evaluating the expression of 195 transcripts annotated in SGD as having a function in meiosis (yeastgenome.org; Cherry et al., 2012). Seventy-six transcripts were significantly altered in [smaug⁻] cells, and a handful of these were differentially regulated in [SMAUG⁺] cells (Figure 5C, Table S2). The most differentially regulated, MUM2, was ~8-fold less abundant in [SMAUG⁺] cells than in [smaug⁻] cells.

MUM2 stood out for several more reasons. First, it is robustly upregulated in vts1Δ cells (Aviv et al., 2006; She et al., 2017). Second, it harbors an SRE element that directly binds to Vts1 (Figure 5D; She et al., 2017). Third, mum2Δ cells do not readily sporulate, giving rise to the gene’s name: muddled in meiosis (Agarwala et al., 2012). Mum2 is a conserved subunit of the MIS (Mum2-Ime4-Slz1) complex, which catalyzes m⁶A methylation of mRNA. In budding yeast, this epitranscriptomic modification increases translation efficiency and is necessary for meiotic progression (Agarwala et al., 2012; Wang et al., 2015).

We measured MUM2 expression by RT-qPCR during meiotic induction. Fourteen hours after transfer to sporulation medium, MUM2 was still repressed in uncured [SMAUG⁺] cells (Figure 5D). However, 24 hours after transfer to sporulation medium, MUM2 expression was comparable between in uncured and cured cells (Figure 5D). SPS1, a well-defined sporulation marker that lacks an SRE, was not differentially expressed at early timepoints. However, its levels were lower in uncured cells at a later time point, likely reflecting the delayed meiotic progression caused by the prion (Figure S4C; Chu et al., 1998).

The early reduction in MUM2 expression in [SMAUG⁺] cells echoed the delayed sporulation linked to this prion. We tested whether enhanced MUM2 expression could counteract the delayed sporulation of [SMAUG⁺] cells, introducing [SMAUG⁺] into strains in which MUM2 was expressed at its endogenous locus, either under the control of its native promoter or a strong constitutive promoter (NOP1; Yofe et al., 2016). [SMAUG⁺] cells expressing MUM2 from its endogenous promoter sporulated poorly. By contrast, constitutive overproduction of MUM2 rescued sporulation (Figure 5E).

We next investigated whether Vts1 binding to MUM2 is necessary for [SMAUG⁺]’s impact on meiosis. We permuted two key nucleotides MUM2’s SRE using a CRISPR/Cas9 system (Anand, 2017; Aviv et al., 2006; Aviv et al., 2003; She et al., 2017). We then scored sporulation in cured and uncured strains harboring either this mutation or a wild type MUM2 sequence. When MUM2’s SRE was mutated in uncured BY4743, sporulation efficiency increased significantly, and was similar to values we observed in cured strains (Figure 5F). By contrast, mutation MUM2’s SRE did not affect sporulation in cells that did not harbor the prion. Combined with our overexpression experiments, these data establish that [SMAUG⁺] controls meiotic commitment in large part via enhanced repression of MUM2.

Lastly, we asked if degradation of MUM2 also mediated the mitotic growth advantage the observed in [SMAUG⁺] cells. If this were the case, deletion of MUM2 in a cured strain, mimicking repression by [SMAUG⁺], should increase growth rate. Indeed, we observed a 1.5-fold increase in the growth rate of cured mum2Δ cells (Figure S4D). MUM2 overexpression had the opposite effect (Figure S4E). These data establish that changes in MUM2 expression can account for the impact of [SMAUG⁺] on mitotic proliferation and meiotic differentiation alike.

Repeated occurrence of an element resembling [SMAUG⁺] in nature

Finally we investigated the breadth of [SMAUG⁺] in nature, examining 26 natural isolates from the Saccharomyces Genome Resequencing Project (Liti et al., 2009), collected from diverse ecological niches. The URA3 gene has been deleted in these isolates (Cubillos et al., 2009), allowing us to cure them of prions by transient expression of a plasmid-borne dominant-negative Hsp70, as described above. Of the 26 strains, 21 yielded successful transformants, enabling us to obtain at least three independent curing events per isolate.

To assay Vts1 activity, we employed the plasmid-borne galactose-inducible GFP-SRE reporter construct described above. Nine of the 21 cured strains exhibited a significant increase in GFP fluorescence, ranging from 15% to 250% (Figure 6A, S5A and S5B). Eight of the remaining cured strains exhibited no significant change in GFP fluorescence (Figure S5C), and three exhibited very modest decreases (Figure S5D). We next transformed the nine cured isolates that exhibited an increase in fluorescence with a GFP-SRE- construct with abolished Vts1 binding. Eight of these strains showed no difference in GFP-SRE- fluorescence upon curing (Figure S6A). Thus, many wild S. cerevisiae strains harbor prion-like elements with the characteristics of [SMAUG⁺].

Figure 6. Natural S. cerevisiae strains harbor [SMAUG⁺]-like elements that impact sporulation.

