Aggregation of the Whi3 protein, not loss of heterochromatin, causes sterility in old yeast cells

See allHide authors and affiliations

Science  17 Mar 2017:
Vol. 355, Issue 6330, pp. 1184-1187
DOI: 10.1126/science.aaj2103

Protein aggregation–mediated aging in yeast

Old age in yeast cells results in insensitivity to mating pheromone. Reduced activity of the histone deactylase Sir2 and consequent alteration of chromatin at mating loci have been implicated in the decreased sensitivity of old cells. However, Schlissel et al. found a different mechanism in the yeast strains that they studied (see the Perspective by Gitler and Jarosz). Proper response to mating pheromone requires arrest of the cell cycle mediated by an RNA-binding protein, Whi3. If aggregation of Whi3 in old cells was inhibited by deletion of a glutamine-rich region that promotes aggregation, loss of sensitivity to mating pheromone was partially prevented, and replicative life span was slightly increased.

Science, this issue p. 1184; see also p. 1126


In yeast, heterochromatin silencing is reported to decline in aging mother cells, causing sterility in old cells. This process is thought to reflect a decrease in the activity of the NAD+ (oxidized nicotinamide adenine dinucleotide)–dependent deacetylase Sir2. We tested whether Sir2 becomes nonfunctional gradually or precipitously during aging. Unexpectedly, silencing of the heterochromatic HML and HMR loci was not lost during aging. Old cells could initiate a mating response; however, they were less sensitive to mating pheromone than were young cells because of age-dependent aggregation of Whi3, an RNA-binding protein controlling S-phase entry. Removing the polyglutamine domain of Whi3 restored the pheromone sensitivity of old cells. We propose that aging phenotypes previously attributed to loss of heterochromatin silencing are instead caused by aggregation of the Whi3 cell cycle regulator.

Budding yeast divide asymmetrically, and each yeast mother cell produces a finite number of daughter cells in its lifetime. This process—yeast replicative aging—has been studied for insights into aging more broadly, because the processes that underlie aging in yeast might be related to factors that underlie aging in other asymmetrically dividing cells (1).

In Saccharomyces cerevisiae, haploid mother cells lose the ability to mate as they age (2). It has been proposed that old mother cells fail to mate as a consequence of a decline in Sir2 function, which would cause loss of heterochromatic gene silencing of the auxiliary mating-type loci HML and HMR (3). Loss of silencing at HML and HMR in old cells has been attributed to the redistribution of Sir proteins to the nucleolus and to a decrease in available Sir2 (46). Furthermore, old cells may be either limited for the Sir2 substrate nicotinamide adenine dinucleotide (oxidized form; NAD+) or exposed to high concentrations of nicotinamide (NAM), an inhibitor of Sir2, resulting in the inactivation of Sir2 in old cells and thus sterility (79).

We characterized transcriptional repression by Sir2 by testing whether transient loss-of-silencing events at HML might precede the complete loss of silencing attributed to the oldest cells. To study silencing at HML in a yeast mother cell, we monitored pedigrees of haploid cells carrying a Cre-based silencing reporter (10). The reporter uses a Cre recombinase gene inserted in place of HMLα2 and a fluorescent reporter inserted at a euchromatic locus elsewhere in the genome (Fig. 1A). Loss of silencing at hmlα2∆::CRE induces a permanent and heritable switch from expressing red fluorescent protein (RFP) to expressing green fluorescent protein (GFP) (Fig. 1A), and the sensitivity of the Cre reporter approaches the sensitivity of single-molecule RNA fluorescent in situ hybridization (10). We manually separated daughter cells from their mothers to analyze pedigrees in two common strain backgrounds, S288c and W303, and observed no loss-of-silencing events in dozens of pedigrees of haploids, diploids, and hybrids (Fig. 1B).

Fig. 1 Loss of silencing is not a feature of yeast aging.