(A) (Left) Heat map of the ratio of the mean GFP intensity in cured vs. uncured natural strains. BY4743 and [smaug⁻] are included as benchmarks. (Right) Violin plots of GFP fluorescence of cured and uncured SGRP strains expressing a GFP-SRE reporter for Vts1 activity. Dotted bars indicate upper and lower quartiles, and solid bars indicates medians. Data from strains with a significant increase in fluorescence upon chaperone curing are shown. P-values are determined by Mann-Whitney U test. N=106–936 cells. (B) Phylogenetic tree of the 21 successfully cured SGRP strains (Liti et al., 2009) denoting the effect of curing on GFP-SRE expression and sporulation.

We next asked whether curing also affected the sporulation of these natural isolates. One strain, isolated from soil, exhibited no change in sporulation when cured (Figure S6B and S6C). Two others, isolated from barrel fermentation and grapes, did not sporulate within 48 hours, but did sporulate to the same degree after 120 hours (Figure S6C). However, five of the eight [SMAUG⁺] strains sporulated more efficiently after curing, ranging from a two- to seven-fold increase (Figure 6B and S6A). These data suggest that [SMAUG⁺], or a substantially equivalent element, arises commonly in both laboratory and natural isolates of S. cerevisiae, and exerts a strong influence on the meiotic developmental program.

Polymorphic [SMAUG⁺] variants in nature

The increased strength of the induced [SMAUG⁺] variant led us to speculate that this prion might form distinct, but stable, activity states, also known as ‘strains’ or ‘polymorphs’. Strain-to-strain variation in curing-dependent expression of the GFP-SRE reporter (Figure 6A) led us to further hypothesize that natural yeast isolates could harbor distinct conformational variants of [SMAUG⁺]. If true, these variants might template the Vts1 protein in different ways. First, we confirmed that lysates from the naturally [SMAUG⁺] BY4743 laboratory strain could seed the condensation of fluorescently-labeled, purified Vts1 protein (Figure 7A; Chakravarty et al., submitted; Figure S7A). In contrast, lysates from selectively cured BY4743 did not seed Vts1 condensates (Figure S7A).

Figure 7. Natural S. cerevisiae strains harbor distinct variants of [SMAUG⁺].

(A) Experimental schema. Lysates from natural S. cerevisiae were added to purified, fluorescently-labeled Vts1. After incubation, the samples were imaged to monitor condensate formation. (B) Representative images of fluorescently labeled Vts1 seeded with lysates from natural strains. A natural [smaug⁻] is included as a control. Scale bar is 5 μm. (C) Condensate formation (quantified by average pixel intensity) vs. the strength of the [SMAUG⁺] activity (from Figure 6A). Blue dots indicate strains that harbor [SMAUG⁺]. Dashed line represents the linear regression line (slope = 0.01 ± 0.003; p = 0.006).

Next, we tested whether lysates of natural yeast isolates exhibiting [SMAUG⁺]-like characteristics could also seed Vts1 condensation. Among these lysates, we observed a range of assembly properties. Lysates from strains isolated from oak bark (YPS606) and stingless bee (UW0PS05–227.2) did not promote formation of large condensates. These strains may either have slower assembly kinetics or express a trans factor that inhibits visible Vts1 condensate formation. Two lysates (from cactus and cactus fruit isolates; USOPS87–2421, USOPS83–787.3) formed condensates that closely resembled each other in size and intensity (Figure 7B). A third lysate (from palm flower nectar isolates; UW0PS03–461.4) seeded assemblies that resembled those generated from BY4743 lysate (Figure 7B and S9A). Lysates from DBVPG6765, a natural strain that did not exhibit a curing-dependent increase in GFP-SRE expression (‘natural control’), did not drive Vts1 condensation (Figure 7B).

Vts1 condensates are readily discernible due to their brightness relative to unassembled protein. As a proxy for condensate formation, we therefore measured the average pixel intensity in each sample. This value correlated with the degree of [SMAUG⁺]-mediated repression of GFP-SRE expression (Figure 7C; p=0.006), suggesting that the activity of a [SMAUG⁺] variant can be predicted by its patterns of self-assembly. VTS1 transcript levels were similar among these natural strains; any modest differences did not correlate with seeding capacity or [SMAUG⁺]-driven phenotypes (Figure S7B). These data establish that natural isolates of S. cerevisiae not only harbor [SMAUG⁺], but also exhibit quantitative differences in the strength of [SMAUG⁺] traits. We conclude that [SMAUG⁺] exists as distinct variants in nature, tuning a fundamental cellular decision.

DISCUSSION

Adaptive anticipation of future environments has been theorized and observed in the laboratory (Mitchell et al., 2009). Yet, our understanding of the molecular mechanisms that govern such behavior in nature remains limited. Using S. cerevisiae as a model, we took advantage of the nutrient dependence of gametogenesis to simulate adaptive prediction in fluctuating environments. These simulations suggested that altering the sensitivity of relationship between starvation and meiosis, via an epigenetic mechanism that is heritable yet reversible, would offer significant adaptive benefit. Guided by these inferences, we identified a heritable prion conformation of an evolutionary ancient RNA binding protein, [SMAUG⁺], that meets these criteria. This prion provides a mechanism for capturing information about the quality of environmental fluctuations and transmitting it to future generations, adaptively modulating the cell fate decisions between mitosis and gametogenesis.

This form of epigenetic regulation is pervasive. [SMAUG⁺] occurs in most laboratory strains of S. cerevisiae, in which common husbandry conditions would favor its enrichment. [SMAUG⁺] is also common in natural S. cerevisiae isolates from diverse ecological niches, where it strongly impacts sporulation efficiency. Thus, [SMAUG⁺] may have repeatedly influenced S. cerevisiae’s ability to adapt to new environments. The absence of [SMAUG⁺] in most SGRP strains suggests that the limited handling of these strains has not itself induced [SMAUG⁺]. These discoveries force a reevaluation of the prevalence and importance of prions. Once thought to be obscure exceptions to the central dogma, these heritable protein conformations may in fact be common in biology, capable of adaptively modulating developmental strategies.

Due to its ubiquity in S. cerevisiae, [SMAUG⁺] is a major contributor to variation in sporulation observed across this species. Our findings have strong implications for functional genomics: between S. cerevisiae strains with a degree of polymorphism similar to two humans, up to ~38% of phenotypic variance is unexplained by genetics (Bloom et al., 2015). [SMAUG⁺] highlights the strong influence that unexplored forms of epigenetics can have on phenotypes of fundamental biological importance. Indeed, [SMAUG⁺] potential interactions between [SMAUG⁺] and other types of protein aggregates that regulate meiosis (e.g. Rim4) raise exciting questions for future study.

The well-studied prion [PSI⁺] can exist as multiple stable conformational variants, each with a different phenotypic strength (Bateman and Wickner, 2013; Tanaka et al., 2004; Tanaka et al., 2006; Toyama et al., 2007). However, because ‘wild’ [PSI⁺] variants are of similar strengths (Halfmann et al., 2012), it has remained unknown whether such variation is exploited in nature. Here, we found that [SMAUG⁺] adopts multiple functionally distinct variants in the wild. Given the adaptive advantage provided by the prion, these variants could be fine-tuned to the frequency of fluctuations inherent to a niche. Together, along with the hyperactivity of [SMAUG⁺] (Chakravarty et al., submitted), these data highlight an emergent dimension of protein structure and function has been largely unappreciated to date, and may be regularly utilized in nature.

In Drosophila, the homolog of Vts1, Smaug, is essential for early development, governing the degradation of transcripts during maternal-to-zygotic transition (Tadros et al., 2007). Here we show that Vts1 controls another developmental process, gametogenesis, in S. cerevisiae. The natural [SMAUG⁺] prion degrades its target transcripts at an elevated rate. Our findings reveal a key biological role for this hyperactivity: repression of transcripts involved early meiosis. The ensuing delay in sporulation is mediated by degradation of the message encoding the mRNA methylase subunit MUM2, a target of Vts1, and itself a critical post-transcriptional regulatory node in meiosis (Agarwala et al., 2012). More than 300 other SREs have been identified in yeast, including in genes unrelated to meiosis (She et al., 2017). Thus, sporulation could be one of many phenotypes regulated by [SMAUG⁺].

Prion induction and loss is often influenced by the environment (reviewed in Harvey et al., 2018). Based on theory (Jablonka et al., 1995; Kussel and Leibler, 2005), one would predict that [SMAUG⁺] could be induced, in a quasi-Lamarckian fashion, by the same conditions in which it is advantageous. Notably in this regard, across the thousands of conditions in which gene expression has been measured in yeast, VTS1 is most upregulated in ‘return to growth’ assays – conditions when S. cerevisiae cells experience nutrient replenishment before irreversibly committing to meiosis (Friedlander et al., 2006; Figure 5A). This natural upregulation would be expected to induce [SMAUG⁺] (by virtue of mass action) as a function of the reproducibility of the environmental fluctuation: as the population experiences regular cycles of nutrient replenishment, more cells would convert to and benefit from [SMAUG⁺]. Indeed, exogenous upregulation is regularly used as a means of robustly inducing prion formation (Chakrabortee et al., 2016; Masison and Wickner, 1995; Wickner, 1994; Wickner et al., 2006). Characterizing how rates of [SMAUG⁺] gain and loss are tuned in different fluctuating environments represents an exciting avenue for future research.

In sum, we have demonstrated that the [SMAUG⁺] prion, pervasive in both nature and the laboratory, serves as a ubiquitous means of adaptive transgenerational memory in S. cerevisiae. [SMAUG⁺] allows S. cerevisiae cells to make ‘wagers’ that were beneficial to their ancestors to anticipate future environmental fluctuations, substantially influencing their growth and differentiation strategies. In addition, this prion regularly explores its conformational space in natural populations, encoding distinct stable polymorphs that amplify the phenotypic diversification sparked by this protein-based genetic element. [SMAUG⁺] controls some of the best studied traits in one of the most widely used model organisms in experimental science. It is present in genetic backgrounds that have been subject to thousands of experiments by laboratories around the world. Yet targeted and ‘omics’ technologies have been blind to its presence and importance. Together our findings portend a much larger and mechanistically diverse ‘prionome’ that is hidden in plain sight, and remains to be unveiled.