(A) Schematic of the CRASH (Cre-reported altered states of heterochromatin) reporter (10). (B) Representative pedigree of 20 sequential daughters from the W303 haploid strain (JRY10774) carrying the CRASH reporter. 1° signifies the colony arising from the first daughter cell, 2° signifies the colony arising from the second, and so forth. (C) Frames depict each division event in a typical pedigree in the S288C haploid strain background (JRY10772). (D) The top panel is a histogram of all cell division events that occurred for 223 pedigrees from a microfluidic experiment using JRY10772. Below, the age of cells at the time that they lost silencing is plotted as a histogram for the 13 pedigrees that lost silencing. Division index refers to the number of buds that each mother produced after the cells were loaded on the microfluidic chip, and the P value was calculated using the Kolmogorov-Smirnov test. (E) Frames depict consecutive daughters of a pedigree as in (C), and nicotinamide (NAM) was added to the medium after ~24 hours.

To measure the frequency of silencing loss as a function of a cell’s life span, we extended the pedigree analysis by using a microfluidic device that traps mother cell’s and separates their buds. We analyzed more than 1500 yeast pedigrees at single-cell resolution and observed 13 loss-of-silencing events (Fig. 1C and movie S1). Furthermore, we found that a cells age did not affect its ability to maintain silencing of HML, and the overwhelming majority of yeast mother cells stopped dividing without even a transient loss of silencing (Fig. 1D). As a control, when Sir2 activity in old cells was inhibited by addition of NAM, all surviving cells lost silencing, suggesting that old cells did not accumulate NAM in amounts that inactivate Sir2 (Fig. 1E and movie S2). Similar results were obtained by analyzing a GFP gene inserted in place of HMLα and using an alternative microfluidic design, indicating that the observation was independent of the reporter and the microfluidic setup used (fig. S1). Previous studies have shown that Sir2 protein levels decrease in cells that are more than seven generations old; however, we found no evidence of a decrease in Sir2 activity at HML (4). It is possible that a decrease in Sir2 levels in old cells could affect other Sir2 complexes, including the nucleolar RENT complex. Inactivating the RENT complex would decrease ribosomal DNA (rDNA) silencing and increase the occurrence of extrachromosomal rDNA circles in old cells, two phenotypes that have been repeatedly observed in old cells (1).

We also purified populations of aged MATα cells from liquid culture and analyzed them for expression of RNA from the HMR locus. Old cells cultured for ~20 generations showed low expression of HMRa1 mRNA relative to cells cultured in the presence of the Sir2 inhibitor NAM, establishing that Sir2-dependent silencing of auxiliary mating-type loci was functional in old cells (Fig. 2, A and B). In addition to analyzing silenced RNA from HMR, we analyzed expression of STE3, which encodes the a-factor pheromone receptor, is expressed in MATα cells, and is repressed in diploids. STE3 expression did not reflect the repressed diploid-like gene expression program that would be expected if mating-type information from HMR were expressed (Fig. 2, A and B).

Fig. 2 Mating-pathway genes did not show diploid-like RNA expression in old cells.

(A) Schematic of mating-factor receptor regulation for a MATα haploid strain. (B) Reverse transcription quantitative polymerase chain reaction (RT-qPCR) for the Sir2-regulated HMRa1 mRNA and the mating type–regulated STE3 mRNA. To account for difficulty in handling small numbers of cells (106 to 107 old cells were prepared for RT-qPCR), we restricted our analysis to direct comparison of age-matched samples for which the RNA preparation conditions were directly comparable. P values were calculated using a two-tailed t test. Results are given relative to those for ACT1 and shown as the sum of the replicates ± SD. (C) Reanalysis of published RNA sequencing data. At left, young cells from two different publications (11, 12) showed repeatable RNA expression. Red dots indicate genes that are regulated by Sir2, as defined by Ellahi et al. (11). Plotted at center is the fold change in expression between old cells and matched young cells from Sen et al. (12) and between sir2∆ cells and a matched wild-type control strain from Ellahi et al. (11). At right, the subset of genes from the center panel that were identified as Sir-regulated by Ellahi et al. are plotted. r, correlation coefficient.

Although the HML and HMR loci are the premier context for studying Sir-based transcriptional repression, RNA sequencing of cells lacking Sir2 (sir2∆) has identified the full complement of genes that are subject to repression by Sir2 (11). Separately, RNA sequencing data have been published from matched young and old cells, including RNA expression data from these Sir2-regulated loci (12). We reanalyzed these data to ascertain whether Sir2-regulated genes show age-associated changes in transcription that could reflect loss of Sir2 function. Genome-wide, we found no evidence that Sir2-dependent gene regulation was related to age-dependent gene regulation (Fig. 2C). Furthermore, telomeric open reading frames repressed directly by Sir2 showed no evidence of an age-dependent increase in transcription (fig. S2). In short, we found no evidence that the transcriptional program in sir2∆ cells was similar to that in old yeast cells.

Having shown that HML and HMR were silenced during aging, we reinvestigated the sterility phenotype reported for old cells (3, 13). We treated young and old MATa cells with various amounts of α-factor pheromone and monitored their ability to arrest in G1 and grow a mating projection (Fig. 3A). Previous experiments that identified an age-associated mating defect used assays sensitized with a low concentration (less than 20 ng/ml) of α-factor (3). Indeed, we found that old mother cells (mean age, 14.3 divisions) responded less efficiently to mating pheromone than did young cells; however, they responded efficiently with pheromone concentrations above 20 ng/ml (Fig. 3B). If the observed loss of mating depended on expression from HML, then deletion of HML would have restored the sensitivity of old cells to the level of young cells, which was not the case (Fig. 3C). Young and old yeast cells in which HMLα2 was deleted were more responsive to α-factor than were wild-type cells (Fig. 3C and fig. S3). Arrest with α-factor did not affect the stability of silencing at HML, indicating that the heightened sensitivity in hmlα2∆ mutants did not reflect transcription from HML during α-factor treatment (fig. S4).

Fig. 3 Old cells required a higher pheromone dose than young cells to form a mating projection.

(A) Schematic of the experimental approach (YPD, yeast extract, peptone, and dextrose). (B) Young and old (on average, 14 divisions old) MATa cells (yYB4172) were purified from 2- and 20-hour cultures, respectively, and their response to pheromone was assayed on agar pads containing indicated α-factor concentrations. The fraction of cells not responding to α-factor at 10 ng/ml increased with age; however, all cells responded to higher concentrations of α-factor. (C) Young and old cells of the yYB6829 (hml∆) strain were tested for pheromone response to α-factor (10 ng/ml). Both wild-type and hml∆ mutant cells lost pheromone sensitivity to a similar extent with age; however, hml∆ cells were more sensitive to pheromone than the corresponding wild-type cells were [yYB4172; data from (B) are repeated for comparison]. All the plots show mean values ± SEM; dots represent independent experiments (n ≥ 30 cells). P values were calculated using a two-tailed t test.

Efficient response to mating pheromone depends on arrest in the G1 phase of the cell cycle. Cells exposed to α-factor for longer than 4 hours escape this cell cycle arrest and become less sensitive to pheromone (14). This adaptation depends on aggregation and subsequent inactivation of Whi3, an RNA-binding S-phase inhibitor, but desensitized mothers produce daughters that are fully sensitive to α-factor (14). Old cells showed a similar asymmetric inheritance of mating competence: The daughters of old cells were more responsive to α-factor than were their mothers (Fig. 4A). To test whether aggregation of Whi3 might explain why old yeast mother cells fail to respond to α-factor, we deleted a glutamine-rich domain required for Whi3 aggregation in cells adapted to α-factor and assayed old MATa cells carrying this deletion for α-factor responsiveness. Deletion of the Whi3 glutamine-rich domain decreased the loss of sensitivity in old cells, indicating that Whi3 aggregation may prevent mating in old cells (Fig. 4B). Live-cell imaging of old yeast mother cells expressing a GFP-tagged Whi3 indicated that old yeast cells did form aggregates of Whi3 (Fig. 4, C and D, and fig. S5). Interestingly, whi3-∆polyQ strains lived slightly longer than wild-type strains, suggesting that aggregation of Whi3 might limit life span (Fig. 4E).

Fig. 4 Formation of Whi3 aggregates contributes to the loss of pheromone sensitivity with age.