STAR METHODS

LEAD CONTACT AND MATERIALS AVAILABILITY

Correspondence and requests for materials should be addressed to Lead Contact Daniel F. Jarosz (danjarosz.aa@gmail.com). All unique reagents generated in this study are available from the Lead Contact without restriction.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

S. cerevisiae strains were obtained from the sources indicated (Key Resources Table). Natural S. cerevisiae isolates were obtained from the Saccharomyces Genome Resequencing Project (SGRP) Strain Set 2 and used as founder strains (Cubillos et al., 2009; https://catalogue.ncyc.co.uk/sgrp-sets). These strains have URA3 deleted.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Critical Commercial Assays
KAPA SYBR Fast qPCR	kapabiosystems.com	Cat# KK4601
TURBO DNA-freeTM Kit	Invitrogen	AM1907
Oligonucleotides
Oligo dT-20 primer	Invitrogen	Cat#18418–020
HO integration check reverse primer: GAG GGTATTCTG G GCCTCCATG	This paper	N/A
HO integration 3’ junction check forward primer: CCTGT AGAAGAGT A1 1 1GTTGT	This paper	N/A
HO integration 3’ junction check forward primer: GAAGCTAAGATTGGTAAGAAG	This paper	N/A
Vts1 forward primer: 1 1 IATGCTAGACACTGCGGG	This paper	N/A
Vts1 reverse primer: AGCTGCTGTAGACGAAACTG	This paper	N/A
TAF10 forward primer: CGTGCAGCAGATTT CACAAC ()	This paper	N/A
TAF10 reverse primer: CGACCTATATTGAGCCCGTATTC	This paper	N/A
MUM2 forward primer TGGATCCCGAGTCATGTTTG	This paper	N/A
MUM2 reverse primer AAGCACCTGGGTCTGATTG	This paper	N/A
SPS1 forward primer GATCGGCCATCGGCATATAA	This paper	N/A
SPS1 reverse primer TGTGCTTAGTCTTGTCGATTCA	This paper	N/A
F primer for generation of pJH2972 with MUM2 gDNA TTATGATCCCCAACATTCGTG MM	This paper	N/A
R primer for generation of pJH2972 with MUM2 gDNA ACGAATGTTGGGGATCATAAGATCA	This paper	N/A
Wild-type SRE homology template: GAAATGTAAACGATAGTCAAACTTAGTAAGAAGAGAATCAGCTGGCTTGGTACTTGATCTTTTGATTTGACCTCATTCTTTTTGCAT ACACGGCTCGTTTGGAATACTGTTGTAAAAATGAATTACATGGCTTATGACTACGACCCCCAGCATTCCTTGGAAACGTCCTTTAACAATTTGGCATTTCATCCCCACCAACAGTCACAG	This paper	N/A
Permuted SRE homology template: GAAATGTAAACGATAGTCAAACTTAGTAAGAAGAGAATCAGGTCGCTTGGTACTTGATCTTTTGATTTGACCTCATTCTTTTTGCAT ACACGGCTCGTTTGGAATACTGTTGTAAAAATGAATTACATGGCTTATGACTACGACCCCCAGCATTCCTTGGAAACGTCCTTTAACAATTTGGCATTTCATCCCCACCAACAGTCACAG	This paper	N/A
Experimental Models: Organisms/Strains
BY4741 MATa	Winston et al., 1995	N/A
BY4742 MATα	Winston et al., 1995	N/A
BY4742 kar1–15	Conde and Fink, 1976	N/A
BY4742 vts1Δ	Giaever et al., 2002	N/A
Natural S. cerevisiae isolates	Liti et al., 2009	https://catalogue.ncyc.co.uk/sgrp-sets
BY4743	Winston et al., 1995 Dharmacon.horizondiscovery.com	Catalog # YSC1050
Recombinant DNA
Hsp70 (K69M) plasmid	Jarosz et al., 2014a	N/A
Hsp70 (K69M) plasmid (Ura-selectable)	This paper	N/A
HO integration mKate2	Wong et al., 2018; Zalatan et al., 2012	N/A
HO integration mNeonGreen	Wong et al., 2018; Zalatan et al., 2012	N/A
Advanced Gateway destination vectors	Alberti et al., 2007	https://www.addgene.org/yeast-gateway/
pJH2972 CRISPR/Cas9 construct	Anand, 2017	https://www.addgene.org/100956/
Software and Algorithms
Matlab	MathWorks, Inc.	https://www.mathworks.com/products/matlab.html
ImageJ	NIH	https://imagej.nih.gov/ij/
Prism	GraphPad Software Inc.	https://www.graphpad.com/
Cell Profiler	Kamentsky et al., 2011	https://cellprofiler.org/
FlowJo	FlowJo, LLC	https://www.flowjo.com/
DeSeq2	Love et al., 2014	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
Leica LAS X Core Image Analysis Software	Leica, Inc.	http://www.leica-microsystems.com/products/microscope-software/
Deposited data
RNA sequencing	This paper	GSE138559
Mendeley Data	This paper	10.17632/kkst3822v4.1

All S. cerevisiae strains were stored as glycerol stocks at −80°C. Before use, strains were revived on YPD. Cultures were grown at 30°C. For growth measurements, cells were inoculated from solid YPD into complete synthetic media containing 2% galactose for pre-growth. Cells were then normalized to OD, = 0.1 and 1 μl was inoculated into the media of interest.

For plasmid transformations, a standard lithium-acetate protocol was used (Gietz et al., 1992). Cells were inoculated into YPD and grown to saturation overnight. Cells were harvested, washed in sterile water, and suspended in a transformation mix (120 μl of 50% w/v PEG 3500, 17 μl of 1M lithium acetate, 24 μl of boiled salmon sperm carrier DNA (2mg/ml), and 17 μl of H₂O containing up to 1μg of plasmid). Cells were incubated at 30°C for 30 minutes, and then at 42°C for 20 minutes, before washing, pelleting, and plating on selective media.

Integration transformations were performed using a standard electroporation protocol (Thompson et al., 1998). Cells were inoculated from solid media into 5 ml of YPD and grown to saturation overnight. The culture was then diluted in 25 ml of YPD and grown to an OD of ~0.5. Cells were then pelleted and suspended in 9 ml of TE buffer and 1 ml of 1 M lithium acetate, and incubated for 45 min at 30°C, mixed with 250 μl of 1 M DTT was added, and cells were incubated for an additional 15 min. Cells were then washed twice with water and once with sorbitol, and then resuspended in 60 μl of cold 1 M sorbitol. Salmon sperm DNA (1.7 μl) and cassette DNA (1μg total) was added to 40 μl of cell mixture, and the mixture was then subjected to electroporation (0.2 cm gap cuvette, 1.5 kV, 24 μF, 200 ohm). Cells were recovered overnight in YPD before plating on selective medium.

METHOD DETAILS

Modeling

In a nutrient-rich environment t₁, both populations undergo exponential growth:

$$x_{\mathit{\text{sporulator}}}=x_{{\mathit{\text{sporulator}}}_{\mathit{\text{initial}}}}{e}^{{\mu }_{\mathit{\text{sporulator}}}{t}_1}$$ (1)

$$x_{\mathit{\text{proliferator}}}=x_{{\mathit{\text{proliferator}}}_{\mathit{\text{initial}}}}{e}^{{\mu }_{\mathit{\text{proliferator}}}{t}_1}$$ (2)

x_sporulator is the fraction of the population that are sporulators; x_proliferator is the fraction of the population that are proliferators. For the first cycle (N = 1), the initial fraction is always 0.5 for each developmental strategy, unless otherwise specified. μ_sporulator and μ_proliferator are the growth rates for proliferators and sporulators, respectively.

In a nutrient-poor environment, t₂, both populations enter the sporulation program. The fraction of the population that form stress-resistant spores is determined by a Hill function with respect of time in nutrient poor environments. A Hill function allows different strains to have different sporulation kinetics while maintaining the same maximum sporulation efficiency – two features we observed in experiments (Figure 2B).

$${\mathit{\text{spores}}}_{\mathit{\text{sporulator}}}=x_{\mathit{\text{sporulator}}}\frac{t_2{}^{k_{2_{\mathit{\text{sporulator}}}}}}{k_{1_{\mathit{\text{sporulator}}}}{}^{k_{2_{\mathit{\text{sporulator}}}}}+t_2{}^{k_{2_{\mathit{\text{sporulator}}}}}}$$ (3)

$${\mathit{\text{spores}}}_{\mathit{\text{proliferator}}}=x_{\mathit{\text{proliferator}}}\frac{t_2{}^{k_{2_{\mathit{\text{proliferator}}}}}}{k_{1_{\mathit{\text{proliferator}}}}{}^{k_{2_{\mathit{\text{proliferator}}}}}+t_2{}^{k_{2_{\mathit{\text{proliferator}}}}}}$$ (4)

spores_sporulator and spores_proliferator are the fractions of the sporulator and proliferators populations, respectively, populations that form spores. k₁ and k₂ are coefficients obtained by fitting a Hill function to the sporulation dynamics observed in Figure 2B.