(A) The left panel shows an example sequence of an old mother cell (green star) and its progeny exposed to α-factor (10 ng/ml). The old mother cell buds instead of responding to pheromone, but its daughters arrest in G1 and form mating projections. Scale bar, 5 μm. Shown on the right is the quantification (mean values ± SEM) of the pheromone response of the first three daughters of pheromone-insensitive old yYB4172 mothers from Fig. 3B. (B) Old and young MATa cells were exposed to α-factor (10 ng/ml). Pheromone insensitivity increased with age in the wild-type strain (3.2-fold between young and old cells), whereas this effect was reduced in whi3-pQ cells (1.9-fold increase). Young cells were about five divisions old, and old cells were between 15 and 20 divisions old. Bars show mean values ± SEM (n > 200 total young cells, n > 170 total old cells). P values were calculated using an unpaired two-tailed t test. (C) Whi3 forms aggregates in old cells. Scale bars, 5 μm. (D) Quantification of the fraction of young and old cells containing Whi3-3GFP aggregates. Each dot represents an independent experiment, bars represent means, and the P value is from a one-tailed t test. (E) Survival curves of wild-type and whi3-∆pQ strains (n = 58 wild-type yYB14326 cells, n = 78 whi3-∆pQ yYB14325 cells). Deletion of the glutamine-rich domain of Whi3 extends life span. The P value was calculated using the log-rank (Mantel-Cox) test.

Our conclusions regarding whether aging affects gene silencing contrast with those of previous work, but the methods that we used were more sensitive and extensive than those available in the past. Although it would be best to repeat analyses with exactly the same strains used previously, those strains have been lost over the years, precluding a direct comparison. Nevertheless, our data establish that age-dependent loss of gene silencing is not a feature of widely used budding yeast strains.

The mechanism by which HML-deleted yeast cells have slightly decreased sensitivity to mating pheromone, independent of transcription at the locus, is unclear. Interruption of the silent HML locus could have an indirect effect on mating-factor sensitivity, perhaps by inducing changes in the three-dimensional architecture of chromosome III that affect expression of genes involved in the α-factor response.

Both yeast and vertebrates are rich in RNA-binding proteins containing low-complexity prion-like domains. Unlike aggregates of typical yeast prions, Whi3 aggregates are sequestered in the mother cell during cell division (14). Aggregation during aging may be an intrinsic liability for yeast memory factors (mnemons) such as Whi3 that encode memory in the form of protein aggregates. In this view, aging-induced aggregation of Whi3 would preclude α-factor–induced Whi3 aggregation as a memory of past unsuccessful mating encounters. It is tempting to speculate that in nature, yeast could benefit from a bona fide differentiation between old and young cells, with some aggregates being beneficial and others not. Understanding why Whi3 aggregates form and are retained in the mother cell during mitosis may shed light on how protein aggregation influences the mitotic inheritance of cytoplasmic factors more broadly.

Supplementary Materials

Materials and Methods

Figs. S1 to S5

Table S1

References (1522)

Movies S1 and S2

References and Notes

Acknowledgments: We thank M. Delarue, N. Azgui, and the University ofCalifornia–Berkeley biotechnology nanofabrication facility for assistance in fabricating microfluidic devices; E. Unal and G. Brar for microscopy support; S. Guetg for providing the hml::GFP reporter; S. S. Lee for advice on microfluidics; the light microscopy center of ETH Zürich (ScopeM); T. Schwarz for technical support; and T. Kruitwagen for critical reading of the manuscript. RNA sequencing data from Ellahi et al. (11) are available from the National Center for Biotechnology Information (NCBI) under accession numbers SRX884375, SRX885291, SRX885292, SRX885297, SRX885304, and SRX885305, and RNA sequencing data from Sen et al. (12) are available from NCBI as series GSE65767. This work was supported by a grant from the National Institutes of Health (GM31105 to G.S. and J.R.), ETH Zürich and the European Research Council project BarrAge (to Y.B.), and an iPHD fellowship from (to M.R.K.).

Stay Connected to Science

Navigate This Article