We define the survival probabilities during the die-off event as being equal for the haploids of sporulators and proliferators.

$$x_{\mathit{\text{sporulator}}}=p_{\mathit{\text{spores}}}{\mathit{\text{spores}}}_{\mathit{\text{sporulator}}}+p_{\mathit{\text{diploids}}}\left(x_{\mathit{\text{sporulator}}}-{\mathit{\text{spores}}}_{\mathit{\text{sporulator}}}\right)$$ (5)

$$x_{\mathit{\text{proliferator}}}=p_{\mathit{\text{spores}}}{\mathit{\text{spores}}}_{\mathit{\text{proliferator}}}+p_{\mathit{\text{diploids}}}\left(x_{\mathit{\text{proliferator}}}-{\mathit{\text{spores}}}_{\mathit{\text{proliferator}}}\right)$$ (6)

After the die-off event, we are left with the resulting populations of individuals for each developmental strategy. This is the starting population for the subsequent cycle. The cycle continues for N nutrient rich, nutrient poor, and die-off phases. For all plots, unless otherwise specified, the simulation was run for N = 10 cycles.

For physiological parameter estimates we used the following observations: 1) relative growth rates for advantageous gene deletions can be as large as μ ≅ 1.1 (Breslow et al., 2008); 2) spores can be ~ 100-fold more resistant to environmental stressors relative to unsporulated diploids (Coluccio et al., 2008); 3) in populations of S. cerevisiae, sporulation occurs over tens to hundreds of hours (Gerke et al., 2006; Mortimer, 2000).

Curing

Strains were cured by transforming an URA3-selectable 2μ plasmid encoding a dominant-negative HSP70 (Ssa1-K69M; Chakrabortee et al., 2016; Garcia et al., 2016; Lagaudriere-Gesbert et al., 2002) under the control of a constitutive promoter (pGPD). Transformants were passaged on selective media five times to allow growth of single colonies. Next, transformants were passaged three times on non-selective media (YPD) to allow for plasmid loss, which was confirmed by the lack of growth on selective media (SD-Ura).

Cytoduction was performed as described previously (Chakrabortee et al., 2016) Strains harboring the kar1–15 allele, which are deficient in nuclear fusion, were converted to petites by growing in YPD + 0.25% ethidium bromide, and then recovered on YPD agar. Respiration incompetence was confirmed by lack of growth on YP-glycerol. Cured and uncured BY4741 (donor strains) were mated to a cured α strain harboring the kar1–15 allele (recipient strain). Heterokaryons were isolated by selecting for strains that were respiration competent and for a MATα auxotrophic marker. Cytoductants were passaged again on SD-Met, and then validated to be haploid. Reverse cytoductions were performed by mating the above cytoductants to cured BY4741 petites, and then selected for mitochondrial respiration and auxotrophic growth on lysine, and validated to be haploids. These cytoductants were mated to cured strains to generate diploids.

Sporulation phenotyping

Strains were inoculated from single colonies on fresh plates into pre-sporulation media (0.8% Yeast extract, 0.3% Bacto-peptone, 0.01% adenine sulfate, 10% dextrose) and grown with rotation at 30°C for 48 hours. At this point, cells were spun down at 1500 g for 3 minutes, washed in H₂O, pelleted, and resuspended in sporulation media (1% potassium acetate, 0.1% yeast extract, 0.05% dextrose, 0.01% amino acid add-back (1:1:5 uracil, histidine, leucine). Cultures were kept on a rotating wheel at 25°C. Unless otherwise specified, cultures were sporulated for 120 hours. DIC imaging was performed at a magnification of 63× on a Leica DMI6000 microscope; two to three images were acquired per sample. For each image, ~100 cells were counted, and spores were scored to calculate the sporulated fraction.

For natural isolates, growth and sporulation was performed in 96 well plates. Plates were Aero sealed and shaken at 500 rpm. Sporulation was scored after 48 hours unless otherwise specified. All other conditions were identical to those for cells sporulated in 5 ml culture tubes (as described above).

RT-qPCR

Five cultures were grown and transferred to sporulation media. At each time point, 1 ml of culture was pelleted and washed with H2O, and pelleted again before flash freezing. RNA was extracted using a standard phenol–chloroform procedure (Collart and Oliviero, 2001). RNA concentration was normalized based on 260 nm/280 nm ratios, and then used to synthesize cDNA using oligo-T primers. qPCR was performed using KAPA SYBR FAST qPCR master mix (Kapa Biosystems). Primers against the housekeeping gene TAF10 were used as controls for relative quantification (Teste et al., 2009). Primer sequences used for probing MUM2 and SPS1 are listed in the Key Resources Table. For measurement of relative VTS1 levels across wild strains, 5 ml of mid-exponential cultures were processed through a near-identical workflow. Primers used for probing relative VTS1 abundance are listed in the Key Resources Table

Library preparation, RNA sequencing, and analyses

For all samples, RNA was extracted from diploid yeast strains (BY4743 and BY4743 selectively cured of [SMAUG⁺]; two biological replicates at each time point for each strain) before (0 h) and after initiation of the sporulation program (14 h) using a standard phenol–chloroform protocol. Libraries were generated using standard kits (NEBNext® Ultra™ II RNA library prep kit Cat# E7770S, poly-A enrichment). Further quality control and quantification was performed by analysis on an Agilent 2100 Bioanalyzer and RT-qPCR (SYBR staining). All samples were sequenced to a read depth of ~10,000,000 (1 × 75 bp) on a single flow cell of an Illumina NextSeq 550™. Quality control of reads was performed using FastQC (Babraham Institute). Reads were de-duplicated and pseudo-aligned against the Saccharomyces cerevisiae strain S288C reference genome assembly R64, and then transcript-level quantification was performed using Kallisto. Differential expression analyses were performed using the DESeq2 package in R. Principal component analyses of most differentially regulated genes and significantly altered transcripts were performed using default settings in the DESeq2 package, which employs the Benjamini-Hochberg approach for approximating the false discovery rate (FDR). RNA-seq data is deposited into Gene Expression Omnibus (GEO; GSE138559).

Protein transformation

Mid-exponential cultures of selectively cured [smaug⁻] haploid strains were transformed with Vts1 condensates generated from purified protein or with BSA (as a control), following a workflow similar to that described in the accompanying manuscript (Chakravarty et al. submitted). In brief, cell pellets were washed and spheroplasted with Zymolyase®−100T (Sunrise Science Products, Cat# 0766555) in SCE buffer (1 M sorbitol, 10 mM EDTA, 10 mM DTT, 100 mM Na-citrate pH 5.8). Following exposure to protein of interest and a co-transformed LEU-selectable plasmid, these spheroplasts were collected and resuspended in 250 μl of SOS buffer (1 M sorbitol, 7 mM CaCl₂, 0.25% yeast extract, 0.5% Bacto-peptone). To avoid disrupting the cell membrane, all manipulations were performed using cut pipette tips with widened apertures. This mixture was incubated at 30°C for 3 h, after which these cells were plated on solid medium (SD-Leu) that was supplemented with 1.2 M sorbitol. Following plating, these cells were overlaid with soft agar (0.8% agar) of an otherwise identical composition, and the plates were incubated at 30°C for 3–5 days. Individual transformants for both sets of transformations were picked and passaged on SD-Leu media (for ~75 generations), mated to a cured strain of opposite mating type, and then subjected to sporulation measurements using workflows described above.

Yeast cell lysate extraction and seeding

Cells from yeast strains in the SGRP collection were harvested from 100 ml cultures in mid-exponential phase by centrifuging at 5000 g for 2 min at 4°C. All subsequent steps were performed at or below 4°C, as stated, and the method for native yeast cell lysis was nearly identical to that described in the accompanying manuscript (Chakravarty et al. 2019). In brief, cells were washed twice with wash buffer (50 mM HEPES-NaOH [pH 7.6], 2 mM EDTA, 0.8 M sorbitol, 300 mM sodium glutamate, 3 mM DTT added fresh immediately before use) and resuspended in ½ packed cell volume of lysis buffer (100 mM HEPES-NaOH [pH 7.6], 0.8 M sorbitol, 950 mM sodium glutamate, 10 mM magnesium acetate, 5 mM DTT, and one Complete Protease Inhibitor Cocktail tablet (Roche Cat# 11836145001; per 5 ml) added fresh immediately before use). This resuspensate was then frozen dropwise in liquid nitrogen, and the resultant beads were stored at −80°C until needed. Yeast cell beads were loaded into a steel vial pre-chilled at −80°C and processed using a CryoMill (Retsch) using the following program sequence: 1.5 min precooling at 5 Hz, 9 cycles of 2 min each at 15 Hz, 30 sec of gap at 5 Hz between each cycle. Next, lysis buffer was added to the mixture resulting from thawing the powder in a chilled conical tube (1/3^rd of the total volume), and the sample was transferred to an Eppendorf tube. This lysate was then centrifuged to remove cell debris, and the resultant supernatant was dialyzed (using a 10 kDa MWCO membrane) against dialysis buffer (50 mM HEPES-NaOH [pH 7.6], 2 mM EDTA, 10 % (v/v) glycerol, 300 mM sodium glutamate, 5 mM magnesium acetate, 3 mM DTT added fresh immediately before use). Total protein concentration was measured by A₅₉₅ using the Bradford assay reagent, and aliquots of the lysates were stored at −80°C. In vitro seeding experiments using these lysates were performed at room temperature for 4 h; the molar ratio of Vts1 from lysate to purified Vts1 was ~1:163 (Ghaemmaghami et al., 2003). Imaging was performed using standard epifluorescence microscopy (Leica DMI6000) using Cy3 filter blocks.

Competition assays

Strains were transformed using an integrating plasmid that incorporated a fluorescent protein (either mNeonGreen [Neon] or mKate2 [Kate]) under the control of a strong constitutive promoter (TDH3) and a hygromycin-selectable marker into the HO locus (Wong et al., 2018; Zalatan et al., 2012). Transformants were selected on YPD + 200 μg/L hygromycin B. To check for proper integration of the integration cassette, PCR was performed on hygromycin-resistant transformants using a cassette nested primer and a flanking primer hybridizing within the HO locus. Fluorescent protein expression was checked on a confocal microscope (mNeonGreen: excitation, 450–490 nm, emission, 500–550 nm; mKate2, excitation 630–640 nm, emission, 690–750 nm). Growth curves of each PCR-confirmed, fluorescent transformant were generated to ensure that exogenous fluorescent protein expression did not result in an obvious growth defect.

Neon- and Kate-expressing strains were pre-grown for 48 hours in complete synthetic media containing 2% galactose and 200 μg/ml hygromycin B. Strains were diluted to an OD of 0.1. Competitions were carried out by mixing a Neon strain and a Kate strain in a 1:1 ratio, and inoculating 1 μl of this mixture into 150 μl of media of interest. Inoculated cultures were grown at 30°C for 24 or 48 hours. At this point, they were diluted to an OD of 0.1 using H₂O. One microliter of this diluted culture was used to inoculate fresh media for the subsequent growth step. The remainder of the diluted culture was fixed using 4% paraformaldehyde for 15 minutes and stored in 1.2 M sorbitol + 0.1 M potassium phosphate at 4°C until flow cytometry analysis. Flow cytometry was performed on a BD LSR II, exciting at 488 nm for Neon and 561 nm for Kate. A total of 10,000 events per sample was collected. Cultures were passaged 5 times total.

To correct for any tag-specific growth effect, control competitions containing the same strains marked with different fluorescent proteins (i.e. cured BY4743 expressing Kate competed with cured BY4743 expressing Neon) were also performed. Any difference in growth observed in these competitions was treated as a tag-specific growth effect and normalized out in analyses of strain-specific growth differences. Selection coefficient was calculated from the slope of a line fitted to the natural log-transformed data (Chevin, 2011).

Vts1 activity

To assay Vts1 activity, we used a construct a galactose-inducible GFP followed by three tandem SREs. In the permuted version, the three tandem SREs were mutated to abolish Vts1 binding. Cells containing the construct were pre-grown in dropout media with raffinose as the carbon source. For BY4743, we used a GFP-SRE construct with a HIS3 selectable marker. For the SGRP collection, we used a GFP-SRE construct with a URA3 marker. After pre-growth for 48 hours, cultures were diluted to an OD of 0.1 and used to inoculate new cultures in dropout S-raffinose + 0.2% galactose to induce expression of GFP-3xSRE. Cells were grown to mid-exponential phase (OD = 0.5; ~12 h). Cells were imaged using a Leica inverted fluorescence microscope with a Hamamatsu Orca 4.0 camera. (GFP excitation: 450–490 nm; emission: 500–550 nm; software: LASX DMI6000B; refraction index: 1.518; aperture: 1.4; exposure time: 750 ms). DIC images of the same fields of view were also collected to define cell edges.

CRISPR/Cas9 mutation

We followed a protocol from the Haber laboratory (Anand, 2017). The guide DNA (TTATGATCCCCAACATTCGT) was designed in to target base pairs 21–40 of the MUM2 coding sequence, upstream of an endogenous PAM sequence, TGG. A homology template that introduced 5 synonymous changes to the guide DNA sequence (CTACGACCCCCAGCATTCCT) and either the wild-type or permuted SRE element was used.

Automated GFP fluorescence quantification

All analyses were performed using established pipelines in CellProfiler. Using the DIC image, cells were defined using the Sobel edge-finding method. Cells that did not have a form factor (4*π*area/perimeter²) greater than 0.5 and an area between 2000 and 10000 pixels (i.e., cells that were unsuccessfully recognized) were ignored. Next, integrated GFP fluorescence intensity was calculated for each successfully recognized cell.

QUANTIFICATION AND STATISTICAL ANALYSIS

Quantification and accompanying statistical tests for all experiments are described in the Results section and figure legends. The Mann-Whitney U test was used to compare measurements between two samples. When the means of more than two samples were being compared, a one-way test of variance was used followed by Sidak’s multiple comparison test. Fisher’s exact tests were used to compare overlap between sets of proteins and genes). p-values < 0.05 were interpreted as reflecting significant differences.

DATA AND CODE AVAILABILITY

All gene expression data collected are deposited in Gene Expression Omnibus (GEO) under accession number GSE138559. Representative microscopy images have been submitted to Mendeley (10.17632/kkst3822v4.1).

Supplementary Material

1

Table S1, Related to Figure 5B. Meiotic transcripts induced in [smaug⁻] and [SMAUG⁺] cells. List of upregulated transcripts during onset of meiosis in [smaug⁻] and [SMAUG⁺] cells.

2

Table S2, Related to Figure 5C. Meiotic transcripts are altered in [SMAUG⁺] cells. Ratio of induction of SGD annotated meiotic transcripts in [SMAUG⁺] cells with respect to [smaug⁻] cells.

• The [SMAUG⁺] prion allows yeast to anticipate nutrient repletion after starvation

• [SMAUG⁺] regulates the decision between cardinal growth and survival strategies

• [SMAUG⁺] is hidden in plain sight in common laboratory yeast strains

• Distinct [SMAUG⁺] variants are widespread in wild yeast populations